Draft
Regulatory Impact Analysis
for the Final Rulemaking on Corrective Action
for Solid Waste Management Units
Proposed Methodology for Analysis
Office of Solid Waste
U.S. Environmental Protection Agency
March 1993
Printed on Recycled Paper
-------
PREFACE
This Regulatory Impact Analysis was prepared by ICF Incorporated (ICF) under the
direction of the Regulatory Analysis Branch, Office of Solid Waste, U.S. Environmental
Protection Agency (EPA). Industrial Economics Incorporated (lEc) wrote Chapters 6, 9, 10, and
12. The contingent valuation survey presented in Chapter 10 was conducted by Dr. Gary H.
McClelland, Dr. William Schulze, and other researchers at the Center for Economic Analysis of
the University of Colorado. Chapter 11 was written by Abt Associates.
The Permits and State Programs Division of the Office of Solid Waste directed the
remedy selection expert panel process and developed much of the supporting analysis for the
simulation of remedy effectiveness. EPA's Office of Research and Development provided
assistance in model selection and development and in development of the risk assessment
methodology.
-------
TABLE OF CONTENTS
Page
EXECUTIVE SUMMARY
1. INTRODUCTION 1-1
1.1 The Need for Regulation 1-1
1.2 Description of Corrective Action Regulatory Impact Analysis 1-5
1.3 RIA Organization 1-6
2. REGULATORY OPTIONS 2-1
2.1 Description of the Baseline 2-1
2.2 Description of the Subpart S Proposed Rule 2-2
2.3 Other Options to be Examined 2-3
3. SAMPLE SELECTION, FACILITY CHARACTERIZATION, AND MODELING OF
RELEASES 3-1
3.1 Approach 3-1
3.2 Results 3-27
3.3 Limitations 3-39
4. REMEDY SELECTION AND MODELING OF REMEDY EFFECTIVENESS ... 4-1
4.1 Background 4-1
4.2 Approach 4-2
4.3 Results 4-14
4.4 Limitations 4-22
5. COSTS 5-1
5.1 Approach 5-1
5.2 Results 5-7
5.3 Sensitivity Analysis 5-23
5.4 Largest Federal Facility Cost Analysis 5-29
5.5 Limitations 5-32
6: OVERVIEW OF BENEFITS 6-1
6.1 Potential Benefits 6-1
6.2 Studies Conducted for this RIA 6-6
7. HUMAN HEALTH BENEFITS 7-1
7.1 Background 7-1
7.2 Approach 7-4
7.3 Results 7-27
7.4 Limitations 7-56
-------
TABLE OF CONTENTS (CONTINUED)
8. ECOLOGICAL BENEFITS 8-1
8.1 Approach 8-1
8.2 Results 8-6
8.3 Discussion 8-14
8.4 Limitations 8-14
9. AVERTED WATER USE COSTS 9-1
9.1 Economic Framework 9-2
9.2 Analytic Approach 9-8
9.3 Results 9-15
9.4 Limitations 9-29
10. NONUSE BENEFITS OF GROUND WATER REMEDIATION 10-1
10.1 Approach 10-2
10.2 Results 10-29
10.3 Sensitivity Analyses 10-31
10.4 Limitations 10-36
11. RESIDENTIAL PROPERTY ANALYSIS 11-1
11.1 Expected Linkages Between TSDFs and Residential
Property Markets 11-1
11.2 Statistical Specifications and Data Employed in
Current Analysis 11-5
11.3 Overview of Findings 11-10
11.4 Conclusions 11-17
12. CHANGES IN THE VALUE OF FACILITIES 12-1
12.1 Economic Framework 12-2
12.2 Analytic Approach 12-7
12.3 Example of Facility Value Benefit Calculations 12-15
12.4 Results 12-17
12.5 Sensitivity Analysis 12-20
12.6 Limitations 12-22
13. COMPARISON OF BENEFITS AND COSTS 13-1
13.1 Introduction 13-1
13.2 Review of Costs and Benefits 13-2
13.3 Comparison of Costs and Benefits 13-6
13.4 Limitations 13-9
BIBLIOGRAPHY
-------
TABLE OF CONTENTS (CONTINUED)
Page
APPENDIX A. DEVELOPMENT OF FACILITY SAMPLE A-l
A.I Creating the Federal Frame A^
A.2 Creating the Non-Federal Sample Frame A"3
A.3 The Full Corrective Action Sample A'8
APPENDIX B. PREDICTING RELEASES AND EXPOSURES WITH MMSOILS B-l
B.I Model Selection B'J
B.2 Parameter Selection and Assumptions B'3
B.3 Model Application Assumptions and Limitations B-13
B.4 Post-MMSOILS Processing B'17
APPENDIX C. SIMULATION OF REMEDY EFFECTIVENESS C-l
C.I Source Control Technologies C-1
C.2 Waste Treatment Technologies c'3
C.3 Ground-Water Remediation Technologies c'7
APPENDIX D. COST ANALYSIS D-l
D.I Expert Panel Cost Estimation D-1
D.2 Additional Analysis of Results D'13
APPENDIX E. HUMAN HEALTH BENEFITS ANALYSIS E-l
E.1 Hazard Identification and Dose-Response Assessment E-l
E.2 Exposure Analysis
E.3 Risk Characterization
APPENDIX F. ECOLOGICAL THREATS: METHODOLOGIES AND
CASE STUDIES F'J
F.I Methodology for Proximity Analysis F'1
F.2 Methodology for Deriving Screening Ecological
Benchmark Levels F-1
F.3 Methodology for Estimating Extent of Contamination F-8
F.4 Proximity Analysis Results F'9
F.5 Concentration-Based Screening Analysis Results F-9
F.6 Time Sequence Results F'9
F.7 Extent of Contamination Results F"9
F.8 Qualitative Case Studies F'17
APPENDIX G. KEY PARAMETERS MATRIX G~l
APPENDIX H. FACILITY AND SWMU DATA FORMS H-l
APPENDIX I. CHARACTERISTICS OF FACILITY AND SWMU
POPULATIONS j'J
I.I Facility Characteristics *'*
1.2 SWMU Characteristics M7
-------
EXECUTIVE SUMMARY
ES.l Background
The purpose of this document is to present a methodology that can be used to estimate
the costs and benefits of site cleanup at hazardous waste treatment, storage, and disposal
facilities regulated under the Resource Conservation and Recovery Act (RCRA). This document
represents the culmination of several years of intensive methodology design and testing work by
the Agency. Although this study evaluates the Subpart S Corrective Action proposal (described
in more detail below), this is only for illustrative purposes. The primary focus and purpose of this
document has been to develop a sound and comprehensive framework that will allow EPA to
evaluate the costs and benefits of various options for approaching site cleanups.
Future work will use the methodology, described in this document, to analyze a wide
range of regulatory alternatives as they develop. It is important to emphasize that there is an
active debate underway among EPA, States, environmental groups, and industry concerning the
RCRA cleanup program and the approach EPA should take in developing final regulations. The
outcome of this policy dialogue and the comments that EPA received on the proposed regulation
will ultimately shape the range of regulatory options that will be considered in the future using
this methodology. The next part of the RIA will reflect comments received on the methodology
and will examine the costs, economic impacts, and benefits of corrective action regulatory
alternative as they emerge.
Following this overview, the Executive Summary presents a brief discussion of regulatory
options in Section ES.l, an overview of the methodology in Section ES.3, results in Section ES.4,
and important limitations to the analysis in Section ES.5.
E.S.I. 1 Nature of the Problem and Legislative/Regulatory History
Prior to the development of stringent hazardous waste management requirements under
the Resource, Conservation, and Recovery Act (RCRA) (42 U.S.C. UU6901 - 6922k), industrial
facilities often managed hazardous wastes and constituents in ways that would not be permissible
currently. These early management practices often resulted in releases to environmental media
resulting in contamination of soils, ground water, surface water, and air. Consequently, many
RCRA treatment, storage and disposal facilities may have "old wastes" (i.e., wastes previously
land disposed) and contaminated media that present current or potential threats to human health
or the environment.
Congress provided authority under RCRA for EPA to promulgate regulations to address
the problems associated with the improper disposal of hazardous wastes and constituents. The
resulting regulations, under 40 CFR Part 264, Subpart F, included requirements for monitoring
and remediating on-site releases to ground water from regulated hazardous waste management
units. In 1984, Congress enacted the Hazardous and Solid Waste Amendments (HSWA), which
significantly expanded corrective action requirements. Sections 3004 (u) and (v) of HSWA
require corrective action for both on-site and off-site releases to all environmental media from all
solid waste management units (SWMUs) at RCRA treatment, storage and disposal facilities.
These requirements were codified in regulations promulgated in 1985 (50 FR 28747) and 1987
(52 FR 45798).
On July 27,1990 (55 FR 30798), EPA proposed the corrective action Subpart S rule.
The proposed rule was developed to replace the existing HSWA rules with a detailed regulatory
** DRAFT - March 26,1993 ***
-------
ES-2
program for implementing corrective action. The Agency is currently engaged in a public debate
on RCRA's cleanup program and will ultimately develop a final rule for corrective action. An
RIA to estimate the costs and benefits of various corrective action regulatory options is a key
element in the development of the final rule. The RIA is discussed in more detail in the next
section.
On February 16,1993 (58 FR 8658), EPA promulgated the Corrective Action
Management Unit (CAMU) and Temporary Unit (TU) final rule. This rule introduced the
concepts of CAMUs and TUs n two new units that will be used for remedial purposes under
RCRA corrective action authorities. The rule defines a CAMU as an area within a facility
designated for the management of remediation wastes generated during the implementation of
specific corrective action requirements, and a TU as a temporary tank and/or container storage
area used solely for the treatment or storage of hazardous remediation wastes during specific
remediation activities. A separate regulatory impact analysis examined the cost savings and
potential effects on human health and the environment associated with the CAMU/TU final rule
[RCRA Docket No. CASF-93J. Due to the timing of the RIA, that portion of the Subpart S
RIA being provided for public comment today reflects the CAMU definition included in the
Subpart S proposed rule, not the definition included in the CAMU/TU final rule.
E.S. 1.2 Development of the RIA
There are significant public concerns associated with the cleanup of RCRA facilities.
Some of the more prominent issues include:
the large universe of facilities with potential releases to the environment, as well
as the threats posed by these releases;
the speed at which cleanups are conducted;
the cost of cleanups at many sites; and
the effectiveness of remediation techniques in achieving cleanup goals in some
circumstances.
Recognizing these concerns, EPA elected to prepare a major RIA to estimate the costs and
benefits of various Subpart S regulatory options and to obtain public comment on them.1
The RIA has used "state of the art" methods to assess the costs and benefits of corrective
action cleanups. This multi-disciplinary approach has employed a broad spectrum of
environmental scientists, economists, and engineers:
extensive involvement of EPA and State implementers, and remediation experts to
help simulate the corrective action remedy selection process;
development, by modeling experts, of a multi-media screening model to estimate
releases and risks;
1 Regulatory impact analyses, mandated by Executive Order 12291, are required for "major"
regulations. Major regulations are defined as those likely to result in (1) annual effects on the
economy of $100 million or more; (2) a major increase in costs or prices for consumers or individual
industries; or (3) significant adverse effects on competition, employment, investment, productivity,
innovation, or international trade. The ultimate version of this RIA will be published with the final
rule and will meet the requirements of Executive Order 12291.
** DRAFT - March 26,1993 **
-------
ES-3
active participation of EPA's risk assessors in ORD to assist with implementation
of recent Agency guidance to better characterize uncertainty in risk assessments;
research by university based economists to develop new and innovative measures
of corrective action benefits; and
use of EPA's geotechnical experts to assess the effectiveness of cleanup
technologies, a research area in which the availability of data is very limited.
These aspects of the methodology are discussed in more detail in the chapters that follow the
executive summary.
The remainder of this executive summary presents EPA's proposed methodology,
highlights primary results, and presents limitations of the RIA.
ES.2 Analytic Approach
ES.2.1 Baseline
The baseline of this analysis represents the requirements under RCRA and other Federal
laws that were in effect prior to the enactment of HSWA. The baseline can also be considered
the "no action" option, since it represents conditions as they would exist with no remediation.
The costs and benefits of other options will be compared with the baseline in the future.
Under the baseline, EPA assumed that land disposal units that received hazardous waste
after July 26,1982 would be remediated under the existing authority of Subpart F of 40 CFR
Part 264. In addition to the existing RCRA cleanup requirements, EPA assumed for this RIA
that certain other Federal requirements, such as those under the Comprehensive Environmental
Response Compensation and Liability Act (Superfund), would be included in the baseline. As a
result, the costs and benefits of complying with these other requirements at facilities subject to
the RCRA requirements are not attributed to the RCRA corrective action program under
HSWA. EPA used the baseline as a starting point for estimating the costs and benefits2 of the
Subpart S proposed rule, which is discussed below.
ES.2.2 Corrective Action Subpart S Proposed Rule
The corrective action rule proposed on July 27,1990 under Subpart S of Part 264
outlined a comprehensive regulatory framework for implementing EPA's corrective action
program under RCRA. The Agency evaluated the proposed corrective action rule by predicting
releases from SWMUs of hazardous waste or hazardous constituents and estimating the costs and
benefits of various options to address those releases. For the purposes of the RIA, EPA
assumed that the Subpart S proposed rule would apply to both permitted and interim status
facilities and that CAMUs would be employed, as appropriate, based on the CAMU provisions
in the proposed rule.3
2 However, EPA did not estimate the costs or benefits of the baseline.
3 The CAMU/TU was published subsequent to completion of most of the analysis for this RIA.
As a result the RIA reflects the CAMU presented in the proposed Subpart S rule, rather than that
in the CAMU/TU final rule.
** DRAFT - March 26,1993 **
-------
ES-4
ES.2.3 Other Options to be Examined Later
This analysis uses the proposed Subpart S rule to showcase the RIA methodology; this
document does not examine other regulatory options. EPA will analyze several regulatory
options and provide the analysis for public comment in the next phase of this RIA. Broad
categories of options the Agency may consider in the next phase include:
The effect of varying the point of compliance where cleanup standards must be
achieved for each medium (e.g., at the unit boundary or facility boundary).
Varying cleanup standards that remedies must achieve (e.g.,to background levels
versus health-based levels). These media cleanup standards represent constituent
concentrations in ground water, surface water, soils, and air.
Implementation of options which would change the implementation strategy for
the program; particularly regarding when and how we address the greatest risks.
For example, remedies that stabilize conditions and minimize risks at facilities in
the short-run.
ES.3 Approach
ES.3.1 Approach to Sample Selection, Facility Characterization, and Modeling of
Releases
As shown in Exhibit ES-1, the first step in development of the RIA was selecting a
sample of facilities for analysis. The sample of facilities was drawn from two groups of facilities:
a Federal facility group and a non-Federal facility group. EPA constructed separate frameworks
for sampling the two groups, using different sampling designs and sample allocations. The final
sample of 79 facilities was comprised of 9 Federal and 70 non-Federal facilities. The Agency
employed a stratified random sample in developing the facility sample in order to avoid bias and
to maximize the precision of the population estimator.
The RIA estimated the costs and benefits of the proposed Subpart S corrective action
rule at the facility level and extrapolated them to the national level. The cost and benefit
analyses required an extensive characterization of each of the facilities in the sample. It was
necessary to characterize a number of factors at each sample facility, including the history of
operations at the site, environmental setting, and extent of existing contamination at the site. To
the extent possible, EPA relied on information from facility-specific reports and databases to
characterize the extent of existing contamination. Where facility-specific data were not available,
EPA relied on other data sources and on professional experience.
Characterizing the short-term extent of contamination and modeling the long-term extent
of contamination were the next steps in conducting the RIA, as shown in Exhibit ES-1. EPA
assessed the current extent of contamination in environmental media (ground water, air, surface
water, and soil) in order to select corrective action remedies and estimate national costs. The
Agency assessed the current extent of contamination using monitoring data where available.
Where monitoring data were not available, and for the purpose of projecting the future spread of
contamination in the short-term, EPA used the MMSOILS model, a multi-media fate and
transport screening model.
DRAFT - March 26,1993 ***
-------
EXHIBIT ES-1
MAJOR RIA STEPS
SELECT SAMPLE OF FACILITIES
CHAPTER 3
CHARACTERIZE AFFECTED FACILITIES
CHAPTER 3
ESTIMATE CURRENT EXTENT
OF CONTAMINATION WITHOUT
CORRECTIVE ACTION
CHAPTER 3
REPLICATE CORRECTIVE
ACTION PROCESS AND
SELECT REMEDIES
CHAPTER 4
PREDICT FACILITY-LEVEL
COSTS AND ECONOMIC
IMPACTS'
CHAPTERS 5,15
ESTIMATE LONG-TERM EXTENT
OF CONTAMINATION
WITHOUT CORRECTIVE ACTION
CHAPTER 3
ASSESS EFFECTIVENESS OF
CORRECTIVE ACTION REMEDIES
IN REDUCING CONTAMINATION
CHAPTER 4
PREDICT FACILITY-LEVEL
BENEFITS
CHAPTERS 7-12
ANALYZE
* OTHER REGULATORY
OPTIONS*
PREDICT NATIONAL COSTS,
ECONOMIC IMPACTS*, AND
BENEFITS
ANALYZE
OTHER REGULATORY
OPTIONS*
(32004-1
ECONOMIC IMPACTS AND OTHER REGULATORY OPTIONS TO BE ANALYZED LATER
-------
ES-6
In order to assess the effectiveness of remedies and estimate benefits, EPA used
MMSOILS to project the extent of contamination over the long term, i.e.,over a 128-year
modeling period. To capture some of the uncertainty in the long-term assessment and comply
with recent EPA risk assessment guidance, the Agency used central tendency and high-end
assumptions for critical waste, release, and fate and transport parameters.4
ES-3.2 Approach to Remedy Selection and Modeling of Remedy Effectiveness
After characterizing the short-term extent of contamination, EPA replicated the
corrective action process and selected remedies, as shown in ES-1. Remedy selection is the step
in the corrective action process in which EPA or an authorized State selects final remedies for
those RCRA facilities requiring cleanup. Actual remedies have been selected at few RCRA
facilities nationwide to date. Lacking facility-specific remedy selection information, the Agency
had to develop a methodology for forecasting the kinds of remedies that would be selected under
the Subpart S proposed rule, in order to estimate the costs and benefits of the rule.
In order to simulate the interactions between EPA and a facility owner/operator, the
Agency developed an innovative approach to predicting remedies and estimating the costs of
corrective action. This approach involved convening two kinds of expert panels: a policy panel
consisting of EPA Regional and State representatives with experience implementing the existing
RCRA corrective action program; and a technical panel consisting of private sector, highly
experienced hazardous waste remediation experts. The technical experts were charged with
developing one or more technical remedies for each facility and estimating the costs of the
remedies. The policy panel represented the role of the regulatory agency in requesting technical
information from the technical panel and in making final remedy selection decisions. The
technical panel represented a facility owner/operator's interest in .developing cost-effective
remedies that would meet the Subpart S regulatory objectives.
The expert panels followed a multi-step procedure in simulating the Subpart S remedy
selection process. First, the Agency presented the panel members with data on the
characteristics of and extent of contamination at the facility and the SWMUs, as well as maps of
the facility. Next, the policy panel reviewed the facility data and developed remedial objectives
for remediating each SWMU of concern (i.e., each SWMU with current or potential releases) at
the facility and all environmental contamination. The completed facility remedy objectives were
presented to the technical panel, which then developed detailed technical options for remediating
the facility based on these objectives. After developing the technical alternatives, the policy
panel reviewed the proposed remedies and made the final remedy selection decision.
After completing the remedy selection process, the Agency simulated the effectiveness of
the selected remedies (see Exhibit ES-1). Modeling of remedy effectiveness, in turn, was
necessary to estimate the benefits of the Subpart S proposed rule. There is significant
uncertainty regarding the long-term effectiveness of corrective action remedies. Consequently,
EPA used available data and expert judgement to develop assumptions about the effectiveness of
4 Data and assumptions for the high-end scenario, while generally resulting in greater constituent
mass available for release to all media pathways, may in some cases inadvertently act to reduce the
mass available to pathways other than ground water. Consequently, the high-end results for surface
water, soils, foodchain, and air pathways that are presented in the human health benefits and
ecological threats analyses may not represent a realistic high-end scenario for those pathways.
*** DRAFT - March 26,1993
-------
ES-7
source control, waste treatment, and ground-water remediation technologies.5 The purpose of
the remedy effectiveness simulation was to assess how effective the remedies would be in
reducing the long-term extent of contamination in environmental media (predicted using the
MMSOILS model) that would occur in the absence of corrective action. Assumptions about
remedy effectiveness were therefore incorporated in the MMSOILS model in order to predict
revised estimates of the extent of contamination with remedies in place.
ES.3.3 Approach to the Cost Analysis
EPA based its corrective action cost estimates for facilities on the remedial cost estimates
.provided by the expert panels. After reviewing each facility in the RIA sample and selecting
remedies to address the contamination at those facilities, the expert panels prepared cost
estimates for each remedial activity. The panels prepared cost estimates for the capital,
operation and maintenance, and investigatory components of each remedial activity. EPA
adjusted the total costs to include design, oversight, and contingency assumptions, and then
discounted the costs using a seven percent discount rate* to account for the timing of
remediations. The cost results from the sample facilities were then extrapolated to the national
level using weights assigned to each sample facility.
ES.3.4 Approach to the Benefits Analyses
EPA estimated the national benefits of corrective action, after modeling the long-term
extent of contamination and simulating the effectiveness of remedies in reducing this
contamination, (see Exhibit ES-1). Benefit measures include human health risk reduction,
averted water use costs, nonuse benefits, and increases in facility values. In addition, EPA
estimated ecological risk under baseline conditions and effects of'facilities on residential property
values. These measures are described below.
Approach to the Human Health Benefits Analysis
EPA estimated baseline human health risks and risk reductions attributable to the
corrective action rule by analyzing detailed facility information, modeling environmental transport
of contaminants to human exposure points using the MMSOILS model, and applying standard
human health risk assessment procedures to estimate cancer risk , noncancer health effects, and
effects from exposure to lead. The Agency conducted the risk assessment in accordance with the
National Academy of Sciences risk framework7 and recent Agency risk assessment guidance.
5 EPA assumed that institutional controls would not be implemented, as discussed below under
the human health benefits approach.
6 "Guidelines and Discount Rates for Benefit-Cost Analysis of Federal Programs". Circular A-
94. Office of Management and Budget. Washington, D.C.: Office of Management and Budget.
October 29, 1992.
'National Academy of Sciences. 1QRV Bisk Assessment in the Federal Government. National
Academy Press, Washington, D.C.
8 U.S.EPA, Risk Assessment Council. Memorandum entitled "Guidance on Risk
Characterization for Risk Managers and Risk Assessors." February 26, 1992.
DRAFT - March 26,1993 **
-------
ES-8
EPA's guidance recommends the use of multiple risk descriptors to characterize the
uncertainty in risk estimates. For the human health benefits analysis, the Agency used all of the
fundamental risk descriptors recommended in the guidance:
Central tendency individual risk,
High-end individual risk,
Individual risk to highly-exposed or sensitive populations, and
Population risk.
The risk descriptors are defined separately for cancer risks and noncancer effects.9 EPA
estimated risks for actual potentially-exposed off-site populations. In addition, EPA estimated
hypothetical on-site individual risks under the assumption that land use would change to
residential or agricultural uses10.
EPA estimated facility-specific risks and extrapolated the results to the national level.
The analysis of baseline risk (i.e., risk in the absence of corrective action) was conducted for all
2,600 facilities projected to require corrective action.11 EPA's analysis of post-remedial risk
focused on the estimated sub-set of 720 facilities projected to have the greatest risk/health effects
of concern or extensive contamination in the baseline. For the 720 facilities, post-remediation
risks were subtracted from baseline risks to estimate benefits in terms of risk reduction.
Due to uncertainties regarding the long-term effectiveness of institutional controls and
engineered remedies, the Agency chose to analyze a "less than 100 percent effectiveness" scenario
in the human health benefits analysis. This scenario relies on estimates of actual effectiveness of
engineered remedies and assumes that institutional controls are not implemented. The Agency
examined this scenario in order to measure the long-term effectiveness of different engineered
remedies in isolation and to assess the potential risks that could result in situations where
engineered remedies fail and institutional controls are not effective and enforceable.12
U.S.EPA, "Guidelines for Exposure Assessment; Notice," FR 22888, May 29,1992.
9 Individual cancer risks are expressed as the probability above background of developing cancer
during a lifetime due to exposure. The population cancer risk describes the number of excess
statistical cancer cases expected in the exposed population over the 128-year modeling period.
Individual noncancer effects are expressed as the hazard index, i.e.,the ratio between the exposure
dose and the effects threshold (i.e., the chronic reference dose). Population noncancer effects are
the number of people over the modeling period with a hazard index greater than one.
10 The expert panels projected that a majority of facilities requiring corrective action would be
likely to change to residential and/or agricultural use in the future. Based on this projection and
given the uncertainty in projecting future land uses, the Agency assumed that all facilities would
potentially change to residential/agricultural used in the future.
11 Results of the remedy selection process are discussed in Section ES.4.1.
12 There is significant uncertainty in the long-term maintenance and enforceability of institutional
controls. This analysis does not reflect the current debate about what constitutes effective and
enforceable institutional controls.
* DRAFT - March 26,1993
-------
ES-9
Approach to the Ecological Benefits Analysis
The Agency conducted ecological risk assessments at a number of facilities subject to the
RCRA corrective action authorities in order to characterize ecological risks under baseline
conditions. The risk assessments were performed in three steps. The first step was a proximity
analysis to characterize environmental settings and potential ecological receptors and to prioritize
facilities for further evaluation. This was accomplished by determining which of the facilities
appeared to be near valuable or vulnerable ecological resources. The second step of the risk
assessment was a concentration-based screening analysis to identify facilities at which projected
concentrations of hazardous substances in surface waters exceed ecological benchmark levels (i.e.,
ambient concentrations above which adverse ecological effects are expected). The third and final
step of the analysis was a more complete ecological risk assessment for three sample facilities to
identify types of risks not accounted for by the other analyses.
Approach to the Averted Water Use Cost Analysis
The remediation required by the corrective action rule would reduce water supply costs in
those cases where the rule resulted in the prevention or reduction of the contamination of
drinking water supplies. In the absence of corrective action, water users would need to treat or
replace the contaminated supply. The Agency used the savings associated with these averted
water use costs to estimate the benefits of the corrective action rule.
EPA analyzed the potentially averted consumer expenditures on water treatment and
replacement of water sources and calculated the resulting changes in consumers' surplus as a
measure of the welfare effects, or benefits, of those savings. The .Agency extrapolated the results
of this analysis to estimate national monetized benefits.
Approach to the Analysis of Nonuse Benefits of Ground-water Remediation
Nonuse benefits of the corrective action rule derive from the values people place on
natural resources unrelated to their own use of those resources. In the case of corrective action,'
nonuse benefits may result from the remediation of ground water, surface water, or soil. In this
RIA, EPA considered only the nonuse benefits from the remediation of ground water, which
include benefits that derive from the knowledge that a given aquifer is in a clean condition and
will remain so for current and future generations (often referred to as altruistic, existence, or
bequest values).
Nonuse values generally do not result in any observable behavior on the part of
individuals. Thus, actions or market transactions cannot be used to infer these values. Instead,
survey research methods (i.e., the contingent valuation methods) are used to elicit nonuse values
from individuals. EPA's approach to estimating nonuse values associated with ground-water
remediation utilized estimates from a recent contingent valuation study of ground water" in
order to value remediation required under the corrective action proposed rule. The Agency
applied estimates generated by the study of households' "willingness to pay" for remediating
13 Gary H. McClelland, William D. Schulze, Jeffrey K. Lazo, Donald M. Waldman, James K.
Doyle, Steven R. Elliott, and Julie R. Irwin. Methods for Measuring Non-Use Values: A
Contingent Valuation Study of Groundwater Cleanup (Draft), prepared for the U.S. Environmental
Protection Agency, Washington, D.C., October 1992.
DRAFT - March 26,1993 ***
-------
ES-10
contaminated ground water to value the nonuse benefits of the corrective action rule. Because
the contingent valuation study did not consider the exact type of ground-water contamination
scenario commonly found at corrective action sites, EPA's calculations constituted a "benefits
transfer" n the application of benefit estimates for a specific nonmarket good to a setting that
differs from that evaluated in the original study. The results of this analysis were extrapolated to
the national level.
Approach to the Residential Property Analysis
EPA estimated economic damages from facilities as expressed in residential property
markets. The Agency used hedonic property value analysis to evaluate the effect of
contaminated facilities on the value of nearby residential properties. The hedonic approach uses
data on all property transactions that take place during a certain time period in an area
surrounding a facility in order to determine the effect of property characteristics (e.g., distance
from a facility) or facility events (e.g., known contaminant releases) on the value of the property.
In this RIA, the results are presented for several sample facilities, but are not extrapolated to a
national level.
Approach to the Site Value Analysis
EPA estimated the benefits that may accrue directly to facility owners due to the site
remediation required under corrective action. These benefits are in the form of increased site
value resulting from remediation of on-site contamination. The Agency employed two scenarios
for increased site values resulting from corrective action. First, under the proposed Subpart S
rule, some facility owners may be required to remediate contamination to levels that make the
property suitable for alternative uses. Any resulting increases in net revenues that accrue to the
owner from this alternative use will usually be a benefit of the rule. Second, corrective action
may affect net revenues where a facility continues to be used as a treatment, storage, or disposal
facility (TSDF), but cleanup increases the useable area on site for current land uses. The
Agency's analysis employed these two scenarios for increased site values, estimating the potential
site value benefits for each sample facility and extrapolating the total benefits to the national
level.
ES.4 Results
ES.4.1 Results of Facility Characterization and Modeling of Releases
The Agency estimates that about 5,800 facilities and 100,000 solid waste management
units (SWMUs) are potentially subject to corrective action. EPA projects that releases to the
environment (i.e., ground water, soil, surface water, and air) that pose a current or potential
threat to human health and the environment would occur at approximately 2,600 (44 percent) of
these facilities. About 80 percent of the 2,600 facilities are projected to have releases to on-site
ground water above action levels, and 30 percent are projected to have releases above action
levels to off-site ground water.14 The maximum areas of on-site and off-site ground-water
14 As discussed in the Subpart S proposed rule, action levels are health and environmental-based
levels determined by the Agency to be indicators for protection of human health and the
environment. Corrective action may be required when concentrations of hazardous constituents
exceed action levels for an environmental medium.
DRAFT - March 26,1993 **
-------
ES-11
contamination at the 2,600 facilities, over the 128-year modeling period, would total about 50,000
acres and 1.8 million acres, respectively. Releases above action levels to on-site soil are
estimated to occur at 68 percent of the 2,600 facilities. Smaller numbers of facilities are
estimated to have releases to surface waters (due to sedimentation during overland transport of
eroded soil and dilution upon release to the surface water) or to air (since many SWMUs are
inactive and the volatile organics that were present in them have already dispersed).
The universe of 100,000 SWMUs at the 5,800 facilities includes approximately 30,000
tanks, 10,000 accumulation areas (e.g., for drum storage), 9,700 surface impoundments, 5,400
landfills, and 2,800 waste piles. About 4,600 of the 100,000 SWMUs are regulated RCRA
Subpart F land disposal SWMUs. About 15,000 of the 75,000 SWMUs located at the 2,600
facilities are estimated to have releases of concern to the environment.
ES.4.2 Results of Remedy Selection and Modeling of Remedy Effectiveness
Based on the projected releases to the environment discussed above, EPA estimates that
2,600 facilities and 15,000 SWMUs would require remediation under the proposed Subpart S
rule. Of the 2,600 facilities, about 2,200 are projected to have releases above action levels to one
or more environmental media. The expert panels specified remedies at the remaining 400
facilities in order to detect and/or prevent potential future releases. Among the 15,000 SWMUs
requiring corrective action, about half of the 5,400 landfills, about 45 percent of the 9,700 surface
impoundments, and about 10 percent of the 30,000 tanks would require remediation.
The post-remediation extent of contamination was estimated for a subset of 720 facilities
that had risks of concern or extensive contamination in the baseline. At these facilities, roughly
80 percent of the 28,000 acres of on-site ground water contamination and 95 percent of the
1,500,000 acres of off-site ground water contamination were projected to be reduced below
action levels during the 128-year modeling period. The median time to remediate contaminated
ground water was estimated to be 115 years for on-site ground-water plumes and 90 years for
off-site plumes. The longer median time to remediate on-site plumes in many cases reflects the
presence of slow-moving ground water or dense non-aqueous phase liquids (DNAPLs), which
impede the effectiveness of ground water remedies. Ground water technologies commonly
employed included french drains and extraction wells.
About 90 percent of the 720 facilities had on-site soil contamination that would be
remediated. EPA estimates that 18 million cubic yards of contaminated soils would be
remediated using a variety of technologies including excavation followed by treatment and
disposal; in-situ treatment (e.g., soil vapor extraction); and capping with clean cover materials.
Predictions of post-remediation reductions in off-site soil and surface water
contamination are uncertain and could potentially range from complete reduction to minimal
reduction. The complete reduction scenario assumes that source control remedies would
generally be effective in reducing the erosion of contaminated soils from SWMUs to surface
waters or off-site soils and that ground water remedies would often eliminate ground water
contaminant discharges to surface waters. Because of current model limitations, MMSOILs
results indicate minimal reductions in off-site soil and surface water contamination. Correction
of these limitations could potentially result in greater estimated reductions in contamination as a
result of remediation.
** DRAFT - March 26,1993
-------
ES-12
ES.4.3 Results of the Cost Analysis
EPA projects a total national present value cost of the Subpart S corrective action
program of about $18.7 billion (in 1992 dollars at a 7 percent discount rate). The Agency
estimates the annual cost (calculated using a 7 percent rate over a 20-year period) at about $1.8
billion. Roughly half of the total cost of corrective action would be incurred by slightly more
than 10 percent of the facilities. Generally, these facilities are very large, with extensive ground-
water contamination, and have a substantial number of SWMUs requiring remediation.
EPA projects that over half of the cost of the corrective action rule would be incurred as
capital costs. The breakdown of the projected total corrective action cost is $10.5 billion for
capital costs, $6.2 billion for operations and maintenance, and $2.1 billion for investigation.
When the costs are broken down by media, ground-water remediation accounts for over 48
percent of the total cost of corrective action. When the costs are broken down by remedial
action, removal/treatment of contaminated media (e.g., soils and ground water) is the most costly
remedial activity, representing 52 percent of the total cost of corrective action.
The estimated corrective action cost per facility for facilities that needed cleanup in the
RIA sample (expressed as a present value) ranges from $0.1 million to $196 million. The
Agency projects a weighted average cost per facility of $7.2 million. The estimated annual cost
per facility ranges from $10,000 to $18.5 million. The weighted average annual cost per facility is
$0.7 million.
The estimated cost of the proposed corrective action rule for federal facilities is estimated
to be $2.1 billion the remaining cost of $16.6 billion will be incurred by privately owned facilities.
When interpreting this distribution of cost there are two driving factors. First, the nine largest
federal facilities are not included in this analysis. This omission alone is not likely to result in a
great underestimate of the costs associated with RCRA Subpart S cleanup. Cleanup at the
largest federal facilities has been moving forward steadily under a number of federal and state
programs that are already in place. For the federal facilities included in the analysis, EPA has
estimated all costs associated with other clean up programs (e.g.,CERCLA) and included those
in the baseline so they are not included as a cost attributable to Subpart S corrective action.
Therefore, EPA does not expect RCRA Subpart S authorities to impose additional costs on
federal facilities beyond those reflected above.
EPA projects the weighted average present value cost per SWMU that needed
remediation at $1.1 million. The estimated cost per SWMU of remediation ranges from $370 to
$58 million. Landfills are projected to have the highest average remediation cost.
Approximately ten percent of the SWMUs incur roughly half of the total corrective action costs.
The chemicals and allied products [Standard Industrial Classification (SIC) 28]" and the
fabricated metals (SIC 34) industries would bear the highest costs of the corrective action rule,
each incurring over 20 percent of the costs of the program. The industry that would incur the
highest average cost per facility is national security [i.e., U.S. Department of Defense (DOD) and
15 Standard Industrial Classifications (SICs) taken from Standard Industrial Classification
Manual. Executive Office of the President, Office of Management and Budget. Washington, D.C.:
Office of Management and Budget. 1987.
DRAFT - March 26,1993 **
-------
ES-13
U.S. Department of Energy (DOE)] facilities in SIC 97]; the weighted average per-facility cost of
remediation in this industry is estimated to be $33 million.16
ES.4.4 Results of the Benefits Analyses
Results of the Human Health Benefits Analysis
Baseline individual risk. Without corrective action, EPA estimates off-site individual
cancer risks or noncancer effects of concern" would occur at from 920 (36 percent) to 1,700(67
percent) of the 2,600 facilities requiring corrective action under central tendency and high-end
assumptions, respectively. Baseline individual risks of concern would result from all pathways
except exposure via inhalation and drinking from a surface water source. For media and
foodchain pathways, the risks/effects of concern at the most facilities would result from exposures
to ground water; from children exposed to soils in off-site fields and agricultural fields; and from
consumption of vegetables. Among the highly-exposed subpopulations, subsistence fanners and
subsistence fishermen would also experience risk/effects of concern at a significant number of
faculties.
In addition to estimating off-site risks, EPA quantified risks to hypothetical individuals
who might reside on current facility sites in the future. Assuming future on-site use, the Agency
predicted baseline cancer risk and noncancer effects of concern would occur at approximately
1,800(71 percent) of the 2,600 facilities.
Baseline population risk. The Agency found that all population risks/effects of concern
were attributable solely to ground-water exposures, mainly at public wells.11 The Agency
estimated population risk via ground water under two scenarios: the first assumed no
testing/treating of public water supplies and no avoidance of water ingestion due to noticeable
taste or odor ("concentrations uncapped"), while the second assumed both testing/treating of
public supplies and avoidance of ingestion based on taste/odor ("concentrations capped").
In the absence of corrective action under the "concentrations uncapped" scenario, the
Agency estimates that approximately 25 million people living near RCRA facilities would
experience adverse health effects and that 21,000 statistical cancer cases would result over the
128-year modeling period. Under the "concentrations capped" scenario, an estimated 900
noncancer effects and 1,200 cancer cases would result during the modeling period. The very
significant difference between the uncapped and capped scenarios results from the fact that
population risk results primarily from exposure via public wells and the assumption that risk-
driving constituents at these wells would generally be tested for. Most of the baseline population
risk is attributable to a single risk-driving facility, with smaller risks contributed by several others.
16 Projected costs exclude those likely to be incurred at the nine very largest DOD and DOE
facilities.
17 For this RIA, the Agency defined a cancer risk of concern as a risk greater than or equal to
one in one million, and a noncancer effect of concern as an exceedance of a noncancer health
effects threshold (i.e., exceedance of a chronic reference dose).
11 Population risks were also calculated for the surface water and air pathways, but, due to lack
of exposure via these pathways, no risks were found.
DRAFT - March 26,1993
-------
ES-14
Baseline Facilities of Concern. EPA estimated that a total of approximately 1,900
facilities (73 percent) would present cancer risks or noncancer effects of concern at RCRA
corrective action facilities under central tendency assumptions. Under high end assumptions,
2,200 facilities (84 percent) would be of concern. These estimates include facilities with off-site
individual risk/effects of concern, facilities with population risk/effects, and facilities with
hypothetical on-site risk/effects of concern. See Exhibit ES-2. Approximately 400 facilities (16
percent) that would be remediated under the Subpart S proposed rule appear not to be of
concern based on the modeling results. The expert panels specified remedies at these facilities in
order to detect and/or prevent potential future releases to environmental media.
Post-Remediation Individual Risk. EPA examined post-remediation reductions in cancer
risk and noncancer effects for a subset of 720 facilities that were found to have significant
risk/effects or extensive contamination in the baseline. Prediction of the post-remediation
reduction in the number of facilities with individual risk/effects of concern is uncertain. This
relates to the uncertain post-remediation levels of off-site soil and surface water contamination,
which affect not only the risk/effects from the soil and surface water pathways, but those from
the foodchain pathway as well. Since a large percentage of facilities were projected to have
individual risk/effects of concern from the soil pathway or the foodchain pathway in the baseline,
assumptions concerning the levels of post-remediation soil and surface water contamination have
a significant effect on post-remediation risk results.19
Post-Remediation Population Risk. EPA estimated ground-water population risk
reductions for two scenarios, one where constituent concentrations were not capped at MCLs or
taste and odor thresholds, and the other where they were capped. For the uncapped scenario,
EPA estimates that there would be a reduction of approximately 13,000 statistical cancer cases
(i.e., about two thirds of baseline cancer cases would be eliminated) and reduction of
approximately 12,000,000persons experiencing noncancer effects (i.e., about half of baseline
noncancer effects would be eliminated). Under the capped scenario, EPA estimates that there
would be a reduction of approximately 400 cancer cases (i.e., about one third of baseline cancer
cases would be eliminated) and reduction of approximately 10 persons experiencing noncancer
effects (i.e., all baseline noncancer effects would be eliminated). Most of the population risk
reductions are associated with public drinking water wells.
The reduction in population risk due to corrective action is less than complete because
EPA assumed that source control, waste treatment, and ground water remedies would generally
be less than 100 percent effective. This is the case at the facility that drives baseline population
risk. The expert panels specified a containment remedy for this facility, features of which (e.g.,
liners) are expected to fail in the future. This facility also has two features that significantly limit
the effectiveness of the ground water remedy: the presence of a karst aquifer and dense non-
aqueous phase liquids (DNAPLs).
Results of the Ecological Benefits Analysis
Ecological risk assessments were conducted at the sample facilities projected to require
corrective action in order to characterize ecological risks under baseline conditions. The effect
of corrective action on reducing ecological risks was not evaluated.
19 Post-remediation extent of contamination results for soil and surface water are discussed in
Section ES.4.2.
DRAFT - March 26,1993 ***
-------
Exhibit ES-2
Percentage of Facilities With Baseline
Cancer Risks and Non-Cancer Effects of Concern
[N = 2,600]
Central Tendency High-End
Risk 61% Risk
64% No Risk
33% No Risk
Facilities With Off-Site Individual
Risk/Effects of Concern '
Central Tendency Only
Risk
29% No Risk
Facilities With Hypothetical On-Site
Individual Risk/Effects of Concern
Central Tendency Only
11% Risk
89% No Risk
Facilities With Off-Site Population
Risk/Effects of Concern
Central Tendency
73% Risk
High-End
84% Risk
27% No Risk
16% No Risk
Facilities With Any Individual or
Population Risk/Effects of Concern
Based on both 1 30- year average and 9- year peak risks
2 Across off-site (1 30-year average and 9- year peakl, on-site, and population risk results
-------
ES-16
The proximity analysis, the first step of the ecological assessment process, characterized
38 of the 52 sample facilities with relatively large acreages of surface water and terrestrial
habitats on-site and within one mile as "higher-risk" (i.e., higher priority for further evaluation).
These higher-risk facilities also tended to be relatively closer to sensitive environments.
Approximately 60 percent (1,500) of the 2,600 facilities in the population likely to require
corrective action would fall in the higher-risk category. The estimated proportion of facilities in
the higher-risk category is greater for Federal facilities (100%) than for non-Federal facilities.
The concentration-based screening analysis indicates that maximum concentrations in
receiving surface waters would exceed ecological benchmark levels at about 144 of the 2,600
facilities under the central tendency scenario and at about 575 of the facilities under the high-end
scenario.30 At one large sample facility, concentrations exceeded benchmark levels under the
central tendency modeling assumptions but not under the high-end modeling assumptions.
Thus, EPA predicts that maximum concentrations would exceed benchmark levels at 578 of the
facilities under at least one modeling scenario.
As mentioned above, the Agency prepared qualitative case studies, the final step in the
ecological assessment, for three facilities in order to identify types of risks that are not accounted
for by the proximity and concentration-based screening analyses. The Agency found that, in
addition to surface-water contamination, sediment and soil contamination are important factors
in determining risk to ecological systems at facilities subject to the corrective action rule.
Results of the Use Value/Averted Cost Analysis
EPA's analysis of averted costs indicates that, in the absence of corrective action, water
use within two miles of the facility could be affected at 358 of the 2,600 facilities potentially
subject to the rule. The Agency predicts that the national benefits associated with averted water
use costs at these facilities could total $5 million, as measured by changes in consumer surplus.
This estimate of national benefits is highly sensitive to the findings for only a few facilities. Two
sampled facility (representing seventy facilities nationally) accounts for over 90 percent of the use
value benefits.
Results of the Analysis of Nonuse Benefits of Ground-water Remediation
Because of questions associated with the reliability of contingent valuation studies for
measuring nonuse values in general, and EPA's application of a particular contingent valuation
study to corrective action sites, the Agency recognizes the uncertainty associated with its
estimates of non-use values. Assuming that the willingness to pay values from the contingent
valuation study are appropriate for application to corrective action ground-water remediation,
EPA estimates that the nonuse benefits of the rule will range from approximately $170 million to
$18 billion, depending on the specific assumptions used. The range of values is driven by two
factors: (1) the number of corrective action facilities assumed to undergo ground-water
20 Predicted exceedences indicate that releases of hazardous substances are likely or highly likely
to result in adverse ecological impacts in nearby surface waters.
21 Data and assumptions for the high end scenario, while often resulting in greater constituent
mass available for release to all pathways, may in some cases inadvertently act to reduce the mass
released to pathways other than ground water. See the discussion in Section ES.3.1. Consequently,
the high-end ecological risk results in some cases may be understated.
** DRAFT - March 26,1993 ***
-------
ES-17
remediation valued by the public, and (2) the number of households assumed to value
remediation of ground water at each of these corrective action facilities. The Agency's estimate
of the most likely value of nonuse benefits is $2.3 billion.
Results of the Residential Property Analysis
For each of the three facilities analyzed, EPA identified statistically significant
relationships between housing prices and distance from the site. Housing prices tend to increase
with distance from the facilities. Changes in housing prices over time appear to be consistent
with the hypothesis that key contamination events can effect housing prices.
The results of this analysis support the assertion that hazardous waste contamination can
impose economic damages that are expressed in residential real estate transactions. At the three
facilities studied, the estimated effect ranged from about $100 to over $18,000 per house,
depending on the facility and regression specification in question.
Results of the Site Value Analysis
EPA projects site value benefits at the facilities that will require some remediation under
die corrective action rule to be about $280 million. Because of uncertainties concerning the
impact of corrective action remediation on the future use of contaminated site areas, the Agency
conducted a sensitivity analysis which suggests the national estimate of benefits could be $610
million. These benefits are dominated by a relatively small number of facilities. On-site
remediation at 470 of the 2,600 facilities produces approximately 80 percent of the national
benefits.
ES.4.5 Results of the Benefit-Cost Analysis
EPA used two common approaches to compare the costs and benefits of the corrective
action rule: benefit-cost analysis and cost-effectiveness analysis. Benefit-cost analysis is a
method of quantifying trade-offs across regulatory alternatives when benefits and costs can be
expressed in a common metric, such as dollars. Cost-effectiveness analysis is used for comparing
costs and benefits that cannot be measured in a common metric. In this RIA, the Agency
employed benefit-cost analysis for the monetized benefits of the rule (i.e., avoided water use
costs, value of the corrective action site, and nonuse values of ground water), and cost-
effectiveness analysis for the non-monetized benefits (i.e., health and ecological risks).
The results of the benefit-cost analysis are presented in Exhibit ES-2. As previously
stated, EPA estimates the total national present value cost of proposed Subpart S corrective
action to be $18.7 billion in 1992 dollars. The majority of the monetary benefits derive from
nonuse values of ground water. The Agency calculated the net benefit of corrective action,
based solely on analyzed economic effects, by subtracting costs from total benefits. The resulting
value is a negative $16.4 billion. This result suggests that total costs are substantially greater than
the total value of monetized benefits. However, several other benefits measures closely related
to the goal of corrective action (e.g., protection of human health and the environment) are not
monetized and therefore not included in the benefit-cost analysis.
*** DRAFT - March 26,1993 **
-------
ES-18
EXHIBIT ES-3
CORRECTIVE ACTION BENEFIT-COST ANALYSIS
Effect of
Corrective
Action
Costs
(billion 1992
$)
S18.7
Benefits (billion 1992 $)
Non-use
values of
ground
water
S2.3
Averted water
use costs
$0.01
Value of the
corrective
action site
(high-end)
$0.28
Sum of
Benefits
$2.6
Net
Benefits
(billion
1992$)
416.1
The results of the cost-effectiveness analysis are displayed in Exhibit ES-3. The first row
presents the total discounted costs of corrective action. The non-monetized benefits of corrective
action, both the total benefits in each category and the benefits per billion dollars, are shown in
the six columns below. The Agency expects corrective action to eliminate individual risks of
concern at about 900 facilities, between 400 and 13,100 cancer cases in the exposed population,
and between 100 and 12 million cases where noncancer toxicity thresholds are exceeded. In
terms of reduced extent of contamination, the Agency expects a minimum of 1,400,000 acres of
contaminated ground water and 18,000,000 cubic yards of contaminated on-site soil to be
remediated. The last row of Exhibit ES-3 presents cost-effectiveness measures n the benefits of
corrective action per unit cost. For every billion dollars spent on the program, the EPA projects
the following benefits: 48 facilities with reduced individual risk; 21 to 711 averted population
cancer cases; 5 to 640,000 averted population exceedences of noncancer thresholds; 75,000 acres
of contaminated ground water remediated; and 950,000 cubic yards of contaminated on-site soil
remediated.
*** DRAFT - March 26,1993 ***
-------
EXHIBIT ES-4
CORRECTIVE ACTION COST-EFFECTIVENESS ANALYSIS
Total Discounted Costs of Corrective Action:
$18. 7 billion
Non-monetized Benefits
Human Health Risk
Facilities
with
Reduced
Individual
Risk
900 facilities
48 facilities
Averted
Population
Cancer
Cases
400 - 13,300
cases
Averted
Population
Exceedences of
Noncancer
Thresholds
100 - 12 million
cases
Ecological
Risks
Facilities with
Averted
Exceedences
of Ecological
Benchmark
Levels
NA"
Extent of Contamination*
Areal
Extent of
Ground-
Water
Cleanup
1,400,000
acres
Benefits Per Billion Dollars
21 -711
cases
5-640,000
cases
NA'
75,000
acres
Extent of
On-site Soil
Remediation
18,000,000
cubic yards
950,000 cubic
yards
Only baseline results are available to date for this measure.
b Results are for only 720 sites nationally and hence may understate total values.
** DRAFT - March 26,1993
-------
ES.5 Limitations
EPA devoted significant time and resources to make this RIA a "state of the art" analysis;
however the questions asked in any RIA methodology development are inherently challenging.
An overview of the essential factors that challenged EPA in the design of the methodology are as
follows:
o The RCRA corrective program is relatively new and data documenting waste
concentrations and the nature and extent of contamination are limited and uncertain.
o The model used in the analysis was chosen in part because the data requirements for the
model were not excessive; however, in order for this analysis to remain nationally
representative, EPA has made a number of simplifying assumptions when we address sites
with complex situations that are beyond the designed capabilities of the model.
o The corrective action RIA is a predictive analysis, intended to provide insights regarding
the likely benefits of corrective action many years into the future.
o Specifically, the RIA methodology focuses on the incremental differences in risk between
sites that have not been cleaned up (baseline scenario) and the same sites after they have
been cleaned up (post-regulatory scenario). The risk assessment of baseline conditions
requires the evaluation of a hypothetical scenario where contamination is allowed to
migrate through the environment and is not cleaned up.
o In order to understand the increment between the baseline and a post-regulatory
scenario, we had to determine the effectiveness of various remedial technologies including
waste treatment, containment, and pump and treat. The long term effectiveness of these
technologies continues as an area of intense discussion yet the RIA methodology was
required to confront these questions and estimate how effective various technologies are
in achieving long term cleanup goals.
o The corrective action RIA methodology balances the need to assess impacts on a national
scale with the consideration that the RCRA remedy selection process is highly site
specific (e.g., requiring the development of detailed specs for pump and treat systems and
containment systems for specific releases of contamination).
Each chapter of the RIA includes a section which elaborates on the technical limitations
specific to the subject area of the chapter in which it resides.
* DRAFT - March 26,1993 *
-------
1. INTRODUCTION
The draft of the corrective action regulatory impact analysis (RIA) being provided for
public comment today presents the methodology used to estimate the costs and benefits of
corrective action at hazardous waste treatment, storage, and disposal facilities. It examines the
baseline ("no action" option) and the corrective action Subpart S proposed rule in order to
illustrate this methodology. This analysis does not represent any conclusion by the Agency about
the appropriate future direction of the corrective action program, but merely acts as a point of
departure for a public dialogue on potential regulatory options for the corrective action program.
The next draft of the RIA will reflect comments received on the methodology and will examine
the costs, economic impacts, and benefits of corrective action regulatory options.
This introduction discusses the nature of the problem being addressed and the legislative
and regulatory history of the corrective action program in Section 1.1, describes the development
of the RIA in Section 1.2, and provides a brief overview of the RIA organization in Section 1.3.
1.1 The Need For Regulation and the Regulatory and Statutory History
1.1.1 Description of the Problem
The Resource Conservation and Recovery Act of 1976 (RCRA)1 was enacted to address
the problem of how to safely manage the huge volumes of municipal and industrial solid waste
generated in the United States. Subtitle C of RCRA originally focused on the safe management
of as-generated hazardous wastes (i.e., wastes that industrial facilities were producing as part of
their ongoing manufacturing or waste treatment processes). Prior to the development of
stringent RCRA hazardous waste management requirements, industrial facilities managed these
wastes in ways that would not be permissible today. These early management practices often
resulted in releases to environmental media resulting in contamination of soils, ground water,
surface water, and air. Consequently, many industrial facilities may have "old wastes" (i.e., wastes
previously mismanaged) and contaminated media that present current or potential threats to
human health or the environment. To address these potential threats, Congress subsequently
expanded RCRA in 1984 to require cleanup (i.e., corrective action) at facilities subject to RCRA
authority. (In a similar manner, the Superfund program, under the Comprehensive
Environmental Response, Compensation, and Liability Act of 1980 [CERCLA], was developed to
clean up hazardous waste problems at abandoned waste sites.)
1.1.2 Pre-HSWA Corrective Action Requirements: Subpart F
Initial RCRA regulation of hazardous waste cleanups was limited in scope, as it
addressed only regulated units,2 on-site releases,3 and releases to ground water. Prior to
passage of the RCRA Hazardous and Solid Waste Amendments of 1984 (HSWA), statutory
1 42 U.S.C. UU6901 - 6992k
2 Regulated units are defined in 40 CFR 264.90(a) as waste piles, surface impoundments,
land treatment areas, and landfills that received hazardous wastes after July 26, 1982.
3 Throughout this RIA, the term "release" refers to the release of hazardous wastes or
hazardous constituents to the environment.
DRAFT-March 24,1993
-------
1-2
authorities for compelling corrective action at facilities regulated under Subtitle C of RCRA
were limited to the following: (1) Section 7003 of RCRA, which provides EPA enforcement
authority to take action where past or present handling, storage, treatment, transportation, or
disposal of any solid or hazardous waste may present an imminent and substantial endangerment
to human health or the environment; (2) Section 3013 of RCRA, which provides authority for
requiring investigations where the presence of hazardous waste or releases of hazardous waste
may present a substantial hazard to human health or the environment; and (3) Section 3004(a),
which requires the EPA Administrator to promulgate regulations that established performance
standards applicable to owners and operators of facilities for the treatment, storage, or
disposal of hazardous waste, as may be necessary, to protect human health and the environment.
Pursuant to Section 3004(a) and other RCRA authorities, EPA promulgated 40 CFR Part 264,
Subpart F, which is discussed below.
Subpart F, established in 1982, created a regulatory program for monitoring and
remediating on-site releases to ground water from regulated hazardous waste management units.
Under current Subpart F regulations, the corrective action requirement (Section 264.100) is the
third step of a three-phase program for detecting, characterizing, and responding to releases from
regulated units to the uppermost ground-water aquifer. The first two phases require ground-
water detection monitoring and additional investigations when a release to ground water is
confirmed.
1.1 J HSWA Corrective Action Requirements
HSWA, passed in 1984, provided the framework for the current RCRA corrective action
program. It created a statutory requirement that a RCRA permit must address all releases
(including off-site releases) to all media from all solid waste management units (SWMUs)
located at the facility. RCRA Sections 3004(u) and 3004(v), added by HSWA, specifically address
such releases at permitted RCRA facilities. Section 3004(u) requires that after November 8,
1984, any permit issued under Section 3005(c) of RCRA to a treatment, storage, or disposal
facility address corrective action for releases of hazardous waste or hazardous constituents from
any SWMU at the facility. Section 3004(v) authorizes EPA to require corrective action beyond
the facility boundary where necessary to protect human health and the environment.
Section 3008(h) of HSWA, the corrective action enforcement authority, provides EPA
with the authority to issue administrative orders or to bring court action requiring corrective
action or other measures, as appropriate, when there is or has been a release of hazardous
constituents from an interim status RCRA facility. The key terms in this provision are
interpreted in a December 15, 1985 memorandum entitled "Interpretation of Section 3008(h) of
the Solid Waste Disposal Act".4 Appropriate information upon which to conclude that there
may have been a release for purposes of Section 3008(h) include, but are not limited to, the
following: data from laboratory analyses (from soil, air, surface water, or ground water samples),
observations recorded during inspections, photographs, and information obtained from facility
records. To exercise the interim status corrective action authority, EPA must first have the
information that there is or has been a release of hazardous waste to the environment at or from
4 Porter, J. Winston and Price, Courtney M. Interpretation of Section 3008fM of the Solid
Waste Disposal Act. Washington, D.C.: U.S. Environmental Protection Agency, December 15,
* * DRAFT-March 23,1993
-------
1-3
the interim status facility. Second, the corrective action or other response measure, in the
judgment of the Agency, must be necessary to protect human health or the environment.
Section 3008(h) authority is not limited to addressing releases from SWMUs. Releases
that are addressed under Section 3008(h) are somewhat different in scope than that of Section
3004(u). Releases under Section 3008(h) encompass the same definition of release under
CERCLA. Therefore, a release is any spilling, leaking, pumping, pouring, emitting, emptying,
discharging, escaping, leaching, dumping, or disposing of hazardous waste into the environment.
In situations where a Section 3008(h) action has been initiated at a facility to address
releases that are not from SWMUs, and where a permit is subsequently issued to the facility,
EPA intends that those actions will be continued under the permit, under the authority of
RCRA Section 3005(c)(3). In addition, any order issued under 3008(h) may include a
suspension or revocation of authorization to operate under Section 3005(e) of this subtitle.
1.1.4 HSWA Codification Rules
EPA promulgated two regulations that codified the statutory language of HSWA. On
July 15,1985, EPA codified the statutory language of the new Section 3004(u) corrective action
authority of HSWA (see 50 FR 28702 and 40 CFR 264.90(a)(2) and 264.101). The July 1985
Codification Rule amended 40 CFR part 264, Subpart F by adding new Section 264.101, which
essentially reiterated the statutory language of Section 3004(u). This Codification Rule defined
the Agency's jurisdiction under the new authorities by interpreting key terms in the statutory
language (e.g., facility, solid waste management unit, and release). This rule also provided the
Agency's interpretation of the authority conferred on it through Section 3008(h) to issue
administrative orders to require corrective action at interim status facilities (i.e., facilities that
have not yet received a RCRA permit).
On December 1,1987, the Agency issued a second Codification Rule, for RCRA Section
3004(v), that modified 40 CFR Part 264 (by adding Parts 264.100(e) and 264.101(c)) and Part
270 of the existing hazardous waste management regulations to implement the new statutory
provisions of HSWA (see 52 FR 45788). This second Codification Rule addressed issues relating
to permit application requirements for SWMUs; corrective action beyond the facility boundary;
and corrective action for injection wells with permits-by-rule.
1.1.5 Subpart S Proposed Rule
The corrective action rule proposed on July 27,1990 [55 FR 30798] creates Subpart S of
Part 264 and is intended to replace the HSWA Codification Rules with a detailed regulatory
program for implementing corrective action under RCRA that includes investigative
requirements, cleanup standards, and implementation procedures.5 The Agency is reviewing
comments received on the proposed Subpart S rule and is developing the Subpart S final rule. A
key element in developing the Subpart S final rule is the Regulatory Impact Analysis (RIA)
examining the costs and benefits of corrective action options. The RIA is discussed in more
detail in Section 1.2 of this chapter.
5 The proposed rule is described in more detail in Chapter 2.
* * * DRAFT-March 23, 1993 * *
-------
1-4
1.1.6 Corrective Action Management Unit/Temporary Unit (CAMU/TU) Rule6
On February 16,1993, EPA promulgated the Corrective Management Unit and
Temporary Unit final rule (CAMU/TU Rule) [58 FR 8658]. This rule established two new units
that will be used for remedial purposes under RCRA corrective action authorities. Both
CAMUs and TUs function solely to manage wastes that are generated at a RCRA facility for the
purpose of implementing remedial actions required at that facility (i.e., remediation wastes, as
defined in the rule). A CAMU is an area within a facility that is designated by the Regional
Administrator under Part 264 Subpart S for the purpose of implementing corrective action
requirements under 264.101 and RCRA Section 3008(h). Remediation wastes may be placed
into a CAMU without triggering the applicability of land disposal restriction (LDRs) or
minimum technology requirements (MTRs). TUs allow owner/operators to treat or store
remediation waste, for a limited period of time, in the unit without complying with RCRA 40
CFR part 264 regulatory standards (i.e., LDRs and MTRs).
The definition of a CAMU in the final rule modified the proposed definition in several
ways:
The new definition does not specify CAMUs as being contiguous areas of
contamination;
Non-land-based units, such as tanks, may be physically located within the
boundaries of a CAMU, however the tank will not actually be a part of the
CAMU;
The new definition specifies that CAMUs are to be used for the purposes of
managing remediation wastes only; and
The definition specifies that CAMUs may be used for corrective actions under
section 3008(h) orders, as well as at permitted facilities under Section 3004(u).
A separate RIA for that rule examined the cost savings and potential effects on human
health and the environment.7 Due to the timing of that RIA, the part of the Subpart S RIA
provided for public comment today reflects the CAMU definition that was included as part of
the Subpart S proposed rule, rather than the definition in the CAMU/TU final rule.
6 Much of the information in this section was taken from Corrective .Action for Solid Waste
Management Units at Hazardous Waste Management Facilities, Proposed Rule (55 FR 30798);
Corrective Action Management Units and Temporary Units, Corrective Action Provisions Under
Subtitle C, Final Rule (58 FR 8658); and EPA's "Environmental Fact Sheet: EPA Issues Final
Rules for Corrective Action Management Units and Temporary Units," January 1993, EPA/530-
F-93-001.
7 The CAMU/TU RIA is available in EPA's RCRA docket.
* * * DRAFT-March 23, 1993 *
-------
1-5
1.2 Description of Corrective Action Regulatory Impact Analysis
1.2.1 The Need for the RIA
There are significant public concerns regarding cleanups at RCRA facilities due to:
The large universe of RCRA facilities with potential releases to the environment;
The potential threats to human health and the environment from releases at these
facilities;
The high costs of cleanups at many sites;
The potential risks posed by remediation techniques that release hazardous
constituents to other media or that allow contaminated media to remain in place;
and
The limited effectiveness of remediation techniques in achieving cleanup goals in
some circumstances.
Recognizing these public concerns regarding the RCRA cleanup program, EPA is in the
process of preparing a major RIA to estimate the costs and benefits of various Subpart S
regulatory options and to obtain public comment on them. The RIA meets the requirements of
Executive Order 12291 (issued February 17, 1981), under which a Regulatory Impact Analysis is
required for every major Federal regulation. The Order defines a major rule as one that is likely
to result in: (1) an annual effect on the economy of $100 million, or more; (2) a major increase in
costs or prices for consumers, individual industries, federal, state, or local agencies, or geographic
regions; or (3) significant adverse effects on competition, employment, investment, productivity,
innovation, or on the ability of United States-based enterprises to compete with foreign-based
enterprises in domestic or export markets. The results of this RIA demonstrate that the final
corrective action rule is likely to be a "major" rule.
1.2.2 RIA Methodology
This RIA used "state-of-the-art" methods to accurately assess the costs and benefits of
corrective action cleanups. This multi-disciplinary approach, which used environmental scientists,
economists, and engineers from EPA, industry, and academia included:
Extensive involvement of EPA and state implementers, and national remediation
experts in simulating the actual remedy selection process.
Significant effort by EPA and private sector modeling experts in developing a PC-
based multi-media fate and transport model to simulate contaminant releases and
estimate potential human health risk and ecological damage at facilities.
Extensive involvement by EPA modeling and exposure assessment experts in
developing "central tendency," "high end," and other modeling assumptions to
better characterize uncertainty and to implement recent Agency guidance on risk
characterization.
* *
DRAFT-March 23, 1993
-------
1-6
Development of new measures of potential benefits of corrective action and
refinement of existing measures, including:
Measurement of individuals' willingness to pay for uncontaminated ground
water through state-of-the-art survey techniques;
Estimation of potential benefits from the avoided costs of replacing
ground water supplies; and
Estimation of impacts on property values in the vicinity of facilities due to
releases.
Assessment of the effectiveness of cleanup technologies commonly used to control
sources of contamination and remediate ground water, an area of research in
which there are very limited data currently available.
13 RIA Organization
The part of the RIA provided for public comment today presents data and methodologies
for estimating costs and benefits, using the baseline (no action option) and the Subpart S
proposed rule for illustration. The next part of the RIA will analyze costs and benefits of various
regulatory options for Subpart S and will also include an analysis of economic impacts.
This RIA is organized into 13 chapters, including this introduction. Chapters 2, 3, and 4
present regulatory options, approach, and process for selecting remedies in the analysis. Chapter
5 summarizes the approach to the cost analysis and presents the results of this analysis.
Chapter 6 provides a conceptual basis for the RIA benefits analyses, which are presented
in chapters 7 through 12. Costs and benefits are compared in chapter 13. The bibliography
concludes the RIA.
A number of appendices (A through J) accompany the RIA and provide background
data, assumptions, and descriptions of the approach for key aspects of the analysis. Appendix A
provides a detailed description of the approach for selecting the facility sample. The procedures
used to predict releases, determine remedial measures, and predict the effectiveness of remedies
are discussed in Appendices B and C. Appendix D provides background information on the RIA
cost analysis. Appendix E provides technical information on the approach and data used in the
human health benefits analysis. The methodologies and case studies used to evaluate the
ecological risk at corrective action facilities are presented in Appendix F. Appendix G
summarizes the parameters for the modeling and risk assessment and Appendix H presents the
forms used to collect facility and SWMU data. Appendix I reviews the characteristics of the
population of facilities subject to corrective action requirements. Finally, Appendix J provides
additional details on the approach for estimating economic impacts (to be completed for the next
RIA draft).
* DRAFT - March 24, 1993
-------
2. REGULATORY OPTIONS
This chapter describes the corrective action regulatory options that EPA examined in the
pan of the RIA provided for public comment today, and reviews some of the potential regulatory
options that may be examined in the next part of the RIA. Section 2.1 describes the baseline for
the analysis; Section 2.2 describes the Subpart S proposed rule; and Section 2.3 presents future
regulatory options that EPA might address after public review of this part of the RIA.
2.1 Description of the Baseline
The baseline of this RIA represents the requirements under RCRA and other Federal
laws that were in effect prior to enactment of the Hazardous and Solid Waste Amendments of
1984 (HSWA). The baseline, or "no action" option, is the scenario against which the costs and
benefits of the other options are compared. By comparing against a common baseline, the
incremental costs and benefits of the rule and regulatory options can be assessed.
Under the baseline, EPA assumed that land disposal units (i.e.,surface impoundments,
waste piles, land treatment units, and landfills) that received hazardous waste after July 26, 1982
would be remediated under the existing authority of 40 CFR Part 264 Subpart F. In addition to
the existing RCRA cleanup requirements under Subpart F, EPA assumed for this analysis that
certain other Federal requirements would be included in the baseline. As a result, the costs and
benefits of complying with these other requirements at facilities subject to the RCRA
requirements are not attributable to the RCRA corrective action program under HSWA. The
other Federal requirements included in the baseline address several situations:
Releases from units or facilities subject to requirements under the Comprehensive
Environmental Response, Compensation, and Liability Act (CERCLA or
Superfund). Although some elements of the corrective action program may
become applicable or relevant and appropriate requirements (ARARs) for
Superfund, the Agency did not examine the potential effect of corrective action
ARARs on the Superfund program.
Releases from units that are not solid waste management units (SWMUs), such as
product storage tanks. These are addressed under CERCLA or Subtitle I of
RCRA. On the other hand, if the source was unknown (e.g., monitoring
information could not pinpoint the SWMUs causing the release), the release was
included in the analysis.
Releases of constituents' addressed under other Federal requirements, particularly
radioactive wastes (addressed under the Atomic Energy Act).
For the purposes of the RIA, EPA did not estimate the costs and benefits of the baseline
corrective action program (i.e., predict the costs and benefits of the RCRA Subpart F program
and CERCLA) but rather focused on the incremental requirements of the corrective action
program under HSWA. Examining the costs and benefits of cleanups that would occur under
Subpart F or other remediation programs was considered to be beyond the scope of this analysis.
** DRAFT- March 24,1993 ***
-------
2-2
2.2 Description of the Subpart S Proposed Rule
2.2.1 Summary of Proposed Rule
The corrective action rule proposed on July 27,1990 under Subpart S of Part 264
outlined a comprehensive regulatory framework for implementing EPA's corrective action
program under RCRA. Together with the National Contingency Plan (promulgated March 8,
1990, 55 FR 8666), the proposed rule defined EPA's overall approach to cleanup of
environmental contamination resulting from the mismanagement of hazardous and solid waste.
As stated in Chapter 1, the proposed rule is intended to replace the July 19851 and December
19872 Codification Rules with a detailed regulatory program for implementing corrective action.
The proposed rule defined both the procedural and substantive requirements associated with
RCRA sections 3004(u) and 3004(v).
The proposed Subpart S rule was intended to promote national consistency in
implementing the corrective action component of the RCRA program, and establish standards to
which states seeking authorization for section 3004(u) corrective action must demonstrate
equivalence. The proposed rulemaking provided a procedural vehicle for the regulated
community and other interested parties to comment on the Agency's regulatory intentions for
this program and outlined the basic framework for the corrective action program for EPA and
authorized states. More specifically, it proposed to codify the procedures for identifying
problems and selecting remedies at RCRA facilities; the standards for cleanup, including the
establishment of cleanup levels; and the standards for managing cleanups and the wastes
generated by cleanups.
2.2.2 Assumptions About the Proposed Rule Used in RIA
For the purposes of the draft of the RIA provided for public comment today, EPA
evaluated the elements of the Subpart S proposed corrective action rule using the assumptions
discussed below. These assumptions were necessary to simplify the analysis and do not represent
EPA policy in implementing the corrective action program. EPA's detailed methodologies for
analyzing the costs and benefits of the proposed rule are provided in subsequent chapters.
EPA assumed that the substantive requirements of the Subpart S proposed rule,
under the authority of RCRA Sections 3004(u) and (v), would be followed in
conducting investigations and cleanups at both permitted and interim status
facilities. This assumption is based on the fact that Section 3008(h) corrective
action orders at all interim status facilities will likely require corrective actions
that are essentially identical to those imposed at permitted facilities under the
final rule.
For the RIA, EPA updated some of the action levels presented in Appendix A of
the preamble to account for more recent information on constituents' toxicity.
Action levels in ground water were based on maximum contaminant levels
(MCLs) promulgated under the Safe Drinking Water Act. For constituents
1 50 FR 28702.
2 52 FR 45788.
* * DRAFT-March 23,1993 * * *
-------
2-3
without MCLs, action levels were based on reference doses (for noncarcinogens)
and slope factors (for carcinogens). Action levels for the other environmental
media were generally based on reference doses and slope factors.
EPA assumed that corrective action management units (CAMUs) would be used,
as provided under the proposed rule. The proposed CAMU provisions would
allow movement or consolidation of hazardous waste within areas designed as
units that will not trigger land disposal restrictions (LDRs) or minimum
technology requirements (MTRs). The proposed CAMU was limited to a
contiguous area of contamination and included only land-based units (e.g., waste
piles, surface impoundments, land treatment areas, and landfills).3
2.3 Other Options To Be Examined
This draft of the corrective action RIA examines only the baseline (no action option) and
the Subpart S proposed rule. After public review and comment on the approach used in this
draft of the RIA, EPA will analyze several regulatory options and provide the next part of the
RIA for public review prior to finalizing the Subpart S rule. The types of options that EPA
might analyze in the next draft of the RIA include:
The point of compliance at which media cleanup standards must be achieved (e.g.,
achieving ground-water cleanup standards throughout the plume boundary vs. the
facility boundary).
Cleanup standards that remedies must achieve (e.g., to background levels vs.
health-based levels). These media cleanup standards represent constituent
concentrations in ground water, surface water, soils, and air.
Conditional remedies or other remedies to stabilize conditions at a facility in the
short-run. This provision would allow, at EPA's or the authorized State's
discretion, an owner/operator to implement remedies over time by addressing the
worst problems first, as long as certain conditions are met.
The timing of cleanups (e.g., immediate cleanup vs. waiting until risks of exposure
are of concern).
Cleanup in situations where achievement of media cleanup standards is technically
impracticable.
1 The definition of CAMU contained in the proposed rule differs somewhat from the
definition promulgated later as part of the final Corrective Action Management Unit and
Temporary Unit rule (58 Federal Register 8658, February 16,1993). This is discussed in more
detail in section 1.1.6.
* * *
DRAFT-March 23, 1993 * * *
-------
3. SAMPLE SELECTION, FACILITY CHARACTERIZATION,
AND MODELING OF RELEASES
This chapter describes the selection and characterization of the sample of RCRA
corrective action facilities. Section 3.1 describes the approach used to select and characterize the
sample of facilities and to characterize the extent of contamination at the facilities. Section 3.2
describes facility characteristics and the estimated extent of contamination at the facilities in the
absence of corrective action. Finally, Section 3.3 presents the limitations to the analysis.
As shown in Exhibit 3-1, this chapter describes the first four major steps followed by EPA
in developing the part of the corrective action RIA being provided for public comment today.
Subsequent chapters of this document describe the remaining steps. The analysis of other
regulatory options will be provided for public comment at a later date.
3.1 Approach
3.1.1 Selecting the Facility Sample
The first step in conducting the RIA consisted of identifying a sample of facilities from
the universe of RCRA facilities that would be subject to the Subpart S proposed rule. The
sample of facilities analyzed in this RIA was drawn from two sample universes or frames: a
federal facility frame and a non-federal facility frame. Because of the different characteristics of
these two frames, the federal and non-Federal samples were then selected from their respective
frames using different methods. This section first discusses the approach taken in selecting the
federal facility sample and then discusses the approach followed in selecting the non-federal
facility sample.1
The 359 federal facilities in the federal facility frame were drawn from the 1990
Inventory of Federal Agency Hazardous Waste Facilities and the Hazardous Waste Data
Management System (HWDMS, which has been supersede by the RCRA Information System
(RCRIS)). These two sources identified 395 federal facilities potentially subject to corrective
action. To verify the accuracy of this preliminary federal facility frame, EPA presented the list of
395 facilities to an interagency work group composed of federal department officials and
requested that the work group identify any facilities that should be eliminated from the universe
due to the fact that they are not subject to RCRA corrective action. As a result of the
verification process, 36 of the 395 federal facilities were eliminated from the list, resulting in a
corrected sampling frame of 359 federal facilities potentially subject to corrective action.
1 Refer to Appendix A for a more detailed discussion of the derivation of the federal and
non-federal sample frames and final samples.
* *
* DRAFT - March 20, 1993 * * *
-------
EXHIBIT 3-1
MAJOR RIA STEPS
SELECT SAMPLE OF FACILITIES
CHAPTER 3
CHARACTERIZE AFFECTED FACILITIES
CHAPTER 3
ESTIMATE CURRENT EXTENT
OF CONTAMINATION WITHOUT
CORRECTIVE ACTION
CHAPTER 3
REPLICATE CORRECTIVE
ACTION PROCESS AND
SELECT REMEDIES
CHAPTER 4
PREDICT FACILITY-LEVEL
COSTS AND ECONOMIC
IMPACTS*
CHAPTERS 5. 15
ESTIMATE LONG-TERM EXTENT
OF CONTAMINATION
WITHOUT CORRECTIVE ACTION
CHAPTER 3
'ASSESS EFFECTIVENESS OF
CORRECTIVE ACTION REMEDIES
IN REDUCING CONTAMINATION
CHAPTER 4
PREDICT FACILITY-LEVEL
BENEFITS
CHAPTERS 7-12
ANALYZE
OTHER REGULATORY
OPTIONS*
PREDICT NATIONAL COSTS,
ECONOMIC IMPACTS', AND
BENEFITS
ANALYZE
OTHER REGULATORY
OPTIONS*
132004-1
ECONOMIC IMPACTS AND OTHER REGULATORY OPTIONS TO BE ANALYZED LATER
-------
3-3
EPA divided the federal facility frame into three strata defined by the estimated
magnitude of potential costs of corrective action:
Very Large Department of Defense/Department of Energy (DOD/DOE) facilities;
Large DOD/DOE facilities; and
Other federal facilities.
Nine DOD and DOE facilities, each of which could incur corrective action costs in excess of $1
billion, were classified as "very large." Twenty-two DOD and DOE facilities, with potential
corrective action costs estimated between $100 million and $1 billion, were classified as "large."
The remaining 328 facilities (largely DOD facilities) were classified as "other."
After defining the sample frame for the federal facilities, EPA selected the sample for
analysis in the RIA based on a variety of methodological considerations. EPA did not include
any of the nine very large federal facilities for two reasons. First, because of their large size and
complexity, their inclusion in the analysis would have greatly reduced the number of non-Federal
and other federal facilities that the Agency would have been able to include in the RIA. Second,
the agencies responsible for these nine facilities are currently expending a high level of effort to
characterize and remediate the facilities, and therefore the agencies are better able to estimate
the potential corrective action costs and benefits at these facilities. EPA plans to work with
these agencies in order to better understand how the RCRA Subpart S program will affect the
activities underway at these very large facilities.
In selecting the sample from the remaining categories in the federal facility frame ("large
DOD/DOE facilities" and "other federal facilities"), EPA weighted the sample with a higher
percentage of large DOD/DOE facilities than are present in the population. These facilities
were over-sampled for two reasons:
This size category is likely to contain a greater proportion of
facilities that will require corrective action; and
Because the costs and benefits associated with large DOD/DOE
facilities are expected to dominate the national estimates, sampling
them at a higher rate will produce a more accurate estimate of
total national costs and benefits.
EPA derived the non-federal facility sample frame of 5,432 facilities from HWDMS and
the Corrective Action Reporting System (CARS) (both systems have now been superseded by the
RCRA Information System (RCRIS)).2 This sample frame represents the population of non-
2 Prepared for U.S. EPA by Research Triangle Institute. Selecting Sample Facilities for the
Corrective Action Regulatory Impact Analysis: Draft Report. Research Triangle Park, NC: June
1991.
* DRAFT « March 20, 1993 * *
-------
3-4
federal hazardous waste facilities subject to the Subpart S corrective action proposed rule. It
includes operating and closing facilities, permitted and interim status facilities, 90-day converters,
and new facilities that have applied for a permit.
Using a one-stage cluster sampling design, EPA sampled the universe of non-federal
facilities across three strata based on facility size and RCRA Facility Assessment (RFA) status:
Large facilities;
Not large facilities with RFAs completed; and
Not large facilities without RFAs.
Facilities in the "large" stratum were identified by EPA Regional officials using best
professional judgment as being the most significant facilities in their Region from a corrective
action perspective. Because the costs and benefits associated with the "large" facilities are
expected to dominate the non-federal sample, they were sampled at a higher rate than their
actual occurrence in the universe to allow for a more accurate prediction.
Facilities classified as "not large" were stratified by RFA status, i.e., whether an RFA had
been completed at the facility. RFA status was used to stratify the sample for two primary
reasons. First, it can indicate the availability of data for facility characterization, as better data
were typically available at facilities with completed RFAs. Second, it can indicate the likelihood
that corrective action will be required; since EPA Regions generally conduct RFAs at the most
environmentally significant facilities first, facilities with completed RFAs were generally more
likely to require corrective action. Thus, EPA believed that "not large with RFA" facilities would
be more likely to require corrective action than "not large without RFA" facilities and that it
would be more feasible to characterize "not large with RFA facilities" fully for analysis in the
RIA. Because the costs and benefits associated with facilities in the "not large with RFA"
stratum were expected to be more significant than those for facilities in the "not large without
RFA" stratum, the "not large with RFA" facilities were sampled at a higher rate than their actual
occurrence in the universe.
This over-sampling approach among strata and the corresponding weighting scheme
developed for the federal and non-federal samples was explicitly developed to correct for bias.
Facilities in all three of the non-federal strata were also profiled according to industry group
using each facility's two digit standard industrial classification (SIC) code. Although stratifying
by SIC code was impractical given the target sample size, the distribution of industries in each
stratum was controlled using a sequential selection procedure that sorted each stratum by SIC
code3. Sorting by SIC code within strata assured a proportional representation of industries in
the sample up to the limitations imposed by the stratum.
3 Additional information, including a breakdown of the non-Federal frame by SIC code, is
presented in Appendix A.
*
DRAFT « March 20,1993 * * *
-------
3-5
Of the 359 federal and 5,432 non-federal facilities identified in the two sampling frames,
EPA randomly selected 9 federal and 70 non-federal facilities, resulting in a total corrective
action sample of 79 facilities. Exhibit 3-2 presents both the number of facilities and solid waste
management units (SWMUs) in the final sample and the population. The number of SWMUs in
the population was estimated using the sample results and facility weights. It is important to
note that this estimate of the SWMU population does not include SWMUs located at 'Very large
DOD/DOE" facilities, as this stratum was not included in the sample. EPA anticipates that these
facilities will have more SWMUs on average than facilities in the other strata due to their size
and the magnitude of environmental problems at these facilities.
Facility Sample Quality Assurance
In developing the facility sample, emphasis was placed on maintaining a stratified random
sampling procedure in order to maximize the precision of the population estimator (given the
available resources). Using a stratified random sample enabled EPA to extrapolate the sample
findings to the population.4 Conducting a pilot phase of the study was an additional quality
assurance measure. The pilot phase provided the opportunity to reevaluate the appropriateness
of the sample size and to check for any potential problems with the sampling frame. Additional
quality assurance steps relating to characterizing the sample and assessing the extent of
contamination are presented in subsequent sections.
3.1.2 Characterizing Affected Facilities
After selecting the sample of facilities for analysis in the corrective action RIA, the next
step involved characterizing each of these facilities as a prelude to estimating the costs and
benefits of the Subpart S proposed rule, as shown in Exhibit 3-1. This facility characterization
provided the information needed for subsequent predictions of the numbers of facilities and
SWMUs that would require corrective action due to releases of hazardous constituents to the
environment. The facility characterization was designed to provide sufficient information on the
sample facilities to support the remedy selection expert panel process, the cost analysis, and each
of the components of the benefits analysis (human health benefits, ecological benefits, averted
water use costs, nonuse benefits of ground-water remediation, and changes in the value of sites).
To support such diverse analytic needs, EPA collected a wide variety of data for each of
the sample facilities: facility operations and history, environmental setting, SWMU characteristics,
extent of existing contamination, and potential receptors. The data were collected from several
sources, focusing primarily on EPA Regional files and state regulatory agency files. The data
elements are described in more detail below.
4 Appendix A provides additional information on the specific steps taken to ensure an
accurate sample.
* * * DRAFT - March 20, 1993 * * *
-------
3-6
EXHIBIT 3-2
NUMBER OF FACILITIES AND SWMUs5 IN THE SAMPLE AND POPULATION BY STRATA
' «ry
large
D05/DOE
Facilities in Population
Population weight per
Sample Facility
9
N/A*
Facilities
SWMUs
0
0
Large
DOD/DOE
Other
Federal
Federal
Total
Large
Not
Large
RFA
SAMPLING FRAME
22
7.3
328
54.7
359
N/A
87
3.3
1.711
63.4
Not Large
No RFA
3,634
213.8
Non-
federal
Total
5,432
N/A
TOTAL
5,791
N/A
SAMPLE
3
300
6
138
9
438
26
1.316
27
786
17
185
70
2,287
79
2,725
POPULATION ESTIMATES
SWMUs'
SWMUs per Facility
N/A
N/A
2.200
100
7.500
23
\ 9,700
27
4,400
51
49,800
29
39400
11
93,800 103,500
17 | 18
5 The number of SWMUs includes Subpart F units, Other SWMUs, and Areas of Concern.
6 Not Applicable
7 The number of SWMUs in the population has been estimated using the sample results and facility weights.
DRAFT - March 20,1993
-------
3-7
The data collection methodology used a hierarchial approach with three major sequential
steps illustrated in Exhibit 3-3. The primary data sources when available consisted of site-specific
information from RFAs, RCRA Facility Investigations (RFIs), CERCLA RODs or RI/FS
reports, RCRA or CERCLA inspection reports, and other site-specific studies. RFI reports
typicallyprovided the most extensive facility-specific information but were seldom available. RFA
reports provided the next best site-specific information and typically provided information on
SWMUs of environmental concern. The use of RFAs for data collection was limited for two
reasons: (1) the approach used to select sample facilities was stratified such that not all of the
non-federal facilities had RFAs, and (2) RFAs vary considerably and do not consistently provide
the same information for all facilities. The CERCLA documents were also generally very useful,
but were not generally available for most facilities.
When additional information was needed after collecting all the available site-specific
documents, EPA relied on more general data sources that could be applicable to the site, such as
regional level information, industry-level information on wastes and operations, and EPA
databases on waste stream composition. Finally, best professional judgment was used when it
was necessary to fill in remaining data gaps after exhausting the available primary and secondary
data sources. The types of data collected and a brief summary of the data sources is presented
below for each information category.
Facility Operations and History
General information on the facility operations and history, were collected to characterize
the broad status of the facility. These data included information on past waste management
practices, enforcement history, RCRA regulatory status (e.g, permitted, interim status, closing, or
operating), financial viability, and whether the facility has other environmental permits (such as
National Pollutant Discharge Elimination Permits (NPDES) under the Clean Water Act). These
data were generally found in RFA and RFI reports, RCRIS, EPA Regional file correspondence,
and other site-specific reports providing background on the facility. Financial viability was
typically determined through Dun and Bradstreet Business Information Reports.
Environmental Setting
A broad range of information was required on the environmental setting of each facility.
Hydrogeologjc information was required to support fate and transport modeling for predicting
the extent of contamination at each facility and included information on the unsaturated and
saturated zone materials, ground-water flow rate and direction, and dimensions of principal
water-bearing units. Climatic information included monthly precipitation, monthly temperatures,
days per month with rainfall, frequency of wind velocities in compass directions, and length of
the growing season. Information on surface hydrology was required to characterize potential
impacts to surface water and included stream locations and flow rates, presence of wetlands, and
location of lakes or coastal zone resources in relation to the facilities. Finally, topographic
information was required to characterize soil runoff pathways and included land slope, soil types,
land cover, and current land use.
* * * DRAFT - March 20,1993 * *
-------
EXHIBIT 3-3
DATA COLLECTION STRATEGY FOR THE CORRECTIVE ACTION RIA
Start
Review facility-
'specific documents (e.g., RFAsT*
RFls, sampling data). Additional
Information needed?
Yes
Review
'Regional level mapsT"
'SIC level data, general waste-^
[specific information (e.g., TSUR/Generatot
WET Model, TCRIA database),
Additional Information^
needed?
No
No
Yes
I
Data
Collection
Complete
Identify remaining data gaps. Use national level data,
Industry-specific studies, or best professional judgement
In conjunction with available-site specific information
to fill data gaps.
-------
3-9
The primary source for these data were site-specific reports such as RFAs, RFIs, and
RCRA ground-water inspection reports. Where these were not available, regional documents
such as USGS water resources reports and state or county ground-water assessments were used
to characterize the broad characteristics of the environmental setting at the regional level. For
some parameters, default values used across all sites were developed based on expert judgment.
SWMU and Waste Characteristics
Among the key data elements were the numbers and characteristics of SWMUs at each
facility. EPA based the identification of SWMUs of concern for this RIA on the definition of a
SWMU from the Subpart S proposed rule. In some cases, potential areas of concern that were
not clearly within the SWMU definition (e.g., product storage tanks) were also characterized to
provide a complete picture of facility contamination. RFA reports were the primary data source
for identifying SWMUs; when they were not available EPA relied on CERCLA inspection
reports, RCRA inspection reports, and general facility documentation.
In order to characterize facility contamination, it was necessary to identity the volumes
and characteristics of wastes in each SWMU. RFA reports, and where available, RFI reports,
provided the best information on SMWU dimensions and waste characteristics. RFAs were
generally sufficient for identifying the number, type, and location of SWMUs at each facility.
Although unit descriptions in RFAs indicated the major contaminants in the waste, RFAs
provided sparse information on constituent concentrations and the total waste volume. For
example, EPA used SWMU-specific information drawn from facility documents to derive
constituent concentrations for about one-third of the units simulated in the analysis. Information
on the remaining two-thirds of the units was drawn from several secondary data sources.
Several databases were used after all pertinent information had been extracted from the
facility-specific data sources. For example, the TSDR/Generator Survey database was a common
source of information on waste types and constituent concentrations. The 1990 Inventory of
Federal Agency Hazardous Waste Facilities (Federal Facility Inventory) database was used to
collect information on federal facilities. The WET Model8 database, TC RIA database9, and
background documents for analyzing Best Demonstrated Available Technologies (BDAT)
developed as part of the Land Disposal Restrictions (LDRs) rulemakings provided general
information on waste composition and physical form.
8 U.S. EPA. Prepared for the Office of Solid Waste, Economic Analysis Branch, by ICF Inc.
The RCRA Risk-Cost Analysis Model Phase in Report. Fairfax, VA: March 1,1984.
9 U.S. EPA. Developed for the Office of Solid Waste by ICF Inc. Toxicitv Characteristic
Regulatory Impact Analysis. Fairfax, VA: March 1990.
* * DRAFT -- March 20,1993 * *
-------
3-10
Extent of Existing Contamination
Another key data element for this RIA was information on the extent of existing
contamination. Because the types of remedies selected at each facility (and consequently the
costs of corrective action) depend on the extent of environmental contamination, it was
important to identify information on existing contamination. This included information on the
nature, extent, and rate of migration of contaminant plumes in ground water; surface water and
sediment contamination; and on-site and off-site soil contamination. It also included
documentation of prior remedial actions taken at the facility. This information was primarily
available through completed RFI reports or Superfund RODs, although these were not available
for most facilities. When the information was not available, it was necessary to predict the
current and future extent of contamination using a multimedia fate and transport model based
on the facility and SWMU information previously collected.10
Potential Receptors
The last category of information included the location and characteristics of potentially
exposed human and environmental receptors. Key data elements included the number and
locations of residential users of ground-water downgradient from the facility, and the number of
users of public water supplies drawing from aquifers potentially affected by the facility.
Information on locations of drinking water intakes in surface water bodies was also collected, as
was information on the current uses of lakes, rivers, wetlands, and marine resources that could
be affected by the facility. The number of individual residences'surrounding the facility was also
characterized in order to determine the potential for exposures to atmospheric contamination.
Information was also sought on sensitive environments that could be adversely affected by
contamination from the facility. These data came from facility documents, USGS quadrangle
maps, and Federal databases with information on public water supply systems.
Facility Characterization Quality Assurance
Standardized facility- and SWMU-specific data forms were developed to ensure quality
and consistency in the data collection process." The data acquisition process used data
collection forms and a prioritized list of data sources to systematically extract and track data
during the data collection and mapping phases. EPA established procedures and decision rules
to outline the procedures for data collection and the method of manipulating data for developing
input data sets for the MMSOILS model. Once data were collected, EPA used a double data-
entry system to electronically compare the two sets of entries and search for inconsistencies.
This system allowed for the efficient identification and correction of data-entry errors.
10
1 The general approach used to characterize existing contamination at sample facilities is
discussed below in section 3.13; the methodology for predicting the extent of contamination over
the long-term is presented in section 3.1.4.
"Sample facility and SWMU data forms are provided in Appendix H.
* * DRAFT - March 20, 1993 * * *
-------
3-11
3.1.3 Characterizing Existing Contamination
As illustrated in Exhibit 3-1, one of the'central steps in the corrective action RIA
methodology involved predicting specific corrective measures at each of the sample facilities that
are expected to require corrective action. In order to predict these remedies, it was necessary to
present extensive information on the nature and extent of existing contamination at the sample
facilities to allow panels of remediation experts to specify potential remedies.12 This section
summarizes the methodology followed in characterizing existing contamination for presentation
to the expert panels.
Under the Subpart S proposed rule, EPA Regions or states are expected to select
corrective action remedies at RCRA facilities based on the results of final RFI and corrective
measures study (CMS) reports submitted by the owner/operator. Accordingly, one of EPA's
objectives in the RIA was to present to the expert panels the kinds of information on
contamination that would be found in a final RFI report. Actual RFI reports typically include
ground-water plume contour maps, tables of sampling results for the principal RCRA
constituents of concern, estimates of the direction and rate of contaminant migration; surface soil
concentrations; surface water and sediment concentrations; air concentrations at receptor points
or the facility boundary; and estimates of contaminants remaining in waste units.
Where these kinds of contamination data were available in site-specific documents such
as RFIs or Superfund RODs, they were presented to the expert panels to represent current
contamination. In the majority of cases, however, most of these data were not available and it
was necessary to simulate the current extent of contamination in order to support the remedy
selection process.
EPA used the MMSOILS model, a multimedia contaminant fate and transport model, to
simulate contaminant release, fate, and transport at sample facilities lacking complete
information on current contamination.13 With the MMSOILS model, EPA was able to estimate
the release of contaminants from the many types of SWMUs found at RCRA facilities (e.g.,
landfills, waste piles, surface impoundments, land treatment units, injection wells, waste pits,
contaminated soil areas, tanks, and sludge lagoons). After simulating contaminant release from
individual SWMUs, the model simulates the subsequent transport of these constituents through
the ground-water pathway (i.e., through the unsaturated and saturated zones), soil erosion
pathway, foodchain/bioaccumulation pathway, surface water pathway, and atmospheric pathway.
Based on information collected during the facility characterization process, EPA
predicted contaminant releases with MMSOILS based on the age of the SWMU, its waste
contents, and the integrity of its containment structures. The modeling horizon allowed the
12 See Chapter 4 for a full discussion of the remedy selection expert panel process.
"An overview of the MMSOILS model and the model selection process conducted by EPA is
presented in section 3.1.5, with additional details provided in Appendix B.
* * *
DRAFT - March 20, 1993
-------
3-12
Agency to simulate releases from units that began operations as far back as 1920; where units
had been in existence prior to that date they were assumed to have begun operation in 1920.
Where units began operation after 1920, they were modeled as opening during their actual
starting year. Because releases in slow moving ground-water systems may take several decades to
migrate to an off-site receptor, this retrospective release horizon can account for current
environmental contamination resulting from old SWMU releases.
In estimating the current extent of contamination, EPA developed results for the assumed
starting year of corrective action, and for the following fifth and tenth years. Providing estimates
of contamination for three time periods enabled the expert panels to consider the rate of
contaminant migration over a ten year period when selecting remedies. It is assumed that RFIs
will typically provide similar information on the rate of contaminant migration.
The starting year of corrective action was determined using National Corrective Action
Prioritization System (NCAPS) rankings. NCAPS is a tool used for roughly ranking the risk
associated with RCRA facilities and for setting relative corrective action priorities. By presenting
a range of start years for corrective action based on facility priority, the RIA reflects Agency
policy whereby the most environmentally significant facilities will be addressed first. The ranking
system evaluates several factors in determining whether a facility is of high, medium, or low
environmental significance:
History of releases;
Hydrogeology of the area;
Route of continuing releases;
Waste types and quantities handled; and
The likelihood of human or environmental exposure.
For the corrective action RIA, the assumed starting years for corrective action are listed below
with the associated NCAPS rankings:
High priority: action assumed to begin between 1992 and 1997;
Medium priority: action assumed to begin in 1997; and
Low priority: action assumed to begin in 2002.
For each facility, the Agency used MMSOILS to generate plume maps depicting plume
boundaries for the contaminant resulting in the greatest areal extent of ground water
contamination as defined by the action level. By presenting the boundary of the largest
contaminant plume at each SWMU, it would be possible for the expert panels to determine the
largest areal extent of ground-water contamination likely to require remediation.
3.1.4 Predicting the Long-term Extent of Contamination
In order to estimate the benefits of the Subpart S proposed rule, the Agency
implemented a methodology for predicting the long-term extent of contamination in the absence
* * * DRAFT - March 20, 1993 *
-------
3-13
of corrective action. As shown in Exhibit 3-1, this long-term estimate of baseline contamination
served as a key input to assessing the effectiveness of corrective action remedies and estimating
each of the components of the benefits analysis: human health benefits, averted water use costs,
nonuse benefits of ground-water remediation, and changes in the value of sites.
The long-term baseline extent of contamination was simulated using the MMSOILS
multimedia release, fate, and transport model. The data sources used for the long-term
MMSOILS modeling were the same as those used in estimating the current extent of
contamination. The long-term modeling methodology was applied to all of the sample facilities
expected to require corrective action, including those facilities with extensive monitoring data.
While the long-term modeling methodology was based on the same general methodology
as was used in estimating current contamination at the sample facilities, it differed in some
important ways. One of the central differences between the long-term and current contamination
estimates involved the different modeling horizons. While both the current and long-term
modeling approaches simulated releases from SWMUs starting in 1920, the duration of the
release simulation differed. While the short term approach only simulated releases from
SWMUs until ten years after the start of corrective action, the long-term modeling approach
simulated releases until the year 2119. This resulted in a maximum potential period for
simulating contaminant releases from SWMUs of 200 years.
Because the baseline period for the corrective action RIA was assumed to begin in 1992,
ambient contaminant concentrations in the various environmental media were simulated from
1992 to 2119, resulting an a modeling horizon of 128 years. This 128-year modeling horizon
served as the temporal basis for estimating the overall benefits evaluated in subsequent chapters
of this RIA. Because the corrective action RIA assumes that corrective measures will be
instituted beginning in 1992 at the earliest, the post-remediation extent of contamination results
(presented in Chapter 4) reflect this same 128-year modeling horizon. Because ambient
concentrations prior to 1992 cannot be affected by the proposed rule, they are not tracked in this
analysis.
Long-term releases from all SWMUs of concern at the sample facilities were simulated
with MMSOILS. MMSOILS can account for the delayed release from SWMUs that had
containment devices such as liners, caps, or secondary containment, as well as releases from units
without any engineering controls. SWMUs of concern were identified as those with hazardous
wastes or constituents in the unit and lacking containment devices that could prevent the release
of these constituents over the long-term modeling horizon. For units with liners or covers, and
for tanks, EPA estimated release rates based on engineering judgment concerning the long-term
integrity of these units.
The Agency recognizes that significant uncertainty is often associated with environmental
fate and transport modeling, human health risk assessment, and other projections involved in this
analysis. To characterize uncertainty for the long-term modeling, EPA developed "central
tendency" projections of releases and benefits and "high-end" projections. The central tendency
* *
* DRAFT - March 20,1993 * * *
-------
3-14
and high-end estimates were developed in accordance with the Agency's guidance on risk
characterization.14
The central tendency case represented EPA's best estimate of the extent of
contamination and benefits. The high-end case used selected high-end SWMU and constituent-
specific input parameters as well as selected high-end modeling parameters to estimate the
potential increase in extent of contamination and benefits as a function of data uncertainty.15
For example, if no uncertainty existed regarding the SWMU dimensions or constituent
concentrations (i.e., exact SWMU dimensions, waste volumes, or waste concentrations were
available from facility-specific documents), EPA used the same values for both the central
tendency and high-end scenarios. However, in cases where SWMU-specific central tendency data
were uncertain, the Agency used available information and best professional judgment to
estimate different high-end values for the parameters. Other parameters were varied uniformly,
rather than on a SWMU-specific basis, for all sample facilities if significant uncertainty was
thought to exist in the approach. For example, because of the variability in ground water
hydrogeology, central tendency ground water concentrations were based on the typical screening
depth of wells while the high-end concentrations were taken from the top of the aquifer, where
concentrations are higher.
3.1.5 Modeling Methodology
The Agency developed a multimedia modeling methodology for predicting both current
and long-term contamination from sample facilities. EPA's Office of Research and Development
(ORD) provided extensive assistance in developing the most scientifically defensible approach
feasible within the constraints of the analysis. The following discussion of the modeling
methodology addresses the approach taken in developing the modeling methodology, presents
the methodology followed for simulating releases from SWMUs and their subsequent fate and
transport in the environment, addresses the aggregation of releases from individual SWMUs to
the facility level, and describes the modeling of karst/fractured geologic conditions and
multiphase flow.
Approach to Developing the Modeling Methodology
The first step in developing the modeling methodology for the corrective action RIA
involved a detailed model selection process. ORD was requested to review and provide
comment on the proposed modeling approach and the recommended use of the MMSOILS
""Guidance on Risk Characterization for Risk Managers and Risk Assessors" memorandum
dated February 26, 1992.
15 Appendix B provides more detailed information on SWMU-specific central tendency and
high-end assumptions. Appendix G presents the specific values used for the central tendency
and high-end parameters.
* * * DRAFT - March 20,1993 * * *
-------
3-15
model for the analysis in order to ensure its scientific acceptability.16 The primary capabilities
that were needed in the model used in the RIA included the capability to estimate multimedia
contaminant release and transport; the ability to simulate contaminant release from a wide
variety of waste management unit types; the ability to simulate the effectiveness of corrective
measures such as caps and liners; ease of use given the tight timeframe for the analysis; and
technical acceptability among the scientific community. MMSOILS was determined to be the
best model currently available that could meet the programmatic and technical objectives of the
Agency in conducting the RIA.
While recommending the use of MMSOILS in the RIA, ORD recommended several
modifications to the model to enhance its acceptability. ORD coordinated a working group of
EPA research laboratories and outside consultants to recommend and implement changes to the
model within a time period that would allow its use in the RIA. Based on this working group, a
number of modifications were made to MMSOILS that improved its capabilities, including
improved simulation of unsaturated zone transport and mixing in the saturated zone, and scale-
dependent contaminant dispersion in the saturated zone.
In addition to recommending changes to MMSOILS, ORD provided guidance on the
collection of chemical parameters for use in the fate and transport modeling. After reviewing
proposed fate and transport parameters for the potential set of 221 RCRA chemicals for which
data had been found, ORD provided data on several key physical/chemical parameters for about
100 of the most commonly observed RCRA hazardous constituents.17
ORD also highlighted several issues concerning the limitations of the modeling approach
and uncertainties. One key limitation concerned the inability of MMSOILS, which incorporates
a porous medium ground-water transport model, to simulate contaminant fate and transport in
karst aquifers.18 To address this limitation, ORD recommended modeling assumptions for
those sample facilities overlying karst aquifers.
ORD also conducted an uncertainty analysis to attempt to place bounds on the results
generated through the MMSOILS modeling approach. They developed a framework for
16 "Selection of Models for the Corrective Action Regulatory Impact Analysis," memorandum
from Loretta Marzetti, Director, Communications, Analysis, and Budget Division, Office of Solid
Waste, to Peter Preuss, Director, Office of Technology Transfer and Regulatory Support, Office
of Research and Development, January 9,1991.
""Review and Recommendations Related to Chemical Data Used in the Corrective Action
Regulatory Impact Analysis," from Gerard Laniak, EPA/ERL-Athens, to Barnes Johnson,
EPA/OSW/CABD, November 7,1991.
I8"Flow and Transport Modeling in Karst Aquifers: A Status Report," delivered by EPA,
Environmental Research Laboratory, Athens, GA, prepared for Office of Solid Waste.
DRAFT - March 20,1993 * *
-------
3-16
evaluating the many potential sources of uncertainty in the modeling analysis and then
implemented the approach using a Monte Carlo version of MMSOILS.19
Overview of Modeling and Post-Processing Methodology
Exhibit 3-4 depicts the main steps that were followed in conducting the long-term extent
of contamination analysis with MMSOILS. As it shows, the first step involved preparing model
inputs from the facility and chemical data collected for each sample facility. After preparing the
model inputs, MMSOILS was executed to simulate contaminant release and fate and transport,
resulting in estimates of ambient concentrations of those hazardous constituents released from
each SMWU.
MMSOILS first estimated the failure of engineering controls and the subsequent release
of constituents from the SWMU being modeled and then simulated the fate and transport of
those constituents in the ground-water, air, surface water, soils, and the foodchain. The way in
which MMSOILS predicts contaminant release from SWMUs, and the subsequent contaminant
fate and transport in the environment, is discussed below. Following the execution of
MMSOILS, the chemical- and SWMU-specific outputs were aggregated, using a post-processing
program developed for the corrective action RIA, resulting in facility-level concentrations at
exposure points. The approach to aggregating concentrations from individual SWMUs to the
facility level in each medium is also discussed below.
Predicting Contaminant Release from SWMUs
MMSOILS was used to simulate releases from SWMUs to each of the five environmental
pathways (i.e., ground water, soil, surface water, foodchain/bioaccumulation, and air). The model
uses different methods to estimate the volume, concentrations, and timing of releases from
different types of SWMUs. While MMSOILS was designed to simulate five basic types of waste
management units (landfills, waste piles, surface impoundments, tanks, and injection wells), it
was also used to simulate a wide variety of less well-defined waste management units. Most of
the SWMUs found at the sample facilities consisted of one of the five units found in MMSOILS,
and the SWMUs could be simulated directly by the model. In order to simulate other kinds of
SWMUs, however, it was necessary to identify an analogous model unit in MMSOILS and
represent the SWMU with that unit. For example, where the SWMU was an area of soil
contamination, it was typically simulated with MMSOILS as an unlined, uncovered landfill.
19 "Proposed Framework to Evaluate Uncertainty in Chemical Fate and Transport Predictions
for RCRA Corrective Action Regulatory Impact Analysis," U.S. EPA, Environmental Research
Laboratory, Athens, GA, September 1992.
* DRAFT - March 20, 1993 * *
-------
EXHIBIT 3-4
overview of MMSOILS Model and Post-Processing Programs
MMSOILS
Model
Post-processing
Programs
Inputs from facility characterization:
- Facility and SWMU data
- Chemical data
Estimation of releases from SWMUs:
- Landfills
- Waste Piles
- Surface Impoundments
- Tanks
- Underground Injection Wells
Modeling of fate and transport of SWMU
contaminants in environmental media:
- Ground Water
- Air
- Surface Water
- Soils
- Foodchain
Aggregation & interpolation of SWMU
media concentrations to obtain annual
facility level extent of contamination
Estimation of baseline human health risks,
ecological risk, and non-use value
Modeling of post-remediation extent of
contamination and risk
Estimation of benefits of corrective
action rule
-------
3-18
The modeling data needs for the five different SWMU types in MMSOILS differed
depending upon the unit type. The release mechanisms and data requirements for each of the
five units are discussed below.20
Landfills and Waste Piles
MMSOILS simulates releases from landfills and waste piles to ground water; to the
atmosphere via paniculate and volatile emissions; to surface water through overland run-off and
erosion and ground-water discharge to streams; and to off-site fields via soil erosion. It simulates
releases to ground water from landfills and waste piles by calculating the volume of infiltration
passing through the unit, and estimating the concentration of constituents leaching from the
waste into the leachate. The model calculates the volume of leachate leaving the unit using a
water balance approach that subtracts outputs (i.e., runoff and evapotranspiration) from inputs
(i.e., precipitation, irrigation, and surface run-on) and accounts for changes in water storage
within the unit, resulting in an estimate of the leachate volume available for release.
MMSOILS can simulate the effect of landfill covers, liners (including multi-layer
systems), and leachate collection systems on the amount of leachate released to ground water
when they are present at the unit, while waste piles are assumed to have no containment
features. Leachate concentration was calculated based on constituent concentrations in the unit
and their aqueous solubilities21. Mass disposed in the unit and leachate releases from the unit
were tracked each year using a mass balance approach.
Both volatile and paniculate emissions from these units to the atmosphere are calculated
based on the waste concentration, the degree of surface cover, and atmospheric data. Volatile
releases are calculated each year as part of the unit mass balance. Paniculate release estimates
use steady-state equations based on average surface concentrations, but releases stop when all
mass has been depleted from the unit. Soil runoff is calculated based on the waste concentration
and the surface water runoff calculated as pan of the water balance. Runoff calculations are
also steady-state, but runoff ends when all mass has been depleted.
Surface Impoundments
MMSOILS simulates leachate release to the ground water and volatilization to the
atmosphere from open surface impoundments. It simulates releases from surface impoundments
using the same basic water balance approach as landfills, although it tracks the waste liquid flow
into the impoundment and effluent discharge from the impoundment. Leachate concentrations
are calculated based on the concentrations of dissolved constituents added to the unit; volatile
emissions are calculated using a standard volatilization model for aqueous solutions of organic
20 For additional information on key modeling assumptions, limitations, and a complete
listing of input parameters refer to appendices B and G.
21 Leachate simulation is discussed thoroughly in Appendix B.
* * * DRAFT - March 20, 1993
-------
3-19
compounds. An annual time series of mass released from the unit is generated for both leachate
release and volatile emissions. No soil or paniculate releases are calculated while impoundments
are active. After the impoundment has been closed (i.e., no additional liquid wastes are added),
the model simulates closure of the unit with waste in place as a landfill, based on the assumption
that 15 percent of the constituent mass has been retained as sediment sludge in the closed unit.
Constituents are assumed to leach from the sludge as in the landfill case, and contaminated
sediments may be transported as dust entrained in the air or in runoff.
Tanks
Releases from tanks are assumed to be limited to the ground-water pathway. For each
tank, a profile of annual releases is determined based on the tank volume, its construction
materials, and its age. The Agency simulated releases based on the results of EPA's Hazardous
Waste Tank Failure Model (HWTFM).22 The HWTFM is a stochastic, Monte Carlo simulation
model that generates release profiles for each of several different tank technologies based on the
probability of events such as overflows, corrosion of tanks and pipes, natural catastrophes, and
accidental spills.23
Injection Wells
MMSOILS simulates failure and release from underground injection wells using
approaches developed for the Oil and Gas Study and Report to Congress.24 Failure is assumed
to take place in the aquifer and is attributed to either grout seal failure or well casing failure.
Using data about the injection rate, depth, and type of well, the model calculates direct releases .
to the aquifer.
Non-Standard Waste Management Units
Where the SWMU was not one of the units described above, EPA determined the most
appropriate way to simulate releases from the SWMU using one of the MMSOILS unit types.
For example, a sludge pit could be modeled as a landfill, or alternatively if it had extensive
standing liquids, it might be modeled as a surface impoundment. Containers of hazardous wastes
were often modeled as tanks, and the failure and release profile from the containers were
estimated using best professional judgment. Some SWMUs could not be modeled due to their
characteristics (e.g., industrial sewer lines, facility-wide runoff ditch systems, and buried gas
22 U.S. Environmental Protection Agency. "Hazardous Waste Tank Risk Analysis, Draft
Report," prepared by ICF Incorporated and Pope-Reid Associates Incorporated, June 1986.
23 See Appendix B for a discussion of tank release profiles.
24 U.S. Environmental Protection Agency/OSW. Technical Support Document. Onshore Oil
and Gas Exploration. Development, and Production: Human Health and Environmental Risk
Assessment. Washington, D.C.: U.S. Environmental Protection Agency, 1987.
* DRAFT - March 20,1993 * *
-------
3-20
cylinders). In addition, releases of complex constituents, such as buried chemical warfare agents
from World War I, were not simulated.
Simulation of Fate and Transport
The MMSOILS model simulates five distinct transport pathways: atmospheric, ground
water, surface water, soil erosion, and food chain/bioaccumulation (see Exhibit 3-5). Modeling of
each of the pathways is briefly described below25.
The Ground-water Pathway
The ground-water pathway is simulated using modules for recharge/runoff, the
unsaturated zone, the mixing zone, and the saturated zone.
The recharge/runoff module calculates the volume of water that percolates into the soil
and moves downward to the groundwater. This is determined using a water balance method to
subtract outputs, such as evapotranspiration and runoff, from inputs, such as precipitation and
irrigation. Values for runoff volume are based on standard methodologies developed by the U.S.
Soil Conservation Service.
The unsaturated zone algorithm is adapted from VADOFT, a subset of the EPA model
RUSTIC,26 and uses a one-dimensional finite difference solution to simulate solute transport
processes, including dispersion, advective flow, linear equilibrium adsorption, and first-order
decay. VADOFT uses the annual time series of releases from the SWMU and generates a
delayed time series that represents the leachate flux at the water table.
The mixing zone calculations estimate the depth of penetration beneath the area of the
unit to establish a three-dimensional source for the saturated zone module (defined by the area
of the waste management unit and the penetration depth). MMSOILS uses a method for
calculating the depth of penetration based on a peer reviewed model (EPACML27) developed
by EPA's Office of Solid Waste.
25
Input parameters for each pathway are listed in Appendix B.
U.S. EPA. Risk of Unsaturated/Saturated Transport and Transformation of Chemical
Concentration fRUSTIO Volume I: Theory and Code Verification. Washington, DC: 1989.
EPA/600/3-89-48a.
U.S. EPA. Background Document for EPA's Composite Model for Landfills TEPACMU.
Washington D.C: February 1990.
* * * DRAFT - March 20,1993 *
-------
EXHIBIT 3-5
EXPOSURE MEDIA, PATHWAYS, AND EXPOSURES ROUTES
Releases
SWMU
On-Slte or
Adjacent
Air
Surface Water
Surface Soil
Vadose Zone
At Point of
Exposure
Air
Biota
Surface Water
Surface Soil
Biota
Groundwater
Exposure
Route
Inhalation
Ingestion
Ingestion
Contact
Inhalation
Ingestion
Contact
Ingestion
Ingestion
Contact
Inhalation
Receptor
F2D028-3
-------
3-22
The module used for the saturated zone is a three-dimensional, quasi-analytic
advection/dispersion model incorporating scale-dependent dispersivity, retardation, and first-order
decay to generate non-steady-state estimates of aquifer concentrations. The time-variant source
is the volume calculated from the unsaturated zone and mixing zone modules. For the ground-
water pathway, plumes were modeled up to five miles when estimating the long-term extent of
contamination. Input parameters are mainly hydrogeologic properties (including unsaturated
depths, soil characteristics, and saturated flow rates) that are generally found in facility-specific
data sources. Transport of contaminants in the saturated zone is also simulated into nearby
streams.
The Atmospheric Pathway
The atmospheric pathway was simulated using a sector-averaged Gaussian dispersion
model based on the Industrial Source Complex Long-Term (ISCLT) model,28 developed by
EPA's Office of Air Quality Planning and Standards (OAQPS). The atmospheric pathway
component of the model considers the release of contaminants from the facility in the form of
vapors and fugitive dust emissions through the following processes:
Volatilization from soils and solid wastes;
Volatilization from a water body or from liquid wastes; and
PM10 and PM30 paniculate emissions due to wind erosion and mechanical
disturbances.
*
These releases vary with environmental conditions (e.g., wind velocities), the waste
management practices at the facility (e.g., waste spreading), and the characteristics of the waste
(e.g., liquid or solid). Once the contaminant is in the atmosphere, it is transported by wind and
dispersed due to turbulence in the flow. Average downwind concentrations were calculated on
an annual basis over a 22.5 degree sector based on central tendency and high-end assumptions
concerning the atmospheric stability, wind velocities, and deposition rates of particulates.
Contaminant concentrations in the atmosphere were estimated at five distances from the facility
boundary corresponding to the centerpoint of each of five concentric rings surrounding the
facility, extending out to a distance of 10 kilometers. The concentrations at these five points
were assumed to be constant throughout each respective concentric ring.
The Surface-Water Pathway
The surface-water pathway evaluates contaminants entering one of two types of receiving
water bodies: a stream or a small lake. For contaminants entering a small lake, the source term
is the contaminated bed sediments resulting from the erosion of contaminated particles (e.g.,
either waste material or soil) from an adjacent waste site. For contaminants entering a stream,
28 U.S. EPA. Industrial Source Complex (ISO Dispersion Model User's Guide - Second
Edition rRevisedV Washington, DC: December 1987. EPA/450/4-38-002a.
* * * DRAFT -- Mareh 20,1993 * * *
-------
3-23
the potential source terms include the erosion of particles with adsorbed contaminants, inflow of
runoff containing dissolved constituents, and the discharge of contaminated ground water into
the stream. Once contaminants have reached a water body, the concentration in the water
column is estimated for each year in the modeling period by assuming that the annual
contaminant loading is completely dissolved and mixed in the stream's flow.
The Off-Site Soil Contamination Pathway
The off-site soil contamination pathway evaluates contaminant movement to soils in off-
site residential fields (e.g., play grounds and back yards) and off-site agricultural fields through
two mechanisms: soil erosion and atmospheric deposition. The model simulates soil erosion
from a site with a simple erosion algorithm used in agricultural applications (Universal Soil Loss
Equation). It then determines off-site soil concentrations based on the delivery fraction of
eroded soil, mixing with off-site soils, and contributions from atmospheric deposition. Soil
concentrations in the fields are based on steady-state equations assuming time-averaged
contributions from surface erosion and atmospheric deposition.
The Food Chain Pathway
The food chain/bioaccumulation pathway evaluates the transport of contaminants from
the facility through other environmental transport pathways to off-site soils where contaminants
can enter the food chain. Examples of environmental transport pathways that MMSOILS
evaluates as sources for the food chain pathway include atmospheric deposition to plants, soil
erosion, and irrigation with contaminated ground water. Based on these sources, the foodchain
pathway examines the accumulation of a chemical within fish, cattle, and the edible parts of
terrestrial plants. A simple representation of bioaccumulation processes that uses
bioconcentration factors and transfer factors is used in the model. The bioconcentration factors
are used to represent the partitioning of a chemical between (1) water and fish, (2) edible parts
of terrestrial plants and soil, and (3) root vegetables and soil moisture. The transfer factors are
used to represent the uptake of chemicals by animals as a function of the mass of chemical
ingested in feed and water.
Aggregation of Long-Term Releases at the Facility Level
Because many facilities had several SWMUs that contributed to off-site contamination,
the Agency developed an approach for aggregating releases from SWMUs to provide facility-
wide estimates of long-term contamination. Because the estimates of current contamination were
presented at the SWMU level to allow the expert panels to determine which individual SWMUs
required corrective action, it was not necessary to aggregate releases for the current
contamination modeling; accordingly this discussion of aggregation techniques is limited to the
long-term modeling approach. Because exposure points were located around the sample facilities
based on media-specific methodologies, each environmental medium employed a different
aggregation approach as described below.
* * * DRAFT - March 20,1993
-------
3-24
Ground-Water Plume Aggregation
The Agency aggregated ground-water concentrations at off-site downgradient points using
one method for private/residential wells and another method for the remaining wells. These
methods may be demonstrated using Exhibit 3-6. For both cases, the aggregation method
assumed that each SWMU lay on the centerline of the grid. This assumption overestimated
exposures in some cases, while underestimating exposures in others. In general, if the size of the
facility was small compared to the size of the grid, the magnitude of the error was small because
the actual distance from the SWMU to the centerline would be small compared to the scale of
the off-site grid.
If the facility was larger, however, this assumption would tend to underestimate exposures
to points that are actually more directly downgradient (e.g., exposures to the public well from
SWMU A in Exhibit 3-6), and it would overestimate exposures if the SWMU and exposure point
were actually farther apart (e.g., exposures to the public well from SWMU B). The Agency
assumed that these errors would generally cancel each other when determining aggregate
exposure concentrations from several SWMUs at a facility. In addition, if the facility was very
large and the SWMUs were not evenly distributed, the Agency focused the grid on the portion of
the facility where the SWMUs were located in order to minimize the error.
For public wells, agricultural wells, and other explicitly defined points, the Agency
determined the downgradient distance (specified in the input file) from the unit to the exposure
point by taking the downgradient distance from the unit to the facility boundary (e.g., distance L,
for SWMU A) and adding the downgradient distance from the facility boundary to the specific
exposure point (e.g., distance Lj). The lateral (i.e., distance perpendicular to the gradient)
coordinate of the exposure point was taken as the distance from the grid centerline to the
exposure point (e.g., distance L,). Because the Agency assumed that the SWMUs lay on the
centerline, the lateral distance was not a function of the SWMU location. To determine
aggregated exposures at the explicitly defined points, a post-processor summed the SWMU-
specific concentrations at each point.
For private wells, average exposures were calculated for each ring instead of at each well.
For these calculations, the distances to the exposure points were not determined according to the
exact well locations, but were dictated by the geometry of the data collection grid (as shown in
Exhibit 3-6). To represent exposure concentrations within any one ring of the grid (e.g. all
points between 1.0 and 1.5 miles downgradient), two points were defined: the first exposure point
(e.g., Cj) lay on the grid centerline at the midpoint of the ring (e.g., at 1.25 miles), and the
second exposure point lay 33.75 degrees off the centerline at the midpoint of the ring. To
calculate average aggregate exposures, post-processing routines summed the SWMU-specific
concentrations for these points (generated by MMSOILS) to give aggregated concentrations at
the points. The routines then averaged these concentrations [(Cl +2xC2)/3] and multiplied the
average concentration by the total ring population to produce an aggregate exposure level for
each ring of the grid.
* * DRAFT - March 20,1993 * * *
-------
EXHIBIT 3-6
GRID FOR SWMU AGGREGATION
(HYPOTHETICAL EXAMPLE)
^^ Facility Boundry
1
SWMi: A
LI
l
1 SWMT'B
!-
i
Miles Downgradient
from
Facility Boundry
Shaded area represents one cell of the grid.
C = exposure points
-------
3-26
Because of symmetry, the offset point represented concentrations on both sides of the centerline.
Note that the exhibit only shows points within one of the six rings, but the post-processor made
separate calculations for each ring.
Aggregation of Atmospheric Releases
Atmospheric concentrations were estimated at the midpoint of each of five concentric
rings surrounding the facility from each SWMU. Concentrations for like chemicals were
summed directly to account for the effects of multiple SWMUs.
Aggregation of Surface Water Releases
Ambient concentrations in surface water were estimated at actual or hypothetical use
points in the surface water body. Concentrations from individuals SWMUs were summed up
across like constituents to account for the effects of multiple SWMUs.
Aggregation of Soil Releases
Ambient concentrations in off-site soils were calculated at actual or hypothetical
agricultural or residential fields. Concentrations from individual SWMUs were summed up
across like constituents to account for the effects of multiple SWMUs.
Modeling of Karst/Fractured Geologic Materials and Multiphase Flow
Karst terrain, fractured bedrock, and multiphase flow present special problems in
modeling contaminant transport in ground water. Karst aquifers are a special form of carbonate
aquifer, differing from other aquifers in that they contain an integrated system of pipe-like
solution channels that act as underground drains for the highly localized transport of water.
Fractured bedrock systems can occur within many different types of rock and are characterized
by an abundance of cracks, joints, faults, and other fissures in the rock. The Agency, for lack of
suitable alternatives, used the same ground water flow equations to model karst terrain and
fractured bedrock as were used for granular porous media. However, for karst terrain, special
parameter assumptions were made.
Multiphase flow is the transport of light non-aqueous phase liquids (LNAPLs) via
formation of a lens of constituent floating on top of the aquifer or transport of dense non-
aqueous phase liquids (DNAPLs) via gravity within or along the bottom of the aquifer. Since
screening level models like MMSOILS generally do not employ the complex algorithms needed
to simulate the behavior of NAPLs, EPA did not model multiphase flow. Instead, NAPLs were
modeled as if they behaved like aqueous leachate releases.29
"Specific assumptions employed in modeling karst/fractured conditions and NAPLs are
presented in Appendices B and G.
* DRAFT - March 20,1993 *
-------
3-27
Quality Assurance
Quality assurance for the RIA modeling effort fell into two main categories: data quality
and model quality. Data quality was assured though many stages during the development of the
corrective action RIA, as discussed in section 3.1.2. Quality assurance of the MMSOILS model
was achieved through a rigorous EPA review. Following this review, many improvements were
made to the model to allow for more flexible and accurate implementation for this analysis. In
addition, EPA applied the model in an iterative fashion to simulate the contamination at each
facility. In cases where monitoring or sampling data were available, EPA calibrated the input
data and assumptions to more closely approximate actual site conditions.
3.2 Results
This section presents estimates of the extent of contamination expected in the baseline
(i.e., if no corrective action were taken). These estimates were generally developed, unless
otherwise noted, from the long-term modeling and post-processing steps reviewed in earlier
sections. For some media, particularly releases to soils on-site, additional data sources, such as
actual soil sampling data, were used to present the extent of contamination.
Out of the total population of about 5,800 facilities subject to the RCRA corrective
action authorities, EPA projects that about 2,600 facilities (about 44 percent of the total
population) are expected to have past, current, or future releases to the environment that will
necessitate corrective action under RCRA Subpart S. Except for'the results of the facility
characterization, the estimates presented in this section focus on these 2,600 facilities and
exclude the remaining 3,200 facilities that are unlikely to require corrective action.30
3.2.1 Facility Characteristics
EPA's characterization of facilities potentially subject to the Subpart S proposed rule
resulted in an extensive set of data.11 This subsection presents three of the key results derived
from the sample:
SWMU characteristics;
Facility size (by area); and
Facility hydrogeology.
"Derivation of the estimate of 2,600 facilities that would require corrective action is
discussed in Chapter 4.
31 See Appendix I for additional information on the facility and SWMU characteristics.
* * DRAFT - March 20,1993 * * *
-------
3-28
SWMU Characteristics
EPA estimated the total number of SWMUs in the population by extrapolating from the
sample results. Exhibit 3-7 presents the estimate of the total numbers of SWMUs in the
population and the average number of SWMUs per facility." Exhibit 3-7 presents the total
number of SWMUs broken down by three categories:
Regulated RCRA Subpart F land disposal SWMUs (i.e., landfill, surface
impoundment, waste pile, or land treatment unit that managed hazardous waste
after 7/26/82);
Other SWMUs (e.g., solid waste landfills, routine and systematic spill areas, other
hazardous waste management units requiring permits or exempt); and
Areas of concern (e.g., one-time spill areas).
EXHIBIT 3-7
TOTAL SWMUs AND AVERAGE SWMUs PER FACILITY
(N = 5,800)
Regulatory Status of
SWMU
Subpart F Units
Other SWMUs
Areas of Concern
TOTAL
Number of SWMUs
4,600
97,000
1,700
104,000'
SWMUs Per Facility
0.79
17
0.29
N/A
* Numbers may not total due to rounding.
Exhibit 3-8 disaggregates the total number of SWMUs by SWMU type. Of the 14 types
of SWMUs listed in the population, an estimated 30 percent are tanks.
32 Note that the sample excluded the nine largest federal facilities and thus underestimates
the total number of SMWUs in the population.
* * * DRAFT - March 20, 1993 * * *
-------
3-29
EXHIBIT 3-8
NUMBERS OF SWMUs BY SWMU TYPE
(N = 5,800)
SWMU Type
Landfills
Land Treatments
Waste Piles
Surface Impoundments
Tanks
Incinerators
Injection Wells
Waste Transfer Stations
Waste Recycling Operations
Spill Areas
Accumulation Areas
Process Sewers
Other SWMUs
Areas of Concern
TOTAL
Number of
SWMUs
5.400
790
2,800
9.700
30.000
1,600
430
1,600
2,300
5,800
10.000
5.000
26,000
1,700
104,000*
Percent of
SWMUs
5.2
076
17
9.3
29
1.5
0.41
1.5
12
5.6
9.6
48
25
1.6
100
* Numbers may not total due to rounding.
Facility Size
During the facility characterization, the Agency determined the dimensions of the
facilities and calculated their total areas. Exhibit 3-9 presents the distribution of facility areas for
the approximate 5,800 facilities potentially subject to the Subpart S proposed rule. The smallest
facility is estimated to cover 0.1 acres, the largest is 8,900 acres, and the mean size is 330 acres.
Nearly 30 percent of all the facilities range between 10 and 100 acres in size.
* * * DRAFT -- March 20, 1993 * * *
-------
2,000
EXHIBIT 3-9
Distribution of Facility Areas
(N=5,800)
(G 1,500
1,000
500
0-1
1,700
1-10 10-100 100-1,000 1,000-10,000 Not Available
Area (acres)
-------
3-31
Facility Hydrogeology
Of the approximately 5,800 facilities in the population, 5 percent (270 facilities) are
expected to have either fractured bedrock or karst terrain. Fractured bedrock is expected to be
present at 250 of the facilities and karst terrain is estimated to exist at 20 facilities. At facilities
with fractured or karst terrain, estimating the contamination flow rate and direction is more
complex than at facilities with porous hydrogeology.
3.22 Baseline Ground-Water Contamination
Releases to ground water of hazardous wastes or constituents are projected to result in
on-site or off-site contamination at almost all of the 2,600 RCRA facilities expected to require
corrective action. Of these facilities, about 2,100 (80%) are projected to have releases above
action levels to on-site ground water. Only about 780 (30%) will have off-site ground-water
concentrations above action levels. Ground-water results are disaggregated by on-site versus off-
site due to the implications for future use, human health risk, and averted cost. Often on-site
ground water is used solely for industrial purposes rather than for residential or drinking water
use.
The expected difference between on-site and off-site contamination is due to several
factors. The most common cause is when the hydrogeologjc conditions and chemical properties
result in transport velocities that are too slow to lead to off-site contamination within the
modeling period. For example, facilities located in the southwestern United States may have
hydrogeologic conditions that slow contaminant transport. Similarly, releases of metals are often
slower-moving in ground water than organic constituents. In other cases, dispersion and
degradation may reduce the ground-water concentrations to values below the action levels before
the plume reaches the facility boundary. Additionally, at some facilities, the contaminant plume
may be intercepted by surface-water bodies or blocked by other geologic features before it is able
to travel off-site.
The extent of ground-water contamination can be expressed in terms of the total area
contaminated above action levels. Slightly different approaches were used to estimate the extent
of off-site and on-site contamination. For on-site contamination, the estimated nationwide extent
of ground-water contamination in the baseline is projected to be approximately 53,140 acres or
83 square miles (over the modeling period). Exhibit 3-10 presents the distribution of facilities
with varying degrees of on-site contamination. As indicated by this exhibit, about half of the
facilities will have one acre or less of on-site ground-water contamination.
* DRAFT - March 20, 1993 *
* *
-------
1,200
1,000 -
EXHIBIT 3-10
GROUND WATER
EXTENT OF BASELINE ON-SITE CONTAMINATION
(N=2,600)*
8 800
600 -
I
I 400
200 -
0
0
1-10 10-100
Plume Area (acres)
100-1,000 > 1,000
*Numbers may not equal due to rounding.
-------
3-33
Estimates for off-site contamination were based on the ground water grid used to
calculate exposures (see Exhibit 3-6). At each year of the simulation period, the Agency
calculated the total contaminated area based on whether concentrations were above action levels
in each ring of the grid. If concentrations exceeded the action level in a particular ring in a
particular year, the entire ring area was counted toward the total area of contamination for that
year. For each facility, the Agency used the maximum area of contamination in any given year
during the 128 year modeling period to calculate the reported extent of contamination. While
fewer facilities are likely to have off-site releases than on-site releases, the area! extent of
contamination is likely to be much greater (about 1,842,000 acres or 2,870 square miles over the
modeling period). Exhibit 3-11 presents the estimated frequency of facilities with varying
amounts of off-site contamination. Although most of the facilities will have contaminated areas
less than 5,000 acres, several are projected to have releases that contaminate almost 20,000 acres
over the modeling period.
About fifty-one constituents are projected to be detected above action levels at facilities
with ground-water contamination. Of these 51 constituents, 19 will be present above action
levels at 70 or more facilities. The predominant contaminants expected to be released to ground
water above action levels are presented in Exhibit 3-12, along with the concentration range
detected relative to the action level for the constituents.
3.2.3 Baseline Soil Contamination
EPA examined releases to both on-site and off-site soils. Volumes of contaminated on-
site soils were estimated based on soil sampling data, data on the size of units from which
releases originated, and expert judgement. Due to limited data on concentration gradients within
the contaminated soils, the Agency generally assumed a uniform concentration existed
throughout the depth of the SWMU. Releases to on-site soils are projected to result in
contamination above soil action levels at about 1,700 (68%) of the facilities likely to require
corrective action. About 20 percent (350 facilities) of these facilities are likely to have in excess
of 1 million cubic feet of contamination, as shown in Exhibit 3-13.
Releases to off-site soils were estimated using the MMSOILS model, which estimates
steady-state soil concentrations in off-site fields based on contaminant contributions from soil
erosion and atmospheric deposition. Off-site soil contamination was projected to exceed action
levels at approximately 200 (eight percent) of the facilities likely to require corrective action.
* * * DRAFT - March 23, 1993
-------
2,000
1,500
on
uu
Vl-l
o
s
-------
3-35
EXHIBIT 3-12
PREDOMINANT CONSTITUENTS ABOVE ACTION LEVELS IN GROUND WATER
(N=2,600)
CONSTITUENT
Chromium
Benzene
Methylene Chloride
Arsenic
Lead
Tetrachloroethylene
Trichloroethylene
Naphthalene
1, 1,2-TrichJoroethane
1, 1-Dichloroethylene
Methyl Chloroform
1, 1-Dichloroethane
1,2-Dichloraethylene
Toluene
Cadmium
Nickel
Aniline
Selenium
Xylenes
CONTAMINANT
CONCENTRATION
TO ACTION LEVEL
RATIO
(mg/1)
1 8,330
1 - 488,680
1 10,830
1 - 7,760
3 - 3.550
1 - 108,210
3-730
20 - 349,640
2 - 11.000
30-640
15 - 190
2-20
1-6
2-2,440
3 - 91,240
3 - 1.570
160-900
6-2060
1 -4
NUMBER OF
FACILITIES
WITH
CONSTITUENT
990
630
490
430
420
370
360
300
230
220
220
220
210
200
150
140
70
70
70
* DRAFT - March 20,1993
-------
EXHIBIT 3-13
EXTENT OF BASELINE SOIL CONTAMINATION
(N=2,600)*
1,000
800
03
OJ
G 600
Li.
'-<
o
i
.o
6
3
400
200
0
830
0
10-20 20-60 60-1,000 1,000-10,000
Thousands of Cubic Feet
*Numbers may not equal due to rounding.
-------
3-37
Tetrachloroethylene is expected to be detected above action levels for on-site soil at 440
facilities and will be the primary constituent found above action level in soil. The most prevalent
constituents expected to be detected above action levels in on-site soil are presented in Exhibit 3-
14.
EXHIBIT 3-14
PREDOMINANT CONSTITUENTS ABOVE ACTION LEVELS IN ON-SITE SOIL
(N=2,600)
CONSTITUENT
Tetrachloroethylene
Trichloroethylene
Chromium
Arsenic
CONTAMINANT
CONCENTRATION
TO ACTION LEVEL
RATIO
(mg/kg)
1 to 100
1 to 10
0.01 to 10
0.1 to 100
NUMBER OF
FACILITIES WITH
CONSTITUENT
440
280
220
220
3.2.4 Baseline Surface Water Contamination
Releases to surface waters are expected to result in contamination above action levels,
under central tendency modeling assumptions, at about 140 (5%) of the 2,600 facilities that are
likely to require corrective action. Concentrations above action levels were measured at the
point at which the contamination intersected any large, off-site surface water body. Off-site
contamination plumes may intersect small creeks or tributaries before reaching the surface water
body where the concentration was measured. In this case, the extent of contamination would be
underestimated due to the exclusion of the contaminant transport that would occur through
smaller surface water bodies. The number of facilities projected to have releases to surface
waters is small relative to the 980 facilities located within a mile of off-site surface waters. The
difference in the number with releases above action level and the number of facilities near
surface waters is due to the dilution of the contaminants that occurs when they are released into
large water bodies and the sedimentation that occurs in overland transport to the water body.
At least seven constituents are projected to be found in off-site surface water bodies
above action levels. These constituents and their concentrations are presented in Exhibit 3-15.33
33 Additional constituents were detected above ecological benchmark levels, as described in
Chapter 8.
* DRAFT - March 23,1993
* * *
-------
3-38
EXHIBIT 3-15
PREDOMINANT CONSTITUENTS ABOVE ACTION LEVELS
IN OFF-SITE SURFACE WATER BODIES
(N=2,600)
CONSTITUENT
Trichloroethylene
2,4-Dinitrotoluene
Tetrachloroethylene
Methylene chloride
3,3'-Dichlorobenzidine
Arsenic
Benzene
MAXIMUM
CONCENTRATION
TO ACTION
LEVEL RATIO
(mg/1)
2
20
70
2
10
10
2
NUMBER OF
FACILITIES WITH
CONSTITUENT
70
60
60
10
5
5
5
3.2.5 Baseline Air Contamination
Although several facilities were estimated to have on-site air releases above action levels
at the SWMU boundary, less than 1 percent of the facilities were projected to have air releases
above action levels at the facility boundary based on central tendency modeling assumptions.
The reason so few facilities have air releases above action levels is because the predominant
constituents in SWMUs are volatile organic compounds (VOCs). The rapid rate of dispersion
and volatilization that occurs with VOC releases to air decreases the likelihood of observing
concentrations above action levels at the facility boundary, particularly for very large facilities.
For example, we evaluated the half lives of roughly 35 VOCs in both soil and air. While the half
lives in soil ranged from 1/2 day to 180 days, averaging 46 days, the half lives in air ranged from
1 hour to 93 days and averaged 20 days. Therefore, VOCs released to air will have volatilized
within a matter of days and will not be detectable. Since many SWMUs are inactive or received
wastes in the past, any air releases that may have occurred from these SWMUs were not
detectable above action levels at the time that either monitoring data were collected or the
releases were modeled.
* DRAFT - March 20, 1993 * *
-------
3-39
3.3 Limitations
This section lists several key limitations inherent in the approach for assessing the extent
of contamination and associated exposures34.
Factors That Might Overstate Extent of Contamination
Conservative assumptions for landfill leachate quality. The use of the Organic
Leachate Model to derive central tendency leachate concentrations for organic
constituents often resulted in leachate concentrations above the solubility limits
for the constituents. Consequently, EPA "capped" leachate concentrations for
these constituents at their solubility limits. This may result in overestimates of the
extent of contamination under the central tendency scenario.
Limited data on "micro" topographic features affecting runoff and on soil volumes
eroded from units. EPA used available topographic maps to chart the overland
How of runoff from facilities to nearby off-site fields. However, due to the scale
of the topographic maps, the effect of "micro" features, such as small ditches and
gullies, on overland flow was not assessed. Due data limitations, EPA used expert
judgement in determining the sediment delivery fraction, i.e., the percent of
contaminated soil eroded from a unit that would reach an off-site field. These
factors could result in overestimates of the transport of constituent mass to off-
site fields.
Conservative assumptions for ground water transport in karst terrain. EPA
assumed for the central tendency scenario that there would be no retardation of
constituents in karst terrain. For the high end, EPA assumed that there would be
no retardation or dispersion. This scenario closely represents the conservative
case of pipe flow of leachates through solution cavities to exposure points. These
assumptions may overstate the concentrations of constituents at exposure points
down-gradient from facilities in karst terrain.
Factors That Might Understate Extent of Contamination
High-end scenario most applicable to pround-water pathway. Data and
assumptions for the high end scenario, while often resulting in greater constituent
mass available for release to all pathways, may in some cases inadvertently act to
34 Additional information on limitations of the MMSOILS model and its application may be
found in Appendix B.
* * *
DRAFT - March 20,1993
-------
3-40
reduce the mass released to pathways other than ground water." Since the
MMSOILS model employs a mass balance algorithm, the increase in mass
released to ground water in the high end may act to reduce the mass available for
release to other media and may reduce the modeled extent of contamination in
those media in the high end. Consequently, the high-end results for surface
water, soils, foodchain, and air pathways that are presented in the human health
benefits and ecological threats chapters of the RIA in some cases may not
represent a realistic high-end scenario for those pathways.
Very large POD/DOE facilities excluded from the analysis. These facilities are
likely to have significant releases, and excluding them could significantly
understate releases to the environment.
Catastrophic events not evaluated. Although the modeling of releases and the
assessment of the extent of contamination includes the possibility of unit failure,
catastrophic and stochastic events are not included in the analysis. For example,
releases from facilities are caused by hurricanes and flooding but these events are
not simulated in the analysis.
Releases from certain SWMUs not simulated. In a few cases, releases from
certain SWMUs were not simulated because of the unique character of the units.
These cases included direct dumping of wastes into lakes, rivers, or wetlands, and
complex waste management practices or units (such as releases from buried
ordnance or chemical warfare agents). Similarly, releases from process sewers
were not simulated because of the complexities involved in simulating failures
from lengthy sewer systems.
Multiphase flow not simulated. The MMSOILS model does not simulate the
multiphase flow that characterizes light and dense non-aqueous phase liquids
(NAPLs), but instead simulates the transport of these constituents as if they
behave like aqueous leachate releases. Simulating NAPLs as if they behave like
aqueous leachate releases underestimates constituent transport rates because the
model is unable to simulate releases from NAPL lenses outside of the unit.
35 The increase in mass released to ground water results from the higher leachate
concentrations assumed in the high end.
* » »
DRAFT -- March 20,1993
-------
3-41
Factors That Have An Indeterminate Effect on Extent of Contamination
General data, model, and knowledge uncertainties. Data uncertainties result from
both random and systematic errors as well as missing data. Random errors, e.g.,
due to inexactness of laboratory and field measurements, may lead to
uncertainties in values of model input parameters. Systematic errors introduced
in the processing of available data can add to uncertainty in modeling results,
although these errors can be minimized with adequate quality assurance and data
review. Model uncertainties result from model selection and model assumptions.
Model selection is based on conceptual and mathematical models of a physical
system and can have a significant impact on the way site-specific analyses are
conducted and on model results. Simplifying assumptions are necessary to
minimize data needs of the model and/or to make the solutions to the model
equations manageable; they may simplify or ignore certain processes. Knowledge
uncertainties result from incomplete understanding of the physical system. For
example, there are processes such as biodegradation which are known to affect
the fate and transport of chemicals in the environment, yet they cannot be
quantified because mechanisms governing the processes are not well understood
or data are unavailable on the rates at which the processes proceed.
Facilities with insufficient information excluded from the analysis. Excluding
these facilities introduces a non-response bias into, the analysis.
Limited monitoring data available for a number of sample facilities. Use of
default data and expert judgement may not accurately characterize the conditions
at the facility.
Limited data on volumes of contaminated soils. The extent of soils contamination
was generally determined based on a limited number of soils samples. As a result,
assumptions were often necessary to define the area and depth of contamination
that would exceed action levels. This could result in overestimates or
underestimates of the extent of soil contamination.
Limited number of modeling points in space and time. The level of detail in
space and time is also limited by computational and data management restrictions.
MMSOILS calculates exposure concentrations at a limited number of points for
each pathway, and these points must be specified before the model is executed.
Simplified approaches used to aggregate multiple releases to ground water. In
addition, EPA designated only one predominant direction of ground-water flow at
each facility to implement this aggregation approach. As a result, the resulting
estimates do not capture processes that occur in reality, particularly releases from
individual units moving in different directions at a particular facility, depending on
the time of year and the spatial arrangement of units.
DRAFT - March 20,1993 *
-------
3-42
Simplified assumptions for karst/fractured geologic conditions. The MMSOILS
model does not truly represent karst and fractured bedrock conditions in so far as
these systems are not characterized by isotropic, porous conditions. The model's
representation of conditions has an indeterminate effect on both the extent of
contamination and the concentrations at down-gradient wells. Special
assumptions were used at these facilities in an attempt to compensate for the
model deficiencies.
MMSOILS subroutines for each pathway not equally precise. The ground water
and atmospheric pathway subroutines are generally more precise than the soils,
foodchain, and surface water pathway subroutines (i.e., they better represent the
underlying environmental fate processes and provide estimates of time-varying
rather than steady-state concentrations).
* * * DRAFT - March 20, 1993
-------
4. REMEDY SELECTION AND MODELING OF REMEDY EFFECTIVENESS
This chapter describes the methodology for selecting remedies at sample RCRA
corrective action facilities and for modeling the effectiveness of these remedies in the corrective
action RIA. It also presents the results of the remedy selection process and effectiveness
modeling. Section 4.1 provides background on the Subpart S corrective action remedy selection
process. Section 4.2 describes the remedy selection expert panel methodology and presents the
methodology for remedy effectiveness modeling. Section 4.3 summarizes the results of remedies
selected at the sample facilities and the post-remediation extent of contamination results from
the remedy effectiveness modeling. Section 4.4 reviews the major limitations of the remedy
selection approach.
4.1 Background
Remedy selection represents the step in the corrective action process where EPA or an
authorized state selects final remedies at those RCRA facilities requiring cleanup. Actual
remedies have been selected at very few RCRA facilities nationwide to date. Accordingly, data
from only a few such remedy selections were available to characterize the kinds of remedies
likely to be selected under the Subpart S corrective action proposed rule. Lacking facility-
specific remedy selection information, the Agency had to develop a methodology for forecasting
the kinds of remedies that would be selected under the Subpart S proposed rule, in order to
estimate the costs and benefits of the program.
In developing a remedy selection methodology for this RIA, the Agency believed it was
critical that it capture the complex interactions between the EPA Region or State1 and the
facility owner/operator in the development of final remedies under the corrective action program.
Only by accounting for this interaction between policy goals and technical feasibility would it be
possible to estimate the kinds of remedies that would actually be specified under the Subpart S
proposed rule. The following summary of the Subpart S remedy selection process serves as a
background to the approach followed.
Proposed section 264.525 of Subpart S specifies the requirements for selecting corrective
action remedies. After a facility has submitted a final corrective measures study (CMS) report
(which presents alternatives for remediating contamination at the facility) to the Region, EPA
selects the final remedy. The remedy selected by EPA and implemented by the owner/operator
must meet the following remedy selection standards in the Subpart S proposed rule:
Protect human health and the environment;
Attain media cleanup standards as specified pursuant to section 264.525 (d) and
(e);
Control the sources of releases so as to reduce or eliminate, to the extent
practicable, further releases that may pose a threat to human health and the
environment; and
1 In the following discussion, EPA is used to represent either EPA or a State authorized for
the Subpart S regulations.
* * * DRAFT - March 20, 1993
-------
4-2
Comply with standards for management of wastes as specified in sections 264.550
- 264.559.
The Subpart S proposal also specifies five decision factors that may be weighted in
selecting remedies:
Long-term reliability and effectiveness;
Reduction of toxicity, mobility, or volume of wastes;
Short-term effectiveness;
Implementability; and
Cost.
Because of trade-offs among the remedy selection factors, different factors may dominate
the remedy selection process at different sites. At some sites the long-term effectiveness of the
remedy may be more relevant than its short-term effectiveness or implementability, while other
facility remedy decisions may be most strongly driven by the importance of reducing waste
toxicity, mobility, and volume. EPA's experience in Superfund has shown that different technical
approaches to remediation may be equally effective but vary widely in cost; in these situations,
cost can be used to discriminate between remedies. In some cases, it may be technically
impracticable to achieve certain cleanup goals, and proposed Subpart S provides EPA with the
authority to require alternative cleanup goals in such cases. The proposed Subpart S rule gives
EPA the authority to specify the final remedy.
Because of the inherent flexibility provided in the Subpart S proposal and the complex
interactions and technical issues that arise during the selection of remedies, the Agency
determined that it would not be feasible to develop a simple set of rules to follow in simulating
remedy selection for this RIA. A simple set of decision rules would not allow the consideration
of the kinds of tradeoffs between cost, effectiveness, timing, and other factors that the Agency
believes will be critical in the selection of site-specific final remedies. Accordingly, an approach
was developed that simulated as realistically as possible the general remedy selection framework
that would be followed under Subpart S. This remedy selection methodology is presented below.
4.2 Approach
This section presents the remedy selection expert panel methodology and the
methodology for simulating the effectiveness of the remedies that were selected.
4.2.1 Remedy Selection Expert Panel Methodology
Remedy selection is a key step in developing RIA cost and benefits results, as shown in
Exhibit 3-1 in Chapter 3. In order to account for the complexity of the decision-making process
when simulating the selection of remedies, EPA developed an approach that relied on panels of
experts to select remedies at the sample facilities. In order to capture the interactions between
EPA and the facility, two kinds of expert panels were convened:
DRAFT - March 20, 1993 * *
-------
4-3
Policy Panels: The policy panels represented the role of the regulatory agency in
setting remedial objectives, in assessing technical information from the technical
panels on the performance of potential remedies, and in making final remedy
selection decisions.
Technical Panels: The technical panels were charged with developing one or more
technical remedies for each facility, based on guidance from the policy panel, and
with estimating the costs of the remedies. The technical panels were encouraged
to develop a range of remedies, including those that would represent the facility
owner/operator's interest in minimizing the costs of any remedy while meeting the
environmental objectives of the regulatory agency.
The policy panels consisted of Regional EPA and State regulatory staff with extensive
experience implementing the corrective action program. They were identified and selected by
officials in EPA's Office of Solid Waste. Each policy panel consisted of six individuals,
representing a variety of EPA Regions and States so as to minimize any potential administrative
or geographic biases. The panel members were also selected to represent expertise in a variety
of technical areas, such as geology, engineering, and risk assessment.
The technical panels consisted of national remediation experts selected for their remedial
design experience. Each technical panel comprised individuals representing several disciplines,
including:
Hydrogeology;
Geology;
Geochemistry;
Soil science;
Civil, chemical, or environmental engineering; and
Chemistry.
The technical experts were identified through a competitive search across many well-
recognized remediation firms in the U.S. Many of the experts had significant RCRA field
experience, and most had extensive experience providing investigation and remediation support
under the Superfund program. Each technical panel consisted of six members selected to
represent in a balanced way the key disciplines listed above.
The remedy selection expert panel sessions were conducted over the course of eight
weeks in 1991 and 1992. The process involved the use of one policy panel and two technical
panels during each of two four-week sessions. The panels evaluated information on the extent of
contamination at 59 out of the 79 sample facilities where corrective action was projected to be
necessary.
Each of the panels during the sessions used a facilitator and a meeting recorder to direct
the discussions efficiently and document the results. When agreement among panel members
* * * DRAFT -- March 20, 1993 * * *
-------
could not be reached on an issue, the facilitator worked to resolve the conflict and reach
consensus among the panel members. If consensus could not be reached in a timely manner, the
facilitator initiated a vote on the issue. Majority vote constituted the resolution of the problem,
with minority opinion documented as a matter of record.
The policy panel evaluated between three and five facilities every day (either setting
remedial objectives at new facilities or selecting final remedies at facilities already addressed by
the technical panels). Because the specification of remedies by the technical panels required
more time per facility, it was necessary to convene two technical panels during each session, each
requiring between one-half to nearly two days to complete a facility, depending on the complexity
of the facility.
In addition to the panel members, facilitators, and recorders, several EPA personnel were
on hand throughout the expert panel sessions providing guidance and interpretation of the
proposed regulation and broader regulatory issues. At the start of the panel sessions, these EPA
personnel conducted a brief training session on the components of proposed Subpart S and the
ground rules for the expert panel process.
The ground rules included the following:
Regulated units subject to the Part 264 Subpart F ground-water monitoring
requirements were not to be addressed as part of the process;
Variances from the land disposal restrictions (LDRs) for contaminated soils would
be addressed according to the Superfund LDR "6A" guidance2;
Panels were to accept the facility characterization information provided to them as
representing actual conditions, even though much of this information was derived
from modeling based on limited site-specific data. Where there were significant
data gaps pertaining to facility information needed in selecting a remedy, the
panels were to make assumptions based on best professional judgement, with such
assumptions documented in the record;
Panels were to focus on national requirements under the Subpart S proposal and
not modify remedies based on current State requirements; and
"Pristine" background concentrations in environmental media were to be assumed
unless otherwise stated.
2U.S. EPA, Office of Solid Waste and Emergency Response, "Superfund LDR Guide #6A,
Obtaining a Soil and Debris Treatability Variance for Remedial Actions," No. 9347.3-06FS,
September 1990.
* * * DRAFT - March 20, 1993
-------
4-5
Agency officials were present throughout the expert panel sessions to resolve specific
questions concerning the interpretation of these ground rules or the applicability of current
Agency policy.
Exhibit 4-1 depicts the expert panel process and illustrates how the expert panels
simulated the proposed Subpart S remedy selection process. In the first step, the panel members
were presented with information characterizing the extent of contamination at each facility. This
information included overviews of historical facility operations, waste generation activities,
permitting and enforcement status, financial condition, environmental releases, and
characteristics of solid waste management units (SWMUs). Description was provided of the
wastes managed in the units and the constituents of most concern in the various media.3 When
available, actual monitoring data were used in characterizing the extent of contamination. For
example, soil samples and ground-water sampling data were available for a number of facilities
that had reached the RFI stage. When monitoring data were not available to estimate current
contamination at a facility, and when predicting future contamination, the MMSOILS model was
used to estimate the extent of contamination4. The panels were also provided with maps
identifying the locations of SWMUs at the facility and delineating contaminant plumes.
For each facility, the policy panel reviewed the data described above and developed
remediation objectives for each SWMU and for the releases of concern to soils, ground water,
surface water, and air. In developing facility-wide objectives, the panels followed the framework
of proposed Subpart S and indicated target cleanup levels that remedies would have to meet,
broad source control objectives, timing objectives, and whether CAMUs or conditional remedies
would be appropriate at the facility. In developing these objectives, the policy panel identified
the extent of current exposures at the site and made assumptions concerning the potential future
use of the site. The policy panel typically expressed remedial objectives as goals rather than
specific technologies.
The remedy objectives for each facility were presented to a technical panel, which then
developed detailed technical options for remediating the facility based on these objectives. In
developing remedies, the technical panels had access to reference materials on treatment
technologies, engineering design information, engineering costs, and, for ground-water extraction
remedies, plume capture computer models. Using these materials, they proposed technical
remedies for each SWMU source control required, for remediating ground water, for excavating
and treating soils, and for remediating any other site problems requiring corrective action.
Where more than one remedial alternative was feasible, the technical panels presented
alternatives for consideration. Finally, the technical panel developed rough cost estimates and
3 EPA selected constituents of most concern based on their concentrations relative to action
levels for environmental media and the distance they had traveled from the point of release.
4See Chapter 3 for a discussion of the MMSOILS model and its use in estimating the extent
of contamination.
* *
DRAFT -- March 20,1993 * * *
-------
Issues
Questions
and Remedies
EXHIBIT 4-1
EXPERT PANEL PROCESS
Site Characterization Inputs
Presentation of Facility and Summation
of Issues
Policy Panel Development
of Remedial Objectives
Technical Panel Development
of Remedial Alternatives
and Rough Cost Estimates
Policy Panel Selection of
Final Remedy
Technical Panel Development
of Detailed Cost Estimates
for Remedy
-------
4-7
qualitatively evaluated the performance of each remedial alternative against the five remedy
selection factors (e.g., long-term effectiveness).
In some cases, the technical panels believed that a remedial objective specified by the
policy panel was not feasible from an engineering standpoint. In such cases, the technical panels
and expert panels would meet to discuss these potential problems. These discussions could result
in the policy panel modifying their objectives or in directions to the technical panel to develop a
remedy that was practicable and protective even though it did not meet the original remedy
objectives.
After receiving the remedial alternatives from the technical panel, the policy panel would
review the proposed remedies and make the final remedy selection decision. In some cases, the
policy panel requested that additional alternatives be evaluated, or requested minor modifications
to a proposed remedy; the technical panel would develop this additional information and submit
it to the policy panel. Based on the final information provided by the technical panel, the policy
panel would select a final remedy for the facility.
After the policy panel selected a final remedy, the technical panel updated the rough cost
estimate to develop a detailed cost estimate for the remedy. These cost estimates were based on
current engineering cost estimation manuals and guidelines and were highly detailed. For
example, to estimate the cost of a ground-water extraction system, the panels would first
determine engineering specifications for the number of wells needed; the well casing depths,
diameters, and materials; the total pumping rate and pump specifications; piping for conveying
the pumped ground water to a treatment facility; the treatment facility design; and an NPDES
permit if required for discharge of the treated ground water. Based on these engineering
specifications, the panels would cost out each of these components of the ground-water remedy5.
Remedy Selection Quality Assurance
EPA monitored the expert panel process to ensure that remedial objectives were
consistent with the provisions of the proposed Subpart S rule. The policy panel and the two
technical panels were provided certain "ground rules" to ensure consistency between panels and
across facilities. Results of the expert panel remedy selection process were reviewed to ensure
that the selected remedies were consistent with the Subpart S proposed rule. Where calculations
were required (e.g., in determining costs), the numbers were double-checked for accuracy.
4.2.2 Modeling of Remedy Effectiveness
As shown in Exhibit 3-1 in Chapter 3, one of the next major RIA steps following
completion of the expert panel remedy selection process was simulation of remedy effectiveness.
5Chapter 5 presents a detailed explanation of the methodology employed for estimating the
costs of corrective action.
* * * DRAFT - March 20,1993 *
-------
4-8
Simulation of remedy effectiveness, in turn, was necessary to estimate the benefits of the Subpart
S proposed rule. EPA's remedy effectiveness approach is discussed below.6
The expert panels selected a wide variety of remedies to address releases to
environmental media. For the purposes of simulating their effectiveness, the remedies were
grouped into the following categories:
Source control technologies (e.g., clay caps that cover wastes left in place and
prevent infiltration of rain water and leaching of constituents, and excavation,
which removes waste or contaminated soils from the ground);
Waste treatment technologies (e.g., incineration, which destroys organic
constituents, and stabilization, which reduces waste permeability and immobilizes
inorganic constituents); and
Ground-water technologies (e.g., extraction wells, which contain the spread of
contaminated ground water and/or remove contaminated ground water from the
aquifer, and slurry walls, which prevent further migration of contaminated ground
water).
For some remediation technologies, e.g., soil vapor extraction, the expert panels often specified
the levels of effectiveness that the technologies were expected to achieve. However, in most
cases it was necessary to independently derive estimates of remedy effectiveness. Empirical data
on the effectiveness of remedies (e.g., data on the timing and magnitude of failures of clay caps)
are very limited, and EPA therefore used available data and expert judgement to develop
assumptions about the performance of remedies over the 128-year time-frame assumed in the
The purpose of the remedy effectiveness simulation was to assess how effective the
remedies would be in reducing the concentrations of contaminants in environmental media,
predicted using the MMSOILS model,7 that would occur in the absence of corrective action (i.e.,
in the baseline). Assumptions about remedy effectiveness were therefore incorporated in the
MMSOILS model in order to predict revised estimates of the extent of contamination with
remedies in place. Exhibit 4-2 summarizes how the technologies in the general categories above
were simulated using MMSOILS.
6 Appendix C contains a more complete discussion of the remedy effectiveness modeling
methodology.
7 MMSOILS was used to estimate the release of contaminants from SWMUs to
environmental media, the fate and transport of the contaminants in the media, and the resulting
concentrations at potential exposure points. See Chapter 3 for a discussion of the model.
DRAFT - March 20, 1993 *
-------
4-9
EXHIBIT 4-2
OVERVIEW OF REMEDY EFFECTIVENESS SIMULATION APPROACH
Source Control Technologies
Liners and
caps
Excavation
Clay, synthetic, and composite liners and caps were simulated directly with
MMSOILS. The thickness and material properties of each layer were simulated
along with estimates of the tuning and magnitude of material failure events.
The timing and extent of waste excavation were simulated directly with MMSOILS.
The percentage of waste specified by the expert panels was eliminated in the year of
excavation.
Waste Treatment Technologies
Stabilization/
solidification
Soil vapor
extraction
(SVE)
Incineration
Landfarming
MMSOILS simulates waste stabilization/solidification by reducing permeability of
wastes containing organic or inorganic constituents and by reducing leachate
concentrations of inorganic constituents; ex-situ stabilization is assumed to be more
effective than in-situ stabilization due to better mixing.
MMSOILS simulates SVE by reducing the contaminant concentrations in the waste
to post-remediation levels specified by the expert panels or to EPA default
effectiveness levels.
On-site incineration was assumed to be 99.9999 percent effective at destroying organic
compounds. As specified by the expert panels, metal-bearing ash would be disposed
either on-site or off-site.
Landfarming was simulated by reducing organic waste concentrations by 95 percent
unless specified differently by the expert panels.
Ground-Water Technologies
Ground-water
extraction
Slurry or
HDPE walls
French
drains
An exponential equation was employed to simulate ground water extraction. The
speed and effectiveness of cleanup depend on the contaminants, their concentration,
and the hydrogeologic setting.
Slurry walls and high density polyethylene (HDPE) barriers were assumed to be
completely effective in granular porous media, when used in conjunction with ground-
water extraction, and ineffective in fractured rock or karst settings.
French drains were assumed to be completely effective in granular porous media.
Off-Site Management of Remediation Wastes
Treatment
and disposal
All off-site management of remediation wastes was assumed to be completely
effective. These remedies were therefore not simulated with MMSOILS.
* * * DRAFT - March 20, 1993
* * *
-------
4-10
The MMSOILS model has the capability to simulate many of the remedies directly,
including many of the source control and waste treatment technologies. Other remedies,
including most of the ground-water remediation technologies, were simulated using a post-
processing program that modified the concentration outputs from MMSOILS. Exhibit 4-3
illustrates the process by which baseline MMSOILS information was modified in simulating
remedy effectiveness.
EPA employed a deterministic approach to simulating remedy effectiveness with
MMSOILS. For example, EPA assumed that all clay caps would be 80 percent effective in
reducing infiltration and leaching upon installation and that this effectiveness would drop to 20
percent after 100 years. The alternative to this deterministic approach would have been a
stochastic approach that incorporated probability distributions for failure events. For example,
varying times and/or magnitudes of failure could have been assigned to different clay caps, and
repeated model runs could have generated expected values for the leaching from the units with
these caps. A stochastic modeling approach was not adopted for simulation of remedy
effectiveness in this RIA because of the resources that would have been required to conduct
multiple model runs and the difficulty in developing probability distributions for failure events.
The sections below describe in more detail the simulation of remedy effectiveness for
source control, waste treatment, and ground water remediation technologies. Assumptions
concerning the effectiveness of off-site management of remediation wastes (e.g., "old wastes" and
contaminated soils) are also presented.
Effectiveness of Source Control Technologies
MMSOILS can directly simulate the installation of covers and liners over units where
wastes remain in place. As described in the MMSOILS documentation8, the water and mass
balance procedures in MMSOILS allow the user to simulate a variety of engineered waste
management units, from unlined and uncovered units to those with RCRA Subtitle C liners and
caps. The model simulates these barriers by reducing liquid flow through them as a function of
their hydraulic conductivity. In addition to accounting for these basic material properties,
MMSOILS also allows the user to simulate periodic failure events for these liner and cover
layers.
8 ni
"MMSOILS: Multimedia Contaminant Fate, Transport, and Exposure Model ~
Documentation and User's Manual," U.S. Environmental Protection Agency, Office of Research
and Development, September 1992, draft.
* * * DRAFT - March 20,1993 * * »
-------
EXHIBIT 4-3
Summary of Remedy Modeling Steps
Run MMSOILS and Post-Processing]
Programs for Baseline
Modify Baseline Modeling Inputs
to Simulate Source Control and
Waste Treatment Technologies:
- Liners
Covers
Excavation
- Soil Vapor Extraction
Soil Washini
| Rerun MMSOILS Model B
Run Ground-Water Remed;
Post-Processing Program
Estimate Post-Remediatio:
Extent of Contamination
-------
4-12
There is considerable uncertainty concerning the degree and timing of failures of clay and
synthetic materials used in liners and caps, but it is generally recognized that these materials do
not remain permanently intact. EPA estimated the timing and extent of failure for these
materials based on current engineering research.9 Clay caps and liners are assumed to have an
initial effectiveness of 80 percent and 70 percent respectively, which would decrease to an
effectiveness of 20 percent and 5 percent respectively after 100 years (if not replaced). The
remedy effectiveness modeling assumed that clay caps are replaced (by adding another layer of
clay) every 100 years; this modified their effectiveness at year 100 to 85 percent. In a simitar
manner, synthetic caps and liners are assumed to have an initial effectiveness that decreases with
time. Synthetic cap effectiveness ranged from 90 to 15 percent (without replacement); with the
RIA assumption of replacement after 100 years, synthetic caps have a final assumed effectiveness
of 90 percent. Synthetic liners have an assumed effectiveness ranging from 85 percent at
installation to 0 percent after 100 years. Composite liners and covers are somewhat more
effective than either clay or synthetic layers alone.
MMSOILS allows the user to simulate waste excavation by removing a specified
percentage of the waste mass remaining in the unit at a pre-specified time. The user can
simulate partial or complete excavation. When simulating waste excavation with MMSOILS,
mass that has previously been released to the unsaturated zone will continue to migrate towards
the saturated zone. If the entire waste mass is excavated, however, no additional constituent
mass will enter the unsaturated zone. For the remedy effectiveness modeling, EPA simulated the
percentage of waste excavated as indicated by the expert panels, which was typically 100 percent.
Effectiveness of Waste Treatment Technologies
MMSOILS can simulate the effectiveness of certain kinds of waste treatment, such as in-
situ or ex-situ solidification/stabilization and in-situ treatment technologies that eliminate waste
mass (e.g., soil vapor extraction).
In simulating solidification/stabilization, MMSOILS allows the user to decrease the
permeability of the waste layer (reflecting the effects of the stabilization process) and to reduce
the concentration of leachate being released from the solidified waste (reflecting the chemical
fixation of metals in the high pH cement matrix). Based on available data on the effectiveness of
these technologies, the RIA methodology assumed that in-situ stabilization decreased the waste
permeability to 1 x 10'7 cm/sec, while the more effective ex-situ stabilization decreased the waste
permeability to 1 x 10"' cm/sec due to the increased mixing of waste with the stabilizing materials.
It was also assumed that leachate concentrations for inorganic compounds would be reduced by
one order of magnitude as a result of chemical fixation processes. Because current engineering
studies suggest that stabilization is generally not effective in reducing the release of most organic
9 U.S.EPA, Office of Solid Waste, "Indexing of Long-Term Effectiveness of Waste
Containment Systems for Regulatory Impact Analysis," draft, December 1992.
* * DRAFT - March 20,1993 * * *
-------
4-13
compounds from wastes, the modeling methodology assumed that organic leachate concentration
would not change with this treatment technology.
In simulating treatment technologies that eliminate constituent mass such as SVE,
specific default treatment efficiencies were estimated for each technology. For example, it was
assumed that SVE was 90 percent effective for removing volatile organic compounds (VOCs), 50
percent effective for semi-volatile organic compounds (SVOCs), and ineffective for polynuclear
aromatic hydrocarbons (PAHs) and inorganic compounds. When the expert panels did not
specify site-specific treatment levels or efficiencies, default assumptions were used. Landfarming
was simulated by reducing organic waste concentrations by 95 percent unless specified differently
by the expert panels.
EPA assumed that on-site incineration would generate treatment residuals, with six order
of magnitude constituent destruction for organics, and that metal-bearing ash would be disposed
of on-site or off-site. EPA assumed that on-site incineration would not create releases of
concern to air, since these emissions are separately regulated under Subtitle C.
Effectiveness of Ground-Water Remediation Technologies
There is considerable uncertainty and debate concerning the effectiveness of ground-
water remediation in restoring contaminated aquifers. Moreover, while approaches have been
developed for using complex numerical computer models to simulate the effectiveness of ground-
water extraction remedies, field results indicate that actual effectiveness can be difficult to
predict and will depend greatly on site-specific heterogeneities and the nature and extent of the
contamination present. Accordingly, the RIA employed a relatively simple analytical method for
simulating the effectiveness of ground-water remediation activities consistent with the general
screening level nature of the analysis. While general in nature, the approaches described below
were selected so as to account for the primary factors believed to impact on the effectiveness of
the remediation.
The expert panels specified several kinds of ground-water remediation technologies at the
sample facilities: ground-water extraction wells; french drains; and slurry walls, HDPE walls, and
other hydraulic containment barriers. The effectiveness modeling for each of these ground-water
remedies was implemented in a different manner and is summarized below.
For ground-water extraction wells and french drains used to reduce contaminant
concentrations in the aquifer, EPA calculated post-remediation concentrations at down-gradient
wells using an analytical equation that takes into account the baseline concentrations, the
extraction system pumping rate, chemical-specific retardation processes, pumping duration, and
aquifer characteristics. The simulation approach was implemented through post-processing
software that modifies the baseline extent of contamination results for the ground-water pathway.
For those exposure wells within the extraction well drawdown area, the post-processing program
calculated a revised ground-water concentration profile. The time required to remediate the
plumes varied among sites. French drains were assumed to be 100 percent effective at
* DRAFT - March 20,1993 * * *
-------
4-14
intercepting ground-water plumes and were assumed to have no effect on existing ground-water
plumes that had migrated downgradient from the drains prior to their installation. The plumes
that had migrated beyond the french drains prior to their installation were assumed to naturally
attenuate unless specifically addressed by the expert panels.
The effectiveness of slurry walls and HDPE walls in preventing the further migration of
plumes was simulated as being dependent on the ground-water setting in which they are installed;
they are also assumed to degrade with time. EPA assumed slurry walls and HDPE barriers to be
completely effective in granular porous media, when used in conjunction with interior ground-
water extraction, and ineffective in fractured rock or karst settings. The complete effectiveness
of slurry walls and HDPE walls used in conjunction with interior ground-water extraction is due
to the inward leakage of ground water induced by the extraction system.
Several of the sample facilities were assumed to be contaminated with dense nonaqueous
phase liquids (DNAPLs). The expert panels often specified that it would be technically
impracticable to restore portions of aquifers contaminated with DNAPLs, though it would often
be possible to contain the DNAPL source within its current bounds in the aquifer. Accordingly,
the RIA assumed three cases for simulating the containment of DNAPLs: complete containment
in granular porous media, 75 percent containment in fractured rock settings, and 50 percent
containment in karst settings.
Effectiveness of Off-Site Management of Remediation Wastes
**
EPA did not simulate the effectiveness of off-site management practices for remediation
wastes (e.g., off-site incineration of hazardous wastes and Subtitle D disposal of non-hazardous
wastes). Instead EPA assumed that current RCRA requirements for management of hazardous
and non-hazardous wastes would ensure complete effectiveness in the treatment and/or disposal
of the remediation wastes sent to off-site facilities.
Effectiveness Modeling Quality Assurance
The quality assurance procedures for remedy modeling were similar to those followed for
the baseline MMSOILS modeling. Checklists were developed to allow staff to extract pertinent
remedy information from the results of the expert panel process. These checklists were then
used to modify the inputs to the MMSOILS model to simulate the remedies.
43 Results
This section summarizes the kinds of corrective measures selected by the remedy
selection expert panels for the sample facilities. It also presents the results of remedy
effectiveness modeling for a subset of sample facilities that had releases resulting in off-site
exposures of concern or extensive contamination of environmental media in the baseline.
* * * DRAFT - March 20,1993 * * *
-------
4-15
4.3.1 Remedies Selected
The expert panels identified a wide variety of corrective measures projected to be
employed under the proposed Subpart S proposed rule. Out of the universe of about 5,800
facilities, it is projected that about 2,600 facilities will conduct a CMS followed by the
implementation of corrective action. Of these facilities, about 2,200 are projected to have
releases above action levels to at least one environmental medium, i.e., ground water, surface
water, air, or soil, as shown in Exhibit 4-4. Approximately 400 facilities that do not have releases
exceeding action levels in environmental media were also found by the expert panels to require
corrective action. For example, they found that SWMUs at some facilities would require source
control in order to prevent future releases to ground water. Similarly, they instituted ground-
water monitoring at some facilities that have not released constituents in order to detect
potential problems in the future.
Ground-water remedies and soil cleanups are expected to be the most prevalent
remediation activities at RCRA facilities, with over 2,000 facilities likely to undergo corrective
action for ground water and/or soil. The RIA projects that relatively few facilities will require
corrective action for air releases, with only slightly more having surface water problems that will
be remediated. This follows from the fact that, as discussed in Chapter 3, very few facilities had
baseline releases of concern to air or surface water.
At the approximately 2,600 facilities projected to be remediated under the Subpart S
proposed rule, the expert panels specified that about 15,000 (20 percent) of the approximately
75,000 SWMUs at these facilities would be remediated. (These 15,000 SWMUs represent 14
percent of the total universe of approximately 100,000 SWMUs.) Exhibit 4-5 illustrates the
numbers of SWMUs by type projected to be remediated under Subpart S. As it shows, about
half of the 5,000 landfills, about 45 percent of the nearly 10,000 surface impoundments, and
about one-third of the waste piles, are projected to require corrective action under Subpart S.
While tanks are the most frequently occurring SWMU (nearly 30,000 tanks at RCRA facilities),
a smaller percentage of them (about 10 percent) are expected to require corrective action.
4.3.2 Post-Remediation Levels of Contamination
This section summarizes the projected levels of post-remediation contamination as
simulated through the remedy effectiveness modeling. These results pertain to a subsample of
720 facilities, expected to present significant risks or have extensive contamination, out of the
2,600 facilities expected to require corrective action. These facilities represent the same subset
of facilities addressed in the post-remediation discussion in Chapter 7, Human Health Benefits,
and are those facilities for which remedy effectiveness modeling was performed. Because of this
smaller subsample, the baseline extent of contamination discussed in this section is a subset of
that discussed previously in Chapter 3.
* * DRAFT - March 20,1993
-------
a;
'G
£
<£
D
o
-------
tyyTTTinTrr * _ c
TOTAL AND REMEDIATED NUMBER OF SWMUs
BY UNIT TYPE [100,000 TOTAL, 15,000 REMEDIATED]
N=2,600
35,000
p 30,000
J 25,000
w
s- 20,000
o
Jj 15,000
g 10,000
2 5,000
0
Total r^l Remediate^
"Misc." includes Areas of Concern, Injection
Wells, Incinerators, Transfer Stations, and
Recycling Operations
-------
4-18
Post-Remediation Ground Water Contamination
Ground-water remediation activities are expected to reduce contaminant concentrations
below action levels for a significant total area of ground-water, as shown in Exhibits 4-6 and 4-7.
As a result of the combined effect of source control activities and ground-water remediations, a
total of over 1 million acres of off-site contamination are expected to be restored to levels below
action levels before the year 2120 at the subset of 720 facilities. At the year 2120, it is predicted
that about 69,000 acres of contaminated ground water will remain off-site at the 720 RCRA
facilities, significantly less than the almost 1,500,000 acres of contaminated ground water
predicted in the baseline at the 720 facilities. For on-site ground-water contamination, more
than 20,000 acres are expected to be remediated by the year 2120, with about 5,000 acres out of
the baseline contamination of about 28,000 acres remaining above action levels.
The time required to remediate both off-site and on-site contaminated ground water is an
important measure of the effectiveness of the remediation activities. The median time to
remediate contaminated ground-water in this subsample of 720 facilities is predicted to be about
90 years for off-site plumes and 115 years for on-site plumes. These results indicate that it will
often take longer to reduce on-site ground water concentrations to action levels than to reduce
off-site concentrations. This trend reflects the fact that plumes in some slow-moving ground-
water systems do not significantly extend beyond the facility boundary, and the slow-moving
nature of the ground-water impedes remediation. The presence of dense non-aqueous phase
liquids (DNAPLs) is another factor that is expected to contribute to less effective on-site
cleanups. The most concentrated pockets of DNAPL are typically located beneath the operating
portions of the facility. In many cases, the expert panels believe it would be technically
impracticable to restore areas surrounding the DNAPL source to action levels, and they believed
that containment of the DNAPL plume on-site would be the maximum remediation practicable.
While not all of the contaminated ground-water is expected to be remediated to action
levels, many of these plumes are expected to contain multiple chemicals, many of which would be
remediated in less than 100 years. Because the effectiveness of ground-water extraction depends
on the mobility of the individual constituents in the plume, their initial concentrations, and
ground-water flow velocity, pump and treat systems may remove some of the chemicals in a
relatively short period of time, while others will not be effectively remediated. As Exhibit 4-8
indicates, about 25 percent of the constituents present in ground-water plumes at the subset of
720 facilities will be remediated to action levels in less than 40 years; about 75 percent of these
constituents are expected to be reduced below action levels within about 100 years. The
remainder may not be remediated due to their lack of mobility (which impedes their removal),
their high initial ground-water concentrations (e.g., contaminants present in DNAPL pockets), or
because they are in hydrogeologjc settings intrinsically difficult to remediate (e.g., karst or
fractured bedrock).
* * * DRAFT - March 20,1993 * * *
-------
EXHIBIT 4-6
EXTENT OF ON-SITE GROUND-WATER CONTAMINATION
BASELINE AND POST-REMEDIATION
N=720
400
250
Baseline
Post-Remediation
EXHIBIT 4-7
EXTENT OF OFF-SITE GROUND-WATER CONTAMINATION
BASELINE AND POST-REMEDIATION
N=720
250
0-1
1-5 5-15
Contaminated Area (thousands of acres)
15-20
Baseline I I Post-Remediation
-------
I
I 0.8
CJ
0.6
5 0.4
3
!/!
= 0.2
o
CB
EXHBIT 4-8
EFFECTIVENESS OF GROUND WATER REMEDIATION
20 40 60 80
Time After Remedy Implementation (years)
100
120
-------
4-21
Post-Remediation Soil Contamination
EPA examined the effect of corrective action remedies on both on-site and off-site soil
contamination. The Agency found that, of the 720 facilities addressed in the remedy
effectiveness modeling, about 650 (90 percent) had on-site contaminated soil that was
remediated by the expert panels. The RIA projects that the 18 million cubic yards of
contaminated soil10 at the 720 facilities would be remediated using several technologies,
including excavation followed by treatment and disposal, in-situ treatment (e.g., soil vapor
extraction), and capping with clean cover.
Approximately 200 facilities, among the 720 facilities examined for the post-remediation
analysis, were projected to have baseline releases to off-site soil above action levels. Predictions
of the post-remediation reduction in soil contamination are uncertain and could potentially range
from complete reduction in contamination to minimal reduction. Implementation of source
control measures at facilities during remediation would effectively eliminate further erosion or air
deposition contributions to off-site soils, and natural attenuative processes would act over time to
reduce any constituent mass remaining in the soils after remediation. Under this scenario, the
reduction in soil contamination would be complete. However, due to its use of the steady-state
algorithm for off-site soils, MMSOILS does not effectively distinguish post-remedial conditions
from pre-remedial conditions. Model results therefore indicate little decline in post-remedial off-
site soil concentrations; reductions in off-site soil concentrations to below action levels were
predicted at only four of the 200 facilities with soil concentrations of concern in the baseline."
This represents the minimal reduction scenario.
Post-Remediation Surface Water Contamination
Out of the subset of 720 facilities analyzed for remedy effectiveness, only 73 facilities (ten
percent) were projected to have releases to surface water above action levels in the baseline.
Predictions of post-remediation reductions in surface water concentrations at these facilities are
uncertain, due, in particular, to the uncertain post-remediation contribution of soil erosion to
streams. As discussed under soil contamination above, a modeling limitation is likely to result in
post-remedial erosion contributions to surface water being overstated. Based on the current
modeling approach, remedy effectiveness for surface waters would be less than complete.
10 This estimate of the volume of contaminated soil does not include any soil that may have
been intermingled with "old waste" from SWMUs.
11 Part of the reason that the model does not effectively distinguish post-remedial conditions
from pre-remedial conditions is that the model currently assumes that soil erosion from units
remediated by capping continues to contribute constituent mass to off-site soils (as well as
surface water). Correction of this limitation of the model would likely result in more significant
estimated reductions in soil (and surface water) contamination due to corrective action.
* * DRAFT - March 23,1993 * * *
-------
4-22
As indicated previously, relatively few corrective measures were instituted directly to
remediate surface water problems. In many cases, however, source control and ground-water
remedies are expected to be effective in reducing surface water contamination. Because overland
runoff of contaminated soils and ground-water discharge to streams are expected to be the
primary sources of surface water contamination at corrective action facilities, source control and
ground-water remediation activities would also prevent releases to surface water. Of these 73
facilities, all are predicted to implement corrective measures (e.g., french drains, caps, and runoff
collection) that would effectively eliminate the transport of contaminants into surface water
bodies. Based on this scenario, remedy effectiveness would be complete.12
Post-Remediation Air Contamination
None of the facilities in the subset of 720 facilities evaluated in the remedy effectiveness
modeling had releases of concern to air in the baseline. Consequently, the effectiveness of
remedies at reducing air risks was not quantified.
4.4 Limitations
Many of the limitations in characterizing facilities and modeling the extent of releases in
Chapter 3 also apply to the selection of remedies and the modeling of remedy effectiveness in
this chapter. Some additional limitations specific to this chapter are presented below.
4.4.1 Factors that are likely to understate remedial effectiveness
' Uncertainty concerning effectiveness of remediation technologies. There is
considerable uncertainty concerning the effectiveness of many remediation
technologies, in particular the long-term effectiveness of caps and liners and the
effectiveness of ground-water extraction remedies. The failure rates for liners,
caps, slurry walls, and HDPE barriers used in the remedy modeling were
conservative estimates based available engineering data and best professional
judgement.
4.4.2 Factors that are likely to overstate remedial effectiveness
* Some remedies assumed completely effective. EPA assumed that off-site
treatment and disposal would be completely effective. EPA also assumed that on-
site incineration would generate treatment residuals (with six order of magnitude
12 Modeled remedy effectiveness would be complete in terms of the constituent mass
dissolved in surface waters. MMSOILS does not model the effect of contaminant releases on
constituent concentrations in sediments in surface water bodies. Source control and ground
water remedies by themselves may not address constituent concentrations in sediments.
DRAFT - March 23, 1993 * *
-------
4-23
constituent destruction for organics), but would not create releases of concern to
air (which are separately regulated under Subtitle C).
4.4.3 Factors that have an indeterminate effect on remedial effectiveness
Limited site-specific data relative to that needed to specify actual remedies. The
actual specification of remedies requires extensive site-specific data and involves
many steps that could not be simulated by the expert panels, including:
bench and pilot-scale treatability studies for determining the most
appropriate waste treatment technologies for wastes, soils, and ground
water;
pump tests to determine local ground-water flow patterns for specifying
proper ground-water extraction well locations and pumping rates; and
ongoing performance evaluation of ground-water extraction remedies
(which may result in changes to the pumping program).
Uncertainty in remedy effectiveness where DNAPLs present. The simple
analytical approach used to estimate the effectiveness of ground-water extraction
remedies does not account for many of the complexities that are likely to affect
the success of remediation. In particular, there is significant uncertainty
concerning the effectiveness of extraction remedies on DNAPL plumes.
Uncertainty in expert panel assumptions. In some cases, the expert panels could
not specify a remedy based on the information provided to them, and it was
necessary for them to make important assumptions concerning the facility
contamination and environmental setting. The representativeness of the remedies
selected in such cases will be dependent on the accuracy of these assumptions.
Use of deterministic modeling approach. The use in this RIA of a deterministic
modeling approach to remedy effectiveness has a generally indeterminate effect
on the model results. Stochastic modeling can be used to provide upper and
lower bounds on model predictions in order to characterize the level of
uncertainty in the results. It is not clear where the results presented in this
chapter would fall within the distribution of results from a stochastic analysis.
DRAFT -- March 23,1993 * * *
-------
5. COSTS
EPA continues to develop the regulatory impact analysis to better cost out options in
the final regulatory development of Subpart S. As such, the Agency has chosen to cost out the
proposed regulation; cost savings are measured as the increment between the cost of the
proposed regulation and the baseline. Cost estimates provided in this draft of the RIA are
preliminary and for illustrative purposes only; any final decision made by the Agency is
contingent upon comments received by the Agency.
This chapter analyzes the potential compliance costs for hazardous waste management
facility owners/operators under the proposed RCRA corrective action regulations. The first
section of the chapter describes the approach for estimating corrective action costs; the second
section presents the results of the cost analysis at the national level and provides some
interpretation of the results; the third section details the sensitivity analyses; the fourth section
reviews the available information on remediation costs at the largest federal facilities subject to
RCRA corrective action requirements; and, finally, the fifth section discusses major limitations to
the cost analysis.
5.1 Approach
The remedial cost estimates provided by expert panels comprised of EPA Regional and
state personnel and remedial contractors were the basis for calculating facility level costs for this
analysis. As described in Chapter 4, the expert panels that were convened to assist with the
analysis of the RCRA corrective action rule initially reviewed each facility in the RIA sample
and selected remedies to address the contamination at those facilities. The panels then prepared
cost estimates for each remedial activity on both a facility-wide and a solid waste management
unit (SWMU) by SWMU basis. After the total costs were adjusted to include design, oversight,
and contingency assumptions, the costs were discounted to provide a total present value cost for
the outlays over the time period of the analysis. All results presented in this chapter are based
on the 1990 proposed rule; other regulatory options that EPA plans to evaluate in the future
may increase or decrease costs relative to the proposed rule.
5.1.1 Scope of Analysis
The costs of regulation can be defined and measured in various ways. Economists define
social cost in terms of the welfare of individuals. The total social cost of a regulation would be
the sum of welfare costs across all individuals in society, measured as the change in consumer
and producer surplus. Measuring social costs requires estimates of all the changes in incomes
* * * DRAFT-March 23, 1993 * *
-------
5-2
and in the supply and prices of goods that result from regulation, and may require general
equilibrium models of the economy.1
Because of the complexity of this kind of analysis, most estimates of the cost of regulation
have measured compliance costs, or the expenditures required by private firms. Compliance
costs should be measured as the opportunity cost of the resources required, or the value
foregone by not using the resources in their next best use. Because this value may be difficult to
measure, many analyses use the accounting cost of resources instead. Since opportunity and
accounting costs are often not the same, this may distort the cost analysis. As an example, a
facility may require land to set up remediation equipment. If the facility owns the land, the
accounting cost of its use is zero. However, the opportunity cost of the land is not zero; it is the
value of potential alternative uses of the land, perhaps from expansion of the firm's productive
activities.2
There are several sources of divergence between the cost analysis in this RIA and an
analysis of the full social costs of corrective action:
EPA does not attempt to use a general equilibrium analysis to estimate regulatory
costs. Corrective action affects a broad spectrum of industries and markets, and
modeling the general equilibrium effects of compliance expenditures in all of
them would be extremely complex.
The analysis does not measure the administrative costs incurred by government
agencies overseeing the regulatory process, including the costs of RCRA Facility
Assessments (RFAs).
EPA uses the prices of remedial technologies, not their costs. The market price
of any good may differ from the true cost of providing it in terms of society's
resources. In competitive markets, price is equal to marginal cost in equilibrium.
However, in non-competitive markets, suppliers may restrict output and raise
prices above marginal cost. If markets for remediation activities are non-
competitive, then price may not be a good measure of the social cost of
remediation.
1 Cropper, Maureen L. and Wallace E. Oates. "Environmental Economics: A Survey,"
Journal of Economic Literature. Vol. 30, (June 1992), pp. 675 - 740. Cropper and Oates give
two examples of general equilibrium analyses of environmental regulation. One, by Hazilla and
Kopp (1990), used a general equilibrium model of the U.S. economy to calculate the social costs
of the Clean Air and Clean Water Acts. The other, by Jorgenson and Wilcoxen (1990),
measured the effects of environmental regulation on U.S. economic growth.
2 See Friedman, Lee S. Microeconomic Policy Analysis. New York: McGraw-Hill Book
Co., 1984; and Stokey, Edith and Richard Zeckhauser. A Primer for Policy Analysis. New York:
W.W. Norton and Co., 1978.
* * DRAFT-March 23, 1993 * * *
-------
5-3
In conducting this analysis, EPA made two additional assumptions that affect the scope of
the cost analysis. First, the analysis addresses only those facilities where corrective action would
occur because of the HSWA corrective action mandate.3 EPA assumed that cleanups mandated
by other federal authorities would have occurred in the absence of HSWA and thus did not
estimate the costs associated with these facilities. In particular, the RCRA Subpart F corrective
action program and the CERCLA (Superfund) program had been addressing and remediating
facilities prior to enactment of HSWA. However, EPA did assess the potential costs of
remediation at the very largest federal facilities subject to RCRA. These costs were derived
primarily from federal agency reports and budget projections (see section 5.4). Facilities subject
to other cleanup authorities (e.g., state cleanup programs that would potentially mandate
remediation in the absence of federal RCRA authority) also were not included in the analysis.
Second, because the Subpart S corrective action RIA estimates the total corrective action
costs resulting from the HSWA corrective action authorities (enacted in late 1984), the analysis
includes estimates of remedial costs incurred by facilities from 1985 to 1992, as well as those
remedial costs projected to take place in the future. The expert panels determined which
activities were attributable to the corrective action program (remediation that occurred after
1984) and estimated costs for these activities in terms of current 1992 dollars. The analysis
assumed that these costs occur in 1992 for all facilities. Because the corrective action program is
relatively new, the remedial costs from 1985 to 1992 are a small portion of the total projected
costs. Therefore no adjustment was made for the opportunity costs of money already expended
on remediation.
The RIA sample consisted of 79 federal and non-federal facilities.4 This stratified
random sample was drawn from the universe of facilities potentially subject to the RCRA
corrective action authorities (i.e., about 5,800 facilities). The cost analysis was designed to
estimate the incremental cost of implementing corrective action at the facilities within this
universe that have a release of concern. As discussed in Chapter 3, the corrective action
universe also includes nine 'Very large" Federal facilities. The RIA sample does not include any
of the very large Federal facilities, and the expert panels did not estimate their remediation costs.
Instead, EPA presents separately in section 5.4 estimates of costs at these facilities derived by the
agencies responsible for them.
5.1.2 Expert Panel Cost Estimates
As discussed in Chapter 4, EPA convened panels of hazardous waste remediation experts
to review and develop remedies for each facility in the RIA sample. The remedial experts
estimated the cost of each remedy based on each remedial technology. The panels specified each
element of the remedies in great detail. For example, cross sectional diagrams, including
materials and distances, were often drawn for each liner, and extraction system designs were
often specified down to the length and diameter of well casing. The detailed remedy designs,
3 See Chapter 2 for a discussion of the baseline in this analysis.
4 The RIA sample is discussed in Chapter 3.
* * * DRAFT-March 23, 1993 * * *
-------
5-4
professional experience, and information from cost-estimating and engineering publications were
used as the basis of the expert cost estimates. Because of their expertise in specific aspects of
remediation (e.g., ground-water extraction or waste treatment technologies), panel members were
able to provide cost estimates that accurately reflect current remediation practices. The panels
also considered facility-specific and regional factors when developing costs. For example, the
cost of drilling ground-water extraction wells was dependent upon site-specific drilling media and
aquifer characteristics. Likewise, the cost of off-site waste incineration was dependent upon the
type of waste being incinerated and how far the facility was from the off-site incinerator. The
panels estimated all costs in current (i.e., 1992) dollars.
Whenever possible, the panels provided unit costs for remedial activities (e.g., the cost of
soil excavation in terms of dollars per cubic yard of soil). In these cases, the panels also
provided volumes of remedial material being managed. Within and across facilities, unit costs
may also vary according to regional and facility-specific characteristics.5
The cost information developed by the expert panels was structured around individual
remedial activities (e.g., piping for ground-water extraction wells) at individual SWMUs. For
each remedial activity, a number of variables were developed:
Total cost of the activity;
Type of cost (capital, O&M, or investigation);6
Timing of the activity (year the activity begins, how long it lasts, and how
frequently it occurs); and
Unit cost ihformation-the preliminary costs used by the expert panels to calculate
the total cost of each activity (e.g., the cost per cubic yard to excavate soil).
For all facilities requiring corrective action, the expert panels provided investigation costs
(i.e., for the RCRA Facility Investigation (RFI) and Corrective Measures Study (CMS)). A few
facilities required a RFI but did not trigger corrective action, and thus did not require a CMS.
EPA calculated RFI costs for these facilities based on the average cost of RFIs conducted at the
other facilities in the sample (adjusted to account for differences in the number of SWMUs per
facility). As estimated by the expert panels, RFI costs are expected to range from $80 thousand
to $16 million per facility, in undiscounted 1992 dollars. Costs to complete CMSs are expected
to range from $20 thousand to $56 million per facility, again in undiscounted 1992 dollars.
5 Appendix D provides a more in-depth discussion of the expert panel cost estimation
process with a focus on unit cost calculations.
6 In this analysis, capital costs included expenditures for equipment, installation of
equipment, and initial remedial activities (e.g., soil excavation). O&M included ongoing
maintenance activities (e.g., electricity for running ground-water extraction pumps). Investigation
costs included RFIs, CMSs, and other investigatory activities (e.g., soil gas surveys).
* * * DRAFT-March 23, 1993 * »
-------
5-5
5.1.3 Timing
Due to the large number of facilities requiring corrective action, full implementation of
the corrective action program will require many years. Facility investigations and remediations
have been underway since enactment of HSWA in 1984 and are ongoing. For this analysis, EPA
assumed that corrective action will occur at the worst facilities first (classified as high priority)
and made timing assumptions based on facility priority:
The National Corrective Action Prioritization System (NCAPS) was used to rank
facilities as high, medium, or low priority;7
For high priority facilities, the RFI will occur between 1992 and 1997 (randomly
assigned);
For medium priority facilities, the RFI will occur in 1997; and
For low priority facilities, the RFI will occur in 2002.
The analysis further assumed that the RFI is completed over two years, the CMS occurs in the
third year, and remedial activities begin in the fourth year. Remedial activities continue for the
period of time specified by the expert panels. In some cases, durations of pump and treat
remediations for ground water were based on modeling results. For consistency with the benefits
analysis, the cost analysis covers the 128-year time period from 1992 through 2119.
The cost analysis additionally incorporated the costs of replacing containment systems
(e.g., caps and HDPE walls will be replaced every 100 years, and slurry walls will be replaced
every 30 years). The replacement assumptions were based on EPA's best current professional
knowledge, as described in Chapter 4. In all cases, the analysis assumed that the replacement
cost was equal to the initial installation cost. Likewise, the analysis included costs of replacing
equipment used in pump and treat remedies (e.g., piping, pumps, treatment facilities) every 20
years, unless otherwise specified by the expert panels. Exhibit 5-1 shows the distribution of
estimated remedial costs over time, disaggregated by priority. Note that some costs occur prior
to completion of the RFI for low and medium priority facilities. These costs reflect remedial
actions that have been implemented prior to 1992. The peaks in the time stream represent the
20, 30, and 100-year equipment replacement cycles for certain remedial measures (e.g., caps).
5.1.4 Discount Rates
Because the costs of corrective action are incurred over many years in the future, EPA
discounted costs to a 1992 present value. By.discounting to a common year, options that are
implemented over differing years can be compared directly. The rationale for discounting is
7 U.S. Environmental Protection Agency, Office of Solid Waste (prepared for EPA by PRC
Environmental Management, Inc.). Draft RCRA Corrective Action Prioritization System
Guidance Document. Washington, D.C.: U.S. Environmental Protection Agency, April 1991.
* * DRAFT-March 23, 1993 * * *
-------
EXHIBIT 5-1
ANNUAL NATIONAL COSTS FOR CORRECTIVE ACTION
BY NCAPS FACILITY PRIORITY [N = 3,500]a
(in billions of 1992 dollars)
$8 -
1
O
o
C/5
c
O
$6 -
S $4
U
C
O
a
C3
'"§
$2 -
$0
1992 2000 2008 2016 2024 2032 2040 2048 2056 2064 2072 2080 2088 2096 2104 2112 2120
Total High Priority Medium Priority Low Priority
3,500 facilities = 2,600 likely to require corrective action and 9^ ' to incur RF1 costs only.
-------
5-7
presented in greater detail in EPA's RIA guidance.8 In addition, to estimate annual costs of the
corrective action rule, EPA calculated annualized costs from the present value estimates (i.e., by
assuming that costs are amortized over 20 years). The 20-year time period was chosen in order
to be consistent with other EPA and OSW RIAs.
Throughout this analysis, a seven percent discount rate was used. Although there is no
single, correct discount rate, the seven percent real discount rate was selected because it
approximates the marginal pretax rate of return on an average investment in the private sector in
recent years. The seven percent discount rate is appropriate for cost analyses of public
investments and regulatory programs that impose costs on the general public. Public investments
and regulations displace both private investment and consumption; the seven percent discount
rate accounts for this displacement.9
5.1.5 Quality Assurance
Due to the large quantity of data processed while conducting this cost analysis, special
attention was paid to guaranteeing the accuracy of the data and minimizing the possibility of
minor errors (e.g., errors of transcription.) Quality assurance for the RIA cost analysis consisted
primarily of internal EPA review of 4 steps: (1) reviewing and adjusting the expert panel cost
information; (2) ensuring that costs were correctly calculated based on unit costs; (3) verifying
that unit costs used by the expert panels were within ranges documented in engineering
literature; and (4) ensuring that data were correctly transcribed prior to analysis. Remedies and
costs developed by the panels were reviewed to guarantee that costs had been estimated for all
appropriate remedial activities. To prevent typographical errors, the data entry process included
many rounds of comparing input data to the expert panel information. Additionally, computer
programs were run to check that anomalous information had not been entered.
5.2 Results
This section presents the results of the cost analysis including the national present value
and annualized costs for the proposed corrective action rule. The results from the sample
facilities were extrapolated to the national level by multiplying facility-level results by facility
weights corresponding to their sample strata. The weights were developed as part of the sample
8 U.S. Environmental Protection Agency, Office of Policy Analysis. Guidelines For
Performing Regulatory Impact Analysis: Appendix C - Analysis of the Choice of Discount Rates.
Washington, D.C.: U.S. Environmental Protection Agency, 1989.
9 Office of Management and Budget. Guidelines and Discount Rates for Benefit-Cost
Analysis of Federal Programs - Circular A-94. Washington, D.C.: Office of Management and
Budget, October 29,1992, page 9.
* * *
DRAFT-March 23, 1993 *
-------
5-8
selection process.10 In addition to the national cost of corrective action, the distributions of
facility and SWMU-level costs are presented.
5.2.1 National Cost of Corrective Action
The total national present value cost of the RCRA corrective action program is projected
to be about $19 billion (in 1992 dollars). These costs are likely to be incurred by about 3,500
facilities, of which 2,600 will require corrective action and 900 will incur RFI costs alone. The
annualized cost (using a 7 percent rate over a 20-year period) is projected to be $1.8 billion.
Since the projected costs are based on information obtained from a sample of the population, it
was necessary to construct confidence intervals around the estimate. The 95 percent confidence
interval ranges from $11 billion to $26 billion.11 The results incorporate additional uncertainties
that have been discussed throughout this RIA (e.g., in modeling fate and transport of releases).
Most of the cost of the corrective action rule is projected to be incurred as capital costs.
Capital costs were defined by EPA in this analysis as including expenditures on equipment and
installation of equipment, as well as the cost of initial remedial activities (e.g., excavation,
clearing, and grubbing). The breakdown of the projected total corrective action cost is 56
percent capital costs, 33 percent operations and maintenance, and 11 percent investigation.
These costs are shown in Exhibit 5-2.12
Results for federal and non-federal facilities
The total present value cost for remediation and investigation of the 2,500 non-federal
facilities and 60 federal facilities triggering corrective action is projected to be $16.6 billion and
$2.1 billion, respectively. While federal facilities are expected to incur roughly 11 percent of the
total present value cost of remediation, they represent only 2 percent of the facilities likely to be
remediated under RCRA authority. The number of federal facilities included in this analysis is
small because most federal facilities are likely to be remediated under CERCLA (or other
federal authorities) rather than under RCRA. As part of this analysis, EPA estimated that 300,
or 84% percent of federal facilities are likely to require remediation, but that 230 of them are
expected to be remediated under CERCLA or other authorities. Note that the nine largest
federal facilities were not included in the analysis of the RIA sample, and are therefore not
included in this cost estimate. The fourth section of this chapter summarizes the available
information for the nine largest federal facilities.
10 For further discussion of the facility sample and weights, see Chapter 3 and Appendix A:
Development of Facility Sample.
11 The confidence interval should be interpreted as follows: for 95 percent of all samples,
the interval will include the actual national cost.
12 Results presented in this chapter were calculated and then rounded (generally to two
significant figures).
* * * DRAFT-March 23,1993 *
-------
EXHIBIT 5-2
TOTAL PRESENT VALUE COST OF CORRECTIVE ACTION [N = 3,500]
(in billions of 1992 dollars using 7% discount rate)
Investigation
$2.0
.
Capital costs include expenditures (in equipment, installation of equipment, and initial remedial activities.
3.500 facilities = 2.600 likely to require corrective action and 900 likely to incur RFI costs only.
-------
5-10
Results for closed and operating facilities
The majority of the total corrective action cost is likely to be incurred by operating, as
opposed to closed, facilities. As shown in Exhibit 5-3, operating facilities are expected to account
for 91 percent of the total cost of remediation. The definition of a "closed" facility used here is
not analogous to the RCRA definition of a "closed" facility. For this analysis, a closed facility is
one at which all industrial or commercial operations have ceased, not that the RCRA-regulated
units at the facility are closed. Likewise, "operating" facilities need not have active hazardous
waste management units to be included in that category.
Results by facility priority
As noted earlier, EPA prioritized facilities into high, medium, and low categories using
NCAPs. The results indicate that medium priority facilities are likely to incur roughly 70 percent
of the total corrective action cost, although they represent only 46 percent of facilities likely to
incur corrective action costs. The breakdown of total corrective action costs by facility priority is
presented in Exhibit 5-4. The corresponding breakdown of facilities expected to incur corrective
action costs (i.e., 3,500) is 22 percent high priority, 46 percent medium priority, and 32 percent
low priority.
Results by media
When the total corrective action cost is divided by the media addressed, ground-water
remediation is expected to account for 48 percent of the total cost of corrective action. Ground
water will generally be remediated using pump and treat systems. These systems include
extraction wells or interception trenches and a treatment system for the extracted ground water
(e.g., an air stripper). Additionally, pump and treat systems must often be operated for a long
period of time, or even indefinitely. In comparison, remediation of other media is generally
completed in one or two years at a lower cost.
The next largest portion of remedial costs involves soil contamination. Excavation,
treatment, sampling, and disposal of soil is projected to account for 22 percent of total corrective
action costs. Only 0.2 percent of the total costs is expected to be allocated to air remediation.
The projected breakdown of the total corrective action cost by media is presented in Exhibit 5-5.
Note that "source control" incorporates expenditures relating to management of waste, debris,
and mixtures of waste and soil. The "soil," "ground water," "surface water," and "air" categories
include only those remedial activities related to the specified media. The "other" category
includes expenditures not related to any one medium.
Results by remedial activity
To further explain the projected costs of corrective action, EPA disaggregated the total
cost by general activity types. "Removal/treatment of contaminated media" is likely to be the
most costly remedial activity, representing 52 percent of the total cost of corrective action. This
remedial activity includes extracting and treating contaminated ground water as well as the cost
of excavating and treating contaminated soil. The projected breakdown of the total corrective
* DRAFT-March 23, 1993 * * *
-------
EXHIBIT 5-3
TOTAL PRESENT VALUE COST OF CORRECTIVE ACTION
BY FACILITY OPERATING STATUS [N = 3,500] a
(in billions of 1992 dollars using 7% discount rate)
Operating Facilities
$17
Closed Facilities
$1.7
Average cost per facility for closed facilities: $1.7 million
Average cost per facility for operating facilities: $ 6.9 million
3,500 facilities = 2,600 likely lo require corrective action and 900 likely to incur RF1 costs only.
-------
EXHIBIT 5-4
TOTAL PRESENT VALUE COST OF CORRECTIVE ACTION
BY NCAPS FACILITY PRIORITY [N = 3,500] a
(in billions of 1992 dollars using 7% discount rate)
High Priority
$4.8
Medium Priority
$13
Low Priority
$0.5
Average cost per high priority facility: $6.2 million
Average cost per medium priority facility: $8.5 million
Average cost per low priority facility: $0.5 million
3,500 facilities = 2,600 likely to require corrective action and 900 ' 'y to incur RFI costs only.
-------
EXHIBIT 5-5
TOTAL PRESENT VALUE COST OF CORRECTIVE ACTION BY MEDIA [N = 3,500]
(in billions of 1992 dollars using 7% discount rate)
Examples:
Pump and Treat
Monitoring
Disposal
Examples:
Soil Excavation
Treatment
Transportation
Disposal
Soil Sampling
Surface Water $0.2
Examples:
Ground Water
9
Other
$2.1
Source Control
$3.2
Capping
Waste Removal
Treatment
Transportation
Disposal
Sampling
Air $0.03
Examples:
Volatile Suppressants
Monitoring
Tarps
Raring
Di version/col lection
Monitoring
Erosion prevention
Investigation
Construction
Permits
Fences
a "Source Control" refers to activities involving waste, debris, or waste/soil mixtures. "Other" refers to those activities not applicable
to a particular media. 3,500 facilities = 2.600 likely to require corrective action and 900 likely to incur RFI costs only.
-------
5-14
action cost by remedial activity is presented in Exhibit 5-6. The "investigation" category includes
RFIs, CMSs, and other investigation of facility contamination. "Containment" includes liners,
slurry walls, and other remedial activities designed to prevent the migration of contamination.
"Monitoring" refers to sampling and testing of soil, air, water, and waste. "Institutional controls"
are structural barriers to prevent people from being exposed to contamination (e.g., fences,
guards, deed notifications).
DRAFT-March 23, 1993 * * *
-------
EXHIBIT 5-6
TOTAL PRESENT VALUE COST OF CORRECTIVE ACTION
BY ACTIVITY TYPE [N = 3,500]a
(in billions of 1992 dollars using 7% discount rate)
Removal/Treatment
of Media
$9.8
Investigation
$2.0
iContainmen^
$1.7 ^Monitoring
$1.6
Capping
$1 *5
Institutional Controls
$0.01
Other $0.1
Disposal
$0.7
Removal/Treatment
of Waste
$1.4
"Removal/Treatment of Media" includes excavation, transportation, and treatment of soil, ground water, surface water, and air. Excavation.
transportation, and treatment of waste, debris, and waste/soil mixtures is included in "Removal/Treatment of Waste."
3.500 facilities = 2.600 likely to require corrective action and 900 likely to incur RFI costs only.
-------
5-16
Results by industry
The highest costs from the corrective action program are likely to be realized by the
chemicals and allied products industry (SIC 28). The chemicals industry is projected to incur 23
percent of the cost, which is roughly proportional to the industry's share of hazardous waste
management facilities (i.e., 19 percent of hazardous waste management facilities are in the
chemicals and allied products industry)." The fabricated metals industry (SIC 34) follows
closely behind chemicals and allied products with a projected 21 percent share of the total cost of
corrective action, representing six percent of hazardous waste management facilities. Federal
facilities (SIC 97) are expected to incur the highest average cost per facility; the weighted
average per facility cost of remediation is projected to be $33 million. The major industries
(grouped by two-digit SICs) that are likely to realize costs are presented in Exhibit 5-7.
5.2.2 Distribution of Facility Costs
In general, a few facilities account for a large portion of the total cost of corrective
action. Roughly half of the total costs of corrective action are likely to be incurred by slightly
more than ten percent of the facilities. These facilities are large (e.g., have a large number of
SWMUs) with extensive ground-water contamination and will have to remediate a substantial
number of SWMUs. Of the approximately 3,000 SWMUs expected to be remediated at the
upper ten percent of facilities, 50 percent are landfills and surface impoundments. The facilities
with the lowest projected remediation costs are likely to be small and require only investigations
or minor remediations.
EPA determined the cost of corrective action for each facility in the RIA sample. The
estimated corrective action cost per facility for remediation only (i.e., not including investigation)
is expected to range from $0.02 million to $180 million (expressed as a present value.) The
weighted average remediation cost per facility is projected to be $6.5 million. The estimated
annual cost per facility is likely to range from $2,000 to $17 million. The weighted average
annual cost per facility is estimated to be $0.6 million.
Exhibit 5-8 presents the cumulative frequency of total present value costs for the
proposed corrective action rule. This exhibit demonstrates that a few facilities are projected to
incur the majority of corrective action costs. Less than 15 percent of facilities likely to trigger
corrective action are expected to incur total per facility costs (i.e., remediation and investigation)
of over $10 million. Total per facility corrective action costs of $1 million to $10 million are
likely to be incurred by 60 percent of facilities triggering. Exhibit 5-9 provides the estimated
corrective action costs for each facility in the RIA sample. Of the 79 facilities in the sample, 59
are expected to incur corrective action costs, and 52 are likely to require remediation. The
exhibit includes the sample weight and four-digit SIC code for each facility.
13 See Appendix I, Exhibit 1-2 for a description of the number of facilities in industries by
two-digit SIC code.
* * * DRAFT-March 23, 1993 * * *
-------
EXHL 5-7
PRESENT VALUE COST OF CORRECTIVE ACTION BY INDUSTRY [N = 3,500]
(in 1992 dollars using 7% discount rate)
$5
f$4
_o
§$3
U
~ $2 -
o
I
-------
EXHIBIT 5-8
CUMULATIVE FREQUENCY OF CORRECTIVE ACTION COSTS
I (X)%
10% of Facilities
8% of SWMUs
409? 60%
Cumulative Percent of Corrective Action Cost
100%
Facilities [N = 3.500]
SWMUs [N = 15,000]
-------
EXHIBIT 5-9
UNWEIGHTED AND WEIGHTED CORRECTIVE ACTION COSTS FOR SAMPLE FACILITIES
(costs in millions of 1992 dollars: discounted using 7% rate)
Observation
Number
1
2
3
4
5
6
7
8
9
10
11
12
J3
14
15
16
17
IS
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
Facility
Weight
3 346
3346
3.346
63 37
213.765
63 37
63 37
3.346
3.346
3.346
63 37
63 37
63.37
63 37
63 37
63 37
3.346
213.765
3 346
63 37
6337
63 37
213 765
63 37
3 346
7.333
54.669
3 346
63.37
3346
6337
3346
63 37
3346
3.346
213 765
213.765
6337
3346
63 37
SIC Code
3000
2911
2860
2834
3699 -
2821
2800
2812
2812
2491
3691
2869
4953
3316
4953
2491
3672
2869
2899
4953
2834
3482
2491
2844
4953
971J
9711
2911
5169
2911
3470
2879
2821
3820
2869
2816
3840
4214
3662
2911
Present Value Costs {Unweighted)
RFI
SO 24
SI 08
SI. 11
SO 17
SO
SO 11
SO 45
SO 20
SO 17
SO 36
SO 11
SO 52
SO
SO 47
SO
SO 06
SO 49
SO
SI 38
SI 58
SO 24
S207
SO 34
SO 21
SO 34
SO
$0
SO 90
SO
SI 58
SI 18
SO 49
SO 05
SO 90
S2.07
$045
SO
SO 49
SI. 72
S0.52
CMS
SO 09
SO 39
SO 04
SO 05
$0
SO 03
SO 12
$007
SO 21
$008
$003
$0 12
$0
SO
SO
SO
SO 03
SO
$0 16
$0 11
SO
SO 31
SO 05
SO 03
SO 09
SO
SO
SO 08
$0
SO 14
SO 17
SO 11
SO 01
$012
SO 16
SO 08
$0
$001
SO 16
$009
Remediation
$17
$9
$49
$3
SO
$044
$22
$2
$21
$20
SO 02
$22
SO
SO
SO
SO
SO 87
SO
$16
. $6
' $0
$45
S4
$2
S4
so
$0
S41
SO
$117
$11
$7
SO 04
$7
$23
$2
SO
$058
$16
$9
Total
Unweighted
PVCost
$17
$11
$50
S3
$0
SO 58
S23
S2
S21
S21
SO 16
S23
$0
$047
SO
$006
$1
SO
S17
S8
SO 24
$48
S4
$2
S4
SO
SO
$42
$0
SI 19
$12
$8
SO 10
$8
$25
$2
$0
SI
SIS
S10
Total
Weighted
PVCost
$57
$35
S166
S201
SO
$37
$1,439
S8
$70
S69
S10
SI ,450
SO
S30
SO
S3
$5
SO
$58
S504
SI5
$3.016
$957
$155
S15
SO
SO
S140
SO
S397
S756
S27
S6
S26
$85
$500
SO
$68
$59
$624
Does not require- investigation or tetnediation
-------
EXHIBIT 5-9
UNWEIGHTED AND WEIGHTED CORRECTIVE ACTION COSTS FOR SAMPLE FACILITIES
(costs in millions of 1992 dollars: discounted using 7% rate)
Observation
Number
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
Facility
Weight
6337
213765
54669
7.333
213.765
3 346
63.37
3346
213 765
3.346
63.37
3 346
213 765
7.333
54.669
3346
3 346
3 346
213.765
«J.37
3.346
213 765
3.346
3 346
213.765
54.669
213 765
6337
6337
213 765
213.765
213 765
3 346
6337!
213.765 i
63.37
54.669
63.37
54.669 i
SIC Code
8221
8730
9711
9711
3354
3760
8221
3480
4953
3728
2911
3339
3568
9711
9711
2911
2879
2491
3827
2491 -
4953
3489
2911
4953
3674
97U
4953
3316
3069
3840
2860
2911
3674
4953
2047
3669
97U
4230
9711
Present Value Costs (Unweighted)
RFI
SO 34
SO 34
S0.21
Sll 03
SO
SO 22
SO 03
SI 27
SO 10
SO 69
$056
SI 81
SO 14
SO
SO
SO 87
SI. 70
SO 39
SO
$0
SI 02
SO 10
SI 48
SI 38
SO
SO
SO. 16
$083
$0.28
S0.28
$0
$241
$048
SO
SO
$038
SO
$052
$0
CMS
$004
$0
$005
S3 49
$0
$003
$002
SO 19
SO
$006
SO 08
SO 33
$003
SO
SO
$0 13
$026
$007
SO
so
SO 15
SO 01
SO 10
SO 05
$0
SO
SO
$0.02
SO 04
SO 06
SO
SO 12
$0.09
SO
SO
SO
so
SO 17
SO
Remediation
S2
SO
Sll
SI81
SO
S36
$2
$20
SO
$2
S5
S12
S3
SO
$0
$47
$39
S14
,$0
. $0
S5
SO 72
Sll
S68
SO
SO
SO
$007
SO 03
S3
so
S6
SO 83
$0
$0
SO
SO
$9
SO
Total
Unweighted
PVCost
$2
SO 34
$12
S195
$0
$36
$2
$21
$0 10
S3
$6
$15
$3
SO
SO
$48
$41
$15
SO
so
$6
$083
$12
$69
SO
SO
$0 16
SO 92
$035
S3
SO
S9
SI
SO
SO
SO 38
SO
S10
SO
Total
Weighted
PV Cost
S125
$73
$632
SI, 433
SO
5120
$156
$72
S21
S9
$382
$49
$586
SO
SO
$162
$136
S49
$0
SO
$20
$177
$42
$232
SO
SO
S35
S58
$22
$644
so
$1,831
$5
SO
$0
S24
SO
$628
$0
Does not rcqaire investigation or remediation
-------
5-21
5.2.3 Distribution of SWMU-level Costs
The weighted average present value cost per SWMU for remediation is projected to be
$1.1 million. The estimated remediation cost per SWMU is likely to range from $370 to $58
million. Exhibit 5-10 shows the distribution of per-SWMU remediation costs (note that the
intervals on this histogram are logarithmic). As seen in this exhibit, the majority of SWMUs (52
percent) are expected to incur remediation costs of $100 thousand to $1 million per SWMU.
When remediation costs are split by SWMU type, landfills and surface impoundments are likely
to account for just over half of corrective action costs. Landfills are projected to have the
highest average remediation cost ($2.6 million per SWMU) and waste transfer stations are
projected to have the lowest average remediation cost ($130,000 per SWMU).
* * DRAFT-March 23,1993 * * *
-------
10,000
EXHIBIT 5-10
DISTRIBUTION OF PER SWMU PRESENT VALUE
REMEDIATION COSTS [N = 15,000]
(in 1992 dollars using 7% discount rate)
8,000
cc
6,000
00
MX
o
1
OJ
X)
£ 4,000
3
2,000
140
dm
$100 to $1,000
7,700
3,500
2,300
900
i
$ i ,000 (o $ i o.ooo $ 10,000 to $ i oo.mo
$100,000(0
$1 million
$1 million to
$10 million
Per SWMU Cost of Remediation
200
$10 million to
$100 million
-------
5-23
The range of per-SWMU costs by unit type is expressed in Exhibit 5-11, and the total
remediation cost attributable to each type of SWMU is displayed in Exhibit 5-12. At the SWMU
level, a few units are expected to dominate total corrective action costs. Approximately 8
percent of the SWMUs are projected to incur roughly half of the total corrective action costs.
Exhibit 5-8 showed the cumulative frequency of per-SWMU remediation costs. The projected
distribution of per-SWMU remediation costs is similar to that of the per facility costs, with
approximately 10 percent of the facilities likely to account for roughly half of the total corrective
action costs.
EXHIBIT 5-11
RANGE OF PER SWMU PRESENT VALUE REMEDIATION COST
BY UNIT TYPE [N = 15,000]
(in millions of 1992 dollars using 7% discount rate)
Unit Type Minimum Maximum Weighted Average
Landfill
Surface Impoundment
Unspecified Unit
Spill Area
Tank
Accumulation Area
Process Sewer
Waste Pile
Land Treatment Unit
Injection Well
Incinerator
Area of Concern
Waste Transfer Station
$0.017
$0.0009
$0.0007
$0.005
$0.0004
$0.0008
$0.048
$0.001
$0.017
$0.66
$0.017
$0.002
$0.13
$39
$58
$13
$20
$2.7 '
$3.5
$2.1
$3.2
$7.3
$0.66
$6.0
$0.63
$0.12
$2.6
$1.2
$1.2
$0.89
$0.26
$0.97
$1.2
$0.38
$0.73
$0.66
$2.0
$0.13
$0.13
5.3 Sensitivity Analysis
The following section discusses the sensitivity analyses that were conducted for the cost
analysis. In this draft, results were assessed only for their sensitivity to the discount rate.
Additional sensitivity analyses may be added when regulatory options are analyzed. This section
also summarized four other studies of corrective action remediation costs conducted in recent
years.
* * * DRAFT-March 23,1993 * * *
-------
EXHIBIT 5-12
TOTAL PRESENT VALUE COST OF CORRECTIVE ACTION
BY TYPE OF SWMU [N = 15,000]a
(in billions of 1992 dollars using 7% discount rate)
Impoundments
$4.3
Injection Wells
Incinerators
Areas of Concern
Transfer Stations
Total = $0.06
Land Treatment Units
$0.2
Waste Piles $0.3
Process Sewers $0.5
Accumulation Areas
$0.7
Dumpster
Sump/Drain
Silo
Septic Tank
Tanks
$0.7
Spill Areas
$1.0
a Some remediation costs are facility-wide and not attributable to individual SWMUs. The additional cost of
facility-wide remediation is $2.1 billion.
-------
5-25
5.3.1 Discount Rates
In order to test the sensitivity of the results of this analysis to the discount rate used,
EPA calculated the total present value cost of corrective action using a four percent and ten
percent discount rate. The total national cost of corrective action is estimated to be $29 billion
using the four percent rate and $14 billion using the ten percent rate. These figures can be
compared to the projected total national cost of $19 billion calculated using a seven percent rate.
The effects of varying the discount rate used in the analysis are summarized below in Exhibit 5-
13.
Because the majority of remediation costs occur after 1997, the discount rate used to
bring the remedial cost stream to a present value has a significant effect on the total costs. A
higher discount rate lowers the present value of costs occurring in the future much more than
does a lower discount rate. The projected timing of the remedial cost stream for corrective
action was presented in Exhibit 5-1. This exhibit shows that the biggest peak of corrective action
costs is expected to occur around the year 2000.
EXHIBIT 5-13
EFFECT OF DISCOUNT RATE ON RESULTS OF COST ANALYSIS
Total National Cost of
Corrective Action
Weighted Average
Cost Per Facility
Remediated
PV
Annual ized
PV
Annual ized
Four Percent
Discount Rate
$29 billion
$2.1 billion
$11.0 million
$0.8 million
Seven Percent
Discount Rate
519 billion
$1.8 billion
$7.2 million
50.7 mil lion
Ten Percent
Discount Rate
$14 billion
$1.6 billion
$5.3 million
$0.6 million
* *
DRAFT-March 23, 1993 * * *
-------
5-26
5.3.2 Comparison to Other Studies
At least four other analyses have examined the costs of remediating contamination at
hazardous waste management facilities:
The proposed corrective action rule RIA (1990)14;
A Chemical Manufacturers Association study (April 1988)1S;
The Oak Ridge National Laboratory Report (September 1991)16; and
A paper by Paul Portney of Resources for the Future.
In this section, EPA provides a comparison of these studies and their results to those of the final
RIA analysis.
For purposes of comparison, the following is a brief summary of the current RIA's cost
analysis methodology and results. The RIA analyzed a sample of 79 facilities subject to
corrective action. It used expert panels of remediation and policy specialists to simulate the
process of remedy selection at the sample facilities under the corrective action rule, and to
prepare cost estimates for the remedies selected. The analysis also included the cost of RFIs.
Cost estimates were discounted to reflect the timing of remediations, using a 7 percent discount
rate. Finally, sample facility cost estimates are extrapolated to the national level, using facility
weights that reflect the sample selection process. EPA projects that the present value of the
national costs of the corrective action rule will be about $19 billion (in 1992 dollars); the average
present value cost per facility will be $7.2 million; and the average present value cost per SWMU
will be $1.1 million.
Proposed Rule RIA
The RIA for the proposed rule estimated the costs of ground-water remediation only. It
analyzed a sample of 65 facilities subject to corrective action which were expected to require
RFIs. Rather than expert panels, it used EPA's Liner Location Model, which models releases,
fate and transport, and remediation using standardized algorithms. The model also estimates
14 Office of Solid Waste, United States Environmental Protection Agency. (Prepared for
EPA by ICF Incorporated.) Regulatory Impact Analysis for the Proposed Rulemakine on
Corrective Action for Solid Waste Management Units. Washington, D.C.: U.S. Environmental
Protection Agency, June 25,1990.
15 Chemical Manufacturers Association. Impact Analysis of RCRA Corrective Action and
CERCLA Remediation Programs. Washington, D.C.: Chemical Manufacturers Association,
1988.
16 U.S. Department of Energy. Costs of RCRA Corrective Action: Interim Report.
Washington, D.C.: United States Department of Energy, 1991. See also Milton Russell, et al.
Hazardous Waste Remediation: The Task Ahead. Knoxville, Tennessee: University of
Tennessee, 1991.
* * * DRAFT-March 23, 1993 * * *
-------
5-27
remediation costs, using unit-cost algorithms based on EPA experience, professional judgement,
and standard construction cost estimation techniques. EPA used the model to simulate four
post-HSWA regulatory alternatives and a pre-HSWA baseline. Per-facility cost estimates were
discounted using a 3 percent discount rate. EPA derived national estimates by multiplying per-
facility estimates by the number of facilities in the universe. .National estimates include RFI
costs, though per-facility estimates do not. The RIA estimated costs for federal and non-federal
facilities separately. The results for the two scenarios most similar to the proposed rule (i.e.,
options B and C) were as follows: the mean present value cost per facility1 ranged from $2.5 to
$23 million for non-federal facilities, depending on the regulatory option, and from $12 million
to $110 million for federal facilities. The Agency estimated total national costs for non-federal
facilities to be $7.4 to $42 billion; for federal facilities, $2.9 to $25 billion. The RIA for the
proposed rule did not estimate per-SWMU costs. These estimates are in 1987 dollars.
Oak Ridge National Laboratory Report
Oak Ridge National Laboratory (ORNL)'s analysis of corrective action costs did not use
sampling to describe the universe of facilities subject to corrective action. Instead, ORNL used
data on hazardous waste facilities from two databases to identify SWMU's potentially in need of
remediation, and assumed that these SWMUs represented the entire universe.1 ORNL used
engineering rules of thumb to assign remedial technologies to each SWMU, and estimated costs
using the Cost of Remedial Action (CORA) model. They did not include investigative costs.
Several scenarios with differing clean-up levels were analyzed: a base case, a less stringent option,
and a more stringent option. In their original analysis, ORNL did not evaluate costs over time
and did not discount their results. They calculated the average cdst per SWMU in the base case
to be $6.4 million, and the mean total cost of corrective action to be $240 billion (1990 dollars,
undiscounted).1' In a September 9,1991 letter to EPA, Bruce Tonn of ORNL reported the
results of an analysis of corrective action costs over time, discounted at various rates. The base-
case total present value cost at a 7 percent discount rate was $89 billion.20
Chemical Manufacturers Association Study
The Chemical Manufacturers Association (CMA)'s study, prepared by Engineering-
Science (ES), estimated the cost of RCRA Corrective Action to the chemical industry only.
17 All costs presented here are incremental costs over the baseline.
18 The National Survey of Hazardous Waste Treatment, Storage, Disposal and Recycling
Facilities (TSDR) and the National Survey of Hazardous Waste Generators (GENSUR).
19 According to ORNL, evidence in the databases they used suggests that they underreport
the universe of SWMUs by 20 to 40 percent; extrapolating to the larger universe yields a total
cost of $290 billion.
20 Bruce Tonn, Oak Ridge National Laboratory, letter to Gary Ballard, U.S. Environmental
Protection Agency, September 9, 1991.
* DRAFT-March 23,1993 * * *
-------
5-28
which ES believed represented 25 percent of all facilities in the corrective action program. To
estimate corrective action costs, ES surveyed 16 CMA plants that reported 147 SWMUs expected
to require RFIs. For comparison, ES also used data from a 1986 CMA survey of 236 plants with
3,411 SWMUs requiring RFIs. ES assigned a degree of remediation to each SWMU in the data
base and estimated per-SWMU costs by SWMU type and degree of remediation required. The
analysis included RFI costs. ES then extrapolated from the samples to the universe of chemical
plants to derive national corrective action cost estimates for the chemical industry. ES presents
total capital costs and present worth costs, but does not report a discount rate. They estimate
capital costs of $5.4 to $9.1 billion dollars (depending on the database used) and present worth
costs of $6.6 to $11 billion. Dividing by ES's estimated numbers of SWMUs requiring RFIs
yields an average capital cost per SWMU of $290 to $770 thousand, and an average present
worth cost per SWMU of $350 to $940 thousand (year of dollars not reported).
Paul Portney Paper
A final estimate of the costs of the corrective action program comes from Paul Portney at
Resources for the Future." Portney cites the uncertainty about the number of SWMUs, and
about the differences between the final and proposed corrective action rules, as impediments to
the analysis of costs. He draws on EPA estimates that there are about 5,700 facilities subject to
corrective action, of which 1,700 will require ground-water remediation and 2,000 soil
remediation. He presents his estimates of the total, undiscounted costs of corrective action as
follows (1990 dollars):
With such great uncertainty about both the number of specific sites to be
remediated and the likely per-site cost, it is obvious that any estimate of overall
corrective action costs must be viewed cautiously. Nevertheless, on the
assumption that 3,000 facilities will require significant corrective actions at some
of their SWMUs, and on the equally heroic assumption that cleanups at these
facilities will cost $30 million each, the RCRA corrective action program may
result in cumulative costs on the order of $90 billion.22
Comparison of Study Results
Exhibit 5-14 summarizes the results presented in this section. The comparison shows the
results of the current RIA to be generally comparable to previous studies, given differences in
assumptions, methodologies and scenarios evaluated. The RIA estimate of total costs is in the
range presented in the RIA for the proposed rule. However, the proposed rule used a lower
discount rate and addressed only ground-water costs. For comparison, the total costs in the
Portney, Paul R., The Economics of Hazardous Waste Regulation," in U.S. Waste
Management Policies: Impact on Economic Growth and Investment Strategies. Washington,
D.C.: American Council for Capital Formation, 1992.
22 Portney, 1992, p.9.
* * * DRAFT-March 23, 1993 » * *
-------
5-29
current RIA at a 4 percent discount rate are projected to be about $29 billion, and the per-
facility costs, $11 million.
ORNL's total costs are higher than the RIA's, though discount rates are the same.
CMA's estimates are lower than the RIA's, but include costs only to the chemical industry. The
current RIA estimates costs to the chemical industry at $4.4 billion, but CMA's definition of the
industry may include different SIC codes. Finally, the RIA's total cost estimate is lower than
Portney's; however, his estimate is undiscounted.
EXHIBIT 5-14
COMPARISON OF CORRECTIVE ACTION RIA RESULTS
TO RESULTS OF OTHER STUDIES
Study
Corrective Action RIA
Corrective Action RIA:
Proposed Rule*
Oak Ridge National
Laboratory
Chemical Manufacturers
Association '
Paul Portney
Discounting
7%
3%
7%
- rate not
reported
none
Total Cost
(Billions)
S19
$10 - $67
$89
$6.6 -$11
$90
Cost Per Facility
(Millions)
$7.2
$2.5 - $110
not estimated
not estimated
not estimated
Cost Per SWMU
(Millions)
$1.1
not estimated
$6.4"
$0.35 - $0.94
not estimated
' Applies to groundwater remediation only.
bFrom undiscounted analysis.
c Applies to the chemical industry only.
5.4 Largest Federal Facility Cost Analysis
The corrective action universe includes nine 'Very large" federal facilities as well as 350
federal facilities classified as "large" and "other." EPA believes that these facilities are most likely
to be addressed under CERCLA, in which case they should be included in the baseline and not
in the analysis of the effects of corrective action. The RIA sample and expert panel cost
estimates did not include these nine very large federal facilities. This section presents a separate
analysis of the facilities, including an assessment of their baseline status and a presentation of
preliminary estimates of remediation costs. In addition to their probable baseline status, the
facilities were analyzed separately for several reasons:
* * *
DRAFT-March 23, 1993
* * *
-------
5-30
The largest federal facilities have large numbers of SWMUs, require extensive
cleanup, and manage complex waste types (e.g., Department of Energy facilities
often manage both radioactive and hazardous waste). Additionally, cleanup is
often conducted under a number of regulatory authorities, obscuring the baseline
status of these facilities.23 Due to these complexities, accurate representation of
these facilities in the RIA sample would have been extremely difficult.
Because evaluation of these facilities by the expert panels would have required a
considerable effort, the technical and policy panels did not examine these
facilities. Instead, EPA obtained budget projections for remediation from
Department of Defense and Department of Energy sources, to give an indication
of the magnitude of potential costs.
EPA and other federal agencies believe that these facilities are best characterized
using data available from the federal agencies themselves, where numerous agency
experts are already analyzing each facility.
5.4.1 Overview
Listed below are the nine very large federal facilities in the corrective action universe.
Seven are Department of Energy (DOE) facilities and two are under the jurisdiction of the
Department of Defense (DoD).
DoD Facilities
McClellan Air Force Base
Rocky Mountain Arsenal
DOE Facilities
Los Alamos National Laboratory
Idaho National Engineering Laboratory (INEL)
Nevada Test Site
Feed Materials Production Center (FMPC)
Hanford Site
Rocky Flats Plant
Savannah River Site
23 In addition, unit costs of remediation may vary significantly between facilities. For
example, an analysis conducted by the Office of Technology Assessment (OTA) determined that
the costs for similar activities can vary significantly among DOE facilities, due to site- and
regional-specific reasons. For example, the cost per cubic yard of excavating soil and sludge
varied from $8 to $260, and the cost per foot of installing a ground-water monitoring well varied
from $150 to $417. (Congress of the United States, Office of Technology Assessment. Complex
Cleanup: The Environmental Legacy of Nuclear Weapons Production. 1991.)
* * DRAFT-March 23,1993 * * *
-------
5-31
5.4.2 Results of Analysis
Budget projections for the largest federal facilities were obtained from DoD and DOE
publications. Because these sites are so complex, the extent of the cleanup required has not yet
been assessed. The available cost estimates are therefore likely to underestimate the true costs
of remediation.
The budget projections obtained for DoD facilities were presented in the Department
documents as lump sums rather than as streams of costs over time. Therefore, the effect of
discounting on the estimates is unknown. In addition, estimates are not in 1992 dollars.
Furthermore, estimates for both McClellan Air Force Base and Rocky Mountain Arsenal
included past expenditures along with future budgetary estimates.24
Budget projections for the DOE facilities were obtained from DOE's environmental
restoration funding estimates from 1991 through 1997.25 ("Environmental restoration" is the
term used by DOE to refer to corrective action activities.)
Exhibit 5-15 provides an overview of the baseline status and the corrective action costs
for the nine largest federal facilities. The remainder of this section discusses the baseline status
and remediation costs for these facilities.
Baseline Status
While all of the nine facilities are subject to RCRA, CERCLA authority also applies to
any federal site. In the absence of the RCRA Subpart S corrective action program, these sites
would likely be remediated under CERCLA provisions. At a number of these sites, CERCLA
has already been invoked,-placing seven sites on the National Priorities List (NPL). For sites on
the NPL, an Interagency Agreement (LAG) between EPA, DOE, and the state is commonly
established. The IAG is intended to evaluate which regulatory authority takes precedence at the
site and provide a single program for corrective action, or to clearly delineate which program will
cover which aspects of the cleanup.
RCRA is the primary regulatory authority at Los Alamos National Laboratories. At
Savannah River, RCRA has been the primary regulatory authority. To date, three consent
orders or decrees have been issued for the Savannah site, requiring DOE to comply with RCRA.
However, the site is also on the NPL, and an evaluation is currently underway to determine
whether CERCLA authority should apply.
24 U.S. Department of Defense. Defense Environmental Restoration Program: Fiscal Year
1991 Report to Congress. Washington, D.C.: Department of Defense, 1992.
23 U.S. Department of Energy. Environmental Restoration and Waste Management. Five
Year Plan. Washington, D.C.: Department of Energy, August 1991.
DRAFT-March 23, 1993 * « *
-------
5-32
Cost Estimates
Exhibit 5-15 also presents the available cost estimates for all nine facilities. These costs
are not discounted. In addition, they are incomplete and therefore likely to be understated. The
preliminary costs total $9.0 billion ($5.3 billion for the DOE facilities and $3.7 billion for the
DoD facilities). These costs range from a low of $211 million for the Nevada Test Site to a high
of $2 billion for the Rocky Mountain Arsenal. The average cost among these nine facilities is
$995 million. The breakdown of total costs by investigation, capital, and operation and
maintenance is not available. However, most of these facilities are in the initial stages of the
cleanup process and a large portion of the costs is likely to be investigative.
5.5 Limitations
5.5.1 Limitations of the RIA Cost Analysis
The cost analysis conducted for the RIA has several major limitations:
May Understate Costs
As discussed above, the analysis did not include some of the social costs of
corrective action. These include government administrative costs, at the EPA and
state levels, that are likely to be substantial. In addition, the costs for the nine
largest federal facilities were not included in the cost estimates. Section 5.4 gives
some indication of the potential magnitude of these costs.
The analysis does not include the opportunity costs for some remedial activities,
that is, the benefits foregone from alternative resource uses. These could include
the cost of using capacity at existing on-site disposal units, the cost of replacement
of units such as non-hazardous waste surface impoundments, and the opportunity
costs to off-site property owners when the use of their land is constrained by
remediation activities and equipment.
RCRA requires facilities subject to corrective action to demonstrate financial
assurance that they will have adequate funds available to complete a cleanup.
Facilities can demonstrate financial assurance in various ways, including insurance,
trust funds, letters of credit, and financial tests. This analysis did not consider the
costs of providing financial assurance, and so may understate the costs of
corrective action. However, EPA expects that most firms will choose a method
involving minimal costs, such as a financial test or letter of credit, and that total
costs will be small.
May Overstate Costs
Using the prices of remedial activities instead of their costs may overstate the
social costs of corrective action. If the remediation industry is not perfectly
competitive, prices may be greater than the marginal cost of providing services.
DRAFT-March 23, 1993 * * *
-------
5-33
EXHIBITS-IS
VERY LARGE FEDERAL FACILITY COST ESTIMATES (IN MILLIONS OF DOLLARS)
Facility
DOE Facilities:
On NPL? Cost
FMPC (Fernald) Y
Hanford Y
INEL Y
Los Alamos N
Nevada Test Site N
Rocky Flats Y
Savannah River Y
POD Facilities:
McClellan AFB Y
Rocky Mtn Arsenal Y
$2,010
$1,180
$475
$550
$211
$412
$414
$1,653
$2,052
Notes
DOE facilities estimates are undisoounted costs in 1990 dollars, expected to be incurred from 1991-
1997. Source: "Environmental Restoration and Waste Management, Five Year Plan." US DOE,
August 1991.
IAG in place integrating RCRA and CERCLA requirements.
It is not certain which regulatory authority will lake the lead at the Nevada Test Site, pending rescoring
of site under revised EPA Hazard Ranking System. Cleanup currently is being conducted to meet both
RCRA and CERCLA requirements.
A Federal Facility Agreement will require that all waste units be evaluated to determine whether they
are regulated under CERCLA. RCRA has been the primary regulatory authority at the Savannah
River Site to date.
Includes $73 million spent as of 1991 plus $1480 million to be spent in the future. Source- "FY 1991
Report to Congress." Environmental Restoration Program, US DOD. February 1992.
Includes $415 million spent as of 1991 plus $1,637 million to be spent in ihe future. Source: "FY 1991
Report to Congress." Environmental Restoration Program, US DOD. February 1992
DRAFT - March 23,1993
-------
5-34
The costs of the corrective action program may be overstated somewhat because
they include at least two types of actions that might have occurred in the absence
of the HSWA corrective action mandate. First, facilities may have voluntarily
chosen to conduct remediations to reduce liabilities or as a good business practice.
Second, state and local laws and regulations may have compelled remediations,
even in the absence of the federal mandate.
The costs of remedial technologies, and in particular new and innovative
technologies, may decrease substantially with time. In addition, new technologies
not currently available may be introduced that allow more cost effective cleanups
in the future. The expert panel costs may therefore overstate remedial costs by
relying on currently available technologies.
The cost analysis is based on the remedies expected to be selected under the
proposed corrective action rule, which includes the proposed approach to
corrective action management units (CAMUs). However, the Agency has issued
the final CAMU rule with provisions that differ from the proposed CAMU
provisions and that is projected to result in greater cost savings than the
proposed CAMU. As a result, the total costs of the corrective action program
projected in this analysis are somewhat overstated. Estimated cost savings are
$0.68 billion.
Indeterminate Effect on Costs
As discussed in Chapter 3, predicting the extent of contamination has inherent
technical limitations; therefore, the remedies chosen by the panel may not truly
predict the corrective action required at the facility.
The selection of remedies was based on more limited data than would be available
from an actual RFI. In some cases, conducting a full RFI may have improved the
quality of remedial decision making.
The timing assumptions may not reflect the actual start dates of the RFIs.
Starting dates of remediation have a large effect on cost due to discounting; costs
that occur earlier have a greater present value than those that occur later.
5.5.2 Limitations of the Largest Federal Facility Cost Estimates.
The following limitations specifically apply to cost estimates for the nine largest federal
facilities.
The cost estimates for the nine largest federal facilities are based on agency
budget projections. Budget estimates are limited by the agency's budget planning
horizon, and costs falling beyond the planning period will not be estimated.
Budget projections will therefore understate true costs. For example, budget
projections for DOE facilities were only provided for activities conducted until
DRAFT - March 23,1993
-------
5-35
1997. However, DOE expects to conduct environmental restoration activities until
2019. It is not clear how far in the future DoD projected remediation costs.
OTA has also noted that an increase of 25 percent is not uncommon between
estimated and actual remediation costs.26 This uncertainty was confirmed by a
General Accounting Office (GAO) report stating that DOE's cost estimates for
Fiscal Years 1991 and 1992 increased about 25 percent between the Five Year
Plans of 1989 and 1990.27
DoD cost estimates included past remediation expenditures as well as projections
of future expenditures, overstating the future costs of corrective action.
26 Congress of the United States, Office of Technology Assessment, 1991, op. cit.
27 General Accounting Office. Nuclear Health and Safety: Long-Term Plans to Address
Problems of the Weapons Complex are Evolving. Washington, D.C.: GAO, 1990.
DRAFT - March 23, 1993
-------
6. OVERVIEW OF BENEFITS
By preventing or mitigating the damages to natural resources caused by releases from
hazardous waste facilities, corrective action provides a variety of benefits. Many of these benefits
are discussed in detail in the subsequent chapters of this Regulatory Impact Analysis (RIA),
which address benefits related to:
Human health risks;
Ecological threats:
Water use costs;
Ground water nonuse values;
Residential property values; and,
Hazardous waste facility values.
This chapter provides an overview of the potential benefits of corrective action (including
benefits not explicitly addressed in this RIA), discusses the conceptual framework for the benefits
analyses, and briefly describes the studies presented in the following chapters.
6.1 Potential Benefits
Determining the potential benefits of corrective action requirements involved a three-step
process:
Identifying the environmental resources that suffer physical, chemical or biological
degradation due to releases from hazardous waste facilities subject to corrective
action;
Identifying the types of benefits associated with preventing or alleviating
contamination of these resources through corrective action; and,
Estimating the value of the benefits.
The following sections discuss each of these steps.
*** DRAFT - March 25, 1993 ***
-------
6-2
6.1.1 Affected Resources
The first step in developing the benefits analyses involved identifying the environmental
resources most likely to be affected by corrective action, and assessing the characteristics and
magnitude of the effects. This step identified the effect of contamination on these resources with
and without corrective action, and indicated the resources on which the benefits analyses should
focus.
As discussed elsewhere in this RIA, releases from hazardous waste facilities can affect a
wide variety of natural resources, such as ground water, soils, air, and surface water.1 The
analysis suggests that releases most often affect ground water and soils, and that the areal extent
of contamination is often greatest for ground water. Therefore, while the benefits analyses
prepared for the RIA address the effects of corrective action on a variety of resources, the
studies often focus on ground water.
6.1.2 Types of Benefits
The second step in the analysis involved identifying the types of benefits most likely to
result from corrective action. The environmental resources affected by releases from hazardous
waste facilities have a variety of attributes which society values. For example, society values
clean ground water because it can be used for drinking water, and may also value simply knowing
that clean ground water exists.
Economists often categorize values as either "use" or "nonuse" values:
Use values stem from the current or future use of resources. These values may
relate to the effect of resource use on human health, or to the use of resources
for recreation, agriculture or other purposes. The use value of ground water, for
example, results from the benefits it provides when used for household,
agricultural, industrial or commercial purposes.
Nonuse (or passive use) values can be more difficult to define. They relate to the
enjoyment of nonconsumptive services; e.g., the pleasure of simply knowing that
something such as clean ground water exists.
Raymond Kopp describes nonuse values as follows in a recent article:
"One can enjoy a particular nonconsumptive service -- for instance, reflecting on
the beauty of the Grand Canyon - without using it, or more generally, without
engaging in any observable behavior. Non-consumptive services, giving rise to
nonuse values, are available to all without the possibility of exclusion. Moreover,
1 Chapters 3 and 4 discuss the extent to which contamination would affect various media
with and without corrective action. Chapter 8 discusses ecological threats.
*** DRAFT - March 25, 1993 ***
-------
6-3
one person's enjoyment does not interfere with the enjoyment of others: and, for
the most part, one may enjoy these services without any monetary expenditures."2
The environmental resources affected by releases from hazardous waste facilities have a
variety of use and nonuse values, as illustrated by the examples in Exhibit 6-1 on the next page.
The list of possible use-related effects is very long; the exhibit provides some examples related to
human health and commercial use. For nonuse values, some economists utilize several
subcategories; in addition to the value placed on simply knowing that a resource exists (existence
value), nonuse values may relate to bequeathing clean resources to the next generation (bequest
value) and expressing concern for the welfare of others (altruistic value). As illustrated by the
exhibit, a particular resource may have several different types of values.
These benefits categories vary in importance. Releases from hazardous waste facilities
will have more significant effects on some categories than on others, and the likely significance of
the effects can be used to set priorities for analysis. For example, because releases often affect
ground water, detailed investigation of ground water-related benefits is important. Because
releases to air are less frequent, related benefits may be relatively small and worthy of less
detailed assessment.
The studies conducted for this RIA focus on the areas where corrective action is expected
to have the most significant effects. Six studies were conducted, five of which consider benefits
related to ground water. The analyses of human health risks, water use costs, residential
property values and facility values all captured benefits related to ground water use (e.g.,
withdrawals), while a separate study addressed ground-water nonuse value. The human health
risk, residential property value and facility value studies also considered the effects of corrective
action on resources other than ground water, while the study of ecological threats focused
primarily on surface water, including wetlands and biota.
6.1.3 Measurement of Benefits
The final phase in the analysis was to estimate the effect of corrective action
requirements on the benefit categories discussed above. This section describes the economic
concepts upon which this measurement was based. These concepts often are applied to develop
monetized estimates of benefits, however, in some cases (particularly when assessing health risks)
the analysis may stop short of determining the dollar value of the effects.
2 Kopp, Raymond J., "Why Existence Value Should be Used in Cost-Benefit Analysis,"
Journal of Policy Analysis and Management. Volume 11, Number 1 (1992), page 123.
*** DRAFT - March 25, 1993 ***
-------
6-4
EXHIBIT 6-1
POTENTIAL BENEFITS OF CORRECTIVE ACTION
Resource Affected
Ground Water
Surface Water*
Upland Soils"
Wetlands
Air
Biota
Benefit Category
Use
Nonuse
Use
Nonuse
Use
Nonuse
Use
Nonuse
Use
Nonuse
Use
Nonuse
Examples of Benefit Category
Withdrawals (household, commercial,
industrial, agricultural)
Existence, bequest, altruism
Withdrawals (household, commercial,
industrial, agricultural)
In-stream (recreational)
Existence, bequest, altruism
Land use (agricultural, recreational,
residential)
Existence, bequest, altruism
' Biological productivity
Recreation
Existence, bequest, altruism
Inhalation (human health)
Existence, bequest, altruism
Recreation
Commercial (fish, shellfish)
Existence, bequest, altruism
Surface water contamination may also affect sediments.
b Upland soils are surface soils; i.e., not wetlands.
*** DRAFT - March 25,1993 ***
-------
6-5
The concepts that underlie benefit analysis are derived primarily from the field of welfare
economics, which focuses on the optimal use of resources to maximize social welfare.3 Within
this framework, the basic building blocks are the preferences of individuals - policies are
evaluated in terms of their effects on individuals' utilities or well being. In other words, this
approach assumes that social welfare is maximized when the net welfare of all individuals is
maximized.
However, information on individual utility levels is rarely available for empirical research,
and economists generally turn to other methods to estimate the welfare effects of public policies.
These methods usually focus on measures of individuals' willingness to pay for a certain type of
change. Willingness to pay can be interpreted as the monetary equivalent of the change in
utility. When goods and services are traded in the market, willingness to pay information can be
derived directly from the observed behavior of individuals. However, for goods such as
environmental quality, where no market exists, willingness to pay information must be derived by
indirect methods or from survey research.
Indirect methods for estimating the value of changes in environmental quality rely on the
observed behavior of people, specifically the choices they make given market prices and
environmental quality. Data on the relationships between environmental quality and goods that
are traded in markets can be used to infer willingness to pay. These indirect methods generally
consider one of three types of relationships between market goods and environmental quality:
Market goods are substitutes for environmental quality;
Market goods are complements of environmental quality; or,
Market goods bundle together environmental quality with other attributes.
Indirect methods that focus on substitutes for environmental quality are generally
referred to as analyses of averting behavior. Such studies look at the willingness to pay to
mitigate or avoid the effects of pollution to infer the value placed on the pollution itself. For
example, expenditures on water treatment can be used to estimate the benefits of reducing water
pollution that poses health risks, and expenditures on cleaning can be used to estimate the
benefits of reducing air pollution that causes soiling.
3 For more detailed information on these concepts and their applications to environmental
policy see: Cropper, Maureen L., and Wallace E. Oates, "Environmental Economics: A Survey."
Journal of Economic Literature. Volume XXX (June 1992), pp. 675 - 740. Freeman, A. Myrick,
The Benefits of Environmental Improvement. Baltimore, MD: Johns Hopkins University Press,
1979. Just, Richard E., Darrell L. Hueth, and Andrew Schmitz. Applied Welfare Economics
and Public Policy. Prentice-Hall: Englewood Cliffs, N.J., 1982. Smith, V. Kerry. "Valuing
Ground Water Resources: A Conceptual Overview." Proceedings of the Clean Water and the
American Economy Conference fdrafO. Sponsored by the U.S. Environmental Protection
Agency and Resources for the Future. October 1992.
*** DRAFT - March 25, 1993 ***
-------
6-6
Alternatively, indirect methods may look at complements for environmental quality. For
example, if clean surface water is associated with increased recreational use, then willingness to
pay for recreational use can be used to indirectly estimate the value of clean water.
A third set of indirect approaches, generally referred to as hedonic methods, look at
goods that include environmental quality as one of several attributes. For example, willingness to
pay for a particular home in part reflects the characteristics of the house itself (e.g., number of
rooms) and of the community (e.g., crime rates). Purchase prices also reflect the quality of the
surrounding environment: all other things being equal, people are likely to be willing to pay
more for a house in an area with relatively high environmental quality than with extensive
pollution. Economists use statistical analysis to separate the effects of these environmental
characteristics from the other factors affecting price.
In contrast to these indirect methods, which focus on observed behavior in the markets
for other goods, direct methods generally involve asking people to report their willingness to pay
for clean resources. These direct methods include contingent valuation techniques, which use
personal or mail surveys to collect relevant data. While contingent valuation methods can be
applied to determine use values, they are most important in the area of determining nonuse
values. Effective indirect methods for measuring nonuse values have not yet been developed
because these values generally are not reflected in observable behavior.
Five of the six corrective action studies use indirect methods to measure the benefits of
corrective action; with the exception of the analysis of nonuse values of ground water, all of the
studies consider the value of other "goods" -- human health, ecological resources, water use and
property value - to determine the potential benefits of corrective action. Two of the studies, the
analyses of ecological threats and of human health risks, do not monetize these values.
6.2 Studies Conducted for this RIA
Six analyses of Benefits were conducted for this RIA. The following sections first provide
an overview of the studies, then describe the inter-relationships among the studies as well as their
relationships to the range of potential benefits identified above in Exhibit 6-1.
6.2.1 Overview of RIA Studies
The benefits analyses conducted for this RIA focused on those areas where the effects of
corrective action are expected to be most significant. Because the regulation will often affect
ground water resources, several of the studies addressed related benefits. Exhibit 6-2 on the next
page summarizes the characteristics of the studies, which are described briefly below.4
4 Detailed information on each study is provided in the subsequent chapters, and the results
are summarized in Chapter 13.
*** DRAFT - March 25, 1993 ***
-------
6-7
EXHIBIT 6-2
SUMMARY OF BENEFITS ANALYSES
Study
Human
Health Risks
Ecological
Threats
Water Use
Costs
Ground-
Water
Nonuse Value
Residential
Property
Values
Hazardous
Waste Facility
Values
Resources
Assessed
Air, soils,
ground water,
surface water
(includes food-
chain effects)
Primarily
surface water
Ground water
only
Ground water
only
All
Soils and
ground water
Benefit
Category
Use value
Not valued
Use value
Nonuse value
Use value
Use value
Type of
Method
Indirect,
substitute
Indirect,
substitute
Indirect,
substitute
Direct,
contingent
valuation
Indirect,
hedonic
Indirect,
substitute
Measure
Individual and
population risks
(number of cases
or exceedences;
not monetized)
Proximity,
comparison to
benchmarks
(not monetized)
Present value of
annual payments
Present value of
monthly payments
for 10 years
Single payment
Present value of
single payment
Time
Period
1992-2119
1992-2119
1992-2119
1992-2119
1983 - 1991
1992-2119
Facilities
Assessed
Full sample
(79 facilities)
Full sample
(79 facilities)
Full sample
(79 facilities)
Full Sample
(79 facilities)
Three
facilities
Full sample
(79 facilities)
DRAFT - March 25,1993
-------
6-8
Human Health Risks
This analysis focused on the extent to which corrective action will mitigate risks to human
health, considering the effects of exposure to carcinogens and non-carcinogens through ground
water, surface water, air and soils, as well as food-chain effects. The study estimated both
population and individual risks from exposure to contaminants from many of these pathways. It
applied a multi-media fate and transport model (MMSoils) to estimate exposures.
Ecological Threats
This analysis considered the potential risks to sensitive environments that may be posed
by contamination in the absence of corrective action. It included two components: (1) an
analysis of the proximity of each sample facility to sensitive environments and; (2) a comparison
of the concentrations of contaminants to ecological benchmarks. The study considered baseline
risks; it did not assess the effect of corrective action on these risks.
Water Use Costs
This study assessed the costs of replacing or treating contaminated drinking water that
would be averted by corrective action, focusing on ground water used for public or private
household supplies. The study applied the economic concept of consumers' surplus to estimate
the benefits of averting increases in these costs.5 It used a case study approach to determine the
effects of contamination on ground-water use in the area surrounding each sample facility.
Ground Water Nonuse Values
This study used the results of a contingent valuation survey to determine peoples'
willingness to pay to remediate contaminated ground water based on nonuse motives. The
results of the survey were applied to each of the sample facilities, using facility-specific data on
the extent of contamination and the surrounding population to determine the effect of corrective
action on the nonuse value of ground water.
Residential Property Values
The property value study used market data on the sales of residences to estimate peoples'
willingness to pay to avoid the perceived costs and risks associated with living close to hazardous
waste facilities. This analysis differed from the other use value analyses in several ways, e.g., it
potentially captured a wider range of the effects of corrective action, focused on past
contamination events (rather than the future effects of contamination), and considered three
facilities rather than the full RIA sample. The study did not, however, separate out the effects
of contamination (which would be prevented by corrective action) from the other effects of the
presence of the hazardous waste facility (which would not be affected by the requirements).
5 The calculation of changes in consumers' surplus is defined in Chapter 10. It takes into
account the changes in consumption that result from price changes.
*** DRAFT - March 25, 1993 ***
-------
6-9
Hazardous "Waste Facility Values
This analysis assessed the changes in the value of hazardous waste facilities that would
accrue as a result of the remediation required by corrective action. For example, remediation of
soils and ground water may provide the property owner with additional usable land, and may
make it possible to use the land for higher valued (perhaps non-industrial) purposes. Data on
values of comparable land parcels in the vicinity of the sample facilities were collected from local
realtors to determine the change in value potentially attributable to corrective action.
6.2.2 Relationships Among Studies
While the studies conducted for this RIA were intended to cover the benefit categories
most likely to be significant, they do not address all of the potential benefits of corrective action.
There is also some overlap across the studies. These issues are discussed below.
Benefits Categories Addressed
The analyses conducted for this RIA cover many, but not all, of the benefits categories
that may be affected by corrective action requirements. Exhibit 6-3 on the following page
provides a simplified summary of the areas covered by the studies, indicating that several of the
analyses address benefits related to ground-water use, while few address the types of
environmental resources that may be damaged less often. Note that a blank cell does not
indicate that benefits could not exist in the affected area; however, in many cases the category
was not subject to detailed study because benefits were not expected to be significant. For
example, available data suggest that surface water use is rarely affected by releases from
hazardous waste facilities because contaminants in surface water are generally below levels of
concern for human health.
Also, the level of coverage of each benefit category varies. For example, the water use
study focuses on ground-water use for public or household supplies and does not address self-
supplied agricultural, commercial or industrial use; the study of ecological threats considers only
baseline effects; the property value study addresses only three facilities; and health risks are not
"Valued" (i.e., monetized). In total, however, the studies address the benefits of corrective action
which were expected to be most significant.
Overlap Across Studies
The six studies completed by EPA each provide different measures of the benefits of the
corrective action requirements. In isolation, each study understates the total benefits of
corrective action because it focuses only on a portion of the effects of the requirements. In the
aggregate, the studies provide a more complete picture of the major benefits of the rule;
however, there is some overlap across the studies, particularly between the analysis of health risks
and water use costs.6
6 In theory, the property value study also overlaps with the other estimates of use value.
However, because this study does not provide a measure of the effects of the corrective action
requirements (separate from the general effects of the hazardous waste facilities), it is not
discussed in detail in this section.
*** DRAFT - March 25, 1993 ***
-------
6-1Q
EXHIBIT 6-3
CATEGORIES OF BENEFITS ADDRESSED BY EACH STUDY
Ground Water
Surface Water
Upland Soils
Wetlands
Air
Biota
Use Value
Nonuse Value
Use Value
Nonuse Value
Use Value
Nonuse Value
Use Value
Nonuse Value
Use Value
Nonuse Value
Use Value
Nonuse Value
Human Health
Risks
.
Ecological
Threats *
Water Use
Costs
,
Ground Water
Nonuse Values
Residential
Property
Values
Facility Values
Note: Dots indicate the areas addressed by each study, while blanks indicate areas that were not assessed.
' Ecological effects may have both use and nonuse value; however, these values were not quantified for this RIA.
DRAFT - March 25, 1993
-------
6-11
The health risk and water use cost studies overlap in that averting actions will prevent
contact with contaminants and hence decrease health risks. For this RIA, EPA did not
undertake a facility-by-facility analysis to determine the extent to which health risks and/or
averting actions are likely at each facility, however, the overlap across the two analyses is likely to
be significant.
At the facilities affected by corrective action, several factors will influence whether people
will undertake averting actions in response to contamination. First, whether they undertake such
actions will depend on whether they are aware of the contamination. Such awareness may occur
if:
The contaminants are detected through monitoring required under the Safe
Drinking Water Act, in cases where public wells are affected.
Community awareness of, or concerns about, contamination from the hazardous
waste facility or from other sources leads to testing of public or private wells.
Contamination is noticeable without testing (e.g., taste or odor thresholds are
exceeded).
Second, even if residents are aware of the contamination, averting actions may not be
undertaken or may not be completely effective in removing health risks.7 In some cases, people
may not be concerned about the risks or may have little information on either risks or effective
avoidance strategies. Risks also may be incurred temporarily while averting actions are
implemented, or may not be completely eliminated if the averting action does not completely
prevent exposure to the contaminants of concern.
Studies of actual contamination incidents indicate that decisions not to undertake averting
action may be relatively frequent even when health risks are well-publicized. For example,
Charles Abdalla studied responses to contamination of public supplies in two Pennsylvania towns.
At one site, he found that 96 percent of the respondents were aware of the contamination, and
that 76 percent of these informed households undertook some sort of averting action.8 At the
second site, 43 percent of the respondents were aware of the contamination, and 44 percent of
the informed households undertook averting actions.'
7 For a discussion of simultaneous analysis of health effects and averted costs, see:
Harrington, Winston, and Paul Portney, "Valuing the Benefits of Health and Safety Regulation."
Journal of Urban Economics. Volume 22 (1987), pp. 101 - 112.
8 Abdalla, Charles W. "Measuring Economic Losses from Ground Water Contamination:
An Estimate of Household Avoidance Costs." Water Resources Bulletin. Volume 26, Number 3
(June 1990), pp. 451 - 463.
9 Abdalla, Charles W., Brian A. Roach, Donald J. Epp, "Valuing Environmental Quality
Changes Using Averting Expenditures: An Application to Groundwater Contamination." Land
Economics. Volume 68, Number 2 (May 1992), pp. 163 - 169.
*** DRAFT - March 25, 1993 ***
-------
6-12
Factors influencing whether averting actions were taken included the nature of information
received on health risks and avoidance activities, household characteristics such as whether
children were present, and trust in public institutions.10
Abdalla also found that the type of action varied. Responses included increased
purchases of bottled water, boiling water before use, hauling water from other sources, and home
water treatment." These averting actions vary in their effects on health risks. For example, use
of bottled water for drinking does not address dermal exposure to, or inhalation of, contaminants
while showering.
In sum, the six studies completed by EPA provide different measures of the benefits of
the corrective action requirements. In isolation, each study understates the total benefits of
corrective action because it focuses only on some of the effects of the requirements. In the
aggregate, the studies provide a more complete picture of the major benefits of corrective action;
however, some potential benefits (such as nonuse values for resources other than ground water)
were not assessed. There is also some overlap across the studies, particularly between the
analysis of health risks and averted water use costs.
10 A study of responses to contamination in suburban Boston found similar factors
influencing decisions to undertake averting behavior; i.e., the level of concern about hazardous
wastes, confidence in local officials, and information on contamination incidents. See Smith, V.
Kerry, and William H. Desvousges, "Averting Behavior: Does it Exist?" Economic Letters.
Volume 20 (1986), pp. 291 -296.
11 The responses to the contamination incidents studied by Abdalla differ somewhat from the
responses considered in the analysis conducted for this RIA, because Abdalla focused on actions
undertaken by private households while they waited for public water authorities to respond to the
contamination. The RIA analysis focuses instead on long-term solutions that are likely to
remove most health risks.
*** DRAFT - March 25, 1993 **
-------
7. HUMAN HEALTH BENEFITS
This chapter describes the human health benefits for the proposed Subpart S corrective
action rule. As discussed earlier in this document, the analysis presented here is intended to
illustrate EPA's proposed methodology and obtain public comment. The draft methodology and
results do not represent any conclusion by the Agency about the appropriate future direction of
the corrective action program.
In Section 7.1 the Agency provides a background discussion on remedy effectiveness
assumptions and on the risk assessment framework underlying the analysis. Section 7.2 presents
an overview of the approach used to assess both baseline risks and risk reduction due to
corrective action. Section 7.3 includes a discussion of the results, and, finally, Section 7.4
describes the limitations of the approach and effects of major uncertainties.1
7.1 Background
7.1.1 Assumptions about Effectiveness of Institutional Controls and Engineered
Remedies
In this RIA, EPA predicts the adverse human health effects that would result, in the
absence of regulation, from long-term exposure to hazardous constituents from solid waste
management units (SWMUs) at RCRA facilities. The Agency also predicts the reduction in
adverse effects, due to the implementation of corrective action remedies, under the Subpart S
proposed rule. In order to conduct this analysis, EPA had to make certain assumptions
regarding the effectiveness of corrective action remedies, as discussed below.
One of the goals of the Subpart S corrective action program is to eliminate human
exposure and risk of concern from releases at RCRA facilities. This would generally be
accomplished through a combination of institutional controls and engineered remedies
implemented as part of the remediation process.
Institutional controls would prevent human exposure to contaminated media and would
include such things as restricting access to facilities and providing bottled water to replace
contaminated water supplies. These controls, where required, would generally be implemented
as the first step in the remediation process. Engineered remedies, such as treatment of
contaminated media and use of caps to prevent leaching of constituents from wastes left in place,
would complement the institutional controls by addressing the source of the releases directly.
Under a scenario of completely effective institutional controls and engineered remedies,
virtually all human exposure and risk from releases would be eliminated. Therefore, under this
"100 percent effectiveness" scenario, the Subpart S proposed rule (as well as other potential
corrective action regulatory options) would result in complete risk reduction.
1 Additional information on the risk assessment approach is provided in Appendix E.
Appendix G provides parameter values for the risk descriptors used in the analysis.
Draft - March 23,1993 ***
-------
7-2
However, in practice, there is some uncertainty about the long-term effectiveness of both
institutional controls and engineered remedies. Institutional controls may not always be
completely effective in the long run due to:
o Failure to implement controls where releases have not been detected. There is
uncertainty in monitoring and detecting releases under certain environmental conditions
(e.g., detecting releases to ground water where complex hydrogeologic conditions are
present).
o Uncertainty in long-term maintenance and enforceability of institutional controls. Long-
term effectiveness requires continuing attention by the owners and operators of facilities
undergoing corrective action and continuing oversight by EPA or authorized states (e.g.,
to prevent holes from developing in fences restricting access to facilities). This analysis
does not reflect the current debate about what constitutes effective and enforceable
institutional controls.
o Potential difficulty in compelling owners and operators to implement institutional
controls. Due to the expense of implementing certain controls or uncertainty about the
origin of the problem, owners and operators may be reluctant to voluntarily implement
controls (e.g., providing alternative water supplies).
Similarly, engineered remedies are believed to be less than completely effective for a variety of
reasons, as discussed earlier in this RIA.2
As a result of these uncertainties regarding long-term effectiveness of institutional
controls and engineered remedies, the Agency chose to analyze a "less than 100 percent
effectiveness" scenario in this chapter. This scenario relies on estimates of actual effectiveness of
engineered remedies and assumes that institutional controls are not implemented. The Agency
examined this scenario in order to measure the long-term cost-effectiveness of different
engineered remedies in isolation and to assess the potential risks that could result in situations
where engineered remedies and institutional controls fail.
7.1.2 Risk Assessment Framework and Guidance
EPA conducted the human health benefits analysis based on both the risk analysis
framework proposed by the National Academy of Sciences (NAS) and recent Agency guidance
on risk characterization and exposure assessment. Elements of risk assessment as defined in
these sources and as applied in this RIA are described below.
2 See Chapter 4 for a discussion of the effectiveness of engineered remedies.
*** Draft - March 23, 1993 ***
-------
7-3
National Academy of Sciences Risk Framework
The NAS3 distinguishes between risk assessment and risk management:
Risk Assessment is the use of factual data to define the health effects due to the
exposure of individuals or populations to hazardous materials and situations.
Risk Management is the process of integrating the results of risk assessment with
both engineering data and social, economic, and political factors in order to select
the most appropriate regulatory or policy alternative.
The human health benefits analysis for this RIA was conducted using a facility-level risk
assessment process. This process employed facility-specific data on wastes, waste management
practices, releases to and fate and transport in the environment, and potential exposure for
human populations to quantify adverse human health effects at the corrective action facilities.
Non-scientific considerations that would more appropriately be a part of risk management (e.g.,
deciding what post-corrective action residual risk levels are acceptable) are not included in this
risk assessment.
There are four key elements in the NAS risk assessment process:
(1) Hazard identification is the determination of whether a particular chemical is or
is not causally linked to particular health effects.
(2) Dose-response assessment is the determination of the relationship between the
magnitude of exposure and the probability of occurrence of the health effects in
question.
(3) Exposure assessment is the determination of the extent of human exposure before
or after application of regulatory controls.
(4) Risk characterization is the description of the nature and often the magnitude of
human risk, including attendant uncertainty.
For this RIA, the Agency tailored a risk assessment approach that relies on the NAS
framework. EPA addressed the first two steps in the NAS risk assessment process by identifying
constituents of concern for modeling at sample facilities and incorporating in the risk assessment
Agency-verified dose-response data for the constituents from EPA data bases, as discussed below
in section 7.2.2. The exposure assessment step of the NAS framework is a multi-step process in
this RIA, involving waste characterization, release analysis, fate and transport analysis, and
3 National Academy of Sciences. 1983. Risk Assessment in the Federal Government.
National Academy Press, Washington, D.C.
*** Draft - March 23,1993 ***
-------
7-4
exposure analysis, as discussed in section 7.2.3. Risk characterization, the final step in the NAS
process, is discussed in section 7.2.4.
The relationship between the NAS risk framework and the corrective action RIA risk
assessment process is shown in Exhibit 7-1.
EPA Guidance on Risk Assessment
In addition to utilizing the NAS framework, the Agency conducted the corrective action
risk assessment in accordance with recently issued EPA guidance on risk characterization and
exposure assessment. Guidance on risk characterization was provided by EPA's Risk Assessment
Council in a memorandum dated February 26,1992 entitled Guidance on Risk Characterization
for Risk Managers and Risk Assessors (hereafter, risk characterization guidance). The
memorandum provided several key principles on risk characterization:
Risk assessment information must be clearly presented, separate from any non-
scientific risk management considerations.
Key scientific information on data and methods must be presented, along with
discussion of confidence and uncertainties.
Risk descriptors that are consistent across Agency programs must be used to
represent the range of different exposure conditions encountered (e.g., central
tendency and high-end individual risk, risk to highly-exposed or sensitive
subpopulations, and population risk).
EPA also recently issued revised guidelines for exposure assessment: "Guidelines for
Exposure Assessment; Notice," FR 22888, May 29 1992 (hereafter, exposure assessment
guidelines). These guidelines establish a broad framework for Agency exposure assessments and
provide guidance on presenting results of exposure assessments and characterizing uncertainty.
Additional detail on the Agency's interpretation of the guidelines in developing the risk
assessment for the corrective action RIA is provided in the sections below.
7.2 Summary of Approach
The Agency estimated human health risks at a sample of facilities4 using detailed facility
information, environmental fate and transport models, and standard human health risk
assessment procedures. The specific approach included the steps discussed below.
4 See Chapter 3 and Appendix A for a description of the facility sample.
*** Draft - March 23, 1993 **»
-------
EXHIBIT 7-1
RISK ASSESSMENT METHODOLOGY
Risk
Assessment
Risk
Management
MAS Risk
Framework
Hazard
Identification
Dose-Response
Assessment
Exposure
Assessment
Risk
Characterization
Key Steps
In Risk
Assessment
for Corrective
Action RIA
Hazard
Identification
and Dose-
Response
Assessment
Waste
Characterization
Release
Analysis
Fate
and
Transport
Analysis
Exposure
Analysis
t
Risk
Characterization
13201* I
-------
7-6
7.2.2 Hazard Identification and Dose-response Assessment
For the hazard identification the Agency compiled a list of common waste constituents
that are known to cause adverse human health effects. EPA has identified hundreds of RCRA
waste constituents for past studies, and has listed these constituents in 40 CFR Part 261
Appendix VII (56 FR 7568, Feb. 25, 1991) and 40 CFR Part 264 Appendix IX (52 FR 25946,
July 9,1987). The Agency has characterized the human health effects, including cancer and
noncancer effects, for about 220 of these constituents. These 220 constituents were selected as
the initial constituents of concern for the risk assessment. Of these, about 100 were identified in
facility investigations and other sources as contaminants of concern at the 52 facilities modeled
for this RIA; this group forms the final set of hazardous constituents considered in this risk
assessment.
Dose-response assessment, as applied in this risk assessment, refers to the examination of
the quantitative relationship between exposure and effects and selection of the appropriate
indicators (toxicity values) that define this relationship. Toxicity values used in this analysis are
cancer slope factors (SFs) and chronic reference doses (RfDs). In general, the SF is the upper
95th percentile confidence limit of the slope of the dose-response curve derived from the
linearized multi-stage model, or is based on a maximum likelihood estimate. A chronic RfD is
defined as an estimate (with uncertainty spanning perhaps an order of magnitude) of a daily
exposure level for the human population, including sensitive subpopulations, that is likely to be
without an appreciable risk of deleterious effects during a lifetime. The RfD is based on the
"no-observed-adverse-effect-lever (NOAEL) or the "lowest-observed-adverse-effect-lever1
(LOAEL) for the most sensitive species and endpoint for exposure to that constituent. For the
dose-response assessment of the approximately 100 constituents, EPA relied on its existing.
standardized, peer-reviewed SFs and RfDs listed in EPA's Integrated Risk Information System
(IRIS, November 1992 update) or the Health Effects Assessment and Summary Tables (HEAST,
March 1992).5 EPA also used IRIS and HEAST to identify the critical health effects associated
with carcinogens and noncarcinogens.
7.2.3 Exposure Assessment
Exposure assessment in this expanded EPA risk assessment approach consists of four
steps: waste characterization, release analysis, fate and transport analysis, and exposure analysis.
For this RIA, the Agency repeated these steps for each RIA sample facility using the best
facility-specific information available. For certain key parameters, the Agency differentiated
between central tendency and high-end scenario values, based on Agency risk assessment
guidance, in order to characterize uncertainty in risk estimates.6 The end result of these steps
5 Exhibit E-l in Appendix E lists all the constituents along with the RfDs and SFs that the
Agency used in this analysis.
6 Further discussion of differences between central tendency and high-end parameter values
appears in Appendices B and G.
*** Draft -- March 23,1993 ***
-------
7-7
was exposure concentrations at exposure points in each pathway. These steps are briefly
described below.7
Waste Characterization
In the waste characterization step, the Agency gathered detailed, facility-specific
information on waste types and waste management practices at a sample of individual RCRA
facilities. Specific types of information collected included:
Number, type, and location of solid waste management units
(SWMUs) at the facility;
Waste types (e.g., physical form) and constituent concentrations;
Relevant waste management practices;
History of releases; and
Data characterizing degree, composition, and areal extent of existing
contamination of environmental media.
One of the most important aspects of the waste characterization was defining quantities
of the waste constituents and their potential to be mobilized into environmental media. The key
waste characterization parameters that were varied between the central tendency and high-end
scenarios include the waste concentration, the waste thickness, and the unit area.
Release Analysis
After determining constituent concentrations in SWMUs, the Agency characterized
existing releases and modeled future releases of these constituents to the environment. The
Agency did this modeling using the MMSOILS model, a multimedia release, fate, transport, and
exposure model. MMSOILS is able to model releases to the environment from five types of
waste management units: landfills, surface impoundments, waste piles, tanks, and underground
injection wells. For each unit selected for modeling, EPA used MMSOILS to simulate
contaminant releases to each environmental medium potentially contaminated. Different
releases for the central tendency and high-end scenarios were estimated by varying several key
parameters. For example, the leachate concentration from landfills and the total release volume
from tanks were increased for the high-end case.
7 Additional detail on waste characterization, release analysis, and fate and transport analysis
is provided in Chapter 3. Additional information on the exposure analysis is provided in
Appendix E.
*** Draft -- March 23,1993 **
-------
7-8
Fate and Transport Analysis
After simulating releases of hazardous constituents to the environment, the Agency
modeled the fate and transport of these contaminants in the environment at each sample facility.
This step accounts for concentration changes at various points in the environment due to
dispersion, degradation, retardation, migration, and intermedia transfer. The course travelled by
a contaminant from the source to reach a location where human exposure could occur is referred
to as an exposure pathway. The Agency identified a variety of exposure pathways that could
possibly contribute to human health risk at the corrective action facilities and modeled fate and
transport of contaminants in five of them: ground water, surface water, air, soil, and the food
chain. Using MMSOILS the Agency estimated concentrations at various locations within each
exposure pathway based on contaminant releases at the facility. The location within an exposure
pathway for potential contact between humans and contaminants is referred to as an exposure
point. At an exposure point, a human receptor may come into contact with the contaminant in
several ways, i.e., by ingestion, inhalation, or dermal contact. These are referred to as the
exposure routes. Thus, exposure points occur wherever humans ingest, inhale, or make dermal
contact with contaminated media. Exhibit 7-2 illustrates the integrated exposure scenario that
would be applicable at each facility, and Exhibit 7-3 summarizes the main elements of an
exposure scenario.8 The Agency's selection of specific exposure points and exposure routes
within an exposure pathway is discussed in the Exposure Analysis section below.
The final outputs from the fate and transport analysis were exposure-point concentrations
in ground water, air, surface water, off-site soils, and the food chain. The Agency estimated
contaminant concentrations at any given exposure point based on-long-term release and fate and
transport modeling, i.e., modeling over a 128-year period from 1992 through 2119.'
The Agency estimated both central tendency and high-end values for exposure point
concentrations by varying several parameters between the two scenarios. For example, the
ground-water hydraulic conductivity, the exposure well depth, and the sediment delivery fraction
to off-site fields were varied between the two scenarios for each facility. As discussed in Chapter
3, the data and assumptions for the high end scenario, while often resulting in greater constituent
mass available for release to all pathways, may in some cases inadvertently act to reduce the mass
released to pathways other than ground water.10 Since the MMSOILS model employs a mass
balance algorithm, the increase in mass released to ground water in the high end may act to
reduce the mass available for release to other media and may reduce the modeled extent of
contamination in those media in the high end. Consequently, the high-end results for surface
8 Note that not all pathways are applicable at all facilities.
9 The application of long- and short-term modeling is discussed in detail in Chapter 3.
10 The increase in mass released to ground water results from the higher leachate
concentrations assumed in the high end scenario.
** Draft - March 23, 1993 ***
-------
EXHIBIT 7-2
EXPOSURE MEDIA, PATHWAYS, AND ROUTES
Rela
1 sw
On-Slte or
.1 ** 1
_,..._. , ( « |
MU 1 4 Surface Soil 1
IVadoaa Tcmft 1
At Point of
Exposure
=Tl Ak 1
-n 1
| Biota |
I
^j Surface Soil ^^___
1
*
*-*
Exposure
Route
Inhalatkxi 1 1
^ Ingesdon |
.^*\ InQBstfon 1
] Contact |
^j
^1 Inhalation 1
^-^*j Ingesaon j
** »i oontaci i
Ilmwaten 1
^*| Ingesdon |
H Inhalation 1 1
«J Receptor I
Fzooni
-------
7-10
EXHIBIT 7-3
DEFINITIONS OF EXPOSURE PATHWAYS, POINTS, AND ROUTES
Element of
Exposure Scenario
Exposure Pathway
Exposure Point
Exposure Route
Description
Course travelled by a contaminant from a source to an
exposure point (e.g., ground water).
Location within an exposure pathway where potential contact
between humans and contaminants could occur (e.g., private
well)
The way in which a human receptor may come into contact
with the contaminant (e.g., ingestion).
water, soils, foodchain, and air pathways that are presented in this chapter in some cases may not
represent a realistic high-end scenario for those pathways.
For ground water, surface water, and air, EPA used MMSOILS to predict a time-varying
concentration profile at any of the exposure points. The term "concentration profile" is used in
this chapter to mean the distribution of concentrations over time (from 1992 through 2119) at a
single exposure point. Concentrations at a given exposure point in ground water, surface water,
or air were predicted for each year during the 128-year modeling period, resulting in a time-
varying concentration profile. Concentrations in soil and the foodchain pathways, on the other
hand, were predicted as steady-state, i.e., a single estimate over time.
Exposure Analysis
In the exposure analysis, the Agency combined concentrations in environmental media (as
modeled in the fate and transport analysis) with assumptions regarding contact and uptake by
human receptors. This analysis proceeded in three main steps:
(1) EPA determined pathway-specific exposure points for which contaminant
concentrations would be predicted (determined prior to and used in the fate and
transport analysis);
(2) For each pathway, EPA identified the relevant exposure routes as well as
exposure (i.e., contact) assumptions necessary to calculate intake via those routes;
and
*** Draft - March 23, 1993
**
-------
7-11
(3) Based on the exposure assumptions and contaminant concentrations at exposure
points, EPA calculated intakes for exposed individuals.
As mentioned above, exposure points are locations where humans are likely to ingest,
inhale, or touch contaminated media. EPA identified separate exposure points for each pathway
included in the fate and transport analysis. In addition, EPA identified separate exposure points
for off-site versus on-site exposures. The Agency determined the locations of these exposure
points from available facility-specific data (e.g., facility documents and topographic maps).11
Off-site exposure points. To estimate off-site risks, EPA identified off-site exposure point
locations at each facility by exposure pathway:
Ground water. For this pathway, EPA assumed exposures to occur wherever
there was a public or private ground-water well within a 90° sector up to 2 miles
downgradient of the facility.12 The 90° sector for which well data were collected
centered on the midline of the ground-water flow direction. EPA defined the
exposure point for calculating individual risk as the well (private or public)
located closest to the facility within the 90° sector. Exposure point locations for
calculating population risk were defined as the center points of six distance ranges
within the sector for private wells, and, where applicable, as the exact locations of
public wells. In the first case, exposure concentrations at all wells within a
specific distance range (e.g., in the 0.5- to 1-mile range) were estimated as being
equal to the concentration at a well located at the center point of that distance
range, on the midline of the sector. Thus, all the people being served by wells
within that distance range were assumed to be exposed to concentrations
occurring at that central point. For public wells, all persons served were assumed
to be exposed at the well location.11
In actuality, exposure to contaminated ground water would occur at
several places in a household, e.g., at a tap or in a shower. These, however, are
not referred to as exposure points in this analysis because concentrations are not
predicted in the fate and transport analysis specifically for these points. Instead,
EPA refers to the ground-water wells and the center-points of the distance ranges
as the exposure points and assumes that concentrations predicted at these points
will be the same at the tap or the shower.
11 Appendix E discusses in detail the methods used to identify specific exposure points and
potentially exposed populations.
12 At some facilities, wells beyond 2 miles were also considered as exposure points; see
Chapter 3.
13 See Chapter 3 for a discussion of the ground-water modeling approach.
*** Draft - March 23, 1993 ***
-------
7-12
Air. EPA first defined the exposure point for individual risk as the location of
the residence closest to the facility, determined from topographic maps of the
area surrounding the facility. Next, data were collected for populations within a
10 kilometer radius of each facility (across 360°). The populations were
subdivided into sectors defined by eight directions and five distance ranges. To
estimate risks to the population residing within a sector, EPA assumed every
person within that sector to be exposed to the concentration at the center point of
that sector, irrespective of the actual direction of the residences from the facility.
Thus, these center points were the exposure points.
Surface water. The Agency determined exposure points for recreational use for
the surface water pathway based primarily on known uses, swimming and/or
fishing, on surface waters downstream of a facility. In most cases, one of these
uses occurred closer to the facility than the other. EPA defined the closer use
point as the exposure point for both swimming and fishing. In cases where
facility-specific information on the distance to these use points was not available,
EPA assumed swimming and fishing to occur at a default distance of 10 meters
downstream from the point where the surface water received contamination from
the facility.
For domestic surface water use, the exposure point for both individual and
population risk (for all affected populations) was defined as the point of intake
identified on the surface water.
Soil. EPA defined two separate exposure points .for calculating individual (adult
and child) and pica child14 risk via the soil pathway. One exposure point is at a
distance from the facility equal to the distance to the nearest off-site field
identified for likely human exposure to soil (e.g., parks, playgrounds, and
residential yards). The second exposure point is located at a distance from the
facility to the nearest off-site agricultural field where human exposure to
contaminants in soil could occur (e.g., agricultural land, or a vegetable garden at
the closest residence).
Food chain. EPA assumed exposures to contaminants bioaccumulated in
vegetables, milk, and beef essentially to occur at the agricultural field identified as
being closest to the facility for the soil pathway. Individual risk is calculated for a
person ingesting the contaminated vegetables, beef, and milk "originating" from
this field. Thus, the exposure point for these pathways, including the subsistence
farmer, is the agricultural field.
Risk due to ingestion of contaminated fish was estimated when a
recreational-use exposure point was identified on surface water contaminated by
14 Pica behavior is discussed in the Exposure Assumptions section, below.
*** Draft - March 23,1993 ***
-------
7-13
releases from a facility. EPA estimated individual risk to subsistence fishermen
based on ingestion of contaminated fish caught from that recreational-use
exposure point.
Hypothetical on-site exposure points. In addition to estimating off-site risk, the Agency
also quantified hypothetical risk under alternative assumptions about future use of on-site land at
the facilities. EPA calculated on-site future risk due to exposures via the ground-water, air, soil,
and foodchain pathways.15 To calculate this risk, the Agency defined exposure points based on
hypothetical future use at several SWMU-specific locations, unique to each pathway:
Ground water. The exposure point for ground water was assumed to be at a dis-
tance of 10 meters downgradient from each SWMU (i.e., at the assumed SWMU
point of compliance). This meant that the exposed individual was drinking from a
well drilled 10 meters from the SWMU.
Air. The exposure point for air was defined as 1 meter from the SWMU
boundary.
Soil. The exposure point for soil varied among facilities because EPA evaluated
this pathway for only those facilities and SWMUs where on-site soil
contamination data were available (Soil data were available for approximately
1,500 facilities). The exposure point in such cases was assumed to be located in
the vicinity of the SWMU represented by the soil contamination data. Exposures
via the foodchain pathway, and to subsistence farmers and pica children were also
assumed to occur at this point.
Exposure routes. For each pathway in the fate and transport analysis, EPA determined
the exposure routes through which a receptor could be exposed. EPA used the same exposure
routes for a particular pathway for both on- and off-site risk estimation as described below.
Ground water. EPA estimated intake of contaminants from ground water as the
sum from three exposure routes: ingestion of contaminated drinking water,
inhalation of volatile compounds during household use of ground water, and
dermal uptake due to direct contact while showering.
Air. The only exposure route that EPA considered for the air pathway was
inhalation of airborne contaminants.
15 The Agency did not evaluate risk via on-site surface water because there are very few
facilities with surface water bodies on site, and, in general, contamination data were not available
for them.
*** Draft - March 23,1993 ***
-------
7-14
Surface water. For recreational surface water use, the Agency calculated intake
as the sum of two exposure routes: dermal absorption via direct contact with
surface water, and incidental ingestion of surface water while swimming.
Soil. Incidental ingestion and direct contact (dermal absorption) exposure routes
were added when EPA calculated intakes due to the soil pathway.
Foodchain. The foodchain pathway accounts for intake due to ingestion of
contaminated vegetables, milk, beef, and fish.
Identifying the exposure routes enabled the Agency to calculate pathway-specific intakes using
exposure assumptions.
Exposure assumptions. In order to quantify on- and off-site human exposure to
contaminants, i.e., to calculate intake at exposure points, EPA relied on pathway- and route-
specific assumptions about:
The extent, duration, and frequency of contact between contaminated media and
human receptors;
The characteristics of a typical human receptor such as body weight;
The portion of the chemical in the environment that is absorbed through the skin
of the receptor16; and
The amount of a contaminant that is transferred from an environmental medium
to a foodchain component (e.g., from surface water to fish, from soil to
vegetable).
Values for these assumptions were used either in the equations to calculate intake at exposure
points or in the MMSOILS model to predict concentrations in the foodchain pathway. Most of
the assumptions used are standard exposure assumptions taken from EPA risk assessment
16 This refers to intake via dermal absorption. Systemic absorption after intake via ingestion
and inhalation was assumed to be 100 percent; see Appendix E for discussion of underlying
assumptions.
*** Draft - March 23, 1993 ***
-------
7-15
guidance documents such as the Risk Assessment Guidance for Superfund (RAGS)17, the
RAGS Supplemental Guidance18, and the Exposure Factors Handbook.19
For both the central tendency and high-end scenarios, EPA used average or 50lh
percentile values for the exposure assumptions (or contact parameters) for all pathways, except
the highly exposed subpopulations.20 For example, EPA estimated cancer risks and noncancer
effects based on a 9-year exposure duration; 9 years is the median residence time at a single
residence according to national statistics.21 Risks from ingestion of ground or surface water
were calculated using an average ingestion rate of 1.4 liters per day, risks from vegetable
ingestion were calculated based on an assumption that the general population would get only 25
percent of their vegetables consumed daily from the contaminated source.
To calculate intakes for highly exposed subpopulations, EPA relied on exposure
parameter values that represent either the 90* or 95th percentile. (In some cases, the reference
sources identified the relevant values only as "high-end," without specifying the percentile that
17 Risk Assessment Guidance for Superfund - Vol.1. Human Health Evaluation Manual (Part
A), Office of Emergency and Remedial Response, EPA/540/1-89/002, Dec. 1989.
18 US EPA 1991. Human Health Evaluation Manual. Supplemental Guidance: Standard
Default Exposure Factors. Office of Emergency and Remedial Response. OSWER Directive:
9285.6-03.
19 USEPA 1989. Exposure Factors Handbook. Office of Health and Environmental
Assessment. EPA/600/8-89/043.
20 Specific values that EPA used for exposure parameters in the exposure analysis are listed
in Exhibit E-3. Appendix G also highlights the differences in exposure parameter values used for
highly exposed subpopulations versus all other pathways.
21 The Agency estimated risk assuming that individuals (other than highly-exposed
populations) are exposed to facility-related contamination for 9 years, i.e., the period that they
are expected to stay near the facility. Both cancer and noncancer risks are estimated based on
this 9-year exposure duration; the averaging time for calculating these two types of risks,
however, differs. Cancer risk is calculated by averaging the intake over a lifetime (averaging time
= 70 years), i.e., the total cumulative dose incurred during the 9-year exposure period is prorated
over a lifetime. This approach for carcinogens is based on the assumption that a high dose
received over a short period of time is equivalent to a corresponding low dose spread over a
lifetime. For noncancer risk, on the other hand, the intake is averaged over the exposure
duration (averaging time = 9 years). All the 9-year exposure durations occur within the
modeling period, such that the 130-year modeling period can be viewed as essentially a
continuum of the 9-year exposure durations. The 70-year averaging time for cancer risk is
related to the lifetime of an exposed individual, and is not connected to the 130-year modeling
period.
*** Draft - March 23,1993 ***
-------
7-16
they represent on the parameter-specific distribution.) These upper range values were used in
order to characterize subsistence lifestyles and atypical situations. Subsistence farmers, for
example, are expected to live at their residence for longer periods of time, and also rely on
homegrown produce more than does the general population. Therefore, EPA assumed the
exposure duration for this subpopulation to be equal to the 90th percentile national residence
time (30 years), and further assumed that 40 percent of the vegetables consumed by this group
come from the contaminated source.
Another highly-exposed subpopulation for which EPA assessed risk is pica children, i.e.,
children exhibiting higher than normal soil ingestion.22 Normal mouthing and subsequent
ingestion of soil is common among children at certain ages (e.g., 1 to 6 year-olds), but some
children are prone to ingesting abnormally high amounts of soil or non-food items. Groups that
are at increased risk for pica behavior are children aged 1-3 years, children from families of low
socioeconomic status, and children with neurologic disorders. For this analysis, EPA used a soil
ingestion rate of 800 mg/day, which represents the upper end of the range of soil ingestion rates
for normal children.
The exposure analysis for this RIA differs in several ways from the methodology used to
calculate action levels provided in the corrective action proposed rule. The proposed rule notes
that action levels were to be used as triggers to identify early in the process the need for
initiating a Corrective Measure Study at a RCRA facility. Because the action levels were to be
used as points of departure, the assumptions used to derive the action levels were simple, and
generally conservative.
Differences in exposure pathways and routes. The proposed rule provides
calculations for action levels for only four pathways (i.e., ground water, surface
water, air, and soil), based on a limited number of exposure routes within these
pathways. Action levels are calculated based on the ingestion route for the
ground water, surface water, and contaminated soil pathways, and the inhalation
route for the atmospheric pathway. For this RIA, EPA examined more pathways
and more routes of exposure within the pathways presented in the proposed rule.
For example, in addition to exposure routes used for action level calculations in
the proposed rule, EPA assessed risks from inhalation of volatile contaminants
from water to indoor air, dermal absorption of contaminants released from water
while showering, dermal absorption and ingestion of contaminants during
recreational uses of surface water, and dermal absorption of contaminants from
soil. In addition, EPA addressed risks from the foodchain pathway (vegetables,
beef, and milk ingestion) and risks to highly-exposed and sensitive subpopulations.
22 Although there are a variety of definitions for pica, which generally refer to ingestion of
non-food items rather than soil specifically, EPA has defined pica as abnormally high soil
ingestion behavior in USEPA 1989, Exposure Factors Handbook. Office of Health and
Environmental Assessment (EPA/600/8-89/043).
*** Draft - March 23,1993 ***
-------
7-17
Differences in exposure assumptions and timing of exposures. The proposed rule
bases calculations of action levels on an assumption of lifetime (i.e., 70-year)
exposure to constant (steady-state) concentrations. Further, the action levels are
based on 95th percentile or high-end values for ingestion rates and exposure
frequencies. For this RIA, EPA calculated intakes based on 9 year exposure
durations and average or 50th percentile values for ingestion rates and exposure
frequencies. Also, EPA calculated intakes based on time-varying concentrations
for ground water, surface water, and air, and estimated risks up to the year 2119
(a total of 128 years).
7.2.4 Risk Characterization
Risk characterization refers to the process of combining the quantitative data from
hazard identification and dose-response assessment with intakes derived during exposure
assessment to obtain measures of adverse health effects. These measures include the lifetime
probability of developing cancer (referred to in this RIA as cancer risk) and magnitude of
exceedence of noncancer thresholds (referred to as a noncancer effect). Risk results were
estimated at the facility level and then extrapolated, based on facility weights, to national
estimates.
Based on requirements noted in both the risk characterization guidance and exposure
assessment guidelines, the Agency used several types of risk descriptors in combination to
conduct and to present the results of the risk assessment. Risk descriptors essentially express the
type and degree of human health effect expected due to exposure .to contamination. For this
RIA, the Agency first defined four types of fundamental risk descriptors. These are the risk
descriptors specified in the risk characterization guidance:
central tendency individual risk;
high-end individual risk;
individual risk to highly-exposed or sensitive subpopulations; and
population risk.
In order to fully describe the results of the benefits analysis, EPA incorporated various
"dimensions" into each of these fundamental risk descriptors, giving rise to several secondary risk
descriptors. These secondary risk descriptors are variations on the fundamental risk descriptor,
depending on the dimension incorporated. These dimensions reflect assumptions about the type
of health effect (cancer risk, noncancer effects, effects due to elevated blood lead levels), physical
location of exposures (on- versus off-site), temporal variation in exposures (9-year peak versus
exposure averaged over a 128-year period), changes in exposures due to water treatment
(concentrations capped at Maximum Contaminant Levels (MCLs) or taste/odor thresholds versus
no capping), and changes in exposure patterns due to remedy implementation (baseline versus
post-remediation). Thus, "off-site central tendency individual cancer risk" would be an example
of a secondary risk descriptor. Exhibit 7-4 summarizes the relationships between the
fundamental risk descriptors and the various dimensions, and the following sections discuss each
risk descriptor and dimension.
Draft - March 23, 1993 ***
-------
7-18
EXHIBIT 7-4
SUMMARY OF RISK DESCRIPTORS USED IN CORRECTIVE ACTION R1A
Risk Descriptor
Off-Site Risk Based on
Current Uses
On-Site Hypothetical Risk
Based on Alternative Future
Uses'
Central Tendency Individual Risk
Cancer
Noncancer
High-End Individual Risk
Cancer
Noncancer
Individual Risk to Highly-
Exposed or Sensitive
Subpopulations
Cancer
Noncancer
Lead4
s
5
Population Risk
Cancer
Noncancer
1 Assumes future residential and/or agricultural uses on site.
2 For the ground water, surface water, and air pathways, estimated peak 9-year and average 130-
year risk based on a time-varying concentration profile.
3 Estimated baseline risk only; did not estimate post-remediation risk.
4 Effects due to exposures to lead are calculated differently from cancer risks and noncancer
effects.
5 Risk calculated based on exposures via the food chain only.
6 Risk to sensitive subpopulations calculated for off-site exposures only.
7 Population risk estimated for the ground water, surface water, and air pathways only.
8 Individual and population risk via ground water estimated for two scenarios, exposure to
ground water constituent concentrations regardless of MCLs or taste and odor thresholds, and
exposure to constituent concentrations not exceeding (capped at) MCLs or taste/odor
thresholds.
** Draft -- March 23, 1993 ***
-------
7-19
Fundamental Risk Descriptors
The risk characterization guidance explains that risk descriptors should be developed
based in part on the exposure distribution within the population of interest. Partly to capture the
variability in exposures, the guidance recommends that EPA risk assessments provide risk
descriptors that include: individual risk to central tendency and high end portions of the risk
distribution; important subgroups of the population such as highly-exposed or highly-sensitive
groups or individuals; and, if known, population risk. The Agency used all these risk descriptors
for this benefits analysis, and they are referred to in this RIA as fundamental risk descriptors.
Central tendency individual risk. The risk estimates under the central tendency scenario
reflect the Agency's assessment of the most likely exposure concentrations (based on constituent
concentration, release, and fate and transport assumptions) resulting from contamination at
corrective action facilities, and the most likely exposure (contact) assumptions (e.g., ingestion
rates, exposure frequency, and exposure duration related to an individual's lifestyle and habits).
Exhibit 7-5 illustrates the Agency's exposure concentration and exposure (contact) assumptions in
developing this risk descriptor.
High-end individual risk. The risk characterization guidance and exposure assessment
guidelines specifically note that EPA risk assessments should also provide estimates of individual
risk that correspond to the high-end portions of the risk distribution. The guidance describes the
high-end risk descriptor as:
"... a plausible estimate of the individual risk for those persons at the upper end of the
distribution. The intent of this risk descriptor is to convey an estimate of risk in the upper
range of the distribution, but to avoid estimates which are beyond the true distribution."
EPA computed this high-end risk by estimating high-end exposure concentrations based
on more conservative assumptions about constituent concentrations in waste and about the rate
of contaminant releases to, and fate and transport in, environmental media.23 These
conservative assumptions lead to a higher exposure concentration compared to the central
tendency scenario, and reflect some of the uncertainty and within-site variability in the factors
used in this RIA to determine exposure point concentrations. To compute high-end individual
risk, EPA used most likely exposure (contact) assumptions (as used for the central tendency
scenario); see Exhibit 7-5. EPA intends for this risk descriptor to describe risks that are
expected to occur in a small but definable "high-end" segment of the subject population.
23
Appendix G provides specific assumptions and parameter values that the Agency varied to
ate the hieh-end exposure.
estimate the high-end exposure.
*** Draft « March 23,1993 ***
-------
7-20
EXHIBIT 7-5
SUMMARY OF EXPOSURE ASSUMPTIONS FOR RISK DESCRIPTORS
Risk Descriptor
Central Tendency
Individual Risk
High-End Individual Risk
Individual Risk to Highly-
Exposed Subpopulations:
Individual Risk to Sensitive
Subpopulations
(Lead Exposure):
Population Risk
Exposure Parameter
Exposure Concentration
Parameters'
Central
Tendency
Values
Assumed
High-end
Values
Assumed
Contact Parameters2
Central
Tendency
Values
Assumed
J
3
NA4
>
High-end
Values
Assumed
NA
1 Examples of exposure concentration parameters include constituent concentration, release
rate, and fate and transport parameters.
2 Examples of contact parameters include ingestion and inhalation rates, exposure frequency,
exposure duration, and percent of food from contaminated source.
3 In addition to contact parameters, toxicity factors (i.e., SFs and RfDs) are also used in
estimating risk. Although the contact parameters used for the central tendency scenario are
most-likely or central tendency estimates, the toxicity factors are more conservative estimates.
For example, the SFs can be the upper 95* percent confidence interval on the dose-response
curve, and RfDs are divided by uncertainty factors of up to 1,000 or more.
4 Not applicable. Effects on blood lead levels of children from lead exposures are estimated
using the Lead Uptake/Biokinetic model, and, therefore, are not based on standard exposure
parameters.
*** Draft -- March 23,1993
-------
7-21
Individual risk to highly-exposed or sensitive subpopulations. For this RIA, EPA
calculated risk to individuals representing subgroups of the population likely to receive
particularly large exposures. These subgroups include subsistence farmers, subsistence fishermen,
and pica children. EPA defined these subgroups as being highly exposed due to their lifestyle
and habits. As described earlier, to calculate risk to individuals in these subgroups, the Agency
used high-end exposure (contact) assumptions (e.g., higher ingestion rates, longer exposure
durations). Assumptions used to calculate the exposure concentrations, however, represented the
Agency's most-likely assessment of factors such as constituent concentrations in waste, release,
and fate and transport (i.e., these were the same as in central tendency individual risk; see
Exhibit 7-5).
The Agency also characterized risk to individuals of a sensitive subpopulation, i.e.,
children up to seven years of age exposed to lead. The Agency determined the incremental risk
to children due to lead contamination by determining the mean blood concentration of lead in
the exposed population and comparing it with a threshold level of 10 ug/dL.24 Lead
concentrations in ground water, surface water, air, soil, and the foodchain were determined based
on the Agency's most-likely assessment of factors related to constituent concentrations in waste,
and release to and fate and transport in environmental media. EPA used these exposure
concentrations as inputs to the Lead-Uptake/Biokinetic (UBK) model to estimate the mean
blood lead levels in exposed children.25 The exposure (contact) assumptions used were those
built into the UBK model.
Population risk. The population risk descriptor used in this RIA is a measure of the
extent of harm for the exposed population as a whole over the 128-year modeling period.
Population risk is derived based on the yearly central tendency individual risk over the modeling
period, i.e., based on the Agency's most-likely estimate of exposure concentrations and exposure
(contact) assumptions. Because the population risk characterization counts numbers of people
affected over the 128-year modeling period, the Agency estimated risks for populations currently
living near corrective action facilities and also for future populations. EPA estimated the size of
future populations based on the size of the current, potentially exposed populations and facility-
specific population growth rates. These growth rates are derived from county-level population
forecasts until 2015; growth from 2015 until 2119 is assumed to occur at a constant rate equal to
the county's average annual growth rate from 1970 until 2015.
The Agency calculated population risk using year-specific individual risks in a cohort-
averaging approach. Each year-specific risk was calculated using exposure concentrations that
24 Note that the range 10-15 ug/Dl is regarded as a "level of concern," warranting attention
from a medical viewpoint, and not a dose level or threshold below which no adverse health
effects would be expected to occur, i.e., it is not strictly parallel to the definition of an RfD; see
also Appendix E.
25 US EPA 1991. "A PC Software Application of the Uptake/Biokinetic Model, Version 0.5"
Office of Health and Environmental Assessment (ECAO-CIN-2178A).
Draft - March 23, 1993
-------
7-22
were averaged over the nine years forming the exposure duration (i.e., average of nine
consecutive yearly concentrations). Population risk in any given year is related not to the total
exposed population in that year but to that portion of the total population that has received the
full nine-year exposure. This portion of the population is defined as the "risk cohort" for that
year. Total population risk for a facility over the 128-year modeling period is equal to the sum
of the cohort population risk from 1992 through 2119.*
Dimensions for Risk Descriptors
As mentioned earlier, EPA incorporated various dimensions into each of the fundamental
risk descriptors in order to fully describe the human health benefits measures. These dimensions
broaden the scope of the fundamental risk descriptors in describing risk. The various dimensions
are described below, and the interplay between the dimensions and the fundamental risk
descriptors is shown in Exhibit 7-4.
Type of health effect. For this RIA, EPA assessed three types of adverse health effects
that could result due to facility-related exposures. Depending on the contaminant, exposure
could lead to increased cancer risk, adverse noncancer (systemic) effects, or increased blood lead
levels that adversely affect neurological function. Thus, EPA used separate risk descriptors for
each type of effect.
Cancer risk. The individual risk descriptor describing cancer effects is the
individual lifetime excess cancer risk, i.e., an expression of the probability above
background of developing cancer during a lifetime due to exposure.27 EPA
calculated the individual excess lifetime cancer risk by adding risks across
contaminants within an exposure route and across exposure routes within an
exposure pathway. The population risk descriptor for cancer effects - population
cancer risk - describes the number of excess cancer cases expected in the exposed
population over the modeling period (i.e., refers to the probabilistic number of
health effect cases).
For this RIA, the Agency considered individual cancer risks greater than 1 x 10"6
(i.e., a one in one million or greater chance of developing cancer over a lifetime)
to be of concern. EPA also considered population cancer risk equal to one or
more statistical cancer cases to be of concern.
26 The approach for calculating cancer and noncancer population risk is described in detail in
Appendix E.
27 An individual generally may be exposed to sources or contaminants unrelated to the
facility-specific contamination being assessed for risk. Cancer risk incurred due to such
exposures is referred to as "background" risk, as opposed to the incremental risk due to facility-
related exposures.
*** Draft - March 23, 1993 ***
-------
7-23
Noncancer effects. The individual risk descriptor for noncancer effects is the ratio
between the dose (i.e., chronic daily intake) over a specified period and the effects
threshold (i.e., the chronic reference dose or RfD). This ratio is referred to as a
hazard quotient. A hazard quotient greater than one indicates an exceedence of
the threshold level, implying that an adverse noncancer effect may occur in the
exposed individual. Because an individual may be exposed to more than one
contaminant via an exposure route, the hazard quotient for each contaminant is
added to form a risk descriptor called the hazard index. The hazard index
accounts for the cumulative effects of multiple contaminants. Further, a pathway-
specific hazard index is obtained by adding the hazard indices across multiple
exposure routes within the exposure pathway. As with the hazard quotient, a
value greater than one for the hazard index indicates cause for concern for
noncancer effects. As a rule, the greater the value of the hazard index above
unity, the greater is the level of concern. Note, however, that the hazard ratio is
not a measure of statistical probability. The population risk descriptor for
noncancer effects is an estimate of the number of people over the modeling
period exposed to concentration levels that correspond to a hazard index greater
than one.
For this RIA, EPA considered a hazard index greater than 1 to be of concern,
and population noncancer effects for one or more persons exposed to levels above
a hazard index of one to be of concern.
Adverse effects due to lead exposure. EPA characterized effects from lead
exposure differently than cancer risks or noncancer effects due to other
constituents. The individual risk descriptor for lead exposure is the geometric
mean blood lead concentration in the target population, as calculated by the
Lead-UBK model. EPA determined effects due to lead contamination from the
facility, accounting for background lead concentrations, by comparing the facility-
specific mean blood lead level results to the 10 ug/dl threshold.28 Blood lead
levels above the threshold were considered levels of concern.
Off-site vs. on-site risk. For each facility in this analysis, EPA identified locations or
points where exposures could occur. These points were identified based on current land use off-
site of the facility, and on hypothetical future on-site land use. EPA calculated and described
risk separately for off-site vs. on-site exposures.
28 A threshold for the noncancer effects of lead is believed to lie within or below the 10-15
ug/dL range. Note, however, that the range 10-15 ug/dL is regarded as a "level of concern,"
warranting attention from a medical viewpoint, and not a dose level or threshold below which no
adverse health effects would be expected to occur.
*** Draft - March 23, 1993 ***
-------
7-24
Off-site risk. In this analysis, EPA focused on risks resulting from modeled off-
site exposures at actual exposure points. Off-site locations where exposures could
occur due to current uses were identified for the ground water, air, surface water,
soils, and foodchain pathways. The Agency described off-site risks via these
pathways using all four of the fundamental risk descriptors, for both cancer and
noncancer effects.
On-site risk. On-site risks are related to future uses of land within the boundaries
of the facility. Based on future land use predictions made by the expert panels,
EPA expects the future use of the majority of the facilities requiring corrective
action to be primarily residential. Often, these facilities will be used for a
combination of residential, industrial, agricultural, and recreational purposes,
varying in terms of projected use by area within the facility. Of the facilities
evaluated by the expert panels, the specific land use projections include:
Projected Land Use Number of Facilities
Industrial 300
Residential 1,400
Residential and Industrial 60
Residential and Agricultural 430
Residential, Industrial, and 6
Recreational
Residential, Industrial, 60
Recreational, and Agricultural
Land use not addressed 300
Because many of the facilities were projected to have residential and/or
agricultural uses and because of the uncertainty in projecting future land uses,
EPA assumed that all facilities would have future residential and agricultural use
for the purpose of estimating on-site risk. That is, EPA examined hypothetical
risks to individuals residing or growing food on-site at some point in the future at
all the facilities requiring corrective action. The future land uses, and therefore
the exposures, were assumed to begin in 1992. EPA calculated on-site risk due to
exposures via the ground-water, air, soil, and foodchain pathways. The Agency
described on-site hypothetical risk using central tendency individual risk
descriptors, including individual risk to highly-exposed subpopulations.
Average vs. peak risk. For exposures via ground water, surface water, and air, EPA
calculated the individual risk based on a time-varying concentration profile. Exposure
concentrations at a given exposure point in any of these media vary from 1992 through 2119 (the
modeling period), leading to time-varying risks (i.e, year-specific risks). These year-specific, time-
Draft - March 23, 1993
-------
7-25
varying risks were calculated using 9-year exposure durations.29 Because of this time
component, EPA used two measures to provide a range of individual risk estimates for the
ground-water, air, and surface water pathways, i.e., average and peak individual risks. Individual
risks from all other pathways are not subdivided into average and peak risks because MMSOILS
models steady state (i.e., constant) contaminant concentrations in these pathways over the
modeling period.
Average 128-year risk. The average risk for off-site exposures was calculated as
the arithmetic mean of all the 9-year risks over the 128-year modeling period,
including risks for years with zero concentrations. The average risk was calculated
differently for the on-site hypothetical individual risk: it was based on a 128-year
average concentration, which approximates the estimate that would be derived
using the off-site average risk method.
Peak 9-vear risk. Peak risk for off-site exposures was calculated as the highest 9-
year risk. The peak risk for the on-site hypothetical individual risk was based on
the highest 9-year average concentration, and is very close to the estimate that
would be derived using the method for off-site peak risk.
In this RIA, EPA used both the 128-year average and 9-year peak risk descriptors for the
ground-water, air, and surface water pathways. Neither measure can be selected a priori as the
single correct measure; therefore, EPA used both measures to provide a range of the risk
estimates for these pathways. Using just the 128-year average may lead to understating the
baseline risks in some cases because high risks occurring during a short time span may be
"averaged out" to levels below concern, i.e., peak risks, in particular noncancer effects, may be
missed. For example, a hazard index could be greater than one for a few years in the beginning
of the modeling period, and then could dip below one for the remaining years. The 9-year peak
risk descriptor would be able to capture this reduction in risk of concern while the 128-year
average would not.
Use of the 9-year peak risk descriptor alone, on the other hand, may lead to
understatement of risk reductions following remediation. This is due to the fact that in some
cases, early in the modeling period, the baseline peak risk and the post-remediation peak risk do
not differ significantly, but after these coincident peaks, the post-remediation risks drop much
more rapidly than the baseline risks. In this case, the 128-year average risk is a better indicator
of the long-run effectiveness of corrective action measured over the entire modeling period.
Risk with exposure concentrations "capped" versus "uncapped." EPA estimated ground
water individual and population risk for two scenarios: exposure to ground water concentrations
not capped at MCLs or taste/odor thresholds, and exposure to constituent concentrations in
ground water capped at MCLs or taste/odor thresholds.
29 See section E.2.4 in Appendix E for details on how EPA calculated intakes and risks from
time-varying concentrations.
*** Draft - March 23, 1993
-------
7-26
Ground water risk - concentrations uncapped. For this risk descriptor, EPA
excluded the effect of MCLs and the taste/odor thresholds from the exposure
assessment. The Agency assumed that public ground-water supplies will contain
concentrations in excess of MCLs, even though many of the constituents have
MCLs which require monitoring and compliance with concentration standards.
Also for many constituents, the Agency determined taste and odor thresholds -
concentrations at which most people can detect the contamination in drinking
water. When taste and odor thresholds are exceeded, most people would avoid
consumption of the water, and municipalities could be prompted to test water
supplies for impurities.
Ground water risk - concentrations capped. For this risk descriptor, EPA
assumed that, for constituents with MCLs and/or taste and odor thresholds,
exposure would not occur at levels exceeding the MCLs, or taste and odor
threshold, whichever was the lowest.
Capped/uncapped risk descriptors are not applicable to any of the other pathways.
Baseline vs. post-remediation risk. To estimate human health benefits attributable to
corrective action, EPA subtracted risk remaining after corrective action from baseline risk. The
two types of risk descriptors that EPA used in this estimation are as follows:
Baseline risk. Baseline risks refer to the adverse health effects on individuals and
populations living near corrective action facilities that would result if corrective
actions were not implemented. Baseline risks are due to exposures to existing
contamination and to potential future releases between 1992 and 2119 (the
modeling time frame for this analysis). The Agency estimated baseline risk for all
the risk descriptors described above.
Post-remediation risk. Post-remediation risks refer to risks remaining after
corrective action is implemented. EPA expects that corrective action would
reduce levels of contamination resulting from existing and future contaminant
releases at RCRA facilities. Risks associated with these reduced levels are the
post-remediation risks. Remediation is assumed to start in the same year for the
risk assessment as for the cost analysis.
Because the remedy selection process was based on central tendency estimates of extent
of contamination and exposure, post-remediation risks are estimated for only the central
tendency individual risk, individual risk to highly-exposed subpopulations, and population risk.
Also, post-remediation risks focus on off-site exposures only. Post-remediation risks are based
on a subset of the facilities requiring corrective action, i.e., a sample representing 720 of the
approximately 2,600 facilities requiring corrective action. This subset was defined primarily by
occurrence of risk of concern or extensive contamination in the baseline.
Draft - March 23,1993 »**
-------
7-27
7J Results
This section presents detailed results of the human health benefits analysis. For this
RIA, the Agency defines human health benefits as the reductions in cancer risk and noncancer
effects resulting from corrective action. Results are presented for baseline risk (i.e., risk in the
absence of corrective action) and for reductions in baseline risk due to simulated implementation
of corrective action remedies.
7.3.1 Baseline Risks
Baseline risks refer to the adverse health effects on individuals and populations living
near corrective action facilities that would result if corrective actions were not implemented.
Baseline risks are due to exposures to existing contamination and to potential future releases
between 1992 and 2119.
EPA presents in this section baseline risks for off-site exposures, and for a hypothetical
future on-site exposure scenario. For off-site exposures, the Agency calculated central tendency
individual risks; high-end individual risks; individual risks to highly-exposed or sensitive
subpopulations; and population risks. EPA calculated only central tendency individual risks and
individual risks to highly-exposed subpopulations for the on-site exposure scenario.
Baseline Off-Site Individual Risks
The Agency predicts from 920 (36 percent) to 1,700 (67 percent) of the approximately
2,600 facilities triggering corrective action to pose baseline risks of concern under the central
tendency and high-end assumptions, respectively.30 Numbers of facilities with risks under each
scenario (i.e., central tendency and high-end) are presented using both average and peak risk
descriptors. Exhibit 7-6 summarizes the total number of facilities with off-site individual risk of
concern across all pathways and across cancer risks and noncancer effects. Columns of the
exhibit show the number of facilities with cancer risks and noncancer effects of concern in the
central tendency and high-end exposure scenarios. Rows show the number of facilities at each of
the levels of concern. The total row (i.e., the bottom row) shows the total number of facilities
with cancer risk or noncancer effects of concern. (The total presented does not double-count
facilities that have both cancer risks and noncancer effects.)
In general, because high-end assumptions produce greater exposure concentrations in
some pathways than do central tendency assumptions, the total number of facilities with high-end
cancer risks or noncancer effects of concern are expected to be greater than the total number of
facilities posing central tendency risks of concern. In some cases, however, within a single risk
range, it is possible for there to be more facilities with central tendency risks of concern than
with high-end risks of concern (i.e., within rows other than totals rows). For example, in Exhibit
30 EPA considered cancer risks of 1 x 10* and above and hazard indices greater than unity to
be of concern.
Draft - Mareh 22, 1993 ***
-------
7-28
EXHIBIT 7-6
NUMBER OF FACILITIES POSING OFF-SITE INDIVIDUAL CANCER
AND NONCANCER EFFECTS VIA ALL PATHWAYS
(n = 2,600)
Risk
Descriptor
| Baseline-Central Tendency
Based on 128-
1 year Average
Based on 9-
Year Peak
Baseline-High-End
Based on 128-
year Average
Based on 9-
Year Peak
Facilities with Individual Cancer Risk
21 x 10'1 to £1
21 x ID* to <1 x 10*
21 x 10* to <1 x 10"
Total 2:1 x 10*
3
210
640
850
3
210
640
850
80
630
610
1300
300
410
610
uoo
Facilities with Individual Noncancer Effects
Hazard Index 2100
210 to <100
21 to <10
Total with
Hazard Index 21
Total FadlUitt of Feting i
Risk/Effects of Concent
0
140
13
160
920
0
140
230
380
*2»
310
560
/ 260
1,100
1,700
310
560
270
1,100
1,700
Subtotals may not add to totals due to rounding.
Interpretation of cancer risks:
21 x 10'2 to S1 denotes a cancer risk greater than or equal to 1 in 100 and less than or equal to 1
21 x 10"4 to <1 x 10* denotes a cancer risk greater than or equal to 1 in 10,000 and less than 1 in 100
21 x 10"* to <1 x 10"* denotes a cancer risk greater than or equal to 1 in 1,000,000 and less than 1 in
10,000
7-6, there are 640 facilities posing central tendency cancer risk within the 1 x 10"6 to 1 x 10"4
range, but there are only 610 facilities posing high-end cancer risk within the same range. This
reflects the different distribution of facilities with risk of concern under the two scenarios. A
portion of the facilities posing cancer risks in the 1 x 10"6 to 1 x 10"* range under central tendency
pose higher risks (e.g., in the 1 x 10"* to 1 x 10-2 range) under the high-end scenario. Thus, the
number of facilities in the 1 x 10* to 1 x 10"* range in the high-end is lower than in the central
tendency scenario.
Off-site individual risks under both sets of assumptions (i.e., central tendency and high
end) include significant numbers of facilities with cancer risk and noncancer effects. Cancer risks
*** Draft - March 22,1993
-------
7-29
of concern are predicted at 850 to 1,300 facilities (33 to 51 percent) under the central tendency
and high-end assumptions, respectively. For noncancer effects, the Agency predicts effects at 160
to 380 facilities (6 to 15 percent) under central tendency assumptions, and at 1,100 facilities (44
percent) under high-end assumptions. By comparing the effects of central tendency and high-end
assumptions, it appears that high-end assumptions nearly double the number of facilities with
risks of concern, and cause an especially large increase in the number of facilities with noncancer
.effects of concern.
Total off-site individual risks include risk attributable to exposures to media and
foodchain pathways and risk to highly-exposed or sensitive subpopulations.31 More detail on
each group of pathways is provided below.
Baseline central tendency and high-end individual risk - media and foodchain pathways.
Among the media and foodchain pathways, the Agency predicts individual risks of concern to
result from all pathways except exposure via inhalation and drinking surface water. Risks via the
air pathway are predicted to be very low or zero because most releases from the SWMUs to air
either finish occurring before, or occur for only a short period in the beginning of, the modeling
period. Risks from drinking surface water are predicted to be very low or zero because relatively
few RIA facilities (i.e., about 400 among the 2,600 facilities) are located near rivers or lakes used
as sources for municipal drinking water. The drinking water intakes were generally located on
sufficiently large rivers of sufficiently far downstream that the predicted contamination is
significantly diluted or degraded, resulting in very low risks. Under central tendency
assumptions, risks of concern due to surface water recreation are expected at 7 facilities. 128-
year average risks of concern due to surface water recreation are expected at 27 facilities under
high-end assumptions. Peak 9-year risks would be of concern at 97 facilities under high-end
assumptions.
Numbers of facilities with cancer risks and noncancer effects of concern from each media
and foodchain pathway are presented in Exhibits 7-7 to 7-10. As these exhibits show, the most
pervasive risks (i.e., the risks of concern at most facilities) appear to be due to ground water
exposure (especially based on 9-year peak risks) (shown in Exhibit 7-7); to children exposed to
soil in off-site fields and agricultural fields (Exhibit 7-9); and to consumption of vegetables
(Exhibit 7-10). As expected, many more facilities have individual risks of concern via ground
water exposures under high-end assumptions (20 to 22 percent of facilities triggering corrective
action) than under central tendency assumptions (3 to 12 percent). Compared to the central
tendency scenario, the high-end often results in higher rates of contaminant release from
SWMUs to all media, but may in some cases may increase the mass released to ground water
and leave less mass available for release to other pathways. One result of this is that adult
individual noncancer effects due to off-site soil exposure would be of concern at fewer facilities
under high-end assumptions (less than 1 percent) than under central tendency assumptions (3
percent).
31 Media pathways are the ground water, surface water, air, and soils pathways; foodchain
pathways refer to fish, vegetable, beef, and milk ingestion; and highly-exposed or sensitive
subpopulations include pica children, subsistence farmers, subsistence fishermen, and children
exposed to lead.
** Draft - March 23,1993 ***
-------
7-30
EXHIBIT 7-7
NUMBER OF FACILITIES POSING OFF-SITE INDIVIDUAL CANCER RISK AND
NONCANCER EFFECTS VIA GROUND WATER PATHWAY -- BASELINE
(CONCENTRATIONS UNCAPPED)'
(n = 2,600)
Risk I) Baseline-Central Tendency
[I Based on 128-
|| year Average
Based on 9-
Year Peak
Baseline-High-End
Based on 128-
year Average
Based on 9-
Year Peak
Facilities with Individual Cancer Risk
£1 x IQ-'to SI
SI x W to <1 x 1CT2
SI x 10-" to <1 i 1O*
Total S 1 x 10*
0
11
73
84
0
14
70
84
7
490
7
500
Facilities with Individual Noncancer Effects
Hazard Index 2s 100
S10 to <100
SI to <10
Total with
Hazard Index SI
Total Facilities Posing :
Risk/Bfecte of
Concern
0
0
11
11
84
0
0
230
230
300
220
280
0
300
510
220
280
73
570
220
280
3
500
570
Subtotals may not add to totals due to rounding.
*** Draft - March 22,1993 **
-------
7-31
EXHIBIT 7-8
NUMBER OF FACILITIES POSING OFF-SITE INDIVIDUAL CANCER RISK AND
NONCANCER EFFECTS VIA GROUND WATER PATHWAY - BASELINE
(CONCENTRATIONS CAPPED)'
(n = 2,600)
Risk
Descriptor
Baseline-Central Tendency
Based on 128-
year Average
Based on
9- Year Peak
Baselme-High-End
Based on 128-
year Average
Based on
9- Year Peak
Facilities with Individual Cancer Risk
2 1 x Iff1 10 £ 1
21 x 10-" to <1 x 102
21 x Wio <1 x 10"
Total 21 x 10*
0
11
73
84
0
11
73
84
0
63
440
500
0
490
77
570
Facilities with Individual Noncancer Effects
Hazard Index 2 100
210to <100
21 to <10
Total with
Hazard Index 21
Tola] Fatalities PfcsSng
RiskfEflecb of Concern
0
0
7
7
«4
0
0
11
11
«4
0
- 0
490
490
500
0
0
490
490
570
Subtotals may not add to totals due to rounding.
***
Draft - Mareh 22, 1993
-------
7-32
EXHIBIT 7-9
NUMBER OF FACILITIES POSING OFF-SITE INDIVIDUAL
CANCER RISKS AND NONCANCER EFFECTS DUE TO SOIL EXPOSURES
IN AGRICULTURAL AND OFF-SITE FIELDS - BASELINE *
(n = 2,600)
Risk
Descriptor
Agricultural Field Soil Exposures
Adult
Central
Tendency
& 1 x 10'2 to S 1
£1 x ID"4 to <1 x 10-
SI x ID* to <1 x 10'
4
Total Six lO*
0
0
130
130
Hazard Index S 100
S10 to <100
SI to <10
Total with
Hazard Index SI
Tbtat Facilities
P0»ng Bisk/Effects
of Co&eerfl
0
0
0
0
130
High-
End
Child
Central
Tendency
High-
End
Off-Site Field Soil Exposures
Adult
Central
Tendency
High-
End
Child
Central
Tendency
Facilities with Individual Cancer Risk
0
0
140
140
0
0
130
130
Facilities with Individual
3
0
210
220
350
0
0
0
0
13d
0
0
140
140
0
3
67
70
0
3
67
70
Noncancer Effects
3
0
210
220
330
a
63
7
70
33 ,
0
0
3
3
TO
0
3
340
350
High-
End
0
67
220
280
0
67
3
70
350
0
0
67
67
350
Subtotals may not add to totals due to rounding.
Draft - March 22, 1993 ***
-------
7-33
EXHIBIT 7-10
NUMBER OF FACILITIES POSING OFF-SITE INDIVIDUAL CANCER RISKS AND
NONCANCER EFFECTS VIA FOODCHAIN PATHWAYS - BASELINE
(n = 2,600)
Risk
Descriptor
Vegetable
Consumption
Central
Tendency
High-
End
Fish Consumption
Central
Tendency
High-
End
Beef and Milk Consumption
Central
Tendency
High-End
Facilities with Individual Cancer Risk
^1 i 10'2 to £ 1
^1 x 10"1 to <1 x 10J
;>! x ID"* to <1 x 10*
Total £1 x 10*
0
130
190
320
7
180
130
320
0
0
7
7
0
10
220
230
0
63
0
63
0
63
130
190
Facilities with Individual Noncancer Effects
Hazard Index S 100
^ 10 to < 100
2*1 to <10
Total with
Hazard Index il
Total Facilities Jtasi&g
RisttEffects
-------
7-34
Pathways with the highest predicted individual cancer risks and noncancer hazard indices
at any single facility are vegetable consumption and exposure to ground water. Exhibits 7-11 and
7-12 show the highest predicted individual cancer risks and noncancer hazard indices at any
single facility estimated among all the exposure pathways. The maximum predicted individual
cancer risk with central tendency assumptions is 7.3 x 10'3, and is attributable to vegetable
consumption. However, under the high-end assumptions, the ground water pathway is projected
to pose the highest cancer risk, i.e., a risk of unity.32 This cancer risk occurs at a facility where
the constituent concentrations in waste were estimated to be very high, and, due to the complex
hydrogeologjc conditions, EPA assumed no retardation and dilution of the constituents, leading
to very high exposure-point concentrations. The highest high-end noncancer hazard index is also
due to ground water exposures at the same facility. The highest noncancer hazard index is
predicted to result from vegetable ingestion (3.4 x 10') under central tendency assumptions, but
would result from ground water exposures (6.7 x 104) under high-end assumptions (Exhibit 7-12).
In general, highest cancer risks and noncancer hazard indices are usually of the same order of
magnitude using either the average or peak risk descriptor.
Baseline individual risk to highly-exposed or sensitive subpopulations. EPA assumed the
existence of highly-exposed or sensitive subpopulations and estimated risk for the central
tendency exposure scenario. For sensitive subpopulations, EPA assessed risk to children up to 7
years of age exposed to central tendency concentrations of lead.
Because of their higher exposure contact potential, individuals of highly-exposed
subpopulations are expected to incur risks of concern at more facilities compared to individuals
exposed via other pathways under central tendency assumptions. .-As shown in Exhibit 7-13, each
of the highly-exposed subpopulations would have risks of concern at over 300 facilities, and
subsistence fishermen appear to be exposed to risk of concern at the largest number of facilities.
EPA expects contamination about 360 (14 percent) of the 2,600 facilities to pose risks of concern
if subsistence fishermen fish in waters downstream of these facilities.
The Agency estimated that children's lead exposures at 70 facilities (about 3 percent of
the facilities triggering corrective action) would be of concern (i.e., result in blood lead levels
above 10 ug/dL). At most of these facilities, lead exposure would occur via the soil and
fppdchain pathways.
Highest predicted cancer risks and noncancer hazard indices at any single facility for each
highly-exposed subpopulation are presented in Exhibits 7-14 and 7-15. Among the highly-
exposed subpopulations, the highest cancer risk levels under central tendency assumptions are
predicted for subsistence farmers (53 x 10'2), and the highest noncancer hazard indices are
predicted for the off-site field soil exposures to pica children (5.9 x 101).
32 This result implies that cancer is certain to occur. The Agency obviously cannot project
the occurrence of health effects with certainty; this result is presented only as indicative of the
magnitude of risks potentially resulting from the high-end assumptions.
*** Draft - March 22. 1993
-------
7-35
EXHIBIT 7-11
HIGHEST ESTIMATED FACILITY-SPECIFIC INDIVIDUAL CANCER RISK BY
MEDIA AND FOODCHAIN PATHWAY
(Baseline -- Central Tendency and High-end)
Pathway
Individual Cancer Risk
Central Tendency
9-year Peak
128-year
Average
Media Pathways
Ground Water
Air
Surface Water-Drinking
Surface Water- Swimming
Off-Site Field-Soil-Adult
Off-Site Field-Soil-Child
Agricultural Field-Soil-
Adult
Agricultural Field-Soil-Child
Food chain-Fish
Food chain-Vegetables
Foodchain-Beef and Milk
'Highest of All Pathways
3.1 x 10J
4.3 x 10''
1.9 x 1Q-10
1.7 i W
3.7 x 10"
5.8 x 10"
2.5 x 10J
3.9 x 10''
2.5 x 10'
4.0 x 10'7
89 x 10-"
1.0 x 10'
3.7 x 10"
5.8 x 10"
2.5 x 10J
3.9 x 10J
Foodchain Pathways
8.6 x 10"'
7.3 x 10's
1.8 x 10"
7 J * «T»
8.6 x 10*
7.3 x 10J
1.8 x 10"
73x10''
High-end
9-year Peak
1.0 x 10°
1.2 x 10*»
4.4 x 10*
1.3 x 10"
5.4 x 10"
8.4 x 10"
5.2 x 10J
8.0 x 10"'
2.7 x 10J
64 x ID"2
62x10"
1x10"
128-year Average
1.0 x 10°
1.7 x lO"10-
2.8 x lO*
9.2 x ID"5
5.4 x 10"
8.4 x 10"
5.2 x 10*
8.0 x ID"5
2 7 x 10J
6.4 x 10'1
6.2 x 10"
1x10*
* Air risks are lower under high-end assumptions than under central tendency assumptions. High-end
assumptions often result in greater constituent mass for release to all media but may also add proportionally
more constituent mass to the ground water pathway, this may in some cases result in less mass available to -
and thus lower risks in - other pathways like air and soil.
Notes: (1) The noncancer hazard index is not segregated by target organ or effect.
(2) Peak and average risks do not vary across the soil or foodchain pathways because of steady state
assumptions.
*** Draft - March 22, 1993
-------
7-36
EXHIBIT 7-12
HIGHEST ESTIMATED FACILITY-SPECIFIC INDIVIDUAL NONCANCER
HAZARD INDEX BY MEDIA AND FOODCHAIN PATHWAY
(Baseline - Central Tendency and High-end)
Pathway
Ground Water
Air
Surface Water-Drinking
Surface Water-
Swimming
Off-Site Field-Soil-
Adult
Off-Site Field-Soil-Oiild
Agricultural Field-Soil-
Adult
Agricultural Field-Soil-
Child
Individual Noncancer Hazard Index
Central Tendency
9-year Peak
128-year
Average
Media Pathways
4.9 x 10°
1.0 x 10"
1 3 x ID''
1.6 x 10°
1.1 x 10'
3.2 x 10'
8.2 x ID'2
23 x W
4.5 x 10'
8.9 x 10"'
5.0 x 10"
1.6 x 10°
1.1 x 10'
3.2 x 10'
8.2 x 10J
23 x 10*
High-end
9-year Peak
128-year
Average
6.7 x 10'
2.2 x 10-5*
3.7 x ID"2
2.8 x 102
U x 10°
4.1 x 10°
2.6 x 101
1.9 x 10'
6.3 x 104
2.2 x 10s*
1.0 x W
1.2 x 102
1.5 x 10°
4.1 x 10°
2.6 x 101
1.9 x 10'
Foodchain Pathways
Food chain-Fish
Food chain- Vegetables
Foodchain-Beef and
Milk
Highest of AUPartiWB)*
1 .5 x 10°
3.4 x 10'
23 x lO4
3.4* SO1
1.5 x 10°
3.4 x 10'
2.3 x 10"1
3.4X101
5.5 x 10'
2.1 x 10'
2.2 x 102
«-7XlO'
5.5 x 10'
2.1 x 104
2.2 x 102
&3XI0'
Air risks are lower under high-end assumptions than under central tendency assumptions. High-end
assumptions often result in greater constituent mass for release to all media but may also add proportionally
more constituent mass to the ground water pathway, this may in some cases result in less mass available to -
and thus lower risks in - other pathways like air and soil.
Notes: (1) The noncancer hazard index is not segregated by target organ or effect.
(2) Peak and average risks do not vary across the soil or foodchain pathways because of steady state
assumptions.
*** Draft -- Mareh 22,1993 *
-------
7-37
EXHIBIT 7-13
NUMBER OF FACILITIES POSING OFF-SITE INDIVIDUAL CANCER RISKS AND
NONCANCER EFFECTS TO HIGHLY-EXPOSED
OR SENSITIVE SUBPOPULAT1ONS - BASELINE *
Risk
Descriptor
S 1 x 10* to £ 1
SI z Wto <1 x 10°
SI x lO-'to <1 x 10"
Total
SI x lO"4
Baseline-Central Tendency
Agricultural Field-
Soil Exposure to
Pica Children
Off-Site Field-
Soil Exposure to
Pica Children
Facilities with Individual Cancer Risk
0
63
280
350
0
70
280
350
Subsistence
Farmer
Subsistence
Fisherman
3
190
130
320
0
7
290
290
Facilities with Individual Noncancer Effects
Hazard Index S 100
2:10 to <100
SI to <10
Total with
Hazard Index SI
Total PaciBtks Posing
tteMBffcctt of
Concern
0
0
0
0
350
<0
67
3
70
350
0
: 7
7
13
930
0
74
7
81
360
Subtotals may not add to totals due to rounding.
*** Draft « March 22,1993 ***
-------
7-38
EXHIBIT 7-14
HIGHEST ESTIMATED FACILITY-SPECIFIC INDIVIDUAL CANCER RISK
BY HIGHLY-EXPOSED SUBPOPULATION PATHWAY
(Baseline Central Tendency)
Highly-Exposed
Subpopulalion
Highly-Exposed
Agricultural Field-Soil
Exposure to Pica Children
Off-Site Field-Soil
Exposure to Pica Children
Subsistence Farmer
Subsistence Fisherman
Highest of All Partway*
Individual Cancer Risk
Central Tendency
Subpopulations
1.0 x 10"
1.1 x 10J
5.3 x 10J
3.9 x 10s
5.3 x 10*
Note: Peak and average risks are identical due to the use of steady state assumptions.
EXHIBIT 7-15
HIGHEST ESTIMATED FACILITY'SPECIFIC INDIVIDUAL NONCANCER HAZARD INDEX
BY HIGHLY-EXPOSED SUBPOPULATION PATHWAY
(Baseline Central Tendency)
Pathway
Highly-Exposed
Agricultural Field-Soil
Exposure to Pica Children
Off-Site Field-Soil
Exposure lo Pica Children
Subsistence Farmer
Subsistence Fisherman
Highest of All ?Wfw*y* j
Individual Noncancer
Hazard Index Central
Tendency
Subpopulations
4.2 x 104
5.9 x 10'
5.6 x 10'
2.0 x 10'
S-dxIO*
Note: Peak and average risks are identical due to the use of steady state assumptions.
***
Draft - March 22, 1993 ***
-------
7-39
Baseline On-Site Hypothetical Individual Risk
In addition to off-site risks, EPA quantified hypothetical risks to individuals that may
reside on current facility sites in the future. Under this scenario, the Agency assumed that
facilities would close in the future and the land would be converted to residential and/or
agricultural use. Assuming these future on-site uses at all sample facilities, EPA predicted
cancer risks and noncancer effects of concern to occur at about 1,800 (71 percent) of the
facilities triggering corrective action. Facilities expected to pose on-site risks include
approximately 1,800 facilities (69 percent) posing cancer risks of concern and 1,600 facilities (60
percent) posing noncancer effects of concern. As shown in Exhibit 7-16, about 850 facilities
would have very high cancer risks (i.e., lifetime individual cancer risk of 1 x 102 or greater) and
about 570 facilities would have very high hazard indices (i.e., greater than 100).
Even though the off-site risk assessment included more exposure pathways than the on-
site risk assessment, the Agency predicted that, under central tendency assumptions, the total
number of facilities with cancer risks and noncancer effects of concern would be greater using
the on-site risk descriptors (1,800 facilities) than using the off-site risk descriptors (920
facilities).33 It is not surprising that the number of facilities with risks of concern on site is
greater than the number of facilities with risks of concern off site, because on-site exposure
points are closer than off-site exposure points to contaminant sources.
Baseline hypothetical individual risk media and foodchain pathways. Pathways posing
on-site risk of concern at the largest number of facilities include ground water exposure (58
percent of facilities), vegetable consumption (55 percent), and soH exposures to children (44
percent). Media and foodchain pathway risk of concern occurs at a large numbers of facilities,
for both cancer and noncancer effects. For example, more than 1,000 facilities are expected to
have cancer risk of concern resulting from ground water exposures, vegetable consumption, and
soil exposures to children. Additionally, about 1,200 facilities are expected to have noncancer
effects of concern resulting from vegetable consumption. Note that if ground water
concentrations are assumed to be capped at MCLs or taste/odor thresholds, there will be a
decrease of approximately 7 percent in the number of facilities with ground water cancer risks or
noncancer effects of concern.
Baseline hypothetical individual risk to highly-exposed subpopulations. For the on-site
risk assessment, the Agency estimated risks for two highly-exposed subpopulations, pica children
and subsistence farmers. Numbers of facilities with baseline risks of concern for these
subpopulations are included in Exhibit 7-16. The Agency predicts that in the absence of
corrective action, there would be 1,100 facilities (44 percent) with on-site individual cancer risks
or noncancer effects of concern for pica children and 1,400 facilities (55 percent) with on-site
individual cancer risks or noncancer effects of concern for subsistence farmers.
33 The analysis of on-site risks excluded surface water-related pathways (i.e., surface water-
drinking, surface water-recreation, foodchain-fish, and subsistence fishing) because there were
few surface water exposure points on site.
»** Draft - March 22, 1993 **
-------
7-40
EXHIBIT 7-16
ON-SITE HYPOTHETICAL EXPOSURES: NUMBER OF FACILITIES WITH CANCER RISK AND NONCANCER EFFECTS OF
CONCERN ~ BASELINE
Risk/
Hazard Descriptor
Media Pathways
Ground Water
Cone.
Uncap.
Cone.
Cap.
Air
Soil-Adult
Soil-Child
Foodchain Pathways
Foodchain-
Vegelables
Food chain-
Beef and
Milk
Highly-Exposed Subpopulations
Soil-Pica
Children
Subsistence
Fanner
Across all
Pathways*
Facilities with Individual Cancer Risk
2lxlOJlo si
alx 10"to <1 xlOJ
2lxIO*to
-------
7-41
Baseline Population Risk
EPA calculated population risks due to off-site exposures via ground water, air, and
surface water. The Agency found that all population risks of concern are attributable solely to
ground water exposures,34 and mainly at public wells.
As explained earlier, EPA estimated ground-water population risk for two scenarios, one
where constituent concentrations were not capped at MCLs or taste and odor thresholds, and the
other where they were capped. For the uncapped scenario, EPA estimates that there would be
approximately 21,000 statistical cancer cases and 25,000,000 persons experiencing noncancer
health effects over the 128-year modeling period in the absence of corrective action. For the
capped scenario, EPA estimates that there would be approximately 1,200 cancer cases and 900
noncancer health effects over the modeling period. The very significant difference between the
uncapped and capped results is due to the fact that population risks result primarily from
exposure via public wells and that most of the constituents that contribute to the risk have
MCLs. The two sets of population risk estimates are shown in Exhibit 7-17.
EXHIBIT 7-17
BASELINE POPULATION RISK DUE TO GROUND WATER EXPOSURE
Risk Descriptor
BASELINE
Population
Cancer Risks1
Population
Noncancer
Effects2
Public
Wells
Private
Wells
Public
Wells
Private
Wells
Risk/effects based on
exposure to
concentrations without
MCL or taste/odor
threshold caps
Risk/effects based on
exposure to
concentrations
capped at MCL or
taste/odor thresholds
21,000
100
25,000,000
6,000
1,100
40
0
11
Notes. (1) Cancer cases over the 128-year modeling period
(2) Number of people with hazard index greater than 1 over the 128-year modeling period
34 See the discussion of baseline individual risk results for an explanation of why the air and
surface water pathways are not projected to present significant risks.
*** Draft - March 22,1993
-------
7-42
An additional finding of the Agency's population risk assessment is that most of the
predicted population risk is attributable to a single high-risk facility, with smaller population risks
contributed by several others. This facility is located just up-gradient from municipal wells
estimated to serve a significant population.
Under the uncapped scenario (i.e., excluding effects of MCLs and taste and odor
thresholds) population cancer risks are projected to occur at 74 (3 percent) of the 2,600 facilities
triggering corrective action. That is, there would be at least one cancer case over the entire
modeling period at these facilities. Population noncancer effects are projected to occur at about
230 (9 percent) of the facilities triggering corrective action. That is, at least one person over the
modeling period would be exposed to contamination above the noncancer threshold (i.e., hazard
index greater than one) at these facilities.
Number of Facilities Potentially Posing Baseline Risk of Concern Across All Descriptors
To summarize the total number of facilities with baseline risks of concern, the Agency
estimated the number of facilities with any individual or population risk/effects (i.e., with cancer
risk or noncancer effects) of concern. Exhibit 7-18 shows percentages of facilities triggering
corrective action that the Agency predicts to pose risks/effects of concern across all descriptors.
In this exhibit, all percentages are based on the "concentrations uncapped" scenario for ground
water exposures. This set of facilities is the union of facilities posing risks/effects of concern for
off-site individuals, populations, and hypothetical on-site individuals.
The pies in the upper left corner of Exhibit 7-18 show the'percentage of facilities with
off-site individual risks/effects of concern under central tendency and high end assumptions.
Under central tendency assumptions, 36 percent of the 2,600 facilities triggering corrective action.
are predicted to pose off-site risk/effects of concern. Under high-end assumptions, the Agency
predicts off-site risks/effects of concern at 67 percent of the 2,600 facilities.
The Agency estimated off-site population risks/effects of concern at 11 percent of the
2,600 facilities triggering corrective action (as shown in the bottom left pie in Exhibit 7-18).
However, since each of the facilities with population risk/effects of concern also poses off-site
individual risk/effects of concern, this set of facilities is a subset of facilities posing off-site
individual risks/effects of concern.
As shown in the upper right pie of Exhibit 7-18, the Agency predicts hypothetical on-site
individual risk/effects of concern at 71 percent of the 2,600 facilities triggering corrective action.
The Agency obtained the union of facilities with hypothetical on-site individual risks/effects and
facilities with off-site individual risk/effects or population risk/effects to determine the percentage
of facilities with at least one type of risk/effect of concern. (This union does not double count
facilities that have on-site and off-site or population risk/effects.) As shown in the lower left pies
in Exhibit 7-18, 73 percent of the 2,600 facilities triggering corrective action are expected to pose
human health risk/effects of concern assuming central tendency assumptions. If high-end
assumptions are used for off-site risk/effects, 84 percent of the facilities pose human health
*** Draft - March 22, 1993 ***
-------
Exhibit 7-18
Percentage of Facilities With Baseline
Cancer Risks and Non-Cancer Effects of Concern
[N = 2,600]
Central Tendency High-End
36% Risk 67% Risk
64% No Risk
33% No Risk
Facilities With Off-Site Individual
Risk/Effects of Concern '
Central Tendency Only
71% Risk
29% No Risk
Facilities With Hypothetical On-Site
Individual Risk/Effects of Concern
i
Central Tendency Only
\\7c Risk
89% No Risk
Facilities With Off-Site Population
Risk/Effects of Concern
Based on both I 30-year average and 9-year peak risks
Across off-site (130-year average and 9-year peak), on-sile, and population risk results
Central Tendency
73% Risk ^77
High-End
84% Risk
27% No Risk
16% No Risk
Facilities With Any Individual or
Population Risk/Effects of Concern
-------
1-44
risk/effects of concern. Given these results, about 16 percent, or 400 of the facilities that would
trigger corrective action are not expected to pose human health risk/effects of concern.
At a majority of these 400 facilities, the expert panels prescribed corrective action with
concern for future exposures via ground water or off-site soils. EPA's modeling, however,
indicated that such future exposures are unlikely, at least within 128 years. Contamination at
some of these facilities exists only in the unsaturated zone, and might not reach the water table
within the modeling period due to extremely slow movement of the water within the pore spaces.
At other facilities, the contaminated soils are not expected to be transported to points where
exposures could occur.
Constituents Driving Baseline Off-Site Risk
Baseline off-site human health risks at corrective action facilities are driven by organic
and inorganic constituents which are associated with a diverse array of adverse health effects.
The Agency identified the main risk driving constituents from among the nearly 100 constituents
found in SWMUs for all of the off-site pathways, separately for cancer risks and noncancer
effects. Risk drivers discussed here are constituents at facilities with the highest predicted
lifetime individual cancer risk or hazard index, i.e., they are the constituents posing highest
individual and population risks, rather than the most frequent constituents at facilities presenting
risks of concern. Exhibits 7-19a and 7-19b present, respectively, the constituents posing the
highest cancer risks and the greatest exceedences of noncancer health effects thresholds, along
with their critical health effects.35 Constituents are listed in descending order of frequency of
occurrence. The exhibit includes central tendency and high-end risk driving constituents for
pathways with baseline risks of concern. In addition, population risk drivers are presented for
the ground water pathway. The Agency identified population risk drivers in ground water based
on the number of cancer cases or noncancer exceedences of threshold levels associated with each
constituent.
The noncancer health effects of common risk-driving constituents include interference
with blood pressure and chemistry, liver and kidney damage, and effects on the central nervous
system. Critical health effects of cancer risk drivers include cancer of the skin, liver, kidneys,
lungs, mammary glands, and adrenal glands. Risk drivers for the ground water pathway
(including population risk drivers) are apparently less important in other pathways (i.e., drive
risks in few pathways other than ground water). Ground water risks are driven primarily by
benzene, 1,1-dichloroethylene, and chloroform. Exposures to benzene are associated with
leukemia. In general, there are differences between constituents that drive risk under central
tendency and high-end assumptions. These differences probably relate to rates of constituent
release into and fate and transport in each medium under each set of assumptions.
35 Types of effects discussed here are those on which the reference dose or slope factor for
the constituent is based. Therefore, these are probably the most sensitive endpoints or effects,
especially in the test species; other effects could also occur in humans.
*** Draft - March 22, 1993 ***
-------
E, , 7-19a
KEY CONSTITUENTS DRIVING NONCANCER EFFECT LEVELS*
PATHWAYS
Constituents Posing
Noncancer Effects
Chromium (VI)
Tetraethyl lead
Chlordane
Pentachlorophenol
Barium
Cadmium
Arsenic
Nickel
Bis(2-ethylhexyl)
phthalate
Tetrachloroethylene
Formaldehyde
Phenol
Methylene chloride
Toluene
Ground Water*
CT* HE' POP*
V V V
V
V
V V V
V V V
SW Swim*
CT HE
V
V
Off-
site/Agric.
Soil'
CT HE
V V
V V
V
V
V V
V
V
Foodchain*
CT HE
V
V V
V V
V
V
V
V V
V
V
V
Off-site/Agric.
Soil
Pica Child"
CT HE
V V
V
V
V
V V
V
V
Subsist
Farmer
CT HE
V
V V
V V
V
V
V
V
Subsist
Fisherman
CT HE
V V
V
V
J
V
V
Critical Health Effects
Central nervous system
effects
Histopathology of liver
and thymus
Liver necrosis
Liver and kidney
pathology
Increased blood pressure
Kidney damage
Skin damage: keratosis
and hyperpigmentation
Reduced body and organ
weight
Increased relative liver
weight
Liver toxicity
Gastrointestinal pathology
Developmental toxicity:
reduced fetal body weight
Altered liver and kidney
weights
Liver & kidney pathology
*** Draft -- March 21,1993 -- Do Not Cite or Quote
-------
EXHIBIT 7-19b
KEY CONSTITUENTS DRIVING CANCER RISK LEVELS*
PATHWAYS
Constituents Posing
Cancer Risk
Pentachlorophenol
Arsenic
Chlordane
(Bis)2-
(ethyihexyl)phthalate
2,3,7,8-Tetrachlorodibenzo-
p-dioxin
3,3-Dichlorobenzidine
Benzene
1,1-Dichloroethylene
Chloroform
Beryllium
Ground Water"
CT*
V
V
V
HE'
V
V
V
V
POP
V
V
V
* Key constituents at the facilities presenting the high
Notes for Exhibits 7-19a and 7-19b:
SW Swim*
CT
V
HE
V
V
Off-site/
Agric.
CT
V
V
V
V
V
Soil'
HE
V
V
V
V
V
Foodchain1
CT HE
V V
V V
V V
V V
V V
V V
V V
Off-site/
Agric. Soil
Pica Child"
CT HE
V V
V V
V
V
V V
Subsist
Fanner
CT
V
HE
*f
*f
V
V
Subsist
Fisherman
CT
V
V
V
V
HE
V
V
V
V
V
Critical Health Effects
Hemangiomas (i.e., benign
tumors of blood vessels)
Skin cancer
Liver cancer
Liver cancer
Multiple cancers
Cancer of mammary gland
Leukemia
Cancer of adrenal gland
Kidney cancer
Gross tumors
est pathway cancer risks.
1 Ingestion of ground water.
b Central tendency, individual risk.
c High-end, individual risk.
d Population risk.
' Dermal absorption from and incidental ingestion of surface water while swimming.
' Exposures to either agricultural or other off-site soils.
1 Includes ingestion of recreationally caught fish, and homegrown vegetables, meat, and dairy products.
b Pica child exposure to either agricultural or other off-site soils.
Draft - March 21, '' - Do Not Cite or Quote
**
-------
7-47
7.3.2 Risk Reduction due to Subpart S Proposed Rule
For this RIA, EPA estimated human health benefits attributable to corrective action by
subtracting post-remedial risks (i.e., risks after corrective action) from baseline risks.36 To
measure these risk reductions, the Agency selected a subset of RIA sample facilities for analysis
of post remediation risk; these facilities represent approximately 720 of the 2,600 facilities
projected to require corrective action.37 The subset of facilities selected for analysis of post-
remedial risk include about 500 facilities with individual or population risk/effects of concern in
any of the media pathways (i.e., ground water, air, soil, surface water) under central tendency
assumptions in the baseline. The remaining facilities included in the post remedial risk analysis
were selected due to significant ground water contamination beyond 2 miles from the facility
boundary. To calculate risk reductions, the Agency compared the number of facilities with off-
site individual and population risk/effects (i.e., cancer risk and noncancer effects) of concern with
and without corrective action.
As discussed in the background section of this chapter, due to the uncertainties regarding
long-term effectiveness of institutional controls and engineered remedies, the Agency chose to
analyze a "less than 100 percent effectiveness" scenario. This scenario relies on estimates of
actual effectiveness of engineered remedies and assumes that institutional controls are not
implemented. The Agency examined this scenario in order to measure the long-term cost-
effectiveness of different engineered remedies in isolation and to assess the potential risks that
could result in situations where engineered remedies and institutional controls fail.
Post-Remediation Reductions in Off-site Individual Risk,
Prediction of the post-remediation reduction in the number of facilities with individual
risk/effects of concern is uncertain and could potentially range from complete reduction to
minimal reduction. This uncertainty relates to the post-remediation levels of off-site soil and
surface water contamination, which affect not only the risk/effects from the soil and surface water
pathways, but those from the foodchain pathway as well. Since a large percentage of facilities
were projected to have individual risk/effects of concern from the soil pathway or the foodchain
pathway in the baseline, assumptions concerning the levels of post-remediation soil and surface
water contamination have a significant effect on post-remediation risk results.38
36 Note that risk reduction is determined only for the central tendency scenario.
37 This is the same subset of facilities that was examined in Chapter 4 to determine the post-
remedial extent of contamination.
38 As discussed in Chapter 4, potential reductions in the extent of off-site soil and surface
water contamination could range from complete to minimal. Considering the effects of remedies
in controlling runoff to off-site soils and of the natural attenuative processes occurring in the off-
site soils would lead to the conclusion that reductions could be nearly complete. On the other
hand, due to current limitations of the MMSOILS model, modeling results indicated minimal
*** Draft ~ March 23,1993
-------
7-48
The results presented below correspond to the "complete reduction" scenario for
individual risk/effects resulting directly or indirectly from contaminated soils and surface water.
Results for the "minimal reduction" scenario, in terms of the number of facilities presenting
risk/effects of concern, would be essentially the same as the baseline results. The primary
difference between the minimal reduction scenario and baseline results would be that under the
minimal reduction scenario, the ground water pathway would no longer present risk/effects of
concern at many of the facilities. However, these facilities would continue to present risk/effects
of concern due to the soil and foodchain pathways.
Under the "complete reduction" scenario, corrective action would reduce off-site
individual risk/effects to below levels of concern at most facilities. A summary of risk reduction
estimates (across pathways) is presented in Exhibit 7-20. Numbers of facilities with risks of
concern are subdivided into facilities with cancer risk of concern and with noncancer effects of
concern, based on both average and peak risks over the 128-year modeling period. The baseline
risk columns show the total number of facilities across all pathways with off-site risks/effects of
concern in the absence of corrective action. The remediation risk columns show the number of
facilities with risk/effects of concern if corrective action were implemented. The risk reduction
columns show the difference between baseline and remediation results (i.e., number of facilities
with risk/effects of concern after remediation subtracted from number of facilities in the
baseline).
EPA's quantitative analysis of post-remediation off-site individual risk focused on a subset
of 720 facilities which posed baseline risk/effects of concern. Of the 720 facilities, 500 pose
average or peak individual risk/effects of concern in media1 pathways under central tendency
assumptions, as shown in Exhibit 7-20. Corrective action would reduce individual risk/effects of
concern at 290 of these 500 facilities. This is a 58 percent reduction in the number of facilities
with risk/effects of concern from the baseline. Most of these reductions are due to the lowering
of risk/effects in the soil and foodchain pathways. In addition, EPA predicts corrective action to
be effective at controlling risk/effects from ground water exposures. Results are presented below
by pathway for ground water, off-site soils, foodchain pathways, and surface water.
Ground water pathway. Risk reductions in the ground water pathway are based on
MMSOILS simulation of remedy effectiveness. The effects of corrective action associated only
with ground water exposures are presented in Exhibit 7-21 for the scenario where concentrations
are not capped. Ground water risk/effects would be reduced below levels of concern at 70 of the
84 facilities with ground water risk/effects of concern in the baseline. Based on peak risks, there
would be risk reduction to below levels of concern at about 220 of the 300 facilities with peak
risks of concern in the baseline. Baseline risks and risk reductions under the cap scenario are
shown in Exhibit 7-22. If ground water concentrations are capped at MCL or taste/odor
thresholds, there would be 84 facilities that pose ground water risk/effects of concern
post-remediation reductions in off-site soil contamination. Reductions in surface water
concentrations are similarly uncertain due to the uncertain reductions in erosion contributions to
surface waters.
Draft - March 24, 1993 ***
-------
7-49
EXHIBIT 7-20
NUMBER OF FACILITIES WITH POST-REMEDIATION INDIVIDUAL OFF-SITE
CANCER AND NONCANCER RISK REDUCTION ACROSS ALL PATHWAYS
(POST-REMEDIATION SOIL PATHWAY RISKS ASSUMED TO BE ZERO)'
(n = 720)
Risk
Descriptor
Baseline Central Tendency
128-year
Average
9-year Peak
Post-Remediation
128-year
Average
9-year
Peak
Risk Reduction
128-year
Average
9-year
Peak
Facilities with Individual Cancer Risk
;> l x 10-2 to z 1
S 1 x W4 to < 1 x W2
S 1 x W to < 1 x 10"4
Total S 1 x W
3
200
290
500
3
210
290
500
0
7
200
210
0
14
200
210
3
190
89
290
3
200
90
290
FacUilies with Individual Noncancer Effects
Hazard Index £ 100
SlOto <100
21 to <10
Total with
Hazard Index S: 1
Tola! Facilities Posing
Risk/Effects of Concern
0
81
10
.- 91
~5QO
0
81
230
310
500
0
11
7
18
2AO
0
11
7
18
21O
0
70
3
73
?90
0
70
220
290
290
Subtotals may not add to totals due to rounding.
*«* Draft -- March 24,1993 ***
-------
7-50
EXHIBIT 7-21
NUMBER OF FACILITIES WITH POST-REMEDIATION INDIVIDUAL OFF-SITE
CANCER AND NONCANCER RISK REDUCTION FOR GROUND WATER*
(CONCENTRATIONS UNCAPPED SCENARIO)
(n = 720)
Risk
Descriptor
Baseline-Central
Tendency
Average
Peak
Post-Remediation
Average
Peak
Risk Reduction
Average
Peak
Facilities with Individual Cancer Risk
£ 1 x 10'J to £ 1
£lx 10" to 100
£10 to <100
Si to <10
Total with
Hazard Index £ 1
Total Facilities Posing
Risk/Effects of Concern
0
0
11
11
44
0
0
230
230
300
0
0
3
3
w .
0
-0
3
3
81
0
0
8
8
70
0
0
220
220
220
Subtotals may not add to totals due to rounding.
*** Draft -- March 24, 1993 ***
-------
7-51
EXHIBIT 7-22
NUMBER OF FACILITIES WITH POST-REMEDIATION INDIVIDUAL OFF-SITE
CANCER AND NONCANCER RISK REDUCTION FOR GROUND WATER'
(CONCENTRATIONS CAPPED SCENARIO)
(n = 720)
Risk
Descriptor
Baseline - Central
Tendency
Average
Peak
Post-remediation
Average
Peak
Risk Reduction
Average
Peak
Facilities with Individual Cancer Risk
& 1 X 10'J to £ 1
2:lx 10" to
-------
7-52
based on either the 128-year average or 9-year peak risk. Using cap scenario results, risk
reduction based on average risk will be the same as in the no cap scenario. Based on peak risk,
however, only 3 facilities will be reduced to below risk/effect levels of concern due to corrective
action (i.e., the measured benefits are less if ground water concentrations are assumed to be
capped at MCL or taste/odor thresholds).
Off-site soil pathway. Off-site soil results reflect the "complete reduction" scenario
discussed above. Under this assumption, baseline soil risk/effects would be reduced below levels
of concern at all of the 414 facilities with soil risk/effects of concern in the baseline.
Surface water pathway. Risk reductions in the surface water pathway potentially include
both reductions in risk/effects posed by consumption of contaminated fish and by incidental
ingestion of and contact with contaminated water during recreational water use. Based on the
complete reduction scenario discussed above, surface water risk/effects (from fish consumption
and swimming) would be eliminated at facilities where erosion was the dominant source of
contaminant to surface water bodies. However, at facilities where ground water discharge to
surface water was the dominant contributor, reductions would depend on the effectiveness of
source control and ground water remedies.39
Foodchain pathway. Foodchain pathway risk/effects result from consumption of
contaminated vegetables, beef and milk, and fish. The effectiveness of remedies in reducing
foodchain pathway risk/effects is related to the effectiveness of remedies in controlling
constituent concentrations in the media pathways. For example, reduction in risk/effects via fish
ingestion is dependant on the effectiveness of remedies in reducing surface water contaminant
concentrations, while reduction in risk/effects via vegetable, beef, and milk ingestion is dependant
on remedy effectiveness for the off-site soil pathway. Under the complete reduction scenario
discussed above, remedies would be completely effective in reducing risk/effects from
consumption of contaminated vegetables, beef, and milk, since soil erosion is the primary source
of constituent mass. The effectiveness of remedies in reducing risk/effects from fish consumption
would depend on the relative contributions of erosion and ground water discharge, as discussed
above for the surface water pathway.
Post-Remediation Reductions in Population Risk
EPA estimated ground-water population risk reductions for two scenarios, one where
constituent concentrations were not capped at MCLs or taste and odor thresholds, and the other
where they were capped. For the uncapped scenario, EPA estimates that there would be a
reduction of approximately 13,000 statistical cancer cases (i.e., about two thirds of baseline
cancer cases would be eliminated) and reduction of approximately 12,000,000 persons
experiencing noncancer effects (i.e., about half of baseline noncancer effects would be
eliminated). Under the capped scenario, EPA estimates that there would be a reduction of
39 The analysis of remedy effectiveness for the 10 facilities, among the subsample of 720
facilities, with baseline risk/effects of concern has not yet been completed.
*** Draft - March 24, 1993 ***
-------
7-53
approximately 400 cancer cases (i.e., about one third of baseline cancer cases would be
eliminated) and reduction of approximately 10 persons experiencing noncancer effects (i.e., all
baseline noncancer effects would be eliminated). Most of the population risk reductions are
associated with public drinking water wells. Population cancer and noncancer risk reductions are
presented in Exhibit 7-23.
Exhibit 7-24 shows how the cancer and noncancer population risks accrue over the
modeling period, both in the baseline and after remedy implementation. A greater portion of
the population risks occur towards the end of the modeling period. The exponential nature of
the cumulative population risk curves is related less to higher concentrations in the future and
more to the growth in the exposed population; the Agency assumed exposed populations to
increase at facility-specific annual growth rates for the entire modeling period. Exhibit 7-24 also
indicates that population risk reductions due to corrective action will occur mainly during the
latter half of the modeling period. Especially for noncancer population effects, remedy
implementation will not decrease exposure concentrations to below threshold levels until about
the year 2100. That is, despite remedy implementation in the year 2000 for the facility that drive
population risk results, EPA predicts people to continue being exposed to hazard indices greater
than unity until about 2100.
The reduction in population risk due to corrective action is less than complete because
EPA assumed that source control, waste treatment, and ground water remedies would generally
be less than 100 percent effective.40 This is the case at the facility that drives baseline
population risk. The expert panels specified a containment remedy for this facility, features of
which (e.g., liners) are expected to fail in the future. This facility .also has two features that
significantly limit the effectiveness of the ground water remedy: the presence of a karst aquifer
and dense non-aqueous phase liquids (DNAPLs).
40 See Chapter 4 for a discussion of remedy effectiveness assumptions.
*** Draft - March 24, 1993 ***
-------
7-54
EXHIBIT 7-23
POST-REMEDIATION POPULATION RISK REDUCTION '
(n = 720)
Risk Descriptor
BASELINE
Population Cancer
Risks1
Population Noncancer
Effects'
POST-REMEDIATION
Population Cancer
Risks1
Population Noncancer
Effects3
Exposure to
concentrations without
MCL or taste/odor caps
Public
Private
Public
Private
21,000
100
25,000,000
6,000
Public
Private
Public
Private
7J700
1
13,000,000
0
RISK REDUCTION
Population Cancer
Risks1
Population Noncancer
Effects'
Public
Private
Public
Private
13.000
99
12,000,000
6,000
Exposure to
concentrations capped at
MCL or taste/odor
thresholds
1.100
40
0
11
750
36
0
0
380
4
0
11
Notes: (1) Subtotals may not add to totals due to rounding
(2) Cancer cases are summed over the 128-year modeling period
(3) Number of people with hazard index greater than unity over the 128-year modeling period
*** Draft -- March 24, 1993 ***
-------
EXHIBIT 7-24
CUMULATIVE CANCER POPULATION RISK
25
1992
2012
2032
2052
2072
2092
2112
YEAR
Baseline Post-remediation
= CUMULATIVE NONCANCER POPULATION RISK
C/5
5
UJ
I
D
2
D
u
D
LU
LU
1992
2012
2032
2052
YEAR
2072
2092
2112
Baseline Post-remediation
Note: Concentrations not capped at MCLs or taste/odor thresholds.
-------
7-56
7.4 Limitations
Where possible, EPA quantified uncertainties by conducting bounding-type analyses (e.g.,
using the high-end and central tendency scenarios for exposure). Many of the limitations and
uncertainties of the human health benefits assessment result from data and modeling limitations,
which are described in Chapters 3 and 4. Limitations related specifically to the Agency's risk
assessment approach are discussed in this section, organized by their likely effect on risk
estimates.
7.4.1 Factors that are Likely to Overstate Human Health Benefits
Uncertainty in remedy effectiveness. EPA cannot predict with great certainty the
effectiveness of the simulated corrective action remedies. Therefore, the post-
remediation risk reductions that are estimated are also uncertain. Remedy
effectiveness is particularly difficult to predict for soil contamination, though is
less uncertain for other pathways (i.e., benefits for other pathways may not be
overstated). Although the Agency expects some soil contamination to be present
for the first few years after remedy implementation due to constituent levels
already present, these levels should be much lower for the remainder of the
modeling period as there will be no additional contribution of constituent mass to
these exposure points. EPA assumed remedies to be completely effective in
eliminating risk from off-site soil contamination, and thus, assumed that risks via
soil and soil-related pathways (e.g., vegetable ingestion) would be lowered to
levels below concern over the long run. This may overstate the human health
benefits.
Conservative assumption underlying on-site hypothetical risk estimates. Several
conservative assumptions may result in on-site hypothetical individual risks being
overestimated. First, exposures are assumed to occur at points that would, as a
rule, lead to the highest risks. This is due to the exposure points being placed in
immediate proximity to SWMUs posing the highest risks (due to factors such as
high waste concentrations or most toxic constituents), or at locations on site
where the ground water plume has maximum concentrations of constituents.
Second, concentrations of constituents in on-site contaminated soils in some cases
do not reflect fate processes, such as erosion, and therefore may be relatively
high.
Addition of cancer risk for constituents with different cancer weieht-of-evidence
classes. Constituents that were used in the cancer risk assessment (i.e., those with
slope factors) do not all have the same likelihood of being human carcinogens.
EPA has assigned different weight-of-evidence classes to describe the level of
confidence that the Agency has in the human carcinogeniciry of a particular
constituent. For example, a class A carcinogen is one that is known to definitely
cause cancer in humans, and a Class C carcinogen could possibly cause cancer in
human, but no definitive human evidence is available. Because of the varying
*** Draft - March 24,1993 ***
-------
7-57
level of confidence in a constituent's carcinogenicity, the addition of cancer risk
across constituents with different cancer weight-of-evidence classes (e.g., adding
risks across class A (known human) carcinogens and class C (possible)
carcinogens) may overstate the true cancer risk. That is, treating chemicals that
are less certain of causing cancer as known human carcinogens may overstate the
true cancer risk.
Inherent conservatism in the EPA-derived slope factors and reference doses. In
general, the cancer slope factors are based on the upper 95th percentile
confidence limit derived from the linearized multi-stage model or based on
maximum likelihood estimates. The reference dose is derived by dividing the "no-
observed-adverse-effect-lever (NOAEL) or the "lowest-observed-adverse-effect-
lever (LOAEL) by an uncertainty factor of either 10,100, or 1000. These are
conservative measures that may lead to an overstatement of the true risk.
Constituent absorption. The Agency assumed that absorption in exposed humans
was the same as that assumed in the derivation of the reference dose or cancer
slope factor from animal data. In reality, absorption in humans may be less
because a portion of the exposure dose would fail to absorb due to excretion,
biochemical breakdown, or deposition in non-target tissues.
Toxicitv to foodchain organisms not assessed . The Agency did not include toxic
effects of constituents on foodchain organisms (i.e., fish, vegetables, and cattle) in
the human foodchain exposure pathway. Foodchain risks may be overstated
because toxic effects to foodchain organisms might prevent human consumption of
foodchain organisms.
7.4.2 Factors that are Likely to Understate Human Health Benefits
Limited modeling time period. Risks are estimated only for exposures occurring
in the 128-year modeling time period. Baseline risks may be understated because
additional exposures could occur beyond the modeling period (e.g., slow-moving
constituents in groundwater could take hundreds or thousands of years to reach
down-gradient wells).
Exposure controls not included. The risk reduction achieved under the Subpart S
proposed rule is likely to be understated since the simulated remedies did not
include exposure controls. In practice, exposure controls would be implemented
as part of the remediation process and would act to eliminate exposures and risk.
High-end scenario most applicable to ground-water pathway. Data and
assumptions for the high end scenario, while often resulting in greater constituent
mass available for release to all pathways, may in some cases inadvertently act to
reduce the mass released to pathways other than ground water. Since the
MMSOILS model employs a mass balance algorithm, the increase in mass
*** Draft - March 24,1993 *»*
-------
7-58
released to ground water in the high end may act to reduce the mass available for
release to other media and may reduce the modeled extent of contamination in
those media in the high end. Consequently, the high-end results for surface
water, soils, foodchain, and air pathways that are presented in this chapter in
some cases may be understated.
Limited data on SWMU locations. Due to limited characterization data for some
facilities, the Agency may not have identified all the SWMUs with potential
releases to environmental media. This would result in risk being understated.
Very large federal facilities not examined. In selecting the sample of facilities to
be examined for the RIA, the Agency chose not to include very large federal
facilities with extensive contamination for which cleanup costs were likely to very
high.41 Exclusion of these facilities underestimates the benefits of the rule
because there would be risk reduction associated with remediation of these sites.
Limited data available on constituents present in SWMUs. The hazard
identification focused on about 100 constituents present in SWMUs of concern at
sample facilities. While these constituents are believed to include the most
abundant and toxic of constituents present, and account for the most serious risks,
there could be additional risks from constituents not included. Harmful
constituents not analyzed could include constituents for which EPA has not yet
compiled useable toxicity data (e.g., RfDs, slope factors) or constituents for which
existing toxicity data has been withdrawn from IRIS or Heast for further review.
Background constituent concentrations not estimated. In calculating risk at each
facility, EPA did not include background concentrations of constituents in the
intake estimates. Although some constituents may have been present at
background levels in the various media, the Agency only determined facility-
related risks, i.e., risks due to facility-related concentrations of constituents.
Background levels could be important for noncancer effects. Facility-related
concentrations may not be sufficient in themselves to exceed the noncancer
threshold levels, but could, as increments to background levels, drive the total
exposure concentration above the threshold. Thus, the RIA method could
understate risks.
Modeling Limited to 5 to 10 constituents per SWMU. A methodological
limitation causing risk to be slightly understated is that EPA chose to model
releases for no more than 10 constituents of concern for each SWMU. The
chosen constituents were those with high concentrations and with available toxicity
data. Because some known constituents were left out of the risk assessment in a
few cases, baseline risks and risk reduction of the rule could be understated.
41 The rationale for this exclusion discussed in Chapter 3.
Draft - March 24, 1993 ***
-------
7-59
Effect of MCLs and taste and odor thresholds on ground water population risk.
To estimate population risks via ground water under the "concentrations capped"
scenario, EPA assumed that exposures at both public and private wells will not
occur to concentrations exceeding the MCLs or taste and odor thresholds,
whichever is the lowest. This is tantamount to assuming that ground water will be
treated to MCLs, or that people would avoid consumption of the water when taste
and odor thresholds are exceeded. While ground water from public wells may be
subject to monitoring and treatment, it is unlikely that ground water from private
wells will be monitored for exceedences of MCLs or treated. Furthermore, taste
and odor problems may not prevent use of the contaminated ground water for
washing and bathing. This may understate the risk under the "concentrations
capped" scenario, particularly risk at private wells.
No assessment of cumulative risks across pathways. EPA did not combine risks
across pathways because of the uncertainty involved in assessing multiple pathway
exposure; this could lead to an underestimation of individual risk.
7.4.3 Factors that have an Indeterminate Effect on Human Health Benefits
Uncertainty in dose-response values. The dose-response assessment used Agency-
approved toxicity data such as reference doses and cancer slope factors.
Uncertainty in the calculation of such toxicity criteria is carried through into the
risk assessment. For example, toxicity criteria for some constituents are based on
low-dose extrapolation of dose-response curves, a process that carries with it some
uncertainty.
Uncertainty in fate and transport simulation. MMSOILS calculations of
constituent concentrations are estimates based on a limited number of data points.
This causes varying degrees of uncertainty in the results it generates depending on
its ability to characterize that pathway. Overall, MMSOILS is capable of more
refined modeling of releases to, and transport in, the ground water and air
pathways than in the soil and surface water pathways. In particular, MMSOILS
does not model on-site soil concentrations. As a result, risks resulting from
surface water and soil and soil-related pathways may be relatively more inaccurate
or imprecise compared to the other pathways.
Uncertainty in some exposure point locations. Exposure points for some pathways
are likely to be relatively more accurate than those for others. For example,
exposure points for the ground water pathway (i.e., ground water wells) are likely
to be relatively accurate because they are discrete points and are easily identified
and verified (EPA verified the location of public wells by contacting the local
water supply authorities, and private well locations were estimated based on
topographical maps). Exposure points for other pathways are more uncertain.
For example, surface water exposure points are the nearest known recreational
use point downstream from the point of contaminant discharge (e.g., boat
*«* Draft - March 24,1993 ***
-------
7-60
launches, swimming spots), or at an assumed point 10 meters downstream from
the discharge point. Where one recreational use was identified, EPA assumed
there to be both fishing and swimming at that point. There is similar uncertainty
for off-site soil exposure points, which are frequently identified from topographic
maps that may be dated. EPA assumes that the nearest soil exposure points can
be identified from these maps and that these maps accurately depict current land
uses, an assumption that may not necessarily be valid. Also, the Agency assumes
that soil exposures would occur at these points, and not at any others.
Uncertainty in long-term population projections and future land use. Current and
future exposed populations are based on current facility-specific data and county-
level population projections through 2015. Annual population estimates for the
modeling period (i.e., 1992-2119) assume that county-level population trends are
representative of trends in the immediate area near sample facilities. In addition,
this approach assumes that the observed demographic and economic trends in
counties with sample facilities are accurate and provide a good basis for modeling
future populations. Furthermore, population projections for areas surrounding
facilities assume constant land use characteristics such as distribution and density
of residential development. This approach may overstate or understate future
populations.
No assessment of the cumulative effect of constituents. A limitation of the risk
characterization is the assumption of no synergistic or antagonistic effects of
multiple contaminant exposures. This may cause overestimation or
underestimation of the risk.
Uncertainty in foodchain-related parameters. Predicting concentrations of
constituents in the foodchain pathway is a multi-step process involving many
parameters. There is uncertainty associated with most of the values used for
these parameters. For example, the soil/feed-to-beef transfer factors for organic
constituents are not based on chemical-specific studies, but instead, are calculated
based on a regression equation using chemical properties (e.g., chemical-specific
octanol-water partition coefficient). Another example is the use of standard
national assumptions for the food ingestion rate of cattle. These assumptions may
either understate or overstate the concentration of the contaminant in beef, and
thus the risk due to consumption of beef and milk. Note, however, that reduction
in concentrations due to processing/preparing produce (e.g., washing, peeling) was
not taken into account, and this is likely to overstate the risk.
Limited data on constituent concentrations in SWMUs. In some cases where
facility investigations or other reports indicated the presence of contaminants but
did not include concentration estimates, the Agency assumed that these
constituents were present at action level concentration levels. These assumed
concentration levels could overstate or understate actual concentration levels and,
thus, benefits.
*** Draft - March 24, 1993
***
-------
7-61
Assumption of proposed Subnart S CAMU. For the purposes of this RIA, the
Agency assumed that the proposed Subpart S CAMU could be employed, where
appropriate, at corrective action facilities. The recently-published CAMU final
rule provides additional flexibility in designing remedies, which could result in
some variation in the remedies selected and in resultant benefits.42
42 See Chapter 1 for a discussion of the CAMU final rule and Chapter 2 for a discussion of
the proposed Subpart S CAMU.
*** Draft - March 24,1993 ***
-------
8. ECOLOGICAL THREATS
8.1 Summary of Approach
Ecological assessments were conducted for the stratified random sample of RCRA
facilities subject to Subpart S corrective action in order to characterize ecological threats under
baseline conditions (i.e., in the absence of the proposed Corrective Action Rule). There are a
number of approaches, applicable to a range of situations encountered in EPA's programs, that
could be used to characterize ecological threats for a national sample of waste-management
facilities.1-"-*5-6-7
The ecological risk analysis for this RIA borrowed from all of these methods and a
combination of approaches was used, including a proximity analysis, a concentration-based
screening analysis, and case-study examples. The combined approach was selected to maximize
1 U.S. Environmental Protection Agency (EPA). Hazard Ranking System: Final Rule. 40
CFR Part 300. Federal Register. December 14, 1990.
2 U.S. Environmental Protection Agency. Standard Evaluation Procedure for Ecological
Risk Assessment. Washington, D.C.: Office of Pesticide Programs. Prepared by Urban, D. J.,
and Cook, N. J., Hazard Evaluation Division. 1986.
3 U.S. Environmental Protection Agency. Permit Writer's Guide to Water Quality-Based
Permitting for Toxic Pollutants. Washington, D.C.: Office of Water Regulations and Standards,
1987. EPA Report No. EPA 440/4-87-005.
4 U.S. Environmental Protection Agency. A Tiered Modeling Approach for Assessing the
Risks Due to Sources of Hazardous Air Pollutants. Research Triangle Park, NC: Office of Air
Quality Planning and Standards, March 1992. EPA Report No. EPA/450/4-92/001. (Although
these guidelines refer to human health risk assessments, a similar document and approach is
currently under development for ecological receptors.)
5 U.S. Environmental Protection Agency. Risk Assessment Guidance for Superfund.
Volume II: Environmental Evaluation Manual. Interim Final. Washington, D.C.: Office of
Emergency and Remedial Response, March 1989. EPA Report. No. EPA/540/1-89/001.
6 U.S. Environmental Protection Agency. Developing a Work Scope for Ecological
Assessments. ECO Update, Intermittent Bulletin, Volume 1, Number 4. Washington, D.C.:
Office of Emergency and Remedial Response, Hazardous Site Evaluation Division, 1992.
Publication 9345.0-051.
7 U.S. Environmental Protection Agency. Ecological Assessment of Superfund Sites: An
Overview. ECO Update, Intermittent Bulletin, Volume 1, Number 2. Washington, D.C.: Office
of Emergency and Remedial Response, Hazardous Site Evaluation Division, 1991. Publication
9345.0-051.
»** DRAFT - March 24, 1993 ***
-------
8-2
use of available data, provide information useful for discriminating relative risk, and allow for
extrapolations from the sample set of facilities to the universe of facilities that may be subject to
corrective action.
The ecological assessments were performed for the 52 facilities from the sample of 79
facilities that triggered corrective action and included three steps. The first step was a proximity
analysis to determine which of the facilities were near valuable or vulnerable ecological
resources, particularly surface waters (e.g., lakes, rivers). The second step was a concentration-
based screening analysis to identify facilities at which releases to surface waters might result in
ambient concentrations of hazardous substances that exceed ecological benchmark levels. The
final step was a more complete ecological risk assessment for three case-studies to identify a
broader spectrum of ecological threats. The approach used in each of the three steps is
summarized in sections 8.1.1 through 8.13, respectively. Details of the methods and results are
presented in Appendix F.
8.1.1 Proximity Analysis
The proximity analysis addressed the potential for hazardous substances released from
facilities to reach valuable or vulnerable ecological resources. The purposes of the proximity
analysis were: (1) to use these analyses to estimate how many RCRA facilities would have
releases that could reach valuable natural resources, and (2) to prioritize sites for further
evaluation in the case-study analysis.
To characterize the ecological value and vulnerability of resources near the 52 sample
facilities, the surrounding land cover was classified in five categories:
Aquatic environments (surface water, wetlands);
Terrestrial habitats (meadow, woodland, forest/field, other natural vegetative
cover types);
Agricultural (cropland, agricultural);
Residential (residential, school); and
Industrial/other (industrial, commercial, bare soil or dirt roads, hard-surface roads,
other paved surfaces).
Aquatic environments and terrestrial habitats were considered ecologically valuable (e.g., for
wildlife and natural biodiversity); the remaining categories were considered less ecologically
valuable. Surface waters were considered most vulnerable because releases to aquatic ecosystems
can spread rapidly over long distances, exposing all components of the ecosystem. Terrestrial
ecosystems were considered less vulnerable, because releases to land tend to remain confined and
expose a fraction of the ecosystem.
A key factor influencing the likelihood of resource contamination is the distance of the
resource to a point of contaminant release. The more distant a resource from a hazardous waste
facility, the less likely it is that contaminants, if released, could reach the resource in toxic
concentrations because of dispersion, degradation, dilution, and adsorption (e.g., to soils).
*** DRAFT - March 24, 1993 ***
-------
8-3
Environments located within one mile of a facility boundary were considered more likely to be
contaminated than environments located further away.
The proximity analysis thus consisted of an evaluation of the land covers on-site and
within a one-mile radius around the facility boundary. Land use and size of facility information
were estimated from the most recent available U.S. Geological Survey (USGS) topographic
maps, county survey maps, and other site information as available. The areas of the one-mile
buffers were estimated based on the size of each facility by assuming the facility is circular and
calculating the area of a "doughnut" with the inner ring at the facility boundary and the outer
ring at a distance of one mile plus the radius of the facility. These data were used to calculate
the combined total acreage of each of the five land-use categories on-site and within one mile of
each facility. Because larger facilities had a larger area within the one-mile radius, larger
facilities had the potential to have larger total acreages of terrestrial and aquatic wildlife habitats
within a one-mile radius.
The distances to the nearest representative of each of the land cover types within one
mile of the facility also were determined. Several types of "sensitive environments" identified in
EPA's Hazard Ranking System (HRS)8 located on-site or within one mile of each facility also
were identified (i.e., wetlands and other federally- or state-protected areas such as national or
state parks and forests, wild and scenic rivers, national estuaries, etc.). The analysis also included
the number of sensitive environments and the distance to the nearest representative of each
sensitive environment on-site or within one mile of each facility. The areal extent of sensitive
environments was not determined.
The stratified random sample of 79 facilities was divided into three relative-threat
categories. The 27 facilities considered unlikely to trigger corrective action were classified as no-
threat. The remaining 52 facilities were divided into two groups higher-threat and lower-
threat - on the basis of environmental setting, emphasizing the total acreage of surface waters
on-site and within one mile.9 Results from the RIA sample then were extrapolated to the
universe of RCRA facilities to estimate the total number of potentially higher-threat facilities in
the RCRA corrective action universe. The total number of acres of surface waters and terrestrial
habitats expected to be on-site and within one mile of the higher-threat and lower-threat facilities
was estimated nationally.
8.1.2 Concentration-based Screening Analysis
The concentration-based screening analysis addressed the intensity and extent of surface
water contamination that might occur from releases at the 52 sample facilities. The analysis
focused on baseline impacts; that is whether adverse ecological effects are likely if corrective
actions were not taken to terminate the releases. This information was used to determine the
8 For scoring sites for possible listing on the National Priorities List (Superfund).
9 Additional details on this procedure are provided with the results in section 8.2.1 and in
Appendix F, section F.I.
*** DRAFT - March 24, 1993 ***
-------
8-4
relative ecological threat associated with each facility. The overall approach used in this analysis
was derived from the approach used to estimate human health risks for this RIA except that
chemical-specific ecological benchmark levels were used instead of human health criteria.
At each facility, solid waste management units (SWMUs) and contaminants of greatest
concern were identified on the basis of volume, concentration, human health toxicity, potential
for release, and professional judgment. Screening ecological benchmark levels for surface water
were derived in order to: (1) identify threshold concentrations above which adverse impacts to
aquatic communities are likely, and (2) provide a reasonably uniform basis for evaluating the
relative ecological threats associated with releases of contaminants to surface water. The starting
point was EPA's chronic ambient water quality criteria (AWQC), which are intended to protect
aquatic communities by protecting 95 percent of the species in an aquatic ecosystem. When a
chronic AWQC was unavailable, benchmark levels were derived from toxicity data in the
literature as described in Appendix F. Using this approach, a benchmark level was derived for
163 (73 percent) of the 220 hazardous substances identified as a contaminant of concern at one
or more facilities in the RIA sample. When possible, bioconcentration factors (BCFs) also were
derived from the literature.10
The MMSOILS model was used to simulate releases of contaminants from SWMUs and
to predict concentrations of each contaminant of concern at the point of discharge from ground
water (or entry of surface runoff) to the nearest surface-water body over a 200-year period (1920
to 2119).11 Exposures were evaluated over a 128 year period, from 1992 to 2119.
For each constituent of concern at each facility, the MMSOILS model was run under two
scenarios: (1) the central tendency scenario representing the most likely releases and magnitudes
of releases, and (2) the high-end scenario representing the most likely worst-case releases. In
most cases, the high-end scenario resulted in higher aqueous concentrations. However, under
certain circumstances, the high-end scenario resulted in lower aqueous concentrations. This issue
is reviewed and discussed in detail in Chapter 3, Section 3.1.2.
At each facility, the maximum estimated aqueous concentration of each contaminant was
compared to its ecological benchmark level using the hazard index approach (i.e., dividing the
estimated environmental concentration by the ecological benchmark level).12 MMSOILS
provides one concentration estimate for each year. The averaging time appropriate for
10 For a more detailed discussion of the derivation of benchmarks for the protection of
aquatic life and data sources for toxicity values and bioconcentration factors, see Appendix F,
section F.2.
11 The MMSOILS model - including two scenarios, central tendency and high-end - is
described in detail in Chapter 3 and Appendices B and G.
12 U.S. Environmental Protection Agency. Review of Ecological Risk Assessment Methods.
Washington, D.C.: Office of Policy, Planning and Evaluation, November 1988. EPA Report No.
EPA/230/10-88/041.
*** DRAFT » March 24, 1993 ***
-------
8-5
comparison with EPA's chronic ambient water quality criteria is four days," which is
substantially less than one year. In other words, adverse sublethal chronic effects might be
expected in an aquatic community following as few as four days of continuous exposure. It was
appropriate, therefore, to compare single annual surface water concentration predicted by
MMSOILS with the chronic ambient water quality benchmarks. If the maximum predicted
annual water concentration exceeded a chronic benchmark, adverse effects were considered likely
for at least one year.
Exposures of organisms to more than one toxic substance at the same time can result in
adverse effects at lower concentrations of each substance than if the organisms were exposed to
only one substance.14 To account for this possibility, a hazard index was calculated for each
facility by summing the hazard indices for each contaminant (an approach which assumes
additive effects of individual hazardous constituents when several constituents are released
together).15 In other words, a hazard index is equal to the sum of the ratios of estimated
environmental concentrations to ecological benchmarks for individual substances. If the hazard
index exceeded 1.0, adverse effects were considered likely.
For the facilities at which the hazard index exceeded one, the bioaccumulation potential
(based on the bioconcentration factor) of constituents that exceeded benchmark levels was noted.
In addition, the time course of exceedances was examined by plotting the ratio of hazardous
substance concentrations to ecological benchmark levels over the 128-year exposure modeling
period.
For facilities at which projected surface water concentrations were above ecological
benchmark levels, the extent of contamination above ecological benchmark levels (i.e., the
distance to the downstream point at which the hazard quotient would equal 1.0) was estimated
for organic constituents using a simple, steady-state decay equation in which the decline in
contaminant concentrations over distance is determined solely by volatilization and degradation
(i.e. it does not account for dilution occurring to the surface water body due to downstream
runoff, tributary, and ground water inputs nor does it account for settling, or biological uptake of
contaminants).16 No other contaminant loss mechanisms could be modeled with this simple
approach.17 Thus, only organic substances to which these decay processes might apply were
13 U.S. Environmental Protection Agency. Guidelines for Deriving Numerical Criteria for
the Protection of Aquatic Organisms and Their Uses. Washington, D.C.: Office of Water
Regulations and Standards, 1986.
14 U.S. Environmental Protection Agency. Guidelines for the Health Risk Assessment of
Chemical Mixtures. 51 Federal Register 34014, September 24, 1986.
15 Ibid.
16 See Appendix F for a description of the equation.
17 See Appendix F, section F3.
*** DRAFT -March 24,1993
-------
8-6
modeled. Metals, which generally do volatilize or "decay", therefore were not modeled. A more
complex model that includes dilution, settling, and biological uptake is needed to determine
extent of contamination for organic compounds and metals with reasonable confidence. Because
the steady-state degradation equation often produced unrealistic results, we capped the estimated
extent at a given facility at twelve miles (the average length of a fourth order stream18).19 The
limitations of the steady-state decay equation suggest that caution must be taken in interpreting
the results of the extent of surface water contamination.
8.1.3 Ecological Risk Assessment Case Studies
Qualitative case studies were performed for three facilities to evaluate additional
exposure pathways that are difficult to model generically. These case studies, which consisted
primarily of more complete and comprehensive proximity and concentration-based analyses,
followed the four-step framework for ecological risk assessment recently developed by EPA's
Risk Assessment Forum.20 Problem Formulation consisted of a qualitative evaluation of
potential contaminant releases, migration, and fate in order to identify contaminants, exposure
pathways, and ecological receptors of concern and select assessment and measurement endpoints.
Ecological Effects Assessment consisted primarily of identifying media- or exposure pathway-
specific ecological benchmark levels for soils, surface water, or sediments. Exposure Assessment
consisted of summarizing existing data on measured exposure point concentrations in various
media of concern. Risk Characterization consisted of evaluating the potential for current and
future adverse effects, and the extent and ecological significance of such effects. Specific
approaches varied among each case study and discussions outlining the conclusions of the risk
assessment are summarized in Appendix F. The case study results were not extrapolated to the
universe of RCRA facilities because the facilities evaluated in the case studies were not a
random sample of all facilities. However, the results of this analysis will provide qualitative
information on the types of ecological risks that may occur at facilities subject to corrective
action that could be missed by our more generic analyses.
8.2 Results
8.2.1 Proximity Analysis
EPA determined that of the 52 facilities likely to trigger corrective action: 38 could be
characterized as higher-threat facilities (most with greater than 100 acres of surface waters at
18 Keup, L.E. Flowing water resources. Water Resources Bulletin. Volume 21,1985, pp.
291-296.
19 Most of the receiving surface water bodies in the sample of RIA facilities were 4th order
streams or smaller.
20 U.S. Environmental Protection Agency. Framework for Ecological Risk Assessment.
Washington, D.C.: Risk Assessment Forum, 1992. EPA Report No. EPA/630/R-92/001.
***
DRAFT - March 24,1993 ***
-------
8-7
each), and 14 as lower-threat facilities (most with less than 100 acres of surface waters).21 There
were two facilities with less than 100 acres of surface water nearby that were categorized as
higher-threat because they were located in areas consisting largely (greater than two thirds) of
natural terrestrial ecosystems. One facility with more than 100 acres of surface waters nearby was
categorized as lower-threat because it was located in a highly industrialized area (greater than
two thirds of the surrounding land use). Exhibit 8-1 summarizes the estimated total acreage of
surface water and terrestrial ecosystems on-site and within one mile for the higher- and lower-
threat facilities. The data indicate that the higher-threat facilities with the larger acreages of
surface waters on-site and within one mile tended to be larger facilities, as expected.
Exhibit 8-2 extrapolates the proximity analysis results to the universe of sites subject to
the proposed corrective action rule based on the weighting factors for each facility category type.
Based on the sample results approximately 1,530 (59 percent) of the 2,570 facilities, likely to
require corrective action, are in the higher-threat category. This represents 26 percent of the
approximately 5,780 facilities subject to the proposed corrective action rule. For all facilities, as
many as one million acres of surface waters and 3.6 million acres of terrestrial habitats nationally
are near RCRA facilities subject to corrective action and maybe threatened by potential releases
of hazardous substances.
8.2.2 Concentration-based Screening Analysis
Exhibits 8-3 and 8-4 present the results of the MMSOILS modeling runs comparing
predicted surface water aqueous concentrations to ecological benchmark levels. Assuming that
the environmental settings for the stratified random sample are representative of the
environmental settings of the universe of RCRA facilities nationwide, Exhibit 8-3 illustrates that
under central tendency fate and transport and waste characteristic assumptions, releases of
hazardous substances to surface waters are predicted to exceed ecological benchmark levels at
approximately 150 (6 percent) of the 2,570 facilities likely to require corrective action. Under
high-end assumptions, releases of hazardous substances to surface waters are likely to exceed
ecological benchmark levels at 570 (23 percent) of the 2,570 facilities likely to require corrective"
action (see Exhibit 8-4).
" For detailed results, see Appendix F, Exhibit F-2.
*** DRAFT -- March 24,1993 ***
-------
8-8
EXHIBIT 8-1
UNWEIGHTED
DISTRIBUTION OF TOTAL ACREAGE OF SENSITIVE ENVIRONMENTS
ON-SITE AND WITHIN ONE MILE
OF LOWER AND HIGHER-THREAT FACILITIES
LIKELY TO TRIGGER CORRECTIVE ACTION
Total Acres
15,000 - 20,000
10,000 - 14,999
5,000 - 9,999
2,500 - 4,999
1,000 - 2,499
500 - 999
100 - 499
< 100
0
Total Facilities
Number of Facilities
Surface Waters
Higher-
threat
-
-
2
4
6
8
16
2
-
38
Lower-
threat
-
-
-
.
.
-
1
10
3
14
Terrestrial Ecosystem
Higher-
threat
2
-
-
9
15
6
6
.
-
38
Lower-
threat
-
-
-
2
2 ,
1
5
4
-
14
Total Facility Size
Higher-
threat
2
5
13
18
-
-
-
-
NA
38
Lower-
threat
-
-
2
9
3
-
-
-
NA
14
"-" indicates zero (0) facilities were in this category.
NA indicates not applicable.
*** DRAFT -- March 24, 1993 ***
-------
8-9
EXHIBIT 8-2
ESTIMATED
NUMBER OF HIGHER- AND LOWER-THREAT FACILITIES
IN RCRA CORRECTIVE ACTION UNIVERSE
Type of Facility
Relative-Threat Category
Higher-Threat
Lower-Threat
No-Threat
Total*'
Facilities
Non-Federal Facilities
Large
Not Large;
RFA Completed
Not Large;
RFA Not Completed
70
760
640
20
380
640
0
570
2,350
90
1,710
3,630
Federal Facilities
Large DOD/DOE
Facilities
Other Federal Facilities
7
55
0
0
15
270
22
330
TOTAL*'
1,530
1,040
3,210
5,780
- Totals may not be exact due to rounding.
*** DRAFT -- March 24, 1993 **
-------
8-10
EXHIBIT 8-3
ESTIMATED NUMBER OF FACILITIES IN THE RCRA UNIVERSE
AT WHICH MAXIMUM PREDICTED CONCENTRATIONS
EXCEEDED ECOLOGICAL BENCHMARK LEVELS
UNDER CENTRAL TENDENCY ASSUMPTIONS
Type of Facility
Relative-Benchmark Exceedance Category
Exceeded
Benchmark
Did Not
Exceed
Did Not
Trigger
Total^
Facilities
Non-Federal Facilities
Large
Not Large;
RFA Completed
Not Large;
RFA Not Completed
20
130
0
70
1,010
1,280
0
570
2,350
90
1,710
3,630
Federal Facilities
Large DOD/DOE
Facilities
Other Federal Facilities
0
0
7
55
15
270
22
330
TOTALS'
150
2,420
3,210
5,780
- Totals may not be exact due to rounding.
*** DRAFT » March 24, 1993 »**
-------
8-11
EXHIBIT 8-4
ESTIMATED NUMBER OF FACILITIES IN THE RCRA UNIVERSE
AT WHICH MAXIMUM PREDICTED CONCENTRATIONS
EXCEEDED ECOLOGICAL BENCHMARK LEVELS
UNDER HIGH-END ASSUMPTIONS
Type of Facility
Relative-Benchmark Exceedance Category
Exceeded
Benchmark
Did Not
Exceed
Did Not
Trigger
Total*'
Facilities
Non-Federal Facilities
Large
Not Large;
RFA Completed
Not Large;
RFA Not Completed
40
320
210
50
820
1,070
0
570
2,350
90
1,710
3,630
Federal Facilities
Large DOD/DOE
Facilities
Other Federal Facilities
7
0
0
55
15
270
22
330
-
TOTALi'
570
2,000
3,210
5,780
- Totals may not be exact due to rounding.
*** DRAFT - March 24, 1993 ***
-------
8-12
For each facility at which projected environmental concentrations exceeded ecological
benchmark levels, the time course of exceedances was examined.22 At more than three quarters
of the facilities where the maximum concentration of at least one substance was predicted to
exceed an ecological benchmark level, the concentration of at least one substance was predicted
to remain above its benchmark level for the entire 128-year exposure period, primarily under
high-end modeling assumptions. The minimum predicted duration of exceedance was 20 years.
Thus, impairment of the aquatic communities would be expected to continue without chance for
recovery for decades in the absence of corrective action at these facilities.
For each sample facility at which projected environmental concentrations exceeded
ecological benchmark levels, the length of river or stream over which the benchmark was
exceeded was estimated. The results indicate that at several facilities, contaminants might exceed
ecological benchmarks for several miles.23 Such exceedances could effectively fragment a river
or stream ecosystem, preventing successful migration of the more mobile species up-stream and
down-stream.24 Less mobile species could be divided into smaller populations on one or both
sides of the stream section, increasing the likelihood that the species would be eliminated from
the stream or river. Streams fragmented with physical barriers have exhibited reduced diversity
of fish species compared with species diversity prior to fragmentation.25
The results of this analysis suggest that for most facilities with a predicted exceedance,
multiple hazardous substances were predicted to exceed ecological benchmark levels for the
entire 128-year exposure period considered. Although bioaccumulation potential tended to be
relatively low at most facilities, predicted releases frequently involved metals and other persistent
hazardous substances and could result in predicted concentrations two to four orders of
magnitude above benchmark levels (substances considered singly or in combination).26
As expected, exceedance of water quality benchmarks for the protection of aquatic life
are more likely under high-end modeling than under central tendency modeling, because the
high-end release predictions generally involved larger quantities of substances and higher
predicted concentrations and sometimes involved a greater duration of predicted releases and
exceedances. Exceedances of water quality benchmarks are most likely for predicted releases to
22 See Appendix F, section F.6 and Exhibit F-4 for additional details of this analysis.
23 See Appendix F, Exhibit F-3.
24 Fish species might avoid chemically contaminated reaches or suffer acute adverse effects if
contaminant concentrations are sufficiently high.
25 Winston, M. R., Taylor, C. M, and Pigg, J. "Upstream extirpation of four minnow species
due to damming of a prairie stream." Transactions of the American Fisheries Society. Volume
120 (1991), pp. 98-105.
26 See Appendix F, Exhibit F-3.
**» DRAFT » March 24, 1993 ***
-------
8-13
surface waters with low or moderately low flow rates (annual flow less than 10 or 100 cubic feet
per second).
8.2.3 Comparison of Proximity and Concentration-based Screening Analyses
The proximity analysis evaluated the potential for contaminants to reach valuable or
vulnerable resources, whereas the concentration-based analysis evaluated the intensity and extent
of contamination. Of the 19 sample facilities at which concentrations were predicted to exceed
ecological benchmark levels, 16 were in the higher-threat category determined by the proximity
analysis alone, while three were in the lower-threat category determined by the proximity
analysis. Releases greater than ecological benchmark levels were predicted to occur at 43
percent of the 38 higher-threat facilities as opposed to 21 percent of the 14 lower-threat facilities
(i.e., releases to surface waters at levels greater than ecological benchmark levels were
approximately twice as likely at facilities in the higher-threat group than at facilities in the lower-
threat group). Thus, relative proximity of facilities to surface waters appears to be an indicator
of relative threat of contamination at hazardous levels. However, the surface water pathway was
the only pathway considered in the screening concentration-based analysis and was the primary
pathway considered in the proximity analysis.
8.2.4 Case Studies
Case studies were prepared for three facilities (A, B and C) to identify types of ecological
threats that are not accounted for by the proximity and screening concentration-based analyses.
Facility A is a large facility that was identified as higher-threat by the proximity analysis, but was
not identified as a higher-threat facility by the concentration-based screening analysis. Facility B
(a large facility) and Facility C (a not-large facility for which an RFA has been completed) were
identified as higher-threat facilities by the concentration-based screening analysis as well as by
the proximity analysis. These three facilities also had relatively extensive site-specific data to
assist in the analysis. Each is described separately below.27
The case studies demonstrated that sediment and soil pathways, as well as the surface
water pathway, can be important factors in determining risk to ecological systems at facilities
subject to the RCRA corrective action rule. Soil contamination was an important factor that
contributed to ecological risk for terrestrial plants and animals at one facility. However, that
facility had not been identified by the concentration-based screening analysis as having an
exceedance, but the proximity analysis and professional judgement had identified it as potentially
at ecological risk. At a second facility contamination was the main ecological threat, and surface
water and aquatic food chain contamination were the primary ecological threats at a third
facility. While the proximity analysis and the concentration-based screening analysis helped to
characterize ecological threats based on the surface water pathway, other means are necessary to
more fully characterize other potentially significant exposure pathways such as aquatic sediments
and food chains and terrestrial soils.
27 More complete summaries of the case study results, approaches, and methodologies are
described in Appendix F.
*» DRAFT - March 24, 1993 ***
-------
8-14
83 Discussion
This analysis suggests that approximately 60 percent of all facilities likely to require
corrective action under the proposed rule are likely to be located relatively near vulnerable
ecosystems such as wetlands, surface water bodies, forests or other terrestrial habitats, or
sensitive environments. As many as one million acres of surface waters and 3.6 million acres of
terrestrial areas may be sufficiently close to these facilities to be at risk from potential releases of
hazardous substances. Modeling results suggest that under baseline conditions, somewhere
between 5 and 25 percent of all facilities subject to RCRA corrective action would be expected
to release hazardous substances to nearby surface waters at levels likely to result in adverse
ecological impacts. Moreover, releases above surface water ecological benchmark levels are
approximately twice as likely at facilities close to surface waters.
Modeling results also suggest that releases of hazardous substances are likely to involve
multiple substances above individual benchmark levels, remain above benchmark levels for many
decades, and remain above benchmark levels for some distance downstream of the point of
release. The case-study examples suggest further that releases may result in contamination of
sediments of aquatic ecosystems and may threaten terrestrial species, including soil-dwelling
invertebrates, small birds and mammals, and fish-eating birds.
8.4 Limitations
Although the overall uncertainty associated with this analysis is difficult to quantify, there
is reason to believe that the analysis has, overall, underestimated baseline ecological risk at the
universe of facilities likely to require corrective action under the proposed rule. Resource
constraints limited the types of approaches that could be taken, the number of exposure
pathways and scenarios that could be evaluated within each approach, and the scope of the
analyses that could be performed. Many of the limitations and uncertainties of the ecological
risk analyses are associated with data availability and modeling assumptions.28 Limitations
related to the ecological risk analyses are discussed in this section, organized by their likely effect
on risk estimates.
8.4.1 Factors That Are Likely to Understate Results
Just one medium of contamination, surface water, was the focus of the proximity
and screening concentration-based analyses. By focusing on surface water
contamination, the analyses evaluated only some of the ways in which releases of
hazardous substances might result in adverse ecological impacts. Other
contaminated media, particularly sediments and soils, are likely to contribute to
ecological risk at the facilities subject to corrective action.
A limited number of environment types were considered. The proximity analysis
focused on a limited number of environment types that could be identified and
28 These limitations are described in Chapters 3 and 4 and in the Appendices.
*** DRAFT - March 24, 1993 ***
-------
8-15
quantified (e.g., those found on U.S.G.S. topographic maps). Wetlands were not
distinguished from surface waters when quantifying land use patterns, and only a
small fraction of the types of areas identified under CERCLA as "sensitive
environments" were addressed.
Not all means for contamination to reach surface waters were modeled. The
analysis was limited to a few of the potential means in which hazardous substances
could reach surface waters (i.e., discharge of contaminated ground water, runoff
from and erosion of contaminated soils). Other potentially significant routes that
were not evaluated included atmospheric deposition to surface waters and direct,
large-scale discharge via flooding of SWMUs or catastrophic failure of
containment systems.
Intermittent effects were not addressed. Reports of fish kills at some facilities in
the RIA sample suggest that intermittent, large-scale releases can result in
significant adverse ecological effects.
Only a limited number of more thorough analyses were conducted. The case
studies more thoroughly evaluated exposure through air, soil, and sediment
pathways, but only three case-study examples were evaluated.
Food chain and sediment pathways could not be addressed thoroughly. Within
surface water systems, the screening concentration-based analysis was limited to
predicted water-column concentrations; risks from exposure to potentially
contaminated sediments or food chain effects were-not evaluated. In surface
waters, many hazardous substances with the greatest potential for significant, long-
term adverse ecological impacts (e.g., most metals, persistent organic compounds
with a high bioaccumulation potential) exhibit low solubility in water, may adsorb
strongly to soil or sediment particles, and are not readily leached from soils.
These substances, although present in SWMUs and releases, often were not
predicted to exceed ecological benchmark levels in the water-column
concentration-based analysis, although their potential risks were evaluated in some
of the case studies. Metals and persistent organic compounds with a high
bioaccumulation potential, when released to surface waters, tend to partition into
sediments, rapidly enter aquatic food chains, and result in prolonged adverse
impacts in aquatic and terrestrial organisms. Potential impacts via food chain
exposures (e.g., to fish-eating wildlife) were addressed only briefly.
For a few SWMUs. constituents were selected based on human health concerns.
For the few SWMUs that contained more than 10 hazardous constituents, some
constituents were eliminated from further evaluation on the basis of low toxicity
to humans without considering ecological toxicity. The likelihood of this selection
screening out ecologically significant contaminants is small, however.
*** DRAFT - March 24, 1993 ***
-------
8-16
8.4.2 Factors That Are Likely to Overstate Results
Ground water influx to surface waters may he overstated for some facilities.
Constituent concentrations in surface waters may be overstated at some facilities
where ground water is an important source of constituent mass to surface water
and conservative ground-water transport assumptions were employed (e.g., at
facilities located over karst aquifers).
Factors other than releases from RCRA facilities may be impairing nearby surface
waters. For the analysis of baseline ecological threats, EPA has assumed that
potential releases from RCRA facilities are the only chemical stressors that may
be impairing nearby aquatic ecosystems. As the case-study analysis indicated,
however, surface waters in the vicinity of RCRA facilities may be contaminated
above ecological benchmark levels by substances from other point and non-point
sources. Thus, some of the RCRA facilities are likely to be responsible for only a
fraction of the contamination existing in nearby surface waters and may only
threaten already impaired ecosystems.
8.4.3 Factors That Have an Indeterminate Effect on Results
Simplified methods were used. Relatively simple methods were used to derive or
identify land use information, ecological benchmark levels, BCFs, and degradation
coefficient (X) values. Data sources were limited to readily available compilations
of toxicity and other parameter values. Generic extrapolation factors were used
to derive benchmark values, rather than quantitative structure-activity
relationships or other more chemical class-specific approaches. Benchmark values
derived using more chemical-specific data may differ from those used in this
analysis by one or more orders of magnitude. Site-specific assessments that might
evaluate all pathways (e.g., sediments) could be performed for only three sites.
The risk characterization method used in the concentration-based analysis and
case-study examples (i.e., a simple hazard quotient/hazard index approach) made
it difficult to go beyond a simple yes/no in characterizing the potential for adverse
ecological impacts. The overall uncertainty associated with the analyses cannot be
quantified.
The ecological benchmarks may be over- or under-protective. Although it is
EPA's intent that the ecological benchmarks represent environmental
concentrations or wildlife intake levels below which no significant adverse effects
are expected, there are many uncertainties associated with extrapolating from one
end point to another (e.g., acute to chronic toxicity) and from one species to
another. For some hazardous substances and exposure pathways, the ecological
benchmarks may be too conservative, and for others, not protective enough.
Bioconcentration and biomagnification depend on site characteristics (e.g., actual
food chain, water conditions, species of concern) as well as on properties of the
chemical. In general, EPA has tried to be conservative, and hence may be
**
DRAFT « March 24, 1993 ***
-------
8-17
overstating threats, but threats also may be understated for some chemicals and
exposure pathways.
*** DRAFT - March 24, 1993 »**
-------
9. AVERTED WATER USE COSTS
This chapter of the RIA addresses the water use costs that may be averted by corrective
action requirements. The required facility remediation will reduce the future costs of water use
in cases where corrective action prevents or decreases the contamination of drinking water
supplies. In the absence of corrective action, water users would need to treat or replace the
contaminated supplies. If related expenditures are averted or mitigated by corrective action, the
savings can be used to estimate some of the benefits of the regulations.
The purpose of this chapter is primarily to present the analytic approach used to assess
averted water use costs. The analysis conducted to-date and reported in this draft chapter is
incomplete. The findings discussed below focus on the area within two miles of each facility;
subsequent analysis will address the area extending two to five miles of each facility.
The analysis of the area within two miles of each facility indicates that:
Of the approximately 5,800 hazardous waste facilities subject to corrective action
requirements (2,600 of which will require remediation), ground-water use could be
affected in the absence of corrective action at about 360 facilities.1 These
facilities are about 14 percent of the facilities requiring remediation.
In the absence of remediation, the present value of water treatment and
replacement costs attributable to contamination from 1992 through 2119 will total
approximately $240 million. Compared to a situation where there was no ground-
water contamination, these costs lead to a loss in Consumers' surplus of $230
million.
Corrective action averts only some of these costs because: (1) several years may
elapse between 1992 and the time when remedies become effective; and (2)
corrective measures are only partially effective in some cases. In particular,
elimination of contamination existing prior to corrective action is often difficult.
Corrective action averts approximately $4.8 million (two percent) of the increase
in water use costs resulting from contamination. The benefits of corrective action
resulting from these averted costs (i.e., the avoided losses in consumers' surplus)
total $4.7 million.
1 This analysis addressed contamination from solid waste management units regulated under
Subpart S only, and focuses on the effects of the proposed corrective action regulations. See
Chapters 1, 2 and 3 for more information on this and related issues.
*** DRAFT March 22,1993 ***
-------
9-2
As noted earlier, the analysis presented in this draft considers the area within two miles
of each facility. Review of available data suggests that about 80 facilities nationally may have
wells located two to. five miles from the facility which are affected by contamination in the
absence of corrective action.2 Subsequent assessment of these additional areas will increase the
estimates of benefits.
The analysis of averted costs focuses on how much people would pay if they were aware
of the contamination and took action to avoid health risks. In reality, some people may choose
to accept the risks associated with using contaminated water, and thus the analysis in this chapter
will overstate actual willingness to pay for averting actions. In addition, contamination may not
always be detected. In both of these cases, information on the resulting health risks is needed to
fully assess benefits related to water use. Chapter 6 discusses the relationship between the
averted cost and health risk analysis in more detail, and presents some information on the extent
to which people have undertaken averting actions in response to actual contamination incidents.
In the following sections, this chapter provides more detailed information on the analysis
of averted water use costs. The chapter first describes the economic framework for the analysis
and the analytic approach, then provides the results of the analysis and discusses related
uncertainties.
9.1 Economic Framework
The framework for this analysis is based on two economic concepts: (1) the use of
averted costs as one measure of the effects of corrective action; and, (2) the use of changes in
consumers' surplus to measure the welfare effects, or benefits, associated with the averted costs.
These concepts are described below.
9.1.1 Use of Averting Expenditures as a Benefits Measure
The use of averted costs as a benefits measure assumes that people would be willing to
spend at least as much to control pollution as they would be willing to pay to avoid the effects of
the pollution. In the case of corrective action, this assumption is equivalent to stating that
people would be willing to pay at least as much for ground-water remediation as they would be
willing to pay for treatment or replacement of contaminated water supplies.
2 The total number of facilities where wells are likely to be contaminated in the absence of
corrective action is approximately 360 because, for most facilities where wells beyond two miles
may be contaminated, wells within two miles are also affected. Available data suggest that on-
site household or public drinking water wells will not be affected at any facility.
*** DRAFT - March 22, 1993 **»
-------
9-3
Whether averting expenditures are equivalent to willingness to pay for environmental
improvements depends on whether the averting action is a perfect substitute for the
improvement. This will not be the case if the averting action has benefits other than reducing
the effects of the pollution, nor will it be the case if the environmental improvement has benefits
other than allowing people to avoid expenditures on the averting action.3
In the case of corrective action, averted water treatment or replacement costs are not an
exact substitute for the environmental improvements resulting from facility remediation. Water
treatment or replacement may provide benefits not provided by remediation, for example,
treatment may improve the taste of the water by removing some naturally-occurring, non-
hazardous contaminants. Replacement options, such as extending public water service to homes
previously relying on private wells, may provide benefits other than avoidance of contamination,
such as reducing the time spent by homeowners on well repair and operation. Public supplies
may also be perceived as safer and more reliable, and as providing more adequate fire
protection.
Because averting expenditures are not always a perfect substitute for environmental
improvements, care must be taken in calculating the expenditures to be included in this type of
analysis. In particular, averting expenditures may include some adjustment costs, and the savings
from reversing the averting action may differ from the costs of preventing it. Investments in
durable goods are one example of this issue. If corrective action prevents a future need to treat
drinking water, then future capital investments can be avoided. If instead corrective action
allows current treatment to be discontinued, the capital investment in the treatment technology
will be recouped only if the plant or equipment has salvage value. In some cases, only the
ongoing (e.g., operations and maintenance) costs may be avoided.
These issues are addressed in various ways in this analysis. First, because the
environmental improvements resulting from corrective action have many benefits other than
averted water use costs, this RIA includes several analyses of other types of benefits. Second,
the averted cost analysis includes only those water use costs related to avoiding the effects of
contamination. The analysis considers the actions needed to maintain drinking water at "safe"
levels, assuming that people will always act to ensure that contaminants are below established
thresholds (i.e., maximum contaminant levels -- MCLs - or health-based action levels when no
MCL exists).4 Finally, adjustment costs are considered by explicitly assessing the effects of
timing. If corrective action occurs before the averting action would otherwise be taken, then the
3 See, for example, Courant, Paul, and Richard Porter, "Averting Expenditure and the Cost
of Pollution," Journal of Environmental Economics and Management. Volume 8 (1981), pp. 321 -
329; and, Bartik, Timothy J., "Evaluating the Benefits of Non-Marginal Reductions in Pollution
Using Information on Defensive Expenditures," Journal of Environmental Economics and
Management. Volume 15 (1988), pp. Ill -127.
4 The health risks that result when averting actions are not taken are discussed in Chapter 7.
*»* DRAFT - March 22, 1993 ***
-------
9-4
analysis assumes that the full costs of the averting action can be avoided. If the averting action is
taken before corrective action occurs, then the analysis considers the extent to which corrective
action is likely to reduce or eliminate related future costs.
9.13. Use of Consumers' Surplus as a Measure of Welfare Changes
Economists measure social welfare based on the utility (or "satisfaction") people gain
from purchasing different bundles of goods and services. This concept explicitly recognizes that
people may be indifferent when choosing among certain groupings of goods. For example, a
consumer may be just as happy with three oranges and two apples as she or he would be with
four apples and one orange.
Because of people's willingness to trade-off among different goods, the value of the
averted costs alone will not accurately measure the change in welfare brought about by pollution
reduction measures. For example, if water treatment costs cause water prices to increase, some
people will reduce their consumption of water in response to the price increase.
Economists generally believe that the concepts of compensating and equivalent variation
are the theoretically correct measures of such welfare changes. These concepts focus on the
amount of money (or change in income) that would be necessary to make an individual
indifferent between the initial situation and the new situation. In particular, compensating
variation is the "amount of money which, when taken away from an individual after an economic
change, leaves the person just as well off as before," while equivalent variation is the "amount of
money paid to an individual which -- if an economic change does not happen - leaves the
individual just as well off as if the change had occurred."5 In other words, compensating
variation focuses on the initial level of utility, while equivalent variation focuses on the
subsequent utility level.
However, estimating these measures requires knowledge of people's level of utility before
and after a price change. Because utility is very difficult to measure directly, economists
generally apply the concept of consumers' surplus in empirical work.6
5 Just, Richard E., Darrell L. Hueth and Andrew Schmitz. Applied Welfare Economics and
Public Policy. Englewood, NJ: Prentice Hall Inc., 1982, pp. 10 -11.
6 Consumers' surplus is an accurate and unambiguous measure of welfare changes only
under certain conditions, e.g., when multiple price changes do not occur simultaneously and
income effects are minimal. However, Willig has demonstrated that "in most applications the
error of approximation will be very small." See: Willig, Robert, "Consumer's Surplus Without
Apology," American Economic Review. September 1976.
*** DRAFT - March 22, 1993 ***
-------
9-5
The economic concept of consumers' surplus is based on the principle that some
consumers benefit at current prices, because they are able to purchase goods at a price that is
less than the amount that they would be willing to pay. If prices increase, the difference between
prices and willingness to pay decreases, and related benefits also decrease. Hence if corrective
action prevents future water price increases, corrective action will provide benefits in the form of
avoided losses in consumers' surplus.
This concept can be illustrated using simple supply and demand curves. Exhibit 9-1
shows a downwards sloping demand curve (D) for water. This curve indicates the quantities of
water demanded at different prices, or, in other words, consumers' willingness to pay for varying
quantities of water. As indicated by the slope of this curve, water demand is likely to decrease as
price rises.
The horizontal line indicates the current price of water. If the marginal costs of
production are constant, this line also represents the supply curve.7 The intersection of the
supply and demand curves represents the amount of water consumed at current prices. The
shaded area above the supply curve but below the demand curve represents the consumers'
surplus that accrues at the current price - because some consumers' willingness to pay exceeds
this price.
Exhibit 9-2 indicates what happens to consumers' surplus when the price of water
increases; e.g., due to the costs of responding to contamination. The supply curve shifts upwards
from SI to S2, reducing both the quantity of water consumed and consumers' surplus. The loss
in consumers' surplus is equal to the shaded area bounded by the two supply curves and the
demand curve.
In the case of corrective action, remediation may have one of two effects on water prices
and hence on consumers' surplus:
If water supplies become contaminated before corrective action occurs,
remediation of the hazardous waste facility may decrease or eliminate the future
costs of responding to the contamination (e.g., reduce treatment costs). In this
case, corrective action leads to a price decrease and hence to a gain in consumers'
surplus.
If corrective action prevents future contamination of water supplies, facility
remediation may eliminate the need for future expenditures to respond to the
contamination (e.g., to treat or replace water supplies). In this case, corrective
action averts a future price increase and hence prevents a future loss in
consumers' surplus.
These gains or averted losses in consumers' surplus estimate the water use-related benefits of
corrective action.
7 Because little information is available on the relationship between water prices and
quantities supplied, the corrective action analysis assumes the supply curve is horizontal. If the
supply curves for water are actually upward sloping (i.e., if marginal production costs increase as
quantity increases), then the analysis will understate total benefits because it excludes changes to
producers' surplus.
** DRAFT - March 22, 1993 ***
-------
EXHIBIT 9-1
WATER SUPPLY AND DEMAND
I
Consumers' Surplus
PI
-SI
Ql
Quantity of Water
PI = Current price of water
SI = Water supply curve
Ql = Current quantity of water demanded
D = Water demand curve
-------
EXHIBIT 9-2
LOSS IN CONSUMERS' SURPLUS
I
e
t
P2
PI
S2
Loss in Consumers' Surplus
Q2
Ql
Quantity of Water
PI - Price of water before contamination
P2 = Price of water after contamination
S1 = Water supply curve before contamination
S2 - Water supply curve after contamination
Ql = Quantity of water demanded before contamination
Q2 = Quantity of water demanded after contamination
D = Water demand curve
SI
D
-------
9-8
9.2 Analytic Approach
In the averted cost analysis, the concepts discussed above were used to assess some of the
benefits of corrective action. The. analysis assessed the expenditures on treatment or replacement
of water sources that may be averted by corrective action, and calculated the resultant changes in
consumers' surplus. A case study approach was used to assess these effects, including detailed
analysis of a representative sample of facilities subject to corrective action.
The analysis considered a sample of 79 hazardous waste facilities selected for this RIA, of
which 52 would require remediation under corrective action requirements.8 Water supply and
demand data for the area surrounding each of these 52 facilities were collected, then detailed
case studies were conducted for those facilities where contamination may affect water use in the
absence of remediation.9 Summaries of the case studies are provided in Section 9.3 below.
There were three basic steps in the analysis:
1. Collect data on current and future water use in the area surrounding each
sampled facility.
2. Determine the change in water price and quantity over time, both with and
without corrective action:
a. determine the number and types of wells affected by contaminants;
b. identify the likely response to contamination for each affected well, e.g.,
treatment or replacement of water supplies;
c. calculate the cost of the response and the effect on the price of water;
and,
d. estimate the change in the quantity of water demanded resulting from the
change in price.
3. Determine the benefits of corrective action, by calculating the change in
consumers' surplus that results from the change in water price and quantity.
These steps are discussed below. The effects of major assumptions on the certainty of the
findings are described at the end of this chapter.
8 Information on this sample is provided in Chapter 3.
9 The analysis reported in this chapter focused on water used for household supplies.
Agricultural, industrial and commercial water use were not assessed.
DRAFT - March 22, 1993 ***
-------
9-9
9.2.1 Data Collection
The first step in the analysis of averted costs was to research current water use and
expected changes in use over time. This task included identifying current well locations,
quantities of water used and water prices. In addition, information was collected on predicted
future demand for water and on plans for developing new water supplies.
Current Water Use
The water use research conducted for this analysis focused on the area addressed by the
ground water fate and transport modeling.10 The modeling assessed a triangular area extending
downgradient five miles from each facility, as illustrated in Exhibit 9-3. This downgradient area
was divided into several sectors as indicated in the exhibit.
To research current water use within this area, local water suppliers, well drillers and
local and state authorities were contacted. Local water supply and planning documents and well
drilling records, as well as topographic maps and ground water assessments developed by the
U.S. Geological Service (USGS), were also reviewed.
These data sources generally provided relatively detailed information on the locations of
currently active public wells, the quantities of water withdrawn, the existing types of treatment
and the relationship of the wells to the overall distribution system. The average price of public
water was calculated from information provided by the water supplier, or by combining the
supplier's fee schedule with 1985 USGS data on county-level per capita public water use.11
For private household wells, available data were generally less detailed because
homeowners usually are not required to register private wells with any central authority. If no
public water supplies were available, the analysis assumed that all homes in the area were using
private wells. If public water was available, information from local water suppliers and
government agencies was used to determine whether some private wells were also in use and to
estimate the number of homes in each sector that relied on private wells.
10
Chapter 3 provides more information on the modeling.
11 The 1985 USGS data referenced in this section was derived from the Aggregated Water-
Use Data System. For consistency with the health risk analysis, the analysis assumed that the
average household included 2.63 persons; the national average in 1990 according to the U.S.
Census.
*** DRAFT - March 22, 1993 '
-------
EXHIBIT 9-3
AREA CONSIDERED IN MODELING
Hazardous
Waste
Facility
3.5 mi
40 mi
4.5 mi
5.0 mi
Sector 1
Sector 2
Sec(or3
Sector 4
Sectors
Direction of
Ground Water
Flow i
Sector 6
Sector 12
-------
9-11
To determine water usage for private well users, 1985 USGS county-level estimates of per
capita use for self-supplied water were used, except in a few cases where local data were
available. For private wells, effective water prices were based on estimates of the annualized
capital and operations and maintenance costs relying on local data to the extent possible.12
Where private well cost data were not available from local sources, an engineer experienced in
water supply development assisted in estimating these costs for the area around each facility.13
Future Water Use
The averted cost analysis covered the same period as the health risk analysis - the 128-
year period extending from 1992 through 2119. The base analysis also applied the same
assumption about future growth as the health risk analysis; i.e., that population in the area near
the facility would increase at the county-wide growth rate predicted from analysis of U.S. Census
data. However, the health risk analysis and the averted cost analysis differed somewhat in their
assumptions regarding future private well use. The health risk analysis assumed that private well
use would increase at the predicted population growth rate. The averted cost analysis found
that, in some areas, continued private well use would be more expensive than switching to public
supplies (even in the absence of contamination), and therfore assumed that private well use
eventually would be discontinued in these cases.14 This was not the case for public wells; both
analyses assumed that currently operating public wells would remain in use, and that new wells
would not be drilled in other areas.
In addition, for the averted cost analysis, site-specific data were reviewed to determine
whether local data lead to different conclusions than the base assumptions described above. For
example, the local area may be growing more or less rapidly than the predicted county-wide rate,
and water suppliers may be planning to extend public systems, drill wells in new areas or close
existing wells. This site-specific analysis considered existing water plans as well as recent
trends.15 Both the base and site-specific analysis assumed that the real price of water remained
12 Throughout this analysis, a seven percent discount rate was used to annualize capital costs
and to calculate present values, consistent with guidance from the U.S. Office of Management
and Budget. See: U.S. Office of Management and Budget, "Guidelines and Discount Rates for
Benefit-Cost Analysis of Federal Programs," Circular A-94, October 29,1992.
13 As needed, private well cost data were provided by Mark L. Wetzel, P.E., Senior Project
Manager, Dufresne-Henry Engineers, Westford, Massachusetts.
14 In these cases, local authorities were contacted to confirm that private well use was
decreasing.
15 Available water supply plans generally focused on the near-term, extending about 30 years
into the future at most.
DRAFT - March 22, 1993 **»
-------
9-12
constant in the absence of contamination, and that there would be no changes in available
technology or in peoples' tastes and preferences.
9.2.2 Effect of Contamination
Assessing the effects of contamination on water prices and use involved four steps: (1)
identifying the number and types of wells affected over the 128-year analysis period; (2)
determining the likely response to contamination; (3) calculating the associated costs and price
changes; and, (4) determining the effects of price changes on water use. These steps were
completed twice, once focusing on the extent of contamination over time in the absence of
corrective action, and a second time focusing on the extent of contamination with corrective
action.
Affected Wells
The central tendency ground water modeling results were used to determine the extent of
contamination both with and without corrective action.16 The averted cost analysis relied on
estimates of the concentrations of contaminants at each current public well location in ten-year
increments over the 128-year analysis period. For private wells, contaminant concentrations were
averaged across each sector where wells are now located, again in ten-year increments. The
averted cost analysis assumed that water users would detect the contamination and take averting
actions once contaminants exceed levels of concern in both the baseline (i.e., without corrective
action) and if corrective action occurs.17
Responses to Contamination
For those wells that would become contaminated, the next step in the analysis involved
determining the most likely response of water users to the contamination, based on site-specific
data.18 The options available to each affected community were researched and the relative costs
were determined. In general, the analysis assumed that the community would choose the lowest
cost option. The analysis focused on long-term water treatment or replacement options that
would prevent human exposure to the contamination. For example, options such as the purchase
16 The modeling results are discussed in detail in Chapter 3.
17 Water users were assumed to respond to contamination when it exceeded the action levels
defined by the proposed corrective action regulations. These action levels are defined in Chapter
2 and presented in Appendix E.
18 At some of the sampled facilities, the modeling results indicated that contaminants
reached wells prior to 1992 in the absence of corrective action, contrary to local authorities who
indicated that no contamination was present. In these cases, the analysis assumed that the
community had already implemented the least cost response to contamination.
DRAFT - March 22,1993 ***
-------
9-13
of bottled water for drinking purposes were not considered, because this option would not
prevent dermal exposure or inhalation during showering. Temporary measures that may be
undertaken while more complete remedies were being implemented also were not assessed.
For public wells, the possible responses usually included two options: the water authority
could replace the contaminated supplies with other sources, or could install treatment. The
analysis found that replacement supplies generally would be provided by connecting to systems
serving neighboring areas, or by increasing the amount of water supplied by uncontaminated
portions of the local system. Treatment options generally included installation of air stripping,
granulated activated carbon or (less frequently) reverse osmosis treatment systems, depending on
the contaminants of concern.
For private well users, the possible responses identified usually included connecting to the
public system or installing point-of-entry treatment. Switching to public supplies in some cases
only involved installing the connection from the house to an existing water main; in other cases,
the public distribution system would need to be extended to serve new areas. Home treatment
options generally included installing granulated activated carbon devices, although in some cases
air stripping or reverse osmosis would be required.
Changes in Water Costs and Prices
The next step in the averted cost analysis involved estimating the cost of the response to
contamination and the effect on local water prices. Cost estimates were based on local data and
information from water planners and engineers, while price changes were calculated assuming
full pass-through of all costs to water consumers.19 In all cases, cost estimates were based on
detailed evaluation of local conditions, and include annualized capital costs (using a seven
percent discount rate) and yearly operations and maintenance costs.
The costs of replacing contaminated water sources with new sources were determined
from local estimates to the extent possible. These replacement cost estimates addressed both the
costs of replacing currently used public wells with alternative sources and of connecting private
well users to the public system. In situations where the area was already served by public water,
local water suppliers generally were able to provide data on the costs of connecting additional
households to the public system. Where an existing public system would be extended into new
areas, or where new wells would be needed, the analysis relied on local data as well as the
19 In reality, flat fees or block rates are often used by water suppliers, and some costs
(particularly capital costs) may be financed by general tax increases or bond issuance. The
assumption used in the analysis may overstate the extent to which increased costs are reflected in
water prices, and hence the resulting change in water demand. The net present value of the
averted costs (as well as the change in consumers' surplus) is reported in Section 9.3 to indicate
the effects of assuming instead that there is no change in water demand.
*»* DRAFT - March 22,1993 ***
-------
9-14
expertise of an engineer experienced in water system design and construction.20 Resulting cost
estimates took into account local factors such as required well depths, the configuration of the
existing distribution network, and any storage or conventional treatment needs (e.g., fluoridation,
water softening or filtration).
Treatment costs for both public and private wells, as well as the most appropriate type of
treatment for removing the contaminants of concern, were estimated in consultation with staff
from EPA's Risk Reduction Engineering Laboratory.21 The analysis considered the effect of
contaminant concentrations and water volumes on the costs of each type of treatment, as well as
location-related factors - particularly state-specific requirements for off-gas controls when
treatment by air stripping is used.
Changes in Water Use
When water prices rise due to treatment or replacement of contaminated water supplies,
consumers are likely to decrease their consumption of water. To estimate the responses of
consumers to water price changes, the averted cost analysis used an estimate of the price
elasticity of demand (-0.3) derived from several studies of residential water use. Elasticity refers
to the percentage change in quantity that results from a percentage change in price, and was
calculated at the mid-point of the price and quantity ranges.
Available studies suggest that elasticities for annual average residential water demand in
the United States generally range from -0.2 to -0.7, with only a few exceptions. In 1984, the
Army Corps of Engineers conducted a detailed review of 50 such studies, considering the quality
of the study design as well as the results.22 They concluded that the most likely average
elasticities are at the low end of the range cited above; i.e., between -0.2 and -0.4. The mid-point
of that range, -0.3, was used in the averted cost analysis. In other words, the analysis assumed
that water demand would decrease by 0.3 percent if prices increased by one percent.
20
Mark Wetzel of Dufresne-Henry Engineers also provided this information as necessary.
21 Treatment cost information, as well as information on the treatment most likely to be
effective for the contaminants from each facility, was provided by Robert Clark and Jeffrey
Adams of EPA. Public supply treatment costs were estimated based on two automated models:
Clark, Robert, and Jeffrey Adams. EPA's Drinking Water and Ground Water Remediation Cost
Evaluation: Granular Activated Carbon. Michigan: Lewis Publishers, 1991; and, Clark, Robert,
and Jeffrey Adams. EPA's Drinking Water and Ground Water Remediation Cost Evaluation:
Air Stripping. Michigan: Lewis Publishers, 1991.
22 U.S. Army Corps of Engineers. Influence of Price and Rate Structures on Municipal and
Industrial Water Use. June 1984.
»»* DRAFT - March 22, 1993 **»
-------
9-15
Because of uncertainty in the elasticity estimate as well as in the extent to which water
cost increases will be passed on to consumers in the form of increased water prices, this chapter
also reports the net present value of the averted costs, with no adjustment for changes in the
quantity of water used.23 This estimate is equivalent to assuming that water use is completely
price inelastic; i.e., that the demand curve is a vertical line.
9.2 J Benefits of Corrective Action
The next step in the averted cost analysis involved determining the benefits of corrective
action, by comparing water prices and use over time in the absence of corrective action to water
prices and use with corrective action. This comparison involved calculating the change in
consumers' surplus attributable to corrective action; i.e., determining the size of the area between
the two supply curves and the demand curve as illustrated earlier in Exhibit 9-2. If pi is the
price of water with corrective action, p2 is the price without corrective action, ql is the quantity
of water demanded with corrective action, and q2 is the quantity of water demanded without
corrective action, the formula for calculating this area is:24
Change in consumers' surplus =
((P2-pl)*q2) + 0.5((p2-pl)*(ql-q2))
The result indicates the loss in consumers' surplus that would be averted by corrective action;
i.e., the water use-related benefits of the corrective action requirements.
Note that the requirements will not necessarily avert all costs related to contamination
from the solid waste management units addressed by corrective action, due to factors such as the
time needed to implement remedies and the difficulties inherent in removing certain
contaminants. In particular, at some facilities the modeling results indicated that contaminants
would reach wells before corrective action was initiated, and subsequent remediation would
reduce concentrations but not eliminate all contamination above levels of concern.
9.3 Results
The analysis of averted costs used the central tendency modeling results and detailed case
studies to estimate the consumers' surplus losses averted by corrective action.25 The analysis
completed to date indicates that contamination would affect water use within two miles at 360
facilities nationwide in the absence of corrective action. Corrective action will decrease water
23 If the change in water use instead was calculated based only on the change in annual
operations and maintenance costs (e.g., assuming that capital improvements are funded by bonds
or taxes rather than through water price increases for public systems), the change in consumers'
surplus would differ somewhat from the estimates presented in this chapter.
24 The analysis assumed that all public water is priced at the average price; the effects of
block rates or other variable price schedules are not estimated.
25 The modeling assumptions and results are discussed in Chapter 3.
DRAFT - March 22, 1993 ***
-------
9-16
treatment and replacement costs by $4.8 million, averting $4.7 million in consumers' surplus
losses in these areas. This estimate of benefits will increase when analysis of averted costs in the
area two to five miles from each facility is completed.
These results are discussed in more detail in the following sections, which describe the
number of facilities with averted costs, report the national results, then summarize the case
studies of affected sample facilities. Section 9.4 discusses the limitations of the analysis and
major sources of uncertainty.
9 J.I Number of Facilities
The number of hazardous waste facilities where contamination may affect water use in
the absence of corrective action depends on both the extent of contamination and the location of
nearby wells. Of the approximately 2,600 facilities requiring remediation nationally, water use
may be affected at about 360 facilities in the absence of corrective action. The factors
influencing the number of facilities with water use effects are illustrated in Exhibit 9-4 and
discussed below.
Exhibit 9-4 provides information on the status of facilities subject to the corrective action
requirements. Of the approximately 5,800 facilities subject to corrective action, 2,600 may
require some remediation. However, ground-water contamination may exceed action levels on-
site at only about 2,100 of these facilities.
Of these 2,100 facilities with on-site ground-water contamination, the modeling results
indicate that about 780 may have off-site ground water contamination exceeding action levels
over the 128-year analysis period in the absence of corrective action. Contaminants may reach
off-site public or private drinking water wells at about 360 of these facilities.26
There are several reasons why drinking water sources may not be affected in the areas
surrounding many of the facilities subject to corrective action. In many cases, surface water
supplies are used, wells are not located downgradient from the facility, or wells are too distant
from the facility to become contaminated.27 Wells also may be drilled deep enough to access
uncontaminated portions of the aquifer. In addition, surface water intercepts and (less
frequently) confined aquifers or other geologic features prevent contaminated ground water from
reaching wells in some areas.
26 No on-site public or private residential wells are expected to be affected by contamination at
any of the facilities.
27 While surface water may also become contaminated in the absence of corrective action,
the modeling results suggest that surface water contamination generally drops below levels of
concern before reaching drinking water intakes.
*** DRAFT - March 22,1993 ***
-------
EXHIBIT 9-4
STATUS OF FACILITIES
5,800 facilities
subject to corrective action
I
2,600 faculties
requiring remediation
(aU media)
I
2,100 facilities
with on-site ground-water contamination
in the absence of corrective action
I
780 facilities
with off-site
ground-water
contamination in the
absence of corrective
action
I
360 facilities
with nearby
wells
contaminated
-------
9-18
9.3.2 National Results
Exhibit 9-5 provides the results for the eight sampled facilities with potential benefits
(within two miles) as measured by the averted cost analysis, as well as the sample multipliers and
national totals.28 These eight sample facilities represent about 360 facilities nationally, and are
the only sample facilities where the modeling results indicate that water use within two miles
would be affected by contamination in the absence of corrective action. The results reflect the
base assumptions; i.e., that water use in the area around the facility increases at the predicted
county-wide population growth rate.29
EXHIBIT 9-5
NATIONAL BENEFITS OF CORRECTIVE ACTION
(within two miles)
(1992 net present value)
Facility
Number
25
46
47
48
69
81
104
132
Results for Sampled Facilities
Averted
Costs
$90
$49,400
$210
none
$2,100
none
$201,800
$37,500
Avoided Loss in
Consumers'
Surplus
$80
$48,100
$180
none
$2,100
none
$201,800
$32,500
National Total:
Sample
Multiplier
3.3
63.4
213.8
63.4
33
3.3
73
33
360 facilities
National Results
Averted
Costs
$290
$3,131,900
$44,900
none
$7,000
none
$1,480,100
$125,500
$4,789,800
Avoided Loss in
Consumers'
Surplus
$270
$3,045,100
$38,100
none
$7,000
none
$1,479,600
$108,700
$4,678,800
The exhibit indicates that the total costs averted nationally by corrective action may total $4.8
million, for the areas within two miles of the facilities subject to corrective action. Related
avoided losses in consumers' surplus are slightly less - approximately $4.7 million -- reflecting
28 Totals in exhibit do not add due to rounding.
29 The analysis of site-specific factors is discussed later in Sections 9.3.3 and 9.4.
** DRAFT - March 22,1993 ***
-------
9-19
consumers' responses to price changes. These estimates will increase once water use beyond two
miles is assessed.30
The costs averted by corrective action, as reported in Exhibit 9-5, are only a portion of
the total costs attributable to contamination. Due to the timing of remediation and the
difficulties inherent in removing contaminants, corrective action is only partially effective in
averting water use costs at some facilities. Exhibit 9-6 reports the total costs attributable to
contamination at each of the facilities where water use is affected, as well as the total consumers'
surplus losses attributable to the contamination.31
EXHIBIT 9-6
LOSSES AVERTED BY CORRECTIVE ACTION
COMPARED TO TOTAL LOSSES ATTRIBUTABLE TO CONTAMINATION
(within two miles)
(1992 net present value)
Facility
Number
25
46
47
48
69
81
104
132
Total Costs Attributable to
Contamination
(1992 - 2119)
Individual
Sample
Facility
$100
$3,773,000
$210
none
$15,400
none
$201,800
$137,300
National Total:
National
Results
(weighted)
$330
$239,097,200
$44,900
none
$51,400
none
$1,480,100
$459,300
$241,133,300
Percent of
Costs
Averted by
Corrective
Action
88 percent
1 percent
100 percent
N/A
14 percent
N/A
100 percent
27 percent
2 percent
Total Consumers' Surplus
Losses Attributable to
Contamination
(1992-2119)
Individual
Sample
Facility
$90
$3,648,500
$180
none
$15,400
none
$201,800
$122,900
National Total:
National
Results
(weighted)
$310
$231,203,400
$38,100
none
$51,400
none
$1,479,600
$411,400
$233,184,100
Percent of
Losses
Averted by
Corrective
Action
88 percent
1 percent
100 percent
N/A
14 percent
N/A
100 percent
26 percent
2 percent
30 Because the national estimates are based on a sample of the affected facilities, there is
some likelihood that the actual effects will differ from these estimates due to sampling error.
31 Totals in exhibit do not add due to rounding.
*** DRAFT - March 22,1993 ***
-------
9-20
As reported in Exhibit 9-6, contamination from the facilities would lead to consumers'
surplus losses of about $233 million nationally from 1992 to 2119, compared to a situation where
there is no contamination. Corrective action averts approximately two percent of the consumers'
surplus losses attributable to this contamination nationally because: (1) several years may elapse
between 1992 and the time when remedies become effective; and (2) corrective measures are
only partially effective in some cases, particularly where remediation begins after contaminants
reach nearby wells. These and related issues are discussed in more detail in the summaries of
the case studies below.
9.3.3 Summary of Case Studies
The averted cost analysis addresses a representative sample of the hazardous waste
facilities subject to corrective action, and is based on detailed case studies of those sites where
water use is affected by contamination. This section briefly summarizes the case studies for the
eight sample facilities where drinking water supplies within two miles of the facility will be
affected.
Exhibit 9-7 on the following page summarizes the findings of each of the case studies
under the base assumptions (i.e., water use increases at predicted county-wide population growth
rates). The findings for each facility are based on detailed analyses of local conditions, which are
summarized below and presented in more detail in the case studies.
Facility 25
In the absence of corrective action, contamination from Facility 25 will eventually reach a
small number of private wells. Corrective action slows the spread of contamination and reduces
the concentration, delaying but not preventing contamination of the wells.
Once the wells become contaminated, residents could choose to install treatment on the
wells or to connect to the public system. The homes are located up to one mile from the existing
.public distribution system, and supplying these homes with public water would require extending
the distribution system, installing service pipeline, paying connection fees, and paying for annual
water usage. Continued private well use would require installing of granular activated carbon
treatment, as well as paying the ongoing costs of well, pump and pipeline maintenance and
replacement.
**» DRAFT - March 22,1993 ***
-------
9-21
1
9-7
SUMMARY OF CASE STUDY FINDINGS
(1992-2119)
Facility
Number
25
46
47
48*
69
81
104
132
Effect of Contamination without Corrective Action
First year in
which wells
arc
contaminated
2092
Prior lo 1992
2069
2012
1997
Prior lo 1992
Private wells:
2000
Public wells:
2042
Prior to 1992
Number of people
served by
contaminated wells
over time
Private
welh'
40-30
None
<10
20-70
50-120
None
20-
11300
40
Public
wells
None
6,500-
34,500
None
None
None
52.500-
1,655,600
5.200-
13,000
None
Response to
contamination
Connect to public
supplies
Install treatment
Install treatment
Connect to public
supplies
Connect lo public
supplies
Close and replace
wells
Private wells:
connect to public
supplies
Public wells:
install treatment
Connect to public
supplies or install
treatment'
Effect of Contamination with Corrective Action
First year in
which wells
are
contaminated
2112
Prior to 1992
No
contamination
Unknown
1997
Prior to 1992
Private wells:
2000
Public wells-
no
contamination
Prior lo 1992
Last year in
which wells are
contaminated
None
(contaminants
present in 2119)
2066
N/A
Unknown
1998
None
(contaminants
present in 21 19)
Private wells:
2007
Public wells-
N/A
2018
Number of people
served by
contaminated wells
over time
Private
wells*
50
None
None
Unknown
50
None
20
20-40
Public
wells
None
6,500-
16,700
None
None
None
52.500-
1,655,600
None
None
Response to
contamination
Connect to public
supplies
Install treatment
N/A
Connect lo public
supplies
Conned to public
supplies
Close and replace
wells
Private wells:
connect to public
supplies
Public wells:
none required
Connect lo public
supplies or install
treatment0
Effect of Corrective Action
Delays connection lo public
supplies
Allows removal of treatment
Prevents need for treatment
None
Fewer homes connect to public
supplies
None
Private wells* none
Public welts' prevents need for
treatment
Allows removal of treatment
' Private well counts in this exhibit assume that private well users do not connect to the public system over time
* Data on the effectiveness of corrective action at Facility 48 are not available, however, residents will connect to the public system regardless of contamination.
' Response depends on well location.
DRAFT - March 22,1993
-------
9-22
Exhibit 9-8 reports the annualized costs of each option at the time when contamination
reaches the wells without and with corrective action.3"3 The costs of continued private well
use in the absence of contamination (i.e., without treatment) are also included for comparison.
EXHIBIT 9-8
ANNUALIZED COSTS OF WATER SUPPLY OPTIONS
FACILITY 25
Without corrective action
(14 homes in 2092)
With corrective action
(18 homes in 21 12)
Continue Private
Well Use without
Treatment
$6,500
$8,400
Continue Private
Well Use with
Treatment
$26,500
$24,100
Connect to Public
Supplies
$13,300
$14,400
In the absence of corrective action, about 14 private wells would become contaminated as
of 2092, and connecting to the public system will be less expensive than continued well use with
treatment. Corrective action delays the onset of contamination to 2112, when 18 private wells
will be affected. In this case, connecting to the public system also will be the least expensive
option for responding to contamination.
Assuming that homeowners connect to the public system once their wells become
contaminated, the net present value (in 1992) of the total costs attributable to contamination in
the absence of corrective action is about $100; the difference between the costs of continued
private well use without treatment and connection to the public system from 2092 through 2119.
Corrective action delays, but does not prevent, these costs. The cost-savings attributable to this
delay total about $90 (1992 net present value). These costs are low because the savings are small
and do not begin to accrue until 100 years in the future.
Estimating the benefits of corrective action requires calculating the avoided loss in
consumers' surplus, as discussed in section 9.2.3. For Facility 25, these benefits accrue only from
2092 to 2112, the time period over which corrective action prevents contamination of the wells.
The relevant comparison over this time period is between continued private well use without
32 All costs in these case study summaries include both annualized capital costs and yearly
operations and maintenance costs, using a seven percent discount rate. All costs are expressed in
1992 dollars (i.e., assuming no change in real costs).
33 Annualized costs for connecting to the public system exclude one-time connection fees,
which are about $200 per household in the area around Facility 25. These connection costs are
included in the subsequent calculations of net present value.
***
DRAFT - March 22,1993
***
-------
9-23
treatment (because corrective action has delayed contamination) and connection to the public
system. By delaying the need to connect to the public system and hence the resulting increase in
water price, corrective action averts consumers' surplus losses of approximately $80 (1992 present
value) in the area around Facility 25.
The above analysis assumes that the population using private wells will increase at the
predicted county-wide population growth rate. In the area around Facility 25, private well use
may remain stable or decrease over time because the population of the immediate area is
declining rather than keeping pace with county-level trends. However, because the benefits of
corrective action are small for this facility, changing the growth rate assumption will have little
effect on the estimates of the national benefits of the regulations.
Facility 46
Facility 46 is located in an area where most households receive public water supplies from
eight wells located about one mile downgradient from the facility. These wells are contaminated
as of 1992, i.e., before corrective action occurs, at which time they serve 6,500 people.
Corrective action reduces contaminant concentrations over time, and, by 2067, causes these
concentrations to drop below levels of concern in the area where the wells are located.
When the public wells became contaminated, the water supplier could have installed
granular activated carbon treatment or connected to a larger nearby public system that relies on
surface water supplies. The costs of these options are reported in Exhibit 9-9. The exhibit
includes incremental costs only; it excludes basic expenditures (for example, to maintain the
town's distribution system) that will be incurred under both options. The treatment costs are the
costs of granular activated carbon treatment, while the costs of connecting to the neighboring
system include the installation of a 2.4 mile transmission main, as well as the incremental
increase in annual charges for water use.
EXHIBIT 9-9
INCREMENTAL ANNUALIZED COSTS OF WATER SUPPLY OPTIONS
FACILITY 46
(in 1992)
Install Treatment
$232,400
Connect to Neighboring
System
$380,400
*** DRAFT - March 22, 1993 ***
-------
9-24
As indicated by the exhibit, treatment is the least expensive response to contamination. In the
absence of corrective action, the costs of this treatment from 1992 through 2119 will total $3.8
million (1992 present value).
Corrective action has little effect on treatment costs from 1992 through 2066, but allows
treatment to be discontinued as of 2067. The 1992 present value of the savings attributable to
corrective action from 2067 to 2119 is $49,400.
These savings lead to a gain in consumers' surplus, as defined in section 9.2.3. The
present value (in 1992) of the gain in consumers' surplus attributable to the savings from
removing the treatment is $48,100.
This analysis assumes that the volume of water withdrawn from the affected wells will
increase at the predicted county-wide population growth rate. However, the area around Facility
46 is growing more slowly than county-wide average, because it does not contain any
undeveloped areas. If there were no growth in water withdrawals over time, the cost savings
attributable to correction action would total $16,800 while the gain in consumers' surplus would
decrease to $16,300.
Facility 47
In the area near Facility 47, two households use private wells which would become
contaminated in 2069 in the absence of corrective action. Corrective action prevents
contamination of these wells.
In response to contamination, residents could install treatment or connect to the public
system. To remove the contaminants, granular activated carbon treatment would be required.
To connect the homes to the public system, the distribution system would need to be extended,
and homeowners would incur the costs of installing service pipeline as well as connection fees
and annual water use.
Exhibit 9-10 reports the annualized costs of each option.14 The costs of continued
private well use in the absence of contamination (i.e., without treatment) are also reported for
comparison. In the absence of corrective action, two private wells would become contaminated
as of 2069. For these homes, installing treatment will be the least expensive response to
contamination.
34 Annualized costs for connecting to the public system exclude one-time connection fees,
which are $1,200 per household in the area around Facility 47.
* DRAFT March 22, 1993 ***
-------
9-25
EXHIBIT 9-10
ANNUALIZED COSTS OF WATER SUPPLY OPTIONS
FACILITY 47
(for two homes in 2069)
Continue Private Well
Use without Treatment
$800
Continue Private Well
Use with Treatment
$3,200
Connect to Public
Supplies
$12,000
Assuming that homeowners would install treatment once their wells become contaminated, the
net present value (in 1992) of the costs attributable to contamination are $210.
Corrective action prevents contamination of these wells and therefore averts these
treatment costs; avoiding an increase in the effective price of water and therefore a loss in
consumers' surplus. The present value (in 1992) of the averted loss in consumers' surplus
attributable to the savings from corrective action is $180. These calculations assume that water
use in the area nearest the facility increases at the county-wide population growth rate. For the
wells affected by contamination from Facility 47, local data are consistent with this assumption.
Facility 48
Although public water supplies are available in the area surrounding Facility 48, a few
nearby households use private wells that would become contaminated in the absence of
corrective action. Contamination would reach these wells in 2012, at which time there would be
seven wells in the area. Data on the extent to which corrective action will prevent this
contamination are not available for this facility.
In this area, connecting to the public system is less expensive than continued private well
use even in the absence of contamination, because the public system already serves the area
where homes now using private wells are located. The per household costs of private well use
(without treatment) are approximately $530 on an annualized basis (including the costs of the
well, pump and pipeline), while connecting to the public system costs about $230 annually
(including the costs of piping and annual water use).35 Therefore, residents are likely to
connect to the public system as their wells deteriorate and need replacement, regardless of
whether contamination exists. In this situation, corrective action will not avert any costs nor
provide related benefits.
35 A one-time permit fee of about $30 is also required for connection to the public system.
DRAFT - March 22, 1993
-------
9-26
Facility 69
Facility 69 is located in an area where most homes receive public water supplies, however,
private wells are currently in use at a new development near the facility. In the absence of
corrective action, 18 of these wells become contaminated as of 1997, and 13 additional wells
become contaminated as of 2022. Corrective action limits the time period over which the wells
closest to the facility are contaminated, and prevents contamination of more distant wells.
When the wells become contaminated, homeowners can install treatment or connect to
the public system. To remove the contaminants, granular activated carbon treatment would be
required. Connection to the public system would require extension of the distribution main as
well as installation of service pipeline to each house. Homeowners also would be charged for
annual water usage, and a one-time connection fee.
Exhibit 9-11 provides the annualized costs of each option as well as the costs of private
well use (without treatment) for comparison.36 Connecting to the public system is the least
expensive option for those homes where wells are contaminated in 1997 in the absence of
corrective action; and is also the least expensive option for those homes affected in 2022.
EXHIBIT 9-11
ANNUALIZED COSTS OF WATER SUPPLY OPTIONS
FACILITY 69
(for 18 homes in 1997)
Continue Private Well Use
Without Treatment
$9,100
Continue Private Well Use
With Treatment
$24,900
Connect To Public
Supplier
$10,700
Assuming that homeowners connect to the public system once their wells become contaminated,
the net present value of related costs is $15,400.
Because corrective action limits the time period over which wells nearest the facility are
contaminated but does not prevent contamination, these homeowners are likely to connect to the
public system regardless of whether corrective action occurs. Corrective action prevents
contamination of more distant wells, allowing these homeowners to avoid the cost of connecting
the public system. The present value of the averted costs for these homes is about $2,100;
resulting in an avoided loss in consumers' surplus also of about $2,100.
36 Excludes one-time connection fee of about $120 per household.
DRAFT - March 22, 1993 ***
-------
9-27
Facility 81
Facility 81 is located in an area served entirely by public water supplies. One well field
used by the public supplier is immediately adjacent to the facility in the downgradient direction.
This well field accounts for about IS percent of total supplies, serving 52,500 persons as of 1992.
These wells are affected by several contaminants as of 1992, i.e., before corrective action occurs.
Corrective action reduces the concentrations of the contaminants, but does not reduce these
concentrations below levels of concern due to the complex hydrogeology in the area.
When the public wells became contaminated, the water supplier could have installed
treatment or closed the wells and increased the water withdrawn from uncontaminated portions
of the system.37 To remove the contaminants affecting these, wells, air stripping treatment with
off-gas controls would be heeded. Replacement of the wells would require expenditures on new
wells, pumps, a pumping station and conventional treatment. Replacement is likely to be
substantially less expensive, as indicated by Exhibit 9-12 below. Over time, treatment costs will
increase as water volume increases, while the incremental costs of replacing the wells will be
negligible. The water'authority will incur the costs of operating and periodically replacing the
wells regardless of where they are located.
EXHIBIT 9-12
INCREMENTAL ANNUALIZED COSTS OF WATER SUPPLY OPTIONS
FACILITY 81
(in 1992)
Install Treatment
$12,793,800
Replace Wells
$317,000
Assuming that the wells were replaced prior to 1992, corrective action will not provide
water use benefits in this area. Because corrective action is not successful in reducing
contaminant concentrations below levels of concern, the water authority is unlikely to re-open
wells in the area even after remediation occurs. Data from local water authorities are consistent
with this finding. The water authority reports that it plans to serve the community's growing
population by developing water sources outside of the area affected by contamination from
Facility 81.
Facility 104
37 Most of the well fields used by the supplier are outside the contaminated area (the service
area covers 165 square miles).
DRAFT - March 22,1993 ***
-------
9-28
Facility 104 is located in an area that relies on ground water for most drinking water
supplies. The public water system serving the area withdraws supplies from over 400 wells, two
of which (serving 2,600 people in 1992) are located about one and three-quarters miles
downgradient from the facility. There is also extensive private well use in the area: about 850
private wells are located within two miles downgradient of the facility in 1992.
In the absence of corrective action, contamination first reaches the area where private
wells are located in 2000 and affects public wells beginning in 2042. Over time, contamination
spreads to affect the entire area within at least two miles downgradient from the facility.
Corrective action limits contamination of the area where the nearest private wells are located to
an eight year period (2000 through 2007), and prevents contamination of more distant areas
including the area where the public wells are located.
However, by the time contamination reaches the nearest private wells, these homes are
likely to be receiving public supplies. The public system is expanding rapidly in this area due in
pan to concerns about contamination from sources other than Facility 104. Therefore, corrective
action is not likely to provide benefits related to private well use.
When contamination affects public supplies, the water system could choose to install
treatment or to close the affected wells and increase the water supplied by well fields located in
uncontaminated areas. Although the latter option might be less expensive, the water supplier has
chosen to install treatment in other well fields that have become'contaminated because of the
extensive contamination in the area. The water authority is therefore likely to choose to install
treatment at the wells affected by contaminants from Facility 104. To remove the contaminants,
air stripping with off-gas controls will be required. The annualized cost of this treatment is
$352,300 in 2042 (when the wells would first become contaminated in the absence of facility
remediation). These costs would be fully averted by corrective action.
The present value of the costs averted by corrective action from 2042 through 2119 is
$201,800, and the related averted loss in consumers' surplus also is about $201,800. This analysis
assumes that water use in the area near Facility 104 will increase at the projected county-level
' population growth rate. Local data are consistent with this assumption.
Facility 132
Some households located downgradient from Facility 132 use private wells, which are
contaminated as of 1992 - before corrective action occurs. Corrective action will eventually
reduce contamination below levels of concern in this area.
When their wells became contaminated, residents could have installed treatment or
connected to the public system. The costs of extending the system vary by location, in part
because two different public systems serve the area. As of 1992,16 nearby private wells are
contaminated. For nine of these households, connecting to the public system will be less
expensive than continued private well use (with treatment) once contamination occurs. The
*** DRAFT - March 22,1993
-------
9-29
remaining seven homes are likely to install treatment and continue to use their wells. The per
household costs of each option are provided in Exhibit 9-13 below. Treatment costs vary
depending on contaminant concentrations, while the costs of connecting to the public system
depend on where the house is located.31
EXHIBIT 9-13
ANNUALIZED COSTS OF WATER SUPPLY OPTIONS
FACILITY 132
(per household in 1992)
Continue Private Well
Use without Treatment
$400
Continued Private Well
Use with Treatment
$1,200 - 1,500
Connect to Public
Supplies
$210 - 3,400
In the absence of corrective action, the net present value (in 1992) of the cost of responding to
contamination (e.g., of switching to public supplies or installing treatment) from 1992 to 2119 is
$137,300.
Remedial activities will gradually reduce contamination below action levels. Those
households that connected to the public system are not likely to disconnect and return to private
well use when contaminants drop below levels of concern, because public supplies will be less
costly. However, the homeowners who installed treatment are likely to discontinue treatment
and accrue related savings. The net present value of these savings is $37,500. These savings lead
to a gain in consumers' surplus of $32,500. These calculations assume that water use in the area
near Facility 132 increases at the county-wide population growth rate; local data are consistent
with this assumption.
9.4 Limitations
Several limitations affect the certainty of this analysis. Issues related to sample selection,
facility characterization, modeling of releases, remedy selection and remedy effectiveness are
discussed in Chapters 3 and 4. The following discussion focuses on the major limitations specific
to the averted cost analysis.
38 Annualized costs for connecting to the public system vary widely depending on the
distance to the existing distribution system. The reported costs exclude one-time connection fees,
which are about $50 to $350 per household depending on where the home is located. These
costs are included in the calculations of net present value.
*
DRAFT - March 22,1993
-------
9-30
9.4.1 Factors That May Overstate Benefits
Two assumptions may cause this analysis to overstate the costs averted by corrective
action. First, the analysis assumed that people would undertake averting actions to eliminate all
risks above regulatory levels of concern. As discussed in Chapter 6, people may choose to incur
health risks rather than undertake averting actions, or may be unaware of the contamination. In
these cases, people would incur health risks rather than averted costs. These risks are reported
in Chapter 7.
Second, as noted in the summaries of the case studies in section 9.3.3, the generic
assumptions about future water use tend to lead to overstatement of benefits in a few cases.
Analysis of site-specific data indicated that, at two of the eight sample facilities where water use
would be affected, water use in the area may be lower than the base estimate if current trends
continue. Factors that could lead to higher water use near the facility were not identified at any
of the sites; e.g., water use was not growing faster than county-level population growth rates nor
were water authorities planning to develop new well fields in the areas proximate to the
hazardous waste facilities. The site-specific data decrease the estimates of benefits (averted
losses in consumers' surplus) by about 60 percent, from $4.7 million to $2.7 million.
9.42 Factors That May Understate Results
The averted cost analysis does not address losses in producers' surplus that may be
averted by the corrective action requirements. If the supply curve is upwards sloping (rather
than horizontal as assumed in this analysis), then water suppliers may accrue benefits from
corrective action. The potential magnitude of these benefits is uncertain both because the
characteristics of water supply curves in the areas surrounding hazardous waste facilities are
unknown and because in the case of public supplies, local regulations may limit the extent to
which suppliers can establish prices that exceed their costs.
9.4.3 Factors That May Have an Indeterminant Effect
Two factors have an indeterminant effect on the analysis. First, several assumptions
related to the calculation of consumers' surplus are somewhat uncertain. The calculation of
consumers' surplus assumed that the price"elasticity of water demand would not vary with the
magnitude of the price change nor with whether private or public supplies were affected. The
calculation also was based on average prices and did not consider marginal costs or variable rate
schedules. Furthermore, the analysis assumed that private well users would adjust water demand
based on both annualized capital costs and yearly operations costs; and, in the case of public
supplies, that all costs (capital as well as operations and maintenance) would be passed onto
consumers in the form of increased prices. Finally, the calculations assumed that price changes
39 In some areas, industrial or other non-residential uses are a high proportion of public
water use.
* DRAFT March 22,1993
-------
9-31
are not large enough to have income effects. In combination, these factors may lead to either
under- or overstatement of benefits.
Second, over the time period covered by the analysis (128 years) water use could change
in ways that cannot be predicted from current trends and hence are not addressed by this
analysis. Unexpected changes in either the technology used or water supply and demand could
lead to higher or lower benefits than estimated in this chapter. The effects of these changes on
the benefits estimates may be relatively small, however, because discounting (to account for the
time value of money) minimizes the effects of changes far in the future.
DRAFT - March 22, 1993
-------
10. NONUSE BENEFITS OF GROUND WATER REMEDIATION
This chapter describes the calculation of nonuse benefits of the corrective action
requirements. Nonuse benefits derive from the values people place on natural resources
unrelated to their own use of the resources; for example, the value placed on simply knowing
that clean ground water exists. In the case of corrective action, nonuse benefits may result from
the remediation of ground water, surface water, or soil, or from other elements associated with
the removal of contamination. This chapter considers only the nonuse benefits from the
remediation of ground water.
The primary purpose of this draft of the RIA is to illustrate the approach EPA has used
to-date. The approach to estimating nonuse values associated with ground-water remediation
involved applying estimates from a recent contingent valuation study of ground water.1 The
McClelland et al. study was developed to inform policy decisions affecting ground-water quality.
As a result, the study generated estimates of household willingness to pay for remediating
contaminated ground water that can be applied, with due care, to valuing nonuse benefits of the
corrective action requirements.
The primary conclusions of this analysis are summarized below:
If the McClelland et al. willingness to pay values are appropriate for application
to corrective action ground-water remediation, the resulting estimates of nonuse
benefits range from $170 million to $18 billion, depending on the specific
assumptions used in the calculations. Within this wide range lies EPA's best
estimate that the nonuse benefits of corrective action for ground water are $2.3
billion.
This wide range of benefits is driven by three factors: (1) the number of facilities
assumed to undergo ground-water remediation; (2) the number of households
assumed to value remediation of ground water at each of these facilities; and (3)
the time at which benefits are assumed to accrue to households.
EPA has asked the Environmental Economics Advisory Committee of the Science
Advisory Board (SAB) to review the McClelland et al. study. The results of this review will have
important implications for the calculations described in this chapter. The SAB review is expected
to be completed later in 1993.
This chapter is divided into four major sections. The first section describes the approach
to estimating nonuse values from ground-water remediation at facilities. The second section
1 McClelland, Gary H., William D. Schulze, Jeffrey K. Lazo, Donald M. Waldman, James K.
Doyle, Steven R. Elliott, and Julie R. Irwin, (prepared for the U.S. Environmental Protection
Agency). Methods for Measuring Non-Use Values: A Contingent Valuation Study of
Groundwater Cleanup (Draft). Washington, D.C.: U.S. Environmental Protection Agency,
October 1992.
DRAFT - March 25, 1993 ***
-------
10-2
presents the results of the calculations and the third section summarizes the sensitivity analyses
conducted. The fourth section discusses limitations of the analysis.
10.1 Approach
This part of the chapter describes the analytic approach used to estimate ground-water
nonuse benefits from the corrective action requirements. The first section defines nonuse values,
and the second section provides a brief overview of current issues relevant to the use of the
contingent valuation method to evaluate environmental nonuse benefits. The third section
summarizes the McClelland et al. ground-water valuation study. The final section presents the
calculations used to apply the McClelland et al. results to facilities subject to the corrective
action requirements, and describes several important uncertainties inherent in this process.
10.1.1 Definition of Nonuse Values
Nonuse values, as defined by Krutilla and Kopp, are derived from the nonconsumptive
services provided by a natural resource.2 These nonconsumptive services are available to all
without the possibility of exclusion; can be enjoyed by an individual without affecting the level of
enjoyment of others; and can usually be enjoyed without any monetary expenditure or
commitment of other resources such as time. As such, nonuse values represent values for a pure
public good which is not traded in any market.1 Nonuse values associated with the remediation
of ground water derive from the knowledge that a given aquifer is in a clean condition and will
be passed on in that state to future generations. Thus, nonuse benefits are not dependent on any
current or future use of the ground water.
Nonuse values generally do not result in any observable behavior on the part of
individuals. Thus, actions or market transactions cannot be used to infer nonuse values. Instead,
survey research methods (the contingent valuation method) are used to elicit nonuse values from
individuals.
10.1.2 Issues in Contingent Valuation
The contingent valuation method (CVM) uses survey research techniques to elicit values
for commodities not typically traded in markets. The technique involves the development of a
hypothetical market where the values expressed by survey respondents are contingent upon the
researcher's description of this market. A well-constructed contingent market provides the
respondent with a description of the good or service being valued, the institutional framework
2 Krutilla, J.V., "Conservation Reconsidered," American Economic Review. Volume 57,
Number 4 (1967), pp. 777-786; and Kopp, Raymond J., "Why Existence Value Should Be Used in
Cost-Benefit Analysis," Journal of Policy Analysis and Management. Volume 11, Number 1
(1992), pp. 123-130.
3 McConnell, K.E., "Existence and Bequest Value." in R. Rowe and L. Chestnut, eds.,
Managing Air Quality and Scenic Resources at National Parks and Wilderness Areas. Boulder,
CO: Westview Press, 1983.
** DRAFT - March 25,1993 ***
-------
10-3
under which the good will be provided, a hypothetical payment vehicle, and an opportunity for
the respondent to express a value for the good. At present, the CVM is the only available
technique for estimating nonuse values.
General Methodological Issues
The ability of CVM to measure environmental nonuse benefits remains a point of debate
in the economics profession. The major issues in the debate over the reliability of CVM are the
ability of the researcher to construct a plausible hypothetical market and the ability of
respondents to express a reliable estimate of the value of the good to them.4 In cases where
individuals have no experience purchasing the types of goods (i.e., public goods) they are asked
to evaluate, it may be difficult for these individuals to conduct the same thought process they use
when purchasing goods in a market.1 Proponents of the method cite the consistency of subjects'
CVM responses with other, market-based determinations of the value of public goods as
evidence that CVM measures the economic value of these goods (i.e., the maximum willingness
to pay for the good).6 On the other hand, opponents of CVM hypothesize that instead of
measuring the willingness to pay for a given good, the method may measure willingness to pay
for moral satisfaction, for a "warm glow" or for a variety of "good causes".7-8
The use of CVM to measure nonuse values can be particularly controversial because,
while respondents may have some experience appraising the use value of goods (e.g., they pay a
4 Mitchell, Robert Cameron and Richard T. Carson, Using Surveys to Value Public Goods:
The Contingent Valuation Method. Washington, DC: Resources for the Future, 1989; and
Diamond, Peter A. and Jerry A. Hausman, "On Contingent Valuation Measurement of Nonuse
Values," in Cambridge Economics Inc., Contingent Valuation: A Critical Assessment, papers
presented at a symposium in Washington, DC, April 2-3,1992.
5 Kealy, Mary Jo, Mark Montgomery, and John F. Dovidio, "Reliability and Predictive
Validity of Contingent Values: Does the Nature of the Good Matter?" Journal of Environmental
Economics and Management. Volume 19 (1990), pp. 244-263.
6 Mitchell and Carson 1989; and McClelland et al. 1992.
7 Kahnemann, Daniel and Jack L. Knetsch, "Valuing Public Goods: The Purchase of Moral
Satisfaction," Journal of Environmental Economics and Management. Volume 22 (1992) pp. 57-
70; and Bumess, H., R.G. Cummings, P.T. Granderton, and G.W. Harrison, "Valuing
Environmental Goods: A Critical Appraisal of the State of the Art," in A. Dinar and D.
Zilberman, eds., The Economics and Management of Water and Drainage in Agriculture.
Boston: Klumer Academic, 1991.
8 Diamond, Peter A. and Jerry A. Hausman, "On Contingent Valuation Measurement of
Nonuse Values," in Cambridge Economics Inc., Contingent Valuation: A Critical Assessment.
papers presented at a symposium in Washington, DC, April 2-3, 1992; and Andreoni, James,
"Impure Altruism and Donations to Public Goods: A Theory of Warm-Glow Giving," The
Economic Journal. Volume 100 (1990), pp. 464-477.
**» DRAFT - March 25, 1993
-------
10-4
monthly water bill for their water use), nonuse values have few if any suitable market proxies.
As a result, a CVM researcher attempting to elicit nonuse values from respondents may provide
information for a good of which respondents were previously unaware, giving the interviewer a
high degree of control over the evaluation process.' Proponents of CVM measurement of
nonuse value argue that CVM practitioners can effectively overcome these objections in two
ways: (1) through extensive pre-testing of alternative forms of a survey; and (2) through
adjustments for measurement error (e.g., testing for the accuracy of subjects' understanding of
the good).10
In view of the debate regarding the reliability of CVM for nonuse values, and faced with
the need to propose a sound approach for estimating natural resource damages under the Oil
Pollution Act, the National Oceanic and Atmospheric Administration (NOAA) in May 1992
commissioned a panel of economists and other experts to review the CVM. The review panel,
chaired by Nobel laureates Kenneth Arrow and Robert Solow, recently concluded that, if applied
in a careful manner and with due consideration of uncertainty, CVM-based estimates can provide
useful information for regulatory decisions." The panel also developed a set of guidelines for
CVM studies, which most existing CVM studies, including the McClelland et al. study, do not
fully meet. EPA's Environmental Economics Advisory Committee will be developing
recommendations regarding the use of the McClelland et al. study as a basis for nonuse value in
light of the findings of the NOAA review panel.
Design of CVM Studies
There are two broad areas in which CVM practice has been refined in recent years: (1)
defining the good to be valued and communicating this information to subjects in a format that is
well-understood; and (2) testing for potential biases that are frequently encountered and
interpreting the results of these tests. The first area primarily involves changes in the design of
the survey instrument, and the second point involves refinements in hypothesis testing for biased
results and adjustment of these results.
Defining the Good and the Market Context: The researcher must develop a survey
instrument that effectively defines the good or commodity to be valued and sets up an effective
hypothetical market. Faced with an unfamiliar good, an awkward payment vehicle, and little
background information to relate the good to other more familiar market transactions, the
respondent is likely to either provide an unrealistic bid or to reject the instrument altogether.
This is a particularly important issue for surveys attempting to elicit nonuse values, where the
9 See Diamond and Hausman 1992.
10 See Mitchell and Carson 1989, McClelland et al. 1992.
11 U.S. Department of Commerce, "Natural Resource Damage Assessment Under the Oil
Pollution Act of 1990 - Appendix I - Report of the NOAA Panel on Contingent Valuation,"
Volume 58 (1993), Federal Register, pp. 4601-4614, January 15,1993.
*** DRAFT - March 25,1993 ***
-------
10-5
problems mentioned earlier regarding unfamiliar!ty with the good are greatest.12 The
respondent must understand the good itself, its substitutes, and what might happen if the good
were provided and if it were not provided.11
One challenge for CVM practitioners is to ensure sufficient context to develop a plausible
hypothetical market without overwhelming the subject.14 There is general agreement as to the
range of issues to address, but there is no single accepted method that applies in all
circumstances. For example, the researcher should consider the method of administration (by
mail, by telephone, or in person), the plausibility of the payment vehicle, and the elicitation
format (e.g., bidding game, payment card, referendum format), but each choice involves a trade-
off. For example, while an in-person interview format may better ensure that survey questions
are understood, this format is expensive and therefore limits the sample size. A payment vehicle
that involves taxes may elicit an emotional reaction in some subjects, but may be better
understood than a "generally higher prices" explanation of how the good will be paid for.
Payment card formats may lead to an anchoring bias if subjects tend to choose values in the
middle of the range of possible bids. A payment card may be preferable to an open-ended
question format, however, if it decreases the rate of non-response. Each of these choices must
be evaluated in the context of the individual study.15
Problems associated with the survey instrument can be reduced through extensive
pretesting. In the pretest phase, researchers can test alternative specifications of the problem
statement, payment vehicle, and willingness to pay elicitation method. Pretesting may involve
some combination of debriefing of subjects to track their thought processes, analysis of the
consistency and rationality of responses for each subject, and analysis of the distribution of
responses among all subjects. Several rounds of pretesting are preferable where possible.
Testing for Bias: To address potential biases, the researcher must design a strategy for
collecting and analyzing data for hypothesis testing. For example, a test for bias associated with
embedding may be used when subjects are asked to value unfamiliar goods. Embedding is found
when a subject can not distinguish the good in question from a broader set of goods. For
example, the good in question may be visibility associated with reduced air pollution, but a
respondent's bid may include the value of decreased health effects associated with the pollution
as well as increased visibility. The degree of embedding may be revealed in debriefing of pretest
subjects, or a survey may ask subjects to re-evaluate and adjust their bids to account for
12 Kopp, 1992; and Rosenthal, D.H. and R.H. Nelson, "Why Existence Value Should Not Be
Used in Cost-Benefit Analysis," Journal of Policy Analysis and Management. Volume 11,
Number 1 (1992), pp. 116-122.
11 See Mitchell and Carson 1989.
14 See McClelland et al. 1992.
15 See Mitchell and Carson 1989.
DRAFT - March 25, 1993 ***
-------
10-6
embedding. Kahnemann and Knetsch, however, argue that subjects may not be fully aware of
their embedding of values, implying that an adjustment based on subjects' responses may only
partially account for this bias.16
Another form of bias that may be present is strategic bias. If subjects perceive that their
responses could influence policy choices to their advantage, they may have incentives to
misrepresent their true preferences (e.g., they may try to "free ride"). Reduction of strategic bias
is one rationale for not revealing the name of the sponsor of the survey to respondents. Mitchell
and Carson state that there is little evidence of strategic bias in CVM responses.17 However,
Diamond and Hausman argue that free riding motives may affect CVM responses, pointing to
evidence from studies of hypothetical versus actual payments to charitable causes.18
Attention to minimizing protest bids at the survey design stage does not imply that the
survey results should not be assessed for bias from protest bids. Protest bids are zero bids or
unusually high bids that may not reflect a respondent's valuation for a good, but more likely
reflect the respondent's unwillingness to go through the desired valuation thought process (i.e.,
lodging a protest against the hypothetical situation). In practice, testing for biases associated
with protest bids in CVM responses is considered good practice and may indicate a need for
adjustment.19 Where it is difficult to distinguish protest bids from statistical outliers, such an
adjustment might be designed to account for both protests and outliers.
These types of adjustments are often the subject of controversy, however, because of the
judgement involved in implementing them. An adjustment for protest and outlier bids might first
involve a decision rule for identifying scenario rejection. Scenario rejection occurs when
respondents do not understand the scenario or do not consider the scenario to be plausible.
Scenario rejection may be indicated by a respondent not completing a survey, not completing
certain items in the survey (e.g., the valuation portions), or giving a markedly low or high bid
(i.e., a protest bid). When a bid is provided, distinguishing scenario rejection from a legitimate
bid may require considerable judgement. Adjustment for outliers and protest bids could involve
one of several measures: devising a means for excluding protest bids from the results (e.g.,
excluding zero protest bids or reporting a "trimmed mean"); calculating a revised or imputed bid
for subjects based on demographic data (e.g., to revise zero bids); or incorporating an adjustment
in the estimation process for a regression model of individual willingness to pay.20
16 See Kahnemann and Knetsch 1992.
17 See Mitchell and Carson 1989.
18 See Diamond and Hausman 1992.
19 See Mitchell and Carson 1989.
20 A regression model of willingness to pay is considered standard practice for CVM results,
to control for the demographic composition of the sample (see Mitchell and Carson 1989).
** DRAFT - March 25, 1993
-------
10-7
The choice of method for reporting and applying results can have a significant effect on
the benefits estimate. The results of alternative adjustments may vary widely, and there is often
no clearly superior method of adjustment that applies in all cases. For example, Balson et al., in
their critique of a study measuring the preservation value of visibility protection in the Grand
Canyon, reported statistical estimators of mean willingness to pay for one visibility improvement
scenario ranging from $0.00 (median) to $10.51 (five percent trimmed mean).21 Mitchell and
Carson cite the need to address outlier data, but no consensus exists on a preferred method.22
Non-response bias is a concern that often is ignored in the application and interpretation
of results.23 There are two types of non-response that are important to consider. Survey non-
response occurs when subjects refuse to participate in the survey. Item non-response occurs
when a subject completes pan of a survey, but neglects or refuses to complete other parts, for
example, the valuation questions. For both types of non-response, the difficult issue is judging
the extent to which the results based on respondents reflect the unknown values held by non-
respondents. Extrapolation of benefits to the total population implies that there are no
important differences between respondents and non-respondents. This issue is of critical concern
not only for application of results to the specific good in the CVM survey, but also for transfer of
the results to other similar goods. This issue is addressed in further detail below when discussing
the calculations for facilities subject to corrective action requirements.
21 Chestnut, Lauraine G. and Robert D. Rowe, Preservation Values for Visibility Protection
at the National Parks. Draft Final Report prepared for the Economic Analysis Branch, Office of
Air Quality Planning and Standards, U.S. Environmental Protection Agency and Air Quality
Management Division, National Park Service, February 16, 1990; and Balson, William E., Jeff
Hausman, and Annette E. Hulse, Navaio Generating Station (NGS^ BART Analysis, report
prepared by Decision Focus Incorporated, Los Altos, CA, for the owners of the Navajo
Generating Station, 1991.
22 See Mitchell and Carson 1989.
23 Edwards, Steven F. and Glen D. Anderson, "Overlooked Biases in Contingent Valuation
Surveys: Some Considerations," Land Economics. Volume 63, Number 2 (1987), pp. 168-178; and
Loomis, John B, The Evolution of a More Rigorous Approach to Benefit Transfer: Benefit
Function Transfer," Water Resources Research. Volume 28, Number 3 (1992), pp. 701-705.
** DRAFT - March 25, 1993
-------
10-8
Implications for the Use of CVM Results
While there are many difficult issues to resolve when using the results of CVM surveys,
the sections above suggest two key issues:
1. Is the description of the commodity and the hypothetical market understandable
to respondents and appropriate for the valuation task at hand?
2. Have appropriate tests for biases such as embedding, strategic behavior, protest
bids and outliers, and non-respondents been incorporated in developing the
results of the study?
The approach to these issues used in the corrective action analysis is discussed in later sections of
this chapter, after the summary of the McClelland et al. study.
10.1.3 Summary of the McClelland et al. (1992) Study
McClelland et al.'s contingent valuation study of the value of ground-water remediation
was conducted from 1990 to 1992. The study was the third in a series of EPA-funded studies
exploring the use of CVM for valuing environmental benefits. Developing reliable methods for
estimation of nonuse benefits was an explicit goal of this research. Although the survey was not
designed solely for application to the corrective action requirements, this application was
explicitly considered in designing the contamination scenario. As a result, there are many
similarities between the hypothetical scenario in McClelland et al.'s survey and the circumstances
at facilities subject to corrective action. There are important differences between the ground- .
water remediation "commodity" described in the survey, however, and the ground-water
remediation that would be accomplished by corrective action. These differences and their effect
on the results of this analysis are reviewed later in this chapter, in Section 10.1.4.
The next few pages briefly summarize the problem statement and survey instrument, pre-
testing and survey administration, tests and adjustments for bias, and results from the McClelland
et al. study. This chapter does not present a complete summary of the McClelland et al. study.
For a thorough understanding of the calculations in this chapter, the reader must be familiar
with the complete report prepared by McClelland and his colleagues and cited previously.
Problem Statement and Survey Instrument
McClelland et al. asked subjects for their "views on what, if anything, should be done to
clean up contaminated ground water which can no longer be used without treatment." After
some introductory information about the nature of ground water and the possible sources and
effects of ground-water contamination, McClelland et al. asked subjects to imagine that they
lived "in a community which has ground-water contamination as the result of a leaking public
landfill." The survey instrument included a schematic diagram on the cover that depicted
ground-water flow and contamination from the landfill. Contaminants were stated to be present
*** DRAFT - March 25, 1993 ***
-------
10-9
in 40 percent of the water used by the community, with the remaining 60 percent of the water
coming from uncontaminated sources. The contamination was described as covering "five acres
underground." The survey instrument stated that "no private company or party is at fault," and
that the contamination was the result of standard disposal practices "that were believed to be safe
at the time."
Information on the expected cancer health risk associated with drinking the contaminated
water also was included. Respondents were informed that "scientists estimate...about 10
additional deaths per million people who drink the water per year" would result. The local
government in this hypothetical situation had concluded the contaminated water was unfit for
drinking or cooking unless it was treated, but that the water could be used for bathing, washing
clothes, or watering lawns.
With this context established, McClelland et al. described five options the community
could adopt as possible responses. These were presented in the following order: complete
remediation, to be paid for through increased water bills of current users; containment, to be
paid for through increased water bills of current users and maintained by future generations;
public treatment of water for users, to be paid for by current and future users of the water
through water bills; home treatment, with each household buying a system for $180 and paying
$25 per month for maintenance; and water rationing, that would require households to cut their
water usage by 40 percent but would not result in any increase in water bills. Subjects were
asked to express their level of satisfaction with each of these options.
At this point, subjects were given a more detailed description of the complete
remediation option and asked to indicate their willingness to pay increased monthly water bills
for the next ten years to achieve the remediation. A payment card was used with 21 values
ranging from $0 to "More than $500." Subjects were told that: the money would only be used in
their own community for the cost of the remediation; the total cost was currently unknown; each
household would pay only a share of this actual cost; and if the program "cost more than people
are willing to pay, the program would not be carried out." In addition, they were told "scientists
are satisfied" the remediation would leave the water contaminant-free and safe to drink, and that
households "moving in or just starting out" would not have to pay (though it was unclear if this
applied to families moving in within the ten-year payment period).
Subjects were given the opportunity to adjust their bids by stating the percentage of their
bid that was just for the complete ground-water remediation program (as opposed to a general
contribution to all environmental causes or a contribution to all environmental and other
worthwhile public causes). They were then asked to partition their adjusted bid among own use
value, use value for others in the community, bequest value, and existence value, and to state the
level of responsibility they felt for helping to clean up such a contamination problem in their
community.
Five versions of the survey were developed and administered. Each version contained the
general background material, the complete ground-water remediation valuation question, and an
DRAFT - March 25, 1993 *»*
-------
10-10
additional set of valuation questions. The first version (called A) asked for valuation of the
containment option. The second version (called B) asked for willingness to pay for remediation
in other communities; the money was to be used to supplement other communities' increased
water bills. The third version (called C) asked for willingness to pay for the public treatment
option. The fourth version (D) asked for WTP for the complete remediation option under 10
percent and 70 percent water shortage scenarios. The fifth version (E) was identical to the
second version, except that no additional context (i.e., background on the actual extent of
national ground-water contamination) was provided. All versions included questions on socio-
economic characteristics of the respondents.
Pretesting and Survey Administration
The survey instrument was subjected to extensive pretesting and review prior to final
administration. In the first stage of this process, preliminary versions of two 24-page "full
context" pretest surveys were administered to five subjects each. These surveys contained more
contextual information and more valuation questions than the final survey described above. The
first stage of pretesting used a verbal protocol methodology. Respondents worked individually
and were asked to think aloud; their comments were recorded and used to test if questions were
being interpreted as the researchers intended. The results of this stage were used to modify the
instrument to improve clarity and to reduce the likelihood of scenario rejection.
The second stage of pretesting involved administering the pretest survey to two small
samples (n=39 and n=41) in late 1990. The survey was administered at a market research
center in Denver. Subjects worked individually and then were debriefed. For the first sample, a
full context survey was administered with preliminary (pre-context) and final revised bids for the
complete ground-water remediation option. This design allowed for within subject testing of the
effect of full information. A second pretest on a new set of subjects did not ask for a
preliminary bid; the results of this pretest were compared with the preliminary bids to evaluate
the effect of different information between samples.
The results of pretesting were used primarily to distinguish the contextual information
respondents actually used from extraneous contextual information, so that the final survey could
be shortened to a manageable length for administration by mail. The original, full context
pretest survey contained extensive information on transport of contaminants through ground-
water; a risk ladder to show relative and absolute risk levels; a preliminary valuation for a one-
year temporary solution (piped in water); and a trust fund payment vehicle that included context
on discounting. These elements were deleted from the final survey instrument because they
either increased scenario rejection (e.g., many respondents thought the trust fund would be used
for other purposes and recorded zero protest bids) or they were extraneous to the valuation
thought process.
The final stage of pretesting included a test of the shortened survey, to ensure that the
limited context still provided enough information for the respondents to complete the valuation
process. This final stage involved 117 randomly selected individuals in a Denver market research
* DRAFT - March 25, 1993 ***
-------
10-11
center. McClelland et al. found that subjects' valuation responses remained stable between the
original pretest and the Denver pretest. This evidence served as the final test that scenario
rejection would not be prevalent in the final survey administration.
The final 12 page survey described above was administered through the mail using the
Dillman Total Survey Design method.24 The survey was sent to a national random sample of
5,000 households, with equally sized and geographically distributed sub-samples for each of the
five versions.25-26 Excluding the 467 bad addresses, the effective sample size was 4,533, of
which 2,874 surveys were returned. The overall response rate was therefore 63.4 percent.
However, only 2,546 surveys were returned with the willingness to pay question answered (for a
response rate of 56.2 percent), and of these surveys only 1,983 included sufficient answers to be
used in the regression analysis of willingness to pay (for a response rate of 43.7 percent). The
effect of non-response on the estimate of nonuse benefits is discussed later in this chapter.
Tests and Adjustments for Bias
McClelland et al. adjusted their data to correct for two sources of bias: (1) embedding;
and (2) high outlier bids. The authors concluded that the extensive pretesting procedure to
minimize scenario rejection meant that few if any protest zeros would be encountered in the final
survey administration. Therefore, no adjustment was made for protest zeros. The authors also
did not consider strategic bias to be a problem.
To adjust for embedding, McClelland et al. used the percentage partitioning of bids
reported by subjects. Most subjects (71 percent) reported that 100 percent of their bid was for
the complete ground-water remediation program alone. No adjustment for embedding was
required for these respondents. Among the remaining respondents, the mean percentage
allocation for the remediation program alone was 42.5 percent; the remaining 57.5 percent of
these bids were for other, more generally defined goods. These respondents' bids were adjusted
to reflect the percentage of their bid that they allocated to the program. The overall effect of
this adjustment was to lower the raw sample mean from $13.98 to $11.57 (a drop of 17
percent).27
24 Dillman, D.A. Mail and Telephone Surveys: The Total Design Method. New York: John
Wiley and Sons, 1978.
25 The geographic distinctions used in this analysis were the ten EPA regions.
26 A sixth sub-sample of 500 version B surveys were sent to zip codes with known ground-
water contamination, to test the effect of proximity to actual sites. As a result, each of the five
main sub-samples consisted of 900 mailed surveys.
27 These bids are stated in terms of a monthly payment for ten years, and are in 1991 dollars.
* DRAFT - March 25, 1993 **
-------
10-12
McClelland et al. used regression analysis to adjust for upward skewness in the final
valuation results (i.e., high outlier bids). The authors estimated a linear regression for each of
the willingness to pay valuation scenarios. The independent variables were income; age; an
education index; a series of variables indicating households with children, race, gender, and
geographic location (EPA region); and a series of variables to capture the subjects' use of,
awareness of, exposure to, and attitudes toward ground-water contamination and remediation.
The dependent variable (monthly household WTP) was adjusted using the Box-Cox
transformation to achieve a normal distribution of error terms. McClelland et al. argue that this
transformation is less subjective than a trimming process to remove outlier bids. The validity of
the method is dependent on an implicit assumption that measurement error, rather than model
error, dominates the residual. McClelland et al. concluded that the pronounced upward skew of
the residuals in the untransformed regression indicate that most of the estimation error is
measurement error. Therefore, they concluded that the Box-Cox transformation, which
normalizes the distribution of residuals, would yield unbiased estimators of willingness to pay.
McClelland et al. estimated a Box-Cox transformation coefficient of 0.15 (a coefficient of
zero would imply a natural log transformation, a coefficient of one would imply no
transformation of the linear model is indicated). The effect of the Box-Cox transformation was
to reduce the predicted mean monthly willingness to pay. The untransformed linear model
predicted a mean willingness to pay for complete ground-water remediation (use and nonuse
value) of $11.57 (incorporating the adjustment for embedding), and the Box-Cox transformed
model yielded an estimate of $7.01 for this value (a drop of about 40 percent).
The combined effect of the adjustments for embedding and high outlier bids was to lower
the mean willingness to pay from $13.98 (the raw mean) to $7.01 for use and nonuse benefits, a
drop of roughly 50 percent. McClelland et al. argued that the lower value "can be viewed as the
appropriate value for policy purposes if all error is assumed to be measurement error." Note
again that this value represents an estimate of the monthly household willingness to pay for a ten
year period (i.e., 120 payments) for ground-water remediation when a 40 percent water shortage
is described. Nonuse value is a portion of this amount.
Nonuse Value
McClelland et al. calculated respondents* nonuse values using three methods. The first
method, which they term the "percent split" approach, used respondents self-reported allocations
to use and nonuse motives. The second method, called the "scenario difference" approach, used
the difference between respondents' values for the complete ground-water remediation option,
which would provide both use and nonuse benefits; and the public treatment option, which would
provide primarily use benefits. The third method, the "extrapolation" approach, involved
extrapolating the value for a zero percent water shortage implied by respondents* bids for the 10
percent, 40 percent, and 70 percent water shortage scenarios. Bids to remediate contamination
that would not create any water use consequences (i.e., would create no water shortages) would
presumably represent pure nonuse values.
** DRAFT - March 25, 1993
-------
10-13
Mean willingness to pay for nonuse benefits was estimated to be $3.49, $2.81, and $3.54
per household per month using the percent split, scenario difference, and extrapolation approach,
respectively. McClelland et al. report that the results for the percent splits and extrapolation
approaches are not statistically different but the scenario difference result is significantly less
than the other two using a statistical test at the five percent level. McClelland et al.
hypothesized that this difference in estimates was the result of subjects attaching a bequest value
to the capital equipment in the public treatment option. They concluded that the portion of
subjects' bids for the public treatment option that represents bequest value should be added to
the scenario difference estimate of nonuse value.
10.1.4 Transferring Estimates from McClelland et al. to Facilities Subject to Corrective
Action Requirements
This section of Chapter 10 describes the calculation of nonuse benefit estimates for the
corrective action requirements based on the McClelland et al. results. Because the McClelland
et al. study did not consider the specific ground-water contamination situation found at facilities
subject to corrective action, these calculations constitute a "benefits transfer." Benefits transfer is
the application of benefit estimates for a specific nonmarket good to a setting that may differ
from that evaluated in the original study.28 A recent series of articles evaluated the accuracy of
benefits transfers for water quality benefits and stressed several key aspects of a defensible
benefits transfer.29 These key aspects include: employing regression results that describe
willingness to pay as a function of socioeconomic characteristics and facility characteristics;
evaluating whether the policy change in the original study closely matches the policy change of
concern; and evaluating whether markets for the good in the original study and the policy
situation are similar in terms of size and availability of substitutes'.
The discussion in this part of Chapter 10 is organized into four pans. The first part
describes the calculations, including the willingness to pay regression and other numerical inputs
used from the McClelland et al. study. The second section compares the ground-water
remediation commodity valued by McClelland et al. to the situation at facilities subject to
corrective action. Based on the key differences in these commodities, the section also discusses
Desvousges, William H., Michael C. Naughton, and George R. Parsons, "Benefit Transfer:
Conceptual Problems in Estimating Water Quality Benefits Using Existing Studies," Water
Resources Research. Volume 28, Number 3 (1992), pp.675-683.
29 Smith, V. Kerry, "On Separating Defensible Benefit Transfers from 'Smoke and Mirrors.',"
Water Resources Research. Volume 28, Number 3 (1992), pp. 685-694. Brookshire, David S.
and Helen R. Neill, "Benefit Transfers: Conceptual and Empirical Issues," Water Resources
Research, Volume 28, Number 3 (1992), pp. 651-655. Boyle, Kevin J. and John C. Bergstrom,
"Benefit Transfer Studies: Myths, Pragmatism, and Idealism," Water Resources Research.
Volume 28, Number 3 (1992), pp. 657-663. Loomis, John B, The Evolution of a More Rigorous
Approach to Benefit Transfer: Benefit Function Transfer," Water Resources Research. Volume
28, Number 3 (1992), pp. 701-705. Atkinson, Scott E., Thomas D. Crocker, and Jason F.
Shogren, "Bayesian Exchangeability, Benefit Transfer, and Research Efficiency," Water Resources
Research, Volume 28, Number 3 (1992), pp. 715-722; and Desvousges et al. 1992.
DRAFT - March 25, 1993 ***
-------
10-14
several caveats and sensitivity analyses of the results. The third section discusses estimates of the
number of households that might be considered the relevant market for ground-water
remediation around facilities subject to corrective action, and proposes sensitivity analyses on this
factor as well. Finally, a brief summary is provided.
Selecting Willingness to Pay Data from McClelland et al.
The first task was to select the appropriate willingness to pay function (and related data)
from the McClelland et al. study, and use this function to calculate per household willingness to
pay estimates for each sample facility where ground water would be remediated. In the selection
of willingness to pay data, differences in the ground-water remediation commodity that would be
provided by corrective action requirements compared to the commodity described in the
McClelland et al. CVM survey were ignored.
As summarized previously in this chapter, McClelland et al. performed several
adjustments to their raw survey data. First, they adjusted for embedding by using the self-
reported percentage partitioning of bids. Second, they adjusted for high outlier bids using the
Box-Cox transformation. The calculations in this chapter began with willingness to pay data that
reflected these two adjustments. As noted previously the combined effect of these two
adjustments was to lower the mean monthly household willingness to pay by approximately 50
percent (from $13.98 to $7.01).
The regression analysis presented in McClelland et al. showed that willingness to pay
(after adjustment for self-reported embedding and high outliers) was influenced by the
respondent's income, age,-education, EPA regional location, and a number of other attitude or
knowledge variables. (See Table 7.7 and related discussion in the McClelland et al. document.)
Data describing the knowledge and attitude variables used by McClelland et al. were not
available for populations around facilities. In addition, it was difficult to develop information on
average household age and education for the areas surrounding facilities. Thus, EPA asked
McClelland and his colleagues to provide a willingness to pay equation estimated using only
household income and EPA regional location (and a constant term) as independent variables.30
To accomplish this, McClelland and his colleagues used the Box-Cox regression presented in
Table 7.7 of their report to calculate a revised willingness to pay bid for each observation. The
requested linear equation was then estimated using these Box-Cox predicted bids. The resulting
equation is shown in Exhibit 10-1 below.31
30
See Loomis 1992, Smith 1992, Desvousges et al. 1992.
31 Provided by William Schulze and Jeffrey Lazo based on data collected for McClelland et
al. (1992).
*** DRAFT - March 25, 1993 ***
-------
10-15
EXHIBIT 10-1
ESTIMATED PARAMETERS FOR WILLINGNESS TO PAY REGRESSION:
40 PERCENT WATER SHORTAGE SCENARIO
Variable Name
INTERCEP
INCOME VD
NORTHEAS
NEWYORK
MIDATLAN
SOUTH
MIDWEST
LAKES
SOUTHWES
MOUNTAIN
WEST
NORTHWES
Explanation
Intercept tern
Household Income (in
$1,000)
EPA Region 1 indicator *
EPA Region 2 indicator
EPA Region 3 indicator
EPA Region 4 indicator
EPA Region 5 indicator
EPA Region 6 indicator
EPA Region 7 indicator
EPA Region 8 indicator
EPA Region 9 indicator
EPA Region 10 indicator
Parameter
Estimate
3.068469
0.066549
0.629767
2.835246
1.289339
-0.194772
Omitted b
1.453938
-0.274252
0.941518
0.712507
0.787813
T-Statistic
(two-tailed test)
6.926
21.724
1.067
5.319
2.463
-0.390
NA
3.033
-0.510
1.442
1.271
1.226
P-Value
0.0001
0.0001
0.2862
0.0001
0.0139
0.6968
NA
0.0025
0.6104
0.1494
0.2039
0.2204
a EPA regional indicators are equal to 1 if respondent lives in that region. 0 otherwise.
b Variable omitted from regression for estimation purposes only.
The regression equation presented in Exhibit 10-1 estimates the total monthly willingness
to pay for a household based on the household's income and EPA regional location. Total
willingness to pay includes the household's use and nonuse values to avoid a 40 percent water
shortage. For the purposes of this analysis, the nonuse portion of this willingness to pay was
derived.
As discussed previously, McClelland et al. applied three approaches to measure nonuse
value. The calculations in this chapter use the scenario difference approach to calculate nonuse
values. The scenario difference approach is based on respondents' evaluations of two public
goods: the complete cleanup scenario, which provides both use and nonuse value, and the public
treatment scenario, which provides the same use value but only very small if any nonuse value.
The NOAA review panel favored this type of method to calculate nonuse values because it
places less burden on the respondent to understand the more abstract scenarios associated with
pure nonuse value. As McClelland et al. pointed out, however, the scenario difference approach
***
DRAFT March 25, 1993
***
-------
10-16
may be biased downward, because respondents may have attached a bequest value to the public
treatment capital equipment that would be passed on to future generations.
To implement the McClelland et al. scenario difference approach, the factors reported in
Table 6.6 of their report were used to adjust the 40 percent water shortage bid to a bid for the
public water treatment program. The adjustment factor reported is 0.5030. Following
McClelland et al., nonuse value was calculated as the difference between willingness to pay for
complete ground-water cleanup and willingness to pay for the public water treatment program.
These calculations are illustrated by the following three steps for Facility 17:
1. Estimate the predicted monthly willingness to pay per household for complete
remediation under the 40 percent water shortage scenario. This calculation was
based on the regression model presented in Exhibit 10-1 above. For each facility,
1989 county level data for mean income from the 1990 Census of Population and
Housing was used, inflated to 1992 dollars using the Gross Domestic Product
deflator.31" The location of the facility indicated the appropriate EPA region.
Facility 17 is located in EPA Region 2, and the mean income in the county where
the facility is located is $57,304 (1992 dollars). Therefore, the monthly household
willingness to pay to remediate a 40 percent shortage was calculated to be:
32 The GDP deflator is from the Economic Report of the President. February 1992, and
from personal communication with the U.S. Department of Commerce, Bureau of Economic
Analysis, February 9,1993.
33 U.S. Department of Commerce, Bureau of the Census. Income data from the 1992
Census of Population and Housing provided by the Housing and Household Economic Statistics
Division, June 9,1992.
DRAFT - March 25, 1993 ***
-------
10-17
n = Intercept + (Income Coefficient * Income in $1,000) +
Region 2 Indicator
3.068469 + 0.066549 ($57.304) + 2.835246
= $9.72 monthly household willingness to pay
2. Estimate the willingness to pay for the public treatment option at each
facility. The adjustment factor for the public treatment option, from Table
6.6, page 150 in the McClelland et al. report, is 0.5030. Therefore, at
Facility 17 the monthly household willingness to pay for public treatment
of water is $4.89.
3. Estimate the nonuse value for each facility as the difference between willingness
to pay for the complete ground-water cleanup and willingness to pay for the
public water treatment option. At Facility 17, the monthly per household
willingness to pay associated with nonuse value is $9.72 - $4.89 = $4.83.
These calculations of nonuse value were completed for each facility in the RIA sample
where (1) ground-water contamination is expected to exceed action levels in the 128-year study
period, and (2) the ground-water remediation (including source control as well as ground-water
treatment) that will be provided under the corrective action requirements is similar enough to
the remediation considered in the McClelland et al. study to allow a meaningful transfer of the
McClelland estimates. Once these individual sample facility calculations were completed, the
sample weights were used to generate a national estimate of nonuse benefits.34
Key Differences in the Ground-water Remediation Commodity
The commodity to be provided by the corrective action requirements generally is similar
to the commodity described in the McClelland et al. CVM survey. Ground-water either will be
or is already contaminated both at the facilities for which nonuse values are assessed (i.e.,
34 Nonuse benefit calculations were also completed using the extrapolation approach. To
implement the extrapolation approach, the factors reported in Table 6.6 of the McClelland
report were used to adjust the 40 percent water shortage bid to a 10 percent and 70 percent
water shortage basis. Next, the constant term of a quadratic equation which describes the curve
formed by the 10, 40, and 70 percent water shortage bids was calculated. This constant term
implied that the bid for a zero percent shortage (nonuse value) would be about 31 percent of the
40 percent shortage bid. Using the scenario difference approach, nonuse value was about 50
percent of the 40 percent shortage bid. Therefore, the national nonuse benefit estimate using
the extrapolation approach would be about 62 percent (31/50) of the base estimate using the
scenario diffference approach reported later in this chapter.
*** DRAFT - March 25, 1993 ***
-------
10-18
hazardous waste facilities subject to corrective action due to ground water contamination) and in
the hypothetical situation described in the survey. In both cases, the ground-water contamination
is expected to reach levels which would present health risks if the ground water is used for
drinking. In addition, in both cases a remedy will be implemented. However, the nature and
extent of contamination, and the remediation to be provided under the corrective action
requirements, differs from that described by the McClelland et at. survey in five ways that may
affect the results of the benefits transfer. As a result of these differences, the monthly nonuse
value calculated above would not be appropriate to use for all facilities considered in this
analysis. These five differences are discussed below.
Contamination Scenario: McClelland et al. used a public landfill as the source of
contamination. The rationale for using a public landfill rather than an industrial facility has two
parts: (1) to provide a hypothetical scenario more familiar to subjects; and (2) to establish a
degree of responsibility among subjects so they accept the idea of paying for the cleanup as part
of their water bills. If an industrial landfill were used as the source, subjects would be more
likely either to bid unrealistically high amounts to "punish" the industrial culprit, or to reject the
scenario.
It is possible that people may view facilities subject to corrective action requirements and
the associated ground-water contamination as different from municipal landfill ground-water
contamination. They may perceive that municipal landfill contamination is more or less
threatening, risky or unpleasant than the ground-water contamination at industrial facilities
where corrective action is required. The effect of these perceptual differences, if any, on
subjects' valuation responses is unknown.
Size of Plume: The McClelland et al. problem statement described a five-acre ground-
water plume. While many of the facilities addressed by this RIA are expected to have ground-
water plumes that do not extend beyond the facility's property boundary, the average size of
plumes for the facilities with ground-water contamination extending beyond the facility
boundaries is expected to be much larger than five acres. Estimated plumes at these facilities are
expected to extend off-site for two miles or more, with areal coverage on the order of thousands
of acres.
People are likely to hold higher nonuse values at facilities where the extent of
contamination is greater, but no data are available to test this hypothesis. This factor may cause
the McClelland et al. results to underestimate the true nonuse value of ground-water remediation
around facilities with large plumes.
Existence of Water Shortage: All versions of the McClelland et al. survey described a
pending water shortage to respondents. In the case of corrective action, about 360 facilities (out
of a national total of about 5,800 facilities subject to the requirements) are expected to cause
contamination of drinking water wells over the 128-year period assessed (1992 to 2119). In the
areas surrounding these facilities, it is more likely that water price increases would result from
the contamination (rather than water shortages).
*** DRAFT - March 25,1993
***
-------
10-19
In theory, the degree of water shortage described to respondents should not affect their
stated nonuse values. However, the nonuse values found by McClelland et al. may have been
different if the surveys had presented a pure nonuse value scenario or a scenario that suggested
price increases rather than shortages. The effects of such changes on nonuse willingness to pay
are unknown.
Effectiveness of Remedy: The McClelland et al. remediation scenario implied that
ground-water remediation would be complete ("scientists are satisfied that water cleaned and
reinjected using these methods will be contaminant-free and safe to drink.")35 However, at
some facilities addressed by this RIA, physical and geologic conditions may make a complete
remediation impossible, and may dictate that the ground-water will not be safe to drink. In
addition, at some facilities other sources of contamination may be present (both natural and
man-made) so that remedies required by the corrective action requirements alone will not be
sufficient to achieve drinkable ground water.
Timing of the Remedy: The McClelland et al. complete ground-water remediation
scenario suggested virtually instant remediation. However, at many facilities subject to corrective
action, remedies will take years to remove contaminants, and the plume may shrink or grow in
the meantime. In addition, remediation may not start immediately; facilities will be addressed
according to priorities established by EPA. These differences in the timing of the remedy may
cause the McClelland et al. results to overstate the true household willingness to pay for
remedies required by the corrective action requirements. For example, respondents may place
little value on the remediation until it is largely complete, and may value remedies achieved
sooner in time more highly than those achieved later in time.
Summary of Key Differences: The five differences discussed above represent important
dimensions on which the McClelland et al. ground-water remediation commodity differs from
that provided by the corrective action requirements. There is currently no way to adjust the
McClelland data for the first three differences (contamination scenario, size of plume, existence
of water shortage) and the direction or magnitude of the overall bias created is unknown. In
view of these three differences, care is required in selecting sample facilities for calculation of
nonuse values. Nonuse values should be calculated only for those sample facilities where the
ground-water remediation commodity is reasonably similar to the commodity valued by
McClelland et al.
Estimates of nonuse benefits were calculated only for those facilities where the ground-
water modeling results indicate contaminants in ground water will exceed action levels on-site or
off-site at some point over the 128-year analysis period in the absence of corrective action. This
group includes 46 sample facilities, representing about 2,100 facilities nationally, where ground-
water contamination exceeds levels of concern (i.e., action levels) as defined by the proposed
35 The statement in quotes is from page five of the facsimile surveys in Appendix D of
McClelland et al.
DRAFT - March 25, 1993 ***
-------
10-20
corrective action regulations.36 At these facilities, the remediation required will have the same
effect as the pump-and-treat remedy evaluated by the respondents in the McClelland et al.
survey, although the remedy may apply different technologies.
There is some question whether respondents would hold nonuse value for remediation of
plumes that are not expected to grow beyond the boundaries of the hazardous waste facility.
Therefore, a second estimate was calculated based on the subset of sample facilities where
ground-water plumes are expected to extend beyond the boundaries of the facility. There are 20
sample facilities that fit this description, representing 780 facilities nationally. The figures
calculated using these two assumptions provide the base nonuse estimate range.
The application of the McClelland et al. data to the corrective action rule also adjusts the
willingness to pay results for the last two differences: effectiveness and timing of remedy. At
facilities where the remedy does not achieve contaminant concentrations below action levels, it is
unlikely that respondents would value the remediation at the amounts suggested by the
McClelland et al. study. Thus, nonuse values were not estimated for facilities where ground-
water remediation will not result in ground water that meets action levels over the 128-year
period assessed." In addition, nonuse values were not calculated for facilities where other
sources are expected to contaminate the ground water.38
To adjust for remedy timing, the analysis used facility-specific estimates of the time
required before the ground-water remedy will be complete, developed by the expert panels who
36 As discussed earlier, the analysis assumes that zero nonuse benefits will accrue for those
facilities where corrective action does not reduce concentrations below levels of concern (i.e., 300
of the 2,100 facilities).
37 It is possible that containment could be achieved at some of the facilities where
remediation is not expected to achieve contaminant concentrations below action levels. In those
cases, the McClelland et al. results for the containment scenario might provide a basis for
calculating benefits. However, applying the results of the containment scenario to measure
nonuse benefits is difficult for two reasons: (1) information on whether containment is achieved
is not available for several facilities; and (2) it is not possible, based on the results of the
McClelland et al. survey, to determine how much of respondents' willingness to pay for the
containment option is attributable to nonuse motives.
38 Available information other sources of ground-water contamination around these facilities
is limited. In the area around one facility in the sample, however, there is extensive
contamination from sources other than the facility. These sources include lawn chemicals and
several Superfund sites in the area. Therefore, nonuse benefits were not calculated at that
facility. Inclusion of this facility would increase the national base estimate by $14 million.
*** DRAFT - March 25, 1993 ***
-------
10-21
selected remedies as pan of the analysis for this RIA.19 The base nonuse value calculations
assumed that nonuse benefits are first realized in the year remediation is completed. This
assumption implies that households will not value ground-water remediation until the remedy is
complete. To provide a sensitivity analysis, the effect of assuming that the benefit is realized
beginning at the initiation of the remedy was also calculated.40
The effect of the timing adjustments for the base calculation can be illustrated for Facility
17 as follows. First, the lump-sum equivalent of the 120 monthly payments can be calculated by
taking the present value at an annual discount rate of seven percent.41 This value represents
the lump-sum willingness to pay for nonuse benefits, which in a later step was multiplied by the
relevant population. However, the effect of the timing adjustment for this lump sum amount can
be illustrated.
For Facility 17, the lump-sum benefit estimate equals $416 per household (the lump-sum
equivalent of 120 monthly payments of $4.83 discounted at seven percent per year). If corrective
action begins at this facility in 1999 and requires 15 years to complete, this value is discounted 22
years back to 1992 at an annual rate of seven percent. The resulting net present value is $94 per
household. This value was not used directly in the calculations, but illustrates the impact of
discounting to adjust for the time needed to complete the remedy.
Estimating the Number of Households
Once facilities were selected for nonuse value calculations,-.the per-household willingness
to pay for ground-water remediation was multiplied by the appropriate number of households to
yield an overall benefit estimate. To determine the number of households, two issues were
considered: (1) the number of households around facilities that would hold a nonuse willingness
to pay for remediation of contaminated ground water (the "extent of market"); and (2) the
handling of survey non-respondents.
39 Note that the estimates of the duration of remediation provided by the expert panels and
used in this chapter were not based on-the modeling results developed for this RIA. The remedy
selection process and remedy effectiveness are described in Chapters 4 and 5.
40 This sensitivity analysis is suggested in part by a recent study that found some evidence
that property value losses associated with proximity to Superfund facilities may have been largely
recovered in the early stages of remediation. Kolhase, Janet E., "The Impact of Toxic Waste
Sites on Housing Values," Journal of Urban Economics. Volume 30 (1991), pp. 1-26.
41 Throughout this analysis, a seven percent discount rate was used to calculate present
values, consistent with guidance from the U.S. Office of Management and Budget. See: U.S.
Office of Management and Budget," Guidelines and Discount Rates for Benefit-Cost Analysis of
Federal Programs," Circular A-94, October 29, 1992.
DRAFT - March 25, 1993 ***
-------
10-22
Estimating the Extent of Market: A critically important issue in benefits transfer involves
the number of households that would value the policy change considered. This is termed the
"extent of market." For this analysis, the extent of the market corresponds to the households
that would derive nonuse benefits from remediation of facilities.42 Unfortunately, there is little
guidance in the existing literature for determining an objective measure of the extent of market
for nonuse values of ground water. McClelland et al. consistently framed their valuation
questions in terms of a landfill located "in your community."43 There is little direct information
in the McClelland et al. report, however, to assess how respondents would define the term "in
your community."44
For the purposes of this effort, two options were considered to characterize the extent of
market: (1) the population of each facility's "community" (i.e., local government unit); and (2)
the population within a given proximity to the facility (i.e., five or ten kilometer radius). The
proximity measure provides a proxy for concern about ground-water remediation. Although
there is significant uncertainty regarding the correct measure of extent of market, using the
population of each facility's local government unit most closely approximates the concept of
"community" used in the McClelland et al. survey. Thus, the base nonuse calculations used the
number of households in each facility's local government unit.
The measure of extent of market must also reflect changes in the number of households
over time. The households that will benefit from corrective action requirements are households
that exist after remediation is under way or completed, but before ground-water contamination
would have been eliminated by natural processes in the absence of the requirements. This
concept is illustrated in Exhibit 10-2. At point T3, ground-water'contamination from facilities
would be eliminated through natural processes and no further benefits would be gained from the
requirements.45
42 See Smith 1992.
43 For example, the pretest survey used the terms "county, city, or community" when asking
subjects about their awareness of ground-water contamination. Although the final survey
instrument did not explicitly state that water users "in your community" would pay for all cleanup
costs, it did state that "the water bills of current users will be increased to pay for a complete
ground-water cleanup."
44 McClelland et al. presently are conducting additional research on the extent of market
issue.
45 For the calculations in this chapter, benefits are calculated through 2119 for consistency
with the other RIA analyses (i.e., T, is assumed to equal 2119). However, in many cases,
ground-water contaminant concentrations remain above action levels as of 2119 in the absence of
corrective action.
DRAFT - March 25, 1993 ***
-------
Exhibit 10-2
HOUSEHOLDS THAT GAIN BENEFITS FROM THE
CORRECTIVE ACTION REQUIREMENTS
Households
t
H
Remedy
Begins
Remedy
Complete
T2
Ground water
would have been
remediated
naturally,
without
requirements
T,
Time
-------
10-24
The base nonuse benefit calculations assumed that benefits accrue to the households
around each facility that exist from time T2 (remedy completed) to time T3. As a sensitivity
analysis benefits were calculated assuming that nonuse values accrue to households from time T,
(remedy begins) to time T3. To be consistent with the framing of the McClelland et al. CVM
survey, each household gains the lump-sum benefit of clean ground water once. Thus, the per
household lump-sum willingness to pay was multiplied by the number of households present in
year T2 (or T, in the sensitivity analysis); and then the additional lump-sum willingness to pay
calculated for the increase in number of households each year from T2 (or T,) to T3 was
added.46
Exhibit 10-3 presents the number of households implied by the local government and
proximity assumptions for the extent of market at sample facilities, incorporating the calculation
of population growth.47 The first column of the exhibit shows the average number of
households within each market measure for the sample facilities where the remedy will achieve
ground-water below action levels through 2119.48 The second column shows the range of
households across the facilities for each market measure. For each of the three measures, the
range of the affected households spans three orders of magnitude, and results from differences in
population density around the facilities. For the local government measure, the wide range also
results from differences in the size of the individual local government units, which are not
uniform. The data in Exhibit 10-3 indicate that the five and ten kilometer proximity measures
are within the same order of magnitude as the local government measure, but the proximity
measures imply a somewhat larger extent of market.
46 The equation for the present value of benefits that accrue over time is:
Benefits - " * (WT/>) " ' ["' - " *
1.07"1992 1.07'*"-'992
Where t is the year benefits begin to accrue, H, is the number of households in year t and WTP
is the lump-sum willingness to pay for nonuse benefits per household. This equation uses a
seven percent discount rate.
47 Population data for places from U.S. Department of Commerce, Bureau of the Census;
1990 Census of Population and Housing, increased to households present when the remedy is
complete using county-level population growth factors projected from Census data. Population
data for five- and ten-kilometer proximity measures were based on 1982 data, updated using the
growth factors derived from county-level Census data.
48 There are about 1,800 facilities, represented by 38 sample facilities, where the remedy is
expected to achieve ground-water concentrations below action levels.
** DRAFT - March 25, 1993
-------
10-H
EXHIBIT 10-3
HOUSEHOLD DATA FOR ALTERNATIVE MEASURES OF
THE EXTENT OF MARKET
Alternative Measures of
Extent of Market
Place where facility is located *
Within ten kilometer radius
Within five kilometer radius
a "Place", as defined by the Bureau of the Census, me
villages; minor civil divisions in the New England st
unincorporated Census-designated places.
Households Present in Year
Remedy is Complete
Mean
18,833
77,371
23,174
Range
123 to 272,803
685 to 323,576
427 to 1 15,485
ans incorporated cities, boroughs, towns, and
ates, New York, and Wisconsin; and
There may be significant problems associated with any assumption about the extent of
market. For example, at some facilities the contamination is expected to cross a local
government border, or the facility itself is located very near a border. While it may be plausible
that all the persons in a given city, including those located several miles away from the
contamination, would express the same willingness to pay for ground-water remediation, it is
implausible that persons located several hundred yards away from the facility but who happen to
live in the next town would have zero willingness to pay.
In addition, while it is plausible that persons far from a facility realize nonuse benefits
from its remediation, it is also plausible that those persons living closest to the facility will be
willing to pay the most for the knowledge that their nearby environment is as clean as possible.
The calculations in this chapter balance these considerations by using the local government unit
population as the base estimate of the extent of market. The sensitivity of the results to this
assumption was analyzed by using the ten kilometer radius as an alternative assumption.
Handling of Non-Respondents: McClelland et al. presented data in their report to
suggest that survey non-respondents may have lower willingness to pay for ground-water
remediation. Some respondents (n=231) returned the survey and answered the willingness to
pay question for a 40 percent water shortage, but did not answer the question designed to test
for embedding or the rest of the survey. McClelland et al. hypothesized that those persons who
did not answer some of the questions may provide information about those subjects who did not
return the survey at all, especially if survey non-respondents failed to respond simply because
they did not have sufficient value for ground-water to take time to complete the survey. The
McClelland et al. report indicated that these 231 "item non-respondents" had a mean willingness
to pay for ground-water remediation of about half that of the 1,983 respondents who answered
**
DRAFT - March 25,1993
***
-------
10-26
all of the questions in the survey ($6.79 vs $14.76). This difference was reported to be
statistically significant. Based on this hypothesis and the valuation of item non-respondents,
McClelland et al. concluded that treating the respondents who completed the survey as
representative of the general population would overestimate the true mean willingness to pay for
ground-water cleanup.
To resolve the uncertainty over survey non-respondents' valuations, Edwards and
Anderson (1987) suggest surveying at least ten percent of non-respondents. This type of research
can be used to determine if non-respondents as a group are fundamentally different from the
sample of respondents. Unfortunately, this additional research could not be performed for this
analysis. In the absence of further information on the characteristics of non-respondents, the
base nonuse calculations assumed that all persons who did not respond have no value for the
good. To provide a sensitivity analysis, alternative calculations assumed that non-respondents
hold nonuse value for ground-water remediation equal to one half that of respondents, consistent
with McClelland et al.'s findings for selected item non-respondents.
The results of statistical tests completed by McClelland and his colleagues were used to
determine the appropriate response rate to use in implementing this approach. Although the
overall survey response rate was 63.4 percent, item non-response reduced the number of persons
with usable responses (i.e., responses suitable for regression analysis) to 43.7 percent of the
persons who received a survey. The raw mean willingness to pay among those respondents who
did not provide usable responses, but who did answer both the willingness to pay question and
the embedding adjustment question (n=332), however, did not differ from the mean for
respondents who did provide usable responses. McClelland et al. performed a statistical test to
confirm this finding. Therefore, McClelland and his colleagues recommended use of the number
of usable responses (n=l,983) plus the number of incomplete responses with evidence of the
same willingness to pay (n=332) to calculate the response rate. This calculation yielded an
effective response rate of 51.0 percent, which was used in the corrective action analysis.
Example Calculation: The first step in accounting for the extent of market and non-
response involves estimating the population present when the remedy is complete. Growth
factors were used to increase the population for the year in which benefits accrue. In addition,
growth factors were used to estimate the incremental population growth in each of the later
years through 2119. Population estimates were divided by the 1990 U.S. average number of
persons per household (2.63) to arrive at the number of households. Next, the household
estimates were multiplied by the present value of willingness to pay per household obtained
previously. The resulting stream of benefit estimates was then discounted back to 1992. Finally,
to account for the non-response bias, the estimate was multiplied by 0.51.
At Facility 17, there are 36,806 persons in 1992 in the local government unit where the
facility is located. Dividing this result by 2.63 yields 13,995 households in 1992. Accounting for
growth between 1992 and 2014, when the remedy is complete, yields an estimated 14,002
households in 2014. The benefit in the year 2014 at this facility is therefore 14,002 households
times $416 or about $5.8 million. The growth factor for this facility indicates the population
DRAFT - March 25, 1993
**»
-------
10-27
decreases for most years after 2014. There are very few additional households added, and very
small additional benefits, after 2014. Discounting the total benefits in 2014 back to 1992 using a
seven percent annual discount rate yields a total benefit of about $1.3 million before adjusting
for non-response. After applying the non-response adjustment factor of 0.51, the nonuse benefits
of remediation at this facility are roughly $671,000.
Summary
The calculations in this chapter are based on several adjustments used to transfer
willingness to pay data from the McClelland et al. survey. Some of the adjustments reflect
methods described in the McClelland et al. report. For example, this analysis relied on the
embedding adjustment and Box-Cox transformation of regression results as used by McClelland
et al. This analysis also used the scenario difference approach for estimating the nonuse
component of willingness to pay bids, as developed by McClelland et al.
Most of the adjustments used in this chapter, however, are unique to the application of
the McClelland et al. results to the corrective action requirements. This analysis used a
regression model for willingness to pay estimated specifically to reflect available data for areas
around affected hazardous waste facilities. In addition, this analysis accounted for several
differences in the commodity valued by respondents in the McClelland et al. survey and the
commodity that will be provided by the corrective action requirements. Finally, to calculate
nonuse benefits this analysis incorporated several assumptions concerning the number of
households that would benefit from the corrective action requirements.
Exhibit 10-4 summarizes the important factors in transferring nonuse values from the
McClelland et al. study to facilities considered in this analysis. In addition, the exhibit describes
the sensitivity calculations completed to determine the effect of alternative assumptions on the
base calculation of nonuse benefits. The results of these sensitivity analyses are reported in the .
next section of this chapter.
DRAFT - March 25, 1993
***
-------
10-28
EXHIBIT 10-4
IMPORTANT FACTORS IN THE BENEFITS TRANSFER
AND DESCRIPTION OF SENSITIVITY ANALYSES
Factor
Assumption Used for
Base Estimate
Expected
Impact
Alternative Assumptions
Used in Sensitivity Tests
Choosing the Number of Facilities
Where Nonuse Benefits Apply
Range of estimates developed using: (1) only
facilities with off-site ground-water
contamination and (2) all facilities with on-site
and/or off-site ground-water contamination
Unknown
None
Differences in Commodity
Contamination Scenario
No adjustment is possible
Unknown
None
Size of Plume
No adjustment is possible
Base estimate may
understate true value
None
Existence of Water Shortage
No adjustment is possible
Unknown
None
Effectiveness of Remedy
Use only those facilities where remediation
achieves concentrations below action levels
Base estimate may
understate true value
None
Timing of Remedy
Assume benefits accrue when remediation is
complete
Base estimate may
understate true value
Assume benefits accrue at
beginning of remediation
Estimating the Number of
Households
Extent of Market
Use number of households in Census "place"
where facility is located
Unknown, could be
substantial
Use number of households
within ten kilometers of
facility
Non-response
Assume non-respondents have zero willingness
to pay
Base estimate may
understate true value
Assume non-respondents
have willingness to pay
equal to half that of
respondents
DRAFT <-ch 17,1993
-------
10-29
10.2 Results
This section of Chapter 10 presents the results of the nonuse benefit calculations. The
base calculation used two alternative sets of facilities. The first estimate was based on about 780
facilities nationally (20 sample facilities) generating nonuse benefits; these are all the facilities
with off-site ground-water contamination. The second estimate was based on about 2,100
facilities nationally (46 sample facilities) with either off-site or on-site ground-water
contamination.
Using these two assumptions for the number of facilities and the other base assumptions
described previously in this chapter, the national nonuse benefits of corrective action range from
$170 million to $470 million.49 The majority of the estimates (about 90 percent of the $470
million estimate) is accounted for by the national nonuse benefits at ten key sample facilities that
represent 930 facilities nationally. Four of these key facilities, accounting for $160 million in
national nonuse benefits, have off-site ground-water contamination; the remaining six facilities
have only on-site ground-water contamination. The derivation of national benefits for these ten
sample facilities is described more fully in Exhibits 10-5 through 10-7.
Exhibit 10-5 provides base data for the key facilities. The data include the mean income,
EPA region, number of households in the place where the facility is located in 1992, the number
of households within ten kilometers of each of the key facilities in 1992, and the year in which
contaminants drop below action levels. In addition, the exhibit presents projections of the
number of households in the year in which the remedy is completed at the individual facility.
Finally, the average percent growth factors for years subsequent to the year benefits accrue are
included.
49 The base assumptions include accrual of benefits after the remediation is complete; non-
respondents have zero willingness to pay for nonuse value; and the extent of market is the
number of households in the Census-defined "place." Because these estimates are based on a
sample, there is some probability that actual benefits will differ from the estimates due to
sampling error.
*** DRAFT - March 25, 1993 ***
-------
10-30
EXHIBIT 10-5
INCOME, EXTENT OF MARKET AND REMEDY DATA FOR KEY FACILITIES
Facility
Number
14
17
44
48
80
84
93
125
162
165
EPA
Region
4
2
9
2
5
4
6
7
5
1
Mean
Household
Income in
County
(1992 $000)
39.701
57304
58.933
55.775
36.317
31.777
30.052
34.830
45.680
49.387
1992
Households
in "Place"
54,163
13,995
12,692
34,747
24,197
13,057
19,036
17,892
10,973
9,911
1992
Households
Within
10km
39,824
209,022
101,267
171,622
245,641
18,870
24,508 .
20,473
31,745
61,455
Year
Remedy
Competed
(End Year)
2030
2014
2028
2030
2035
2003
2030
2030
2030
2030
Number of
Households
in "Place"
in End Year
127,852
14,002
24,076
60,185
29,071
13,779
20,860
26,616
13,071
9,927
Number of
Households
Within
10 km in
End Year
94,004
209,135
192,100
297,270
295,123
19,914
26,856
30,456
37,817
61,559
Average Annual
Household Growth
After End Year
(percent)
2.06%
-0.20%
2.00%
1.21%
0.27%
0.98%
0.36%
0.97%
-0.07%
0.53%
DRAFT-March 25, 1993
-------
10-31
Exhibit 10-6 presents the results of the calculations for predicted monthly household
willingness to pay for ground-water remediation under the 40 percent water shortage and public
water treatment scenarios, and the predicted nonuse value calculated using the scenario
difference approach, for key facilities. As shown in Exhibit 10-6, the monthly household
willingness to pay for nonuse benefits at sample facilities is about 50 percent of the mean
willingness to pay for a 40 percent shortage at each facility. The monthly nonuse value estimate
is highest at Facility 17, because of the relatively high income of residents near the facility and
the facility's location in Region 2. (The Region 2 indicator in the willingness to pay regression
has the largest coefficient among the regional indicator variables.) Exhibit 10-6 also provides the
results of the present value calculations for each of the key facilities. The present value was
calculated using a seven percent discount rate.
Exhibit 10-7 summarizes the calculation of national nonuse benefits. The first column of
this exhibit was calculated by multiplying the per household lump sum value (last column of
Exhibit 10-6) by the number of households in each facility's Census place in the first year
benefits accrue (as shown in Exhibit 10-5). This figure was then discounted back from the year
the remedy is complete to 1992 to yield the result in the first column. The second column
provides the present value of the stream of benefits that will accrue in subsequent years up to
2120. The third column is the sum of the previous two columns, and the fourth column was
calculated as 51 percent of the third column, to adjust for non-response bias. The next column
presents the sample weight for each facility; this weight was multiplied by the value in the fourth
column to yield the total national nonuse benefits derived from the sample facility.
10J Sensitivity Analyses
To evaluate the effect of varying key assumptions, three sensitivity calculations were
completed. The results of these sensitivity analyses are summarized in Exhibit 10-8. In addition
to the assumption about the number of sample facilities where nonuse benefits are expected to '
accrue (which defines the range of the base estimate), the estimate is also sensitive to the
assumption about when benefits accrue. Using the alternative assumption that benefits accrue at
the beginning of remediation, the estimate increased to a range of $970 million to $2.3 billion.
These alternative estimates are three to five times larger than the base estimate. The extent of
market assumption is also important. Use of a ten kilometer radius around each facility
increased the benefit estimate to a range of $630 million to $2.3 billion, or a factor of between
two and five times the base estimate. Using an alternative assumption to account for non-
response bias yields an estimate about 1.5 times higher than the base estimate.
Exhibit 10-8 also presents the impact of all alternative assumptions taken simultaneously.
Using all the alternative assumptions yielded an estimate of from $6.3 billion to $18 billion, or
about 38 times the range of the base estimate. Therefore, varying key assumptions yields an
overall range of estimates of $170 million to $18 billion. Within this wide range the Agency
concludes that the best estimate of nonuse benefits would be $2.3 billion. This estimate assumes
that benefits accrue from the 2,100 facilities with on-site and off-site contamination. Also, this
*** DRAFT - March 25, 1993 ***
-------
10-32
estimate results from relaxing the conservative assumptions in the base calculation to assume that
benefits begin to accrue at the start of remediation instead of after remediation is complete.
DRAFT - Mareh 25, 1993 ***
-------
10-33
EXHIBIT 10-6
WILLINGNESS TO PAY CALCULATIONS FOR KEY FACILITIES
(1992 dollars)
Facility
Number
14
17
44
48
80
84
93
125
162
165
40 Percent
Shortage
WTP/Monlh
$5.52
$9.72
$7.70
$9.62
$6.94
$4.99
$4.79
$5.39
$7.56
$6.98
Public Water
Treatment
Program
WTP/Month
$2.77
$4.89
$3.87
$4.84
$3.49
$2.51
$2.41
$2.71
$3.80
$3.51
Nonuse
Value
WTP/Month
$2.74
$4.83
$3.83
$4.78
$3.45
$2.48
$2.38
$2.68
$3.76
$3.47
Lump-Sum
Equivalent of
Monthly WTP*
$236
$416
$330
$412
$297
$214
$205
$231
$324
$299
a Capitalized at 7 percent per year for 120 months
DRAFT-March 25, 1993
-------
10-34
EXHIBIT 10-7
NATIONAL NON-USE VALUE BENEFITS
AT KEY FACILITIES
(1992 dollars)
Facility
Number
14
17
44
48
80
84
93
125
162
165
Discounted Benefits
in End Year
$2,308,000
$1,315,000
$695,000
$1,894,000
.$471.000
$1,398,000
$327,000
$469,000
$324,000
$227,000
Total for ten sample facilities
Total for all facilities
Discounted
Benefits in All
Subsequent Years
$950,000
$1,000
$274,000
$393,000
$19,000
$185,000
$17,000
$75,000
$21,000
$0
Facility Benefits
(Before
Adjusting for
Non-Response)
$3,258.000
$1,315,000
$968,000
$2.287,000
$490,000
$1,583,000
$345,000
$545,000
$344,000
$227,000
Facility Benefits
Assuming
Non-Respondent's
WTP = 0
$1,662,000
$671,000
$494,000
$1,166.000
$250,000
$807,000
$176.000
$277,000
$175,000
$116,000
Sample
Weight
63.4
63.4
63.4
63.4
63.4
63.4
54.7
213.8
63.4
213.8
930
2,100
Total
National
Nonuse
Benefits
$110 million
$43 million
$31 million
$74 million
$16 million
$51 million
$9.6 million
$59 million
$11 million
$25 million
$430 million
$470 million
DRAFT-March 25,1993
-------
EXHIBIT 10-8
RESULTS OF SENSITIVITY ANALYSES
(1992 dollars)
Description of Estimate
Assumptions
Total National
Nonuse Benefits
Including Only Facilities
With Off-Site
Contamination
(780 facilities nationally)
Including All Facilities
With On-Site and/or Off-
Site Contamination
(2,100 facilities nationally)
Base Estimate
Benefits accrue after remediation is
complete; non-respondents have
zero willingness to pay for nonuse
value; extent of market is
households in Census "place"
$170 million
S470 million
Difference in Timing of
Benefits
Base assumptions except benefits
accrue at start of remediation
$970 million
$2,300 million
Alternative Adjustment for
Non-response
Base assumptions except non-
respondents have willingness to pay
equal to half that of respondents
$250 million
$700 million
Alternative Extent of Market
Base assumptions except market
equals all households within It)
kilometers of the facility
$630 million
$2,300 million
All Sensitivity Analyses
Simultaneously
Benefits accrue at start of
remediation; non-respondents have
willingness to pay equal to half that
of respondents; 10 kilometer radius
extent of market.
$6,300 million
$18,000 million
DRAFT-March 25,1993
-------
10-36
10.4 Limitations
In addition to issues related to sampling error, several other limitations affect the
certainty of this analysis. Issues related to sample selection, facility characterization, modeling of
releases, remedy selection and remedy effectiveness are discussed in Chapters 3 and 4. The
discussion below focuses on the major limitations specific to the nonuse value analysis discussed
in this chapter.
10.4.1 Factors That May Understate Results
The McClelland et al. study indicates that the application of the own community results
may underestimate the total national willingness to pay for corrective action remedies. In
addition to the own community ground-water cleanup question, McClelland et al. asked 2,085
subjects for their valuation of a national cleanup program that would be used to supplement
communities* own willingness to pay to remediate ground-water contamination. Sixty-two
percent of the 1,117 respondents who answered this question gave non-zero valuation for the
national program. McClelland and his colleagues are continuing research to provide further
information on whether these results are applicable for nonuse value around facilities.
Another factor that may bias the results downward is the impracticality of applying
McClelland et al.'s containment scenario results to facilities where corrective action does not
reduce concentrations below action levels, but does contain the plume. Unfortunately, it is not
possible to determine the nonuse component of respondents' bids for the McClelland et al.
containment scenario. As a result, this analysis assumes that at the 300 facilities where the
remedy does not achieve ground-water concentrations below action levels over the 128-year study
period, no nonuse benefits will result. If containment is achieved at some of these facilities,
nonuse benefits could be greater.
10.4.2 Factors That May Overstate Results
Some facilities subject to corrective action requirements may be located in relatively close
proximity to other such facilities. For example, in one area in New Jersey, there are at least 33
hazardous waste facilities within a ten-mile radius. The McClelland et al. survey, however,
describes a community with a single source of contamination. In multiple-facility communities,
the valuation of the remediation may not increase proportionately with the number of facilities,
as the analysis in this chapter implicitly assumes. For example, a community with two facilities
that require remediation may not value the remediation of both at twice the level of remediation
of one facility.
As discussed earlier in this chapter, there is limited knowledge of other sources of
ground-water contamination (e.g., Superfund sites) near the sample facilities. If other sources do
exist near the sample facilities, and affect the same ground-water resources, then remediation
required under the corrective action rule may not yield ground-water contaminant concentrations
below action levels. The analysis in this chapter excludes one sample facility where ground water
DRAFT - March 25, 1993 ***
-------
10-37
is contaminated by other sources (e.g., the use of lawn chemicals). To the extent that the
analysis includes other facilities where corrective action will not lead to clean ground water, it
may overstate the benefits of corrective action.
10.4J Factors That Have an Indeterminate Effect
The sensitivity analyses described in the previous section consider sources of error
associated with the transfer of benefits from the McClelland et al. study. However, this analysis
does not consider any underlying error that may be inherent in the basic willingness to pay data
taken from the McClelland et al. study. Some economists have argued that values from
contingent valuation studies should be considered to have accuracy within about an order of
magnitude, but not all economists agree with this assessment.50 The McClelland et al. study
does not include estimates of the uncertainty in the willingness to pay results reported, or in the
willingness to pay function provided for this analysis.
50 See Chestnut and Rowe 1990.
*** DRAFT - March 25, 1993 **»
-------
11. RESIDENTIAL PROPERTY ANALYSIS
This chapter presents an analysis by EPA to estimate the economic damages from
treatment, storage, and disposal facilities (TSDFs) as expressed in residential property markets.
The analysis evaluates the effect of contaminated TSDFs on the value of nearby residential
properties. Section 11-1 discusses the expected linkages between TSDFs and the behavior of
residential property markets. Section 11-2 presents the methodology that formed the basis of the
analysis and describes the data that were assembled. Three TSDFs were the target of EPA's
evaluations. Section 11-3 presents the findings of EPA's analyses and, in particular, an overview
of the estimated economic damages derived from statistical models of the relationships between
TSDFs and housing prices. Section 11-4 draws conclusions about the relevance of this analysis to
the current rule.
11.1 Expected Linkages between TSDFs and Residential Property Markets
11.1.1 Overview of Hedonic Analysis
Though valuable, many aspects of the environment are not traded on markets, making it
difficult to quantify their value. Within the realm of environmental benefit-cost analysis, it is a
common practice to deal with this problem by inferring the value that consumers place on some
particular good from their purchasing behavior towards other related goods, for which there are
explicit values because they are traded on markets. Houses can be seen in this light. In
scrutinizing a particular house, potential homebuyers have to evaluate whether its bundle of
characteristics (size, number of rooms, location, age, etc.) satisfy their preferences at an
acceptable price. Location is the most important in determining' the household's exposure to
environmental amenities (e.g., natural beauty) and disamenities (e.g., poor air quality, toxic
contamination from adjoining land uses, noise).
Consequently, locational preferences embodied in housing choices provide a link between
a purchased commodity (housing) and environmental effects. What homebuyers pay for a house
in part reflects the value they place on environmental amenities and disamenities that come with
the house. It is through statistical techniques that one tries to determine what portion of the
home purchase payment is attributable to specific effects, such as the influence of a nearby
TSDF. Generally, the analysis of the value of different attributes of a given good is known as
hedonic analysis, and in this case, as hedonic property value analysis.1
11.1.2 Relationship to Corrective Action
In this context, it is important to ask whether the rule will result in changes that affect
homeowners' perceptions of any negative effects from hazardous waste facilities. Corrective
action is a blunt instrument - remediation may affect more facets of a facility than just the risks
it poses. Some of these facets (unsightliness, noxious but not toxic fumes) may be very important
1 For more detail on hedonic analysis, see Raymond B. Palmquist. "Hedonic Methods."
Chapter in Measuring the Demand for Environmental Quality. John B. Braden and Charles D.
Kolstad, Editors. Amsterdam: North Holland. 1991.
**
DRAFT - March 23,1993
-------
11-2
to nearby homeowners and more readily detected than risks per se. For example, at one facility
considered in EPA's analysis, there were two solid waste management units of concern, both of
which were surface impoundments. One of these surface impoundments might be a candidate
for soil capping or a RCRA cap under corrective action. The reduction of fumes and the
improved landscaping that would come from this remedy could reduce the negative impacts on
nearby households and therefore on property values. If corrective action reduces negative
impacts on property values, independent of whether any risks have been reduced, then clearly
homeowners are better off.
The goal of corrective action is to cleanup past, present, and potential releases of
hazardous wastes at facilities regulated by RCRA. The corrective action provisions cover large
releases as well as small releases. In its analysis, EPA examined public perception of TSDFs
regulated under RCRA. The public is alerted to the disamenities of living near a hazardous
waste facility by reports of contamination and by odors wafting from the facility, as well as
smoke, fire, or unsightly physical structures. A reduction of such disamenities will result from
the implementation of corrective action.
11.1.3 Lessons from Past Studies
The hedonic approach has been applied to environmental problems before. A large
number of studies have evaluated the impact of air pollution on property values; a few have
evaluated the impact of hazardous waste sites. The hazardous waste site studies, mostly funded
by EPA in the early and mid-1980's, focused primarily on uncontrolled waste sites that were
already Superfund sites or were similar to Superfund sites. The primary intent of these studies
was to examine the benefits of the Superfund program. In all of these studies, distance from a
given site provided one measure of the differential effect of the site on housing in its vicinity.
The record from earlier hedonic studies of Superfund-like sites is mixed. Some studies identified
significant effects; others found only weak correlations. This uneven record can be explained in
part by differences in the state of knowledge at the time of any given study. Later studies drew
upon the more successful techniques of earlier ones
The settings for TSDFs range from the heart of densely populated New Jersey to rural
Alabama or remote Prudhoe Bay in Alaska. The property value approach is not feasible for
measuring the value of TSDF environmental impacts in all of these settings. From a review of
past studies, certain common features stand out which carry great weight in determining whether
a property value approach will work or not.
Conducting a hedonic study on sites in a metropolitan area is a virtual necessity. In
order to characterize the hedonic function for a housing market, it is important to have adequate
data on housing transactions. Even more important for measuring the impacts of a hazardous
waste facility is the av.iilability of housing sales within proximity of the site. Several of the earlier
studies had very few observations within one-half of a mile, where one would expect there to be
the greatest impacts. Densely populated areas are more likely to generate more sales in close
proximity. In rural areas, the fact that the population is sparse compounds the likelihood that
the number of housing sales will be low.
DRAFT - March 23, 1993
-------
11-3
In an analysis of the corrective action rule, the influence of ground-water contamination
deserves special attention. Ground-water contamination affects the welfare of individuals in
several ways. Some of these effects, such as potential health risks, can be linked to housing
choices; others, such as existence values cannot. Even where they can be linked, describing the
connections has been challenging because not only are physical processes involved, such as the
movement of contamination underground, but also human behavior, and most importantly,
human perceptions.
It is tempting to think, for example, that the toxic risks from contamination should be
modelled in a manner analogous to the way it would be modelled in a risk assessment -
characterize the source and nature of the contaminant, its movement through the environment,
the nature of human exposure, and, ultimately, the resulting risks. This approach is not
satisfactory because it presumes a level of sophistication in analysis and information which most
individual homeowners are unlikely to have. It is more likely that individuals will be much more
simplified in their assessment of the effects from contamination. Accordingly, relatively
sophisticated characterizations of the influence of contamination have not outperformed simpler
approaches in empirical property value studies. As a result, a large part of the modelling of the
influence of contamination in these studies can be summarized in terms of three attributes -
proximity, timing, and area-wide or systemic effects.
Where a single, identifiable source of contamination is involved, proximity of each house
to this source has become the common measure of the impact of contamination on housing
prices. The expected relationship is that housing prices increase with distance. Proximity has
been depicted in various forms in statistical specifications of the relationship between housing
prices - using linear measures of distance, nonlinear measures (including an exponential decline
function intended to mimic the fate of contaminants in the environment), and linear spline
functions (which allow price effects to vary over discrete intervals from the site). A review of
property value studies to date, which have mostly focused on landfills, Superfund sites, and
hazardous waste management facilities, indicated housing prices increased by $900 to $18,500 per
mile of distance from the contamination source, with a median value of $3,800 (1991 dollars).2
These estimates reflect not only the influence of ground-water contamination and other forms of
contamination but the other aforementioned disamenities as well. When limited to studies where
ground-water contamination was identified as a major factor, the range became $3,000 to $9,000
per mile.5
Identifying impacts on property values also appears to be sensitive to timing. When a
distinctive contamination episode takes place (such as the discovery of long-standing leaks at a
facility or an accident which leads to immediate contamination), new information is provided to
2 This range was based upon calculations made using estimates from different specifications
generated in four studies: Kohlhase, 1988; Michaels and Smith, 1990; Michaels et. al., 1992;
Smith and Desvousges, 1986; and Thayer et. al., 1991. See Bibliography for full citations.
t 3 These calculations were based upon results in Kohlhase, 1988 and Michaels and Smith,
1990.
DRAFT - Mareh 23,1993 ***
-------
11-4
the public about the nature of that facility's operations. Subsequently, public perceptions of the
potential environmental impacts from a facility are likely to change and these can translate into
property value effects. Prior to such an episode, the facility might have been expected to have
some negative impact on local property values but after the episode, the negative impact could
be magnified. In most of the past studies of hazardous waste sites, negative effects were
identified after but not prior to a key contamination event.
Finally, though proximity is a powerful proxy for some effects of contamination sources
on property values, it does not capture certain area-wide effects associated with contamination.
The contamination of public drinking wells, as occurred in the 1970's at the Woburn Wells
Superfund site in Massachusetts, is one example of where proximity may not be as important as
systemic effects. Since these wells provided 25 percent of the drinking water for the town of
Woburn, more than the households living near the contaminated sites were affected. A property
value model which tried to take these sorts of systemic effects into account found no economic
damages for Wobum per se, but did find such damages for two other sites in Ashland and
Acton, Massachusetts, where ground-water contamination was an important factor in their
designation as Superfund sites. The estimated damages for Acton and Ashland were $4.3 million
and $47.6 million, respectively.
11.1.4 Motivation for New Analysis
For this RIA, EPA initiated a new effort to evaluate the effect of TSDFs on property
values. The earlier studies provide a useful foundation for examining the effects of contaminated
waste sites on property values but they were not deemed sufficient for calculating the types of
damages to be addressed by corrective action. They focused on actual Superfund sites or other
uncontrolled hazardous waste sites, not on TSDFs. TSDFs operate under Federal hazardous
waste regulations which became much more stringent during the 1980s, partially in response to
public interest in avoiding future Superfund-like problems. Many of the practices that led to
Superfund problems have been prohibited at TSDFs. Consequently, the waste facilities operating
today should pose much less of an environmental threat than the universe of hazardous waste
facilities did 10-15 years ago. One would expect then that the estimated damages from
contamination at federally regulated TSDFs would be smaller than what was observed in the
.earlier studies of Superfund sites. However, this possibility depends on how public perceptions
might have changed during this period.
Public perceptions could be very different today than they were at the time of the earlier
studies. The public now has 12 years of experience with Superfund and several years of
experience with a more stringent RCRA. This experience implies changes in their perceptions of
the magnitude and extent of public health and environmental problems that may stem from waste
facilities. To the degree that public concerns have grown with this experience, one would expect
the negative impacts to have increased. Conversely, if the public was more confident that
government was making more effective efforts to control these problems, impacts on property
values should be less prevalent. Either tendency could be in effect today.
Finally it is uncertain how stable the negative property value effects are over time. They
could grow as public concern persists. They could dissipate as remediation takes place, but this is
DRAFT - March 23,1993
-------
11-5
not necessarily the case if it takes complete cleanup to rebuild public confidence. Conducting a
new hedonic study of TSDFs offered the opportunity to take advantage of data sets which are
richer in housing transactions that took place after contamination events at TSDFs in the early to
mid-1980s.
11.2 Statistical Specifications and Data Employed in Current Analysis
The hedonic approach uses data on all property transactions during a certain time period
in an area surrounding a TSDF. EPA's analysis uses ordinary least squares regression to
estimate the price-distance relationship, controlling for variation in housing characteristics and
for a time trend. In order for the estimated effect of distance on price to reflect effects of the
TSDF accurately, the effect of omitted variables on price must be uncorrelated with distance
from each site. For example, suppose housing prices are determined by equation [1] (a "hedonic
price equation"):
g (p) = a +b-D + c-X + k-T + «, [1]
where p is the price of the house; g is an increasing function; D is distance from the TSDF in
miles; X represents other housing characteristics measured in the data set; 7 is a time trend; and
u, the residual, reflects the effect of any relevant omitted variables, such as housing and
neighborhood characteristics for which data are not available. In this analysis, g (p) was set equal
to either/? or log p. Regression of g (p) on D, X, and 7" will consistently estimate b if « is
uncorrelated with D. The regression coefficient b should be positive if households are willing to
pay more for houses further removed from a TSDF.
The timing of public knowledge about contamination at TSDFs was expected to play an
important role in the influence of TSDFs on housing prices. Earlier studies addressed this issue
by using model specifications that allow the effect of distance on price to change after a key
event (e.g., the time when contamination was publicized, or the time of a public intervention).
Most of the sites initially targeted in this study showed high potential for such a key event. This
approach involved estimation of a regression equation such as [2], which allows the price-distance
relationship to change after a key event:
g (p) = a + a-PERIOD + b-D + p-D-PERlOD + c-X + k-T + u, [2]
where PERIOD is an index variable set to 0 before the event and 1 after. Before the event, the
derivative of g (p) with respect to D is b; after the event, it is b + /3. The coefficient on the
event indicator, a, should be negative if the event generally depresses all house prices regardless
of distance. Furthermore, if distance from the facility is valued even more after the event, then /3
should be positive. Finally, specifications where g (p) = p and g (p) = \ogp are referred to as
linear and semilogarithmic specifications, respectively.
11.2.1 Data Requirements
EPA chose three TSDFs for analysis based on a number of criteria. First, a facility had
to be in a metropolitan area to insure that it was located near large residential communities.
DRAFT - March 23,1993
-------
11-6
Initially, 30 facilities clustered in 12 metropolitan areas were selected. Second, being able to
obtain data from other sources for the area (on property values and housing characteristics) was
necessary, which eliminated certain metropolitan TSDF clusters. Third, the need to use
information from other sources on TSDFs to identify sites which had a history of contamination
or which had a key contamination event. Of the six facilities remaining at this stage of the
selection process, three with the greatest potential for meeting other criteria (significant numbers
of housing transactions from nearby properties, major contamination at the facility) were chosen
for the most in-depth study.
All three facilities are located in heavily populated metropolitan areas but they vary
significantly in their size and corrective action requirements. Facility #1 is a commercial,
hazardous waste TSDF. The operation encompasses unlined surface impoundments, an
incinerator, treatment facilities, and waste storage tanks. The facility has been in operation for
more than 30 years. Expert review of this facility recommended that corrective action be
implemented for 30 percent of the Solid Waste Management Unit (SWMU) area (6.9 of the 23
acres of surface impoundments). Facility #2 is a large oil refinery which generates a wide variety
of hazardous wastes, including oil-contaminated solids, waste caustics and solvents, metal-bearing
wastes, and PCBs. It has been in operation for 90 years and occupies several thousand acres.
Expert review of this facility recommended that corrective action be implemented for
approximately 60 percent of all SWMU areas. The recommended actions have high potential for
reducing unsightliness and odors since the remedies involve installing a cap or a cover. Facility
#3 is a former chemical facility, where, at various times over a three decade period, chemical
production, wax refining, plastics and polyester resin manufacturing have taken place. This
facility is three acres in size. No expert review of potential remedial actions has been conducted
to date.
Hedonic property value analyses use data on individual homes. In cross-sectional hedonic
studies, the data for each house, encompass the last sales price and date, measures of certain
physical characteristics of the house, and measures of community characteristics. Observations
for more than one time period can also be pooled. In hedonic housing studies of hazardous
waste facilities, information on the location of each house relative to a waste facility is included
in most econometric specifications. The housing data for the analyses were obtained from two
sources. One source provided information based on property transfer information from local
government records. The information included address, sale value, and date of sale. Data were
available for homes that sold between 1982 and 1991, and included multiple sales of the same
residence. The second data source provided information based on tax assessor records of all
houses assessed in 1989. The information included address, sale value, date of sale, and data on
characteristics of the homes. The most relevant characteristics were the square footage, number
of bedrooms, number of bathrooms, lot size, age of the building, the number of stories, and the
presence of a pool.
For Facilities #1 and #2, the housing data were supplemented with information on the
socioeconomic characteristics of the neighborhoods where the houses were located. In hedonic
analysis, including this information is necessary where there is substantial heterogeneity since
differences in neighborhood characteristics can induce differences in housing values. Further
motivation for including socioeconomic characteristics comes from earlier research experience
*** DRAFT - March 23,1993
-------
11-7
which indicates that individuals respond differently to the same information about environmental
events. Some people care a lot and some care very little, at least as expressed in subjective
measures of risk. Differences in certain socioeconomic characteristics have been observed to be
correlated with such differences in subjective risk.4 While it was not possible to describe each
homeowning household, it was possible to use Census data to describe groups of households at
the block group level. This is the lowest level of aggregation used by the Census and it
encompasses 300 to 700 households in the cases of Facilities #1 and #2. These Census data
provided information on community characteristics (total population, racial composition, housing
stock occupancy status, housing tenure, and age composition for 1980 and 1990 and household
income for 1980).
In Exhibit 11-1, descriptive statistics are provided for three socioeconomic characteristics
(percentage of households who are non-white, percentage renter, and household size) of the
populations living Census block groups in the vicinity of Facilities #1 and #2 in 1980 and in
1990.s These statistics suggest that community characteristics exhibit enough heterogeneity -
across distance or time -- to warrant further attention in the regression specifications.
Furthermore, the populations around each facility may vary enough from each other to affect the
magnitude of the estimated impacts on property values since their underlying preferences may be
substantially different.
11.2.2 Methodology for Estimating Economic Benefits
To estimate the benefits of hypothetical interventions that would reduce the adverse
environmental effects of a TSDF, changes in property values were simulated using alternative
hedonic model specifications. In order to form such predictions 'from the hedonic regressions, the
analysis characterized hypothetical interventions in terms of factors that enter into the hedonic
analysis: distance and time. First, interventions were considered which would achieve reductions
in perceived risks and disamenities equivalent to moving each house a certain distance further
away from the TSDF. Next, interventions were considered which would reduce perceived risks
and disamenities to their level prior to a key event.
4 See, for example, G. H. McClelland, W. D. Schulze, and B. Hurd, The Effect of Risk
Beliefs on Property Values: A Case Study of a Hazardous Waste Site, manuscript, University of
Colorado (1989) or, V. K. Smith and W. H. Desvousges, The Value of Avoiding a LULU:
Hazardous Waste Disposal Sites, Review of Economics and Statistics. Vol. LXVIII, No. 2 (1986).
5 The statistics show how these characteristics vary among block groups both with distance
from the TSDFs and over time. Each block group typically consists of 1,000-2,000 people and
500 homes, but none within the vicinity of either of these two facilities sites exceeds 8,000 people
or 5,000 homes. The block group appears to be an appropriate level of aggregation because it is
small enough to reflect the demographic profiles of individual communities.
DRAFT - March 23,1993
-------
11-8
Intervention Modeled as Increase in Distance
Using the estimated coefficients from hedonic regressions, it is possible to compute the
benefits of hypothetical interventions that would achieve reductions in perceived risks and
disamenities equivalent to increasing the distance between each house and the TSDF. To be
more precise, suppose that the hedonic regression is estimated over the area within four miles of
the TSDF. To calculate the benefits of moving each house a distance d further away from the
TSDF, two predicted prices were computed for each house in EPA's dataset which lay within a
four-mile radius. First, to predict the price in the absence of intervention, the actual values of
distance D and housing characteristics X are substituted into the estimated regression equation
and T is set equal to January 1,1992. Then, to predict the price after the intervention, the same
calculation is performed except that distance is set equal to either D + d or four miles,
whichever is smaller. (This is equivalent to assuming that outside the radius of four miles over
which the hedonic regression is estimated, price does not vary with distance from the TSDF.)
These two predictions are aggregated over all houses within the four-mile radius. The difference
between the predicted aggregate property values with and without intervention provides an
estimate of the benefits of intervention.
DRAFT - March 23,1993 **
-------
11-9
Exhibit 11-1
TSDF Facility #1
Range
(miles)
0-1
1-2
2-3
3-4
Census
Year
1990
1980
1990
1980
1990
1980
1990
1980
Obs
68
68
613
613
2021
2021
810
810
Percent Non-
white
Mea Std Min-
n Dev Max
16 3 11-17
18 4 10-21
13 3 4-18
9 2 1-21
13 4 3-56
8 2 5-33
12 5 3-56
8 3 5-33
Percent Renter
Mean Std Min-
Dev Max
25 6 22-37
70 0 70-70
18 10 7-54
80 6 48-95
25 14 11-100
73 17 0-86
35 22 6-100
70 22 0-91
Household Size
Mean Std Min-
Dev Max
3.12 0.21 2.7-3.2
3.28 0.27 2.8-3.4
2.99 0.55 1.7-3.8
3.10 0.29 1.9-3.4
2.86 1.57 1.8-30.5
2.93 0.51 1.2-7.6
2.58 1.73 1.4-30.5
2.60 0.50 1.2-7.6
TSDF Facility #2
Range
(miles)"
1-2
2-3
3-4
Census
Year
1990
1980
1990
1980
1990
1980
Obs
67
67
332
332
805
805
Percent Non-white
Mean Std Min-
Dev Max
51 22 8-95
43 24 8-97
52 13 8-98
34 -14 11-98
46 18 26-97
46 28 16-98
Percent Renter
Mean Std Min-
De Max
V
54 9 30-75
49 10 26-59
48 13 29-76
52 13 27-79
35 19 7-94
64 22 9-96
Household Size ..
Mean Std Min-
Dev Max
2.72 0.68 1.8-6.2
2.40 0.31 1.9-3.6
2.97 0.36 1.7-3.5
2.62 0.32 1.8-3.2
2.65 0.52 1.6-3.4
2.77 0.36 1.8-3.6
" From the TSDF.
Note: For any given year and distance range, the difference between the means of 1980 and
1990 are statistically significant for all cases except Household Size in the 3-4 mile
range at Facility #1 and Percentage Non-white in the 3-4 mile range at Facility #2.
»** DRAFT - March 23, 1993
-------
11-10
A slightly different approach is to assume that the intervention provides the same protectioi.
to all houses within a certain distance. This hypothetical intervention can be simulated by increasing
the distance between the TSDF and each house in the four-mile radius to exactly four miles. The
benefit calculations involved are the same except that to predict the price after the intervention,
distance is set equal to four miles.
Intervention Modeled as Reversal of Event
To estimate the benefits of reversing the adverse environmental effects of a key event
hedonic regressions of the form [2] were estimated. The coefficients are used to predict the price of
each house (a) as if the event had not happened, and (b) given the event. To form these
predictions, the actual values of distance D and housing characteristics X are substituted into the
estimated equation, T is set equal to January 1, 1992, and PERIOD is set equal to 0 (for the
prediction without the event) or 1 (for the prediction with the event). The benefit estimates were
aggregated over all houses in our data set within the four-mile radius. Again, the difference between
the predicted aggregate property values under the two scenarios provides an estimate of the benefits
of the intervention.
11J Overview of Find ings
11.3.1 Hedonic Property Value Estimates
Statistically significant relationships between housing prices and distance to each of the thr*"
TSDFs have been identified. Housing prices tend to increase with distance from the facilities.
Changes in housing prices over time appear to be consistent with the hypothesis that key
contamination events can affect housing prices. The case that contamination at a particular point in
time affects housing prices appears to be strongest for Facility #1. Prior to this contamination
event, an insignificant or negative relationship was found between housing prices and distance to the
facility.
The strong price-distance relationships discovered at all three facilities contrasts with much of
the experience embodied in earlier work of this type, focused especially on Superfund sites in the
early 1980s, which found it difficult to identify such impacts. However, the estimated specifications
still exhibit a large amount of unexplained variation in housing prices. To investigate this concern
further, EPA's analysis tested several alternative specifications at two of the facilities (#1 and #2).
Following recommendations from the applied economics literature, these alternatives were based
upon the inclusion or exclusion of variables which capture "nuisance" effects in a regression model.4
In the case of Facility #1, this exercise did not significantly alter the estimated effects of distance on
housing prices. In the case of Facility #2, there was greater fluctuation across specifications,
especially with the inclusion of different combinations of variables describing neighborhood
characteristics. In light of their greater robustness, the results for Facility #1 are discussed in
greater depth.
6 F. T. Denton, "Data Mining as an Industry, Review of Economics and Statistics. Vol.
LXVII (February, 1985).
DRAFT » March 23,1993
-------
11-11
Exhibit 11-2 shows the results of various linear regressions for housing transactions taking
place near Facility #1 between 1983 and 1991.7-8 Coefficient estimates are presented with
corresponding t-statistics beneath in parentheses. Each specification controls certain housing
characteristics (lot size, an indicator when lot size information was missing, building size, number of
bedrooms and bathrooms, an indicator when a pool was present, age of the house), time indicators,
and distance (alone and interacted with a time period indicator). Additional analysis conducted on
this facility indicated that events at the facility may have influenced the price of houses. Mid-1986
was identified as an influential threshold since significant distance effects were not estimated prior to
this time. This timing effect may have been directly related to citations of the facility and to public
announcements regarding the need to close the facility at that time. EPA's analysis employed a
period indicator to distinguish between the time periods before and after mid-1986.
The first regression does not include any sociodemographic variables. This "baseline"
specification forms a basis for comparison to the other six regressions which include various
sociodemographic variables. In the baseline specification, the effect of distance on housing prices is
significant only after mid-1986. This specification implies an increase in property values of $11,790
per mile of distance from the facility.9
Although the F statistic for this specification rejects the hypothesis of no linear association,
the R2 statistic indicates that a large amount of variation is left unexplained. For this reason,
alternate specifications were estimated to determine how robust the coefficients of particular interest
(those associated with distance formulations) were. The absence of coefficient estimates for any
variable indicate that a variable was not included in a particular specification. While EPA
investigated every combination of the demographic variables, only selected regressions are presented.
The second regression, in linear form in Exhibit 11-2 and in semilogarithmic form in Exhibit
11-3, yielded the smallest benefit estimates.10 The third, fifth, and seventh include all three
demographic variables and combinations of their interaction terms. The fourth and sixth regressions
include the renter variable and combinations of its interaction terms. The sixth yielded the largest
benefit estimates in both forms. This specification implies an increase in property values of $13,735
per mile from the facility after mid-1986.
7 Exhibit 11-3 shows the results for analogous semilogarithmic specifications. These are
offered for comparison but are not discussed directly.
8 The housing price data were converted to constant year-end 1991 dollars using the CP1 for
homeowners' costs in the local area. Because the dependent variable has been deflated by this index,
the effects of distance and housing characteristics are measured in real terms.
9 This result is based on adding the coefficient on distance (-3.83) to the coefficient on the
interaction between distance and the time period indicator (15.62). The difference (11.79) is
measured in units of $1,000.
10 The ranking of benefit estimates was based upon the results for interventions equivalent to
moving each house one-mile.
DRAFT--March 23, 1993 ***
-------
11-12
These regressions demonstrate that the addition of the demographic variables did not nav<
large impact on the variables of interest or the benefit estimates, even though the demographic
variables were statistically significant in many cases. As in the baseline regression, the distance -
period indicator cross product remained significant at the five percent level. In addition, the
coefficients on the housing characteristics proved fairly robust. Note that changes in the coefficients
on distance and on the distance-period indicator interaction term are not directly interpretable in
specifications where socioeconomic characteristics are also interacted with distance. They must be
analyzed in connection with any terms where distance is interacted with a demographic variable (e.g.,
DRAFT - March 23,1993 **
-------
11-13
Exhibit 11-2
Facility #1, 1983-1991
Number of Observations:
Prices)
3475
Linear Specifications (House Price; $1000; 1991
Independent Variables
Intercept
Lot dau missing indicator
Lot size (acres)
Building size (sq.ft )
Number of bedrooms
Number of bathrooms
Presence of pool
Age (days)
Time Trend (days)
Period Indicator
(=0 before 1986 5; = 1 after)
Distance x period indicator
Distance (miles)
Non-white population (%)
Number of rented homes (%)
Avg. number of persons/home
Distance x non-white pop
Distance x rented homes
Distance x a vg. persons/home
Distance x period indicator
x non-white pop
Distance x period indicator
x rented homes
Distance x period indicator
x avg. persons/home
R2
F
1
-213.51
(-6.519)
13.11
(3.052)
0.54
(0.082)
0.07
(13.551)
-0.69
(-0.26)
-1.18
(-0.281)
-4.35
(-0.829)
-2.07e-03
(-6.594)
3.IOe-02
(9.739)
-63.35
(-4.153)
15.62
(2.791)
-3.83
(-0.768)
0.18
69.13
2
-201.06
(-5.952)
13.02
(3.028)
0.34
(0052)
0.08
(13.607)
-0.50
(-0.189)
-1.18
(-0-28)
-4.36
(-0.832)
-2.04C-03
(-6 474)
3.IOe-02
(9.727)
-6445
(-4.213)
12.90
(2.102)
-4.53
(-0.906)
-4.05
(-1-537)
1.09
(1.033)
0.18
58.70
3
-156.31
(-3.84)
12.42
(2.887)
0.53
(0.08)
0.08
(13.701)
-1.99
(-0.73)
-1.07
(-0.253)
-5. II
(-0 975)
-1.72e-03
(-5.061)
276e-02
(8.033)
-72.03
(-4.628)
1440
(1.906)
-3.10
(-0.609)
226.47
(1.998)
-4407
(-1.52)
-11.55
(-2.758)
6994
(-1.598)
4.09
(0.403)
3.42
(2-225)
0.18
45.60
4
-110.23
(-2 483)
12.18
0-834)
1.21
(0.184)
0.07
(13.467)
-1.22
(-0.457)
-1.80
(-0 429)
-4.42
(-0 843)
-1.77e-03
(-5.296)
2.7U-02
(8.023)
-84.61
(-5.108)
23.3*5
(3.837)
-22.21
(-2 61)
-168 58
(-3.155)
47.64
(2.68)
0.18
59.61
5
-39.32
(-0.536)
12.18
(2.828)
1.13
(0.173)
008
(13.501)
-1.30
(-0 479)
-1.89
(-0.446)
-4.32
(-0823)
-I.7U-03
(-5.061)
2.57e-02
(7.21)
-8585
(-4.981)
23.73
(3.719)
-40.06
(-2 148)
77.26
(0.317)
-186.85
(-32)
-20.52
(-1.015)
-6.46
(-0.078)
52.80
0.718)
5.70
(0.856)
018
45.83
6
-100.68
(-2 26)
11.97
0785)
047
(0.071)
0.07
(13.571)
-1.46
(-0 548)
-1.75
(-0.417)
-4 14
(-0 789)
-1.67e-03
(-4 973)
2 71e-02
(8024)
-91.99
(-5-46)
37.16
(4.399)
-35.73
(-3.482)
-185 33
(-344)
77.17
(3.549)
-27.95
(-2.355)
0.18
55.83
7
-43.77
(-0.596)
12.01
(2.791)
0.53
(0.081)
0.08
(13 686)
-1.87
(-0.688)
-168
(-0.397)
-4.08
(-0777)
-].6le-03
(-4 698)
260e-02
(7.3)
-92.06
(-5.278)
34.26
(3541)
-47.43
(-2451)
6.96
(0.028)
-201 .93
(-3.445)
-14.79
(-0727)
114.67
(1.23)
73.52
(3.183)
0.55
(0.08)
118.43
(-2.366)
-19.39
(-1.555)
4.10
0-462)
0.19
39.59
-------
11-14
Number of Observations: 3475
Exhibit 11-3
Facility #1, 1983-1991
SemUog Specifications (Log(Housing Price); $1000; 1991 Prices)
Intercept
Lot data missing indicator
Lot size (acres)
Building size (sq.ft.)
Number of bedrooms
Number of bathrooms
Presence of pool
Age (days)
Time Trend (days)
Period Indicator
(=0 before 1986.5; = ! after)
Distance x period indicator
Distance (miles)
Non-white population (%)
Number of rented homes (%)
Avg number of persons/home
Distance x non-white pop
Distance x rented homes
Distance x avg. persons/home
Distance x period indicator
x non-white pop
Distance x period indicator
x rented homes
Distance x period indicator
x avg. persons/home
R2
p
1
2.62
(15.172)
0.07
0.295)
-0.04
(-1.264)
3.49e-04
(11.972)
-0.01
(-0.37)
0.04
(1.696)
-0.02
(-0.707)
-1.46e-05
(-8.805)
l.90e-04
(11.333)
-0.23
(-2.829)
0.06
(2.174)
4.10e-03
(0.156)
0.22
86 68
2
2.64
(14.781)
0.07
(3.285)
-0.04
(-1.269)
3.49e-04
(11.966)
-0.01
(-0.363)
0.04
(1.693)
0.02
(-0.71)
-1.46e-05
(-8.758)
1.90e-04
(11.326)
-0.23
(-2.842)
0.06
(1.859)
3.35e-03
(0.127)
-4.20e-03
(-0.302)
l.64c-03
(0.295)
0.22
73.32
3
2.89
(13 437)
0.07
(3.143)
-0.04
(-1.238)
3.57e-04
(12.153)
-0.01
(-0.949)
0.04
(1.682)
-0.02
(-0.871)
-1.25e-OS
(-6.939)
1.7U-04
(9417)
-0.26
(-3.169)
0.07
(1.684)
4.80e-03
(0.178)
0.62
(1.042)
-0.17
(-1.103)
-0.03
(-1-243)
-0.07
(-0.322)
-0.01
(4.195)
0.01
(0.83)
0.22
56.78
4
3.36
(14.365)
0.07
(3.017)
-0.04
(-1.124)
3.45e-04
(11.856)
4.01
(-0.595)
0.03
(1.517)
-0.02
(-0.705)
-1 .26e-05
(-7.18)
1.64c-04
(9 192)
-0.38
(-4.377)
0.12
.X3-759)
' -0.13
(-3.001)
-1.23
(-4.383)
0.36
(3.834)
0.22
7547
5
3.69
(9.54)
0.07
(3.016)
-0.04
(-1.117)
3.50e-04
(11.909)
-0.01
(-0 719)
0.03
(1.513)
4.02
(-0.697)
-1.24e-05
(-6.96)
1.56e-04
(8.305)
-0.39
(-4.242)
0.12
(3.617)
-0.22
(-2.245)
0.57
(0.44)
-1.30
(-4.225)
-010
(4.954)
4.06
(4.14)
0.38
(3.7)
0.03
(0.852)
0.22
57.82
6
3.43
(14.625)
0.07
(2.951)
4.04
(-1.282)
3.50e44
(12.014)
4.01
(4.724)
0.03
(1.536)
4.02
(4.63)
-1.19e4S
(-6.735)
1.64e44
(9.2)
444
(-4.919)
022
(4.993)
4.23
(-4-331)
-1.36
(-4.785)
0.58
(5.033)
4.21
(-3.289)
0.22
71.05
3.65
(9-432)
0.07
(2-967)
4.04
(-1.261)
3.55e44
(12.04)
4.01
(4.867)
0.03
(1.536)
4.02
(4617)
-1.18e4S
(-6.522)
1.59e44
(8445)
443
(-4.695)
0.22
(4.32'
4.2'.
(-2.882)
0.09
(0.072)
-1.39
(-4.512)
-5.99e42
(4.558)
0.29
(0.589)
0.57
(4.709)
0.01
(0.315)
4.26
(4.986)
4.19
(-2.912)
7.37e43
(0.84)
0.22
49.80
-------
11-15
distance times rented homes, in regression specification number 4). The benefit estimates
themselves are a simpler indicator of total changes in the estimated relationship between housing
prices and distance than are the coefficients. As is discussed below, the benefits estimated from
these regressions were no more than 35 percent greater or 6 percent less than the baseline's,
indicating only modest changes in the price-distance relationship across specifications and therefore
relatively robust results.
The inclusion of demographic variables did influence the significance of the distance
coefficients. The distance coefficient is not significantly different from zero in the first (baseline)
specification but is significant at the five percent level in the fourth, fifth and seventh regressions and
at the one percent level in the sixth regression. The changed significance of this coefficient indicates
that the addition of demographic variables has helped to establish the price - distance relationship
more firmly. By contrast, investigation of similar regressions for Facility #2 has shown less
consistency in the relationship between the demographic variables and housing prices. Although the
property values increase more per mile from Facility #2 than from Facility #1, this finding is
sensitive to the inclusion of different demographic variables in the regression.
Of all the demographic variables, the percent of renters in an area produces the most
consistent results. The descriptive statistics in Exhibit 11-1 help explain this phenomenon. From
1980 to 1990, the percentage of rented homes dropped dramatically within four miles of Facility #1.
It is clear that this wholesale shift towards home ownership can be associated with a change in the
housing price structure of the area. It is important to note, however, that while this shift might have
raised prices, it also suggests that the population in 1990 was very different from the one in 1980.
The population inhabiting the area in 1990 may have stronger preferences for environmental safety.
Exhibit 11-3 shows the analogous regressions for the semi-logarithmic specifications. Many
of the above observations apply to these results, including little change in the R2 statistic, the
robustness of the housing characteristic coefficients, and the significance of the distance, period
indicator, and demographic variables and their cross products. The benefit estimates based on these
regressions were no more than 24 percent greater or 6 percent less than the baseline, again
indicating only small changes in the price - distance gradient. As is discussed below, the benefits
estimated from both the linear and semi-logarithmic specifications are comparable, and there is no
clear indication that one functional form is preferable to the other.
EPA's investigation of the linear and semi-logarithmic specifications included regressions on
datasets from which extreme observations had been removed. The exclusion of observations has
been tested only on an experimental basis. Using arbitrarily defined extreme values, a very small
portion of the data was removed (approximately two percent). From this exclusion, the Rz statistic
improved by as much as 20 percentage points in the linear case. Despite this radical change in the
goodness of fit of the linear specification, the benefit results remained stable. For example, the
benefit estimate for regression number 6, the one generating the largest benefit estimates, fell by
only ten percent, to $12,318 per mile. These results provide added evidence of the potential
robustness of the regression estimates with respect to reflecting the impact of Facility #1 on property
values.
DRAFT - March 23, 1993
-------
11-16
11.3.1 Benefit Estimates
Several pieces of evidence indicate that the current set of estimated property value effects are
relatively robust. Still, the preliminary nature of EPA's investigations suggests that the current
estimates of benefits should be taken as indicative rather than conclusive. They are presented
primarily to show that there is likely to be a measurable impact from the types of facilities
considered in these analyses on the value of nearby residential properties. For all three facilities,
indications are that the impacts could be substantial. In this section are summarized the findings for
each facility, with the greatest attention to Facility #1, where the regression estimates are most
consistent.
For Facility #1, benefits were calculated based upon hypothetical interventions equivalent to
moving each house one mile further from the facility, to moving each house to a distance of four
miles from the facility, and to reversing the "key event" at the facility which changed the valuation of
distance. For the one-mile and four-mile interventions, these results were calculated for each of the
seven regression specifications presented in Exhibit 11-2 and Exhibit 11-3. For the "event-reversal"
intervention, the results were calculated from regressions employing these same specifications but
using only a subsample of the sales data around the event (limited to the 1984-1988 timerrame).
The resulting per house benefit estimates are presented in Exhibit 11-4.
For the one-mile equivalent intervention, the estimated benefits per house range from $6,765
to $9,715 (in constant 1991 dollars). The interpretation that can be given to these figures is that
they represent the expected average change in the value of a house sold in the 1983-1991 timeframe
had the intervention taken place prior to all sales (i.e., prior to 1983). Since some of the sales
included in the 1983-1991 data include transactions prior to the time of the "event," before which
there was only a small or even a negative effect of distance from the facility on property values, anu
since houses already three to four miles away from the facility experienced little effect from this
hypothetical intervention, the estimated per house benefits are smaller than the $11,790 to $13,735
increase per mile cited above. This range applies only to sales after mid-1986 and to full one-mile
increments.
For the four-mile equivalent intervention, the estimated benefits per house range from
$10,702 to $14,099. Although the hypothetical intervention appears to be four times greater than the
previous one, the estimated benefits are not four times greater since very few houses will receive the
full effect equivalent to be moved four miles. Eighty percent of the housing transactions in the
sample were located two to four miles from the facility already. For the event-reversal intervention,
the estimated benefits range from $8,012 to $25,973.
It should be noted that the per house estimates are consistent with the results of earlier
studies. The average damages per household found in the site-specific studies (i.e, those which
estimate proximity effects) ranged from $730 to $35,000 (median: $6,300)." Conservatively
11 This range was based upon estimates reported in Harrison and Stock, 1984; McClelland,
Schulze, and Kurd, 1989; Michaels and Smith, 1990; and Thayer, Albers, and Rahmatian, 1991.
See Bibliography for full citations.
** DRAFT - March 23,1993 ***
-------
11-17
applying the figures based upon the one-mile intervention to all owner-occupied homes within two
miles of the facility in 1990, implies benefits ranging from $17 million to $21 million. Exhibit 11-5
provides a breakdown, by distance from the facility, of estimated benefits per house and aggregate
benefits from the one-mile and four-mile equivalent interventions. The "No. of Observations" refers
to the number of transactions in the sales data set which were associated with houses within a given
one-mile ring. The aggregate benefits were calculated using the number of actual occupied homes or
the number of owner-occupied homes within each one-mile ring, as of the 1990 Census. This exhibit
illustrates that less conservative assumptions about the effectiveness of an intervention reducing the
property value effects of Facility #1 could translate into much larger aggregate benefits than those
presented above.
For Facility #2, based upon similar hypothetical interventions as Facility #1, the estimated
benefits per house range from $119 to $13,789 for the linear specifications of the one-mile
intervention, and $276 to $18,309 for the same specifications of the four-mile intervention. These
results are based upon all transactions within four miles of the facility. However, once restricted to
houses within two miles only, some of the specifications imply negative benefits. For the one-mile
intervention, the range of estimated impacts is $3,577 to $18,923. Applying these estimates to the
population of owner-occupied homes within two miles of this facility implies benefits ranging from
$11 million to $59 million. However, once the aggregation extends to all owner-occupied houses
within three miles, the aggregate benefits range from nearly $2 million to $200 million. Clearly these
results are very sensitive to the regression specification employed. For Facility #3, only a narrow
range of specifications was evaluated. The resulting benefit estimates range from $2,000 to $3,000
per house, depending upon the intervention considered.
11.4 Conclusions
This chapter reports the results of an exploratory analysis to examine the effects of
contamination at TSDFs on the values of nearby residential properties. The analysis employs the
framework of hedonic housing analysis, which has been applied to contaminated sites in other studies
but where the statistical results have been mixed, ranging from small and insignificant effects to large
and significant impacts on residential properties. The results from this analysis generally made a
stronger case than those of earlier studies, supporting the assertion that hazardous waste
contamination can impose economic damages that are expressed in residential real estate
transactions. At the three sites studied, the estimated effect ranged in value from approximately
$100 to $18,000 per house, depending on the facility and regression specification in question. Using
the results of the most robust specifications, aggregate benefits from an intervention that would
mitigate the effects of a TSDF on nearby properties were estimated to be on the order of $20
million.
The property value results may contribute to an improved understanding of the economic
damages imposed by TSDFs, but the analysis should still be viewed as a work-in-progress. The
results are exploratory and not yet conclusive. They derive from housing market behavior at three
TSDFs only and are not intended to represent the entire universe of TSDFs and their potential
impacts on real estate markets. Nonetheless, the estimated benefits coincide with findings in
separate studies of contaminated sites and residential property markets. Furthermore, in this study,
DRAFT - March 23, 1993 ***
-------
11-18
much more extensive datasets were employed over a wider array of sites than in earlier studies. 1
current effort appears to have generated more robust estimates of property value impacts as a rest
The remedial activities to be initiated in response to the corrective action regulation have the
potential to mitigate the negative impacts that TSDFs impose on nearby residents. While this
analysis has indicated that those negative impacts can be substantial, it is still difficult to determine
the degree to which these impacts can be mitigated by corrective action. Still, in cases where
corrective action activities transform the operation and appearance of a TSDF in a major way, it is
expected that the range of estimated benefits presented here would provide a relevant gauge against
which the economic costs of those activities can be measured.
DRAFT - March 23,1993 ***
-------
11-19
Exhibit 11-4
Facility #\
Period: 1983-1991
Observations = 3,475
Aggregate Property Value
No Intervention
Moving Each House 1 Mile
Moving Each House To 4 Miles
Benefit of Intervention
Moving Each House 1 Mile
(Benefit Per Home)
Moving Each House To 4 Miles
(Benefit Per Home)
Even Revena] latervealion
Under Pre-Event Conditions
Under Post-Event Conditions
Change in Property Value with
Intervention and Changes in
Community Demographics
(Benefit Per Home)
Linear Specifications
(SI. 000)
1
603.294
628,296
642.645
25.002
7.195
39.351
11.324
829.019
738.764
90.255
25973
2
603.286
626.795
640.474
23,509
6.765
37.188
10.702
826.420
736.172
90,249
25.971
3
603.321
633.173
650.252
29.851
8.590
46.931
13.505
765,010
704.396
60,614
17443
4
603,336
636,534
651,836
33,198
9.553
48,500
13.957
762,047
706,231
55,816
16.062
5
603,324
634,982
649,965
31.658
9.110
46,641
13.422
699,412
671,571
27,841
8012
6
603.322
637.083
652.314
33,761
9715
48.993
14099
775.037
711,912
63,125
18 165
7
603.317
636,068
651.392
32,750
9.425
48,075
13.835
709.448
675,553
33,895
9754
* DRAFT -- March 23, 1993
-------
Exhibit 11-5
Facility ftl
Period: 1983-1991
Linear Specification (in $1000)
Regression
Distance: 0-1 miles
No. of Observations:
1
2
3
4
5
6
7
Distance: 1-2 Miles
No. of Observations:
1
2
3
4
5
6
7
Distance: 2-3 Miles
No. of Observations:
1
2
3
4
5
6
7
1 Mile: Benefit
Per Home
68
8.350
8.164
8.401
10.725
11.626
10.994
13.530
610
8.467
8.187
10.173
7.873
7.913
7.878
7.934
2019
7.076
6.609
8.501
9.456
9.050
9.533
9.283'
4 Mile: Benefit
Per Home
27.840
27.227
27.836
35.371
38.561
36.159
45.154
20.014
19.334
24.094
18.967
18.927
19.022
19.022
10.038
9.375
12.100
12.853
12.236
12.862
12.458
1 Mile: Aggregate
No. of Owner
Occupied Homes
2,238
2,188
2.251
2,874
3,116
2,946
3,626
15,536
15,022
18,667
14,446
14,520
14,455
14,559
65,728
61,395
78,965
87,835
84.062
88,549
86,226
Benefits Based on:
No. of
Occupied Homes
2,881
2,817
2,898
3,700
4,011
3,793
4,668
23,275
22,505
27,965
21,642
21,752
21,655
21,811
98,532
92,036
118,375
131,672
126,016
132,743
129,260
4 Mile: Aggregate
No. of Owner
Occupied Homes
7,461
7,297
7,460
9,479
10,334
9,691
12.101
36,725
35.477
44.212
34,805
34,730
34,906
34,906
93,243
87,081
112,397
1 19,389
113,659
119,475
115,719
Benefits Based on:
No. of
Occupied Homes
9,605
9,393
9,603
12,203
13,304
12,475
15,578
55,017
53,148
66,234
52,141
52,029
52,292
52,292
139,779
130,541
168,493
178,974
170,385
179,103
173,472
-------
Exhibit 11-5,
Facility #1 (Continued)
Linear Specifications (in $1000)
Regression
Distance: 3-4 Miles
No. of Observations:
1
2
3
4
5
6
7
Total, all homes
No. of Observations:
1
2
3
4
5
6
7
1 Mile: Benefit
Per Home
804
6.198
5.741
7.352
10.666
9.664
11.147
10.260
3501
7.141
6.715
8.526
9.483
9.043
9.643
9.355
4 Mile: Benefit
Per Home
6.198
5.741
7.352
10.666
9.664
11.147
10.260
11.240
10.622
13.405
13.853
13.322
13.994
13.732
1 Mile: Aggregate
No. of Owner
Occupied Homes
33,873
31,373
40,180
58,289
52,812
60,916
56,071
117,375
109,978
140,063
163,444
154,509
166,867
160,482
Benefits Based on:
No. of
Occupied Homes
54,693
50,656
64,876
94,116
85,272
98,357
90,534
179,381
168,013
214,114
251,130
237,050
256,548
246,273
4 Mile: Aggregate
No. of Owner
Occupied Homes
33,873
31,373
40,180
58,289
52,812
60,916
56,071
171,302
161,227
204,249
221,962
211,535
224,987
218,797
Benefits Based on:
No. of
Occupied Homes
54,693
50,656
64,876
94,116
85,272
98,357
90,534
259,094
243,738
309,206
337,434
320,989
342,227
331,877
-------
12. CHANGES IN THE VALUE OF FACILITIES
This chapter of the RIA addresses the benefits of hazardous waste facility remediation
required under corrective action that may accrue directly to facility owners. These benefits are in
the form of increased facility value resulting from remediation of on-site contamination. This
analysis estimates the facility value benefits at the national level for those facilities requiring
some remediation under corrective action.
The purpose of this chapter is primarily to illustrate the approach used to assess the
facility value benefits. The chapter reports the preliminary results of the analysis of the proposed
corrective action requirements. The analysis will later be refined and expanded to address
regulatory options.
The primary conclusions of this analysis are summarized below.
The estimated national facility value benefits are $280 million. Approximately
2,200 of the 2,600 facilities requiring remediation contribute to these estimated
facility value benefits.
Most of these benefits are found at a relatively small number of facilities. On-site
remediation at about 470 facilities, roughly 18 percent of the facilities requiring
remediation under corrective action, produces approximately 80 percent of the
total national benefits.
Sensitivity analysis on the area of facilities affected by contamination suggests that
the national estimate could be as high as $610 million.
The following sections present the analytic framework for evaluating facility value
benefits, including the economic theory that serves as the foundation for this study and the
specific calculations required. Then, the results of this analysis for the sample facilities and
development of national estimates are discussed. Finally, the limitations inherent in the
methodology and their implications for the estimate of facility value benefits are described.
*** DRAFT - March 25, 1993 ***
-------
12-2
12.1 Economic Framework
12.1.1 Theory of Property Values
Corrective action requires owners of permitted hazardous waste treatment, storage and
disposal facilities (TSDFs) to remediate on- and off-site contamination. The calculation of the
costs of complying with the corrective action requirements and of the benefits to the surrounding
community are relatively well understood in theory.1 However, the economic literature and
previous benefit-cost analyses have not addressed how to appropriately treat benefits that accrue
directly to the facility owner.
For the purposes of benefit-cost analysis, the proper comparison is between the economic
welfare of all affected individuals with and without corrective action requirements, holding all
other things constant. The economic welfare associated with commercial property ownership
depends on the streams of future costs and revenues associated with the facility. If an
environmental regulation causes costs to increase or revenues to decrease, then it reduces the
welfare of the facility owner - and these effects should be included as part of the costs of the
regulation. The opposite is also true: increases in revenues or decreases in costs enhance the
welfare of the owner and should be considered as benefits of the regulation.
The theory of property values states that the market value of a commercial property will
be equal to the discounted present value of the anticipated flow of net revenues (that is,
revenues minus costs) realized from operating the facility in its most valuable use. The key word
here is "anticipated" -- money already spent on existing structures and improvements represents
sunk costs and is irrelevant for the calculation of facility value. For example, the market value of
an apartment building is not necessarily equivalent to the amount originally spent to purchase it,
rather it is the present value of anticipated future rent receipts minus the expected costs of
operations.
In addition, economic theory states that property value is based on the facility's "highest
value" use, i.e., the use that will accrue the largest net revenues. Because owners generally wish
to maximize their wealth, the current use of the property will generally be the same as its best
use given current and expected future facility, market and regulatory conditions. For example, if
managing hazardous wastes is not (or is no longer) the most profitable use of the land, the owner
is likely to switch businesses regardless of federal requirements.
One strategy for measuring the cost of regulatory requirements is to calculate the
reduction in the discounted present value of the stream of net revenues, that is, the difference in
the present values of net revenues with and without the requirements. A different strategy is to
examine reductions in the market value of the facility itself once the requirements are imposed.
Economic theory suggests that these two measures should be equivalent, using the logic
described above.
1 These costs and benefits are addressed elsewhere in this RIA.
DRAFT - March 25, 1993 **
-------
12-3
Economists generally believe that the imposition of regulatory requirements that restrict
the use of a facility or require costly actions which the owner would not otherwise undertake
cannot make the facility more valuable to the owner. The logic is straightforward - if
expenditure of one dollar increases the value of the facility by more than one dollar, owners
would voluntarily undertake these expenditures without regulation. However, one can envision
situations where the expenditure of one dollar increases the facility value by a positive amount
but by less than one dollar. In such cases, the increase in facility value should be included either
as an offset to the regulatory requirement's cost or as a benefit of the requirement.
12.1.2 Application to Corrective Action1
The impact of corrective action requirements on the value of a facility that is permitted
to treat, store, or dispose of hazardous waste is illustrated in Exhibit 12-1. The value of such a
facility at any time is the net present value of all anticipated future costs and revenues associated
with the use of the facility. The original value (V°) of the facility decreases with the imposition
of corrective action requirements, because future expenditures must increase by the cost of
remediation (C).3 Once the owner completes the remediation, the value of the facility rebounds
to the original market level because additional future expenditures are no longer required. Note
that this increase from the value after corrective action back to the original, pre-regulation value
is not a benefit of the corrective action requirements, but rather is simply a measure of the
owner's required compliance expenditures.4 The increase in the value of the facility is exactly
offset by the facility owner's expenditure on remediation.
In addition, if remediation also causes future net revenues'to rise, this anticipated net
revenue increase will be reflected in the facility value, increasing it beyond V°. In such a case,
the facility value after remediation would be greater than the original market value before
corrective action. In theory, however, if this increase is greater than the clean up costs, the
facility owner would have undertaken the remediation to maximize net revenues even without
corrective action requirements. This increase in facility value above the original value (shown as
the shaded area (VB) in Exhibit 12-1) represents a benefit of the requirements imposed by
corrective action and is the focus of this analysis.
The representation of change in facility value over time in response to corrective action in
Exhibit 12-1 is simplified for ease of discussion. The actual shape of the graph may vary
significantly across facilities. For example, the imposition of corrective action requirements may
2 The analytic approach presented in Sections 12.1.2 and 12.1.3 of this chapter was
developed in collaboration with A. Myrick Freeman III of Bowdoin College and Raymond P.
Palmquist of North Carolina State University.
3 TSDF owners must remediate even if they plan to close and change businesses.
4 Because the imposition of the requirements caused the initial drop in market value for the
facility, one cannot count the resulting increase in facility value back to previous levels as a
benefit of the requirements.
*** DRAFT - March 25,1993 ***
-------
EXHIBIT 12-1
FRAMEWORK FOR FACILITY VALUE BENEFITS ANALYSIS
Facility Value
(dollars)
- C
Imposition of
CA
Requirements
Remedy
Completed
Time
V°
VB
Facility value prior to the imposition of corrective action requirements
Increase in facility value benefit
Cost of corrective action remediation
-------
12-5
lead to a decrease in facility value over time as the RCRA Facility Investigation (RFI) is
completed and the full extent of contamination is realized, rather than the immediate decrease
shown in the exhibit. Similarly, the rebound of facility value that accompanies remediation may
occur gradually as the remedy is announced and progresses toward completion. The magnitude
of the remedial cost in comparison to the facility value also may vary. In fact, the cost of
remediation may exceed the value of the facility, resulting in the facility value dropping below
zero prior to remediation. None of these potential variations in the impact of corrective action
on facility value over time affect the analysis in this chapter. The decrease in facility value
resulting from corrective action requirements is always exactly offset after remediation has been
completed.
In general, there are two ways for corrective action to result in increases in facility values
(VB). First, corrective action may affect net revenues in cases where a facility continues to be
used as a TSDF or for another industrial purpose. The contamination at these facilities primarily
affects Solid Waste Management Units (SWMUs), soils and ground water, and may or may not
influence how the facility itself is used. However, where use of a portion of the facility is
restricted due to health risks or physical limitations, facility value may be reduced. By requiring
remediation and making that area safer or more productive, corrective action may increase future
net revenues.
Second, under corrective action some facility owners may be required to remediate to
levels that make the property suitable for alternative uses (e.g., residential or commercial use)
that would not have been possible prior to remediation. If an alternative use in fact will occur
given the location of the facility, local market conditions and other factors, then any resulting
increases in net revenues will be a benefit of the corrective action requirements ~ subject to the
earlier caveat that increases in value in excess of the costs of corrective action indicate that
facility owners are likely to clean up even in the absence of federal requirements.
In practice, it can be difficult to separate out changes in facility productivity or use that
can or cannot be attributed to the requirements imposed by corrective action, and to determine
the value of these changes. For example, one facility considered in this analysis is a military base
that' is closing.5 Some proportion of the value of the change in use for this facility is attributable'
to defense policy, not to the corrective action requirements. At other facilities, part of the
remediation is being conducted under other authorities (e.g., state programs or CERCLA) and
only a portion of the resulting change in facility value is attributable to corrective action.6
5 Note that the ownership of the facility does not affect how facility value is calculated. In
theory, the value of a facility is the same whether it continues under current ownership or is sold.
In addition, the approach used to estimate the costs and benefits of the corrective action
requirements is the same regardless of whether the facility is federally or privately-owned; i.e.
owned by taxpayers, a corporation or an individual.
6 This analysis is focused on changes in facility value related to SWMUs that are affected by
Subpart S corrective action. However, when considering the impacts of ground-water
contamination on facility use, it was not possible to distinguish between the contributions of units
being remediated under Subpart S and those being remediated under other authorities.
*** DRAFT - March 25, 1993 ***
-------
12-6
12.1.3 Numerical Examples
As noted above, the value of a facility decreases by the cost of remediation after the
imposition of corrective action requirements, and this decrease in value is clearly attributable to
the requirements. If the required remediation also will increase future net revenues beyond the
pre-corrective action level, these incremental net revenues are attributable to the requirements
and should be included as either an offset to costs or as benefits in related benefit-cost analyses.
This situation is illustrated below.
Suppose that, for a given facility:
V° = Facility market value before the imposition of
corrective action requirements = $100
C = Present value of the cost of complying with
corrective action requirements= $15
then,
V° - C = Facility market value after imposition of corrective
action requirements, but before remediation = $85
Prior to the imposition of corrective action requirements the facility's value is V° or $100,
the present value of net revenues at the facility. Then the requirements are imposed with a
present value of C or $15. The value of the facility immediately'drops to $85 to reflect the
owner's liability for the remediation cost. The owner has no choice but to comply, so he or she
spends C and the facility value returns to V° or $100.
However, if net revenues for the facility increase after remediation is completed, the
facility owner gains a benefit. For example, suppose that the owner can achieve net revenues of
$110 after compliance with corrective action requirements. This provides $10 more in net
revenues than the $100 possible before remediation (VD = $10). Because compliance with the
requirements has created a higher value for the facility than was possible in the facility's pre-
compliance state, the $10 gain is a benefit of corrective action.
Note that by changing the example slightly, one can illustrate cases where there would be
no increase in facility value attributable to the imposition of corrective action requirements. For
example, if post-remediation net revenues are the same as those before corrective action was
required (V° or $100), no benefits would accrue. Alternatively, if the present value of net
revenues after remediation were $120, a $20 increase over the value before remediation (V°) and
a $5 increase above the costs of remediation (C), then the owner would clean up even without
the corrective action requirements. The increase in value in this case could not be attributed to
corrective action because the remediation would be caused by the desire to maximize wealth, not
by the implementation of federal requirements.
DRAFT - March 25, 1993 ***
-------
12-7
12.2 Analytic Approach
The analyses that comprise this RIA are based on a sample of 79 hazardous waste
facilities, 52 of which will require some remediation under corrective action.7 The sampled
facilities represent roughly 5,800 facilities nationwide, of which approximately 2,600 will require
remediation. The costs of complying with corrective action requirements have been estimated by
convening expert panels to evaluate the potential scope and cost of remedial activities at these
facilities.8
As discussed above, benefits related to changes in facility value will result from increases
in future net revenue streams (VB in Exhibit 12-1). The theory suggests that such increases in
value can have two components:
Owners may find that facility remediation provides additional usable land on-site,
thereby increasing the net revenue stream from the facility (V,B). Once
remediated, this land becomes available for maintaining, improving or expanding
facility operations. Without remediation, an owner may have to use another
portion of the facility or even purchase new land to meet these needs.
An alternative use of the property may become possible after corrective action
remediation (V2B). A change to an alternative use would occur only if the
anticipated net revenues under the new use were larger than those under the
current use.
s
Separate approaches for evaluating the benefits from these two types of changes in
facility value were developed for the purposes of this analysis. First, the value to the owner of
having increased usable land at the facility was estimated as the market value of the area of
SWMUs and any surrounding soil contamination that would be remediated under corrective
action requirements.9 Here, it was assumed that use of land at the facility for continued
industrial purposes would be impaired only by surface contamination (i.e., SWMUs and
surrounding soil contamination) - ground-water contamination does not influence an owner's
ability to use the facility.
Second, where higher value alternative uses (e.g., commercial, residential or recreational
development) seemed plausible, the potential increase in sale value of the area of the facility
affected by contamination served as a measure of the increased facility value. For this pan of
7 See Chapter 3 for a discussion of this sample.
8 See Chapters 4 and 5 for a discussion of the expert panel facility evaluations.
' The area of SWMUs being remediated was determined from the summary RFIs and expert
panel reports prepared for this RIA, as described in Chapter 4. When available, the full area of
soil contamination around a SWMU that required remediation was included in the analysis.
*** DRAFT - March 25, 1993 »**
-------
12-8
the analysis, it was assumed that the use of land at the facility would be impaired by both
ground-water and surface contamination.10
The facility value benefits from increased usable land (V,B) occur once surface
remediation is complete. The benefits from higher value alternative uses (V2B) are realized only
after the ground-water remediation is complete. Together, these two types of benefits represent
an estimate of the total corrective action benefits related to increases in facility value (i.e., V,B +
V2B = VB).n The specific methodologies for completing these two elements of the analysis are
described in greater detail below.
12.2.1 Benefits Related to Increased Usable Land
Corrective action facility value benefits resulting from increases in usable land were
estimated from the market value of areas of SWMUs and surrounding soil contamination
requiring remediation. The benefits can be expressed as:
PI x 0, = V,B (1)
Where:
P, = Price of comparable replacement industrial land
Q, = Area of surface contamination (i.e., SWMUs and surrounding soil
contamination) where remediation allows for increased use
V,B = Added facility value related to increased usable land
In general, the price of replacement land (P,) for each facility was based on market prices
quoted by local realtors for parcels of comparable industrial land.12 To the extent possible,
these parcels were similar to the area of surface contamination, having equivalent proximity to
10 Converting a facility to an alternative use requires investment to change its configuration.
The presence of ground-water contamination can greatly reduce the willingness to make such an
investment, especially if there are substitute properties with no ground-water pollution that could
be used for similar purposes.
11 When combining these two types of benefits, each is discounted over the appropriate time
horizons to provide comparable net present values.
12 Market prices were based on actual sale prices when available. Asking prices were used
only if no sale prices were available. At some facilities, no market values could be determined.
For example, there were facilities for which no comparable land sales could be found because
they are located in isolated areas where there is little demand for property, or the on-site area
affected by contamination was significantly larger than other parcels in the vicinity of the facility.
Under these circumstances, the assessed value of the land (without buildings) was used.
*** DRAFT - March 25,1993 ***
-------
12-9
town centers, and access to transportation (i.e., rail, roads and waterways) and utilities (i.e.,
sewer, water and electricity). Available access to transportation was determined from facility
maps provided in the RFIs. The determination of access to utilities was less exact, relying on the
location of the surface contamination relative to facility operations as an indication of the
proximity of utility hook-ups. For example, when an area of surface contamination was located
in a remote area of a large facility, the replacement price was based on unimproved industrial
land with poor access to public roads and utility hook-ups. Replacement values for areas of
surface contamination located near the operating area of a facility were based on prices for
improved industrial land with easy access to roads and utilities."
The area of SWMUs and surrounding soil contamination where remediation will increase
future use (Q,) was determined from the RFIs and expert panel reports. These documents
provided the following information on each SWMU: status (i.e., active or inactive), size, location
and recommended remedial action. For each facility we summed the areas of those SWMUs
meeting the following criteria:
Remediation is required under corrective action;
The specific remedy selected by the expert panel will result in the area of
the SWMU and surrounding soil contamination becoming usable (as
defined below); and
The area of the SWMU and surrounding soil contamination is large
enough to provide a meaningful increase in usable=,area at the facility.
Usable area includes inactive SWMUs and surrounding soil contamination where
remediation through excavation, land farming, soil vapor extraction and/or capping was
recommended by the expert panel. While not all of these units may be suitable for unrestricted
future use, they were considered as part of the estimate of usable areas because of their potential
for limited use, even if only as a parking lot or buffer zone. Inherent in this approach is the
assumption that owners cannot use these areas prior to remediation. Therefore, the areas are of
no value until they become usable after remediation.
The area of active SWMUs and surrounding soil contamination was included when
remedial action would allow for their continued use. This decision rule was based on the
premise that, in the absence of remediation, an owner would need to close and replace the unit.
Through completing the remedial action, the SWMU and surrounding area remains usable and
the land that would have been used to replace the unit becomes available for other purposes.
However, if the remedial action at an active SWMU includes closing the unit, there would be no
net gain in usable land, under the assumption that it would need to be replaced.
13 Improved industrial land includes land that is ready for construction. Utility hook-ups are
available, but no structures have yet been built.
*** DRAFT - March 25, 1993 ***
-------
12-10
Values of Qa among the 52 sample facilities ranged from zero to 306 acres. There was
no increase in usable land after corrective action remediation at 12 facilities.14 The increase in
usable land exceeded 100 acres for only three facilities.
These values of Q, represent one estimate of the increase in usable area at facilities.
However, it is possible that the decision rules described above underestimate Q,. Specifically,
remediation of certain configurations of SWMUs and soil contamination at facilities may enable
the use of a portion of the facility larger than the area of SWMUs and associated soil
contamination. A sensitivity analysis was conducted to explore the potential impacts of larger
increases in usable facility area on the estimated facility value benefits. The results of this
analysis are presented in Section 12.5.
The facility value benefit (V,B) is the product of the replacement cost for comparable
land and the increase in usable area, discounted to reflect the time required to complete surface
remediation. There are two components to the time required to complete remediation ~ the
estimated start date for corrective action remediation and the time required to implement the
remedy. For the purposes of the RIA, remedial activities were assumed to begin at facilities
between 1996 and 2005.15 Information on the time required to complete SWMU and soil
remediation at each facility was obtained from the expert panel reports.16 A discount rate of
seven percent was used to calculate the net present value of facility benefits.17
12.2.2 Benefits Related to Changes in Facility Use
With the imposition of corrective action requirements, some facility owners may
remediate to levels that make the property suitable for alternative uses -- i.e., commercial,
residential or recreational development. According to economic theory, the facility use will
change only if this alternative use has a higher value than the current industrial use. In this case,
the facility value after remediation will not simply rebound to the pre-requirements value (V°),
but will increase to some higher level. This increase (V2B) can be expressed as:
" Total areas of surface contamination being remediated at 8 of these 12 facilities were too
small to be usable (e.g., the removal of a 100 square foot drum storage pad). There was no
surface remediation required under corrective action at three facilities. The remediation selected
for the remaining facility -- a RCRA cap across large portions of the facility - did not allow for
future use.
15 See Chapter 5 for a discussion of these assumptions.
16 See Chapters 4 and 5 for a discussion of the expert panel facility evaluations. A default
value of one year was used at five sites where the expert panel did not estimate the time to
complete surface remediation.
17 U.S. Office of Management and Budget, "Guidelines and Discount Rates for Benefit-Cost
Analysis of Federal Programs," Circular A-94, October 29,1992.
*** DRAFT - March 25, 1993 ***
-------
12-11
(P2 - P,') x Q2 = V2B (2)
Where:
P,' = Price of comparable industrial land18
P2 = Price of land under alternative use
Q2 = Area of the facility (SWMUs and surrounding soil contamination,
non-overlapping ground-water plumes plus contiguous areas) that
becomes suitable for alternative use after remediation
V2B = Added facility value related to change in use
The first step in estimating the facility value benefits related to changes in use involved
determining the potential post-remedial uses for a facility from the following factors:
Current operating status of the facility, the expectations for future operations, and
the implications of remediation for continued operations;
Surrounding land use;
Overall development trends and zoning restrictions in the vicinity of the facility;
and
Need for any major additional physical changes at the facility to make it suitable
for an alternative use - e.g., demolition of buildings or removal of pavement.
Information on the surrounding land use, the effect of remediation, and the physical
changes that may be required to support an alternative use was available from the RFIs, expert
panel reports, and facility maps. Information collected for the averted cost analysis presented in
Chapter 9 provided insights into development trends. Additional information on the
development trends in the vicinity of facilities was obtained from local realty offices.
Potential alternative uses were identified for 25 of the 52 sample facilities.19 Based on
clean-up goals, surrounding land uses and conversations with local realtors about development
trends and any zoning restrictions, the following future uses are anticipated at the sample
facilities: residential use at 12 facilities, commercial use at 7 facilities, either residential or
commercial use at 4 facilities, recreational use at 1 facility, and either residential or recreational
18 P,' may not be equal to P, from the increased usable land part (V,B) of the analysis. The
prices quoted for land varied with parcel size. Given that areas affected by ground-water
contamination may be quite large, discounts in land pricing may drive these prices (P,') lower
than P,.
19 Note that these uses may differ from the future use assumptions applied by the expert
panels in the remedy selection process, because these uses are derived from the economic
analysis discussed in this chapter rather than the requirements of corrective action.
» DRAFT - March 25, 1993 »**
-------
12-12
use at 1 facility. The primary reasons for not selecting alternative uses at the remaining 27
facilities were: (1) facility configuration or remedial clean-up goals make the facility unsuitable
for commercial, residential or recreational uses; or (2) there is little demand for development of
alternative uses in the vicinity of the facility. For example, there are several industrial facilities
located in areas where the presence of a number of vacant lots in the vicinity suggests a low
potential for developing alternative facility uses.
For any of these potential alternative uses to occur, the price of land for the new use (P2)
must be higher than the price of land under the current use (P, ').m The price of land for
alternative uses was based on market prices quoted by local realtors for comparable land under
the same type of use. To the extent possible, these quotes were based on properties that were
similar to the facility area affected by contamination in terms of size, proximity to town centers,
and access to major transportation routes and utilities. P2 was higher than P,' for only 8 of the
25 sample facilities where changes in use appeared plausible.21 Benefits related to changes in
facility use were estimated for these facilities. At the remaining facilities, the value of alternative
uses were either equal to or less than the value of current uses. In either case, there is no
economic incentive for changing the facility use.
The facility area that becomes suitable for an alternative use after remediation (Q2) was
based on the area of surface contamination where remediation allows for increased use (Q,) plus
the on-site area of ground-water contaminant plumes. In addition, the area considered was
broadened beyond the actual locations of the surface contamination and plumes to a contiguous
area that encompasses all of the surface contamination and plumes. This larger area was defined
as Q2 -- the area that will become available for alternative use after remediation.
Two different patterns of facility contamination were encountered in reviewing the
facilities to determine Q2. In the first pattern, illustrated in Exhibit 12-2(a), the surface
contamination and ground-water contaminant plumes were concentrated in one portion of the
facility. For these facilities, Q2 was represented by a subdivision of the facility that encompassed
the entire area that will be affected by the remediation. Exhibit 12-2(b) shows the second
pattern, where facility contamination is wide-spread. Since no reasonable subdivision of the
facility encompasses the contamination, Q2 is equal to the entire area of the facility.22
20 Otherwise, the owner would not benefit from changing the use of the facility, and it would
continue under the current use.
21 No comparable lots were available for one large facility where future recreational use is
anticipated. For the purpose of this analysis, the facility value under an alternative use as a park
was assumed to be twice the current assessed property value. It is unlikely that this assumption
greatly affects the overall estimate of site value benefits, because this facility contributes only
three percent of the total benefits.
22 A key assumption here is that a higher value alternative use will not be impaired by any
contamination that may remain at the facility after corrective action remediation - i.e.,
contamination that is not subject to corrective action requirements. Such facility conditions
could prevent the change of use because of concerns over potential liability or health risks.
DRAFT - March 25, 1993 ***
-------
EXHIBIT 12-2
DETERMINATION OF AREA SUITABLE FOR HIGHER
VALUED ALTERNATIVE USES (Q2):
PATTERNS OF FACILITY CONTAMINATION
(a)
0
Facility Boundary
Area of S WMUs and surrounding
soil contamination
Extent of Ground Water
Contamination Plumes
Subdivided area of facility
for which change of use
benefits accrue
(b)
-------
12-14
Exhibit 12-3 illustrates the differences between the increase in usable facility area (Q,),
the facility area that becomes suitable for alternative use (Q2) and the total area of the facility
for the eight facilities where there are higher valued alternative uses. Total facility areas for
these facilities ranged from 3.1 to 1,273 acres. The values of Q, ranged from zero to 80 acres,
representing a maximum of 18 percent of the total facility area. Q2 was significantly larger than
Q, for each of the facilities, ranging from 2.5 to 435 acres. For three of the eight facilities, Q2
was equal to the total facility area. The potential impact of using alternative decision rules to
determine Q2 on the national estimate of facility value benefits is discussed in Section 12.5.
EXHIBIT 12-3
COMPARISON OF TOTAL FACILITY AREA TO INCREASED
USABLE AREA (Q,) AND AREA THAT
BECOMES SUITABLE FOR ALTERNATIVE USES (OJ
Facility
Number
2
17
39
41
46
63
114
125
Total Facility
Area
(acres)
325
3.1
15
72.4
1,273
14
435
55
Increased
Usable Area
(Q,)
Acres
5
0.2
0
0
31.5
1
80
0
Percent of
Total
Area
1.5%
6.5%
0%
0%
2.5%
7.1%
18%
0%
Increased
Usable Area
(Qj)
Acres
125
3.1
2.5
35
100
14
435
24
Percent of
Total
Area
39%
100%
17%
48%
7.9%
100%
100%
44%
The facility value benefit from change in use (V2B) is the product of the facility area that
becomes available for a higher value use and the price differential between land under this
higher value use and the current facility use. The net present value of this benefit was calculated
using a seven percent discount rate over the time required to fully complete corrective action
remediation at the facility -- i.e., both ground-water and surface remediation.23 The time
periods used for discounting these change in use benefits included the same delays in initiating
23 At one facility, where the higher value use is a recreational park, change of use benefits
are not contingent upon completion of ground-water remediation. Benefits for this facility were
discounted over the time required to complete surface remediation.
**» DRAFT - March 25, 1993 ***
-------
12-15
remediation that were used in the first pan of the analysis. The time required to fully remediate
ground water once activities have begun at each facility were obtained from preliminary remedy
duration estimates in the expert panel reports.24
12.2.4 Total Facility Value Benefits
The estimate of the total facility value benefits (VB) is the sum of the present value of the
increased usable land benefits and change in use benefits (V,B + V2B). To check that these
benefits were attributable to corrective action, they were compared to the projected cost of
facility remediation.25 The facility value benefits were less than the corresponding costs of
facility remediation at each of the sample facilities. Under these circumstances, facility owners
would not undertake remediation in absence of regulatory requirements, and increases in facility
values were considered to be a benefit of corrective action.
The final step in the analysis was to develop national estimates of facility value benefits
from the sample values. The national estimate of facility value benefits, representing
approximately 2,600 facilities, was obtained by applying the sampling weights to the individual
estimates of benefits for the 52 sample facilities.26
12.3 Example of Facility Value Benefit Calculations
To illustrate how facility value benefits were calculated, the derivation of benefits for one
facility -- Facility 46 -- is described in this section. This facility accounts for roughly nine percent
of the estimated national facility value benefits.
Facility 46 encompasses an area of roughly 1,300 acres, of-which 10 percent is covered
with buildings and other paved surfaces. Approximately three-quarters of the facility is meadow
and woodlands, some of which is used for storage and product testing. The surrounding areas
are primarily residential and agricultural.
Corrective action is required at this facility to address 16 SWMUs of concern and two
ground-water plumes. The total area of these SWMUs and surrounding soil contamination that
may represent increased usable facility area after remediation (Q,) is 31.5 acres.27 These
SWMUs are mostly waste piles and landfills that will be excavated, stabilized and/or capped.
The facility value benefit from increased usable land was based on the assumptions that:
(1) these areas of surface contamination are not usable in their present condition; and (2) this
land would become usable once the proposed remedial activities outlined in the expert panel
24 See Chapters 4 and 5 for a discussion of the expert panel facility evaluations.
25 See Chapter 5 for a discussion of the costs of corrective action remediation.
26 See Chapter 3 for a discussion of the sampling weights.
27 This area is the sum of the area to be remediated for 15 of the 16 SWMUs. One SWMU
-- a storage tank - was not included in this total because the area being remediated was too
small to provide additional usable land.
*** DRAFT - March 25, 1993
-------
12-16
reports were completed. Therefore, the benefit was calculated as the price of replacing these
areas of surface contamination.
A local realtor familiar with industrial property values quoted a price (P,) of $7,000 per
acre for larger parcels (i.e., more than 30 acres) of unimproved industrial land in the vicinity of
the plant. Using Equation 1 in Section 12.2.1, the increased usable land benefit is then:
$7,000 per acre x 31.5 acres = $220,500
These benefits must be discounted to calculate the net present value in 1992. Corrective
action remediation at Facility 46 is assumed to begin in eight years from 1992, in 2000. Adding
another year to complete surface remediation, the benefits will be realized in nine years. Using a
discount rate of seven percent, the net present value of V,B is about $120,000.
The primary factors considered in identifying the potential alternative uses of this facility
were the predominantly residential and agricultural use of surrounding land and the high
percentage of open space at this facility. Both commercial and residential use were considered
potential alternative uses for this facility.
The incremental facility area that may become suitable for alternative commercial or
residential use (Q2) is 100 acres. This area represents a subdivision of the facility that
encompasses the 31.5 acres of surface contamination described above, the 40 acre areal extent of
ground-water contamination and the intervening area as shown in Exhibit 12-2.
Discussions with a second local realtor familiar with land values revealed that prices were
significantly higher for commercial than for residential land. Commercial land prices in the more
desirable areas can be as high as $1.50 to $2.50 per square foot. However, the area in the
immediate vicinity of this facility is reportedly not particularly attractive for business
development, and prices would tend to be somewhat lower -- $1.00 per square foot or $43,560
per acre. Residential land prices in this area range from $10,000 to $25,000 per acre. Therefore,
the price under the highest value alternative use (P2) is $43,560 per acre, the price for
commercial use. Because this price is higher than the price of this land under the current
industrial use (P,'), the post-remedial alternative use is higher in value than the current use.28
Using Equation 2 in Section 12.2.2, the facility value benefits related to this change in use (V2B)
can be calculated as follows:
($43,560 per acre - $7,000 per acre) x 100 acres = $3,656,000
28
Note that P, equals P,' at this facility, because the unit price for industrial land is
constant across the range of areas considered in estimating the two types o£ facility value benefits
-- 31.5 to 100 acres.
*** DRAFT - March 25, 1993 ***
-------
12-17
In discounting these benefits, the time required to complete ground-water remediation
must be considered.29 As cited above, remediation at Facility 46 is assumed to begin eight years
from 1992, in 2000. According to the expert panel report for this facility, ground-water
remediation is anticipated to occur over a period of 30 years. Adding these two periods
together, the benefits are estimated to occur in 38 years. Using a discount rate of seven percent,
the net present value of V2B in 1992 is about $280,000.
The total facility value benefits ( VB = V,B + V2B) of roughly $400,000 are less than the
estimated $47.6 million in corrective action remediation costs at this facility. This demonstrates
that the facility owner would not undertake remediation in the absence of corrective action
requirements. Therefore, the increase in facility value can be attributed to the imposition of
corrective action requirements and is one benefit of corrective action.
The final step in estimating the facility value benefits is to weight the facility-specific
benefits to reflect the national benefits. The sample weight for Facility 46 is approximately 63.
This facility therefore contributes about $25 million to the national estimate of facility value
benefits.
12.4 Results
Exhibit 12-4 summarizes the estimated corrective action facility value benefits. The
national estimate of this benefit category is $280 million.10 These benefits were derived from
approximately 2,200 of the 2,600 facilities where remediation will be required under corrective
action. Over 80 percent of the benefits - roughly $230 million - are due to the anticipated
increase in usable area after remediation. The small contribution of changes in facility use to
these benefits is primarily due to the small number of facilities where there are higher value
alternative uses - roughly 420 facilities or 19 percent of those with benefits.11
29 The suitability of this facility for commercial development is contingent upon completing
ground-water remediation. Therefore, the benefits related to change in use for this facility were
discounted over the anticipated time to complete this activity.
30 Because this national estimate is based on a sample of the affected facilities, there is some
likelihood that the actual benefits will differ from this estimate due to sampling error.
11 Potential alternative uses at an additional 1,100 facilities nationally are equal or lower in
value to current industrial use.
*** DRAFT - March 25, 1993 ***
-------
12-18
EXHIBIT 12-4
ESTIMATED CORRECTIVE ACTION FACILITY VALUE BENEFITS
(1992 dollars)
National Estimates
of Benefits
Number of
Facilities
with Benefits
2,200
Benefits from
Increase in
Usable Land
$230 million
Benefits from
Change in Use
$49 million
Total Facility
Value Benefits
$280 million
* Total does not match sum due to rounding.
Information on the ten sample facilities contributing over 90 percent of the national
estimate of facility value benefits is presented in Exhibit 12-5. The columns in this exhibit show
the individual inputs into the calculation of the benefits from both increased usable land and
change in facility use. The sum of the sample weights (column two) for these 10 facilities
indicates that they represent 965 of the roughly 2,200 facilities with facility value benefits. The
years to complete surface remediation (column three) for each facility is measured from a
baseline of 1992 and includes both the estimated starting date for remediation and the estimated
time to complete the remedial activities. The estimated benefits from increased usable land
(land price multiplied by the area remediated) have been discounted at a rate of seven percent
over this time period. The years to complete ground-water remediation (column four) are
provided for Facilities 46 and 125. These are the only two of these ten facilities where there are
higher value alternative uses that are contingent upon ground-water remediation. Change of use
benefits for these facilities have been discounted over the period required to complete this
activity at these particular facilities. Ground-water quality does not affect the analysis of benefits
at the remaining eight facilities.32 The benefits for each sample facility in this exhibit have been
multiplied by the sample weights to produce estimates of the national benefits.
The estimate of facility value benefits is dominated by a relatively small number of
facilities. As shown in Exhibit 12-5, the largest single contributor to national benefits (Facility
44), representing 63 facilities, accounts for 26 percent of the estimate. Approximately 80 percent
of the facility value benefits occur at about 470 facilities - roughly 22 percent of the facilities
contributing to the total estimate of facility value benefits.
32 The change of use benefits for Facility 114 are based on recreational use. Benefits from
this alternative use are not affected by ground-water quality and can be realized once soil
remediation has been completed. Therefore, these benefits have been discounted over the 6-year
period required to complete surface remediation at this facility.
*** DRAFT - March 25,1993 ***
-------
12-19
IT 12-5
KEY FACILITIES RESPONSIBLE FOR ESTIMATED
NATIONAL CORRECTIVE ACTION FACILITY VALUE BENEFITS
(1992 dollars)
Facility
Number
44
60
46
18
125
29
167
114
47
122
Subtotal for ten key
facilities
Total for all
facilities'
Facility
Sample
Wright
63
3
63
63
214
63
214
3
214
63
965
2,200
Yean to
Complete
Surface
Remediation*
7
7
9
5
9
9
9
6
9
28
Yean to
Complete
Ground-Water
Remediation*
38
38
Facility Value Benefits from
Increase in Usable Land
Price under
Current
U»e (P,)
($ per acre)
$261,360
$130,680
$7,000
25,000
NA
$37,500
$5,000
$7,800
$40,000
$15,850
Area
Remediated
(Q,)
(acres)
7.0
2140
31.5
21.0
00
150
168
800
13
- 300
National
Estimate of
Benefits'
$72.000,000
$58.000,000
$7,600,000
$24,000,000
NA
$19,000,000
$9.800.000
$1.400.000
$5,800,000
$4,500,000
$200,000.000 ,
$230,000,000
Facility Value Benefits
from Chance In Use
Price Differential
Under Alternative
Use(P,.P,-)
(S per acre)
$36,560
$57,121
$7^00
Area
Changin
(Use
(Q,)
(acres)
1000
240
4350
National
Estimate
of Benefits'
$18,000,000
$22,000,000
$ 7.500,000
$48,000,000
$49,000,000
Total
National
Facility
Value
Benefits'
$72,000,000
$58,000,000
$25,000,000
$24,000.000
$22,000,000
$19,000,000
$9,800,000
S9.000.000
S5.800.000
$4,500,000
$250.000.000
$280.000,000
Cumulative
Percent of
Total
Benefits
26%
47%
56%
65%
73%
80%
83%
87%
89% x
90%
Notes-
Time to complete surface remediation includes any delay in initialing surface remediation plus the duration of the remedy.
Time to complete ground-water remediation includes any delay in initialing ground-water remediation plus the duration of the remedy.
P, O, Facility Weight, discounted at seven percent over number of years to complete surface remediation.
(Pi-P| ') QI Facility Weight, discounted at seven percent over years to complete ground-water remediation. Benefits for Facility 114 were discounted over the years required to complete surface
remediation because change to alternative (recreational) use is not contingent on ground-water remediation
Totals may not add due to rounding.
Approximately 2,200 of (he 2,600 facilities where remediation is required under corrective action contribute to the total facility value benefits of $280 million.
DRAFT -March 25,1993
-------
12-20
12.5 Sensitivity Analysis
As discussed previously, there is some uncertainty inherent in the estimation of both the
facility area where remediation allows for increased use (Q,) and the area where remediation
enables higher valued alternative uses (Q2). This section presents the results of a sensitivity
analysis of alternative definitions of these two areas.
In the preceding analysis the increased usable facility area Q, was limited to SWMU
areas plus any surrounding areas of soil contamination that would be remediated. It is possible
that SWMU and soil remediation at a facility could enable the use of a portion of the facility
that is larger than the specific area remediated. This may occur if one or more SWMUs restrict
the only possible access to a larger portion of the facility. Alternatively, if several SWMUs are
located in close proximity, remediation may increase the use of the SWMUs themselves plus the
intervening areas.
Any increase in Q, will lead to a proportional increase in facility value benefits from
increased usable land. Therefore, if Q, is uniformly doubled across all facilities to account for
the potential increased use of intervening areas after remediation, the facility value benefits from
increased usable area will also double, increasing from roughly $230 million to $460 million.
The facility area that becomes suitable for a change in use after remediation (Q2) is based
on an area that encompasses the SWMUs and associated contaminated soils plus the area of on-
site ground-water plumes. This area is generally significantly larger than the increased usable
area. In fact, for three of the eight sample facilities with higher valued alternative uses, Q2 was
equal to the entire facility area. For the remaining five facilities where Q2 is less than the full
area of the facility, it is possible that facility contamination may affect a larger area than was
assumed in the preceding analysis.
Alternative larger Q2 values can be developed for these facilities to measure the potential
impact of this factor on the estimate of facility value benefits. Exhibit 12-6 presents both the
preceding and alternative Q2 values, as well as the increase in facility value benefits associated
with using the alternative values. The alternative estimates for Q, are equal to the full facility
area for four of the five facilities. However, due to the size of Facility 46 - 1,273 acres - the
alternative value is based on a subdivision of the facility that is larger than that used originally.
Use of these alternative values for Q2 would increase the facility value benefits by approximately
$100 million, raising the benefits of change in facility use from nearly $50 million to $150 million.
Nearly all of the increased benefits are associated with two sample facilities - Facilities 46 and
125.
DRAFT - March 25, 1993 »**
-------
12-21
EXHIBIT 12-6
INCREMENTAL IMPACT OF ALTERNATIVE ESTIMATES OF AREA SUITABLE
FOR CHANGE OF USE (QJ ON FACILITY VALUE BENEFITS
(1992 dollars)
Facility
Number
2
39
41
46
125
Original Estimate
ofQ,
(acres)
125
2.5
35
100
24
Alternative
Estimate of Q,
(acres)
325
15
72.4
500
55
Incremental Increase above Baseline Estimate
Increase in Change of
Use Benefits
$140,000
$24,000
$720,000
$71,000,000
$29,000,000
$100,000,000'
* Total does not match sum due to rounding.
The combined effect of using alternative values for Q, and Q2 is shown in Exhibit 12-7.
Total estimated facility value benefits increase by approximately .$330 million, from $280 million
to $610 million. The majority of this increase -- $230 million or nearly 70 percent -- is due to the
higher values for increased usable area at facilities. The remaining $100 million increase in
benefits is related to alternative values for Q2 at five sample facilities.
EXHIBIT 12-7
SUMMARY OF SENSITIVITY ANALYSIS OF INCREASED USABLE
AREA (Q,) AND AREA SUITABLE FOR ALTERNATIVE USE (Q,)
(1992 dollars)
Source of Facility
Value Benefits
Increased Usable Area
Change in Use
Total Facility Value Benefits
Base Benefits Estimate
$230 million
$49 million
$280 million1
Increased Benefits Estimate
$460 million
$150 million
$610 million
* Total does not match sum due to rounding.
*** DRAFT March 25,1993
»**
-------
12-22
12.6 Limitations
There are several major limitations and sources of uncertainty in this analysis. Some of
these are associated with sampling, data availability and modeling assumptions. These limitations
are described in Chapters 3 and 4. Limitations related specifically to the analytical approach
used to estimate facility value benefits are discussed in this section.
12.6.1 Factors That May Overstate the Benefits
A key assumption used to evaluate facility value benefits in this analysis was that
contaminated areas, or areas affected by nearby contamination, have no value for current or
alternative uses until they are remediated. In theory, the calculation of facility value benefits
should account for the current value of contaminated areas. However, it was not possible to
incorporate current value into the analysis with the available resources. This means that the pre-
remediation value of the land affected by contamination was conservatively set at zero. To the
extent that the area affected by contamination could in fact be used for some purpose (e.g., a
parking lot or buffer zone) prior to remediation, the analysis overstates benefits by not
accounting for this baseline value. The magnitude of this effect on benefits at a facility depends
on the value of this pre-remediation use and the size of the area being considered.
A second source of uncertainty that may overstate the benefits is that this analysis did not
consider the impact of other sources of contamination that will not be addressed by corrective
action. This approach may lead to an optimistic prediction of the potential for converting the
facility to alternative uses, which would overstate the facility value benefits. After the completion
of corrective action remediation, the continued presence of contamination that is not addressed
by corrective action or other authorities may prevent future alternative uses of a facility. The
remaining facility contamination may pose unacceptably high health risks for commercial,
residential or recreational uses. Alternatively, contamination may raise concerns over liability for
future remediation that discourage potential investors from committing the financial resources
required to convert the facility to an alternative use. The impact of this uncertainty on the
estimate of total facility value benefits is limited by the fact that higher value alternative uses
contribute $49 million --18 percent - to the total benefits.
12.6.2 Factors That May Understate the Benefits
As described earlier, there is uncertainty in the estimation of both the facility area where
remediation allows for increased use and the area where remediation enables higher valued
alternative uses. The decision rules used in this analysis may have underestimated these areas,
and consequently, underestimated the facility value benefits. The sensitivity analysis presented in
Section 12.5 suggests that the approach used could have underestimated the benefits by as much
as $330 million.
*» DRAFT - March 25, 1993 **
-------
12-23
12.6.3 Factors That Have an Indeterminant Effect on Benefits
The price of land used to calculate the facility value benefits in this analysis is a source of
uncertainty. Relying on market prices creates two potential limitations. First, the prices
obtained are based on current conditions in the real estate market. There are many examples of
dramatic shifts in local real estate markets over the past twenty years - including unforseen
revitalization of defunct industrial areas and vice-versa. Some of the areas where corrective
action facilities are located may experience such changes at some time in the future, but it is not
possible to predict such changes - whether increases or decreases in value -- and incorporate
them into this analysis. The potential impact of this limitation on the benefits estimate is
mitigated by the fact that any such gains or losses in facility value are likely to occur in the future
and will therefore be discounted in the present value calculations.
Second, comparable land prices were obtained through contacting local realtors. In
working with realtors to price comparable land, we attempted to specify key facility parameters
affecting the property value, such as proximity to town centers; terrain; availability of water,
sewer, and other utility hook-ups; and access to road, rail and water transportation facilities.
Despite these efforts, a wide range in potential prices was quoted for some of the facilities. In
these cases, the average price was used to estimate facility value benefits.
*** DRAFT - March 25, 1993 ***
-------
13. COMPARISON OF BENEFITS AND COSTS
This chapter compares the costs and benefits of the proposed Subpart S corrective action
rule. The chapter is organized into four sections. Section 13.1 is an introduction describing the
approach used for this analysis . Section 13.2 summarizes each of the measured costs and
benefits of the rule. Section 13.3 compares all of the costs and benefits of the rule. Section 13.4
discusses some limitations of the analysis.
13.1 Introduction
As has been previously stated, the purpose of this RIA is to report on the methodology
developed to help the program evaluate the costs and benefits of a variety of options before a
final Subpart S rule is completed. To illustrate the results of these methodologies, EPA has
chosen to quantify, to the extent, possible the potential costs and benefits of the proposed
Subpart S corrective action rule. The final form and content of the Subpart S rule depends on
comments received by the Agency and on the results of further analysis that remains to be done.
The costs and benefits of the proposed rule have been quantified in previous chapters of this
RIA. In this chapter, these benefits and costs are compared to one another. The methodologies
developed and the results that derive from those methodologies will be more meaningful when a
more complete array of regulatory options are analyzed.
To assess the proposed rule, we use two common approaches to the comparative analysis
of costs and benefits of regulatory proposals. The first of these tools, benefit-cost analysis,
compares benefits and costs when they are expressed in a common metric (e.g., dollars). The
second tool, cost-effectiveness analysis, is useful for comparative analyses of costs and benefits
that are not measured in a common metric. Although the costs and some benefits of corrective
action are measured in dollars, benefits like reductions in human health risk in this analysis are
expressed in non-monetary units (e.g., areal extent of contamination avoided). The Agency used
benefit-cost analysis to evaluate monetized benefits alone(e.g., avoided water use costs, nonuse
values of ground water), and cost-effectiveness analysis to evaluate all benefits, including non-
monetized benefits (e.g., health risks and ecological risks).
The benefit-cost and cost-effectiveness analyses presented in this chapter cover only those
benefits and costs that the Agency was able to quantify. Potential benefits to society that could
not be reasonably quantified are not considered; these are described in the limitations section of
this chapter. One potentially significant benefit of corrective action, impact on residential
property values was estimated for a small number of case-study facilities, but not at the national
level. These estimates are not included in the benefit-cost analysis. Results of the residential
property value case studies are briefly discussed in section 13.2.2 of this chapter.
**
DRAFT - March 24, 1993
-------
13-2
13.2 Review of Costs and Benefits
Each of the costs and benefits of corrective action studied by EPA for this RIA has been
described in earlier chapters:
Costs:
Costs to facilities subject to corrective action (Chapter 5);
Monetized Benefits:
Averted water use costs (Chapter 9);
Nonuse values of ground water (Chapter 10);
Value of the corrective action site (Chapter 12);
Non-monetized Benefits:
Reduced extent of contamination (Chapter 4); and
Human health risk reductions (Chapter 7);
Ecological risk reductions (Chapter 8);
Residential property values (Chapter 11).
This section briefly summarizes the key findings for each major cost and benefit. Costs
are summarized in section 13.2.1 and benefits are summarized, m section 13.22.
13.2.1 Costs
The costs of corrective action are measured as the total present value of conducting
corrective action. These costs include the costs of conducting facility investigations, the capital
costs of remedial activities, and operations and maintenance costs. Remedial costs for each
sample facility and each solid waste management unit (SWMU) were estimated by expert panels
assembled by EPA. Remedial costs estimated by the expert panels were then adjusted to include
design, oversight, and contingency assumptions, and were discounted to reflect the timing of
remediations, using a seven percent discount rate. There are two major results of the cost
analysis:
The total national present value cost of corrective action is
estimated to be S18.7 billion (1992 dollars).1
The annualized cost is estimated to be S1.8 billion per year.2
'Discounted over the 128-year modeling period. See Chapter 5.
2Costs were amortized over 20 years. See Chapter 5.
DRAFT ~ March 24, 1993
-------
13-3
Chapter 6 presents more detailed cost results including disaggregation of total costs by type of
cost (i.e., capital costs, operations and maintenance, investigations), facility type (i.e., Federal vs.
non-Federal), environmental medium, industry, and SWMU type.
13.2.2 Benefits
The benefits of corrective action measured in the RIA are summarized below. EPA
monetized three kinds of benefits: nonuse values of ground water, averted water use costs, and
value of the corrective action site. Human health risk, ecological risks, and extent of
contamination were not monetized. The section concludes with a discussion of property value
benefits for which the Agency conducted facility-level case studies but did not estimate national
benefits.
Monetized National Benefits3
Use Value of Ground Water: EPA assessed the costs that would be incurred to replace
or treat contaminated drinking water sources if remediation did not occur. By preventing water
supply treatment or replacement, corrective action would avert potential price increases. In
addition, corrective action could lower water prices by making currently contaminated water
sources usable. The Agency measured these potential benefits as the increase in consumers'
surplus that would be expected with corrective action. To estimate these benefits, the Agency
collected site-specific data on current and future water uses, related water uses to current and
potential ground-water contamination, and determined the effects of expected ground-water
contamination (with and without corrective action) on water supplies and prices. The study
found that the avoided losses in consumers' surplus associated with averted water use costs would
total $5 million.
Nonuse Values of Ground Water; The study used the results of a survey to determine
people's willingness-to-pay to remediate contaminated ground water, focusing on nonuse values.
Nonuse values are the values people place on natural resources that are unrelated to their own
use of the resources. In the case of ground water, this could mean simply the knowledge that
clean ground water exists. Although the Agency studied the nonuse values of ground water only,
corrective action would provide nonuse benefits for surface water, soil, and air as well. There is
considerable uncertainty in the reliability of willingness-to-pay studies, but the Agency estimates
that the nonuse benefits for corrective action range from $170 million to $18 billion (1992
dollars). The Agency's estimate of the most likely nonuse value is $2.3 billion.
Value of the Corrective Action Site: The analysis also assessed the changes in the value
of the corrective action site resulting from corrective action. Two possible scenarios for
increased site values were included in the analysis. First, the Agency identified facilities with
potential alternative site uses (i.e., commercial, residential, or recreational) that would provide
owner/operators with greater site value than current uses. Second, the Agency identified
'All monetized benefits are presented in present value terms.
DRAFT - March 24, 1993
-------
13-4
facilities for which cleanup would increase the usable area on-site for current land uses. The
potential site value benefits for each sample facility were then compared to the total costs of
cleanup at that facility. If the benefits exceeded the costs, the analysis assumed that the property
owner would remediate even without the corrective action rule, and did not count the increase in
site value as a benefit of the rule. If the benefits were less than the costs, the owner would not
remediate without the rule, and the increase in site value was counted as a benefit of the rule.
For these facilities, benefits were then extrapolated to estimate national level site value benefits.
The analysis showed that the potential future values of corrective action sites could be
significant at about $280 million. A sensitivity analysis on the land area of facilities affected by
contamination suggests that the national estimate could be as high as $610 million.
Non-monetized National Benefits
Extent of Contamination: This analysis estimated the long-term extent of contamination
in environmental media. Short-term contamination was profiled through a combination of
available monitoring data and modelling results. The long-term extent of contamination was
predicted using the MMSOILS model. Assumptions about remedy effectiveness were then
incorporated to predict revised estimates of the extent of contamination with remedies in place.
The reduction in the extent of contamination found contribute to the benefits of the rule.
The Agency's analysis found that, at a minimum, contaminated ground water underlying
1,400,000 acres and 18,000,000 cubic yards of contaminated soils would be remediated by
corrective action. These estimates are minimum estimates because remedy effectiveness
simulations were limited to only about 720 facilities requiring corrective action.
Human Health Risks: This analysis focused on the extent to which corrective action will
mitigate risks to human health, considering the effects of exposure to carcinogens and non-
carcinogens through various pathways. To assess human health risks, the Agency modeled
releases of contaminants from SWMUs and modeled fate and transport of contaminants in the
environment using facility-specific data. Estimated concentrations of constituents at human
exposure points were then used to estimate individual cancer and non-cancer risks for each
human exposure pathway at each facility, and to estimate population risks and risk to individuals
of highly-exposed or sensitive subpopulations. Individual risks and risks to highly-exposed or
sensitive subpopulations were measured as the number of facilities nationwide with individual
cancer or non-cancer risks above levels of concern. Two population risk descriptors were used.
The first, for cancer risk, describes the number of excess cancer cases expected in the exposed
population. The second, for non-cancer risk, is an estimate of the number of people exposed to
contaminant levels exceeding thresholds for non-cancer toxicity.
The human health benefits analysis suggested that in the absence of corrective action
there would be significant individual risks (including risks to highly exposed or sensitive
subpopulations) at 917 of the 2,600 facilities expected to trigger corrective action.
Implementation of the rule would eliminate individual risks of concern at all but 18 of these
facilities. The Agency's analysis of population risks predicts that there would be 1,200 to 21,100
excess cancer cases without corrective action, and 800 to 7,700 cases with corrective action. In
DRAFT - March 24, 1993
-------
13-5
addition, there would be an estimated 900 to 25 million exceedances of non-cancer toxicity
thresholds without corrective action, and 800 to 13 million with corrective action.
Ecological Risks; This analysis considered the potential risks to sensitive environments
(such as wetlands) and wildlife that may be posed by contamination in the absence of corrective
action. The ecological risk assessment was composed of three separate analyses: the proximity
analysis, the concentration-based analysis, and ecological risk assessment case studies of three
corrective action facilities. The proximity analysis summarized the distance of facilities from
sensitive environments and showed the areal extent of sensitive environments near the facilities.
Based on the proximity analysis, the Agency classified approximately 60 percent of the facilities
as having potentially "high" ecological risk.
Because the proximity analysis did not provide conclusions about the effect of corrective
action, the Agency conducted a concentration-based analysis of constituent-specific risks to
aquatic organisms. In this analysis, constituent concentrations in surface water were compared to
ecological benchmark levels to determine the number of facilities with ecological risks.
Nationwide, the Agency expects exceedances of ecological benchmark levels at 144 facilities (5.6
percent) in the absence of corrective action.
In the third ecological risk analysis, the Agency conducted detailed ecological risk
assessment case studies for three RIA sample facilities. In these case studies, the Agency
identified specific ecological risks for all potential ecological receptors at each facility. The case
studies showed that existing contamination could pose significant risks to a variety of aquatic and
terrestrial ecological receptors. The facilities selected for case studies were facilities with
available ecological information and suspected ecological risks. Thus, these case studies are not
necessarily representative, of ecological risks at all corrective action facilities and cannot be
extrapolated to national levels.
Case Study Benefits
Residential Property Values: The study used market data on the sales of residences to
estimate people's willingness-to-pay to avoid perceived costs and risks from living close to
contaminated sites and to purchase "insurance" against future risks. The residential property
analysis used regression analysis to estimate the relationship between housing prices and
distances to corrective action facilities.. The Agency controlled for two additional factors
affecting property values: household characteristics and the effect of key events where
contamination became publicly known.
This analysis was limited by data availability, and the Agency chose to study property
value impacts on a case study basis only. Thus, the results of the analysis are indicative of
property value impacts, but they are not conclusive. For the three facilities studied, the Agency
found statistically significant relationships between housing prices and distances from RCRA
facilities requiring corrective action. Housing prices tend to increase with distance from the
facilities. In addition, the analysis showed that changes in housing prices over time are consistent
with the hypothesis that contamination events can affect housing prices. These results suggest
DRAFT « March 24, 1993
-------
13-6
that corrective action would prevent adverse impacts on residential property values by reducing
existing levels of contamination and preventing future contaminant releases.
13 J Comparison of Costs and Benefits
This section uses simple quantitative analysis of costs and benefits to evaluate the overall
effects of corrective action. It presents a benefit-cost analysis (section 13.3.1) that evaluates costs
and monetized benefits, and a cost-effectiveness analysis (section 13.3.2) that evaluates non-
monetized benefits.
13 J.I Benefit-Cost Analysis
The benefit-cost analysis compares the total costs of corrective action to the sum of the
monetized benefits. The purpose of the benefit-cost analysis is to determine whether the
monetary benefits of corrective action provide a reasonable return on costs, as required by
Executive Order 12291. However, because this analysis includes only the monetized benefits of
corrective action, the Agency cannot evaluate the corrective action rule using the results of the
benefit-cost analysis alone. The Agency will also consider the non-monetized benefits evaluated
in the cost-effectiveness analysis.
All monetized effects of corrective action included in the benefit-cost analysis are total
values covering the 130-year modeling period. The estimated total values for costs and benefits
are discounted to net present values in 1992 dollars using a real discount rate of seven percent.
Discounting is necessary to enable comparison of costs and benefit occurring in different years.
The discount rate of seven percent is used because it approximates the marginal pre-tax rate of
return on an average investment in the private sector in recent years.4 The Agency uses total
net present values rather than annualized values because annual costs and benefits are expected
to vary significantly over time, and it may be misleading to present average annual values of each
cost and benefit measure.
The benefit-cost analysis is presented in Exhibit 13-1. The total expected cost of
corrective action under the proposed rule is $18.7 billion. In comparison, expected monetary
benefits of corrective action total $2.6 billion. The majority of the monetary benefits result from
nonuse values of groundwater. The net benefit of corrective action, based solely on analyzed
economic effects, is calculated by subtracting costs from total benefits. The resulting value is
negative $16.1 billion. This result indicates that total costs are substantially greater than the total
value of monetized benefits. However, the benefit measures most closely related to the goal of
corrective action (protection of human health and the environment) are not monetized and are
not included in the benefit-cost analysis. To gauge the overall effect of the rule, the Agency
must compare both the non-monetary and monetary benefits to the costs. The cost-effectiveness
of the non-monetary benefits is examined in the following section.
4 "Guidance and Discount Rates for Benefit-Cost Analysis of Federal Programs". Circular
A-94. Office of Management and Budget. Washington, D.C.: Office of Management and
Budget. October 29, 1992.
*** DRAFT » March 24, 1993
-------
13-7
EXHIBIT 13-1
CORRECTIVE ACTION BENEFIT-COST ANALYSIS
Effect of
Corrective
Action
Costs
(billion
1992 $)
$18.7
Benefits (billion 1992 $)
Nonuse
values of
ground
water
$2.3
Averted
water use
costs
$.01
Value of the
corrective
action site
$0.28
Sum of
Benefits
$2.6
Net
Benefits
(billion
1992 $)
- $16.1
13.3.2 Cost-effectiveness Analysis
The cost-effectiveness analysis relates non-monetized benefits of corrective action to total
costs. The approach is to divide appropriate descriptors of these benefits by costs to show the
benefits gained per unit cost (e.g., human cancer cases prevented per billion dollars). Cost-
effectiveness is most useful for comparing multiple regulatory alternatives because it shows which
alternative offers the greatest amount of benefit for equal units of cost. However, in this case
the cost-effectiveness analysis only evaluates two alternatives: the proposed rule versus no action.
Thus, the rule must be evaluated by judging whether the overall benefits of corrective
actionmonetized and non-monetizedare worth the costs.
Benefits included in this cost-effectiveness analysis include reduced human health risks,
reduced ecological risk, and reduced extent of contamination. Three descriptors of reduced
human health risks are used: number of facilities with reduced individual risks of concern, the
number of averted population cancer cases, and the number of exceedances of non-cancer
toricity thresholds.5 The analysis uses one descriptor of ecological risk reduction, the number of
facilities with averted exceedances of ecological benchmark levels. Descriptors of reduced extent
of contamination are reduced areal extent of ground-water contamination and reduced areal
extent of on-site soil contamination. These descriptors do not include the benefits of reduced
contamination in air, surface water, or off-site soils.
5The individual risk descriptor does not include individual risks calculated using high-end
scenario and hypothetical on-site risks.
DRAFT - March 24, 1993
-------
13-8
Cost-effectiveness is measured by dividing each benefit descriptor by total cost to show
benefits per billion dollars spent. All measures of costs and benefits used in the cost-
effectiveness analysis reflect the total value measured over the 130-year modeling period.6
Results of the cost-effectiveness analysis are presented in Exhibit 13-2. The first row shows the
total present value costs of corrective action. The non-monetized benefits of corrective action
are shown below in six columns, in terms of both the total benefits in each category and the
benefits per billion dollars. The Agency expects corrective action to eliminate individual risks of
concern at 899 facilities, to eliminate between 400 and 13,100 cancer cases, and to eliminate
between 100 and 12 million cases where non-cancer toxicity thresholds are exceeded.7 In terms
of reduced extent of contamination, the Agency expects a minimum of 1,400,000 acres of
contaminated ground water and 18,000,000 cubic yards of contaminated on-site soil to be
remediated. The last row of Exhibit 13-2 contains cost-effectiveness measures, the benefits of
corrective action per unit cost. For every billion dollars spent on the program, the Agency
predicts that the following benefits will result: 48 facilities with reduced individual risk; 21 to
averted population cancer cases; 5 to 640,000 averted population exceedances of non-cancer
thresholds; 75,000 acres of contaminated ground water remediated; and 950,000 cubic yards of
contaminated on-site soil remediated. In addition to these non-monetized benefits, each billion
dollars of corrective action expenditures would produce $139 million in monetized benefits.
711
'A potential limitation of the analysis is that costs are discounted, but some benefits are not.
See section 13.4.
7 The range of estimates reflects two risk reduction scenarios. In the first, EPA excludes the
effects of MCLs and taste/odor thresholds. In the second, EPA calculates baseline risk and risk
reduction assuming that averting behavior will occur if MCLs or taste/odor thresholds are
exceeded, and exposure and baseline risk will fall. The second scenario corresponds to the lower
end of the risk reduction estimates.
DRAFT - March 24, 1993
-------
13-9
EXHIBIT 13-2
CORRECTIVE ACTION COST-EFFECTIVENESS ANALYSIS
Total Present
Value Estimate of
Costs of Corrective
Action (7%, 1992
$):
Non-monetized
Benefits
Total Benefits
Benefits Per
Billion Dollars
$18.7 billion
Human Health Risk
Facilities
with
Reduced
Individual
Risk
899
facilities
48
facilities
Averted
Population
Cancer
Cases
400 13.100
cases
21 -711
cases
Avoided
Non -Cancer
Effects
100- 12
million cases
5 - 640.000
cases
Ecological
. Risks
Facilities
with Averted
Exceed an ces
of Ecological
Benchmark
Levels*
NA
NA
Extent of Contamination*
Area!
Extent of
Ground
Water
Cleanup
1,400.000
acres
75.000
acres
Extent of
On-site Soil
Remediation
18,000,000
cubic yds
950,000
cubic yds
Ecological risk reductions have not been estimated, only baseline risks are currently available.
b Results are for only 720 sites nationally for which remedy effectiveness analysis was performed and hence
understate the total values.
13.4 Limitations
The principal methodological limitation of the cost-benefit comparison for this RIA is
that not all of the benefits could be monetized. The objective of cost-benefit analysis is to
describe all of the impacts of a proposal in a common metric, usually dollars, so that the net
benefits of the proposal can be determined. If the net benefits are positive, the proposal is
viable. If there are other proposals as well, the one with the greatest net benefit should be
selected. However, when some of the benefits of a proposal cannot be monetized, then cost-
benefit analysis does not support such decision rules. Since the impacts are not all expressed in
the same terms, costs and benefits cannot be aggregated, and costs cannot be subtracted from
benefits to yield net benefits. Instead, some impacts must be expressed in non-monetary terms
and the decisionmaker must make a judgment about whether the benefits of a proposal justify its
costs. This limitation is particularly critical for this analysis because human health and ecological
risk reductions, some of the most important corrective action benefits, could not be expressed in
monetary terms.
DRAFT - March 24, 1993
-------
13-10
Another methodological limitation of the analysis is that total costs are discounted, while
non-monetary benefits are not discounted. This means that costs occurring late in the time
period are given less weight, while all non-monetary benefits, whenever they occur, are weighted
equally. Overall, the effect is that total costs are given less weight in the analysis relative to total
benefits.
Another source of uncertainty results from limitations of the data and approaches used to
estimate each of the costs and benefits of corrective action. These are described in earlier
chapters.
An important limitation of the cost-benefit comparison relates to the scope of the costs
and benefits included in the RIA." For example, as discussed above, though residential property
values were analyzed at a case-study level, they could not be incorporated in the national
estimates of benefits. Four other potentially significant effects of corrective action not
considered in the RIA are discussed below:
Effects on recreational resources;
Effects on agricultural production;
Effects on international trade; and
Distributional effects.
Corrective action could benefit recreation by reducing contaminant concentrations in
recreational areas near corrective action facilities. For example, corrective action could benefit
recreational fisheries by reducing fish kills from surface-water contamination. In addition,
corrective action could eliminate or prevent fishing bans in areas adversely affected by releases.
Corrective action could also provide benefits related to other recreational uses of surface waters,
such as swimming and boating, by eliminating use bans in contaminated areas. Although these
benefits are not evaluated in this RIA, they are partially captured in the human health benefits
analysis, which showed that corrective action would reduce risks to subsistence fishermen by
reducing the level of contaminants in fish.
Corrective action could potentially benefit agricultural production by increasing
availability of water for irrigation or livestock. In the absence of corrective action, ground-water
wells near corrective action facilities used to irrigate crops or water livestock could be closed due
to contamination. If wells were closed, agricultural activities supported by the wells would be
adversely affected or perhaps eliminated. If the contaminated wells were not closed, production
of livestock and crops could be adversely affected by toxic effects of uncontrolled contamination.
In addition to these effects, corrective action could increase agricultural production by making
facility properties suitable for agricultural use after closure.
"Chapter 6 discusses all of the potential benefits of corrective action, and those which are and
are not addressed in the RIA. See Section 6.12.
* DRAFT - March 24, 1993
-------
13-11
A frequently cited cost of environmental regulation that was not considered in this RIA is
the competitive disadvantage to U.S. firms engaged in international trade. Environmental
regulation increases the production costs of U.S. firms. Competing firms operating in countries
with less stringent environmental regulations will have a cost advantage, and they may increase
their market share at the expense of U.S. firms. The Agency believes that corrective action
would not favor foreign firms operating in the U.S. over domestic firms, because the program
would apply to all TSDFs, regardless of ownership. However, the Agency has not attempted to
evaluate the expected impact of the rule on U.S. competitiveness with firms operating abroad.
Another issue not addressed in the RIA is the distribution of costs and benefits of
corrective action, including the geographic and inter-generational distributions. Corrective action
addresses contamination at specific sites in the U.S., and many of its benefits accrue to the
persons who are located near these sites. The costs of the program fall on the firms affected,
and ultimately on their suppliers (including their workers), their stockholders, and their
consumers. There is no reason to expect that the people who bear the costs are the same people
that receive the benefits. The inter-generational distribution issues are determined by the
distribution of costs and benefits across time. The costs of the program accrue immediately,
while the benefits continue into the future. A decision to undertake corrective action would
impose costs on present generations to provide benefits, in part, to future generations. Not
undertaking corrective action would save the present generation its costs, but deny future
generations its benefits. Though these distributional issues are important, they were beyond the
scope of this RIA.
DRAFT « March 24,1993
-------
BIBLIOGRAPHY
Abdalla, Charles W. "Measuring Economic Losses from Ground Water Contamination: An
Estimate of Household Avoidance Costs," Water Resources Bulletin. Volume 26, Number 3
(June 1990), pp. 451-463.
Abdalla, Charles W., Roach, Brian A., and Epp, Donald J. "Valuing Environmental Quality
Changes Using Averting Expenditures: An Application to Groundwater Contamination," Land
Economics. Volume 68, Number 2 (May 1992), pp. 163-169.
Adler, K., Anderson, R.C., Cook, Z., Dower, R., and Ferguson, A.R. The Benefits of
Regulating Hazardous Waste Disposal: Land Values as an Estimator," Report by the Public
Interest Economics Center to the U.S. Environmental Protection Agency (1982).
Andreoni, James. "Impure Altruism and Donations to Public Goods: A Theory of Warm-Glow
Giving," The Economic Journal. 1990, Vol. 100, pp. 464-477.
Atkinson, Scott E.; Crocker, Thomas D.; and Shogren, Jason F. "Bayesian Exchangeability,
Benefit Transfer, and Research Efficiency," Water Resources Research. 1992, Vol. 28, No. 3, pp.
715-722.
Baes, C.F. Ill; Sharp, R.D.; Sjoreen, P.L.; and Herman, D.W. TERRA: A Computer Code for
Simulating the Transport of Environmentally Released Radionuclides through Agriculture. Oak
Ridge, TN: Oak Ridge National Laboratory, November 1984. ORNL5785.
;
Balson, William E.; Hausman, Jeff; and Hulse, Annette E. Navajo Generating Station fNGS}
BART Analysis. Report prepared by Decision Focus Incorporated, Los Altos, CA for the owners
of the Navajo Generating Station. 1991.
Bartik, Timothy J. "Evaluating the Benefits of Non-Marginal Reductions in Pollution Using
Information on Defensive Expenditures," Journal of Environmental Economics and Management.
1988, Vol. 15, pp. 111-127.
Beyer, W.N. Evaluating Soil Contamination. Washington, D.C.: U.S. Department of Interior,
U.S. Fish and Wildlife Service. Biological Report Volume 90 (1990), pp.16-17.
Boyle, Kevin J. and Bergstrom, John C. "Benefit Transfer Studies: Myths, Pragmatism, and
Idealism," Water Resources Research. 1992, Vol. 28, No. 3, pp. 657-663.
Brookshire, David S. and Neill, Helen R. "Benefit Transfers: Conceptual and Empirical Issues,"
Water Resources Research. 1992, Vol. 28, No. 3, pp. 651-655.
Bumess, H.; Cummings, R.G.; Granderton, P.T.; and Harrison, G.W. "Valuing Environmental
Goods: A Critical Appraisal of the State of the Art," in A. Dinar and D. Zilberman eds. The
Economics and Management of Water and Drainage in Agriculture. Boston: Kluwer Academic,
1991.
DRAFT - March 23, 1993 *
-------
BIB-2
Chemical Manufacturers Association. Impact Analysis of RCRA Corrective Action and
CERCLA Remediation Programs. Washington, D.C.: Chemical Manufacturers Association, 1988.
Chestnut, Lauraine G. and Rowe, Robert D. Preservation Values for Visibility Protection at the
National Parks. Draft Final Report prepared for Economic Analysis Branch, Office of Air
Quality Planning and Standards, U.S. Environmental Protection Agency and Air Quality
Management Division, National Park Service. February 16,1990.
Clark, Robert and Adams, Jeffrey. EPA's Drinking Water and Ground Wafer Remediation Cost
Evaluation: Granular Activated Carbon. Michigan: Lewis Publishers, 1991.
Clark, Robert and Adams, Jeffrey. EPA's Drinking Water and Ground Water Remediation Cost
Evaluation: Air Stripping. Michigan: Lewis Publishers, 1991.
Congress of the United States, Office of Technology Assessment. Complex Cleanup: The
Environmental Legacy of Nuclear Weapons Production. Washington D.C.: Congress of the
United States, Office of Technology Assessment, February 1991.
Courant, Paul and Porter, Richard. "Averting Expenditure and the Cost of Pollution," Journal of
Environmental Economics and Management. 1981, Vol. 8, pp. 321-329.
Cropper, Maureen L. and Wallace, E. Gates. "Environmental Economics: A Survey," Journal of
Economic Literature. Vol. 30, (June 1992), pp.675-740.
*Cummings, Ronald F.; Brookshire, David S.; and Schulze, William D. ed. Valuing
Environmental Goods: An Assessment of the Contingent Valuation Method.
Davis, et al. Quantitative Estimates of Soil Ingestion in Normal Children between the Ages of 2
and 7 Years: Population-based Estimates Using Aluminum. Silicon, and Titanium as Soil Tracer
Elements. Archives of Environmental Health, March/April 1990, Vol. 45, No. 2, pp.113-122.
Deck, Leland B. and Chestnut, Lauraine G. "Benefits Transfer: How Good is Good Enough,"
Paper presented at the 1992 Association of Environmental and Resource Economists Workshop
on benefits transfer held in Snowbird, UT. June 3-5, 1992.
Denton, F.T. "Data Mining as an Industry," Review of Economics and Statistics. Vol. LXVII
(February 1985).
Desvousges, William H.; Naughton, Michael C; and Parsons, George R. "Benefit Transfer:
Conceptual Problems in Estimating Water Quality Benefits Using Existing Studies," Water
Resources Research. 1992, Vol. 28, No. 3, pp. 675-683.
Diamond, Peter A. and Hausraan, Jerry A. "On Contingent Valuation Measurement of Nonuse
Values," in Cambridge Economics Inc. Contingent Valuation: A Critical Assessment. Papers
presented at a symposium in Washington, D.C., April 2-3,1992.
* * DRAFT - March 23, 1993
* * *
-------
BIB-3
Dillman, D.A. Mail and Telephone Surveys: The Total Design Method. New York: John Wiley
and Sons, 1978.
Edwards, Steven F. and Anderson, Glen D. "Overlooked Biases in Contingent Valuation
Surveys: Some Considerations," Land Economics. 1978, Vol. 63, No. 3, pp. 168-178.
'Environmental Research and Technology (ERT). The Land Treatabilitv of Appendix VIII
Constituents Present in Petroleum Industry Wastes. Environmental Research and Technology
(ERT), February 1984.
Freeman, Myrick A. III. The Benefits of Environmental Improvement. Baltimore, MD: Johns
Hopkins University Press, 1979.
Freeman, Myrick A. III. The Benefits of Environmental Improvement (revV Baltimore, MD:
Johns Hopkins University Press, 1991.
Freshwater Foundation. Economic Implications of Groundwater Contamination to Companies
and Cities. Washington D.C.: Freshwater Foundation, 1989.
Friedman, Lee S. Microeconomic Policy Analysis. New York: McGraw-Hill Book Co., 1984.
*GeoServices Inc. Liner/Leak Detection Rule Background Document. GeoServices Inc., April
1987.
Harrington, Winston, and Portney, Paul. "Valuing the Benefits of Health and Safety
Regulation," Journal of Urban Economics. Volume 22 (1987), pp. 101-112.
Harrison, D., Jr. "Housing Values and the Willingness to Pay for Hazardous Waste
Regulations," manuscript, John F. Kennedy School of Government, Harvard University (1983).
Harrison, D., Jr. and Stock, J. "Using the Hedonic Housing Method to Estimate the Benefits of
Hazardous Waste Cleanup," Volume I, part 1, of "Research and Demonstration of Improved
Methods for Carrying Out Benefit-Cost Analyses of Individual Regulations," Report for the U.,S.
Environmental Protection Agency (1984).
ICF Incorporated. Focus Chemicals for the Clean Air Act Amendments Great Waters Study.
Research Triangle Park, NC: U.S. Environmental Protection Agency, Office of Air Quality
Planning and Standards, draft report August 15,1991.
Just, Richard E.; Hueth, Darrell L.; and Schmitz, Andrew. Applied Welfare Economics and
Public Policy. Englewood Cliffs, NJ: Prentice-Hall Inc., 1982.
Kahnemann, Daniel and Knetsch, Jack L. "Valuing Public Goods: The Purchase of Moral
Satisfaction." Journal of Environmental Economics and Management. 1992, Vol. 22, pp. 57-70.
* *
DRAFT - March 23,1993 * *
-------
BIB-4
Kealy, Mary Jo; Montgomery, Mark; and Dovidio, John F. "Reliability and Predictive Validity of
Contingent Values: Does the Nature of the Good Matter?" Journal of Environmental Economics
and Management. 1990, Vol. 19, pp. 244-263.
Keup, L.E. "Flowing water resources," Water Resources Bulletin. 1985, Vol. 21, pp. 291-296.
Kohlhase, Janet E. The Impact of Toxic Waste Sites on Housing Values." Journal of Urban
Economics. 1991, Vol. 30, pp. 1-26.
Kopp, Raymond J. "Why Existence Value Should be Used in Cost-Benefit Analysis," Journal of
Policy Analysis and Management. 1992, Vol. 11, No. 1, pp. 123-130.
Kopp, Raymond J.; Rosenthal, D.H.; and Nelson, R.H. "Why Existence Value Should Not be
Used in Cost-Benefit Analysis," Journal of Policy Analysis and Management. Volume 11,
Number 1 (1992), pp. 116-122.
Krutilla, J.V. "Conservation Reconsidered." American Economic Review. 1967, Vol. 57, No. 4,
pp. 777-786.
LaGoy, P.K. Estimated Soil Ingest ion Rates for Use in Risk Assessment. Risk Analysis, 1987,
Vol.7, No. 3, pp.355-359.
Loomis, John B. "Expanding Contingent Value Sample Estimates to Aggregate Benefit
Estimates: Current Practices and Proposed Solutions," Land Economics. 1987, Vol. 63, No. 4, pp.
396-402.
Loomis, John B. "The Evolution of a More Rigorous Approach to Benefit Transfer: Benefit
Function Transfer," Water Resources Research. 1992, Vol. 28, No. 3, pp. 701-705.
Mayer, F.L., Jr., and Ellersieck, M.R. Manual of Acute Toxiciry: Interpretation and Data Base
for 410 Chemicals and 66 Species of Freshwater Animals. Washington, D.C.: U.S. Department of
Interior, Fish and Wildlife Service, 1986.
McClelland, Gary H.; Schulze, William D.; Lazo, Jeffrey K.; Waldman, Donald M.; Doyle, James
K.; Elliott, Steven R.; and Irwin, Julie R. Methods for Measuring Non-Use Values: A
Contingent Valuation Study of Ground Water Cleanup. Draft report prepared by the Center for
Economic Analysis, Department of Economics, University of Colorado, Boulder for the Office of
Policy, Planning and Evaluation, U.S. Environmental Protection Agency. October 1992.
McClelland, G.H., Schulze, W.D., and Hurd, B. The Effect of Risk Beliefs on Property Values:
A Case Study of a Hazardous Waste Site," manuscript, University of Colorado (1989).
McConnell, K.E. "Existence and Bequest Value," in R. Rowe and L. Chestnut, eds. Managing
Air Quality and Scenic Resources at National Parks and Wilderness Areas. Boulder, Co.:
Westview Press, 1983.
» * * DRAFT - March 23, 1993
-------
BIB-S
Michaels, R.G. "A Discrete and Continuous Choice Valuation of a Hazardous Waste Site's
Removal," Ph.D. dissertation, Vanderbilt University (1987).
Michaels, R.G. "Prospects and Pitfalls: An Evaluation of Hedonic Housing Analysis Applied to
Superfund Sites," Paper presented at the Annual Meeting of the American Real Estate and
Urban Economics Association, New York, NY (1988).
Michaels, R. Gregory and Smith, V. Kerry. "Market Segmentation and Valuing Amenities with
Hedonic Models: The Case of Hazardous Waste Sites," Journal of Urban Economics. 1990, Vol.
28, pp. 223-242.
Michaels, G., Byrnes, R., Giordano, P., Lin, W., and Wojcik, W. "Hazardous Waste Facilities
and Residential Property Values," Draft Report by Abt Associates Inc. to the U.S.
Environmental Protection Agency (1992).
Mitchell, Robert Cameron and Carson, Richard T. Using Surveys to Value Public Goods: The
Contingent Valuation Method. Washington D.C.: Resources for the Future, 1989.
National Academy of Sciences. Risk Assessment in the Federal Government. Washington D.C.:
National Academy Press, 1983.
New York State Department of Environmental Conservation. Niagara River Biota
Contamination Project: Fish Flesh Criteria for Piscivorous Wildlife. Albany, NY: Division of
Fish and Wildlife,Bureau of Environmental Protection, 1987. Technical Report 87-3.
Nicolson, Walter. Microeconomic Theory. Florida: Dryden Press, 1989.
Office of Management and Budget. Guidelines and Discount Rates for Benefit-Cost Analysis of
Federal Programs - Circular A-94. Washington, D.C.: Office of Management and Budget,
October 29, 1992, page 9.
Palmquist, Raymond B. "Hedonic Methods," chapter in Measuring the Demand for
Environmental Quality. John B. Braden and Charles D. Kolstad, Editors. Amsterdam: North
Holland. 1991.
Portner, J. Winston and Price, Courtney M. Interpretation of Section 3008fhi of the Solid Waste
Disposal Act. Washington, D.C.: U.S. Environmental Protection Agency, December 15, 1985.
Portney, Paul R., The Economics of Hazardous Waste Regulation," in U.S. Waste Management
Policies: Impact on Economic Growth and Investment Strategies. Washington, D.C.: American
Council for Capital Formation, 1992.
Press, W.H.; Flannery, B.P.; Teukolsky, S.A.; and Vetterling, W.T. Numerical Recipes in C.
Cambridge: Cambridge University Press, 1988.
* * DRAFT -- March 23, 1993 *
-------
BIB-6
Rosenthal, D.H. and Nelson, R.H. "Why Existence Value Should Not Be Used in Cost-Benefit
Analysis." Journal of Policy Analysis and Management. 1992, Vol. 11, No. 1, pp. 116-122.
R.S. Means Company. MEANS Site Work Cost Data. Kingston, Massachusetts: R.S. Means
Company. Published annually.
Russell, Milton, et al. Hazardous Waste Remediation: The Task Ahead. Knoxville, Tennessee:
University of Tennessee, 1991.
Schulze, W., McClelland, G., Kurd, B., and Smith, J. "Improving Accuracy and Reducing Costs
of Environmental Benefit Assessments: Volume IV, Estimating Benefits for Toxic Waste
Management: An Application of the Property Value Method," 1986. Draft Report, U.S.
Environmental Protection Agency contract #CR812054-01-1.
Smith, V. Kerry, and Desvousges, William H. "Averting Behavior: Does it Exist?" Economic
Letters. Volume 20 (1986), pp. 291-296.
Smith, V. Kerry. "On Separating Defensible Benefit Transfers from 'Smoke and Mirrors'." Water
Resources Research. 1992, Vol. 28, No. 3, pp. 685-694.
Smith, V.K. and Desvousges, W.H. The Value of Avoiding a LULU: Hazardous Waste
Disposal Sites," Review of Economics and Statistics. Vol. LXVIII, No. 2 (1986).
Smith, V. Kerry. "Valuing Ground Waster Resources: A Conceptual Overview." Proceedings of
the Clean Water and the American Economy Conference (draft). Sponsored by the U.S.
Environmental Protection Agency and Resources for the Future. October 1992.
Smolen, Gerald E.; Moore, Gary; and Conway, Lawrence V. "Hazardous Waste Landfill Impacts
on Local Property Values," The Real Estate Appraiser. April 1992.
Sobotka & Company. Experience Curves. Innovation and Diffusion: Application to RCRA.
Washington, D.C.: Sobotka & Company, April 1992.
Stokey, Edith and Zeckhauser, Richard. A Primer for Policy Analysis. New York: W.W. Norton
and Co., 1978.
Sutherland Ronald J. and Walsh, Richard G. "Effect of Distance on the Preservation Value of
Water Quality," Land Economics. 1985, Vol. 61, No. 3, pp. 281-291.
Thayer, M., Albers, H., and Rahmatian, M. "The Benefits of Reducing Exposure to Waste
Disposal Sites: A Hedonic Housing Value Approach," manuscript (1991).
Tonn, Bruce, Oak Ridge National Laboratory, letter to Gary Ballard, U.S. Environmental
Protection Agency, September 9, 1991.
* DRAFT - March 23,1993 * * *
-------
BIB-7
U.S. Army Corps of Engineers. Influence of Price and Rate Structures on Municipal and
Industrial Water Use. Washington D.C.: U.S. Army Corps of Engineers, June 1984.
U.S. Department of Commerce, Bureau of the Census. Data from the 1992 Census of
Population and Housing provided by the Housing and Household Economic Statistics Division
for Industrial Economics, Inc., June 9,1992.
U.S. Department of Commerce. GDP deflator from personal communication with the Bureau of
Economic Analysis, February 9,1993.
U.S. Department of Commerce. "Natural Resource Damage Assessment Under the Oil
Pollution Act of 1990 - Appendix I - Report of the NOAA Panel on Contingent Valuation,"
Volume 58 (1993), Federal Register, pp. 4601-4614, January 15, 1993.
U.S. Department of Defense. Defense Environmental Restoration Program: Fiscal Year 1991
Report to Congress. Washington D.C.: U.S. Department of Defense, February 1992.
U.S. Department of Defense. Installation Restoration Program Cost Estimate. Washington D.C.:
U.S. Department of Defense, September 1991.
U.S. Department of Energy. Chemical Data Base for the Multimedia Environmental Pollutant
Assessment System (MEPAS): Version 1 (Draft for External ReviewV Richland, WA: External
review draft prepared for the U.S. Department of Energy Assistant Secretary, Office of
Environmental Audit, by Pacific Northwest Laboratories, September 1989.
U.S. Department of Energy. Costs of RCRA Corrective Action: Interim Report. Washington,
D.C.: Washington, D.C.: U.S. Department of Energy, 1991.
U.S. Department of Energy. Environmental Restoration and Waste Management. Five Year
Plan. Washington, D.C.: U.S. Department of Energy, August 1991.
U.S. Environmental Protection Agency. A Compendium of Technologies Used in the Treatment
of Hazardous Wastes. Washington D.C.: U.S. Environmental Protection Agency, September
1987.
U.S. Environmental Protection Agency. A PC Software Application of the Uptake/Biokinetic
Model. Version 0.5. Washington, D.C.: U.S. Environmental Protection Agency, Office of Health
and Environmental Assessment. ECAO-CIN-2178A.
U.S. Environmental Protection Agency. A Tiered Modeling Approach for Assessing the Risks
Due to Sources of Hazardous Air Pollutants. Research Triangle Park, North Carolina: U.S.
Environmental Protection Agency, Office Air Quality Planning and Standards, March 1992. EPA
Report No. EPA/450/4-92-001.
U.S. Environmental Protection Agency. Background Document for EPA's Composite Model for
Landfills rEPACMLV Washington D.C: U.S. Environmental Protection Agency, February 1990.
* DRAFT -- Mareh 23, 1993
-------
BIB-8
U.S. Environmental Protection Agency. Data Base Analyzer and Parameter Estimator fDBAPE)
Interactive Computer Program User's Manual. Washington, D.C.: U.S. Environmental Protection
Agency, 1990. EPA/600/3-89/083.
U.S. Environmental Protection Agency. Dermal Exposure Assessment: Principles and
Applications. Washington D.C.: U.S. Environmental Protection Agency, 1992. EPA-600/8-
91/011B.
U.S. Environmental Protection Agency. Developing a Work Scope for Ecological Assessments.
ECO Update, Intermittent Bulletin, Volume 1, Number 4. Washington, D.C.: U.S. Environmental
Protection Agency, Office of Emergency and Remedial Response, .Hazardous Site Evaluation
Division, 1992. Publication 9345.0-051.
U.S. Environmental Protection Agency. Draft RCRA Corrective Action Prioritization System
Guidance Document. Washington, D.C.: U.S. Environmental Protection Agency, April 1991.
U.S. Environmental Protection Agency. Ecological Assessment of Superfund Sites: An Overview.
ECO Update, Intermittent Bulletin, Volume 1, Number 2. Washington, D.C.: U.S. Environmental
Protection Agency, Office of Emergency and Remedial Response, Hazardous Site Evaluation
Division, 1991. Publication 9345.0-051.
U.S. Environmental Protection Agency. EPA Guidelines for Implementing the Regulatory
Flexibility Act. Washington, D.C.: U.S. Environmental Protection Agency, 1992.
U.S. Environmental Protection Agency. Engineering Bulletin: In Situ Steam Extraction
Treatment. Washington, D.C.: U.S. Environmental Protection Agency, May 1990. EPA/540/2-
91/005.
U.S. Environmental Protection Agency. Engineering Bulletin: Soil Washing Treatment.
Washington, D.C.: U.S. Environmental Protection Agency, September 1990. EPA/540/2-90/017.
U.S. Environmental Protection Agency. Estimating Exposure to 2.3.7.8 - TCDD. U.S.
Environmental Protection Agency, Office of Health and Environmental Assessment, March 1988.
EPA/600/6-88/005A.
U.S. Environmental Protection Agency. Exposure Factors Handbook. Washington D.C.: U.S.
Environmental Protection Agency, Office of Health and Environmental Assessment, 1989.
EPA/600/8-89/043.
U.S. Environmental Protection Agency. Flow and Transport Modeling in Karst Aquifers: A
Status Report. Athens, Georgia: U.S. Environmental Protection Agency Environmental Research
Laboratory.
U.S. Environmental Protection Agency. Framework for Ecological Risk Assessment.
Washington, D.C.: U.S. Environmental Protection Agency, 1992. EPA Report No. EPA/630/R-
92-001.
* * * DRAFT - March 23, 1993
*
-------
BIB-9
U.S. Environmental Protection Agency. Guidance on Assessment. Criteria Development, and
Control of Bioconcentratable Contaminants in Surface Waters. Prepared by the National
Effluent Toxicity Assessment Center, Environmental Research Laboratory Duluth; Office of
Water Enforcement and Permits; Office of Water Regulations and Standards; and Office of
Health Effects Assessment Cincinnati, June 1989.
U.S. Environmental Protection Agency. Guidance for Establishing Target Cleanup Levels for
Soils at Hazardous Waste Sites. Washington, D.C.: U.S. Environmental Protection Agency, 1988.
U.S. Environmental Protection Agency. Guidance on Remedial Actions for Contaminated
Ground Water at Superfund Sites. Washington, D.C.: U.S. Environmental Protection Agency,
1988.
U.S. Environmental Protection Agency. Guidance on Risk Characterization for Risk Managers
and Risk Assessors. Memorandum to EPA Assistant Administrators and Regional
Administrators. Washington D.C.: U.S. Environmental Protection Agency, February 26, 1992.
U.S. Environmental Protection Agency. Guide to Treatment Technologies for Hazardous Wastes
at Superfund Sites. Washington D.C.: U.S. Environmental Protection Agency, March, 1989.
U.S. Environmental Protection Agency. Guidelines for Deriving Numerical Criteria for the
Protection of Aquatic Organisms and Their Uses. Washington, D.C.: U.S. Environmental
Protection Agency, Office of Water Regulations and Standards, 1986.
U.S. Environmental Protection Agency. Guidelines For Performing Regulatory Impact Analysis:
Appendix C - Analysis of the Choice of Discount Rates. Washington D.C.: U.S. Environmental
Protection Agency, 1989.
U.S. Environmental Protection Agency. "Guidelines for the Health Risk Assessment of Chemical
Mixtures." 51 FR 34014, September 24,1986.
U.S. Environmental Protection Agency. Handbook on In Situ Treatment of Hazardous Waste-
Contaminated Soils. Washington D.C.: U.S. Environmental Protection Agency, January 1990.
EPA/540/2-90/002.
U.S. Environmental Protection Agency. Handbook - Remedial Action at Waste Disposal
Facilities. Washington D.C.: U.S. Environmental Protection Agency, 1985.
U.S. Environmental Protection Agency. Hazard Ranking System: Final Rule. 40CFR Part 300.
Federal Register. December 14,1990.
U.S. Environmental Protection Agency. Hazardous Waste Land Treatment. Washington, D.C.:
U.S. Environmental Protection Agency, April 1983.
U.S. Environmental Protection Agency. "Hazardous Waste Tank Risk Analysis, Draft Report."
prepared by ICF Incorporated and Pope-Reid Associates Incorporated, June 1986.
DRAFT - March 23,1993
-------
BIB-10
U.S. Environmental Protection Agency. Health Effects Assessment and Summary Tables.
(HEAST, March 1992).
U.S. Environmental Protection Agency. Human Health Evaluation Manual. Supplemental
Guidance: Standard Default Exposure Factors. Washington D.C.: U.S. Environmental Protection
Agency, 1991. OSWER Directive #9285.6-03.
U.S. Environmental Protection Agency. Indexing of Lonp-Term Effectiveness of Waste
Containment Systems for a Regulatory Impact Analysis (draft). Washington, D.C.: U.S.
Environmental Protection Agency, Office of Solid Waste, December 1992.
U.S. Environmental Protection Agency. Industrial Source Complex flSO Dispersion Model
User's Guide - Second Edition (Revised^. Washington, D.C.: U.S. Environmental Protection
Agency, 1987. EPA/450/4-38-002a.
U.S. Environmental Protection Agency. Innovative Treatment Technologies: Overview and
Status - Preliminary Draft. Washington, D.C.: U.S. Environmental Protection Agency, March
1991.
U.S. Environmental Protection Agency. In-Situ Treatment of Hazardous Waste-Contaminated
Soils. Washington, D.C.: U.S. Environmental Protection Agency, January 1990.
U.S. Environmental Protection Agency. Integrated Risk Information System fIRlSV (IRIS,
November 1992 update).
U.S. Environmental Protection Agency. Interim Guidance for Dermal Exposure Assessment.
Washington D.C.: U.S. Environmental Protection Agency, 1991. EPA/600/8-91/011 A.
U.S. Environmental Protection Agency. Interim Sediment Criteria Values for Nonpolar
Hvdrophobic Organic Compounds. Washington, D.C.: U.S. Environmental Protection Agency,
1988.
U.S. Environmental Protection Agency. MMSOILS: Multimedia Contaminant Fate. Transport.
and Exposure Model - Documentation and Users Guide. Draft. Washington, D.C.: U.S.
Environmental Protection Agency, September, 1992.
U.S. Environmental Protection Agency. Permit Writer's Guide to Water Quality-Based
Permitting for Toxic Pollutants. Washington, D.C.: U.S. Environmental Protection Agency, .Office
of Water Regulations and Standards. 1987. EPA Report No. EPA 440/4-87-005.
U.S. Environmental Protection Agency. "Proposed Framework to Evaluate Uncertainty in
- Chemical Fate and Transport Predictions for RCRA Corrective Action Regulatory Impact
Analysis." Athens, Georgia: U.S. Environmental Protection Agency Environmental Research
Laboratory, September 1992.
* * * DRAFT - March 23,1993
-------
BIB-11
U.S. Environmental Protection Agency. Regulatory Impact Analysis for the Proposed
Rulemaking on Corrective Action for Solid Waste Management Units. Washington, D.C.: U.S.
Environmental Protection Agency, June 25,1990.
U.S. Environmental Protection Agency. "Review and Recommendations Relate to Chemical
Data Used in the Corrective Action Regulatory Impact Analysis," memorandum from Gerard
Laniak, EPA/ERI^Athens, to Barnes Johnson, Office of Solid Waste. November 7,1991.
U.S. Environmental Protection Agency. Review of Ecological Risk Assessment Methods.
Washington, D.C.: U.S. Environmental Protection Agency, Office of Policy, Planning, and
Evaluation, November 1988. EPA Report No. EPA/230/10-88/041.
U.S. Environmental Protection Agency. Risk Analysis of TCDD Contaminated Soil.
Washington, D.C.: U.S. Environmental Protection Agency, Office of Health and Environmental
Assessment, November 1984. EPA/600/6-84-031.
U.S. Environmental Protection Agency. Risk Assessment Guidance for Superfund Vol. I.
Human Health Evaluation Manual (Part AV Washington D.C.: U.S. Environmental Protection
Agency, December 1989. EPA/540/1-89/002.
U.S. Environmental Protection Agency. Risk Assessment Guidance for Superfund. Volume II:
Environmental Evaluation Manual. Interim Final. Washington, D.C.: U.S. Environmental
Protection Agency, Office of Emergency and Remedial Response, March 1989. EPA Report No.
EPA/540/1-89/001.
U.S. Environmental Protection Agency. Risk of Unsaturated/Saturated Transport and
Transformation of Chemical Concentration (RUSTICS Volume I: Theory and Code Verification.
Washington, DC: U.S. Environmental Protection Agency, 1989. EPA/600/3-89-48a.
U.S. Environmental Protection Agency. "Selection of Models for the Corrective Action
Regulatory Impact Analysis," memorandum from Loretta Marzetti, Director, Communications,
Analysis and Budget Division, Office of Solid Waste, to Peter Preuss, Director, Office of
Technology Transfer and Regulatory Support, Office of Research and Development, January 9,
1991.
U.S. Environmental Protection Agency'. Selecting Sample Facilities for the Corrective Action
Regulatory Impact Analysis: Draft Report. Washington, D.C.: U.S. Environmental Protection
Agency, June 1991.
U.S. Environmental Protection Agency. Seminar Publication - Corrective Action: Technologies
and Applications. Washington D.C.: U.S. Environmental Protection Agency, September 1989.
EPA/625/4-89/020.
U.S. Environmental Protection Agency. Standard Evaluation Procedure for Ecological Risk
Assessment. Washington, D.C.: U.S. Environmental Protection Agency, Office of Pesticide
Programs. Prepared by Urban, D.J., and Cook, N.J... Hazard Evaluation Division. 1986.
DRAFT -- March 23, 1993
-------
BIB-12
U.S. Environmental Protection Agency. Subsurface Contamination Reference Guide.
Washington D.C.: U.S. Environmental Protection Agency, October 1990.
U.S. Environmental Protection Agency. Summary Regulatory Impact Analysis/Background
Information Document for the Development of Subtitle C Location Standards under Section
3004 (o)f7) of RCRA. Washington, D.C.: U.S. Environmental Protection Agency, January 26,
1990.
U.S. Environmental Protection Agency. Summary of Review and Evaluations of the Technical
Approach to Corrective Action Regulatory Impact Analysis: Fate and Transport. Athens,
Georgia: U.S. Environmental Protection Agency, Office of Research and Development,
Environmental Research Laboratory, 1992.
U.S. Environmental Protection Agency. Sunerfund Chemical Data Matrix. Washington, D.C.:
U.S. Environmental Protection Agency, datafile compiled in April 1991.
U.S. Environmental Protection Agency. Superfund Exposure Assessment Manual, Draft Report.
Washington, D.C.: U.S. Environmental Protection Agency, 1987.
U.S. Environmental Protection Agency. "Superfund LDR Guide #6A, Obtaining a Soil and
Debris Treatability Variance for Remedial Actions." Washington, D.C.: U.S. Environmental
Protection Agency, Office of Solid Waste, No. 9347.3-06FS, September 1990.
U.S. Environmental Protection Agency. Technical Guidance Document: Final Covers on
HayarHnm Waste Landfills and Surface Impoundments. Washington D.C.: U.S. Environmental
Protection Agency, July 1989. EPA/530-SW-89-047.
U.S. Environmental Protection Agency. Technical Support Document. Onshore Oil and Gas
Exploration. Development, and Production: Human Health and Environmental Risk Assessment.
Washington, D.C.: U.S. Environmental Protection Agency, 1987.
U.S. Environmental Protection Agency. The RCRA Risk-Cost Analysis Model Phase III Report.
Washington, D.C.: U.S. Environmental Protection Agency, March 1,1984.
U.S. Environmental Protection Agency. The Superfund Innovative Technology Evaluation
(Srra Program: Technology Profiles. Washington D.C.: U.S. Environmental Protection Agency,
November 1990. EPA/540/5-90/009.
U.S. Environmental Protection Agency. Database developed for Toxicity Characteristic
Repulatorv Impact Analysis. Washington, D.C.: U.S. Environmental Protection Agency, March
1990.
U.S. Environmental Protection Agency. Toiicitv Characteristic Regulatory Impact Analysis.
Washington, D.C.: U.S. Environmental Protection Agency, 1990.
* * DRAFT - March 23,1993 * *
-------
BIB-13
U.S. Environmental Protection Agency. Water Quality Criteria to Protect Wildlife Resources.
Corvallis, OR: U.S. Environmental Protection Agency, Office of Research and Development,
Environmental Research Laboratory, December 1989. EPA Report No. EPA/600/3-89/067.
U.S. General Accounting Office. Nuclear Health and Safety: Long-Term Plans to Address
Problems of the Weapons Complex are Evolving. Washington D.C.: U.S. General Accounting
Office, September 1990.
Water Resources Council. "Procedures for Evaluation of National Economic Development
(NED): Benefits and Costs in Water Resources Planning (Level C), Final Rule." 44 Federal
Register 72892-997. December 14,1979.
Willig, Robert. "Consumer's Surplus Without Apology." The American Economic Review.
September 1976.
Winston, M.R.; Taylor, C.M.; and Pigg. J. "Upstream Extirpation of Four Minnow Species Due
to Damming of a Prairie Stream." Transactions of the American Fisheries Society. 1991, Volume
120, pp.98-105.
DRAFT « March 23, 1993
------- |