United States Office of Water EPA815-B-98-002
Environmental Protection (4607) November 1998
Agency
&EPA REGULATORY IMPACT ANALYSIS
FOR THE STAGE 1
DISINFECTANTS/DISINFECTION
BYPRODUCTS RULE
-------
gulatory Impact Analysis
for the
age 1 Disinfectants/Disinfection Byproducts Rule
PREPARED FOR:
U.S. ENVIRONMENTAL PROTECTION AGENCY
Office of Ground Water and Drinking Water
Ms. Valerie Blank, Task Order Project Manager
Ms. Maggie Javdan, Technical Manager
PREPARED BY:
THE CADMUS GROUP, INC.
4900 Seminary Road
Alexandria, VA 22311
SCIENCE APPLICATIONS INTERNATIONAL CORPORATION
1710 Goodridge Drive
Mclean, VA 22102
US EPA CONTRACT: 68-C6-0059
Work Assignment: 118-6
November 12,1998
-------
Regulatory Impact Analysis for the
Stage 1 Disinfectants/Disinfection Byproducts Rule
Errata sheet
1. Chapter 5, page 5-3
Exhibit 5.1 "Summary of Costs under the Stage 1 DBPR"
Change title to read "Summary of Costs under the Stage 1 DBPR ($000)"
2. Chapter 5, page 5-22
Exhibit 5.11 "Regulatory Flexibility Act Cash Flow Analysis for Small Ground Water
Systems"
Add "($000)" to the headers of the following columns: Total Revenue, Op. Exp., Net
Total Rev., Total DBF Cost, Jncr. DBF Op. Exp, Post-DBF Net Rev.
-------
-------
Contents
Executive Summary:
ES.l Protection of Public Health . ES-1
ES.2 Exposure ES-2
ES.3 Health Hazards ES-2
ES.4 Risk Assessment and Uncertainty ES-2
ES.5 Compliance Costs and Treatment Effectiveness ES-3
ES.6 National Benefit/Cost Comparisons ES-3
ES.7 Household Benefit/Cost Comparisons ES-3
ES.8 Central Tendency Estimates of Net Benefits ES-4
ES.9 Consideration of Alternatives ES-4
ES.10 Conclusion ES-6
Chapter 1: Introduction 1-1
1.1 Introduction 1-1
1.2 Description of the Issue 1-1
1.3 Regulatory History 1-3
1.4 Public Health Concerns to be Addressed 1-4
1.5 Summary of the Rule 1-5
1.6 Environmental Justice 1-6
1.7 Unfunded Mandates Reform Act Analysis 1-6
1.8 Regulatory Flexibility Analysis 1-7
Chapter 2: Consideration of Regulatory Alternatives 2-1
2.1 Chronological Review of Regulatory Options Considered 2-1
2.2 Options with Complete Cost and Benefit Analyses 2-3
Chapter 3: Baseline Conditions 3-1
3.1 Introduction 3-1
3.2 Industry Profile 3-1
3.3 Influent Water Quality 3-4
3.4 Existing Treatment Characterization 3-12
3.5 Risk Assessment and Benefit Analysis 3-12
Chapter 4: Benefits Analysis 4-1
4.1 Introduction 4-1
4.2 Health Risks from Exposure to DBPs 4-1
4.3 Exposure Assessment 4-6
4.4 Baseline Risk Assessment Based on TTHM Toxicological Data 4-7
4.5 Baseline Risk Assessment for Bladder Cancer Based on
Epidemiological Data 4-10
4.6 Baseline Risk Assessment for Other Cancers Based on
Epidemiological Data 4-14
4.7 Baseline Risk Assessment for Reproductive and Developmental Health Effects
Based on Epidemiological Data 4-14
-------
4.8 Exposure Reduction Analysis 4-16
4.9 Expected Benefits from Reduction in Exposure to DBFs 4-17
4.10 Monetization of Bladder Cancer Health Endpoints 4-18
4.11 Range of Potential Monetized Benefits from Reducing Bladder Cancer . . 4-21
Chapter 5: Cost Analysis 5-1
5.1 Introduction 5-1
5.2 The Stage 1 DBPR and New Data 5-1
5.3 Estimated National Costs of the Stage 1 DBPR 5-2
5.4 Compliance Treatment Forecast 5-4
5.5 Estimated System Costs of the Stage 1 DBPR 5-11
5.6 Small System Impacts—Regulatory Flexibility Act Analysis 5-15
5.7 Combined Effect of the Stage 1 DBPR and the IESWTR 5-25
Chapter 6: Assessing Net Benefits of the Stage 1 DBPR 6-1
6.1 Alternative Approaches for Assessing Benefits of the Stage 1 DBPR .... 6-1
6.2 Overlap of Benefit and Cost Estimates 6-1
6.3 Minimizing Total Social Losses Analysis 6-3
6.4 Breakeven Analysis 6-10
6.5 Household Cost Analysis 6-14
6.6 Decision-Analytic Model 6-20
Chapter 7: The Economic Rationale for Regulation 7-1
7.1 Introduction 7-1
7.2 Statutory Authority for Promulgating the Rule 7-1
7.3 The Economic Rationale for Regulation 7-1
References
-------
Exhibits
Executive Summary
ES.l Summary of Costs under the Stage 1 DBPR ES-8
ES.2 Cumulative Distribution of Annual Average Systems Costs and Household
Costs for All Systems ES-10
ES.3 Detail of Cumulative Distribution of Annual Average Costs per System ES-12
ES.4 Cumulative Distribution of Annual Household Costs of All Systems ES-14
ES.5 Characteristics of Water Systems . ES-16
Chapter 1: Introduction 1-1
1.1 Typical Treatment Plant Model (Conventional) 1-2
Chapter 3: Baseline Conditions 3-1
3.1 Number of Systems that Disinfect by Source and Size 3-2
3.2 Number of Households 3-3
3.3 Cumulative Distribution of TOC Concentrations in Source Waters 3-6
3.4 Cumulative Distribution of Alkalinity in Source Waters 3-6
3.5a Percentage of TOC Removal Required under the Stage 1 DBPR 3-7
3.5b Systems within TOC and Alkalinity Parameters as a Percentage of Total Systems 3-8
3.5c Systems Meeting and Not Meeting Removal Targets within TOC and
Alkalinity Parameters . . . . 3-8
3.6 Systems Meeting and Not Meeting TOC Removal Targets 3-9
3.7 Systems Not Meeting TOC Removal Targets 3-10
3.8 Enhanced Coagulation Matrix Distribution of TOC Removal 3-11
3.9 Cumulative Distribution of TTHM Concentration within Distribution System ... 3-14
3.10 Cumulative Distribution of HAAS Concentration within Distribution System .... 3-14
3.11 Plant-Level Concentration of TTHMs and HAA5 within Distribution System .... 3-15
Chapter 4: Benefits Analysis 4-1
4.1 Steps in the Risk Assessment Process for Cancer 4-2
4.2 Potential Health Effects from Disinfectants and Disinfection Byproducts from
Laboratory Animal Studies 4-4
4.3 Population Potentially Exposed to DBPs 4-7
4.4 Baseline Cancer Incidence 4-9
4.5 Summary of Epidemiology Studies for Bladder Cancer 4-12
4.6 Bladder Cancer Epidemiology and Toxicology: Comparison of Estimates
Made in 1994 and 1998 4-13
4.7 Summary of Epidemiology Studies for Colon and Rectal Cancer 4-14
4.8 Summary of Epidemiology Studies for Reproductive and
Developmental Effects 4-15
4.9 Summary of Stage 1 DBPR Benefits 4-18
-------
Chapter 5: Cost Analysis .".. 5-1
5.1 Summary of Costs underthe Stage 1 DBPR 5-3
5.2 Summary of Costs under the Proposed Stage I DBPR in 1998 Dollars 5-4
5.3 Summary of Costs under the Proposed Stage 1 DBPR in 1992 Dollars 5-4
5,4 Comparisons of Large System Compliance Treatment Forecasts 5-7
5,5 Compliance Treatment Forecast by Type of Treatment 5-10
5.6 Annual Treatment Costs (Capital and O&M) of the Stage 1 DBPR by
Treatment Type 5-12
5,7 Monitoring Activities Required to Comply with the Stage 1 DBPR 5-13
5.8 Small Entities Affected by the Stage 1 DBPR 5-18
5.9 Summary of Annual Monitoring Activities and Estimated Burden ., 5-18
5.10 Regulatory Flexibility Act Cash Flow Analysis for Small Surface Water Systems 5-21
5.11 Regulatory Flexibility Act Cash Flow Analysis for Small Ground Water Systems 5-22
5.12 Cost Impact of Current and Expected Rulemakings 5-26
Chapter 6: Assessing Net Benefits of the Stage 1 DBPR 6-1
6. la Conceptual Overlap of Estimated Benefits and Costs of the Stage 1 DBPR 6-2
6.1b Overlap of the Ranges of Estimated Benefits and Costs of the Stage 1 DBPR .... 6-2
6.2 Stage 1 DBPR Minimizing Total Social Costs Analysis 6-5
6.3a Residua] Damages Monte Carlo Simulation Summary: No Action Scenario 6-6
6.3b Residual Damages Monte Carlo Simulation Summary: Stage 1 DBPR 6-6
6.3c Residual Damages Monte Carlo Simulation Summary: Strong Intervention 6-7
6.4 Stage 1 DBPR Minimizing Maximum Loss Analysis 6-9
6.5 Breakeven Analysis Willingness-to-Pay (WTP) Summary 6-11
6.6 Breakeven Analysis Cost-of-Illness (COI) Summary 6-12
6.7 Percent Reduction in Exposure Needed to Break Even by Baseline Number of
Attributable Cancer Cases 6-13
6.8 Cumulative Distribution of Annual Household Costs under the Stage 1 DBPR ... 6-16
6.9 Summary of the Number of Households and Percentage of Total Households
in Each Cost Category 6-17
6.10a Cumulative Distribution of Annual Costs per Household in Small Systems 6-18
6.1 Ob Cumulative Distribution of Annual Costs per Household in Large Systems 6-19
6.11 Density Function of Implementation Costs 6-23
6.12 Density Function of Exposure Reduction 6-23
6.13 Density Function of PAR Estimates 6-24
6.14 Cumulative Distribution of Predicted Net Benefits 6-24
-------
ACKNOWLEDGMENTS
This document was prepared for the U.S. Environmental Protection Agency, Office of Ground
Water and Drinking Water (OGWDW) by Science Application International Corporation (SAIC)
(Contract No. 68-C6-0059) and its subcontractor, The Cadmus Group, Inc. Overall planning and
management for the preparation of this document was provided by Maggie Javdan of OGWDW
and Tom Carpenter of SAIC.
EPA acknowledges the valuable contributions of those who wrote and reviewed this document.
They include: John Cromwell of Hagler Bailly Services, Inc.; James Albright, Curtis Haymore,
Rosemarie Odom, and Elena Ryan of The Cadmus Group, Inc.; Tom Carpenter of SAIC; Stig
Regli, Michael Messner, Maggie Javdan, Michael Cox, Melonie Williams, and John Bennett of
U.S. EPA. EPA also thanks the following external peer reviewers for their excellent review and
valuable comments on the draft manuscript: Charles Abdalla (University of California at
Berkeley), Gunther Craun (G. Craun and Associates), and Joe Jacangelo (Montgomery Watson).
-------
Executive Summary
ES.l Protection of Public Health
The primary mission of the Environmental Protection Agency (EPA) is to safeguard human health and
the environment. This document addresses the expected impacts—both improvements to public health
and the costs to industry and consumers—of one EPA regulation that will make water safer to drink,
Many water systems treat their water with a chemical disinfectant to prevent disease. The public health
benefits of common disinfection practices in preventing infectious diseases from microbiological
contaminants are significant and well-recognized. Disinfection, however, may pose risks of its own.
Disinfectants and their byproducts have been associated with potential health risks that include cancer
and reproductive and developmental effects. EPA has identified ways to significantly lessen these
potential risks at reasonable costs. To implement these changes, EPA is publishing a final Stage 1
Disinfectants/Disinfection Byproducts Rule (Stage 1 DBPR) that contains the new requirements for
water systems and this Regulatory Impact Analysis (R1A), which documents the costs and benefits of the
rule.
In exercising its responsibility to protect public health, EPA must often make regulatory decisions with
less than complete information and with uncertainties in the available information. This is because a
public health risk is often first identified as a "potential" health risk. At this level of understanding, it is
sometimes not clear whether the risk will, in fact, materialize. And, it is often the case that the risk could
materialize at varying degrees of severity ranging from trivial to significant. Preventive decisions can be
difficult because they often have to be made before all the facts are known. If action is delayed to obtain
a perfect understanding, it may be too late to prevent the damage. But if action is premature and
over-zealous, it can drain resources from other beneficial public health measures. Thus, a keen sense of
balance is required in each decision.
In the classic paradigm of public health decisionmaking, it is necessary to decide upon a prudent course
of action despite confounding factors. The decision process consists of weighing the available evidence
to gain as much insight as possible into expected or possible health outcomes while also weighing the
costs and technological realities of available responses. At one end of the spectrum, a "No Action"
option might be justified when the balance of health evidence suggests low exposure, low probability,
and low severity while the response technologies imply high costs and limited effectiveness. At the
opposite extreme, urgent and forceful action might be warranted when the health evidence suggests high
exposure, high probability, and high severity while the response technologies have modest costs and
good effectiveness. The Stage 1 DBPR lies in the middle of this spectrum. On balance, however, EPA
believes the weight of evidence suggests there is sufficient exposure, probability, and severity on the
health side to warrant a public health decision to accept the cost and technology impacts of the rule in
order to obtain the projected exposure reduction. Highlights of this balancing analysis are summarized in
the following discussion.
Stage 1 DBPR Final R1A ES-1 November 12, 1998
-------
effects of alternative technologies that have not yet been fully studied. In addition, avoiding such shifts
minimizes the need for capital expenditure until the risk from DBFs is better understood. Only 6.5
percent of utilities are projected to have to change technologies in order to comply with the Stage 1
DBPR.
ES.10 Conclusion
In the final analysis, the various benefit/cost comparisons developed in this RIA are quite useful in
assisting the balancing and weighing analyses that must be performed to support public health
decisionmaking. While the uncertainties prevent the various approaches to economic analysis from
producing definitive or deterministic answers, these analyses are nonetheless very informative. Based on
a careful weighing of the projected costs against the potential quantified and non-quantified benefits,
EPA has determined that the benefits of the rule justify its costs.
Stage / DBPR Final RIA
ES-6
November 12, 1998
-------
Exhibit ES.l
Summary of Costs under the Stage 1 DBPR
Notes
Costs for the Stage 1 DBPR are estimates of what the rule may cost over 20 years, expressed as an
annual figure. The rule includes requirements for additional treatment, construction of new facilities,
where necessary, and monitoring compliance. Both States and utilities bear the costs of the rule, although
State costs compose a relatively minor portion of the annual cost. These costs were estimated by
engineers and economists familiar with equipment, process, and labor costs using available data and
expert judgement.
Costs are provided for large and small surface and ground water systems. The exhibit tabulated the total
estimated cqst (over 20 years) of purchasing and implementing required treatments. These "capital costs"
are then annualized (multiplied by a capitalization factor), the "cost of capital," that determines what the
cost might be per year. The breakout of annualized costs are based on a 7 percent cost of capital, which is
the rate required by the Office of Management and Budget for benefit/cost analyses. Also shown are the
annualized capital costs using the 3 percent and 10 percent cost of capital. The annualized capital costs
(Row B) are then added to the operation and maintenance (O&M) costs (Row A) to derive the total
annual treatment costs of the rule (Row C).
The cost of a required treatment usually varies by the scale of the treatment. Therefore, for most
treatments, costs are estimated separately for each size category of system (in terms of the number of
people served by the system). Costs are usually expressed as dollars per 1,000 gallons of water ($/kgal).
These costs represent the marginal costs to systems to change treatment practices. This analysis estimates
the number of systems in each size category that might have to modify their treatment to meet the MCLs
or that might have to implement an approved technology. In addition to the capital costs of the treatment
techniques implemented, each system will incur annual O&M costs. The annual capital costs (annualized
over 20 years at each of the three costs of capital) are added to the O&M costs to estimate an annual cost
of the Stage 1 DBPR.
Each utility must monitor their own compliance with EPA regulations (Row E), and the State must
review this compliance (Row G). There are also costs ("start-up costs") associated with implementation
of a new regulation, for both utilities (Row D) as well as States (Row F), These annualized start-up costs,
added to the annual monitoring the treatment costs, form the basis for the total national compliance costs
(Row H).
At a 7 percent cost of capital, the Stage 1 DBPR is expected to result in annual costs of $701 million. At
3 percent, the annual costs are an estimated $626 million. At 10 percent, the annual costs are an
estimated $756 million.
Stage I DBPR Final RIA ES-7 November 12, 1998
-------
Exhibit ES.l
Summary of Costs under the Stage 1 DBPR ($000)
A
B
A+B
D
E
F
G
C+IHE+F+G
^m
Utility Costs
Treatment Costs
Total Capital Costs
Annual O&M
Artnualized Capital Costs
Annual Utility Treatment
Monitoring and Reporting Cost
Start-Up Costs
Annual Monitoring
Small Large Subtotal
$ 242,652 $ 554,564 $ 797,216
23,068 201,308 224,376
22,786 62,355 85,141
S 45,855 S 263,663 S 309,518
82 39 12!
10,867 14,619 25,485
Small Large Subtotal
$997,537 $528,539 $1,526,076
83,910 54,243 138,153
94,403 50,046 144,449
$178,313 $104,289 5282,602
946 36 982
.38,803 26,326 65,129
State Costs
Start-Up Costs
Annual Monitoring
Total Annual Costs at 7 Percent Cost of Capital
1 Total Annual Costs at 3 Percent Cost of Capital
^^^^^^B Total Annual Costs at 10 Percent Cost of Capital
Total
S 2,323,292
362,530
229,590
$ 592,120
1,103
90,615
4,099
13,243
$701,180
$ 626,484
$755,773
Stage 1 DBPR Final RIA
ES-8
November 12, 1998
-------
Exhibit ES.2
Cumulative Distribution of Annual Average Systems Costs and Household Costs for All
•Systems
Notes .
Average annual cost per system for all surface and ground water systems is displayed in Exhibit ES.2,
Because each system will implement a differenttreatment technique and will undertake different
monitoring activities depending on its current water quality characteristics, most systems will incur
different annual costs under the Stage 1 DBPR. Additionally, while 12,988 systems will have to modify
their treatment techniques to meet the requirements of the rule, all 76,051 systems will have to perform
annual monitoring. Thus, 12,988 systems will incur both treatment and monitoring costs, and 63,063
systems will incur only monitoring costs.
Under the Stage 1 DBPR household will face increases in annual costs, since at a minimum, all systems
are required to monitor. As shown in the cumulative distribution of households affected by the rule,
however, a large number (95 percent) of households may face an estimated maximum increase in cost of
$12 per year ($1 per month). In other words, 110 million household may incur no more than a $1
increase in their monthly costs and most substantially less. Slightly more than 3 million households (3
percent) may face an increase in costs of between $12 and $60 per year ($1 to $5 per month). The highest
estimated costs is approximately $400 per year, and less than 2 percent of households may incur costs
ranging from $60 a year to $400 a year ($5 to $33 per month).
Stage 1 DBPR Final RIA ES-9 November 12, 1998
-------
Exhibit ES.2
Cumulative Distribution of Annual Average System Costs and
Household Costs for AH Systems
Annual Average Cost per System
Cumulative Percent of Systems
100%
90% .
80% .
70% -
60% -
50% -
40% -
30% -
20%
10%
0%
$- $1,000 $2,000 $3,000 $4,000 $5,000 $6,000 $7,000 $8,000 $9,000
Annual Average Cost per System ($OOOs)
Annual Average Cost per Household
•o
o
1
o
3
0>
§
Q.
|
O
inmc.
90% .
80% .
70% .
60% .
50% .
40% .
30% .
20%
10%
0% .
(£"" 10$ ftr month (99tb percmtile)
1$ per month (95tb perf*ntile)
$- $50 $100 $150 $200 $250 $300 $350 $400 $450
Annual Cost per Household
Stage 1DBPR Final RIA
ES-JO
November 12, 1998
-------
Exhibit ES.3
Detail of Cumulative Distribution of Annual Average System Costs
Notes
Exhibit ES.3 displays the cost per system for large and small surface and ground water systems. The
methodology for estimating these costs is identical to that for Exhibit ES.2.
For small systems, the highest costs may be incurred by those systems employing membrane technology.
Fifty percent of small surface water systems may incur average annual costs of less than $3,000. The
highest average annual costs is $450,000 incurred by approximately 10 systems. Fifty percent of small
ground water systems may incur average annual costs of less than $500, with the highest average annual
costs being incurred by approximately 100 systems ($181,000).
As with small systems, the highest costs in large systems may be incurred by systems using membranes
as their treatment technique. Fifty percent of large surface water systems may incur average annual costs
of less than $500. Approximately 3 systems may face costs of $7 million per year. Fifty percent of large
pound water systems may face average annual costs of approximately $10,000. Approximately 2
systems may face costs as high as $9 million.
Stage 1 DBPR Final RIA ES-11 November 12, 1998
-------
1
1
*
1
?
*•—
K>
f
N
»-••«
i
Small Surface Water Systems Small Ground Water Systems
100%
90%
1 80%
S
8, 70%
i CO
"o 60% ,
**
| 50% .
a! 40% .
» 30% .
P8
| 20% .
3
° 10%
0%
s
r
r
$50 $100 $150 $200 $250 $300 $350 $400 $450 $500
Annual Average Cost per System (SOOOs)
Large Surface Water Systems
1fKWt n| :
sr><«, --^
I «0% /
it I
" 70% If
. 70%
w
"S 60% .
I ' 50% .
£ 40% .
1 30% .
g 20% .
i
0 10% -
0% -
i
I
$- $1;000 $2,000 $3,000 $4,000 $5.000 S6.000 $7,000 $8.000 59,000
I
Annual Average Cost per System (SOOOs)
Exhibit ES.3
Cumulative Distribution of Annual Average Costs per System
-------
Exhibit ES.4
Cumulative Distribution of Annual Household Costs for All Systems
-Notes
Exhibit ES.4 displays the annual increase in cost per household for large and small surface and ground
water systems. Below each graph of the cumulative distribution for all systems in that size category, a
detail of the 90th to 100th percentiles is displayed. The methodology for estimating these costs is identical
to that for Exhibit ES.2.
Seventy percent of household served by small surface water systems may face a monthly increase of no
more than $1 per month under the Stage 1 DBPR. Ninety-nine percent of households may incur no more
than a $10 increase in monthly costs.
In large surface water systems, 98 percent of households may face an increase of no more than $1 per
month in expenses. Almost 100 percent may face an increase of no more than $10 per month.
Most households served by small ground water systems, 91 percent, may face an increase of no more
than $1 per month. Ninety-six percent may face an increase of no more than $10 per month.
Ninety-five percent of households served by large ground water systems may face no more than $ 1 of
increase monthly cost. Ninety-nine percent may face a monthly cost increase of no more than $10. These
results are summarized below.
Small surface water systems
Large surface water systems
Small ground water systems
Large ground water systems
Total
Total
Households
4,267,000
71,378,000
15,671,000
24,174,000
115,490,000
SO to $1
per month increase
71 percent
3,009,000 households
98 percent
69,870,000 households
91 percent
14,245,000 households
95 percent
22,969,000 households
95 percent
11 0,093,000 households
SI to SI 0
per month increase
28 percent
1,204,000 households
2 percent
1,489,000 households
5 percent
755,000 households
4 percent
939,000 households
4 percent
4,382,000 households
$10 to S33
per month increase
1 percent
54,000 households
0.03 percent
20,000 households
4 percent
67 1,000 households
1 percent
266,000 households
1 percent
1,01 1,000 households
Stage I DBPR Final RIA
ES-13
November 12, 1998
-------
5*
•8
to
=§
^
•8
tS
fcfet
t
3
o-
1
100%
•8 90%
1 80%
| 70%
J 60%
o
c 50%
o 40%
93%
& 92% .
3
I 91%
O
90%
$
Small Surface Water Systems
^— ' 1 OS per month (98th percentile)
A 1$ per month (72nd percentile)
j
/
(
/
$50 $100 $150 $200 $250 $300 $350 $400 $4
Annual Cost per Household
_^-~----~--~~~~~im~~
^^
f
/
\
$50 $100 $150 $200 $250 $300 $350 $400 $4
Annual Cost per Household
50
!
50
| 100%
90%
es
5 80%
ti
o
X 60%
"5
•g 50%
S 40%
Q.
* 30%
2 20%
i 10%
' O
0%
$
100%
99%
•g 98% .
jC
8 97%
3
I 96% -
"8
«•• 95%
H 94%
Q.
« 93%
'&
£ 92%
i 91%
O
90%
$
L
f
i
i
/
/
Small Ground Water Systems
f 10$ per month (97th percentile)
S per month (91st percentile)
$50 $100 $150 $200 $250 $300 $350 $400 $4
Annual Cost per Household
/
f
^^^
^^^
$50 $100 $150 $200 $250 $300 $350 $400 $4,
Annual Cost per Household
50
50
Q
: —
3
£L
ft
I 2.
55
C
*•*.
9
O
"•*!
>.
'§ W
e x
n S
o •*
55. W
«» C/3
MM '
f&
a
sr
E-
5"
«3
3
C/3
3
en
-------
1
1
^
Large Surface Water Systems
100%
„ 90%
i TJ
1 80% .
SB
§ 70% .
o
5 60% .
o
S 50% .
01
| 40%
a.
§ 30% .
S 20% .
3
I 10%
o
0% .
„_, * _
^ 10$ per month (100th percentile)
\
IS per month (98th percentile)
$- $50 $100 $150 $200 $250 $300 $350 $400 $450
Annual Cost par Household
100% r
«.' 99%,-
Ji 98%
§ 97% .
o
J 96% .
o
c 93% .
8
« 94%
a.
s, y o /o - —
- 92% .
a
O
nn%
^
$- $20 $40 $60 $80 $100 $120 $140 $160
Annual Cost per Household
Large Ground Water Systems
100%
90%
in
1 80%
| 70%
I 60%
•s
£ 50%
| 40%
| 30%
S 20%
i 10% ,
O
0% .
$
"Q-
& -*~~ "10S per month (100th percentile)
IS per month (95th percentile)
$50 $100 $150 $200 $250. $300 $350 $400 $450
Annual Cost per Household
100%
99%
| 98% .
w 97% .
X 96% ,
•8
•g 95% .
S 94%
a.
| 93% .
I 91%"
U
90%
^—-— -~
/
\
I
^>1—-
f
$- $20 $40 $60 $80 $100 $120 $140 $160
Annual Cost per Household
Exhibit ES.4
Cumulative Distribution of Annual Costs per Household in Large Systems
-------
Exhibit ES.5
Characteristics of Water Systems
Notes
Exhibit ES.5 summarizes several key characteristics of the water systems analyzed in this R1A.
While small ground water systems are the most numerous, most people are served by large surface water
systems. Additionally, 68 percent of surface water systems will have to modify their treatment, though
only 12 percent of ground water systems will have to implement a new treatment technique.
Exhibit ES.5
Characteristics of Water Systems that Disinfect
Surface Water Systems
System Size
25-100
100-500
500-1 K
1K-3.3K
3.3K-10K
10K-25K
25K-50K
50K-75K
75K-100K
100K-500K.
500K-1M
>1M
Total
Number of
Systems
1,046
1,010
845
1,103
1,161
569
328
157
108
175
43
15
6,560
Number or Plants
1,046
1,010
845
1,103
1,161
569
328
157
216
350
86
30
6,901
Number of
Systems to Modify
Treatment
732
707
592
772
813
347
200
96
66
107
26
9
4,466
Number of
Systems to
Monitor Only
314
303
253
331
348
222
128
61
42
68
17
6
2,094
Number of
Households
21,000
88,000
265,000
926,000
2,966,000
4,361,000
5,986,000
5,043,000
5,125,000
17,246,000
18,834,000
14,783,000
75,644,000
Stage 1 DBPR Final R1A
ES-16
November 12, 1998
-------
Ground Water Systems
System Size
25-100
100-500
500- IK
1K-3.3K
3.3K-10K
10K-25K
25K-50K
50K-75K
75K-IOOK
lOOK^SOOK
500K-1M
>1M
Total
Number of
Systems
30,476
22,934
6,508
5,882
2,371
866
288
78
29
53
5
1
64,491
Number of Plants
30,476
22,934
6,508
5,882
2,371
!,299
432
156
58
159
15
3
70,293
Number of
i
Systems to Modify !
Treatment '
3,721
2,800
795
718
290
130
43
12
4
8
1
0
8,522
Number of '
Systems to
Monitor Only
26,755 ,
20,3 i 4
5,713
5,164
2,081
736
245
66
25
45
4
1
60,969
Number of
Households
623,000
2,009,000
2,043,000
4,938,000
6,058,000
6,638,000
5,256,000
2,505,000
1,376,000
5,223,000
2,191,000
986,000
39,845,000
Stage 1 DBPR Final RIA
ES-17
November 12, 1998
-------
1: Introduction
1.1 Introduction
This document analyzes the impact of the final Stage 1 Disinfectants/Disinfection Byproducts Rule
(Stage 1 DBPR). Executive Order 12866, Regulatory Planning and Review., requires EPA to estimate the
costs and benefits of the Stage 1 DBPR in a regulatory impact analysis (RIA) and to submit the analysis
in conjunction with publishing the final rule.
This RIA provides background on the rule, summarizes the key components, discusses alternatives to the
rule, and estimates costs and benefits to the public and State governments. This chapter summarizes the
technical and regulatory issues associated with the rule. It explains the nature of disinfection byproducts
(DBFs), identifies the public health concerns addressed by the rule, and summarizes the key components
of the rule.
The subsequent chapters are intended to meet the requirements of the Executive Order by responding to
specific analytical questions. Chapter 2 reviews alternative approaches considered as the rule was being
developed. Chapter 3 presents public water system data and discusses the changes utilities would have to
make as a result of this rule to establish a baseline of information for use in the following three chapters.
Chapter 4 examines the rule's potential benefits, reviewing epidemiological and toxicological data.
Chapter 5 presents an estimate of the costs to implement the rule. Chapter 6 explores different
approaches to confirm positive net benefits. The analysis concludes in Chapter 7 with an examination of
the economic rationale for regulating DBFs.
This rule is part of a larger process of improving drinking water quality through the development of a
series of related rules. Each rule is accompanied by several analyses, required either by law or executive
order. The analyses include reviewing impacts on small systems, examining unfunded mandates imposed
by this rule, and determining whether minority or low-income populations are disproportionately
affected by the requirements of this rule.
1.2 Description of the Issue
There are over 76,000 utilities (public water systems) in the United States that disinfect their water.
Utilities are supplied through either ground water sources (tapped through wells) or surface water
sources (lakes, reservoirs, and rivers). Ground water systems greatly outnumber surface water systems,
although most people are served by a small number of large surface water systems.
Since most water is not pure enough to be ingested directly from the source, utilities usually apply some
form of contaminant control. Disinfection is one important and widespread (but not universal) practice
used to meet the public health goal of providing safe water to the public. Utilities disinfect drinking
water supplies by adding chemicals to kill or inactivate microbial contaminants. Exhibit 1.1 shows a
schematic of a typical conventional treatment plant.
Stage 1 DBPR Final RIA 1-1 November 12, 1998
-------
Exhibit 1.1
Typical Treatment Plant Model (Conventional)
Range of Sites for Primary Disinfectant
Secondary
Disinfectant
Water
Source
Rapid
Mix
Flocculation Sedimentation
Backwash
Recycle
Filter
Clearwell Distribution System
Potential Changes in Treatment Processes Under the Stage 1DBPR
Disinfection remains a critical treatment process for addressing microbial contamination. Although
chemical disinfection has been used for decades, EPA's Science Advisory Board (SAB), an independent
panel established by Congress, cited drinking water contamination as one of the highest ranking
environmental risks as recently as 1990. The SAB reported that microbiological contaminants (e.g.,
bacteria, protozoa, viruses) are likely the greatest remaining risk-management challenge for drinking
water suppliers.
Disinfection, however, poses health risks of its own. Byproducts may result from chemical interactions
between DBF precursors in water and chemical disinfectants in plants and distribution systems of public
water systems. Source water often carries substantial levels of organic material that, when mixed with
disinfectants, form new compounds. Some of these byproducts, including those that are the subject of
this rule (total trihalomethanes—TTHM—and five haloacetic acids—HAAS), are potentially associated
with health risks, such as cancer and reproductive and developmental effects. However, because
disinfection is effective in reducing microbial contamination, reducing disinfection to decrease DBFs can
increase the risk to the public from microbial contamination. This is known as a "risk-risk tradeoff."
Plants can use the following changes in treatment processes to reduce the level of DBFs:
> Choice of disinfectants;
* Sequence of disinfectant application;
*• Timing and duration of disinfectant application; and,
*• Choice of equipment.
The analysis of this risk-risk tradeoff is a continuing process. The inconclusiveness of past scientific
research has made the development of regulations difficult. New research concerning water quality
standards and DBFs will continue to improve the quantification of health risks. Recent research results
concerning the health risks associated with DBFs (discussed further in Section 2.2) supports the
development of the Stage 1 Disinfectants/Disinfection Byproducts Rule. Previous regulatory efforts
concerning this problem are outlined in the following section.
Stage ] DBPR Final RIA
1-2
November 12, 1998
-------
13 Regulator)'History
The primary responsibility for regulating the quality of drinking water lies with EPA. The Safe Drinking
Water Act (SDWA) establishes this responsibility and defines the mechanisms at the Agency's disposal
to protect public health. EPA sets water quality standards by identifying which contaminants should be
regulated and establishes which levels of the contaminant are to be attained by utilities. -
To regulate a contaminant, EPA first establishes a maximum contaminant level goal (MCLG)5 which
establishes the contaminant level at which no known or anticipated adverse health effects occur. MCLGs
are unenforceable health goals.,EPA then sets an enforceable maximum contaminant level (MCL) as
close as technologically possible to the MCLG. If it is not feasible to measure the contaminant, a
treatment technique is specified.
Additionally, EPA identifies maximum concentrations of disinfectant residual concentrations and sets
maximum residual disinfectant level goals (MRDLGs) and maximum residual disinfect levels (MRDLs).
Residual levels of disinfectants are maintained in the distribution system following treatment in order to
assure microbial safety all the way to the customer's tap. Like MCLGs and MGLs, MRDLGs are
unenforceable while MRDLs are enforceable.
For utilities, compliance with a regulation means not exceeding the MCL. However, when MCLs are not
economically or technologically feasible, an approved treatment technique can be used. A treatment
technique requirement is a regulatory approach that specifies a technology that reduces exposure to
contaminants to the extent feasible.
Several drinking water regulations predate the current regulatory effort, including rules controlling levels
of trihalomethanes, total coliform, and microbial pathogens. The first of these, the 1979 Total
Trihalomethane (TTHM) Rule, set an interim MCL of 0.10 mg/L (100 yug/L), based on an annual
average. Trihalomethanes have long been recognized as potential carcinogens. The 1979 TTHM Rule
applies only to utilities (using either ground or surface water) serving 10,000 or more people that
disinfect their water. The Total Coliform Rule (1989) applies to all utilities. It regulates the levels of
coliform bacteria permissible in drinking water systems. Coliform bacteria also serve as indicators for
other microbes that may be pathogenic. In 1989, EPA also promulgated the Surface Water Treatment
Rule (SWTR), the primary control for microbial pathogens in surface water. The SWTR established
treatment technique requirements for Giardia lamblia, viruses, and Legionella, and included regulations
for all utilities using surface waters (or ground water sources under the direct influence of surface water).
To address the complex issues associated with regulating DBPs, EPA launched a rule-making process in
1992 and convened a regulatory negotiation advisory committee (RegNeg) under the Federal Advisory
Committee Act (FACA), representing a range of stakeholders affected by possible regulation. The
RegNeg Committee met repeatedly over a period of 10 months and arrived at a consensus proposal for
taking progressive steps toward addressing both DBPs and microbial pathogens. The 1992 consensus-
building process resulted in the three following regulatory proposals—
1) A staged approach to regulation of disinfectants and DBPs (referred to as the Stage 1 and Stage 2
DBPRs) incorporating MCLs, MRDLs, and treatment technique requirements;
Stage 1 DBPR Final RIA 1-3 November 12, 1998
-------
2) A companion Interim Enhanced Surface Water Treatment Rule (IESWTR) designed to improve
control of microbial pathogens and prevent inadvertent reductions in microbial safety as a result
of DBF control efforts, and;
3) An Information Collection Rule (ICR) to collect information necessary to reduce many key
uncertainties prior to subsequent negotiations regarding the Stage 2 DBPR requirements.
In 1997, a similar FACA process was implemented with the Microbial-Disinfectants/Disinfection
Byproducts (M-DBP) Advisory Committee. The M-DBP Committee convened to analyze new data
available since 1994, review previous assumptions made during the RegNeg process, and move the rule
forward on the expedited schedule mandated under the 1996 Amendments to the SDWA. The efforts of
this committee resulted in the drafting of the Stage 1 DBPR.
1.4 Public Health Concerns to be Addressed
EPA's main mission is the protection of human health and the environment. When carrying out this
mission, EPA must often make regulatory decisions with less than complete information and with
uncertainties in the available information. When making regulatory decisions, EPA believes it is
appropriate and prudent to act to protect public health when there are indications that exposure to a
contaminant could present a risk to public health, rather than take no action until risks are unequivocally
proven.
In regard to the Stage 1 DBPR, EPA recognizes that the assessment of public health risks from
disinfection of drinking water currently relies on inherently difficult and preliminary empirical analysis.
On one hand, epidemiologic studies of the general population are hampered by difficulties of design,
scope, and sensitivity. On the other hand, uncertainty is involved in using the results of high-dose animal
toxicological studies of a few of the numerous byproducts that occur in disinfected drinking water to
estimate the risk to humans from chronic exposure to low doses of these and other byproducts. In
addition, such studies of individual byproducts cannot characterize the entire mixture of disinfection
byproducts in drinking water. While recognizing these uncertainties, EPA continues to believe, for the
reasons cited below, that the Stage 1 DBPR is needed for protection of public health from exposure to
DBFs.
A fundamental component in risk assessment is the number of people that may be exposed to a particular
parameter of concern. In this case, there is a very large population potentially exposed to DBFs via
drinking water in the U.S. Over 200 million people in the United States are served by public water
systems that apply a disinfectant (e.g., chlorine) to water in order to provide protection against microbial
contaminants. While these disinfectants are effective in controlling many harmful microorganisms, they
combine with organic matter in the water and form DBFs, some of which may pose health risks. One of
the most complex questions facing water supply professionals is how to minimize the risks from these
DBFs and still control microbial contaminants. Because of the large number of people exposed to DBFs,
there is a substantial concern for any risks that may be associated with DBFs.
Numerous toxicological studies have been conducted with regard to the public health endpoint or
symptoms of concern have shown several DBFs to be carcinogenic in laboratory animals (such as
bromodichloromethane, bromoform, chloroform, dichloroacetic acid, bromate, and MX). Some DBFs
have also been shown to cause reproductive or developmental effects in laboratory animals (such as
Stage 1 DBPR Final RIA 1-4 November 12, 1998
-------
chlorite and certain haloacetic acids). While many of these studies have been conducted at high doses,
EPA believes the studies provide evidence that DBFs present a potential public health problem that needs
to be addressed.
In the area of epidemiology a number of additional studies have also been completed investigating the
relationship between exposure to chlorinated drinking water and cancer. These studies have suggested an
association, albeit uncertain, between bladder, rectal, and colon cancer and exposure to chlorinated
drinking water. Several epidemiology studies have also been completed evaluating the association
between exposure to chlorinated drinking water and reproductive and developmental effects. While there
are fewer of these studies than for cancer, more recent, better designed studies have suggested an
association between early term miscarriage and neural tube defects and exposure to drinking water with
elevated trihalomethane levels.
While EPA recognizes there are data deficiencies in the information on the health effects from DBFs and
the levels at which adverse health effects occur, EPA believes the weight-of-evidence represented by the
available epidemiological and toxicological studies on DBFs and chlorinated surface water support a
potential hazard concern and warrant regulatory action at this time. Because of this deficiency, EPA
believes the incremental two-stage approach agreed upon during the RegNeg process is prudent and
necessary to protect public health and meet the requirements of the 1996 SDWA.
In conclusion, because of the large number of people exposed to DBPs and because of the different risks
that may result from exposure to DBPs, EPA believes the Stage 1 DBPR is needed to further prevent
potential health effects from DBPs (beyond that controlled for by the 1979 TTHM Rule). This is in
agreement with the recommendations of the RegNeg for the 1994 proposed rule and the M-DBP
Advisory Committee, that while additional information is needed for the Stage 2 DBPR, especially on
health effects, the Stage 1 DBPR is currently necessary to reduce risks from DBPs.
1,5 Summary of the Rule
The Stage 1 DBPR uses a combination of new MCLs, MRDLs, and a treatment technique requirement to
improve control of disinfectants and DBPs. The rule applies to all utilities defined as community or non-
transient/non-community systems that treat their water with a chemical disinfectant. (Community
systems are public water systems that regularly serve at least 25 year-round residents; non-transient/non-
community systems generally include businesses and other fixed establishments, such as schools in
remote areas.) The 1ESWTR, promulgated concurrently with the Stage 1 DBPR, will further control for
microbial contamination and prevent increases in microbial risk. These rules were developed in tandem
since microbial contamination and disinfection are directly related. Both rules will be promulgated in
November 1998.
In the Stage 1 DBPR, EPA establishes MCLGs and MCLs for previously unregulated byproducts (except
in the case of TTHMs). EPA is setting MCLGs of 0 for chloroform, bromodiehloromethane, bromoform,
bromate, and dichloroacteic acid, and MCLGs of 0.06 mg/L for dibromochloromethane, 0.3 mg/L for
trichloracetic acid, and 0.8 mg/L for chlorite. In addition, EPA is setting MRDLGs for chlorine and
chloramines at 4.0 mg/L and 0.8 mg/L for chlorine dioxide.
The Stage 1 DBPR sets a new, more restrictive MCL for TTHMs at 0.08 mg/L (80 /ug/L). EPA is adding
MCLs for HAAS of 0.06 mg/L (60 /ug/L), for bromate of 0.01 mg/L, and for chlorite of 1.0 mg/L. In
Stage 1 DBPR Final RIA 1-5 November 12, 1998
-------
addition to these byproduct MCLs, EPA is setting MRDLs for chlorine and chloramines of 4.0 mg/L and
0.8 mg/L for chlorine dioxide.
EPA identifies several technologies that utilities can use to meet the MCLs and MRDLs. These include
using alternative disinfectants, such as ozone, or alternative treatment practices, such as enhanced
coagulation/enhanced softening or membrane filters.
Enhanced coagulation (or enhanced softening in systems that soften their water) is specified as a
treatment technique for systems with source water qualities that exceed certain parameters, (e.g., Total
Organic Carbon—TOC—is above 2.0 mg/L) unless certain exception criteria are met. Enhanced
coagulation is when systems increase their use of a coagulant, such as alum, to improve the removal of
precursors and compounds that react with disinfectants to form DBFs. Precursors are generally identified
as total organic carbon (TOC) and bromide. One purpose of the enhanced coagulation and softening
requirements is to control for DBPs not controlled through compliance with the MCLs. Another purpose
is to decrease the reliance on alternative disinfection practices to help comply with the MCLs for DBPs
and MRDLs for disinfectants.
1.6 Environmental Justice
National drinking water regulations apply uniformly to utilities, and although not all have to modify
treatment or operations to reach a particular standard, all must comply with the water quality standards as
promulgated. Thus, the level of protection is consistent across all populations served by utilities.
Traditionally developed environmental justice analyses are, therefore, not appropriate in this case.
One indicator that the concerns and issues of affected communities, including sensitive populations, are
included in the Stage 1 DBPR was the undertaking of the RegNeg and M-DBP processes to craft the
regulation. Both committees were chartered under the FACA and included a broad cross-section of
regulators, the regulated communities, industry, and consumers. Extensive discussion on setting levels
that provided the maximum protection feasible took place, and the final consensus on recommendations
to EPA for the Stage 1 DBPR considered issues of affordability, equity, and safety,
1.7 Unfunded Mandates Reform Act Analysis
Title II of the Unfunded Mandates Reform Act (UMRA) of 1995, P.L. 104-4, establishes requirements
for Federal agencies to assess the effects of their regulatory actions on State, local, and tribal
governments and the private sector. Under UMRA section 202, EPA must prepare a written statement
including a benefit/cost analysis, for proposed and final rules with Federal mandates that may result in
expenditures to State, local, and tribal governments, in the aggregate, or to the private sector, of $ 100
million or more in any one year.
Because EPA believes that this rule may result in expenditures of $100 million or more for State, local,
and tribal governments, in the aggregate, or the private sector, in any one year, it has prepared Unfunded
Mandates Reform Act Analysis for the Stage 1 Disinfectants/Disinfection Byproducts Rule to accompany
this RIA. This document reviews the benefit/cost analysis, estimates potential disproportionate budgetary
effects, and summarizes State, local, and tribal government input. The analysis identifies the selected
regulatory options as the least costly, most cost-effective, and least burdensome that accomplish the
objectives of the Stage 1 DBPR.
Stage 1 DBPR Final RIA 1-6 November 12, 1998
-------
1.8 Regulatory Flexibility Analysis
The Regulatory Flexibility Act provides that if a rule has a significant impact on a substantial number of
small entities, its proposal must be accompanied by a Regulatory Flexibility Analysis (RFA) to be made
available for public comment. Under current policy, EPA regards any impact as a significant impact and
any number of small entities as a substantial number. Thus, a Regulatory Flexibility Analysis is clearly
required for the Stage 1 DBPR. The Regulatory Flexibility Analysis can be incorporated within other
analyses—as is the case here—so long as it is clearly stated how the requirements are being met.
Both Advisory Committees sought to provide quantitative characterization of small system impacts
throughout the RegNeg process. The RegNeg and M-DBP Committees evaluated regulatory alternatives
that span the complete range of considerations required by Agency guidance for implementation of the
Regulatory Flexibility Act, encompassing extended timetables, performance versus design standards,
exemption-based alternatives, and relaxed standards for small entities. The discussion in Chapter 5.6,
Small System Impacts—Regulatory Flexibility Analysis, summarizes the small system impact analysis,
regulatory alternatives relevant to small systems, and impact mitigation measures considered in the
RegNeg process.
Stage ! DBPR Final RIA 1-7 November 72, 1998
-------
2: Consideration of Regulatory Alternatives
2.1 Chronological Review of Regulatory Options Considered
2.1.1 Alternative Development Process
The central requirement of regulatory analyses under Executive Order 12866 is to perform an analysis of
net benefits and to consider the regulatory alternatives in light of a criterion of maximizing net benefits.
This chapter discusses the regulatory alternatives considered.
The 1994 Disinfectants/Disinfection Byproducts Rule (DBPR) proposal attempted to balance the control
of health risks from compounds formed during drinking water disinfection against the risks from
microbial organisms to be controlled by the Interim Enhanced Surface Water Treatment Rule (IESWTR).
The 1997 modifications sought the same balance but were enhanced by new data and the 1997 Microbial
Disinfectants/Disinfection Byproducts (M-DBP) Advisory Committee process. Although in many
aspects the 1994 proposal and the 1997 rule are similar, important differences exist.
Regulatory impact analysis (RIA) was a major focal point of the RegNeg and M-DBP Technologies
Working Groups (TWGs). The TWGs (involving stakeholder representatives) developed consensus
analyses of the impact of regulatory alternatives throughout the negotiating process. Representatives of
the TWGs typically presented regulatory impact analysis briefings at the beginning of negotiating
committee meetings. As the consensus process progressed, the TWG impact analyses progressed through
a series of alternatives proposed and modified by the negotiating committee.
The impact analyses developed by the TWGs covered all of the major regulatory alternatives considered.
Impacts were characterized in terms of both cost and effects on public health. Because health impacts are
less quantifiable, the analysis tended toward a cost-effectiveness framework rather than a cost-benefit
framework. The scope of analysis performed by the TWG nonetheless encompassed all of the general
substance of regulatory impact analysis required under EO 12866 and current EPA guidelines.
In general, the TWGs provided evaluations of the specific regulatory alternatives. Analysts prepared the
cost estimates based on agreed upon assumptions and provided the estimates to the TWGs and
Committees for review and feedback. Often, the cost estimates provoked discussion and debate, with the
TWGs and Committees asking for further research and refinements of the estimates before reaching a
consensus on the proposed regulation.
At each phase of the process, the Committees reviewed the findings and analyses of the TWG and further
refined the proposal. As a result, a variety of alternatives were discussed and costed. A chronological
review of these alternatives provides an understanding of the goals and direction of the EPA proposal.
The RegNeg Committee and TWG focused initially on surface water systems that filter, but do not
soften. This was selected primarily because it is perhaps the most relevant category to choose for detailed
study of regulatory alternatives to control DBFs. There are approximately 5,600 water systems in this
category, serving more than 130 million people. While this category represents only about 8 percent of
all community and non-transient non-community water systems, at the time of proposal this category
~Stage 1 DBPR Final RIA '' ~1 ; '' November 12, 1998
-------
was projected to incur about 50 percent of total capital costs. In addition, this category represents about
80 percent of the total population served by surface water systems.
The preliminary analysis focused on two options, Option 1 and Option A. Option 1 proposed a total
trihalomethane (TTHM) maximum contaminant level (MCL) of 80 and a total haloacetic acid (HAA)
MCL of 60 for large water systems (and a simple TTHM standard of 100 for small systems). Option A
called for the use of precursor removal technology to reduce the level of total organic carbon (TOC).
Alternative levels of TOC were considered, ranging from 4,0 to 0.5. The presumption behind Option A
was that DBF MCLs would be established in a manner that would be consistent with meeting the TOC
target; i.e., the TOC target would be the driving force and would drive compliance towards precursor
removal technology. Potential adverse health risks associated with alternative disinfectants would thus be
avoided.
After additional analysis by the RegNeg Committee, two additional options, or hybrids (Option A), were
added to the mix: the 80/60/4 and 80/60/5 options represented an attempt to merge concepts of TOC
removal and MCLs of 80 for TTHM and 60 for five HAAs (HAAS). These also represented the first
detailed considerations of a staged approach to DBF regulation.
Option 1 (100/80/60) and the two hybrids under Option A (80/60/4 and 80/60/5) were carried forward
after a review of the reductions in exposure and a comparison of national costs arising from the options.
Option 1 would have required treatment changes in 45 percent of plants, whereas the 80/60/4 option
would have required changes in 56 percent of plants and the 80/60/5 option would require changes in 43
percent of plants.
National cost estimates developed at the time indicated that total capital cost of the three Stage 1 options
ranged from $3.7 billion for Option 1 to $8 to $9 billion for Option A. The small (serving populations of
less than 10,000) systems' share of the national capital cost of the Stage 1 options ranged from $0.8
billion for Option 1 to $3.1 to $3.2 billion for Option A (i.e., the 80/60/4 suboption). Reduced exposure
to TOC was considerable under Option A hybrids. Option 1, however, does not reduce TOC levels. The
major cost difference between Option 1 and the Option A hybrids stems from the requirement to reduce
TOC.
The discussion on alternatives began to center on the evaluation of the two-stage approach to DBF
regulation, according to consensus reached by the RegNeg Committee. The Stage 1 proposal represented
a compromise between the 100/80/60 (Option 1) concept and the 80/60/4 (Option A hybrids) concept. A
treatment technique requirement for "enhanced coagulation" would apply to all systems with effluent
TOC above 2.0 mg/L to reduce overall TOCs, and this would be coupled with Stage 1 MCLs of 80 and
60, for TTHMs and HAAS, respectively.
The total cost and compliance forecasts presented to the RegNeg Committee at the end of the process
included $4.4 billion for Stage 1 and $10.5 to $11.2 billion for a possible Stage 2. At the time, total
annualized costs were estimated at $1.1 billion for Stage 1 and $2.43 to $2.6 billion for the Stage 2
option (i.e., the cumulative costs from Stage 1 through Stage 2). These costs and compliance forecasts
served as the starting point for the M-DBP Committee three years later.
The M-DBP Committee convened in 1997 to review assumptions and new data and to lay the
groundwork for the promulgation of the final rule in November of 1998. As described in Chapter 5, costs
Stage I DBPR Final R1A 2-2 November 12, 1998
-------
were modified based on new unit cost estimates and revised assumptions about the compliance treatment
forecast.
One topic of discussion of the M-DBP Committee was a change to the enhanced coagulation model. The
rule requires systems treating surface water (or ground water under the direct influence of surface water)
and using conventional treatment or precipative softening to remove DBF precursors by enhanced
coagulation or enhanced softening. The removal of TOC is to be used as a performance indicator for
DBF precursor removal. Removal targets for subject systems are described in the rule by a matrix of
influent raw TOC and alkalinity levels (the "3-X-3 matrix," Exhibits 4.5a, 4.5b, 4.5c).
Extensive research on key elements of the proposed enhanced coagulation requirements in recent years
led the M-DBP Committee to recommend additional exceptions, the primary on being if utilities had
raw-water Specific Ultraviolet Light Absorbence (SUVA—an indicator of the humic content of the
water) of equal or less than 2.0 L/mg-m. This exception, among others, was intended to limit enhanced
coagulation requirements to only those waters where DBF precursors would be effecting removal, and
thereby also to limit costs for the utilities and their primary agencies.
Based on recent research on enhanced softening (removal of certain levels of TOC or of 10 mg/L
magnesium hardness), the M-DBP Committee recommended changes to the 1994 proposal. In particular,
proposed TOC removals were modified at systems with high alkalinity, and lime softening plants would
not be required to perform lime soda softening or to lower alkalinity below 40 mg/L.
Additionally, the M-DBP Committee reviewed the significance of predisinfection on treatment. The
Stage 1 DBPR as proposed originally would not have allowed utilities required to use enhanced
coagulation or enhanced softening to take credit for compliance with disinfection requirements in the
1989 Surface Water Treatment Rule (SWTR) or the IESWTR prior to removing required levels of
precursors, unless they met specified criteria. Analysis by the M-DBP Committee indicated that most
utilities using enhanced coagulation, as required by the treatment technique provision, would be able to
meet the MCLs for TTHM and HAAS while maintaining their existing disinfection practice. This
analysis also indicated that significant precursor removal and DBF reduction could still be achieved with
predisinfection left in place. Also, systems would incur large capital costs to remain in compliance with
disinfection requirements if predisinfection credits were disallowed. The Committee, therefore,
recommended that EPA continue to allow credit for compliance with applicable disinfection
requirements for disinfectants applied at any point prior to the first customer, consistent with the existing
provisions of the 1989 SWTR.
2.2 Options with Complete Cost and Benefit Analyses
National compliance costs and projected benefits were estimated for all elements of the final rule with
cost implications. These cost and benefit projections follow in Chapters 4, 5, and 6.
Stage 1 DBPR Final R1A 2-3 November 12, 1998
-------
3: Baseline Conditions
3.1 Introduction
To develop forecasts, of the economic and financial impacts of regulatory alternatives on the water
supply industry, and ultimately on customers, for the Stage 1 Disinfectants/Disinfection Byproducts Rule
(Stage 1 DBPR), it is necessary to develop a baseline—a characterization of the industry and its
operations—before considering the effect of any regulatory option. This chapter reviews this baseline in
three sections:
> Industry Profile-—describing the water supply industry that is subject to the rule;
»• Influent Water Quality Characterization—describing the quality of the water the industry has to
work with; and,
» Existing Treatment Characterization—describing what the industry currently does with the
water.
The baseline is not an encyclopedic review of the industry, source waters, and practices. Instead, the
baseline is at a level of detail and precision appropriate to the needs of subsequent analyses and the
decisions that were under consideration. Characteristics that were important to the decisions being made
were given careful treatment; those distinctions that were unlikely to result in significant differences or
affect decisions about the proposed rule were considered, though not in great detail.
This baseline derives from analyses that accompanied the 1994 Stage I DBPR package and considers
new data, where available. The process involved knowledgeable stakeholders and incorporated the latest
research; therefore, the data used in the analyses is accepted as the best available. Further, although new
data sources will eventually permit a more refined understanding of the industry, differences that would
significantly affect the results of this regulatory impact analysis are not anticipated.
EPA is presently developing a standardized set of baseline information for use in regulatory impact
analyses under a separate effort. However, for purposes of the final Stage 1 DBPR and Interim Enhanced
Surface Water Treatment Rule (IESWTR) packages, these estimates were not yet available to the
Microbial-Disinfectants/Disinfection Byproducts (M-DBP) Committee (and are not yet available).
Therefore, the baseline does not draw on this project.
3.2 Industry Profile
Data on utilities and their capacity to achieve treatment levels were analyzed to develop the national
compliance cost model. Data inputs include the total number of systems to which the provisions would
apply, households and populations served by these systems, average and maximum system flow rates,
and applicable costs of capital, labor, and operations and maintenance. Utilities are characterized by
whether or not they are able to achieve compliance with the rule and, if not, which practices they will
need to modify in order to comply.
Stage 1 DBPR Final JUA 3-1 November 12, 1998
-------
3.2.1 Total Number of Systems and System Size
The two features that most distinguish water suppliers are the source of their water (ground or surface)
and size of their systems (as measured by the number of people served). In general, there are about 10
ground water systems for every surface water system, and many more small systems than large systems.
Ground water systems primarily serve populations fewer than 10,000 people. For example, about three-
quarters of surface water systems serve populations fewer than 10,000, whereas over three-quarters of
ground water systems serve fewer than 500 people. These characteristics are summarized in Exhibit 3.1.
A total of 6,561 surface water systems and 69,491 groundwater systems are estimated to be affected by
this rule. The total of 76,052 systems are a mixture of publicly and privately operated systems. Analysis
of the Federal water system database, the Safe Drinking Water Information System (SDWIS), in 1991 as
well as a number of other data sets, established the base number of systems for the regulatory impact
analysis. Since the number of small systems is decreasing due to consolidation with larger systems, data
on water systems changes frequently, and is difficult to establish with specificity at any time.
Exhibit 3.1
Number of Systems that Disinfect by Source and Size
System Size
(population served)
25-100
100-500
500-1,000
1,000-3,300
3,300-10,000
10,000-25,000
25,000-50,000
50,000-75,000
75,000-100,000
100,000-500,000
500,000-1,000,000
1,000,000 or more
Total
Number
Surface Water
1,046
1,010
845
1,103
1,161
569
328
157
108
175
43
15
6,560
of Systems
Ground Water
30,476
22,934
6,508
5,882
2,371
866
288
78
29
53
5
1
69,491
Source: 1994 Stage 1 DBPR RlA
Larger systems, obviously, serve more households and deliver more water (Exhibit 3.2). The largest 292
systems (fewer than one half of one percent of all systems) serve fully half of the households in the
country. Because the variability in system size is so important for cost and operational considerations,
the baseline includes 12 size categories.
Stage 1 DBPR Final RlA
3-2
November 12, 1998
-------
3.2.2 Average System Flow Rates
Average system flow rates are integrated into the national compliance cost model in determining
household costs. Average and maximum system flows, expressed in millions of gallons per day (MOD),
were developed separately from the cost model, but are key components in generating unit costs (EPA,
June 24, 1998). The 1996 Water Industry Database (WIDE) contains a higher value for the largest
(1,000,000 people or more) system size category (127,8 million gallons per year versus 98.6 million
gallons per year) than the data sources used for the bulk of the cost estimation in this analysis. Cost
summaries presented in Chapter 5 reflect the lower flow rate. For purposes of comparison, the higher
flow rate is used to calculate costs at the 7 percent cost of capital and is displayed at the end of Appendix
C.
Exhibit3.2
Number of Households
System Size
(Population Served)
25-100
100-500
500-1,000
1,000-3,300
3,300-10,000
10,000-25,000
25,000-50,000
50,000-75,000
75,000-100,000
100,000-500,000
500,000-1,000,000
1 ,000,000 or more
Total
Number of Systems
31,522
23,944
7,354
6,985
3,532
1,435
616
235
137
228
48
16
76,052
Average Flow/System
(000 gallons/year)
2,044
8,760
31,390
83,950
255,500
766,500
1,825,000
3,212,000
4,745,000
9,855,000
43,800,000
98,550,000
Number of Households
644,000
2,097,000
2,308,000
5,864,000
9,024,000
10,999,000
11,242,000
7,548,000
6,501,000
22,469,000
21,024,000
15,769,000
115,489,000
Note: Detail may not sum due to independent rounding.
3.2.3 Cost of Capital
A cost of capital rate of 7 percent was used to calculate the unit costs for the national compliance cost
model. This rate represents the standard discount rate preferred by the Office of Management and Budget
(OMB) for benefit/cost analyses of government programs and regulations.
In addition to the 7 percent rate, unit costs were generated using both a 10 percent and 3 percent rate and
evaluated using the national cost model. The 10 percent cost of capital rate provides a link to the 1994
Stage 1 DBPR cost analyses and is assumed to be a reasonable estimate of the cost to utilities to finance
capital purchases that may be called for under the rule.
Stage 1 DBPR Final RIA
3-3
November 12, 1998
-------
The exhibits of cost estimates presented in Chapter 5 reflect the 7 percent rate. The 10 and 3 percent rates
are presented in the cost summary exhibit (Exhibit 5.5) for purposes of comparison. Costs presented in
the analysis are expressed in 1998 constant dollars.
3.2.4 Unit Costs
Unit cost estimates are an integral part of the calculation of national compliance costs for the Stage 1
DBPR. They are an estimation of the marginal cost of complying with the rule based on a dollar amount
per 1,000 gallons of water. Both capital and operating and maintenance costs for each treatment option
have been estimated (EPA, June 24, 1998). These costs were estimated by engineers and economists .
familiar with equipment, process, and labor costs using available data and expert judgment. Unit costs
were calculated at 3, 7, and 10 percent costs of capital. Unit costs estimates are included in Appendices B
through D. For detail of the assumptions on deriving the unit costs for this R1A, refer to the July 1998
Cost and Technologies document.
3.2.5 Costs of Labor
Labor rates in the national compliance cost model are used primarily to estimate costs to utilities and
States for DBF monitoring and reporting. A labor load rate, representing fringe payments, indirect costs,
and general and administrative costs, was multiplied by the direct labor rate. This rate was originally
estimated at 15 0 percent of the direct labor rate (1.5 load), but current Department of Labor statistics
indicate that a lower, 140 percent, rate (1.4 load) is more accurate. The 1.4 load rate was used in the final
calculations.
3.3 Influent Water Quality
Part of the regulatory baseline is the nature of the source waters that the industry uses. The quality of the
source waters, and perhaps more important, the varisition in the source waters, determines what is needed
and practical to consider as treatment alternatives. An encyclopedic review of characteristics is not
needed, but just those parameters key to subsequent analyses.
The rule requires surface water systems using conventional treatment processes (and ground water
systems under the direct influence of surface water and ground water systems that disinfect) to modify
these processes if current DBP formation exceeds the Maximum Contaminant Levels (MCLs) established
in the rule, as well as implement enhanced coagulation if influent organic content exceeds a threshold.
The 1996 Water Industry Database (WIDB) served as the source of influent water quality
characterization, including system process information, influent total organic carbon (TOC), alkalinity,
effluent TOC, and DBFs.
The enhanced coagulation treatment technique uses system TOC data to determine whether and how
systems must comply with the technique. TOC removal, an indicator of the effectiveness of enhanced
coagulation, is measured as the difference between influent TOC and effluent TOC as a percentage of
total influent TOC. There are two parameters of influent water quality that drive the analysis—TOC and
alkalinity. TOC removal targets are based primarily on the perceived feasibility of the treatment
technology to consistently remove TOC without prohibitive cost or level of effort. These measures are
used to categorize systems into levels of the needed percent removal of TOC. Available data on these
parameters derive from surveys of large surface water systems, but the data are presumed to be
representative of all the surface water systems affected by the rule. The reduction of TOC is used as one
Stage 1 DBPR Final RIA 3-4 November 12, 1998
-------
measure for eventual byproduct reduction and so is important for the analysis of benefits, as well as
establishing the degree of needed treatment.
Systems are not required to comply with the enhanced coagulation technique if influent TOC values are
below 1.7 mg/L (2.0 ing/L with an assumed 15 percent buffer). Above 1.7 mg/L, measurements of
influent TOC concentrations in the 221 systems used as a baseline extend to as high as 26 mg/L,
although few samples had levels above 6.8 mg/L (Exhibit 3.3). Alkalinity has a less concentrated
distribution (Exhibit 3.4).
Stage 1 DBPR Final RJA 3-5 November 12, 1998
-------
Exhibit 3.3
Cumulative Distribution of TOC Concentration in Source Waters
(Based on Data for Large Surface Water Systems)
TOC (mg/L)
Exhibit 3.4
Cumulative Distribution of Alkalinity in Source Waters
(Based on Data for Large Surface Water Systems)
100 150 200
Alkalinity (mg/L)
Stage 1 DBPR Final RIA
3-6
November 12, 1998
-------
The percent removal targets of TOC required under the Stage 1 DBPR are presented in Exhibit 3.5a.
These targets are based on the feasibility of TOC removal given the removal technology within the
specified TOC and alkalinity parameters (mg/L). Subsequent analyses categorize systems into nine
groups in the form of a "3-X-3 matrix" (Exhibit 3.5b and 3.5c) using the relationship of systems' TOC
and alkalinity. The systems are not evenly divided among the nine groupings; most systems are in the
lower range of both TOC and alkalinity. The seatterplot of TOC and alkalinity for each of the systems
shows the variability across both parameters and, for reference, shows the divisions used in subsequent
analyses (Exhibit 3.6). Those systems not currently meeting the TOC removal target are presented in
Exhibit 3,7. The cumulative distributions for each cell of the 3-X-3 matrix are presented in Exhibit 3.8.
These distributions show the removal targets for TOC, the percent reduction in TOC removal for the
systems in the specific cell, and the systems meeting and not meeting the TOC removal target.
Exhibit 3.5a
Percentage of TOC Removal Required under the Stage 1 DBPR*
If pi
> 1.7 to < 3.4
3.4 to 6.8
> 6.8
^••••1
o
35%
45%
50%
ant alkalinity (mg/L)
25%
35%
40%
is...
15%
25%
30%
...then plants must remove this percentage of TOC**
* Removal targets are based on the feasibility of TOC removal in each category.
** Percent TOC Removal = [(Influent TOC - Effluent TOC)/Influent TOC]
Stage 1 DBPR Final RIA
3-7
November 72, 1998
-------
Exhibit 3.5b
Systems within TOC and Alkalinity Parameters as a Percentage of Total Systems
Tots
22
Influent TOC (mg/L)
si Systems:
1 (100%)
> 1.7 to < 3.4
3.4 to 6.8
>6.8
Alkalinity (mg/L)
<60
0
Systems: 3 1
Percent of Total: 14.0%
Systems: 33
Percent of Total: 14.9%
Systems: 2
Percent of Total: 0.9%
60 to 120
Systems: 46
Percent of Total: 20,8%
Systems: 43
Percent of Total: 19.5%
Systems: 5
Percent of Total: 2.3%
> 120
Systems: 22
Percent of Total: 10.0%
Systems: 29
Percent of Total: 13.1%
Systems: 10
Percent of Total: 4.5%
Source: Calculated from 1996 WIDB Data
Exhibit 3,5c
Systems Meeting and Not Meeting Removal Targets
within TOC and Alkalinity Parameters
Total Systems:
221
Influent TOC (mg/L)
> 1.7 to < 3.4
3.4 to 6.8
>6.8
Alkalinity (mg/L)
<60
Systems: 3 1
meeting target: 14
not meeting target: 17
Systems: 33
meeting target: 1 8
not meeting target: 15
Systems: 2
meeting target: 2
not meeting target: 0
60 to 120
©
Systems: 46
meeting target: 20
not meeting target: 26
Systems: 43
meeting target: 9
not meeting target: 34
Systems: 5
meeting target: 2
not meeting target: 3
>120
Systems: 22
meeting target: 1 0
not meeting target: 1 2
Systems: 29
meeting target: 1 7
not meeting target: 12
Systems: 10
meeting target: 10
not meeting target: 0
Source: 1996 WIDB Data
Stage 1 DBPR Final RIA
3-8
November 12, 1998
-------
OQ
Exhibit 3.6
Systems Meeting and Not Meeting TOC Removal Targets (Based on Data for Large Surface Water Systems)
(Does not include systems with TOC < 1.7 mg/L)
1
1
1
Uj
November
N
1
0
1.7
3.4
5.1
6.8
8.5
10.2
£ 11.9
•£ 13.6
= 15.3
17
18.7
20.4
22.1
23.8
25.5
27.2
60
Alkalinity
120 180
240
300
»» t A «» A A
*'-<«"* "
•" . .
•
•
rf*"*** ^ **
* * * ^ *
» ^ * «*A
** •
•
» *
^
#
•
»Sy
ASy.
A A
I«A
•
•
stems Meeting TOC F
stems Not Meeting TC
A »
* *
(em ova I
)C Removal
-------
I
I
r
Ki
Exhibits.?
Systems Not Meeting TOC Removal Targets (Based on Data for Large Surface Water Systems
(Does not include systems with TOC < 1.7 mg/L)
Alkalinity
60
120
180
240
300
O
o
3
tf
0
1.7
3.4
5.1
6.8
8.5
10.2
11.9 .
13,6 .
15.3 .
17
18.7 .
204
22.1 .
238 -
25.5 .
27.2 -
4 A A A
A* *4** *
*AA
*i A
A * **
***** ?*
-~H* *A*
A * \ *
** * *
A j* *A *
A
A A
* A A
~A
A<
A A
A
.A A
p
A
A Systems Not Meeting TOC Removal
-------
Cj>
1
*«.
1
t]
i
3
u>
*••
k—
§
I
!
Kj
'**•
^0
X>
Co
— — Meeting Removal
- — ~— Not Meeting Removal
— Percent TOC Removal Target
100%
B0%
70%
- W%
i i«
30%
10%
0
100% ,
80%
80%
70% ,
H 50% .
2
£ 40% .
30% .
20% .
10% ,
9% .
O Alkalinity < 60
1 .7 mg/L < TOC < 3.4 mg/L
J
/ 35 %
/*
I ' __- — I
/
^ _/
-""'" 1 \
^ \
% 1G% 20% 30% 40% 50% " 80% 70% 60% 80% 100
Percent TOC Removal
O Alkalinity < 60
3.4 mg/L < TOC < 6.8 mg/L
/
( t
/ 45% I
J ^— -~ — ' ' '
f ^£" !
i r
^ )
^ r i
j I \
/ f i
0% 10% 20% 30% 40% 50% 60% 70% 80% $0% 100
Percent TOG Removal
^p Alkalinity < 60
i(xm..,_. I9G>Mm!L
901
•01
70S
4! 301
» 40?
301
t, -
» ! 50%i
*.
f, :
f, i
30% i
tm, }
°* !
Exhibit 3.8
Enhanced Coagulation Matrix Distribution of TOC Removal
(Based on Data for Large Surface Water Systems)
60 < Alkalinity* 120
1.7 mg/L < TOC < 3,4 mg/L
80%
S0%
40%
30*
20%
10% \y
0%
25%
10% 20% 30% 40% 50% 80% 70% 80% 90% 100%
Percent TOC Removal
60 < Alkalinity < 120
' 3.4•mail 120
1.7 mg/L < TOC < 3.4 mg/L
0% 10% 20% 30% 40% 50% 60% 70% 80% SQ% 100%
Percent TOC Removal
O
Alkalinity > 120
3.4 mg/L 120
TOC > 6.8 mg/L
o% §0% ico%
0% 10% 20% 30% 40% 50% 6014 70% 60% 90% 100%
Percent TOC Removal
10% 20% 30% 40% 50% 60% 70% 80% 00% 100%
Percent TOC Removal
0% 10% 20% 30% 40% 50% 60% 70% 60% 90% 100%
Percent TOC Removal
-------
3,4 Existing Treatment Characterization
Two parameters—TTHMs and HAAS—are important byproducts of treatment that the Stage ] DBPR
aims at reducing. Establishing a baseline concentration for these parameters is needed to estimate the
benefits of reductions in their concentrations. This baseline characteristic is a moving target because both
disinfection practices and byproduct control strategies may have changed in the wake of the
implementation of the Surface Water Treatment Rule and Total Coliform Rule in the early 1990s. In
addition, some systems may have begun to implement changes in anticipation of the IESWTR and Stage
1 DBPR, as these regulatory proposals have been on the table for a number of years.
The WIDE serves as the source for TTHM and HAA5 data for surface water systems for use in this RIA.
The distribution of TTHMs is relatively even across the range (Exhibit 3.9). The distribution of HAAS (a
much smaller data set) is shown in Exhibit 3.10. Subsequent analyses distinguish between systems that
are above certain levels of either parameter. The scarterplot of TTHMs against HAAS (Exhibit 3.11)
segments those systems that meet the levels 64 /ug/L for TTHMs or 48 /ug/L for HAAS from those
systems that exceed either level. These breakpoint levels are used in subsequent analyses and are set at
20 percent below the Stage 1 DBPR MCLs of 80 Mg/L and 60 /^g/L, respectively. This buffer provides
some range for variability taking into account the need for systems to reliably meet compliance targets.
The TWG set the assumed buffer for TTHM and HAAS compliance slightly higher than the buffer for
TOC removal (20 percent versus 15 percent) to reflect the greater variability and uncertainty of
controlling TTHMs and HAAS,
Within each of the nine TOC/alkalinity categories described for the enhanced coagulation technique,
different TOC removal levels apply. Part of the baseline is to characterize the distribution of systems
within each of the nine cells of the 3-X-3 matrix and to identify the number that do not meet the required
levels (See Exhibits 3.5, 3.6, and 3,7) and by how much (Exhibit 3.8). As noted above, Exhibit 3.8
displays the cumulative distributions of those systems meeting and not meeting the TOC removal levels.
The degree to which those not meeting the targets must achieve compliance is shown by the distance
between the cumulative distribution and the target level appropriate to each category. Because the TOC
removal targets are compliance targets, the analysis treats a measured removal of 25 percent (for
example) as though it were actually 21 percent (that is, 25 percent multiplied by 0.85 to allow a 15
percent buffer) in developing compliance forecasts. The levels are shown here for reference.
These data are not a census of affected plants, but the relationships described above are representative of
the universe of the surface water industry and are used in subsequent analyses.
3.5 Risk Assessment and Benefit Analysis
Assessing the benefits of reducing exposure to disinfection byproducts requires performing a risk
assessment to determine the health effects due to exposure to DBFs in drinking water, the reduction in
the health effects produced by the Stage 1 DBPR, and then assigning a value to those reductions. Risk
assessments require information on the health effects, toxicity, and exposure. Benefits analysis require
information on the value of reducing health and other potential damages.
3.5.1 Health Effects and Toxicity
Several sources were used to assess the health effects and hazards posed by DBPs in drinking water.
Available baseline toxicological and epidemiological data is discussed in detail in Chapter 4 and is
Stage I DBPR Final RIA 3-12 November 12, 1998
-------
largely derived from "Summaries of New Health Effects Data" in the EPA drinking water docket (EPA,
October 1997), information contained in the 1998 Notice of Data Availability, and the supporting
information to this document. The TTHM occurrence information from WIDB discussed earlier is also
used to assess exposure and changes to exposure.
3.5.2 Benefits Analysis
To estimate the benefits of reducing the health damage attributable to DBFs, the monetary valuations for
two health endpoints—fatal bladder cancer cases and nonfatal bladder cancer cases—were considered.
Chapter 4, Section 4.5 discusses fully the source and derivation of the values for fatal and nonfatal
bladder cancer used throughout this R1A. In addition, Appendix H also discusses the assumptions about
the bladder cancer incidence, fatality rate, and trends in population that might affect the benefits
projections.
Stage / DBPR Final RIA 3-13 November 12, 1998
-------
Exhibit 3.9
Cumulative Distribution of TTHM Concentration within Distribution System
(Based on Data for Large Surface Water Systems)
TTHM
Exhibit 3.10
Cumulative Distribution of HAAS Concentration within Distribution System
(Based on Data for Large Surface Water Systems)
40 50 60
HAAs(ng/L)
Stage 1 DBPR Final RIA
3-14
November 12, 1998
-------
Co
I
I
2
o-
Exhibit3.il
Plant-Level Concentration of TTHMs and HAAS within Distribution System
(Based on Data for Large Surface Water Systems)
100
90
80
70
60
en
3-
to
50
40
30
10
«-«
23% are above targets
11'('!J'" K FW,! *' ^ y"T ' S¥u- f Y/ii^S^iii^I^'~~ *£" "~--^'"l"°ri'J'""g j -^ ''»-* " ""* ' * ^fe^*-*' " ?™^^" , ! 1"" —^-sr ^ f
77% are below targets
^ p A
,B,,',V "fu,' ,
1*4^1,', ,X
* *
10
Source: WIDB Data
20
30
40
TTHMs
50
60
70
80
90
-------
4: Benefits Analysis
4.1 Introduction
The benefit derived from the promulgation of a drinking water standard has often been measured by the
health damages (medical costs and productivity losses) that will be avoided as a result of the
enforcement of the standard. This is an incomplete concept of the true economic benefit. The complete
concept of the economic benefit of drinking water standards consists of the total value to the consumer of
the reduced health risk. The total value includes not only the avoidance of health damages, but also the
avoidance of the pain and suffering associated with the health endpoint and the disutility associated with
risk and uncertainty (i.e., the risk premium). This larger conceptual framework goes beyond valuing out-
of-pocket medical costs and lost time to include the value consumers place on avoiding pain and
suffering and the disutility associated with risk and uncertainty, captured in the consumer's "willingness-
to-pay" for the change (Freeman, 1979). To the extent possible, the analysis in this chapter focuses on
quantifying and valuing the willingness-to-pay to avoid health damages, using out-of-pocket costs as a
substitute only if the more complete value is unavailable.
The potential economic benefits of the Stage 1 DBPR are derived from the increased level of protection
to public health and decreased level of potential health risks such as cancer and adverse
reproductive/developmental effects from disinfection byproducts (DBFs). As discussed below, there are
significant uncertainties in the available data to assess health risks associated with exposure to DBFs.
Because of these uncertainties, this R1A presents five alternative approaches to assess net health benefits
or cost effectiveness of the Stage 1 DBPR. The health benefit of a drinking water standard is a reduction
in risk—i.e., a decrease in the likelihood of potential health damage that would translate to economic
benefits. The analysis of uncertainties that enter into the assessment of health damage is critical, and it is
a central theme that is carried through this R1A.
Based on a consideration of the five alternative approaches discussed in Chapter 6, there is a reasonable
basis to believe that the Stage 1 DBPR will produce positive net benefits and is superior to the
alternatives of no action or stronger intervention. It is also important to stress that the benefits that have
been quantified in this chapter are based on human bladder cancer cases only. Other potential benefits
from this rule could include other cancers (e.g., colon and rectal) and adverse reproductive and
developmental effects. Data were inadequate for quantifying these benefits, however. Since economic
benefits inherently derive from a reduction in risk, these qualitative benefits should be considered when
evaluating the Stage 1 DBPR.
4.2 Health Risks from Exposure to DBPs
Risk assessment is an integral element of benefit/cost analysis and environmental decision making. It is
used to characterize and estimate the potentially adverse health effects associated with exposure to
environmental agents and to understand potential benefits. It follows a standard methodology employed
within EPA and the Federal government and is generally organized by the paradigm put forward by the
National Academy of Sciences/National Research Council (1983; 1994). Risk assessment is based on
analysis of scientific data to determine the likelihood, nature, and magnitude of harm to public health
associated with particular agents, and involves four types of analysis: hazard identification, dose-
response assessment, exposure assessment, and risk characterization. In the case of DBPs, the RIA
Stage 1 DBPR Final RIA 4.] November 12, 1998
-------
focuses on the potential bladder cancer hazard associated with exposure to DBFs. Exhibit 4.1 illustrates
the steps in a traditional risk assessment process for characterizing the potential human cancer associated
with DBFs in drinking water.
Exhibit 4.1 Steps in the Risk Assessment Process for Cancer
HAZARD EXPOSURE RISK
IDENTIFICATION ASSESSMENT CHARACTERIZATION
Toxicity
(dose-response
relationships)
X
Exposure
Health effects
(# of cases of cancer)
Population Size and Distribution
Ingestion/Dose Human Intake Factors
Concentration of DBFs in Finished
Drinking Water Supply and Available
for Human Consumption
DBFs Formed (or Removed) During
Treatment or in the Distribution
System
Concentration of DBFs in Source
Water
In the case of Stage 1 DBPR, it is only possible to perform dose response assessments for a few
individual DBFs—chloroform, BDCM, bromoform, and DBCM—based on laboratory animal studies.
Health research of other DBFs and mixtures of DBFs is continuing but not yet sufficient to perform a
dose-response assessment.
Stage 1 DBPR Final RIA
4-2
November 12, 1998
-------
EPA's main mission is the protection of human health and the environment. When carrying out this
mission, EPA must often make regulatory decisions with less than complete information and with
uncertainties in the available information. EPA believes it is appropriate and prudent to err on the side of
public health protection when there are indications that exposure to a contaminant could present risks to
public health.
The National Research Council (NRC) noted in 1983, and in 1994, that uncertainties are inherent in risk
assessment because scientific knowledge is not typically complete regarding the health risks of particular
agents, and thus, default assumptions must be made in risk assessment. This is the case with potential
health risks associated with exposure to DBFs in disinfected drinking water. In its 1994 report, the NRC
supported the continued use of default assumptions as a reasonable way to deal with uncertainty and
recommended that EPA explain the science and policy considerations underlying the appropriate default
assumptions in a risk assessment.
In regard to the Stage 1 DBPR, EPA acknowledges that the assessment of public health risks from
disinfection of drinking water currently relies on inherently difficult and incomplete empirical analysis.
On one hand, epidemiologic studies of the various populations are hampered by difficulties of design,
scope, and sensitivity. On the other hand, uncertainty is involved in using the results of high-dose animal
toxicological studies of a few of the numerous byproducts that occur in disinfected drinking water to
estimate the risk to humans from chronic exposure to low doses of these and other byproducts. Such
studies of individual byproducts cannot characterize the entire mixture of DBPs in drinking water. While
recognizing these uncertainties, EPA continues to believe that the Stage 1 DBPR is needed for protection
of public health from exposure to potentially harmful DBPs. There was agreement among the members
of the regulatory negotiating committee on the need to take steps to reduce exposure to DBPs. There is
also general agreement among the scientific community that the risk associated with disinfected drinking
water and DBPs can not be reliably quantified at this time. Under the Executive Order 12866, EPA must
conduct an RIA. It should be understood that the quantitative analyses presented in this chapter are done
so in support of the RIA Executive Order and to provide some reasonable basis for projecting potential
health risks.
4.2.1 Hazard and Dose-Response Assessment: Toxicology
Since the discovery of chlorination byproducts in drinking water in 1974, a number of studies in
laboratory animals have been conducted. As depicted in Exhibit 4.2, several key DBPs, including
trihalomethanes (THMs), such as chloroform, bromodichloromethane (BDCM), and bromoform, have
been shown to produce cancer in 2-year rodent bioassays. Certain haloacetic acids (HAAs), such as
dichloroacetic acid, also have been reported to cause cancer in animal studies. Several DBPs, including
chlorite, DCA, trichloroacetic acid (TCA), and BDCM, have been shown to cause reproductive or
developmental effects in laboratory animals. A few DBPs have been identified as potentially causing
other health problems such as nervous system effects in laboratory animals (e.g., DCA). EPA thus
believes that these toxicological studies provide supporting evidence that DBPs present a potential public
health problem that must be addressed.
Stage 1 DBPR Final MA 4-3 November 12, 1998
-------
Exhibit 4.2
Potential Health Effects from Disinfectants and Disinfection Byproducts
from Laboratory Animal Studies
Contaminants
Health Effects 11]
Disinfectants
Chlorine dioxide
neurodevelopmental, hemolytic, reproductive
Trihalomethanes
Chloroform
Bromodichloromethane
Chlorodibromomethane
Bromoform
cancer, liver and kidney toxicity, developmental
cancer, liver and kidney toxicity, developmental
cancer, liver and kidney toxicity
cancer, liver and kidney toxicity, reproductive, developmental
Haloacetic Acids
Dichloroacetic acid
Trichloroacetic acid
Dibromoacetic acid
Bromochloroacetic acid
Bromoacetic acid
cancer, liver toxicity, developmental, reproductive, neurotoxicity
cancer, liver toxicity, developmental
reproductive, developmental
reproductive, developmental
developmental
Inorganic DBFs in Stage 1 DBPR
Bromate
Chlorite
cancer, kidney toxicity, reproductive
neurodevelopmental, reproductive, hemolytic
Aldehydes
Formaldehyde
Acetaldehyde
cancer, developmental [2]
cancer, developmental [2]
Other
MX
cancer [3]
[1] Health effects summarized in: 1) preamble to the 1994 proposed Stage 1 DBPR (59 FR 38668) and the criteria documents
accompanying the proposed rule (USEPA, 1993; 1994a; 1994b; 1994c; 1994d; 1994e; 2) preamble to the 1997 Notice of Data Availability
{62 FR 59388) and in "Summaries of New Health Effects Data" (USEPA, 1997); and in the preamble to a 1998 Notice of Data Availability
(63 FR 15674) and in several assessment documents mat accompanied the Notice (USEPA, 1998a; 1998b; 1998c,
[2] Integrated Risk Assessment System (IRIS),
[3] Komulainen, et al., 1997.
To date, EPA has established cancer assessments for seven DBFs, as reported in the 1994 Proposed Stage
1 DBPR, the 1997 and 1998 NODAs, and the Integrated Risk Information System (IRIS). A health
assessment on a given chemical is included in IRIS after a comprehensive review of all available health
data by U.S. EPA scientists from several Agency program offices, including the Office of Research and
Stage 1 DBPR Final MA
4-4
November 12, 1998
-------
Development. The information in IRIS
includes a weight-of-evidence evaluation of
whether the chemical has the potential to be
a human carcinogen and, generally, a dose-
response assessment. An RfD and RfC may
also be available for noncancer toxicities.
The dose-response assessment involves
describing how the frequency of an adverse
effect changes with the amount of exposure
to a substance. The sidebar summarizes the
DBF animal cancer information contained
in the DBF proposed rule, the NODAs, and
IRIS,
The cancer assessments presented in
Exhibit 4.2 rely on animal studies
conducted at DBF exposures much higher
than those found in drinking water. Some
studies (e.g., for BDCM, bromoform, and
DBCM) did not use the most relevant route
of human exposure (i.e., drinking water) but
rather delivered the DBF to the animals via
corn oil gavage. Thus, several
extrapolations are required to project human
risk (e.g., from high to low doses, from
nonhuman species to human beings, from
one route to another route of exposure).
Each extrapolation may introduce
uncertainty into the assessment.
Summary of DBF Ca
Chemical
Bromoform
Bromodichloromethane
Chloroform3
Dibromoehloromethane
Dichloroacetic Acid
Trichloroacetic Acid
Bromate
ncer Risk Asst
Human
Carcinogen
Assessment1
Probable
Probable
Probable
Possible
Probable
Possible
Probable
>ssments
Dose
Response
Assessment2
2.3 X 10-'
(IRIS, 1994)
1.8X10-6
(IRIS, 1994)
1.7 X10-7
(IRIS, 1994)
2.4 X 10-*
(IRIS, 1994)
Not available
(1998 NODA)
Not available
(IRIS, 1994)
2X 10'5
(1998 NOD A)
1 Classified under EPA 1986 Cancer Risk Assessment Guidelines
2 Dose Response information is the Drinking Water Lifetime Unit
Risk (risk per ^g/liter)
3 Under Agency review
These assessments also use the Agency's default assumption of low-dose risk (i.e., linear extrapolation)
to extrapolate from the high doses used in animal studies to the anticipated low environmental human
exposures because the mode of carcinogenic action is not understood for most DBFs at this time. EPA
continues to believe, as discussed in the 1998 NODA, that the issues underlying a nonlinear approach for
estimating the carcinogenic risk associated with lifetime exposure to chloroform via drinking water is
well founded. However, based on several policy and scientific issues raised in public comments from the
1998 NODA, EPA believes it is important that additional deliberations with EPA's Science Advisory
Board be completed on the questions of a nonlinear approach. Therefore, the low-dose, linear-dose
response assessment for chloroform will be used in this RIA. Although cancer assessments are available
for several key DBFs, it is important to note that cancer data are lacking for the majority of DBFs. Thus,
a comprehensive assessment of DBF cancer risks is not possible.
Research into cancer effects of other DBFs and health effects is ongoing. The Stage 1 DBPR is expected
to reduce health effects associated with Total Trihalomethanes (TTHMs), five Haloacetic Acids (HAAS),
chlorite, and bromate through the setting of maximum contaminant levels (MCLs). Other DBFs will be
controlled by these MCLs, as well as the enhanced coagulation treatment technique. Health damages that
may be reduced include cancer, reproductive and developmental effects, and neurotoxicity.
Stage I DBPR Final RIA
4-5
November 12, 1998
-------
4.2.2 Hazard Assessment: Epidemiology
Cancer epidemiological data provides valuable information that contributes to the overall weight-of-
evidence evaluation on the potential human health hazards from exposure to chlorinated drinking water.
Approximately 30 cancer epidemiological studies have been conducted over the past 20 years to examine
the association between exposure to chlorinated water and cancer, including several new studies
published since the 1994 proposal (EPA, 1994b, 1997; 1998). These studies have reported small relative
risks for bladder, rectal, and colon cancer incidence for populations consuming .chlorinated drinking
water for long periods of time (EPA 1994, 1997, 1998).
Several epidemiology studies have been completed evaluating the association between exposure to
chlorinated drinking water and reproductive and developmental outcomes (EPA, 1994; 1997; 1998),
While there are fewer studies than for cancer, more-recent, better-designed studies have suggested an
association between exposure to drinking water with elevated THMs and adverse reproductive and
developmental outcomes. In particular, a study by Waller, et al., (1998) suggests an association between
consumption of drinking water containing high concentrations of THMs with an increased risk of early
term miscarriage. Another recent report by Klotz and Pyrch (1998) in New Jersey, reported a small
increased risk of neural tube defects associated with consumption of drinking water containing high
levels of TTHMs. However, no significant associations were observed with individual THMs, HAAs,
and haloacetonitriles.
4.2.3 Hazard/Risk Characterization
As conveyed in the 1994 proposal, the interpretation of the epidemiological studies on chlorinated
drinking water remains controversial. EPA believes that causality has not been established between
exposure to chlorinated drinking water and adverse health effects based on epidemiological studies. As
discussed later, EPA acknowledges that the epidemiological and toxicological data are limited for
making quantitative inferences regarding exposure to DBFs and disease. Nevertheless, EPA believes that
the overall weight-of-evidence (i.e., epidemiological findings plus lexicological results) have sufficient
merit to support a public health concern and thus the need to reduce exposure to DBFs in drinking water.
4.3 Exposure Assessment
A large portion of the U.S. population is potentially exposed to DBFs via drinking water. Over 200
million people in the U.S. are served by PWSs that apply a disinfectant (e.g., chlorine) to water in order
to provide protection against microbial contaminants. While these disinfectants are effective in
controlling many harmful microorganisms, they combine with organic and inorganic matter in the water
and form DBFs, some of which may pose health risks. One of the most complex questions facing water
supply professionals is how to minimize the risks from these DBFs and still control for microbial
contaminants. Because of the large number of people potentially exposed to DBFs, there is a substantial
concern for any health risks that may be associated with exposure to DBFs.
Several factors are necessary to assess the exposure to DBFs: the size of the population potentially at
risk; the method and rate of ingestion; and the concentration of DBFs in drinking water. Because DBFs
are formed in drinking water by the combination of disinfectants with organic compounds, the
population at risk is identified as the population served by drinking water systems that disinfect. Based
on recent Safe Drinking Water Act Information System (SDW1S) data, Exhibit 4.3 contains the estimated
population served by each of the four system categories. Based on recent information, it was assumed
Stage I DBPR Final RJA 4-6 November 12, J 998
-------
that all surface water systems disinfect and a portion of ground water systems disinfect (95 percent by
population for large systems and 83 percent by population for small systems). Approximately 239
million persons are estimated to be served by water systems that disinfect and are potentially exposed to
DBFs. This widespread exposure represents over 88 percent of the total U.S. population (270 million).
The route of exposure is through drinking disinfected tap water. The general adult population is assumed
to consume nearly 2 liters of water per day (which represents the 84th percentile) (Haas and Rose, 1995).
Exhibit 4.3 Population Potentially Exposed to DBPs
Large Surface Water
(> 10,000 population)
Small Surface Water
(< 10,000 population)
Large Ground Water
(> 10,000 population)
Small Ground Water
(< 10,000 population)
TOTAL
Percent of Population
Receiving Disinfected
Population Served Water
141,297,000 100%
17,232,000 100%
56,074,000 95%
32,937,000 83%
Population Receiving
Disinfected Water
141,297,000
17,232,000
53,270,300
27,337,710
239,137,010
In general, little data are available on the occurrence of DBPs on a national basis. Although there is
sufficient occurrence data available for key THMs in large water systems to develop a national
occurrence distribution for that subset of systems, data are limited for small water systems. Similarly,
some occurrence data for HAAS are available for large surface water systems but not small surface water
and ground water systems. Thus, the development of a national distribution capturing all system sizes
and types is problematic. Data are also lacking on the co-occurrence of the mix of DBPs found in
drinking water.
4.4 Baseline Risk Assessment Based on TTHM Toxicological Data
As shown in Exhibit 4.4, a quantitative risk assessment based on laboratory animal studies can be
performed using the dose-response information on certain THMs (found in IRIS and the supporting EPA
assessment documents 1994,1997, 1998). These assessments, however, capture only a portion of the
potential risk associated with DBPs in drinking water. It is not possible, given existing toxicological and
exposure data, to gauge how much of the total cancer risk associated with the consumption of chlorinated
drinking water is posed by TTHMs alone. An assessment of certain key THMs, however, should provide
some estimation of the potential human risk, albeit limited.
As discussed in Section 4.2.1, performing the risk assessment based on TTHM toxicological data
requires making several assumptions and extrapolations (from a nonhuman species to humans, from high
doses in the laboratory study to lower environmental exposures, and from a nondrinking water route to
the relevant route of human exposure). Assumptions are also made about the occurrence of TTHMs and
the individual DBPs. EPA has derived a weighted average TTHM baseline concentration for use in the
Stage 1 DBPR Final RIA
4-7
November 12, 1998
-------
exposure assessment (described in detail in Appendix .G). The mean weighted average baseline TTHM
concentration is estimated at 43.55 Mg/L, with a 25th percentile of 4] .2 /ug/L and a 75* percentile of 45,9
,ug/L, as modeled using a Monte Carlo simulation. It should be noted that this range does not capture the
full variability of TTHMs in all systems. Instead, it captures the distribution around the weighted average
(central tendency), which is an adequate value for risk assessment.
Occurrence data from an EPA DBF field study indicate that chloroform is the most common THM (in
general, about 70 percent of total THMs), with bromoform being the least common (1 percent).
Bromodichloromethane has an occurrence of approximately 20 percent, with dibromochloromethane
comprising the final 8 percent. These proportions are used to divide the average TTHM concentration
into the concentration for the four individual compounds. It is important to understand that this study was
biased towards systems with potentially high DBF levels and towards systems that were low in bromide.
In systems with higher bromide levels, the relative percentages of the different THMs would shift to the
more brominated species.
Two estimates of risk factors are used to estimate the cancer incidence. The first set of lifetime unit risk
factors is from EPA's IRIS system and EPA (1998) and represents the upper 95 percent confidence limit
of the dose-response function. The second estimate of lifetime unit risk is the maximum likelihood
estimate used in the 1994 analysis that represents the central tendency of the dose-response function
(Bull, 1991). The annual unit risk is calculated by dividing the lifetime risk by a standard assumption of
70 years per lifetime.
To calculate the annual incidence of cancer due to consumption of TTHMs in drinking water, the annual
drinking water unit risk is multiplied by the number of units, in this case the concentration of TTHMs in
/-ig/L, broken out into individual THMs based on the proportions presented above. Exhibit 4.4 contains
the resulting estimated annual cancer cases due to TTHMs in drinking water.
Based on these cancer risk estimates derived from laboratory animal studies, the annual number of
cancer cases attributable to TTHMs is approximately 100. Using the maximum likelihood estimates, the
number of cancer cases is about 2. For the purposes of the analyses that follow, a range of zero to 1 - 100
possible baseline cancer cases is assumed to be attributable to TTHMs based on existing toxicological
data.
Stage I DBPR Final R1A 4-8 November 12, 1998
-------
Exhibit 4.4
Baseline Cancer Incidence
Based on Modeled TTHM Concentration and Toxicological Data
Annual Cases = Population Exposed (persons) X DBF Concentration (/ug/L) X Annual Risk Factor
(cases/persons/year/^g/L)
Total Population (served by systems that disinfect) (Appendix G)
239,137,010
Pre-Stage 1 Population-Weighted Average
(See Appendix G)
Percent
TTHMs
Chloroform
Bromodichloromethane
Dibromochloromethane
Bromoform
100%
70%
21%
8%
1%
Concentration
Mg/L
43.55
30.49
9.15
3.48
0.44
Drinking Water Unit Risk Factors (from
Chloroform
Bromodichloromethane
Dibromochloromethane
Bromoform
IRIS system and central tendency of dose-response)
IRIS (Upper 95% Confidence Interval)
Lifetime
(unit risk ,ug/L)
1.70E-07
1 .80E-06
2.40E-06
2.30E-07
Annual Risk
2.4E-09
2.6E-08
3.4E-09
3.3E-09
Maximum Likelihood Estimate
(from Bull, 1991)
Lifetime Annual Risk
(unit risk Mg/L)
n/a L4E-10
n/a 3.3E-10
n/a 3.3E-10
n/a 1.8E-10
Annual Cancer Incidence Based on Toxicological Data (cases/year)
Chloroform
Bromodichloromethane
Dibromochloromethane
Bromoform
Total
IRIS (Upper 95% Confidence Interval)
17.7
56.2
28.6
0.3
102.8
Maximum Likelihood Estimate
1.0
0.7
0.3
0.0
2.0
Stage 1 DBPR Final R1A
4-9
November 12, 1998
-------
4.5 Baseline Risk Assessment for Bladder Cancer Based on Epidemiological Data
Epidemiological studies can be used to assess the overall population risk associated with a particular
exposure. Since the late 1970s, epidemiological investigations have attempted to assess whether
chlorinated drinking water contributes to the incidence of bladder, colon, rectal, and other cancers.
Several studies have reported a weak association between bladder cancer and exposure to chlorinated
drinking water but a causal relationship has not been confirmed (Freedman, et al., 1997).
A 1992 analysis presented an aggregate meta-analysis of the published epidemiology literature relating
to water chlorination and cancer (Morris, et al., 1992). The analysis identified ten articles published
between 1966 and 1991 that evaluated exposure to chlorinated water and cancer at the level of the
individual (called case-control studies). The analysis evaluated various cancer sites, the most frequent
being bladder and colon (seven articles each), followed by stomach, rectum, and pancreas (six articles
each). The study found that there were elevated risks associated with bladder and rectal cancer (odds
ratio of 1.21 for bladder and 1.38 for rectal). These summary odds ratios were used to generate estimates
of the number of cases of cancer within the general population that could be prevented by eliminating
exposure to chlorinated drinking water (i.e., 10,000 cases per year).
During the regulatory negotiation, some negotiators supported using an estimate of over 10,000 cancer
cases per year linked to exposure to chlorinated water and its associated byproducts based on the meta-
analysis and supporting evidence of carcinogenicity from toxicological studies. Others argued that the
national baseline incidence of cancer attributed to DBFs may be less than 1 case per year, based on
maximum likelihood estimates of toxicological risk associated with the THMs. Deriving toxicological
risk estimates for the other DBFs is not possible because of the lack of occurrence and dose-response
data. The 1994 regulatory negotiation and draft Regulatory Impact Analysis determined that until more
extensive epidemiological and toxicological studies have been completed, it is not possible to draw
definitive quantitative conclusions regarding the extent of cancer and non-cancer health effects from
exposure to DBFs beyond the broad range of less than 1 to 10,000 cancer cases.
Subsequent review of the meta-analysis indicated that the estimate of cancer cases had limited utility for
risk assessment purposes for several methodological reasons. Problems included sensitivity to reasonable
changes in analytical methods and the addition or deletion of one study and evidence of publication bias
within the body of literature. Based on these issues, EPA has decided not to use the Morris, et al., meta-
analysis to estimate the potential benefits from the Stage 1 DBPR.
Several cancer epidemiological studies examining the association between exposure to chlorinated
surface water and cancer were published subsequent to the 1994 proposed rule and the 1992 meta-
analysis. In general, these new studies are better designed than the studies published prior to the 1994
proposal. The new studies include incidence of disease, interviews with the study subjects, and better
exposure assessments. More evidence is available on bladder cancer for a possible association to
exposure to chlorinated surface water than other cancer sites.
Based on the better-designed studies, a range of potential risks was developed through the use of the
population attributable risk (PAR) concept. Epidemiologists use PAR (also referred to as attributable
fraction, attributable portion, or etiologic fraction) to quantify the fraction of disease burden in a
population (e.g., bladder cancer) that could be eliminated if the exposure (e.g., chlorinated drinking
water) were absent. PAR provides a perspective on the potential magnitude of risks associated with
various exposures under the assumption of causality. For example, the National Cancer Institute
Stage 1 DBPR Final R1A 4-10 November 12, 1998
-------
estimates that there will be 54,500 new cases of bladder cancer in 1997, If data from an epidemiological
study analyzing the impact of consuming chlorinated drinking water reports a PAR of 1 percent, it can be
estimated that 545 (54,500 X 0.01) bladder cancer cases in 1997 may be attributable to chlorinated
drinking water.
For the purposes of this RIA, EPA has chosen to estimate cancer risk for chlorinated drinking water
using PAR to provide a basis for the benefit/cost analysis. While EPA recognizes the limitations of the
current epidemiological data base for quantitative risk assessment, EPA considers the data base
reasonable for performing an RIA, as it does not require proof of causality before the determination of
regulatory benefits. To that end, EPA selected studies for inclusion in the quantitative analysis if they
contained the pertinent data to perform a PAR calculation and met all three of the following criteria:
1. The study was a population-based, case-control, or cohort study conducted to evaluate the relationship
between exposure to chlorinated drinking water and incidence of cancer cases, based on personal
interviews;
2. The study was of high quality and well designed (e.g., adequate sample size, high response rate,
adjusted for known confounding factors); and,
3. The study had adequate exposure assessments (e.g., residential histories, actual THM data).
Using the above criteria, five bladder cancer studies were selected for estimating the range of PARs.
*• Cantor, etal,, 1985;
> McGeehin, etal,, 1993;
•» King and Marrett, 1996;
»• Freedman, et al., 1997; and
» Cantor, etal., 1998.
Exhibit 4.5 contains a summary of these five bladder cancer studies.
The PARs from the five bladder cancer studies ranged from 2 percent to 17 percent. These values were
derived from measured risks (odds ratios) based on the number of years exposed to chlorinated surface
water. Because of the uncertainty in these estimates, it is possible that the PAR could also include zero.
The uncertainties associated with these PAR estimates are likely to be large due to the common
prevalence of both the disease (bladder cancer) and exposure (chlorinated drinking water).
This PAR range would pertain to the U.S. population of bladder cancer cases if the study populations
selected for each of the cancer epidemiology studies were reflective of the entire population that
develops bladder cancer; if the percentage of those cancer cases in the studies exposed to chlorinated
drinking water were reflective of the bladder cancer cases in the U.S.; if DBFs were the only carcinogens
in these chlorinated surface waters; and if the relationship between DBFs in chlorinated drinking water
and bladder cancer were assumed to be causal.
Stage / DBPR Final RIA 4-11 November 12, 1998
-------
Exhibit 4.5 Summary of Epidemiology Studies for Bladder Cancer
Study
Description
Summary of Results
Interpretation
Cantor, et al.
(1998)
Case-control study of association
between bladder cancer and
consumption of chlorinated
surface water
- Little overall association
between bladder cancer risk and
exposure to chlorinalion
byproducts
- Bladder cancer risk increased
with exposure duration
Opposite trends were found in
males and females. Total lifetime
and average lifetime TTHM
levels show all risk increases are
apparently restricted to male
smokers.
Cantor, et al.
(1987)
Case-control study of association
between bladder cancer and
consumption of chlorinated
surface water
- Odds ratio for all Whites with
over 59 years of exposure is 1.1
(Confidence Interval: 0.8-1.5)
- Odds ratio for nonsmokers is
2.3 (Confidence Interval: 1.3-
4.2)
- Odds ratio for current smokers
is 0.6 (Confidence Interval: 0.3-
1.2)
Majority of water systems
contained less than
THMs.
McGeehin, et
al.(1993)
Case-control study of association
between bladder cancer and
consumption of chlorinated
surface water
- Odds ratio for bladder cancer
with over 30 years of exposure is
1.8 (Confidence Interval: 1.1-
2.9)
- Odds ratio for cases consuming
over 5 glasses of tap water per
day is 2.0 (Confidence Interval:
1.1-2.8)
Level of total THMs, residual
chlorine, or nitrates not
associated with bladder cancer
risk controlling for years of
exposure.
Freedman, et
al. (1997)
Nested case-control study of
association between bladder
cancer and consumption of
chlorinated drinking water
- Odds ratio for bladder cancer
using 1975 measure of exposure
is 1.2 (Confidence Interval: 0,9-
1.6)
- Slight gradient of increasing
risk with increasing duration
noted only among smokers
Further stratification by gender
showed elevated odds ratios to
be restricted to male smokers.
King and
Marrett{1996)
Case-control study of association
between bladder cancer and
consumption of chlorinated
surface water
- Bladder cancer risk increased
with years of exposure
- Odds ratio for bladder cancer
for 30 years of exposure
compared to 10 years is 1.41
(Confidence Interval: 1.09-1.81)
- Bladder cancer risk increased
with years of exposure
- Risk increases by 11 percent
with each l,QOO^g/L THMs-
years
Statistically signficant only for
lengthy exposures. Results
provide no support for an
interaction between volume of
water consumed and years of
exposure to THMs level > 49
Based on the estimate of 54,500 new bladder cancer cases per year nationally, as projected by the
National Cancer Institute for 1997, the number of possible bladder cancer cases per year potentially
associated with exposure to DBFs in chlorinated drinking water is estimated to range from zero to 1,100
(0.02 X 54,500) to 9,300 (0.17 X 54,500) cases. In making these estimates it is necessary to assume that
these bladder cancer cases are attributable to DBFs in chlorinated surface water, even though the studies
examined the relationship between chlorinated surface water and bladder cancer. This derived range is
not accompanied by confidence intervals, but the confidence intervals are likely to be very wide. EPA
Stage 1DBPR Final RIA
4-12
November 12, 1998
-------
believes that the central tendency (i.e., mean) is a reasonable estimate of the potential range of risk
suggested by the selected epiclemiological studies. Exhibit 4.6 contains a summary of the risk estimates
from the 1994 draft RJA and the estimates derived from the more recent analysis.
It should be noted that an alternative analysis based on odds ratios was conducted to derive a range of
plausible estimates for cancer epidemiologic studies. This analysis was also based on bladder cancer
studies (the five studies cited above in addition to Doyle, et al., 1997). For the purpose of this exercise,
the annual U.S. expected number of 47,000 bladder cancers cited by Morris, et al., (1992) was used to
calculate estimates of the cancers prevented. The number of cancers attributable to DBF exposure was
estimated not to exceed 2,200-9,900 per year. Given the uncertainty in the epidemiology studies, EPA
believes that this range is similar to the 1,100 - 9300 PAR range and used the PAR range for this R1A.
Exhibit 4.6
Bladder Cancer Epidemiology and Toxicology:
Comparison of Estimates Made in 1994 & 1998
1994 Estimates
1998 Estimates
Number of New Bladder Cancer Cases/Year
Number of Estimated Deaths Due to Bladder
Cancer/Year
approx. 50,000
did not state
54,500
12,500
Attributable to DBFs in Drinking Water
Data Source
Causality
Percent Attributable to DBPs
Number of Cancer Cases Attributable to DBPs
Estimated Using Toxicological Data
Estimated Using Epidemiologieal Data
> 15 studies
No
did not state
less than 1*
over 10,000
5 studies that meet specific
criteria
No
2% to 17%
1* to 100**
Zero to 1,100-9,300
* Based on maximum likelihood estimates of risk from THMs
** Based on IRIS 95* percent Confidence Interval estimates of risk from THMs
Interpreting the Risk Results
The current benefits analysis is structured in roughly the same manner as that presented in the 1994
RIA—the baseline cancer risks could lie anywhere from 0 to 1-100 cases per year based on toxicological
data; and 0 to 1,100-9,300 cases per year based on epidemiological data. Consequently, the task is to
assess the economic benefit of the final Stage 1 DBPR in the face of this broad range of possible risk.
Stage 1 DBPR Final RIA
4-13
November 12, 1998
-------
4.6 Baseline Risk Assessment for Other Cancers Based on Epidemiological Data
The scientific literature indicates that exposure to DBFs may be related to other health effects besides the
bladder cancer quantified in the above analyses. Some epidemiology studies have indicated a weak
association (Odds ratio: 1.5-2.0) between consumption of chlorinated drinking water and cancer of other
sites besides the bladder, namely colon and rectal cancer, while other studies have shown no association.
Several population-based, case-control studies have been published that evaluate the association between
consumption of chlorinated drinking water and colon or rectal cancer. Exhibit 4.7 summarizes key
epidemiology studies for colon and rectal cancer.
Exhibit 4.7 Summary of Epidemiology Studies for Colon and Rectal Cancer
Study
Description
Summary of Results
Interpretation
Cragle, et al.
(1985)
Hospital-based, case-control
study of association between
colon and rectal cancer and
exposure to THMs
- Increased risk in those persons
60 years and older with greater
than 15 years of exposure to
chlorinated water
- Increased risk in those persons
greater than 70 years with any
duration of exposure
Results could be misinterpreted
because of common disease and
common low exposure
prevalence.
Young, et al.
(1987)
Case-control, interview study of
association between colon and
rectal cancer and exposure to
THMs
- Odds ratio for all variables
uniformly close to 1.0
Majority of water systems
contained less than
THMs.
Doyle, et al.
(1997)
Prospective cohort study to
evaluate the association between
cancer incidence and drinking
water source and chlorinated
byproducts.
- Increased risk of colon cancer
in women who used municipal
surface water sources in
comparison with women who
used municipal ground water
sources.
Hildesheim, et
al. (1997)
Population-based, case-control
study of the association between
ehlorination byproducts and
colon and rectal cancer
- A significant increase in risk
associated with durations of
chlorinated surface water and
colon cancer was not reported.
- Indicated an association
between rectal cancer and
chlorinated surface water.
EPA believes that the association between exposure to chlorinated drinking water and colon and rectal
cancer, while possibly significant, cannot be determined at this time because of the limited data for these
cancer sites (EPA, March 31,1998).
4.7 Baseline Risk Assessment for Reproductive and Developmental Health Effects
Based on Epidemiological Data
Epidemiological studies have also indicated that consumption of chlorinated drinking water could be
linked to various reproductive and developmental adverse health effects. Epidemiology studies have
evaluated the impacts of chlorinated drinking water on somatic parameters (e.g., birthweight, body
length, cranial circumference, and neonatal jaundice), premature births, intrauterine growth retardation,
Stage 1DBPR Final RIA
4-14
November 12, 1998
-------
increased risk of miscarriage, and neural tube defects (EPA, March 31, 1998). One recent study reported
an elevated odds ratio, generally between 1.5 to 2.1, for the association of neural tube defects with
TTHMs (Klotz and Pyrch, 1998). Another study reports that consumption of tap water containing high
concentrations of THMs, specifically BDCM, is associated with an increased risk of early term
miscarriage (Waller, et al., 1998). Exhibit 4.8 summarizes some of the epidemiological studies on the
potential reproductive and developmental health effects possibly associated with chlorinated drinking
water and DBFs.
Exhibit 4.8 Summary of Epidemiology Studies for Reproductive and Developmental Effects
Study
Kramer, et al.
(1992)
Aschengrau, et
a!. (1993)
Bove, et al.
(1992 a and b)
Savitz, et al.
(1995)
Description
Population-based case-control
study to determine if high levels
of chloroform and other THMs
are associated with low
birthweight, prematurity, and
intrauterine growth retardation
Case-control study of the
association between drinking
water quality and a variety of
birth defects
Cross sectional and case-control
study by the New Jersey
Department of Health of the
association between drinking
water contaminants and birth
weight and birth defects
Population-based, case-control
study of potential risk of
miscarriage, preterm delivery,
and low birth weight based on
water source, amount of water
consumed, and TTHM
concentration
Summary of Results
- Increased risk of intrauterine
growth retardation (Odds ratio:
1 .8; 95% Confidence Interval:
1 .-1-2.9) at THM > lO^g/L
- Slightly increased risk of low
birth weight (Odds ratio: 1.3;
95% Confidence Interval: 0.8-
2.2)
- Higher frequency of stillbirths
correlated with chlorination and
lead levels
- Elevated risk for low-term
birth weight (Odds ratio: 1 .29;
95% Confidence Interval: 1.08-
1.5)
- Increased risk of central
nervous system defects (Odds
ratio: 2.6; 95% Confidence
Interval: 1.48-4.6)
- Increased risk of central neural
tube defects (Odds ratio: 2.98;
95% Confidence Interval: 1.25-
7.1)
- Increased risk of cardiac defects
(OR 1 .44, 95% Cl: .97-2. 1 )
Interpretation
Authors indicate results are
preliminary and should be
interpreted with caution.
Results are preliminary.
Results are useful for hypothesis
generation; should be interpreted
with caution and may be subject
to confounding factors.
"These data do not indicate a
strong association between
chlorinated byproducts and
adverse pregnancy outcome, but
given the limited quality of the
exposure assessment and the
increased miscarriage risk in the
higher exposure group, more
refined evaluation is warranted."
Stage 1 DBP-R Final R1A
4-15
November 12, 1998
-------
Study
Description
Summary of Results
Interpretation
Klotz and
Pyrch(1998)
Case-control study of neural tube
defects and TTHM exposure
- Increased risk levels, generally
between 1.5 and 2.1, for the
association between neural tube
defects with TTHMs
- Statistically significant results
at highest THM exposures (> 40
ppb) and limited to those
subjects in which there was no
other malformation (Odds ratio:
2.1; 95% Confidence Interval
I.I-4.0)
- No clear relationship for HAAs
or HANs
Study adds to weight-of-
evidence concerning the
potential adverse reproductive
health effects from DBFs.
Waller, et al.
(1998)
Population-based study of early
term miscarriage and exposure to
THMs.
- Increased risk of early term
miscarriage associated with high
TTHM exposure in home tap
water (drinking 5 or more glasses
per day of cold home tap water
or drinking any amount of tap
water containing at least 75
Mg/L) (Odds ration: 1.8; 95%
Confidence Interval: 1,1-3.0)
- Increased risk from BDCM
exposure (Odds ratio: 3,0; 95%
Confidence Interval 1.4-6.6)
Study adds to weight-of-
evidence concerning the
potential adverse health effects
from DBFs, but does not prove
that exposure to TTHMs and
BDCM causes early term
miscarriages.
As with the other reported adverse outcomes from the epidemiology studies, there is considerable debate
in the scientific community on the significance of these and earlier findings. While the new
epidemiology studies add to the database on the potential reproductive and developmental effects from
DBFs, the results are inconclusive and do not support quantification of benefits at this time. These
uncertainties, however, need to be considered when evaluating regulatory alternatives from a public
health standpoint.
4.8 Exposure Reduction Analysis
During the 1994 RegNeg, the DBPRAM model (Appendix K) was used to estimate the changes in
exposure due to the provisions of the proposed Stage 1 DBPR. The DBPRAM was a Monte Carlo
simulation model of influent variability combined with a treatment model to predict treatment
performance. The DBPRAM used the Water Treatment Plant model and its chemical equations to
estimate the formation of DBPs given a range of influent waters and various compliance choices. The
DBPRAM estimated that in the average (median) system the proposed Stage 1 DBPR would result in a
reduction in TTHMs of 33 percent, in HAASs of 29 percent, and TOC of 12 percent in large surface
water systems.
As discussed in Chapter 3, review of new data available from the 1996 replication of the WIDB survey
indicated that some of the assumptions underlying the DBPRAM modeling work, drawn from the 1988-
90 WIDB survey, no longer reflect existing baseline conditions. In addition, some of the provisions of
the rule have changed. These changes are outside the sensitivity of the DBPRAM modeling apparatus.
Stage 1 DBPR Final RIA
4-16
November 12, 1998
-------
Considering this new data, EPA undertook an alternative "desktop" analysis to predict exposure
reductions due to the current Stage 1 DBPR based on present baseline conditions. This analysis consists
of a sequence of data and assumptions that lead to quantitative assessment of exposure reduction. While
it is considered the best assessment that can be made at this time, it is necessary to recognize the
substantial uncertainties inherent in such analysis. Appendix G contains a full description of the
methodology and assumptions used to assess the change in exposure resulting from the Stage 1 DBPR.
EPA used the current concentration of TTHMs as a marker to measure the exposure to the range of DBFs
because data are available on the baseline occurrence and formation of TTHMs. There are limited data
on the total mix of byproducts in drinking water. Therefore, the reduction in TTHMs is assumed to
reflect the reduction in exposure to all DBPs. To determine the change in exposure, it is necessary to
estimate the Pre-Stage 1 baseline TTHM concentration and the Post-Stage 1 TTHM concentration. The
difference in the Pre- and Post-Stage 1 TTHM concentrations reflect the potential reduction in TTHMs
and thus in DBPs.
EPA calculated the Pre-Stage 1 TTHM concentration for the four system categories (large and small
surface water, and large and small ground water) and then derived a weighted-average concentration
based on the population served by systems that disinfect within each system category. The Pre-Stage 1
TTHM weighted-average concentration is 43.55 /ug/L, with a modeled 25* percentile of 41.2 Mg/L and a
modeled 75* percentile of 45.9 Mg/L. This distribution represents the variability around the calculated
weighted average, not the full variability of possible TTHM measurements in all systems.
The Post-Stage 1 TTHM weighted average concentration is estimated at 32.9 /ug/L, with a modeled 25th
percentile of 30.9 /ug/L and a modeled 75* percentile of 34.9 /ug/L. Again, this distribution represents the
variability around the weighted average, not the full variability of the underlying TTHM values.
The resulting reduction in exposure as modeled by the Monte Carlo simulation is 24 percent at the mean
with a 25* percentile of 18 percent and a 75* percentile of 30 percent. Please refer to Appendix G for a
full explanation and presentation of results.
4.9 Expected Benefits from Reduction in Exposure to DBPs
The economic benefit of a drinking water standard is a reduction in risk—i.e., a decrease in the
likelihood of health damage. The Stage 1 DBPR is expected to reduce exposure to DBPs by
approximately 24 percent, thereby reducing the likelihood of the health damages described previously,
including the potential risk of bladder cancer, colon cancer, rectal cancer, and developmental and
reproductive effects. Sufficient data, however, is available to quantify and monetize only the benefits
associated with the reduction in bladder cancer. The discussion of monetization of health effects and net
benefits that follows includes only those benefits associated with reducing the risk of bladder cancer. In
deciding whether and how to regulate, it is essential to consider all potential benefits in the protection of
public health and safety, including both the quantifiable benefits and qualitative benefits (Exhibit 4.9).
Stage 1 DBPR Final RJA 4-17 November 12, 1998
-------
Exhibit 4.9 Summary of Stage 1 DBPR Benefits
Category
Analytical Approach
Bladder Cancer Benefits
Fatal Bladder Cancer
Nonfatal Bladder Cancer
Other Cancer Sites
Reproductive Effects
Developmental Effects
Total Benefits
Monetized
Monetized
Qualitative
Qualitative
Qualitative
4.10 Monetization of Bladder Cancer Health Endpoints
Monetary valuations were derived for two health endpoints: fatal bladder cancer cases and nonfatal
bladder cancer cases. The following discusses the source and derivation of the values for fatal and
nonfatal cancer used throughout the regulatory impact analysis.
Bladder cancer is a disease in which cancer (malignant) cells are found in the bladder. Bladder cancer
affects approximately 50,000 individuals in the United States each year. An estimated 54,500 new cases
were expected in 1997. Of these, approximately 11,700 were expected to result in death. Bladder cancer
risk increases with age (over 65) and is much more prevalent in men than women.
Four types of treatment are used for bladder cancer:
*• Surgery (taking out the cancer or removing the bladder in an operation. If the bladder is
removed, a new way for the patient to store and pass urine must be made);
* Radiation therapy (using high-dose x-rays or other high-energy rays to kill cancer cells and
shrink tumors);
*• Chemotherapy (using drugs to kill cancer cells); and
*• Biological therapy (using the body's immune system to fight cancer).
Bladder cancer is one of the first cancers associated with industrialization, due most likely to organic
chemical, solvent, and dye exposure. There is also evidence that bladder cancer risk increases with
increases of fluid intake. However, the largest risk factor in the development of bladder cancer,
responsible for as many as 60 percent of the cases, is cigarette smoking.
Stage 1 DBPR Final RIA
4-18
November 12, 1998
-------
4.10.1 Willingness-to-Pay to Avoid a Fatal Bladder Cancer Case
In regulatory impact analyses, it is common lo use an average willingness-to-pay (WTP) (or willingness-
to-accept) value—derived from either revealed preference or stated preference approaches—as the basis
for monetizing small changes in risk, known as the "value of statistical life (VSL)" (Chestnut and
Alberini, 1997). One recent study, The Benefits and Costs of the Clean Air Act, 1970 to 1990, derived a
distribution of VSL estimates based on 26 individual studies selected as appropriate for policy use
(Chestnut and Alberini, 1997). The distribution was lognormally distributed with a mean of $4.8 million
and standard deviation of $3.2 million at a 1990 price level, truncated at an upper value of $13.5 million
(Chestnut and Alberini, 1997). For the purposes of the benefit evaluations in the Stage 1 DBPR RIA, the
distribution of values was updated to a June 1998 price level by multiplying the distribution by an update
factor of 1.25 through a Monte Carlo simulation.1 The resulting distribution has a mean of $5.6 million
and a standard deviation of $3.16 million. The actual distribution generated by the Monte Carlo
simulation, capped at $16.87 million ($13.5 million X 1.25), is used consistently throughout the DBF
benefits analysis. The results of the updated VSL simulation appear in Appendix H-2.
4,10.2 Valuation of Nonfatal Bladder Cancer Case
The complete valuation of the nonfatal cancer case measures the WTP to avoid a nonfatal case of bladder
cancer. Presumably, the WTP would exceed the medical costs of the illness to include the premium for
risk aversion and the value of avoiding the pain and suffering associated with the treatment of bladder
cancer, including chemotherapy, radiation, and removal of the bladder. A review of the available WTP
literature did not reveal any studies that measured the WTP to avoid bladder cancer, specifically. A
distribution of values derived through a contingent valuation study of the WTP to avoid chronic
bronchitis is used a substitute for the WTP to avoid nonfatal bladder cancer. As an alternative, a cost-of-
illness (COI) value derived directly from medical costs and lost productivity is also estimated and used
in the benefits analysis to compare the results. It is important to note that either the substitute WTP
measure or the COI measure is used to value nonfatal bladder cancer, not both.
Derivation of Cost-of-Illness for Nonfatal Bladder Cancer
The COI estimate for nonfatal bladder cancer cases consists of two costs: the costs of medical treatment
and the cost of lost productivity. The treatment for bladder cancer usually involves surgery, alone or in
combination with other treatments, in 90 percent of the cases. Preoperative chemotherapy alone or with
radiation before cystectomy (bladder removal) has improved some treatment results (American Cancer
Society, 1998).
The treatment costs for bladder cancer were derived from a study of Medicare payments for patients with
bladder cancer. The study found that average payments for bladder cancer were $57,629 in 1990 dollars
(Riley, et al., 1995). Medicare payments cover only part of medical treatment costs. One article reports
that an additional 4.1 percent of medical costs are paid by private insurance companies and an additional
3 percent are paid as out-of-pocket expenses by the patient (Dried and Scheffler, 1992). The total
treatment costs are estimated by multiplying the Medicare costs by 1.071 to take into account
out-of-pocket costs (3 percent) and private insurance costs (4.1 percent) for a total treatment cost of
1 Consumer Price Index, al! items, all consumers, June 1998/1990 average = 163.0/130.7 = 1.25. Assumed to reflect 1998 price
levels.
Stage 1 DBPR Final R1A 4-19 November 12, 1998
-------
$61,721 in 1990 dollars. This value is then updated to a current value of $91,964.2 This value is in the
same range as values for comparable cancer sites calculated and used to estimate benefits of the
Superfund program (EPA, June 1994).
The same Superfund report is the source of the estimate for productivity losses. Using rectal cancer as a
substitute for bladder cancer, the report estimates that the average expected number of productivity loss
days over the first 2 years of treatment is 283 days (EPA, June 1994). Assuming a value-per-day loss of
$101.92, the total value of productivity losses is estimated at $28,843 (Bureau of Labor Statistics, 1998).
The resulting total CO1 used in the benefits analysis is the treatment cost ($91,964) plus the productivity
loss ($28,843) for a total of $121,000. The calculations, assumptions, and estimates for the COI value
appear in Appendix H-3.
Derivation of Willingness-to-Pay for Nonfatal Bladder Cancer
As stated previously, there is no reported value in the literature for the WTP to avoid a nonfatal bladder
cancer case. One study, however, derived WTP to avoid a case of chronic bronchitis through a contingent
valuation survey that measured risk-risk tradeoff (Viscusi, et al, 1991). The study asked participants to
compare the risk of chronic bronchitis with the risk of a fatal auto accident to produce a relative
valuation. The results "...suggest that the risk of a chronic bronchitis case is worth 32 percent of the
comparable risk of death, as measured by the median tradeoff rate" (Viscusi, et al., 1991). The study also
measured a risk-dollar tradeoff by comparing the risk
reduction for chronic bronchitis or an auto accident
fatality against a cost-of-living increase. The result is
a distribution of values representing the WTP to
avoid a case of chronic bronchitis with a median of
$457,000 in 1990 dollars. The value derived for the
auto fatality was $2.29 million in 1990 dollars, which
is well within the value of statistical life distribution,
although at the lower end.
The distribution of values for the WTP to avoid
chronic bronchitis reported in the study is updated
through a Monte Carlo simulation by multiplying by
a Consumer Price Index factor of 1.25. The resulting
distribution, with a median value of $535,600, a
mean value of $587,500, and truncated at $1.5
million, is used throughout subsequent analyses. The
results of the Monte Carlo simulation are presented
in Appendix H-4.
The WTP to avoid chronic bronchitis is not a perfect
substitute for the WTP to avoid a case of bladder ,
cancer, though it appears to be a reasonable
approximation for the purposes of benefit assessment
(see sidebar). Some of the attributes of chronic
Applicability of Using WTP to Avoid Chronic
Bronchitis for WTP to Avoid Bladder Cancer
1. Both have similar long-term quality of life health
implications, including:
» Using medical equipment for the rest of life (for
chronic bronchitis, wearing small portable oxygen
tank; for bladder cancer, wearing a bag to store
and pass urine),
* Limiting recreational and job-related activities,
* Visiting doctors regularly and taking medication,
and
* Experiencing periods of depression.
Net Impact on WTP: About the same
2. Chronic bronchitis is associated with more
obvious lingering implications, such as shortness of
breath and more frequent chest infections,
Net Impact on WTP: Chronic bronchitis higher
3. Bladder cancer has more severe acute health
effects, including major surgery and undergoing
radiation or chemotherapy treatments (with attendant
side effects).
Net Impact on WTP: Bladder cancer higher
1 Consumer Price Index, medical care, all consumers, June 1998/1990 average = 242,0/162.8 = 1.49. Assumed to reflect 1998
price levels.
Stage 1 DBPR Final RJA
4-20
November 12, 1998
-------
bronchitis and bladder cancer are quite similar, such as using a respirator for chronic bronchitis and using
a bag to store and pass urine for bladder cancer. Chronic bronchitis may be associated with more severe
chronic effects, but bladder cancer is associated with more severe acute effects. In addition, a comparison
of the COI for chronic bronchitis and bladder cancer reveals that the COIs are similar: $95,000 for
chronic bronchitis and $121,000 for bladder cancer (Cropper and Krupnick, 1989).
4.11 Range of Potential Monetized Benefits from Reducing Bladder Cancer
The range of potential benefits from the Stage 1 DBPR can be calculated by applying the monetary
values for fatal and nonfatal bladder cancer cases to the estimated number of bladder cancer eases that
will be reduced by the rule. The following assumptions are used to estimate the range of potential
benefits:
* An estimate of the number of bladder cancer cases attributable to DPBs in drinking water ranges
from 0 to 9,300 annually;
*• A 24 percent reduction in exposure to DBFs (using reduction in TTHMs as a proxy for reduction
for all DBFs) due to the Stage 1 DBPR (75 percent Confidence Interval of 18 to 30 percent) will
result in a 24 percent reduction in bladder cancer cases;
*• A value per statistical life (VSL) saved for fatal bladder cancer is represented by a distribution
with a mean of $5.6 million; and,
>• A WTP to avoid a nonfatal case of bladder cancer is represented by a distribution with a mean of
$587,500.
Using the low end of the risk range of 0 bladder cancer cases attributable to DBFs results in a benefits
estimate of $0. To calculate the high end of the range, the estimated 9,300 attributable cases is multiplied
by the percent reduction in exposure to derive the number of bladder cancer cases reduced (9,300 X 0.24
= 2,232 bladder cancer cases reduced). Assuming that 23 percent of the bladder cancer cases end in
fatality and 77 percent are nonfatal, the number of fatal bladder cancer cases reduced is 513 (2,232 X
0.23) and the number of nonfatal bladder cancer cases is 1,719 (2,232 X 0.77). Based on the valuation
distributions described above, the estimate of benefits at the mean associated with reducing these bladder
cancer cases is approximately $4 billion. It should be noted that these estimates do not include potential
benefits from reducing other health effects (other cancers; reproductive and developmental effects) that
cannot be quantified at this time. While the low end of the range cannot extend below $0, it is possible
that the high end of the range could extend beyond $4 billion if the other reductions in risk could be
quantified and monetized. No discount factor has been applied to these valuations, although there is
likely to be a time lag between compliance with the rule and realization of benefits.
Given this wide range of potential benefits and the uncertainty involved in estimating the risk attributable
to DBPs, EPA undertook five different approaches to assessing the net benefits of the Stage 1 DBPR to
assist decision-makers in choosing a regulatory alternative for addressing the public health risk of DBPs
in drinking water. These approaches are described in Chapter 6 and should be considered both
individually and in the aggregate in support of rulemaking to protect the public's health and safety.
Stage 1 DBPR Final RJA , 4-21 November 12, 1998
-------
5: Cost Analysis
5.1 Introduction
This chapter estimates the total national costs of complying with the Stage 1 Disinfectants/Disinfection
Byproducts Rule. It discusses which elements of the rule incur costs, on what basis those costs are
estimated, and how they are aggregated. Chapter 6 compares the cost estimates with the potential
benefits of the rule.
The cost estimation for the Stage 1 DBPR combines information from existing data sources with
technical assumptions based on expertise developed by the Microbial-Disinfectants/Disinfection
Byproducts (M-DBP) Advisory Committee and its Technologies Working Group (TWG). These
estimates are the result of an iterative process that was continually updated by new data and modified
assumptions. Where necessary, a chronology of the decisions that formed a particular estimate is
discussed.
5.1.1 How This Chapter Is Organized
This chapter describes how these estimates were derived from previous analyses, changes in the Stage 1
DBPR, and review of recent studies. Section 5.2 presents an overview of the process and available new
data. Section 5.3 discusses DBF treatment effectiveness and costs. Section 5.4 more fully describes the
national cost estimates. Finally, -Section 5.5 addresses the impacts of the rule on small systems,
consistent with analytical requirements under the Regulatory Flexibility Act, Additional documentation
on the analyses and cost estimates in this chapter are included in Appendices A through E.
5.2 The Stage 1 DBPR and New Data
A regulatory impact analysis was developed in 1994 in support of the 1992-1993 Regulatory Negotiation
(RegNeg) process that produced the proposed Stage 1 DBPR. The results of the 1992-1993 RegNeg
process are summarized in Chapters 1 and 2. Since the rule was proposed, some new sources of data have
become available that were used to update the forecasts made in the 1994 RIA. In addition, there have
been several revisions incorporated into the final Stage 1 DBPR (and into the companion Interim
Enhanced Surface Water Treatment Rule—IESWTR) that have effects on national cost estimates.
The major revisions in the rule that produced changes in the national cost forecast include the
following—
>• Allowance of credit for disinfection prior to the point of coagulant addition;
>• Re-definition of total organic carbon (TOC) removal requirements for enhanced coagulation; and
»• Re-specification of minimum disinfection requirements for the IESWTR.
Stage 1 DBPR Final RIA , 5-1 November 12, 1998
-------
Exhibit 5.2
Summary of Costs under the Proposed Stage 1 DBPR in
1998 Dollars (1.129 Inflation Factor), 10 Percent Cost of Capital ($000)
Treatment Costs
Total Capital Costs
Annual O&M
A nnualized Capital Costs
Annual Utility
Treatment Costs
Monitoring and Reporting Cost
Start-Up Costs
Annual Monitoring
State Costs
Slart-Up Costs
Annual Monitoring
Total Annual Costs
Surface Water Systems
Small Larie Total
677,400 2,258,000 2,935,400
56,450 333,033 389,503
104,99' 233,703 338,700
161,447 566,758 728,205
85 50 134
18,371 15,190 33,561
Ground Water Systems
Small Large Total
1,241,900 790,300 2,032,200
99,352 63,22-1 162,576
188,5-13 88,062 2~6,60S
287,895 152,415 440,310
722 35 807
62,027 23,408 85,435
All Systems
4,967,661)]
~*_ ~~ J
552,0*1
616,434
$1,168,515
942
118,9»S
4,058
13,593
51,306,103
Exhibit 5.3
Summary of Costs under the Proposed Stage 1 DBPR in
December 1992 Dollars, 10 Percent Cost of Capital ($000)
Treatment Costs
Total Capital Costs
Annual O&M
Annualiicd Capital Costs
Annual Utility
Treatment Costs
Monitoring and Reporting Cost
Start-Up Costs
Annual Monitoring
State Costs
, Start-Up Costs
Annual Monitoring
Total Annual Costs
Surface Water Systems
Small Large Total
600,000 2,000,000 2,600,000
JO, 000 293,000 345,000
93,000 207,000 300.000
143,000 502,000 645,000
Ground Water Systems
Small Large Total
1,100,000 700,000 1,800,000
88,000 56,000 144,000
167,000 78.000 245,000
225,000 135,000 390,000
AH Systems
4,40(1,000
489,000
546,000
$1,035,000
834
105,399
3,594
12,040
$1,156,867
5.4 Compliance Treatment Forecast
The compliance treatment forecast is the basis of the cost analysis. The compliance treatment forecast is
the culmination of the analysis of systems, their treatment practices, and changes to those practices
required by the Stage 1 DBPR. There are four major categories of systems affected by this rule: large
Stage I DBPR Final RIA
5-4
November 12, 1998
-------
surface water, small surface water, large ground water, and small ground water systems. There is a
different compliance treatment forecast for each category.
This section will review several key contributions to the final Stage 1 DBPR compliance treatment
forecast. First, analyses conducted to support the rule when it was first proposed in 1994 are reviewed in
the light of new data. The compliance treatment forecast for large surface water systems is discussed, as
significant changes in this forecast affect the estimated costs of the rule. Changes in small system
forecasts are also reviewed. Finally, the impact of the enhanced coagulation treatment technique on the
overall compliance treatment forecast is examined. Exhibit 5.5 displays the entire compliance treatment
forecast at the end of this section.
5.4.1 Comparison with Previous Analyses
This section compares the analysis conducted in 1994 (when the rule was first proposed) with the
assumptions about treatment practices and effectiveness that are the basis for the current rule. The
forecast of how many systems must make changes in their treatment practices to comply with the rule
underpins the cost estimates. An extensive cost analysis was prepared for the 1994 RlA, and it is useful
to briefly review how it was developed and how this cost analysis differs.
The 1994 RIA was supported by an elaborate modeling apparatus known as the DBF Regulatory
Analysis Model (DBPRAM). The DBPRAM, which was actually a collection of analytical models, used
Monte Carlo simulation techniques to produce national forecasts of compliance and resulting DBF
exposure reduction for different regulatory scenarios. The model is described in Appendix K.
One of the first activities of the TWG in 1997 was to revisit the modeling tools and re-examine the
results with new assumptions regarding the effectiveness of enhanced coagulation in the presence of
predisinfection. A central component of the DBPRAM apparatus is the Water Treatment Plant model.
Initial investigations concluded that the manner in which predisinfection is characterized in the Water
Treatment Plant model makes it impossible to distinguish the effects of the changes in the Stage 1
DBPR, since the model makes simplifying assumptions about the point of predisinfection and restricts
marginal analysis of shifting this point. In the 1994 analysis, the point of predisinfection did not matter
since the proposal called for elimination of IESWTR credit for predisinfection and the analyses assumed
predisinfection would be eliminated.
The "3-X-3 matrix" (Exhibits 3.5a, 3.5b, and 3.5c) is used to define whether and to what extent systems
must adopt enhanced coagulation, by dividing systems into nine possible categories based on influent
TOC and alkalinity characteristics and identifying removal targets for each category.
The major role of the DBPRAM model in the 1994 RIA was to help verify assumptions for a compliance
treatment forecast. The driving factor in the 1994 RIA became the degree to which water systems would
have to cross over the threshold from standard treatment technologies to more expensive technologies
such as GAC, ozone, chlorine dioxide, and membranes. Focusing on this feature, the M-DBP TWG
designed an approach to re-evaluating the 1994 national cost analysis by re-evaluating the manner in
which newly available information and changes in the proposed rules would affect this advanced
technology threshold in the compliance treatment forecast.
Two sets of data were provided to the TWG that documented levels of TOC, TTHM, HAAS, and
predisinfection practices for groups of water systems. The 1996 Water Industry Data Base (WIDB) data
Stage 1 DBPR Final RIA 5-5 November 12, 1998
-------
set provided data for 308 water systems nationwide. The AWWSCo data set provided 2 years of data
(1991 and 1992) for 52 plants located primarily in the Northeast and Midwest.
Using these two data sets and experience and knowledge of these particular plants, the TWO was able to
undertake a plant-by-plant assessment of the prospective compliance choices of the plants likely to
change treatment practices under the Stage 1 DBPR. By computing the percentage of systems forecast to
require the more expensive advanced treatments, it was possible to see how results compared with the
1994 RIA. This analysis is detailed in the next section.
5.4.2 Compliance Treatment Forecast for Large Surface Water Systems
The review of the previous analysis formed the basis for determining whether the compliance treatment
forecast had changed since 1994 and, if so, in which ways. A sub-group of the M-DBP TWO consisting
of individuals familiar with the 1994 DBPRAM analyses as well as the WIDB and AWWSCo data sets
performed the re-evaluation of the compliance treatment forecast based on the rule changes. They made
case-by-case evaluations of each water system in the data set for which TTHM or HAAS exceeded 64
Aig/L or 48 jug/L, respectively. These numbers are design targets for maximum contaminant levels
(MCLs) of 80 fj.g/L and 60 /ug/L, reflecting the need for utilities to build in an operational safety margin
of 20 percent.
Exhibit 5.4 presents a side-by-side comparison of compliance treatment forecasts for large water systems
developed for the 1994 Stage 1 DBPR RIA, the 1998 Stage 1 DBPR R1A (using 1996 WIDB data), and
the 1991 /92 AWWSCo data.
The compliance treatment forecast developed for the 1994 RIA using the DBPRAM (Column 2 of
Exhibit 5.4) indicates that 17 percent of systems would adopt advanced treatments (ozone, chlorine
dioxide, GAC, or membranes) in order to comply with the Stage 1 DBPR MCLs. In many instances, the
adoption of advanced technologies was forecast as a result of the companion requirements of the
IESWTR to increase disinfection to assure a less than 10"4 (1 in 10,000) annual risk level for giardiasis,
the illness associated with the protozoa Giardia.
Since the 1994 proposal, the IESWTR requirement to achieve a 10"4 risk level for giardiasis has been
replaced with a "disinfection benchmark" requirement intended to preserve the status quo of disinfection
practices. Systems are required under the IESWTR to establish a profile of DBP data prior to a change in
their disinfection practices. As a result, there are expected to be fewer systems (6.5 percent) having to
adopt advanced technologies. In addition, probable compliance choices can be evaluated based on the
existing treatment configuration and performance rather than having to first predict the effects of
changes in disinfection, as was done with the DBPRAM.
Stage 1 DBPR Final RIA 5-6 November 12, 1998
-------
Exhibit 5.4
Comparisons of Large System Compliance Treatment Forecasts
Treatment
(A) No Further Treatment
(B) Chlorine/Chloramines
(C) Enhanced Coagulation +
Chloramines
(D) Enhanced Coagulation +
Chlorine
(E) Ozone, Chlorine Dioxide,
GAC, Membranes
Total*
1998 Stage 1 DBPR RIA
(Analysis 1996 WIDE)
Percent Number
(1)
39.0%
16.6%
19,0%
19.0%
6.5%
100-0%
544
231
265
265
90
1,395
1994 Stage 1 DBPR RIA
Percent
Number
(2)
27.7%
2.9%
9.7%
43.0%
16.6%
100.0%
386
41
136
600
232
1,395
Analysis of AWWSCo
1991-1992 Data
Percent
Number
(3)
22.0%
28.0%
35.0%
15.0%
100.0%
307
391
488
209
1,395
* Detail may not add to total due to independent rounding.
The TWO reviewed the data for 73 of 3081 systems in the 1996 WIDB data set (24 percent) that had
either TTHM > 64 ^g/L or HAAS > 48 ^g/L (Exhibit 5.4, Column 1, Rows B and E). The systems were
evaluated at a plant level, incorporating multiple plant compliance strategies where applicable and other
data such as that available from the 1996 ICR plant schematics (Exhibit 5.4, Column 1, Row E). Based
on the case-by-case analysis of this sample, the TWG predicted that 20 of the 73 systems would require
advanced technologies (ozone, chlorine dioxide, GAC, membranes) in order to comply with the proposed
MCLs. This equates to 6,5 percent (20/308) (Exhibit 5.4, Column 1, Row E), The TWG assigned another
51 systems (16.6 percent) to a compliance category consisting of various combinations of relatively low
cost strategies, such as moving the point of predisinfection and using chloramines (Exhibit 5.4, Column
l,RowB).
5.4.3 Enhanced Coagulation Treatment Technique
The TWG did not forecast of the number of systems in the WIDB data set that would have to install
enhanced coagulation in compliance with the treatment technique portion of the Stage 1 DBPR. Because
several years have passed since the 1992-1993 RegNeg process, it is likely that some water systems have
already moved ahead with implementation of enhanced coagulation. Indeed, it is probably the case that
some systems were achieving enhanced coagulation even before it was given its name during the
RegNeg process. In order to complete a compliance treatment forecast of the final Stage 1 DBPR, it is
necessary to know what proportion of the universe is already achieving enhanced coagulation and what
proportion will have to employ enhanced coagulation.
1 Percentages reported here differ from those computed earlier by members of the TWG due to a correction in the
denominator. Previous calculations used 399 systems as a denominator, but since 91 of them did not report TTHM
or HAAS data, they should not be included in the computations. In addition, a 1998 quality review of the 1996
WIDB conducted after the TWG met resulted in different denominators and numerators in each category (see
Exhibit 4.4), although the results of the analysis are consistent with TWG findings.
Stage 1 DBPR Final RIA
5-7
November 12, 1998
-------
The 1996 WIDE data are the best available source of information from which to develop enhanced
coagulation estimates. For this analysis, large surface water systems served as the universe of total
surface water systems. The 1996 W1DB provides data on influent TOC, effluent TOC, and alkalinity by
plant as well as TTHM and HAAS data by system. Using this information, an assessment of the extent to
which enhanced coagulation is already in place has been developed. The estimates can be developed
through the following sequence of steps and assumptions.
Assumption: It is reasonable to match plant-level TOC with the system-level TTHM and HAA5
data.
Assumption: All compliance targets require an operational buffer. In meeting TTHM and HAAS
MCL targets of 80 /^g/L and 60 Mg/L, utilities are projected to allow a 20 percent buffer
to ensure consistent compliance and will attempt to achieve levels of 64 /-ig/L and 48
Mg/L, respectively. In meeting TOC removal targets, utilities will design their systems
allowing a 15 percent buffer. Thus, the 2.0 mg/L TOC trigger for enhanced coagulation
becomes a design target of 1.7 mg/L.
Step 1: Sort the percent of systems that are below all compliance targets. These systems
meet the compliance requirement under the rule (TTHMs < 64 Mg/L; HAAS < 48 yug/L;
and TOC < 1.7 mg/L) and will not need to take any additional action. The computation
estimates that JO percent of all systems take no action.
Step 2: Calculate the percent of systems that are below both 64 ^g/L and 48 /ug/L, but have
source water TOC > 1.7 mg/L. This is the universe of systems that would not have to
use enhanced coagulation to meet MCLs but that might have to use it to meet the
treatment technique requirements of the rule. This sorting results in an estimate that 62
percent of all systems fall in this category. The next step determines what portion of
these 62 percent are already using enhanced coagulation.
Total Systems: 221
100%
Influent TOC (mg/L)
> 1.7 to < 3.4
3.4to6.8
>6.8
Alkalinity (mg/L)
<60
Systems: 31
14,0%
Systems: 33
14.9%
Systems: 2
0.9%
60 to 120
Systems: 46
20.8%
Systems: 43
19.5%
Systems: 5
2,3%
>120
Systems: 22
10.0%
Systems: 29
13.1%
Systems:
4.5%
Information is estimated from the 1996 W]DB and represents the percentage of the total 221 systems.
Step 3: Sort all systems from Step 2 into the 3-X-3 matrix (above). Calculate the number of systems
in each of the nine cells of the matrix (categorizing by influent TOC and alkalinity) used to
define enhanced coagulation requirements. In this version of the matrix, the number of systems
Stage I DBPR Final RIA
5-8
November 12, 1998
-------
in each row is calculated using a buffer. For example, on the basis of a threshold of 1.7 mg/L
TOC, systems incorporate a 15 percent buffer to consistently meet 2.0mg/L.
Step 4: Further sort each matrix cell to determine the proportion of systems that meet or
do not meet the enhanced coagulation TOC removal target for the cell. Each cell has
a removal target; systems either exceed or fall short of this target. For sorting, each
system's TOC removal (based on the 1996 WIDE) has a 15 percent buffer applied to the
removal value (TOC removal value X 0.85 = TOC removal value with buffer). The
number of systems is tabulated in the 3-X-3 matrix.
Step 5: Calculate the proportion of systems meeting enhanced coagulation requirements as
a percentage of all systems in the 3-X-3 matrix. The proportion of systems meeting
enhanced coagulation TOC removal targets with buffer in each cell of the matrix is
weighted by the proportion of all systems that fall into the cell. These weighted values
are combined across the nine cells to produce a single composite estimate of the
proportion of all systems that already meet the enhanced coagulation TOC removal
targets. This computation produces an estimate of 46 percent (of systems in the 3-X-3
matrix).
Step 6: Apply *ne results from the 3-X-3 matrix analysis to the total universe of systems.
The number of systems that might have to meet enhanced coagulation targets (due solely
to the treatment technique requirement) is 62 percent of all systems (from Step 2). From
Step 5, the proportion of systems that already meet the enhanced coagulation target
removals is 46 percent. Multiplying these two figures together yields an estimate that 29
percent of all systems already meet enhanced coagulation targets and will require no
further compliance action,
Step 7: Add the "no action" systems together. Summing the 10 percent of systems that have
TOC < 1.7 mg/L (from Step 1) to the 29 percent that have TOC > 1.7 mg/L and TTHM
and HAAS < 64 ^g/L and 48 ptg/ L respectively, and that already meet the enhanced
coagulation targets (from Step 6), yields an estimate that 39 percent of systems will have
to take no further action in order to comply with the treatment technique requirement in
the rule.
Step 8: Establish the number of systems that need to comply with the treatment technique.
Add the 39 percent (from Step 7) to the 23.1 percent of systems that were assigned to
compliance choices by the TWG based on their need to meet the 80 ^.gfL and 60 yug/L
MCLs (16.6 percent chlorine/chloramines (CI2/NH2C1)) and 6.5 percent other advanced
technologies), yields a total of 62 percent. That leaves 38 percent unaccounted for. These
38 percent are assumed to be the systems that will have to change treatment and incur
costs in order to meet enhanced coagulation TOC removal targets. As a default
assumption, 50 percent of these systems currently use chlorine as their primary
disinfectant and 50 percent use chloramines.
The resulting compliance treatment forecast for large surface water systems is summarized in Column 1
("Analysis of 1996 WIDE Data") of Exhibit 5.4.
Stage 1 DBPR Final RIA 5-9 November 12, 1998
-------
A parallel case-by-case analysis was performed by members of the M-DBP TWG using the AWWSCo
1991-92 data representing 52 systems and summarized in Column 3 ("Analysis of AWWSCo 1991-1992
Data") of Exhibit 5.4. The results differ and potentially reflect a number of factors: 1) more adverse DBF
control conditions in the waters represented in this data set and 2) a predisposition to chloramines in this
data set.
The full compliance treatment forecast for both surface water and ground water systems is displayed in
Exhibit 5.5.
Exhibit 5.5
Compliance Treatment Forecast by Type of Treatment-
Surface and Ground Water Systems
Treatment
So Further Treatment
CI2/NH2CI
Enhanced Coagulation
EC and CI2/NH2C1
Oz/NH2Cl
EC and Oz/NH2Cl
ECandGACIO
EC and GAC20
Chlorine Dioxide
Membranes
Total*
No Further Treatment
CI2/NH2C1
Oz/NH2Cl
Membranes
Total*
Surface Water Systems
Small Systems < 10,000
# Systems % Systems
1,549 30.0%
826 16.0%
1,983 38.4%
465 9.0%
184 3.6%
0 0.0%
0 0.0%
0 0.0%
0 0.0%
157 3.0%
5,165 100%
Large Systems > 10,000
# Systems % Systems
544 39.0%
232 16.6%
265 19.0%
265 19.0%
29 2.1%
29 2.1%
2 0.2%
2 0.2%
22 1.6%
4 0.3%
1,395 100%
All Systems
# Systems % Systems
1,577 24.0%
1,058 16.1%
2,248 34.3%
730 11.1%
213 . 3.3%
29 0.4%
2 0.0%
22 0.0%
22 0.3%
161 2.5%
6,560 100%
Ground Water Systems
Small Systems < 10,000
# Systems % Systems
59,847 88.0%
5,403 7.9%
0 0.0%
2,921 4.3%
68,171 100%
Large Systems * 10,000
# Systems % Systems
1,122 85.0%
119 9.0%
26 2.0%
53 4.0%
1,320 100%
All Systems
# Systems % Systems
60,969 88.0%
5,522 7.8%
26 0.0%
2,974 4.3%
69,491 100%
* Detail may not add to total due to independent rounding.
5.4.4 Compliance Treatment Forecast for Ground Water Systems
The compliance treatment forecast for ground water systems did not change from the 1994 proposal
because there were no changes in the MCLs, and the enhanced coagulation requirements do not apply to
ground water systems.
Stage 1 DBPR Final RJA
5-10
November 12, 1998
-------
5.4.5 Compliance Treatment Forecast for Small Surface Systems
The changes in the compliance treatment forecast described so far address compliance in large surface
and ground water systems. Small systems face a different set of compliance choices because the current
TTHM standard of 100 /j.g/L does not currently apply to them and they are therefore "starting from
scratch" in applying DBF controls.
Compliance treatment forecasts for small surface water systems are based on the assumption that the
same percentage of small systems would have to resort to expensive, advanced technologies (ozone or
membranes) as predicted for large systems (6.5 percent). This is based on the implicit assumption that
the character of source waters is roughly the same in both small and large systems.
The split between ozone and membranes is about 50:50 overall, but is predicted to differ within the small
system size categories. The very small categories (serving less than 500 people) are projected to rely
primarily on membranes, whereas the larger small categories (serving 500 to 10,000 people) are
predicted to rely much more heavily on ozone. Moreover, a heavier reliance on ozone is predicted in
these size categories than in the 1994 RIA because of the differences in the 1ESWTR. In the 1994
proposal, the IESWTR was projected to require higher levels of inactivation. This, in turn, limited the
applicability of ozone due to concerns for generation of higher levels of biologically assimilable organic
carbon that could exacerbate biofilm problems in distribution systems and due to increased bromate
formation. Since the current IESWTR requires no increase in inactivation levels, these concerns that
previously limited the predicted use of ozone are less constraining.
It is believed that the percentage of small systems that will comply with enhanced coagulation is roughly
comparable to that assumed for large systems (49 percent large versus 47.4 percent small). It is further
believed that about 16 percent of systems will comply with the simple act of adding chloramines and that
30 percent of systems will have to take no compliance action at all.
5.5 Estimated System Costs of the Stage 1 DBPR
The estimated cost of compliance with the provisions of the Stage 1 DBPR is a function of compliance
options from the compliance treatment forecast and the unit costs of each option. The greatest portion of
the costs arises from compliance with the treatment options of the rule. Additional contaminant
monitoring, implementation, and State costs form a smaller portion of the total costs. This section
summarizes the analysis providing a national cost of compliance.
5.5.1 Estimated Cost of Treatment
Exhibit 5.6 displays the annual treatment cost (7 percent cost of capital) associated with the rule.
Estimated costs are broken down in several ways. Costs differ by type of water system being regulated,
in this case small ground water systems, large ground water systems, small surface water systems, and
large surface water systems.
Stage 1 DBPR Final RIA 5-11 November 12, 1998
-------
Exhibit 5.6
Annual Treatment Costs (Capital and O&M) of the Stage 1 DBPR
by Treatment Type (SOOOs at 7 Percent Cost of Capital)
Treatment
Chlorine/Chloramines
Enhanced Coagulation/Chlorine
Enhanced Coagulation/Chloramines
Ozone/Chlonunines
Enhanced Coagulation/Ozone/Chloramines
Enhanced Coagulation/GACIO
Enhanced Coagulation/GAC20
Chlorine Dioxide
Membranes
Total
Surface Water Systems
Small Large
$2,643 $7,482
16,369 90,408
5,323 98,971
10,870 10,924
0 20,917
0 2,623
0 7,272
0 5,722
10,649 19,345
$45,854 $263,664
Ground Water Systems
Small Large
$10,180 $2,153
0 0
0 0
0 6,054
0 0
0 0
0 0
0 0
168,132 96,083
$178,312 $104,290
Total
$22,458
106,777
104,294
27,848
20,917
2,623
7,272
5,722
294,209
$592,120
The estimated costs presented in this RIA incorporate the modified Stage 1 DBPR compliance treatment
forecast. The model permits replication of the cost estimates for water systems by multiplying forecast
percentages by the number of systems within a size category, and then multiplying the result by unit
costs and annual average flows. These costs constitute that portion of total costs attributed to
implementation of treatment options.
The 1994 cost estimates differ from costs estimated in this RIA. Exhibits 5,2 and 5.3 present
corresponding costs estimates completed as part of the 1994 economic analysis. The total annual cost for
surface water systems in the 1994 RIA was $645 million per year in 1992 dollars; for ground water
systems the total was $390 million. Cumulatively, treatment costs equaled $1,035 million. The same
cumulative figure updated to 1998 dollars is $1,168 million.
The 1994 calculations were based on a cost of capital of 10 percent; 1998 figures are calculated at a 7
percent cost of capital. Using the final compliance treatment forecast and applying current assumptions
and unit costs, 1998 costs for treatment equal $327 million for surface water systems and $317 million
for ground water systems at the 10 percent cost of capital. The cumulative total is $644 million, or a
reduction of $391 million over the 1994 figures.
The reduction in costs comes from the review of the compliance treatment forecast and new unit costs in
many categories. One issue is the proportion of the estimated cost in the 1994 economic analysis that was
attributable to enhanced coagulation. While enhanced coagulation by itself is not very expensive in terms
of the cost per household, it can add up to a large sum nationally when it is broadly implemented. In
1994, enhanced coagulation alone accounted for $307 million of the estimated total $728 million (1998
dollars) for surface water treatment cost, or about 42 percent.
Stage I DBPR Final RIA
5-12
November 12, 1998
-------
In comparison, enhanced coagulation in this RJA is responsible for $211 million. Enhanced coagulation
accounts for 36 percent of this estimated total, or 10 percent less than in 1994. Two major factors cause
this drop in estimated costs: 1) halving of the number of systems estimated to employ advanced
technologies (e.g., ozone, GAC), and 2) assuming that some systems have already implemented
enhanced coagulation.
The above compliance treatment forecasts and cost estimates are subject to considerable uncertainty, due
to the difficulty in establishing compliance scenarios. Supporting tables and cost model outputs are
included in Appendices A through D of this document
5.5.2 Estimated Cost of Monitoring and State Implementation
Exhibit 5.1 summarizes the monitoring and State implementation cost estimates. Monitoring costs are
divided into start-up and annual costs. Start-up costs have been annual ized over the same 20-year period
as used for capital cost annualization. Unit cost estimates for the costs of annual sampling were
originally estimated in 1994; these unit costs have been adjusted for inflation to 1998 dollars.
All systems will be required to monitor under the Stage 1 DBPR. Systems will monitor for influent water
quality parameters (TOC and alkalinity), disinfectant residuals (chlorine, chloramines, chlorine dioxide)
and DBFs (TTHM, HAAS, bromate, chlorite). The extensive monitoring described in the rule ensures the
effectiveness of the treatment regime employed. Exhibit 5.7 summarized which systems will have to
perform monitoring for the different contaminants being regulated under the Stage 1 DBPR.
Exhibit 5.7
Monitoring Activities Required to Comply with the Stage 1 DBPR
System Size
(population served)
TOC Routine/Reduced
Alkalinity Routine/Reduced
TTHM Routine/Reduced
HAAS Routine/Reduced
Bromate Routine/Reduced
Chlorite Daily [2]
Chlorite Monthly J2]
Chlorine
Chlorine Dioxide ]2]
Chloramines
Total Costs ($1998) (OOOs)
Small
Surface Water
Systems
/
/
/
/
/
[3}
13]
SI 0,867
Large
Surface Water
Systems
/
/
ill
/
/
/
/
131
/
131
$14,619
Small
Ground Water
Systems
/
/
S
S
$38,803
Large
Ground Water
Systems
m
/
/
s
s
$26,326
[1] Large systems are already monitoring under the 1979 TTHM Rule.
[2] Only required for systems that use chlorine dioxide.
[3] Already monitoring under the Surface Water Treatment Rule.
Stage 1 DBPR Final RIA
5-13
November 12, 1998
-------
Monitoring requirements in the rule specify sites and sampling frequency and serve as the basis for the
cost estimates. The monitoring cost model factors the frequency and number of samples per site, the
number of sites per system, and the time (burden hours) and cost for each sample. The total costs and
number of samples are calculated annually and do not reflect the staging of the requirements for large
and small systems.
This analysis includes several assumptions that, when taken together, conservatively estimate the costs.
Routine monitoring is the base monitoring activity; reduced monitoring can be applied if a system meets
certain water quality targets. This estimate assumes all systems perform routine monitoring in lieu of
estimating the number of systems that could apply at some point for reduced monitoring.
Surface water systems serving fewer than 75,000 people are assumed to have one sampling site per
system; larger systems are assumed to have two sites. For ground water systems, multiple wells drawing
from the same aquifer are considered to be a single plant. Ground water systems with multiple wells
drawing from different aquifers are considered multiple plants. To reflect this, systems serving fewer
than 10,000 people are assumed to have one aquifer, those serving between 10,000 and 50,000 people
one and a half aquifers, for those serving between 50,000 and 100,000 people two aquifers, and for those
serving at least 100,000 people three aquifers.
The full cost calculation and list of assumptions for monitoring are displayed in Appendix E-4.
Stage 1 DBPR Final RIA 5-14 November 12, 1998
-------
5.6 Small System Impacts—Regulatory Flexibility Act Analysis
The Regulatory Flexibility Act provides that if a rule has a significant impact on a substantial number of
small entities, its proposal must be accompanied by a Regulatory Flexibility Analysis (RFA) to be made
available for public comment. Under current policy, EPA regards any impact as a significant impact and
any number of small entities as a substantial number. Thus, a Regulatory Flexibility Analysis is clearly
required for the Stage 1 DBPR. The Regulatory Flexibility Analysis can be incorporated within other
analyses—as is the case here—so long as it is clearly stated how the requirements are being met. The
specific RFA requirements are as follows.
1. Explain why the Agency is considering taking action.
Since most water is not pure enough to be ingested directly from the source, utilities usually apply some
form of contaminant control. Disinfection is one important practice used to meet the public health goal of
providing safe water to the public. Utilities disinfect drinking water supplies by adding chemicals to kill
or inactivate microbial contaminants.
Disinfection, however, poses health risks of its own. Byproducts may result from chemical interactions
between DBF precursors in water and chemical disinfectants in plants and distribution systems of public
water systems. Source water often carries substantial levels of organic material that, when mixed with
disinfectants, form new compounds. Some of these byproducts, including those that are the subject of
this rule (TTHM and FLAA5), are potentially associated with health risks, such as some cancers and
reproductive and developmental effects.
However, because disinfection is effective in reducing microbial contamination, reducing disinfection to
decrease DBFs can increase the risk to the public from microbial contamination. This is known as a
"risk-risk tradeoff."
Due to the inconclusiveness of past scientific research and the "risk-risk tradeoff," the development of
regulations is difficult. However, recent research results concerning the health risks associated with
DBFs supports moving ahead with the development of the Stage 1 DBPR.
While recognizing these uncertainties, EPA believes, for the reasons cited below, that the Stage 1 DBPR
is needed for protection of public health from exposure to DBFs.
1) There is a large population potentially exposed to DBFs in drinking water in the U.S.
2) Since the discovery of chlorination byproducts in drinking water in 1974, numerous
toxicological studies have been conducted that have shown several DBFs to be carcinogenic in
laboratory animals. Some DBFs have also been shown to cause reproductive or developmental
effects in laboratory animals. While many of these studies have been conducted at high doses,
EPA believes the studies provide evidence that DBFs present a potential public health problem
that needs to be addressed.
3) Numerous epidemiology studies have been completed investigating the relationship between
exposure to chlorinated drinking water and cancer. These studies have suggested an association,
albeit small, between bladder, rectal, and colon cancer and exposure to chlorinated drinking
water.
Stage I DBPR Final RIA 5-15 November 12, 1998
-------
4) EPA recognizes there are data deficiencies in the information on the health effects from DBFs
and the levels at which the health effects occur, but believes the weight-of-evidence represented
by the available epidemiological and toxicological studies on DBFs support a potential hazard
concern and warrant regulatory action at this time.
5) Because of the large number of people exposed to DBFs and because of the different risks that
may result from exposure to DBFs, EPA believes the Stage 1 DBPR is needed to further prevent
potential health effects from DBPs,
2. State succinctly the objectives of, and legal basis for, the proposed rule.
To address the complex issues associated with regulating DBPs, EPA launched a rule-making process in
1992 and convened a RegNeg Advisory Committee under the Federal Advisory Committee Act (FACA),
representing a range of stakeholders affected by possible regulation. The RegNeg Committee met
repeatedly over a period of 10 months and arrived at a consensus proposal for taking progressive steps
toward addressing both DBPs and microbial pathogens. The 1992 consensus-building process resulted in
the three following regulatory proposals—
1) A staged approach to regulation of DBPs (referred to as the Stage 1 and Stage 2 DBPRs)
incorporating MCLs, MRLDs, and treatment technique requirements;
2) A companion IESWTR designed to improve control of microbial pathogens and prevent
inadvertent reductions in microbial safety as a result of DBP control efforts, and;
3) An 1CR to collect information necessary to reduce many key uncertainties prior to subsequent
negotiations regarding the Stage 2 DBPR requirements.
In 1997, a similar FACA process was implemented with the Microbial-Disinfectants/Disinfection
Byproducts (M-DBP) Advisory Committee. The M-DBP Committee convened to analyze new data
available since 1994, review previous assumptions made during the RegNeg process, and move the rule
forward on the expedited schedule mandated under the 1996 Amendments to the Safe Drinking Water
Act (SDWA). The efforts of this committee resulted in the drafting of the Stage 1 DBPR.
The Stage 1 DBPR uses a combination of new MCLs, MRDLs, and a treatment technique requirement to
improve control of disinfectants and DBPs. The rule applies to all utilities defined as community or non-
transient/non-eommunity systems that treat their water with a chemical disinfectant. (Community
systems are public water systems that regularly serve at least 25 year-round residents; non-transient/non-
community systems generally include businesses and similar fixed establishments.)
In the Stage 1 DBPR, EPA establishes MCLGs and MCLs for previously unregulated byproducts (except
in the case of TTHMs). EPA is setting MCLGs of 0 for chloroform, bromodichloromethane, bromoform,
bromate, and dichloroacetic acid, and MCLGs of 0.06 mg/L for dibromochloromethane, 0.3 mg/L for
trichloracetie acid, and 0.8 mg/L for chlorite. In addition, EPA is setting MRDLGs for chlorine and
chloramines at 4.0 mg/L and 0.8 mg/L for chlorine dioxide.
The Stage 1 DBPR sets a new, more restrictive MCL for TTHMs at 0.08 mg/L (80 /^g/L). EPA is adding
MCLs for HAAS of 0.06 mg/L (60 ^ug/L), for bromate of 0.01 mg/L, and for chlorite of 1.0 mg/L. In
Stage 1 DBPR Final RIA 5-16 November 12, 1998
-------
addition to these byproduct MCLs, EPA is setting MRDLs for chlorine and ehloramines of 4.0 mg/L and
0.8 mg/L for chlorine dioxide.
EPA identifies several technologies that utilities can use to meet the MCLs and MRDLs. These include
using alternate disinfectants, such as ozone, or alternative treatment practices, such as enhanced
coagulation/enhanced softening or membrane filters.
3. Describe, and where feasible, estimate the number of small entities to which the proposed rule
will apply.
Exhibit 5.8 summarizes the small entities that will be affected by the Stage 1 DBPR. All systems will be
required to perform monitoring activities under the rule, though not all systems will have to modify their
treatment techniques to comply with the MCLs and MRDLs. Of 5,165 small surface water systems, the
majority, 70 percent, will have to modify their treatment techniques. Of 68,171 small ground water
systems, 12 percent will have to modify their treatment techniques.
Most of these small systems required to modify treatment will use ehloramines, a low-cost treatment
technique, to treat their water, though a small portion, 4 percent, will have to treat using higher-cost
membrane technology. While all systems will have to monitor under the rule, these costs generally tend
to be lower than modifying treatment.
4. Describe the projected reporting, recordkeeping, and other compliance requirements of the rule,
including an estimate of the classes of small entities that will be subject to the
requirements and the type of professional skills necessary for preparation of reports or
records,
As previously stated, all small systems will have to perform monitoring under the Stage 1 DBPR.
Monitoring consists primarily of sampling water for precursors, DBFs, and residual chemicals from the
disinfection process. Exhibit 5.9 summarizes the average sampling activities and the estimated burden
for complying with the rule. While ground water systems have to take many more samples to monitor
chlorine and ehloramines, this sampling requires relatively little time compared to TTHM and HAAS
sampling. The analysis of water samples must be conducted at EPA-certified laboratories.
This information is displayed in greater detail in both Appendix E-4 and in the accompanying document,
Information Collection Request for the Stage 1 DBPR.
Stage 1 DBPR Final RIA 5-17 November 12, 1998
-------
Exhibit 5.8
Small Entities Affected by the Stage 1DBPR
System Size
(population served)
Surface water systems
25-100
100-500
500-1,000
1,000-3,300
3,300-10,000
Ground water systems
25-100
100-500
500-1,000
1,000-3,300
3,000-10,000
Totals
Estimated Number of
Systems to Modify
Treatment and Monitor
732
707
592
772
813
3,721
2,800
795
718
290
11,940
Estimated Number of
Systems to Monitor
Only
314
303
254
331
348
26,755
20,134
5,713
5,164
2,081
61,397
Total Number of
Systems
1,046
1,010
845
1,103
1,161
30,476
22,934
6,508
5,882
2,371
73,336
Exhibit 5.9
Summary of Annual Monitoring Activities and Estimated Burden
System Size
(population served)
Surface water systems
25-100
100-500
500-1,000
1,000-3,300
3,300-10,000
Ground water systems
25-100
100-500
500-1,000
1,000-3,300
3,000-10,000
Totals
Estimated Number of
Systems to Monitor
1,046
1,010
845
1,103
1,161
30,476
22,934
6,508
5,882
2,371
73,336
Total Estimated Annual
Samples
37,200
36,800
35,800
46,800
49,200
247,200
250,600
76,900
152,800
150,300
1,083,600
Total Estimated Annual
Burden (hours)
19,000
18,600
18,900
24,700
26,000
71,700
64,700
19,300
31,300
27,400
321,600
Stage 1 DBPR Final RIA
5-18
November 12, 1998
-------
5, Identify, to the extent practicable, all relevant Federal rules that may duplicate, overlap, or
conflict with the proposed rule.
The IESWTR, promulgated concurrently with the Stage 1 DBPR, will further control for microbial
contamination and prevent increases in microbial risk. These rules were developed in tandem since
microbial contamination and disinfection are directly related. Both rules will be promulgated in
November 1998.
The IESWTR is intended to improve control of pathogens such as Cryptosporidium as well as assure no
significant increase in microbial risk as systems act to meet the new DBF MCLs under the Stage 1
DBPR. With the exception of a sanitary survey requirement that applies to all surface water systems, the
IESWTR applies only to public drinking water systems using surface water or ground water under the
direct influence of surface water (GWUD1) as a source, using rapid granulated filtration treatment
technology, and serving 10,000 or more persons.
Major features of the rule include a new MCLG for Cryptosporidium, limitations on turbidity, a
disinfection benchmark and, for all surface water systems or GWUDI systems, a sanitary survey
requirement. In addition, the IESWTR adds Cryptosporidium to the definition of GWUDI and to
watershed control requirements for unfiltered systems, as well as requiring that newly constructed
finished water reservoirs be covered.
The discussion below summarizes the small system impact analysis, regulatory alternatives relevant to
small systems, and impact mitigation measures considered in the RegNeg process.
5.6.1 Small System Impact Quantification
Throughout the rule development process, small systems were defined as those serving fewer than
10,000 people. This definition was used because there is an existing SDWA standard of 100 /ug/L for
TTHMs that applies only to systems serving more than 10,000 people. Surface and ground water systems
serving fewer than 10,000 people are presently unregulated with respect to DBFs. There are, as a result,
two different baseline conditions from which water systems will approach additional DBF control.
The major type of impact is the requirement to install and operate water treatment equipment to meet
specific standards of quality in the delivered water. These requirements pertain primarily to systems that
actually treat or disinfect their water. Systems that purchase treated water from another source may see
an increase in their wholesale costs, but a data base sufficient to track all the wholesale treated water
transactions in the country does not exist. Impacts are therefore evaluated in terms of the systems that
treat water. The data with which to characterize the capacities and flows of these facilities does exist and
provides an adequate basis for assessing total capital and operating costs.
It is estimated that there are a total of 76,051 community and non-transient non-community water
systems that treat water. Of these, an estimated 73,336 (96 percent) serve fewer than 10,000 people.
Despite their overwhelming dominance in terms of industry structure, these systems provide water to
only 22 percent of the total population served by public water supplies.
The 73,336 small system universe consists of 68,171 small ground water systems and 5,165 small
surface water systems. Ground water has historically been inexpensive to develop and has been of
relatively good quality, requiring little treatment for microbial contaminants. This accounts, in part, for
Stage 1 DBPR Final RIA 5-19 November 12, 1998
-------
the proliferation of small ground water systems across the country. Of particular note, most ground
waters have much lower levels of TOC than surface waters and are, therefore, much less susceptible to
DBF formation. Ironically, the lower levels of TOC make TOC removal with coagulation less cost-
effective and may cause systems to have to resort to more expensive technologies of GAC and
membranes if precursor removal is necessary.
Impacts On Small Groiindwater Systems
Of the total 68,171 small groundwater systems, it is estimated that 8,323 (12 percent) will have to modify
treatment to comply with the Stage 1 DBPR. The TWG forecast that 5,403 (8 percent) systems will
comply with the very inexpensive technology of chloramines while 2,921 (4 percent) systems will
require more expensive membrane treatment systems. This will result in $998 million in total capital
costs of treatment.
Impacts On Small Surface Water Systems
Of the 5,165 small surface water systems, it is estimated that 3,616 (70 percent) will have to modify
treatment to comply with the Stage 1 DBPR. The TWG forecast that 3,459 systems (67 percent of the
total) will comply with cost-effective combinations of enhanced coagulation, chloramines, and ozone.
Another 157 (3 percent) systems will require more expensive membrane treatment systems. This will
result in $243 million in total capital costs of treatment.
System operators are assumed to project system upgrades using a least-cost algorithm, one in which
technologies that provide similar levels of protection or compliance are evaluated on a cost basis, with
selection biased towards the least expensive. Unit costs for ozone treatment, for example, exceed those
for membranes in the smallest two size categories. These systems are expected to select membranes for
treatment. Conversely, when ozone costs are more favorable than membranes, systems more heavily
select ozone. The compliance forecast used to estimate costs incorporates this assumption.
The highest portion of small surface water system costs are projected in the largest small system size
category (3,300-10,000 people served). This is due to several factors, including the large absolute
number of systems in this category, the increased system flows, and high membrane costs. Most systems
in this category using an advanced technology are assumed to use ozone, rather than install membranes.
5.6.2 System-Level Impacts on Cash Flow and Viability
Exhibits 5.7 and 5.8 present cash flow impact analyses of the major small system compliance scenarios
described above disaggregated across several size categories of small water systems. The percentage
increase in total operating expenses and the ratio of the increase in operating costs to net operating
revenue are good measures of the impact of a regulation on the cash flow—and therefore on the
economic viability—of a small entity.
There is no hard and fast rule, or threshold, for evaluating these indicators. The major benefit of these
indicators is that they permit a display of the pattern of impacts involved in a regulatory compliance
scenario. It is clear from these indicators that the impacts will be significant where the more expensive
technologies are involved and where the systems are the smallest. It is equally clear that the pattern of
impacts is very unequal; there are many small systems for which the impact of compliance under these
scenarios will be comparatively small, due to better source water with fewer TOC precursors.
Stage 1 DBPR Final R1A 5-20 November 12, 1998
-------
As stated previously, the impact of DBF regulation is inherently a function of unique, site-specific
circumstances. The results in Exhibit 5.10 and 5.11, based on the 1996 Community Water Systems
Survey, reflect this pattern. The Stage 1 DBPR will leave many small ground water systems untouched,
will affect small surface water systems most intensely, and will produce a mixture of relatively low-cost
and relatively high-cost outcomes among-the small systems that will have to modify treatment in order to
comply. The pattern of the impacts results from the raw water characteristics; there is no systematic
aspect of the rule that singles out small systems as compliance targets.
On the surface, the more severe impacts on cash flow illustrated in the exhibits imply potential threats to
the viability of these hardest-hit small water systems. The RFA typically uses such measures to assess
the risk of small business failures resulting from regulatory proposals. In many economic sectors, the
failure of a small business results in the closure of the business. In the drinking water arena, however,
interpretation is not so straightforward. Most often, it is not acceptable to discontinue provision of water
service to a community. The small system segment of the water industry is constantly undergoing
restructuring activity wherein new ownership and institutional arrangements replace old ones. Currently
there is an increased level of such restructuring activity and the projection is that this trend will continue.
The trend results in part from SDWA compliance pressures. There is a wide range of interpretation
regarding whether this trend is beneficial and promising or detrimental.
Exhibit 5.10
Regulatory Flexibility Act Cash Flow Analysis for Small Surface Water Systems ($000)
Systems Complying by C12/NH2CI, Enhanced Coagulation', Oz/NH2CI, or Combination
Pop. per
System
25-100
101-500
500-1 K
1K-3.3K
3.3K-10K
No.
Systems
663
650
583
761
801
Total
Revenue
$10
120
120
304
840
Op.
Exp.
$15
57
103
232
650
Net
Total
Rev.
($5)
63
17
• 72
190
Total
DBF
Cost
$535
1,260
3,499
8,333
21,578
Incr.
DBF
Cost
$1
2
6
11
27
Post-
DBP Op.
Exp.'
$16
59
109
243
677
Post-DBF
Net Rev.
($6)
61
11
61
163
Increase
Op. Exp.
(%)
5,5
3.4
5.9
4.7
4.1
DBF
Cost as
% Net
Rev.
(19.4)
3.1
33.9
15.3
14.2
Systems Complying by Membranes
25-100
101-500
5004 K
1K-3.3K
3.3K-10K
69
57
8
11
12
$10
120
120
304
840
$15
57
103
232
650
($5)
63
17
72
190
$475
1,690
885
2,418
5,182
$7
30
111
220
432
$22
87
214
452
1,082
($12)
33
(94)
(148)
(242)
46.7
52.6
107.8
94.8
66.5
(140.0)
47.6
652.9
305.5
66.5
Stage 1 DBPR Final R1A
5-21
November 12, 1998
-------
Exhibit 5.11
Regulatory Flexibility Act Cash Flow Analysis for Small Ground Water Systems
Systems Complying by C12/NH2C1, Enhanced Coagulation, Oz/NH2Cl, or Combination
Pop. per
System
25-100
101-500
SOO-1K
1K-3.3K
3.3K-10K
No,
Systems
2,415
1,818
516
466
188
Total
Revenue
$22
57
84
194
617
Op.
Exp.
$2
26
72
170
509
Net Total
Rev.
$20
31
12
24
108
Total
DBF Cost
$3,552
3,027
1,043
1,105
1,453
Incr.
DBF
Cost
$1
2
2
2
8
Post-
DBF Op,
Exp.
$3
28
74
172
517
Post-
DBF Net
Rev.
$19
29
10
22
100
Increase
Op. Exp.
(%)
50.0
7.7
2.8
1.2
1.6
DBF
Cost as
% Net
Rev.
5.0
6.5
16.7
8.3
7.4
Systems Complying by Membranes
25-100
101-500
500-1K
1K-3.3K
3.3K-10K
1,306
983
279
252
102
$22
57
84
194
617
$2
26
72
170
509
$20
31
12
24
108
$8,989
29,361
29,200
55,240
45,341
$7
30
105
219
445
$9
56
177
389
954
$13
1
(93)
(195)
(337)
350.0
115.4
145.8
128.8
87.4
3S.O
96.8
875.0
912.5
412.0
The issue of regulatory impacts as a contributor to restructuring trends was considered in evaluating
smail system impacts. A preliminary issue discussed by the TWO concerned the fact that the cost
forecasts are based upon the assumption that small systems facing a regulatory requirement to install
expensive treatments have no choice but to do so. This ignores the role of prospective compliance costs
as incentives that may promote alternative choices by water systems and market responses by equipment
and service providers.
Faced with extreme costs, many small systems may elect to connect to a another water system nearby.
Half of all small water systems are located within metropolitan regions where distances between water
systems may not present a prohibitive barrier, given the cost of the alternative. Small systems driven to
consolidation as a result of these regulations will do so primarily because it is their least-cost approach to
compliance. The impact estimates that assume comprehensive installation of treatment equipment on a
small scale may also be an overestimate of the actual impact.
The prospective expenditures for DBF control are of such magnitude that a market response from
manufacturers and service providers seems likely. Increased demand for O&M services could create
broader markets for contract O&M services and allow for these services to deliver at lower cost.
Although least-cost optimization by water systems and cost-reducing market adaptations by suppliers
may place downward pressures on costs, the cost estimates need to first be developed without dilution by
cost-minimizing assumptions in order to give the correct signal to trigger these very reactions at the start.
These trends and market forecasts are, therefore, not included in the cost estimates.
Stage 1 DBPR Final RIA
5-22
November 12, 1998
-------
There are differing views regarding induced restructuring of small systems. On one hand, there are
thousands of small water systems that are, by the usual financial measures, very viable enterprises under
current circumstances. These systems would be forced to entertain notions of consolidation or other
forms of restructuring only as a result of regulations. Small system advocates note correctly that all of
the above impact analysis considers the impact of only one set of regulations on these small entities,
whereas the total force behind small system failure and/or restructuring is a result of the cumulative
impact of all SDWA regulations. Notably, the most significantly affected category of systems—those
treating surface water—are presently faced with the impacts of the Surface Water Treatment Rule and
will have to also face the impacts of an Interim Enhanced Surface Water Treatment Rule. These impacts
are not included in the exhibits.
Conversely, it is also true that there are thousands of small water systems that are persistent violators of
current drinking water standards—accounting for 90 percent of all violations, including violations of
fundamental protective requirements such as the coliform standard. These persistent violations may, in
effect, indicate that business failure has already occurred in these systems in terms of performance.
Experience shows that when financial data are available for such systems, they confirm that compliance
default and financial default are correlated.
Public health officials have long been concerned about the ability of small systems to adequately treat
surface water. Increased knowledge of microbial risks and recent experiences with outbreaks of
waterborne disease have strengthened these convictions. That some small surface water systems are
likely candidates for restructuring and likely to show significant impacts from DBF regulation is not a
surprise. It is a manifestation of a larger trend stemming from new understandings in this area of public
health protection.
Considering this broader context, the assessment of the net effects of SDWA-induced restructuring enters
a grey area. Some of this restructuring activity must be attributed to an inevitable baseline change in the
business environment of the industry. The change happens to involve water quality issues that are the
subject of regulation. Because regulation is imperfect, there is some potential to force more than an
optimal level of restructuring through over-regulation, but it is not clear where that line should be drawn.
In addition, it is possible to make regulatory decisions that are underprotective, in which case the costs
are borne in terms of health effects. The level that is optimal, or most cost-effective, for the nation as a
whole may not be optimal for all systems affected, especially when there are broad differences in costs
resulting from differences in the scale of operations. A potential solution to this paradox is to consider
the alternative of having different standards of protection for large systems versus small systems. This
option was proposed and considered in the RegNeg deliberations, as described in the next section.
5.6.3 Small System Regulatory Alternatives
One of several participants in the RegNeg process representing the interests of small water systems
disagreed with the consensus proposal that had been developed for the Stage 1 DBPR and subsequently
withdrew. The issues that led to the withdrawal were manifest in the discussion of several regulatory
alternatives. Three alternatives were involved.
*• Option 1: TTHM standard of 100 /^g/L for systems serving fewer than 10,000 people; and
TTHM standard of 80 /ug/L and RAA5 standard of 60 /ug/L for systems
serving more than 10,000 people.
Stage 1 DBPR Final RIA 5-23 November 12, 1998
-------
>• 80/60/4 Opt.: TTHM and HAAS standards of 80 //g/L and 60 /ug/L with an additional standard
of 4.0 mg/L for TOC; applicable to all water systems.
> Final Rule: TTHM and HAAS standards of 80 /^g/L and 60 ,ug/L coupled with a
treatment technique requirement mandating enhanced coagulation in systems
with influent TOC greater than 2.0 mg/L.
The essence of the Stage 1 DBPR is to apply the most inexpensive form of TOC removal — enhanced
coagulation— to a large segment of surface water systems, yielding virtually the same exposure profiles
as if the more expensive technologies were applied to a smaller segment of the surface water universe at
the extreme end of the spectrum of water quality conditions. The rule is much more cost-effective due to
reliance on the less-expensive technology and also because it considers a combination of technologies
(EPA, 1994).
The difference in the cost estimates for the different options results, in part, from the cost-effectiveness
of the treatment technique approach and also from methodological changes that consider combinations of
technologies.
There were several other issues discussed concerning this option as an alternative to the Stage 1 DBPR.
These are briefly summarized as follows.
> There was some support for a TTHM standard of 1 00 /4g/L for small systems as a reasonable
first step. This will enable small systems to draw on the experience of larger systems in meeting
this level. Going to lower levels of TTHMs, extending regulation to haloacetic acids, and
introducing TOC removal were seen as steps into more foreign territory. On this basis, a TTHM
standard of 100 /ug/L for small systems was supported by some as a better interim step.
* The argument favoring the rule as proposed over the alternate proposal discussed above
consisted of at least three points: 1) that cost impacts would not be significantly different, as
outlined above; 2) that the experience with the current TTHM MCL of 100 Mg/L suggests that a
head-long rush to chloramines and possible compromising of microbial protection could result
from simple extension of this standard to small systems; and, 3) that since a best available
technologies concept that begins with TOC removal through enhanced coagulation is the
direction in which the later regulations may be headed, it might be a waste of small system
resources to structure Stage 1 in a manner that encourages them to select different and
potentially incompatible control strategies.
> There was some discussion of the issue of creating a double standard of public health protection
by having a less stringent standard for small systems. Some small system negotiators made the
point that unaffordable regulations that push small systems into non-compliance also result in a
double standard of public health protection. However, the alternative of a small system TTHM
MCL of 100 Mg/L was considered only in the context of an interim Stage 1 DBPR target,
In summary, the debate over the Stage 1 DBPR covered two of the important categories of alternatives to
be considered in an RFA: performance versus design standards, and relaxed standards for small entities.
In traditional effluent-oriented, or externality-oriented, areas of environmental economics, performance
standards are often more economically efficient than design standards, and relaxed standards for small
Stage 1 DBPR Final RIA 5-24 November 12, 1998
-------
entities may be economically justifiable. In this application to protection of drinking water supplies, all
the reverse conclusions hold.
The treatment technique approach is shown to be more cost-effective than the MCL approach for TOC
removal. Moreover, an MCL approach to byproduct control could trigger compliance choices that might
have to be reversed by subsequent regulations.
5.6.4 Small System Impact Mitigation
Two major strategies for mitigation of impacts on small systems were considered by the RegNeg
Committee: 1) extended timetables for compliance, and 2) variances and exemptions. These are
discussed in this section.
Extended Compliance Timetable
The Stage 1 DBPR incorporates an extended timetable for small system compliance. While the
compliance date of the rule requirements for large surface water systems is in 2001, compliance by small
surface water systems is not required until 2003. The compliance date for all ground water systems is
2003. The extended timeframe for smaller systems to achieve compliance is to allow for simultaneous
compliance with the Long Term 1 Enhanced Surface Water Treatment Rule (LT1) and the Ground Water
Disinfection Rule and their associated compliance deadlines. Additionally, it allows small systems to
wait for the capital improvements that will come from implementation in large systems and a lower cost
of implementation given a competitive market for the technologies used.
Variances and Exemptions
There was extensive discussion of the prospects for small system relief through exercise of the variance
and exemption provisions of the SDWA.
5.7 Combined Effect of the Stage 1 DBPR and the IESWTR
Because the Stage 1 DBPR and IESWTR were developed in tandem to address the risks of DBFs while
not compromising protection against microbial contaminants, it is important to examine the combined
effects of both rules as well as those expected to be implemented in the next several years.
While the IESWTR may impose additional costs to large surface water systems beyond those described
in this chapter for the Stage 1 DBPR, these systems may see greater benefits as well. The anticipated
impact of both rules at a 7 percent cost of capital is summarized in Exhibit 5.12.
Stage 1 DBPR Final R1A 5-25 November 12, 1998
-------
Exhibit 5.12
Cost Impact of Current and Expected Rule-Makings
System Types
Small Surface Water
Large Surface Water
Small Ground Water
Large Ground Water
Subtotal
States
Totals
Current and Expected Rules
D/DBP
Stage I ($000)
$56,804 .
278,321
218,062
130,651
$ 683,838
17,342
$701,180
Interim
ESWT ($000)
$0
291,165
0
0
$291,165
15,556
S 306,721
Other
Rule-makings Planned
Stage 2 DBPR
Long-term ESWTR 1 (LT1)
Stage 2 DBPR
Long-term ESWT 2 (LT2)
Stage 2 DBPR
Ground Water Disinfection
Stage 2 DBPR
Ground Water Disinfection
Stage 1 DBPR Final RIA
5-26
November 12, 1998
-------
6: Assessing Net Benefits of the Stage 1 DBPR
6.1 Alternative Approaches for Assessing Benefits of the Stage 1 DBPR
In light of the scientific uncertainties surrounding the risk estimates and limited data availability, EPA
explored several alternative approaches to assessing the net benefits of the Stage 1 DBPR:
* Overlap of Benefit and Cost Estimates;
»• Minimizing Total Social Losses;
*• Breakeven Analysis;
* Household Costs; and
>• Decision-Analytic Model.
6.2 Overlap of Benefit and Cost Estimates
One method to characterize net benefits is to compare the relative ranges of benefits and costs.
Conceptually, an overlap analysis tests whether there is enough of an overlap between the range of
benefits and the range of costs for there to be a reasonable likelihood that benefits will exceed costs. In a
theoretical case where the high end of the range of benefits estimates does not overlap the low end of the
range of cost estimates, a rule would be difficult to justify based on traditional benefit/cost rationale
(although it may be based on law).
For the Stage 1 DBPR, the two overlap analyses show that there is substantial overlap in the estimates of
benefits and costs (Exhibit 6. la and 6.1b). The exhibits portray the range of quantified benefits extending
from zero to over $4 billion.1 The zero end of the range of estimated benefits represents the possibility
that there is essentially no health benefit from reducing exposure to DBPs. The $4 billion end of the
range assumes a 24 percent reduction in the incidence of 9,300 annual bladder cancer cases attributable
to DBPs. The high end of the benefits range could potentially exceed $4 billion by a large amount if the
non-quantified benefits are included.
1 Refer to Section 4.11.
Stage 1 DBPR Final RIA 6-1 November 12, 1998
-------
Exhibit 6,1 a Conceptual Overlap of the Estimated Benefits and Costs of the Stage 1 DBPR
.-"
•i.'-V >;. '»'••„•,' ""•• , -j
|
'
";
Exhibit 6.1b Overlap of the Ranges of the Estimated Benefits and Costs
of the Stage 1 DBPR
1
| (Range of Estimated Co
! 1
1 !
1 '' I
| | |Ra«8
1 !
I !, ,
Kange 01 tsiimaieu jaeneii
its
: of Estimated Net Benefits
1 II
-2-10 1 2 3 45
$ Billions
The range of cost estimates is significantly smaller, ranging from $500 million to $900 million annually
(based on the central tendency estimate plus or minus the standard deviation). Although cost estimates
have uncertainty, this degree of uncertainty is of little consequence to the decisions being made given the
scale of the uncertainty in the benefits.
Interpreting the Results
The overlap analysis shows that there is substantial basis for the Stage 1 DBPR. Benefits exceed the
costs over a wide range of the possible estimates of each. More detail on the nature of this overlap is
provided in the Breakeven Analysis later in this chapter.
Stage 1 DBPR Final RIA
6-2
November 12, 1998
-------
6.3 Minimizing Total Social Losses Analysis
Minimizing Total Social Losses analysis, sometimes called "minimizing regrets" analysis, is a decision-
aiding tool that is suited for use in situations where it is impossible to pin down the exact nature and
extent of a risk. The basic premise of Minimizing Total Social Losses analysis is to estimate total social
costs for policy alternatives over a range of plausible risk scenarios. The actual, or "true" risk is
unknowable, so instead this analysis asks what range and levej of risks could be true, and then evaluates
the total costs to society if that particular risk turns out to be the "true" value. Total social costs include
both the cost to implement the policy option, plus costs related to residual (i.e., remaining) health
damages at that risk level after implementation of the policy option.
The decision-maker compares the total costs of the policy alternatives for each potential risk scenario.
The policy alternative with the lowest total costs within a risk scenario represents the "minimal loss"
alternative for that scenario. That is, if that risk scenario turns out to be true, the decision-maker has
made the right decision, the one that minimizes costs. If the decision-maker has chosen one of the other
policy options and that risk scenario turns out to be true, you have incurred "losses" in the form of the
social costs higher than the lowest-cost alternative. These potential losses are calculated for a range of
potential risk scenarios. The decision-maker can then evaluate the policy options based on the potential
losses if he or she guesses wrong, trying to avoid the greatest potential losses across the range of risk
scenarios.
The following discussion defines the steps in developing a minimizing social losses analysis for the
Stage 1 DBPR.
Step 1: Define the policy options to be considered
The Regulator)'Negotiation (RegNeg) Committee considered a range of alternatives as described in
Chapter 3, before settling on a staged-rule approach. For the purposes of this analysis, the No Action
(i.e., leaving the current DBF regulatory requirements as they are) is compared against the Stage 1 DBPR
alternative and a stronger policy intervention option as represented by the RegNeg placeholder
provisions for the Stage 2 DBPR.
Step 2: Define the range of plausible risk scenarios to evaluate
Based on the toxicological data, 1 to 100 cancer cases attributable to DBPs was chosen as the low end of
the plausible risk range with 10,000 cases (based on the meta-analysis and PAR analysis) as the high end
of the plausible risk range. To illustrate break points and trends within the 1 to 10,000 range,
intermediate points of 1,000, 2,500, 5,000, and 7,500 cancer cases are also calculated.
Step 3: Define Implementation Costs of Stage 1 DBPR for policy alternatives
In the case of the No Action alternative, the total implementation costs of the Stage 1 DBPR are $0. The
revised costs described in Chapter 5 are used as the total implementation costs for Stage 1 ($701.18
million annual cost). The implementation cost estimates for the stronger intervention alternative have not
been revised since the 1994 RIA. For the purposes of this analysis, the Stage 2 placeholder costs from the
1994 RIA have been updated to a 1998 price level to $2.892 billion and are used as an estimate for the
stronger intervention option (EPA, 1994). These estimates are used in Exhibit 6.2.
Stage 1 DBPR Final RIA 6-3 November 12, 1998
-------
Step 4: Estimate Residual Health Damages
The residual health damages are the number of cancer cases remaining after the implementation of each
action. After the implementation of the "No Action" alternative, obviously all of the potential cancer
cases remain. The cancer cases are expected to be reduced, however, by implementation of the Stage 1
DBPR and the stronger intervention alternative. The mean reduction in DBFs (as estimated using data for
TTHMs as a proxy for all DBFs) estimated in the exposure reduction analysis (see Appendix G) is 24
percent for Stage 1 and 40 percent for Stage 2 (see Appendix G-4). Exhibits 6.3a, 6.3b, and 6.3c contain
the estimated residual number of cancer cases for each risk scenario and alternative action.
Step 5: Monetize Residual Health Damages
Health costs associated with the residual cancer cases then need to be calculated. Based on current
fatality rates, it is assumed that 77 percent of bladder cancer cases are nonfatal, valued at the willingness-
to-pay (WTP) to avoid a cancer case, and 23 percent of the cases result in fatality, valued at the value per
statistical life (VSL).
Stage 1 DBPR Final RIA 6-4 November 12, 1998
-------
I
§'
Exhibit 6.2
Stage 1 DBPR Minimizing Total Social Costs Analysis
(Billions of Dollars, 1998 Price Level)
No Action
Cost of DBP Rule Option
Residual Health Costs 1
Total Social Costs
Excess Social Losses
Stage 1
Cost of DBP Rule Option
Residual Health Costs 1r2
Total Social Costs
Excess Social Losses
Strong Intervention
Reg N eg Stage 2 Placeholder)
Cost of DBP Rule Option
Residual Health Costs 1'3
total Social Costs
Excess Social Losses
<1 Cancer
Case
$ 0
$ 0
$ 0
$ 0
$ 0.701
$ 0
$ 0,701
$ 0.701
$ 2.892
$ 0
$ 2.892
1 2.892
100 Cancer
Cases
$ 0
S 0.176
$ 0.176
$ 0
$ 0.701
$ 0.131
$ 0.832
$ 0.656
$ 2.892
$ 0.106
$ 2.998
$ 2.823
1,000
Cancer
Cases
$ 0
$ 1.755
$ 1.755
$ 0
$ 0.701
$ 1.335
$ 2.036
$ 0.281
$ 2.892
$ 1.053'
$ 3.945
$ 2.190
Risk Scenarios
2,500
Cancer
Cases
$ 0
$ 4.388
$ 4.388
$ 0.352
$ 0.701
$ 3.335
$ 4.036
$ 0
$ ••• 2.892
• $ 2.633
$ - 5.525
$ 1.489
>
5,000
Cancer
Cases
$ 0
$ 8.776
$ 8.776
$ 1.405
$ 0.701
S 6.670
$ • 7.371
$ 0
$ 2.892
$ 5.266
$ 8.158
$ 0.787
7,500
Cancer
Cases
$ 0
$ 13.164
$ 13.164
$ 2.458
$ 0.701
$ 10.005
$ 10.706
$ 0
$ 2.892
$ 7.899
$ 10.791
$ 0.085
10,000
Cancer
Cases
$ 0
$ 17.552
$ 17.552
$ 4.129
$ 0.701
$ 13,340
$ 14,041
$ 0.617
T"_8i92"j
$ 10.531
$ 13,423
$ 0
1 Mean values from Monte Carlo Simulation
2 Assumes 24 percent reduction in exposure (see Appendix G)
3 Assumes 40 percent reduction in exposure (see Appendix G)
C
-------
Exhibit 6.3a
Residual Damages Monte Carlo Simulation Summary: No
(Billions of Dollars)
Mo Action
Percent Reduction
Total Cancer Cases Avoided
Number of Fatal Residual
Cancer Cases
Number of Nonfatal Residual Cancer
Cases
Total Cases
Mean Dollar Damages from
Fatal Residual Cancer Cases
Mean Dollar Damages from Nonfatal
Residual Cancer Cases
Total Dollars
Calculation
A
I!
C
= (A x B)
D
= ((A - C) x 23%)
E
= ((A - C) x 77%)
F
= (D + E)
G
= D x lognormal dist.,
mean $0.5600 bil.
H
- E x dist., median
$0.0006 bil.
I
= G Dist. + H Dist.
Action Scenario
Baseline Cancer Cases
100 !,000
0% 0%
0 0
23 230
77 770
100 1,000
$0.130 $1.302
$ 0.045 $ 0.454
$0.176 $1.755
2,500
0%
0
575
1,925
2,500
$3.254
$1.134
$4.388
5,000
0%
0
1,150
3,850
5,000
$ 6.508
$ 2.268
$ 8.776
7,500
0%
0
1,725
5,775
7,500
$9.762
$ 3.402
$13.164
10,000
0%
0
2,300
7,700
1.0,000
$ 13.016
$4.536
$ 17.552
NOTE: Mean values resulting from the Monte Carlo simulation may not precisely match mathematically derived values.
Exhibit 6.3b
Residual Damages Monte Carlo Simulation Summary:
(Billions of Dollars)
No Action
Percent Reduction
Total Cancer Cases Avoided
Number of Fatal Residual
Cancer Cases
Number of Nonfatal Residual Cancer
Cases
Total Cases
Mean Dollar Damages from
Fatal Residual Cancer Cases
Mean Dollar Damages from Nonfatal
Residual Cancer Cases
Total Dollars
Calculation
A
B
C
= (A x B)
D
= ((A - C) x 23%)
E
= ((A - C) x 77%)
F
G
= D x lognormal
dist., mean $0.5600
bil.
H
= E x dist., median
$0.0006 bil.
I
= G Dist + H Dist.
Stage 1 DBPR
Baseline Cancer Cases
100
24%
24
17
• 59
76
$ 0.096 $
1,000
24%
240
175
585
760
0.990
$ 0.035 $ 0.345
$0.131 $
1.335
2,500
24%
600
437
1,463
1,900
$ 2.473
- $0.862
$3.335
5,000 7,500
24% 24%
1,200 1,800
874 1,311
2,926 4,839
3,800 5,700
$4.946 $7.419
$1.724 $2.585
$ 6.670 $ 10.005
10,000
24%
2,400
1,748
5,852
7,600
$ 9.892
$ 3.447
$ 13.340
NOTE: Mean values resulting from the Monte Carlo simulation may not precisely match mathematically derived values.
Stage 1 DBPR Final RIA
6-6
November 12, 1998
-------
Exhibit 6.3c
Residual Damages Monte Carlo Simulation Summary: Strong
(Billions of Dollars)
Mo Action
Percent Reduction
Total Cancer Cases Avoided
Number of Fatal Residual
Cancer Cases
Number of Nonfatal Residual Cancer
Cases
Total Cases
Mean Dollar Damages from
Fatal Residual Cancer Cases
Mean Dollar Damages from Nonfatal
Residual Cancer Cases
Total Dollars
Calculation
A
B
C
= (AxB)
D
- ((A - C) x 23%)
E
= ((A - C) x 77%)
F
= (D + E)
G
= D x lognormal
dist., mean $0.5600
bil.
H
= E x dist., median
$0,0006 bil.
I
= G Dist. + H Dist.
Intervention
Baseline Cancer Cases
2,500
40%
40
. 14
46
60
$ 0.079
$ 0.027
$0.106
1,000
40%
400
138
462
600
$0.781
$ 0.272
$ 1.053
2,500
40%
1,000
345
1,155
1,500
$1.952 1
5,000
40%
2,000
690
2,310
3,000
» 3.905
$0.680 $1.361
$ 2.633
£ 5.266
7,500
40%
3,000
1,035
3,465
4,500
$ 5.857
$2.041
$ 7.899
10,000
40%
4,000
1,380
4,620
6,000
$7.810
$ 2.722
$10.53!
NOTE: Mean values resulting from the Monte Carlo simulation may not precisely match mathematically derived values.
Cancer cases that end in fatality are valued at a VSL estimate derived from previous EPA efforts. The
result is a distribution of values represented by a lognormal distribution with a mean of $5.6 million and
standard deviation of $3.16 million (updated to current price level). Nonfatal cancer cases can be valued
as the WTP to avoid a nonfatal case of bladder cancer or the less complete concept of the cost of the
illness (COI—treatment costs and lost productivity). A WTP value specifically for avoiding a case of
bladder cancer was not available in the current literature. As a substitute, a distribution of WTP values
derived from a study of chronic bronchitis is used in this analysis (mean value of $587,500 per nonfatal
case). Please refer to Section 4.10 and Appendix H for a complete explanation of the derivation of the
monetary values for the health endpoints.
The monetary values of the residual health damages are calculated using a Monte Carlo simulation
drawing from the distributions for the VSL and WTP to avoid nonfatal bladder cancer. The resulting
mean values of the simulation can be found in Exhibits 6.3a, 6.3b, and 6.3c. Appendix I contains the full
results of the Monte Carlo simulation.
Step 6: Calculate Total Social Costs
The total social costs for each alternative and plausible risk scenario are calculated by adding the cost of
the rule option and the residual health costs and are presented in Exhibit 6.2 (page 6-4).
Stage J DBPR Final RIA
6-7
November 12, 1998
-------
Step 7: Determine least-cost alternative and social losses
Looking down each column in Exhibit 6.2, the policy option (i.e., row) with the lowest total social costs
in that column is. the "least-cost" option. In other words, if the risk estimate for that column turns out to
be the "true" risk, the option with the lowest total social costs is the "right" choice (i.e., the one that
minimizes costs). Each of the other two policy options incur excess costs, referred to as social losses. To
calculate the social loss associated with a given policy option, simply subtract the cost of the least-cost
option from the cost of the other option.
In Exhibit 6.1, the least cost option for each risk scenario is indicated by the grey border. Taking the
example of 5,000 cancer cases, the least cost option at $7.371 billion total social costs is the Stage 1
DBPR. If the number of cancer cases attributable to DBFs turns out to be 5,000 per year, then choosing
the Stage 1 DBPR option is the correct choice, the one that minimizes costs to society. Choosing the No
Action alternative results in social losses of $1.405 billion ($8.776-$7.371) and choosing the Strong
Intervention results in social losses of $0.787 billion ($8.158-S7.371).
Interpreting the Results
If there were perfect information on which risk scenario is closest to the "true risk" attributable to DBFs,
the choice of options would be easily apparent from Exhibit 6.2, Even if there were information that
allowed the assigning of probabilities to each risk scenario, a calculation of expected values could
identify the least-cost option based on the probability that each risk scenario might be true. In the case of
cancer and DBFs in drinking water, the state of the science does not currently allow conclusively
choosing a risk scenario (column) or accurately assigning probabilities. Fortunately, further analysis can
help identify preferences among the policy options (rows).
Minimizing Maximum Losses
When it is impossible to narrow the range of plausible risks, decision theory suggests using an approach
that minimizes the maximum loss (Hillier and Lieberman, 1990). In plain English, the option that cuts
losses and minimizes downside risk should be preferred. To do this, a look across the rows in Exhibit 6.4
identifies the value of the largest potential social loss in that row. For the No Action alternative, the
largest potential Joss is at the 10,000 cancer case risk scenario ($4,129 billion). For the Stage 1
alternative, the largest potential loss is at the less than 1 cancer case risk level ($0.701 billion). For
Strong Intervention, the largest loss is also at the less than 1 cancer case level ($2.892 billion).
With the Stage 1 DBPR option, the largest possible loss is $0.701 billion, but the largest possible loss is
over 4 times as much with Strong Intervention ($3 billion) and almost 6 times as much with No Action
($4 billion). In light of the scientific uncertainty regarding risk, choosing Stage 1 minimizes the
maximum loss.
The 1994 RegNeg and 1997 M-DBP Committees implicitly applied this type of "minimizing maximum
loss" framework when developing and evaluating the DBF regulatory options. The RegNeg and M-DBP
Committees recognized that they could not narrow the potential range of cancer risk (1 to 10,000 cases)
or develop a central tendency for the risk. Instead, they developed a regulatory option (Stage 1) that
minimizes the maximum potential loss across the range of risks.
Stage 1 DBPR Final RIA 6-8 November 12, 1998
-------
a
a
I
Exhibit 6.4
Stage 1 DBPR Minimizing Maximum Loss Analysis
(Billions of Dollars, 1998 Price Level)
No Action
Cost of DBF Rule Option
Residual Health Costs 1
Total Social Costs
Excess Social Losses
Stage 1
Cost of DBP Rule Option
Residual Health Costs1'2
Total Social Costs
Excess Social Losses
Strong Intervention
RegNeg Stage 2 Placeholder)
Cost of DBP Rule Option
Residual Health Costs 1% 3
Total Social Costs
Excess Social Losses
Risk Scenarios
<1 Cancer
Case
$ 0
$ 0
$ 0
$ 0
$ 0.701
$ 0
$ 0.701
$ 0.701
$ 2.892
$ 0
$ 2.892
$ 2.892
100 Cancer
Cases
$ 0
$ 0.176
$ 0.176
$ 0
$ 0.701
$ 0.131
$ 0.832
$ 0.656
$ 2.892
$ 0.106
$ 2.998
$ 2.823
1,000
Cancer
Cases
$ 0
$ 1.755
$ 1.755
$ 0
$ 0.701
$ 1.335
$ 2.036
$ 0.281
$ 2.892
$ 1.053
$ 3.945
$ 2.190
2,500
Cancer
Cases
$ 0
$ 4.388
$ 4.388
| 0.352
$ 0.701
$ 3.335
$ 4.036
$ 0
$ 2.892
$ 2.633
$ 5.525
$ 1.489
5,000
Cancer
Cases
$ 0
$ 8.776
$ 8.776
$ 1.405
$ 0.701
$ 6.670
$ 7.371
$ 0
$ 2.892
$ 5.266
$ 8.158
$ 0.787
7,500
Cancer
Cases
$ 0
$ 13.164
$ ,13.164
$ 2.458
$ 0,701
$ 10.005
$ 10.706
$. 0
$ 2.892
$ 7.899
$ 10.791
$ 0.085
10,000
Cancer
Cases
,_— -— |
$ 17.552 j
$ 17.552 j
J$ 4£29j
$ 0.701
$ 13,340
$ 14.041
$ 0.617
$ 2.892
$ 10.531
$ 13.423
$ 0
f
Oo
1 Mean values from Monte Carlo Simulation
2 Assumes 24 percent reduction in exposure (see Appendix G)
3 Assumes 40 percent reduction in exposure (see Appendix G)
Gray border represents maximum excess social loss for each alternative action (row).
STAGE 1 MINIMIZES THE MAXIMUM EXCESS SOCIAL LOSS.
-------
6.4 Breakeven Analysis
Breakeven analysis represents another approach to assessing the benefits of the Stage 1 DBPR given the
scientific uncertainties. Breakeven is a standard benchmark of cost effectiveness and economic
efficiency and is essentially the point where the benefits of the Stage 1 DBPR are equal to the costs.
Normally, the benefits and costs of an option are calculated separately and then compared to assess
whether and by what amount benefits exceed costs. In the case of the Stage 1 DBPR, independently
estimating benefits is difficult, if not impossible, because of the 10,000-fold uncertainty surrounding the
risk. Instead, the breakeven analysis works backwards from those variables that are less uncertain. In this
case, implementation costs for the rule and the monetary value associated with the health endpoints are
used to calculate what baseline risk and risk reduction estimates are needed for the rule to just pay for
itself in avoided health damages associated with bladder cancer.
The first step in the breakeven analysis is to calculate the number of bladder cancer cases that would
need to be avoided for the benefits of avoiding those cases to be equal to the cost of the rule. The simple
calculation is to divide the annual costs of the rule ($701.18 million) by the value per cancer case to
derive the number of cancer cases needed to cover the costs of the rule. The value of a cancer case differs
based on whether the cancer case ends in fatality. Fatal cancer cases are valued at the VSL with a mean
of $5.6 million and a standard deviation of $3.16 million, as mentioned earlier and described in depth in
Appendix H. It is assumed that 23 percent of all bladder cancer cases are fatal and are, therefore,
assigned the VSL.
For the nonfatal bladder cancer cases, comprising 77 percent of the total, two alternative methods for
determining the value are used to calculate the breakeven point. The first, described in a previous
section, uses a value from a study that estimates the WTP to avoid a case of chronic bronchitis as a
substitute for the WTP to prevent a nonfatal case of bladder cancer (a distribution with a mean value of
$587,500). The second method estimates the COI for a nonfatal cancer case, including treatment costs
and lost productivity. The COI, estimated at $121,000, is an incomplete measure because it does not
include the value for pain and suffering or risk aversion, but involves less uncertainty because it is
derived directly from the costs of medical care. Appendix H contains a more detailed discussion of the
derivation of the WTP and COI estimates for nonfatal bladder cancer.
A Monte Carlo simulation was used to develop a distribution of the number of cancer cases avoided
necessary to breakeven. Results can be found in Appendix J.
At the median, using the WTP value and a cost of capital rate of 7 percent, the Stage 1 DBPR would
need to avoid 438 bladder cancer cases per year to break even, of which 101 are assumed to be fatal and
337 are nonfatal (Exhibit 6.5). Using the COI value, the Stage 1 DBPR would need to avoid 574 bladder
cancer cases per year to break even (132 fatal and 442 nonfatal) (Exhibit 6.6).
Stage 1 DBPR Final RIA 6-10 , November 12, 1998
-------
Exhibit 6.5
Breakeven Analysis Willingness-to-Pay (WTP) Summary
If you assume the following:
Implementation cost of the rule
Value per statistical life saved (mean of iognormal distribution, standard
deviation of $3.16 million)
Willingness-to-pay to avoid a case of bladder caocer (median)
Percent of fatal bladder cancer cases
Percent of nonfatal bladder cancer cases
Total number of cancer cases prevented to break even (at median)
Fatal cancer cases
Nonfatal cancer cases
Reduction in exposure necessary to reach breakeven at given baseline
attributable risk
1,000 baseline attributable cases requires
5,000 baseline attributable cases requires
10,000 baseline attributable cases requires
Baseline attributable risk necessary to reach breakeven at given
exposure reductions
30 percent reduction in exposure
24 percent reduction in exposure
18 percent reduction in exposure
3% Cost of Capital
$626,48 million
$5.6 million
$587,500
23 percent
77 percent
391
90
301
39 percent reduction
8 percent reduction
4 percent reduction
1,300 attributable cases
1 ,630 attributable cases
2,170 attributable cases
7% Cost of Capital
$701. 18 million
$5.6 million
$587,500
23 percent
77 percent
438
101
337
44 percent reduction
9 percent reduction
4 percent reduction
1,460 attributable cases
1,825 attributable cases
2,435 attributable cases
The breakeven number of cases provides only part of the information needed to assess under what beliefs
the Stage 1 DBPR will break even. Two other factors, the baseline number of attributable bladder cancer
cases and the percent reduction in exposure due to the Stage 1 DBPR, combine to give us the number of
cancer cases avoided by the rule. In general, these two factors have an inverse relationship with respect
to the breakeven point: the higher the baseline number of cases, the lower the reduction needs to be to
break even. Conversely, the lower the baseline number of cases, the higher the reduction in risk needs to
be. Exhibit 6.7 contains a graph that addresses the question: "What range of baseline risk and risk
reduction due to the Stage 1 DBPR would you need to reach these breakeven cancer cases?"
Exhibit 6.7 displays two breakeven lines assuming a 7 percent discount rate, one calculated with WTP
and one with COI to value nonfatal cancer cases. Each point on the line represents the combination of
baseline attributable cancer cases and percent reduction in exposure needed to produce the breakeven
cases. For WTP, the combination of baseline number of cancer cases and percent reduction in exposure
at each point on the dark, solid line produces 438 cancer cases avoided. For example, at 1,000
attributable cancer cases, the associated percent reduction is approximately 44 percent. At 5,000 cancer
cases, the percent reduction is 9 percent. For the COI (dark, dashed line), each point on the line produces
574 breakeven cancer cases. Graphically, at any combination of baseline risk and percent reduction in the
area to the right of the lines, the Stage 1 DBPR would exceed breakeven (benefits would exceed costs)
and at any combination in the area to the left of the lines, the costs of compliance would exceed the
benefits (counting only the those attributable to the reduction of bladder cancer risk).
Stage 1 DBPR Final MA
6-11
November 12, 1998
-------
Exhibit 6.6
Breakeven Analysis Cost-of-Illness (COI) Summary
If you assume the following;
Implementation cost of the rule
Value per statistical life saved (mean of lognormal distribution, standard
deviation of $3.16 million)
Cost-of-illness for a case of bladder cancer
Percent of fatal bladder cancer cases
Percent of nonfatal bladder cancer cases
Total number of cancer cases need to prevent to break even (at median)
Fatal cancer cases
Nonfatal cancer cases
Reduction in exposure necessary to reach breakeven at given baseline
attributable risk
1,000 baseline attributable cases requires
5,000 baseline attributable cases requires
10,000 baseline attributable cases requires
Baseline attributable risk necessary to reach breakeven at given
exposure reductions
30 percent reduction in exposure
24 percent reduction in exposure
1 8 percent reduction in exposure
3% Cost of Capital
$626.48 million
$5.6 million
$121,000
23 percent
77 percent
513
118
395
5 1 percent reduction
10 percent reduction
5 percent reduction
1,710 attributable cases
2, 1 40 attributable cases
2,850 attributable cases
7% Cost of Capital
$70 1.1 8 million
$5.6 million
$121,000
23 percent
77 percent
574
132
442
57 percent reduction
1 2 percent reduction
6 percent reduction
1,915 attributable cases
2,390 attributable cases
3,190 attributable cases
Three estimates of percent reduction are of particular importance. The exposure reduction analysis
described earlier estimated that the Stage 1 DBPR will reduce exposure by 24 percent, with a 25th
percentile value of 18 percent and a 75th percentile value of 30 percent. If the reduction in exposure is
assumed to be 24 percent, the baseline number of attributable bladder cancer cases necessary to reach
breakeven (at a 7 percent discount rate) would be 1,825 assuming a WTP measure and 2,390 assuming a
COI measure. A 30 percent reduction results in breakeven attributable cases of 1,460 for WTP and 1,915
for COI. At 18 percent reduction, the breakeven attributable cases rises to 2,435 for WTP and 3,190 for
COL
Exhibit 6.7 also displays the breakeven points at a 3 percent social discount rate. Assuming an annual
implementation cost of $626.5 million (at 3 percent annualization), the Stage 1 DBPR would need to
prevent 391 bladder cancer eases (at the median) using a WTP measure and 513 cases using a COI
measure. At a 24 percent reduction in exposure, this translates into 1,630 attributable cases for WTP and
2,140 for COI.
Stage 1 DBPR Final R1A
6-12
November 12, 1998
-------
c
o
1
I
CD
0)
o
Q.
X
UJ
,2
+s
o
•a
o
o:
«*-!
O
o
0.
100%
90%
80%
70%
60% 4
50%
40%
30%
20%
10%
Exhibit 6.7
Percent Reduction in Exposure Needed to Break Even by Baseline Number of Attributable Cancer Cases
1,000
2,000
3,000
4,000 5,000 6,000
Baseline Attributable Cancer Cases
7,000
8,000
9,000
10,000
- - 7% - Cost of Illness
•7%-Willingness-to-Pay - - 3% - Cost of Illness
•3% - Willingness-to-Pay
-------
Interpreting the Results
Using these two different measures to value nonfatal bladder cancer cases and percent reductions due to
the Stage 1 DBPR with a mean of 24 percent and a range of 18 to 30 percent, the number of bladder
cancer cases attributable to DBFs needs to be at least between 1,460 to 3,190 for the rule to break even at
a 7 percent rate. In terms of the attributable risk, these values translate into population attributable risk
(PAR) numbers of between 2.7 to 5.9 percent. In other words, if the actual PAR is between 2.7 and 5.9
percent and exposure reduction is between 18 and 30 percent, the Stage 1 DBPR will be cost-effective.
These breakeven PAR values are well within the low to mid end of the range of PARs suggested by the
epidemiological data (2 to 17 percent). Based on this breakeven analysis, there seems to be a reasonable
likelihood that the Stage 1 DBPR will be cost-effective with the potential benefits at least equaling the
costs.
6.5 Household Cost Analysis
A fourth approach for assessing the net benefits of the Stage 1 DBPR is to calculate the costs per
household for the rule. Household costs provide a common-sense test of benefit/cost relationships and
are another useful benchmark for comparing the WTP to reduce the possible risk posed by DBFs in
drinking water. It is essentially a household level breakeven analysis. It works backwards from the costs
to ask whether the implied amount of benefits (WTP) needed to cover costs is a plausible amount.2
An estimated 115,490,000 households are located in service areas of systems affected by the Stage 1
DBPR. Of these households, 71 million (62 percent) are served by large surface water systems.
Approximately 4.2 million (4 percent) are served by small surface water systems. Large ground water
systems served 24 million households (21 percent) and small ground water systems serve 15.7 million
households (13 percent).
All of the households served by systems affected by the Stage 1 DBPR will incur some additional costs,
even if the system does not have to change treatment to comply with the proposed rule, for monitoring.
The costs calculated below include both monitoring and treatment costs.
The cumulative distribution of household costs for all of the systems and by each system type is
displayed in Exhibit 6.8. The distributions show that the large percentage of households will incur small
additional costs, with a small portion of systems facing higher costs. At the highest end of the
distribution, approximately 1,400 households served by surface water systems in the 25-100 population
size category switching to membrane technology will face an annual cost increase of $400 per year ($33
per month).
The households have been sorted into three cost categories for the ease of comparison (Exhibit 6.9). The
first category includes households with a cost increase of less than $12 per year, less than $1 per month.
The second category contains households with costs greater than $12 per year, but less than $120 per
year ($10 per month). The third category includes households with cost increases greater than $120 per
year to $400 per year ($33 per month).
The calculations assume that all increases in costs are ultimately borne by consumers, either through direct charges
from the utilities or through increased prices of goods passed on by producers. This assumptions overstates the cost to affected
households because some of the price increases of goods may be borne by consumers outside of the service areas affected by the
Stage 1 DBPR and not all increases are passed along to consumers, depending on the price elasticity of demand.
Stage I DBPR Final RIA 6-14 November 12, 1998
-------
Across all system categories, 95 percent of the households (110.1 million) fall within the first category
and will incur less than $1 per month additional costs due to the Stage 1 DBPR. An additional 4 percent
(4.4 million) are in the second category at between $1 and $10 per month cost increase and 1 percent (1.0
million) are in the highest category ($10 to $33 per month).
For households served by large surface water systems (Exhibit 6.10), 98 percent will incur less than $1
per month, 2 percent will incur between $1 and $10 per month, and 0.03 percent will incur greater than
$10 per month. The highest cost ($125 annually, $10 monthly) is faced by households served by systems
in the 10,000 to 25,000 population size category implementing membrane technology.
For households served by small surface water systems (Exhibit 6.10), 71 percent will incur less than $1
per month, 28 percent will incur between $1 and $10 per month, and 1 percent will incur greater than $10
per month. The highest cost ($400 annually, $33 monthly) is faced by households served by systems in
the 25-100 population size category implementing membrane technology.
For households served by large ground water systems (Exhibit 6.10), 95 percent will incur less than $1
per month, 4 percent will incur between $1 and $10 per month, and 1 percent will incur greater than $10
per month. The highest cost ($125 annually, $10.40 monthly) is faced by households served by systems
in the 10,000 to 25,000 population size category implementing membrane technology.
For households served by small ground water systems (Exhibit 6.10), 91 percent will incur less than $1
per month, 5 percent will incur between $1 and $10 per month, and 4 percent will incur greater than $10
per month. The highest cost ($357 annually, $30 monthly) is faced by households served by systems in
the 25-100 population size category implementing membrane technology.
In the small proportion of systems where household costs exceed $10 per month, these results are driven
by the assumption that advanced technologies, such as membranes or ozone will be the selected
treatment, as noted above. Additionally, two points must be made: 1) many of these systems may find
less expensive means of compliance (e.g., use of point-of-use devices, selection of alternative source
water, purchased water, or consolidation with other systems); and 2) if these systems do install advanced
technologies such as membranes, they may reap additional water quality and/or compliance benefits
beyond those associated with DBFs. For example, because membranes are so effective, systems that
install membranes are likely to incur no significant compliance costs for future rulemakings.
Interpreting the Results
Given the uncertain nature of the risks associated with DBFs, household costs provide a common sense
estimate of WTP to reduce the risks: Would the average household (95 percent of households) be willing
to pay less than $1 per month ($12 per year) to reduce the potential risks posed by DBFs? Would a small
percentage of households (4 percent) be willing to pay between $1 per month ($12 per year) and $10 per
month ($120 per year) to reduce the potential risks posed by DBFs? Would the remaining small
percentage of households (1 percent) be willing to pay between $10 per month ($120 per year) and $33
per month ($400 per year) to reduce the potential risks posed by DBFs?
WTP studies are not available to directly answer these questions. Taking the $1 per month figure as a
measure of implied public health benefit at the household level, it is useful to ask what benefits can be
identified that could balance a $1 per month expenditure. First, it is entirely possible that there is much
more than a dollar-a-month's worth of tangible health benefit based on the reduced risk of bladder cancer
Stage I DBPR Final R1A 6-15 November 12, 1998
-------
alone. Second, the broad exposure to DBFs and the myriad possible health effects involved offer the
possibility that there are significant additional health benefits of a tangible nature.
Finally, however, the preventive weighing and balancing of public health protection provides also a
margin of safety—a hedge against uncertainties. Recent survey research conducted in the drinking water
field provides compelling empirical evidence that the number one priority of water system customers is
the safety of their water. There is no doubt, given the uncertainties, that part of the public health benefit
of the Stage 1 DBPR is the intangible benefit of having an additional margin of safety. Although
definitive economic research has not been performed to investigate the extent of household WTP for
such a margin of safety, there is very strong evidence from conventional customer survey research
implying a demand for this benefit.
Exhibit 6.8
Cumulative Distribution of Annual Household Costs under the Stage 1 DBPR
Cumulative Percent of Households
100%
tf IOS per month (99th pcrccnlile)
\\\
fifl%, I I 1S P*r month (95th percentile)
70%
/
B0%
50%
40%
30%
20%
10%
0%
•
$- $50 $100 $150 $200 $250 $300 $350 $400 $450
Annual Cost per Household
Stage 1 DBPR Final RIA
6-16
November 12, 1998
-------
Exhibit 6,9
Summary of the Number of Households and Percentage of Total Households in Each Cost Category
Total
Large Surface Water
Small Surface Water
Large Ground Water
Small Ground Water
All Systems
Number of Percent of
Households Total
115,490,000 100%
71,378,000 61.8%
4,267,000 3.7%
24,174,000 20.9%
15,671,000 13.6%
$0 -$12 per Year
Cost Per Household
Number of Percent of
Households Total
110,093,000 95%
69,870,000 60%
3,009,000 3%
22,969,000 20%
14,245,000 12%
$12,01 -$120 per Year
Cost Per Household
Number of Percent of
Households Total
4,387,000 4%
1,489,000 1%
1,204,000 1%
939,000 1%
755,000 1%
$120.01 -$400 per Year
Cost Per Household
Number of Percent of
Households Total
1,011,000 1%
20,000 0.02%
54,000 0.05%
266,000 0.2%
671 ,000 0.6%
Summary of the Number of Households and Percentage of Households in Each Cost Category by System Type
Total
Large Surface Water
Small Surface Water
Large Ground Water
Small Ground Water
All Systems
Percent of
Number of System
Households Category
115,490,000 100%
71,378,000 100%
4,267,000 100%
24,174,000 100%
15,671,000 100%
$0 -$12 per Year
Cost Per Household
Percent of
Number of System
Households Category
110,093,000 95%
69,870,000 98%
3,009,000 71%
22,969,000 95%
14,245,000 91%
$12.01 -$120 per Year
Cost Per Household
Percent of
Number of System
Households Category
4,387,000 4%
1,489,000 2%
1,204,000 28%
939,000 4%
755,000 5%
$120.01 - $400 per Year
Cost Per Household
Percent of
Number of System
Households Category
1,011,000 1%
20,000 0.03%
54,000 1%
266,000 1%
671,000 4%
Stage 1 DBPR Final RIA
6-17
November 12, 1998
-------
5"
*•-«
b
^
^
a
5,
t^
5
1
1
-------
5?
*
1
**i
1
1
*
fc^
^0
r
Q"
3
N
1
Large Surface Water Systems Large Ground Water Systems
100%
m 90% .
13
J 80% .
§ 70% .
o
J 60%
o
c 50% ,
0)
fe 40%
Q.
S> 30%
£ 20%
3
1 10% .
u
0% .
, ,i (jf -•
*' 10$ per month (100th pcrccntilc)
\
IS per month (98th percentile)
1
S- $50 $100 $150 $200 $250 $300 $350 $400 $450
Annual Cost per Household
100%-,.-
u 99% ..
2
J2 98%
§ 97% .-
o
J 96% -
o
c 95%
e
o 94%
0.
g 93% -
5 92% _
a 91% -
O
90%
' "
s^
1
'
•$- $20 $40 $60 $80 $100 $120 $140 $160
Annual Cost par Household
i
\
100%
90%
1 80%
| 70%
I 60%
"5
•g 50%
o
| 40%.
« 30%
4j 20% -
1 10%
o
0%
%
4-
jfr r~ "lOS per month (100th percentile)
IS per month (95th percentile)
$50 $100 $150 $200 $250 $300 $350 $400 $450
Annual Cost per Household
100%
99%
M
•S 98%
£
S 97%
3
O
I 96% -
| 95% .
v
SJ 94%
0.
a 93%
| 92%^ .
1 91%
U
»0">A
^-"*"*"
— — """
/
/
$- $20 $40 $60 $80 $100 $120 $140 $160
Annual Cost per Household
Exhibit 6.1 Ob
Cumulative Distribution of Annual Costs per Household in Large Systems
-------
6.6 Decision-Analytic Model
Total social cost of the proposed rule is a function of the annualized implementation cost, the number of
bladder cancer cases that are attributable to DBFs in chlorinated water, and the effectiveness of the rule
in reducing bladder cancers through reduced DBFs in drinking water. The baseline cost, the cost of not
promulgating the rule, depends only on the number of bladder cancers attributable to DBFs. Of the three
parameters, the best known is the implementation cost and the least-known is the number of attributable
bladder cancer cases.
In the approach described below, uncertain information is modeled as probability functions. Expected
total social costs are derived for the No Action and Stage 1 DBPR alternatives, revealing that the rule is
superior. Finally, indifference points are determined. The indifference points show levels of the unknown
parameters for which the No Action and Stage 1 DBPR alternatives would have equal total social costs.
Characterizing Uncertain Information
Probability functions have been constructed to represent a set of reasonable assumptions about the three
uncertain parameters. Exhibit 6.11 represents assumptions regarding implementation cost. The
distribution is Gaussian (normal) with a central value of $0.701 billion and a standard deviation of
$0.105 billion. The coefficient of variation (standard deviation divided by the mean) is 15 percent.
Exhibit 6.12 represents the estimated assumptions regarding the effectiveness of the Stage 1 DBPR in
reducing exposure to DBPs. Assuming that the reduction in bladder cancer risk is proportional to
reductions in TTHM concentration, these figures represent the effectiveness in reducing bladder cancer.
This distribution is also Gaussian but has a mean of 0.24 and a 25 percent coefficient of variation. This
mean value is EPA's estimate of the reduction in exposure derived in Appendix G.
The uncertainty in attributable bladder cancers is not as simple. Approximately 54,500 new bladder
cancer cases are diagnosed each year, but the fraction due to DBPs is largely unknown. As described
earlier, since causality has not been proven, there may, in fact, be no bladder cancer cases due to DBPs.
In this analysis, EPA has assumed a 20 percent probability that these DBPs do not cause bladder cancer.
Based on the PAR estimates derived from the recent epidemiological studies described earlier, the range
of PARs is estimated at between 0 percent to 20 percent. An upper bound of 20 percent for the PARs was
deemed reasonable since all the calculated PARs in the 2 to 17 percent range were based on the central
tendency estimate for each study. Exhibit 6.13 utilizes a uniform distribution to represent total
uncertainty over that range, but allows for a 20 percent probability that the PAR is 0. The expected value
of the PAR under this set of assumptions, denoted as E(PAR) is 8 percent.
Computing Total Social Cost
Using the values per cancer case described in previous sections, one can use the probability functions to
derive the expected total social costs. In the equations below, fPAR(PAR), fi^conCC), and fR^ctionO*)*
denote the probability density functions for PAR, rule implementation cost, and percent reduction,
respectively. The expected total social cost (in billions of dollars) of the No Action alternative is derived
first:
E(Cost NoActJon)=J 54500 cases * $1.750/1 OOOcases * PAR * fPAR(PAR) dPAR
Stage 1DBPR Final RIA 6-20 November 12, 1998
-------
E(Cost No Acrion)=J 95.4 * PAR * fPAR(PAR) dPAR
E(Cost NoAction)= 95.4 * E(PAR)=$7.63 billion
A bit more complex is the estimation of expected total social cost of the Stage 1 DBPR:
E(Cost s,age ,)=]> f,mpicosl (c) dc + 95.4 J J PAR * (1-r) fPAR(PAR) freduction(r) dr dPAR
E(Cost stage ,)= E(c) + 95.4 J PAR * (1 -E(r)) fPAR(PAR) dPAR
E(Cost stagc ,)= E(c) + 95.4 * E(PAR) * (l-E(r))
E(Cost Slage ,)= 0.702 + 95.4 * .08 * 0.76 = $6.5 billion
Estimating the Indifference Points
The three key uncertain parameters are implementation (cost), attributable bladder cancers (PAR), and
the effectiveness of the rule (r). Maintaining any two of these attributes as uncertain (represented by the
probability functions described earlier), we ask what level of the third attribute would be needed to make
the expected cost of the No Action equal to the expected cost of the rule. The indifference points derived
through this analysis are as follows:
Parameter
Implementation Cost (c)
Attributable bladder cancers
Rule Effectiveness
Indifference Point
$1.831 billion
PAR of 3. 07
9.21 percent reduction
The indifference points for implementation cost and rule effectiveness are unlikely levels. An
implementation cost greater than $1.831 billion would favor the No Action alternative, but the
probability of a greater cost is virtually zero. Similarly, an effective reduction of less than 9.21 percent
would favor the No Action alternative, but the probability of such a low effectiveness is less than 0.01
percent, under the probability assumptions. New information on the cost or effectiveness, such as would
be produced by new research or more accurate engineering or cost models, is not likely to make the No
Action a more attractive alternative.
In contrast, the indifference point for the PAR, 3.07 percent, appears to be a reasonably likely level. The
probability of a lesser PAR is 32 percent, which includes the assumed 20 percent allowance that the PAR
may equal 0. New information could be quite valuable, especially if it would establish or reject causality.
Although new information would be imperfect, the expected value of perfect information can be viewed
as the upper bound on what should be spent for imperfect information. Derived using numerical
integration, the expected value of perfect information on PAR is approximately $200 million. While this
far exceeds the cost of typical epidemiological studies, the inability of such studies to test causality may
Stage 1 DBPR Final RIA
6-21
November 12, 1998
-------
render a planned study's value far less that its cost. Still, the large expected value of perfect information
suggests that a more in-depth value-of-information analysis could be beneficial,
Interpreting the Results
This approach calculated expected total social costs for the Stage 1 DBPR and No Action alternatives
under a set of reasonable assumptions. Uncertain parameters were modeled and their impacts on the
decision were carefully evaluated and considered.
Under the given set of assumptions, the results indicate that the choice of the Stage 1 DBPR over the No
Action alternative is not sensitive across the reasonable range of possible values for both implementation
cost or percent reduction in exposure. However, the choice is sensitive to the assumed values for the
attributable bladder cancer cases. Exhibit 6.14 displays the distribution of the estimated Stage 1 DBPR
net benefits given the above input assumptions. The results suggest that there is a one-in-three chance
that net benefits could be negative and a two-in-three chance that the net benefits could be positive.
Stage I DBPR Final RIA 6-22 November 12, 1998
-------
Exhibit 6.11
Density Function of Implementation Costs
1
0,8
£» 0.6 -
c
0)
Q Q4
0.2
0
$
0
$250 $500 $750 $1,000 $1,250 $1,500
Implementation Costs (Millions $)
Exhibit 6.12
Density Function of Exposure Reduction
5% 10% 15% 20% 25% 30% 35% 40%
Percent Reduction in Exposure
45%
Stage 1 DBPR Final RIA
6-23
November 12, 1998
-------
Exhibit 6.13
Density Function of PAR Estimates
n RD -
U.UU "
n AR
n AO
U.'tVJ
n ^i
U.OO "
SK 0 ^0
w n o^ .
£ W.t.v>
J 1
n on
0.
00 0.02 0.04 0.06 0.08 0.10 0.12 0.14 0.16 0.18 0.
PAR
20
inn"/, _.
Q0% -
an% .
70%
m
— Rn%
**
O*yO%
O 40% -
Q. °
•sn% -
70% -
10% -
no/.
-3
C
52 -5
'umu
/
/
/
»1
Exhibit 6.14
Cumulative Distribution of Predicted Net Benefits
$0 $1 $2 $3
Net Benefits (Billions $)
$4
$5
$6
Stage 1 DBPR Final R1A
6-24
November 12, 1998
-------
7: The Economic Rationale for Regulation
7.1 Introduction
This chapter of the analysis discusses the economic rationale for choosing a regulatory approach to
address the public health consequences of drinking water contamination. The economic rationale is
provided in response to Executive Order Number 12866, Regulatory Planning and Review, which states:
[E]ach agency shall identify the problem that it intends to address
(including, where applicable, the failures of the private markets or public
institutions that warrant new agency action) as well as assess the
significance of that problem (Section 1, b(l)).
In addition, OMB guidance dated January 11, 1996, states that "in order to establish the need for the
proposed action, the analysis should discuss whether the problem constitutes a significant market failure"
(p.3). Therefore, the economic rationale laid out in this section should not be interpreted as the agency's
approach to implementing the Safe Drinking Water Act (SDWA). Instead, it is the agency's justification,
as required by the Executive Order, for a regulatory approach to this public health issue.
7.2 Statutory Authority for Promulgating the Rule
The 1996 reauthorization for the Safe Drinking Water Act (SDWA) mandated new drinking water
requirements. EPA's general authority to set Maximum Contaminant Level Goals (MCLGs) and the
National Primary Drinking Water Rule (NPDWR) was modified to apply to contaminants that "may have
an adverse effect on the health of persons," are "known to occur or there is a substantial likelihood that
the contaminant will occur in public water systems with a frequency and at levels of public health
concern," and for which "in the sole judgment of the Administrator, regulation of such contaminant
presents a meaningful opportunity for health risk reductions for persons served by public water systems"
(1996 SDWA, as amended).
The 1996 Amendments also require the promulgation of the Interim Enhanced Surface Water Treatment
Rule (IESWTR) and a Stage 1 Disinfectants/Disinfection Byproducts Rule (Stage 1 DBPR) by
November 1998. In addition, the 1996 Amendments require EPA to promulgate a Final Enhanced
Surface Water Treatment Rule and a Stage 2 DBPR by November 2000 and May 2002, respectively.
7.3 The Economic Rationale for Regulation
In addition to the statutory directive to regulate disinfection byproducts, there is also economic rationale
for government regulation. In a perfectly competitive market, market forces guide buyers and sellers to
attain the best possible social outcome. A perfectly competitive market occurs when there are many
producers of a product selling to many buyers, and both producers and consumers have complete
knowledge regarding the products of each firm. There must also be no barriers to entry in the industry,
and firms in the industry must not have any advantage over potential new producers. Several factors in
the public water supply industry do not satisfy the requirements for a perfect market and lead to market
failures that require regulation.
Stage 1 DBPR Final RIA 7-1 November 12, 1998
-------
First, water utilities are natural monopolies. A natural monopoly exists when it is not economically
efficient to have multiple suppliers competing to build multiple systems of pipelines, reservoirs, wells,
and other facilities. Instead, a single firm or government entity performs these functions under public
control. Under monopoly conditions, consumers are provided only one level of service with respect to
the quality attribute of the product, in this case drinking water quality. If they do not believe the margin
of safety in public health protection is adequate, they cannot simply switch to another water utility.
Second, there are high information and transaction costs that impede public understanding of the health
and safety issues concerning drinking water quality. The type of health risks potentially posed by trace
quantities of drinking water contaminants involve analysis and distillation of complex toxicological data
and health sciences. EPA is currently in the final stages of developing the Consumer Confidence Report
rule that will make water quality information more easily available to consumers. The Consumer
Confidence Report rule will require community water systems to mail their customers an annual report
on local drinking water quality. However, consumers would still have to analyze this information for its
health risk implications. Even if informed consumers are able to engage utilities regarding these health
issues, the costs of such engagement-transaction costs (measured in personal time and commitment)
present another significant impediment to consumer expression of risk preference.
SDWA regulations are intended to provide a level of protection from exposure to drinking water
contaminants that would not otherwise occur in the existing market environment of public water supply.
The regulations set minimum performance requirements for all public water supplies in order to protect
all consumers from exposure to contaminants. SDWA regulations are not intended to restructure market
mechanisms or to establish competition* in supply; rather, SDWA standards establish the level of service
to be provided in order to better reflect public preference for safety. The Federal regulations remove the
high information and transaction costs by acting on behalf of all consumers in balancing the risk
reduction and the social costs of achieving this reduction.
Stage J DBPR Final R1A 7-2 November 12, 1998
-------
References
58 FR 51735. Executive Order 12866. "Regulatory Planning and Review," October 4, 1993,
59 FR 38668. Preamble to the 1994 proposed Stage 1 DBPR,
62 FR 59388. Preamble to the 1997 Notice of Data Availability.
63 FR 15674. Preamble to the 1998 Notice of Data Availability.
American Cancer Society. 1998. "Cancer Facts and Figures." http://www.cancer.org/
Amy, G., M. Siddiqui, K. Ozekin, H. W. Zhu and C. Wang. 1995. "Empirically Based Models for
Predicting Chlorination and Ozonation By-Products; Haloactie Acids, Chloral Hydrate, and Bromate."
EPA Draft Report Number Cx 819579, Cooperative Agreement: Univerisity of Colorado and EPA.
Aschengrau A, S. Zierler and A. Cohen. 1993. "Quality of Community Drinking Water and the
Occurrence of Late Adverse Pregnancy Outcomes." Arch. Env. Health. 48:105-113.
Bove F, M. Fulcomer, J. Klotz, J. Esmart, E. Dufficy, R. Zagraniski and J. E. Savrin. 1992b. "Public
Drinking Water Contamination and Birthweight, and Selected Birth Defects: A Case-Control Study
(Phase IV-B)." New Jersey Department of Health. May 1992. 23.
Bove F, M. Fulcomer, J. Klotz, J. Esmart, E. Dufficy, R. Zagraniski and J. E. Savrin. 1992a. "Public
Drinking Water Contamination and Birthweight, Fetal Deaths, and Birth Defects; A Cross-Sectional
Study (Phase IV-A)." New Jersey Department of Health. April 1992.
Bried, E.M. and R.M. Scheffler. 1992. "The Financial Stages of Cancer in the Elderly." Oncology.
Supplement: 6(2).
Bull, RJ. and F. Kopfler, 1991. "Health Effects of Disinfectants and Disinfection By-Products." Denver,
Colorado: American Water Works Association Research Foundation.
Cantor, K.P., R. Hoover and P. Hartge. 1985. "Drinking Water Source and Bladder Cancer: A Case-
Control Study." In: R.L. Jolley, RJ. Bull and W.P. Davis, eds. Water Chlorination: Chemistry,
Environmental Impact and Health Effects, 5(1).
Cantor, K.P., C.F. Lunch, M. Hildesheim, M. Dosemeci, J. Lubin, M. Alavanja and G.F. Craun. 1998.
"Drinking Water Source and Chlorination Byproducts I. Risk of Bladder Cancer." Epidemiology: 9.
Chestnut, L.G., A. Alberni and D. Mills. 1997. "Monetary Valuation of Human Mortality Risks in
Cost-Benefit Analyses of Environmental Programs."
Stage 1 DBPR Final R1A References November 12, 1998
-------
References
Cragle, D.L., CM. Shy, R.J. Struba and EJ. Siff. 1985. "A Case-Control Study of Colon Cancer and
Water Chlorination in North Carolina." In: R.L. Jolley, R.J. Bull and W,P. Davis, eds. Water
Chlorination: Chemistry, Environmental Impact and Health Effects. Chelsea, MI.: Lewis Publishers, Inc.
Cropper and Krupniek, 1989. "The Social Costs of Chronic Heart and Lung Disease." Resources for the
Future Discussion Paper. QE 89-16-REV.
Doyle, T.J., W. Sheng, J.R. Cerhan, C.P. Hong, T.A. Sellers, L,H. Kushi and A.R. Folsom. 1997. "The
Association of Drinking Water Source and Chlorination By-Products with Cancer Incidence Among
Postrnenopausal Women in Iowa: A Prospective Cohort Study." American Journal of Public Health:
87(7).
Freedman, M., K.P. Cantor, N.L. Lee, L.S. Chen, H.H. Lei, C.E. Ruhl and S.S. Wang. 1997. "Bladder
Cancer and Drinking Water: A Population-Based Case Control Study in Washington County, Maryland."
Cancer Causes and Control.
Freeman, A.M. 1979. "The Benefits of Environmental Improvement: Theory and Practice." Resources
for the Future. Johns Hopkins University Press.
Haas, C.N. and J.B. Rose. 1995. "Developing an Action Level for Cryplosporidium." Journal of the
American Water Works Association: 89(9).
Hildesheim, M.E., K.P. Canbor, C.F. Lynch, M. Dosemeci, J. Lubin, M. Alavanja and G.F. Craun. 1997.
"Drinking Water Source and Chlorination Byproducts: Risk of Colon and Rectal Cancers."
Epidemiology: 9(1).
Hillier, F.S. and G.L. Lieberman. 1990. Introduction to Mathematical Programming. McGraw Hill.
King, W.D., and L.D. Marrett. 1996. "Case-Control Study of Bladder Cancer and Chlorination By-
Products in Treated Water (Ontario, Canada)." Cancer Causes Control: 7.
Klotz, J.B. and L.A. Pyrch. 1998. "A Case-Control study of Neural Tube Defects and Drinking Water
Contaminants." New Jersey Department of Health and Senior Services. Sponsored by Agency for Toxic
Substances and Disease Registry.
Komulainen, H., V. M. Kosma, S. L. Vaittinen, T. Vartianen, E. Kaliste-Korhonen, S. Lotjonen, R. K,
Tuominen and J. Tuomisto. 1997. "Carcinogenicity of the Drinking Water Mutagen 3-Chloro-4-
(D5chloromethyl)-5-Hydroxy-2(5H)-Furanone in the Rat." 848-856.
Kramer, M. D., C. F. Lynch, P. Isacson and J. W. Hanson. 1992. "The Association of Waterborne
Chloroform with Intrauterine Growth Retardation." Epidemiology. 3:407-413.
Stage J DBPR Final R1A References November 12, 1998
-------
References
Krasner, Stuart. April 1997. "Issue Paper on Enhanced Coagulation to the M-DBP Advisory
Committee."
Krasner, Stuart. April 9, 1997. "Issue Paper on MCLs for DBPs."
McGeehin, M.A., J.S. Reif, J.C. Becher and E.J. Mangione. 1993. "Case Control Study of Bladder
Cancer and Water Disinfection Methods in Colorado." American Journal of Epidemiology. 138.
Morris, R.D., A.M. Audet, I.F. Angelillo, T.C. Chalmers and F. Mosteller. 1992. "Chlorination,
Chlorination By-Products, and Cancer." American Journal of Public Health.
National Cancer Institute. 1997. http://www.nci.nih.gov/
National Research Council. 1983. "Risk Assessment in the Federal Government: Managing the Process."
Washington, DC.: National Academy Press.
Office of Management and Budget; Guidance. "Economic Analysis of Federal Regulations Under
Executive Order 12866." January 11, 1996. p. 3.
>
Riley, G.F., A.L. Potosky, J.D. Lubits and L.G. Kessler. 1995. "Medicare Payments from Diagnosis to
Death for Elderly Cancer Patients by Stage at Diagnosis." Medical Care: 33(8).
Savitz, D. A., K. W. Andrews and L. M. Pastore. 1995. "Drinking Water and Pregnancy Outcome in
Central North Carolina: Source, Amount, and Trihalomethane Levels." Environ. Health Perspectives.
103(6), 592-596.
Unfunded Mandates Reform Act of 1995 (UMRA), P.L. 104-4. "Unfunded Mandates Reform Act for the
D/DBPR."
U.S. Bureau of Labor Statistics. 1998. http://www.bls.gov/
U.S. Environmental Protection Agency. July, 1998. "Technologies and Costs for Control of Disinfection
By-Products." Office of Ground Water and Drinking Water. Standards and Risk Reduction Branch.
Standards and Risk Management Division.
U.S. Environmental Protection Agency. June 24, 1998. "Cost and Technology Document for Controlling
Disinfectants and Disinfection Byproducts." Washington, DC.: Office of Ground Water and Drinking
Water Resource Center.
U.S. Environmental Protection Agency. March 31,1998. DBP Notice of Data Availability. Federal
Register.
Stage 1 DBPR Final RlA References November 12, 1998
-------
References
U.S. Environmental Protection Agency. 1998a. "Quantification of Cancer Risk from Exposure to
Chlorinated Water." Office of Science and Techology, Office of Water. Washington, DC. March 13,
1998.
U.S. Environmental Protection Agency. 1998b. "Health Risk Assessment/Characterization of the
Drinking Water Disinfectant Byproduct Chlorine Dioxide and the Degradation Byproduct Chlorite."
Office of Science and Techology, Office of Water. Washington, DC. March 13, 1998.
U.S. Environmental Protection Agency. 1998c. "Health Risk Assessment/Characterization of the
Drinking Water Disinfectant Byproduct Chloroform." Office of Science and Techology, Office of Water.
Washington, DC. March 13, 1998.
U.S. Environmental Protection Agency. 1997. "Summaries of New Health Effects Data." Office of
Science and Technology, Office of Water. Washington, DC. October 1997.
U.S. Environmental Protection Agency. 1997. "The Benefits and Costs of the Clean Air Act, 1970-
1990." Prepared for U.S. Congress.
U.S. Environmental Protection Agency. 1994. "Health Care Costs and Productivity Losses Due to
Superfund."
U.S. Environmental Protection Agency. 1994. "Regulatory Impact Analysis of Proposed
Disinfectant/Disinfection Byproduct Regulations." Washington, DC. EPA-###-R-93-###.
U.S. Environmental Protection Agency. Undated. Introduction to IRIS.
http://www.epa.gov/ngispgm3/iris/intro.html/
Viscusi, W.K., W.A. Magat and J. Huber. 1991. "Pricing Environmental Health Risks: Survey
Assessments of Risk-Risk and Risk-Dollar Trade-Offs for Chronic Bronchitis." Journal of
Environmental Economics and Management: 21.
Waller, K., S.H. Swan, G. DeLorenze and B. Hopkins. 1998. "Trihalomethanes in Drinking Water and
Spontaneous Abortion." Epidemiology: 9(2).
Yannick, R. and L.A. Ries. 1994. "Cancer in Older Persons. Magnitude of the Problem—How do We
Apply What We Know?" Cancer.
Young, T.B., D.A. Wolf and M.S. Kanarek. 1987. "Case-Control Study of Colon Cancer and Drinking
Water Trihalomethanes in Wisconsin." International Journal of Epidemiology: 16(190).
Stage J DBPR Final RIA References November 12, 1998
------- |