United States
Environmental Protection
Agency
Air Pollutant Emission
Standards and Guidelines for
Municipal Waste Combustors:
Economic Analysis of
Materials Separation Requirement
(Program Approach)
IV-A-28
Office of Air Quality EPA 450/3-91-006
Planning and Standards November 1990
Research Triangle Park NC 27711
-------
CONTENTS
Section Page
1 Summary 1-1
2 Approach 2-1
3 Results 3-1
4 Limitations of the Analysis 4-1
5 References 5-1
ii
-------
1. SUMMARY
Costs were estimated for a subset of 171 municipal waste combustors
(MWC's) (out of a total population of 280 planned and existing MWC's) that
were assumed to incur incremental costs as a result of the federal
regulations. The excluded MWC's were either located in states with existing
recycling programs and goals or were below a predetermined size cutoff. The
costs were computed using various assumptions about avoided disposal costs
and scrap revenues. The final costs for the rule ranged from $58 million per
year to a savings of $345 million per year.
The separation costs estimates were dominated by collection and
processing costs for recyclables, although administrative costs were included
in the totals. The cost savings resulted principally from a combination of
scrap revenues, avoided landfilling costs, and avoided trash collection
costs. The modeling scenarios were designed to show variations in the
avoided cost of trash collection and disposal as well as scrap revenues.
Scrap revenues did not vary dramatically, given the model assumptions; the
difference between the two scrap scenarios was about $43 million per year.
The disposal/collection assumptions did make a substantial difference,
however. The alternative assumptions accounted for a $182 million difference
in avoided disposal costs for existing MWC's and a $177 million difference in
avoided trash collection costs.
Costs per ton of waste combusted (after the rule is in place) ranged
from $2 per ton to a savings of $12 per ton. The cost per ton of waste
diverted ranged from $5 per ton diverted to a savings of $29 per ton
diverted.
1-1
-------
2. APPROACH
2.1 MUNICIPAL WASTE COMBUSTOR DATA
The population of municipal waste combustors (MWC's) was divided into
two groups for this analysis: existing MWC's and planned MWC's. The
database included 213 existing MWC's (units in service or that will be in
service sometime in 1991). These 213 units were located by city and state.
Similar information was developed for planned MWC's, but the designs,
locations, and capacities of these units were less certain. The planned
units were matched to model plant characteristics and then scaled up to a
total of 67 units of varying technologies and throughput.
Overall, the data included 280 MWC's with a municipal waste throughput
(capacity times utilization factor) of 53.4 million tons per year.
2.1.1 Expanded Service Areas for Existing MWC's
Since MWC's may have financial or energy output obligations that assume
a certain throughput, it is important for them to maintain the pre-recycling
level of operation. One way to maintain throughput is to expand the area
served by the combustor; if more homes send waste to the unit, a specified
fraction can be removed for recycling and the combustor would still be able
to maintain its operating level.* As explained below, the waste diversion
analysis concluded that in order to meet the 25 percent source separation
target, 28.6 percent of the waste generated must be diverted." Given a
28.6 percent diversion, 71.4 percent of the waste collected will still be
burned. For a 100-ton-per-day (TPD) unit then, 100 = 0.714 * Q where Q is
the necessary size of the new, expanded service area. If the new service
*This assumption ignores any change in the heat content of each ton of waste
after recyclables have been removed.
**The discrepancy results from the regulation's limit on the amount of yard
waste that can count toward the 25 percent goal. As proposed, yard waste
could account for no more than 10 percent of the credit; since assumptions
concerning municipal waste composition yield a larger share of yard waste,
the additional material is still diverted, it simply does not contribute to
the 25 percent target.
2-1
-------
area grows to collect 140 TPD, then 40 tons (28.6 percent) can be diverted
and 100 tons will still be burned.
In general, the service areas of existing MWC's must increase by
40 percent. Thus, the original throughput of the 280 MWC's was increased
from 53.4 million tons per year to 67.7 million tons per year, with the
increase only affecting the 213 existing MWC's.
2.1.2 MWC's That Will Not Incur Incremental Costs from the Federal
Requirement
Municipal waste combustors in several states would be forced to
implement substantially similar programs because of state laws; the federal
requirement would impose no incremental costs on MWC's in those states. Nine
states (plus the District of Columbia) would qualify on these grounds. Each
of these 10 jurisdictions have laws requiring at least a 25 percent recycling
rate by 1995. States with lower targets or with similar targets after 1995
were not exempted. This assumption dramatically affected the population of
MWC's analyzed in this study. The 10 jurisdictions contained 88 MWC's, or
31 percent of the 280 original units. Most of the 88 were existing units
(66) and the remaining 22 were planned.
Second, MWC's with capacities of less than 40 TPD were excluded from
compliance with the requirement. This assumption was made in anticipation of
a similar exemption of the final rule. The assumption excluded an additional
21 existing MWC's from the economic impact analysis. Overall, the analysis
focused on 171 MWC's that would incur incremental costs from the
source-sepacation requirement. These 171 MWC's collected just over
40 million tons of MSW per year.
2.1.3 Components of the Wastestream
The composition of MSW collected was a critical factor in this analysis.
The waste composition determined how much diversion was possible given
varying styles of recycling programs; revenues from the sale of collected
materials also depended on the material types. Many communities and even
some states have conducted waste composition studies that sample incoming
loads at landfills and then separate, weigh, and possibly measure the volume
of various waste components. In the research supporting this analysis and in
support of EPA's 1990 Waste Characterization study (the Franklin data)l, many
sampling studies were compiled describing the composition of the
wastestream.2 Waste composition data were not available for the MWC's in the
2-2
-------
database nor are site-specific studies available for a significant share of
the MWC's.
Because of the detail in the composition data in the Franklin study and
because of the complexity involved in modeling site-specific waste profiles,
Franklin's 1988* waste composition data was applied to the MWC's to estimate
site-specific quantities of various components of municipal solid waste
(MSW). t While waste composition can vary substantially based on the mix of
industrial, commercial, and residential generators, climate, socioeconomic
status, education, etc., reliable sources of regional waste composition
estimates are not available that would be appropriate as the basis for
extrapolation to particular regions of the country.
Franklin1 estimated the composition of three different categories of
MSW: MSW generated, materials recovered from the wastestream, and materials
discarded (the residual of generation minus recovery). Table 2-1 summarizes
the national estimates of several recoverable components of MSW, along with
the estimates for total tons discarded nationally.
Yard waste accounts for a significant share of the recoverable materials
(47 percent of the 65.7 million tons that is potentially recoverable) and
20 percent of total discards. Paper and paperboard products included are
newsprint and corrugated and account for about 33 percent of the recoverable
materials identified. The remaining materials are glass, metal cans, and
selected plastic bottles.
Two important assumptions underlie use of these data. First, these
national averages were applied to individual MWC service areas to compute
MWC-specific estimates of the quantity of yard waste, glass containers, etc.
collected in each wasteshed. Second, only certain recoverable materials were
the focus for this analysis; the selection was based on judgment that these
materials would likely be part of a comprehensive recycling program designed
to achieve fairly high rates of diversion (e.g., 25 percent), considering
their prevalence in MSW, the existence of scrap markets, and precedent for
collection of the materials in: existing programs. This is not to say that
other materials might not be part of recycling programs or that every one of
these commodities would be; instead, a representative mix of materials was
selected and it was assumed that recycling programs would collect these
materials. Assumptions about how the materials are collected are described
2-3
-------
TABLE 2-1. SUMMARY OF RECOVERABLE MATERIALS MODELED IN THE MWC ANALYSIS
Waste component
Percentage of
MSW disposed
(%)
Quantity disposed
Nationally in 1988
(million tons)
Yard waste
19.7
30.7
Corrugated boxes
8.1
12.6
Container glass
6.3
9.9
Old newsprint
5.7
8.9
Steel cans
1.5
2.2
Aluminum cans
0.4
0.7
PET soft drink bottles
0.2
0.3
HDPE milk bottles
0.2
0.4
Subtotal
42.1
65.7
Other products/wastes
57.9
90.3
Totals
100
156
Sources: Franklin Associates, Ltd.*
2-4
-------
later in this chapter under the descriptions of the individual recycling
programs modeled.
2.1.4 Diversion Rates
Several factors determine how much of the material disposed could
actually be collected and recycled. First, materials that are contaminated
may not be recyclable; water-soaked newsprint or corrugated cardboard, for
example, may be worthless because of contamination. Similarly, combinations
of recyclable and nonrecyclable materials may prevent recycling. Whether
because of contamination or inconvenience, however, only a fraction of the
quantity of these materials will be available for recovery.
A second factor to consider is the willingness of residents to
participate in recycling programs. Even if a high percentage of newspaper
generated is easily separated and recycled, only those households that
participate in the recycling program will contribute to the diversion of
material from the MWC.
A final consideration is the waste and contaminants that are mixed in
with the collected materials and are ultimately discarded. Labels, closures,
moisture and dirt may all be mixed with recyclables and will eventually find
their way into a landfill or MWC, so this is a third adjustment necessary to
calculate the net quantity of waste diverted from the MWC.
2.1.5 Research
Assumptions regarding recoverable fractions and participation rates are
summarized in Table 2-2. A number of sources on potential recovery rates of
various materials were reviewed and used to derive two separate estimates of
recoverable percentages (to take account of contamination, losses,
inconvenience) and participation rates (which vary by the type of recycling
program and extent of participant motivation). Most sources report
effectiveness of recycling programs in terms of total diversion; these two
components were separated in order to allow modeling different participation
rates inherent in different styles of collection programs (e.g., curbside
versus drop-off programs). These are uncertain estimates and will vary
dramatically based on the characteristics of individual communities.
Nevertheless, it is important to adjust total discards of the materials to
reflect realistic expectations of actual materials recovery.
As shown in the Table 2-2, recoverable fractions range from 60 to
90 percent. Net diversion rates (the product of recoverable fractions and
2-5
-------
TABLE 2-2. SUMMARY OF MATERIAL RECOVERY ESTIMATES
Waste component
Percentage of
MSW disposed
(%)
Percentage
recoverable3
(%)
Net diversions
given 80 percent
participation
(%)
Yard waste
19.7
90b
72
Corrugated boxes
8.1
75C
60
Container glass
6.3
87d
70
Old newsprint
5.7
00
70
Steel cans
1.5
60f
48
Aluminum cans
0.4
909
72
PET soft drink bottles
0.2
75&
60
HDPE milk bottles
0.2
601
48
Totals
42.1
aFraction of materials recoverable from household waste stream correcting for
losses, contamination, inconvenience, etc. Several of these estimates were
derived by dividing estimates of net diversion by the participation rate.
&NRDC estimate from testimony before the EPA Materials Separation Workshop,
February 15, 1990, p. 9.
cMay be high for households, but consistent with commercial corrugated
recycling rates.
^Considers NRDC's 65 percent net diversion rate and higher rates achieved in
other types of programs (e.g., deposit laws).
eSlightly above the 65 percent net diversion rate cited as a target by NRDC
and as a projection by Andover International Associates, Resource Recycling,
April 1990, pp. 70+.
^Given an 80 percent participation rate, the net diversion would represent
about a 50 percent Increase af)ove current recycling (just over 30 percent
for steel beverage cans).
^Assumes virtually all cans are recoverable. Diversion rate of 72 percent is
well below that of some deposit states.
^Based on a net diversion rate of over 50 percent which as been achieved in
deposit programs.
"ฆVery limited experience with recovery of these containers.
2-6
-------
participation rates) range from 48 percent for milk bottles (for which there
is little precedent for recycling thus far) up to 72 percent for yard waste
and aluminum cans.
The notes to the table explain the assumptions and sources behind the
recoverable fractions. In some cases, estimates were obtained from secondary
sources, but in other cases estimates were based on judgment or derived from
a source that cited the overall diversion rate. Given a 70 percent net
diversion rate and an 80 percent participation rate, for example, the
recoverable fraction must be 0.87, or 87 percent.
An 80 percent participation rate was selected as applicable to recycling
programs that provide curbside pickup of materials.* According to the
sources reviewed (further discussion of the recycling programs that were
modeled appears later is this chapter), participation rates for curbside
programs ranged from 3 to 98 percent. In the data reviewed, 49 programs
reported participation rates and 14 of them cited rates of greater than
80 percent, so this participation rate is viewed as an optimistic, but
realistic, estimate.
The effect of these recoverable and participation rates is to decrease
the total quantity of waste that Is diverted from MWC's as a result of
recycling. Table 2-3 integrates the information on waste components with the
net diversion rates to compute the share of MSW diverted from MWC's. Given
implementation of a recycling program that covered all of these materials, an
80 percent participation rate would divert nearly 29 percent of the MSW
disposed. Jhe "extra" diversion above the 25 percent target is discussed
next.
2.1.6 Adjustment to the Diversion Rates
Two adjustments were made to the diversion rates: the first
incorporated the 10 percent cap on credit for yard waste recycling and the
second accounted for contaminants collected with recyclables that are
ultimately discarded rather than recycled.
The regulation stipulates; that regardless of how much yard waste is
collected, it can only contribute 10 percentage points of the 25 percent
*Given the high separation target, curbside collection is the most likely to
provide the high participation rates necessary to meet the target. Drop-off
or buy-back centers may be sufficient in some settings or for some
materials, but participation tends to be much lower.
2-7
-------
TABLE 2-3. SUMMARY OF DIVERSION ESTIMATES FOR MWC's
Waste component
Percentage of
MSW disposed
(*)
Percentage of
MSW Diverted bv Recvclina
With
Total 10 percent cap
diversion on yard waste
(*) (*)
Yard waste
19.7
14
10
Corrugated boxes
8.1
5
5
Container glass
6.3
4
4
Old newsprint
5.7
4
4
Steel cans
1.5
0.7
0.7
Aluminum cans
0.4
0.3
0.3
PET soft drink bottles
0.2
0.1
0.1
HDPE milk bottles
0.2
0.1
0.1
Totals
42.1
29
25
Notes;
Percentage diverted is based on net diversion rates shown in Table 2-2.
Totals may not add because of rounding.
2-8
-------
target: other materials must make up the remaining 15 points. Calculations
of diversion rates and participation demonstrate that over 14 percent of the
wastestream could be diverted from yard waste recycling, so the actual share
diverted will exceed 25 percent. Table 2-3 also summarizes the percentage of
MSW diverted once the cap on yard waste is included. The analysis assumes
that the additional material would still be collected and the appropriate
costs and offsets, such as avoided disposal costs, would still accrue to the
community.
The second adjustment made was for contamination of recyclables.
Moisture, dirt, labels, or other nonrecyclable materials often are included
with recyclables. Data from 72 operating and planned material recovery
facilities (MRF's) that handle a mix of recyclables indicate that the average
waste factor is 9.7 percent.3
The adjustment did not affect the recycling target since all the
material collected in the recycling program was originally separated and
would count toward the 25 percent goal. In computing the tonnage that would
avoid trash collection and disposal costs, the quantity of material collected
was reduced by 9.7 percent. A similar adjustment was made before computing
scrap revenues.
To summarize the adjustments to waste quantity recycled, the total
quantity of recyclable materials currently sent to MWC's was first estimated.
Adjustment factors were applied to these estimates to reflect the ease of
recovery (the recoverable fraction). Then participation rates were used to
further adjust the quantity of material collected. The final estimate of
waste diverted from landfills is adjusted by a waste factor that equals
approximately 10 percent of the weight of material collected. Recycling
programs are designed to collect the total quantity of recyclables taking
into account the recoverable fraction and participation rates. Avoided
disposal costs are computed based on the tonnage ultimately diverted from
MWC's, which includes the waste factor.
2.1.7 Recycling Programs and Costs
The model developed provided several alternative recycling program
options. The options included in the MWC analysis were:
Curbside collection programs for all the modeled materials except
yard waste;
2-9
-------
The MRF's that may be combined with a curbside program to provide
for separation and processing of collected materials; and
Compost programs for yard waste that include curbside collection
and centralized processing of the material.
Various combinations of these three programs were also modeled. Given the
assumed participation rates associated with each and the mandatory 25 percent
recycling target, only tWo program scenarios could achieve the desired
diversion rate: compost plus curbside collection, and compost plus curbside
collection and an MD.F, Either option would meet the 25 percent target, even
taking into account the 10 percent cap on yard waste credit.
Unit cost data were compiled from secondary sources for curbside, MRF's,
and compositing programs. All costs in the model were based on reported data
from existing programs and were expressed in annualized costs per ton of
recyclables collected assuming 260 operating days per year. The costs
reflect annual operating costs and amortized capital costs assuming a three
percent real discount rate.* Where the reported capital expenses were listed
as depreciation charges, the depreciation charges were used for the annual
capital costs. Finally, all costs were scaled to 1989 dollars.
2.1.8 Methodology for Estimating Recycling Program Costs
Cost estimates for each type of program were derived from statistical
analyses of empirical data from existing programs. Since recycling programs
can vary tremendously, the relationship between cost and several cost
determinants for each type of program were analyzed. The factors analyzed
were the size, design, level of technology, and collection method of the
recycling program. For example, the cost of a curbside collection program
would depend on size 1n terms of the quantity of material handled and the
number of households served. Also, program design considerations such as the
materials collected, pick-up frequency, and the degree of household
separation (commingled or separated) affect cost. Beyond this, factors that
contribute to the effectiveness of a program, such as whether participation
is voluntary or.mandatory, must also be considered.
After some preliminary analysis, the primary recycling program cost
determinant was found to be the program size in tons of materials handled.
*The amortization period was 20 years for buildings and 7 years for most
other costs (e.g., trucks and storage bins).
2-10
-------
While other factors may also be significant, the data were insufficient to
allow estimation of their effect on costs with any degree of confidence.
Therefore-, the relationship between the program size and cost was analyzed
for all the program types. From these relationships, cost functions were
determined relating the program's cost to its size in tons per day.
To define each program's cost function, regression analysis was used to
test the ability of waste throughput to predict recycling program costs.
Each program was tested with two functional forms: a linear relationship,
and a linear relationship in logarithmic form. The better of the two cost
predictors was used in the model, and if neither was a good predictor of
cost, the cost was set to the sample average in dollars per ton.
2.1.7 Curbside Collection
Curbside collection programs involve collection of recyclables at the
point of generation. Households typically place materials in bins or bags at
the curb on specific recycling days or on trash days. The materials
collected in the curbside programs were assumed to be container glass,
newsprint, aluminum and steel cans, certain PET and HDPE bottles, and
corrugated cardboard (see earlier section in Chapter 2 for a description of
the recyclables assumptions). The hauler was assumed to collect the material
and transport it to a central facility where the materials are unloaded.
Materials may be sold as they are collected, or they may be processed at an
MRF.
Cost data were compiled for curbside programs from secondary sources
ranging from trade journal articles to municipal reports on curbside
collection. Although there are more than 1,500 curbside programs in
operation across the United States,* reliable data on their costs are scarce.
Of the 60 curbside programs initially evaluated, the cost data necessary to
calculate a total cost per ton were only found for 31 programs. The final
database included programs serving populations ranging from 1,186 to
500,000 people.
The cost of curbside collection programs is positively related to the
tons of material collected. Increasing the quantity of recyclables collected
results in a larger program with higher capital costs, as well as higher
operating costs due to the increased labor needed for collection, processing,
and administration. Curbside programs do demonstrate economies of scale,
however. Although the total program costs increase with throughput, the cost
2-11
-------
per ton declines because capital and labor can be used more efficiently in
the larger programs.
The regression analysis resulted in a curbside collection cost function
of:
Y = 132.6 * Q-0.235
where Y represents the annualized cost per ton and Q represents the quantity
of recyclables collected in TPD. Program sizes in the database range from
0.26 to 22.6 TPD, yielding a program costs per ton range of $218 to $64. For
larger programs (greater than 22.6 TPD), costs were assumed to remain flat at
$64 per ton because data were inadequate to justify an assumption that
economies of scale would continue to decrease the cost per ton.
2.1.8 Material Recovery Facilities (MRF's)
In this analysis, MRF's were used to provide a centralized point for
processing recyclables collected in a curbside program. The MRF's separate
and prepare materials for marketing to end users such as paper mills or glass
plants. Their primary function is to improve the marketability of collected
materials (by cleaning, separating, baling, etc.) and therefore command
higher prices for the recyclables.
Cost data were compiled from a MRF database that reported considerable
information on operating characteristics as well.5 The database included
entries for 40 existing MRF's and 64 planned MRF's. However, complete data
on capital and operating costs were available for only 51 of those
facilities. If available, tipping fees (the price charged to process
recyclables) were used as the indicators of MRF costs. If tipping fees were
not reported, cost estimates were based on reported capital and operating
costs. The final database included 51 programs ranging in size from 2 to
385 TPD.
A statistical analysis revealed that for MRF's, size was a poor
predictor of cost. Since no reliable function relating size and cost could
be derived, the MRF cost was set to the sample average in dollars per ton.
The observed costs per ton ranged from $1.56 to $71.46 per ton, with an
average cost of $21.79 per ton. Therefore, the estimate of MRF cost was set
at a flat rate of $21.79 per ton.
2-12
-------
2.1.9 Centralized Composting
It was assumed that yard waste was collected at the curb and processed
in centralized composting facilities. Composting programs vary in the types
of yard waste collected and the level of technology employed. Higher
technology approaches use machines to turn and aerate the compost, whereas
the lower technology approaches involve no machinery. Although both
approaches produce the same result, the lower technology approach requires
more land and time to yield the same quantity of compost.
Cost data were compiled from secondary sources, including EPA studies,
municipal studies, and trade journals. Although more than 600 centralized
composting programs operate in the United States,^ reliable data were
available for just 50 programs. Of these 50 programs, complete cost data
were available for only 10 programs.
The cost of composting programs is positively related to the quantity of
yard waste collected. Similar to curbside collection of recyclables,
increasing the quantity collected results in both higher capital and
operating costs. However, composting programs also demonstrate economies of
scale, causing the cost per ton to decline with increasing quantity
collected.
Regression analysis resulted in a centralized composting program cost
function of:
Y = 60.30 - (1.66 * Q)
where Y represents the annualized cost per ton and Q represents the quantity
of yard waste collected in TPD. Program sizes in the database ranged from
0.45 to 25 TPD, yielding a program costs per ton range of $60 to $19. For
larger programs (greater than 25 TPD), it was assumed that costs would remain
flat at $19 per ton.
2.1.10 Selection of Program Alternatives
The assignment of MWC's to each type of program was determined by the
net cost of the.various program options (i.e., program costs net of scrap and
disposal offsets as appropriate). The model's objectives was to achieve the
25 percent target in the least expensive way. MWC's were assigned to either
compost plus curbside or compost plus curbside plus MRF, whichever was
cheaper. The MRF may have been a good investment if scrap prices were
relatively strong in the region and further processing of the scrap would
2-13
-------
increase revenues enough to cover the capital and operating costs of the MRF.
The modeling is discussed in more detail later in the chapter.
2.1.11 Limitations
Although there are a large number of recycling programs in the
United States, reliable and complete cost data are available for only a small
number of programs. In addition to the small number of programs for which
reliable data are available, it is uncertain how representative the database
sample is of the total population. Both of these factors contribute to the
uncertainty of the cost function estimates. These and other limitations are
discussed in more detail in Chapter 4.
2.1.12 Other Costs
The model also included costs for two other activities resulting from
the regulation: measuring compliance with the 25 percent target, and
administering the program. These costs were imposed uniformly on MWC's based
on several simplifying assumptions.
Measurement costs were computed by assuming that all MWC's would install
electronic scales to weigh incoming loads. The MWC would keep records of
past and current acceptance rates to document changes in the unit's
throughput and records of materials collected to show that the diversion rate
was being achieved. Costs were included only for the purchase, maintenance,
and operation of the scale; no recordkeeping costs were added. Annualized
costs were set at $3,800 per year assuming a 3 percent discount rate.ฎ
Administrative costs were based on an annual cost assumption of $35,000
per 25,000.population in the service area." Waste throughput was converted
to population using a generation rate of 0.68 tons per person per year.
2.1.13 Avoided Costs
Establishment of a source separation and recycling program would provide
savings to the community in three different areas:
Avoided landfilling costs for waste shifted from land disposal to
existing MWC's, assuming that existing MWC's would seek to maintain
the current MWC throughput by expanding the area from which waste
was collected. Communities woufd save the cost of landfilling
waste that was shifted to the MWC.
*Costs were based on site visits conducted in New Jersey.
2-14
-------
Savings resulting from "downsizing" planned MWC's. Because over
28 percent of the planned MWC throughput would be diverted through
recycling, the planned MWC's could be redesigned smaller than
originally intended, saving on construction and operation costs.
Savings on trash collection at all MWC's. The waste diverted from
the combustor by the recycling program would not be collected as
regular trash, thus saving on collection costs.
2.1.14 Landfill Savings at Existing MWC's
The most important assumption involved in computing this savings was the
modeling assumption that existing MWC's would expand their service areas to
collect more waste than they do already. All of the "new" waste brought into
the service area was assumed to be drawn from nearby landfills, so the
avoided cost should be appropriate to those neighboring landfills.
Estimates of the full cost of disposal developed by the Office of Solid
Waste (OSW) in support of the Regulatory Impact Analysis for revised criteria
for municipal solid waste landfills were used to estimate the avoided
landfilling costs.? jhe cost estimates included both the current or baseline
cost of disposal plus incremental costs attributable to the revised criteria.
The costs were specified for seven different size categories of landfills and
for five alternative design scenarios that address the range of designs
likely to be required after the regulations are effective. Weighted average
costs were computed based on the predicted incidence of each landfill design.
Averaging costs across size categories, however, was deemed unreliable
because of the broad range in costs across sizes (i.e., the substantial
economies scale in landfill costs). Disposal costs per ton varied by as
much as a factor of 10 between the largest landfill size category and the
smallest. For this analysis, therefore, disposal cost savings were estimated
under two different scenarios, each using a different aggregation of these
costs across sizes.
A landfill-by-landfill average of costs resulted in a disposal cost
estimate of $62.60 per ton. If a tonnage-weighted average was used (each ton
of waste disposed counts equaTly instead of each landfill unit), the disposal
cost estimate was $20.24.8 The landfill-based average is simply too high,
given the huge fraction of waste disposed in the largest landfills. In
addition, MWC's are likely to be located in high population, high density
areas where larger landfills are more typical.
2-15
-------
To define a range of disposal cost savings for this analysis, the
average of these two estimates ($41.42 per ton) was used as an upper bound
and the tonnage-weighted estimate of S20.24 per ton was used for the lower
bound. The costs were computed as the savings per ton of waste diverted from
the landfills. The total tons diverted equalled only those recyclables that
were collected and sold; the fraction of the recyclables that were not
marketable (the contaminants) were eventually returned for disposal, so no
savings was credited.
These estimates assumed that the full cost of disposal was "avoidable,"
that is, removing a ton of waste from a landfill would save the full cost per
ton of disposal in that landfill. In fact, only a portion of the cost may be
avoidable. Nevertheless, this range was used to bound the disposal cost
savings for this analysis. The results were quite sensitive to these
assumptions, as shown in Chapter 3.
2.1.15 "Downsizing" Savings at Planned MWC's
MWC's that are still in the planning stage could accommodate the
projected decrease in throughput most efficiently by simply reducing the
scale of the project. For this analysis, this approach was assumed to be
adopted by all planned MWC's. The savings were estimated by estimating the
total annualized cost of constructing and operating planned MWC's, given
compliance with the air emission requirements, and then computing the savings
in these costs if waste flows were reduced.
The savings was computed in terms of tons of waste diverted from the
planned MWC's and was set at $37.59 per ton. As with the avoided landfilling
costs, the unit savings was multiplied by the actual tons of recyclables
diverted and sold. Contaminants in the recyclable mix did not produce
savings.
2.1.16 Trash Collection Savings at all MWC's
Any material collected and sold under a recycling program would no
longer be collected by a trash hauler, so savings were computed for each ton
diverted and sold, whether the;waste was destined for an existing MWC or a
planned unit.
The estimated savings per ton was computed in two stages. The first was
to identify the current cost of trash collection and the second was to
compute the portion of those costs that could be avoided if waste volume
declined. The current cost of trash collection was computed from empirical
2-16
-------
data of collection costs from a national survey.9 y^e survey was not an
exhaustive one, as costs were extracted for collection services for only
63 communities across the country serving populations ranging from 2,000 to
nearly 1 million. A significant relationship between community size and cost
was not apparent, so an average was calculated for the 63 communities of
$45.30 per ton (in 1989 dollars). The actual data ranged between costs of
$11.80 per ton and $83.60 per ton.
In this analysis, it was assumed that only a fraction of the trash
collection cost was avoidable. Since many aspects of the collection program
would remain unchanged (e.g., the routes still must service the same area and
the same stops), only certain elements of the program's cost could be saved
by reducing the average amount of waste collected per stop. Only a limited
amount of data were available describing collection costs and none of the
sources explicitly dealt with savings from reduced waste flows.
In the absence of independent data on avoided trash collection costs,
the avoidable percentage of trash collection cost was linked to the avoidable
percentage of disposal costs. Two different percentage factors were used to
adjust costs. In the short run, it was assumed that only variable costs
could be avoided; across the landfill sizes analyzed, the avoided cost
percentage was 24 percent or $10.87 per ton when applied to the trash
collection cost. Over the long term, savings would be possible on capital
equipment as well; given the cost functions for landfills and a 25 percent
waste diversion, the avoided cost percentage was 59 percent or $26.73 per ton
when applied to trash collection costs. As with the avoided disposal and
downsizing savings, the savings per ton were applied only to recyclables that
were collected and sold. Using these landfill-based percentages for
computing avoided trash collection cost introduced uncertainty in the
estimates, since the approach implies that the cost structure for collection
and disposal are similar.
2.2 SCRAP REVENUE
The model also included scrap revenue for most of the materials
collected by the recycling programs. Scrap prices for various commodities
were taken from region-specific price summaries reported regularly in
Recycling Times. The only exception was for yard waste/compost, which was
2-17
-------
assigned no market value. Price data were compiled for the first 7 months of
1990 only.
These scrap prices cover seven regions of the country and represent
quotations provided by various purchasers of scrap materials during the given
period (i.e., prices they paid). The sampling is not random and no
statistical conclusions can be reliably drawn from the data, but they do
provide indications of regional price levels and trends over time. If a
range of prices was quoted for a given region, the newspaper reports the
highest and lowest quotes they receive. In addition, data on scrap prices
across the country are not readily available, so this provided a quick and
inexpensive source of the price information.
2.2.1 Modeling Approach
Scrap prices for each material were computed for two different markets.
The first market was prices paid by "processors," defined by Recvclino Times
as scrap dealers, brokers, scrap yards, and municipal centers. The second
set of prices was for "end users/mills," which are mills, foundries,
factories, or other plants where the material is actually reused. This
material was assumed ready to enter the plant for recycling as opposed to the
material sold to processors, which is likely to undergo further processing
and consolidation before it is shipped to an end user.
In the analysis of recycling progress, these two markets were used as
proxies to indicate the differences in potential revenue to programs that
sell materials to intermediaries for further processing and those that
process the material themselves and sell to end users. In the model,
communities choosing the curbside recycling option sell to intermediaries and
earn the processor prices, while communities with curbside plus a MRF earn
the end-user prices.
In addition to the different markets, scrap revenues were computed for
seven materials based on two different scrap price scenarios. A "current"
scrap price scenario was derived from the average scrap prices offered for
each material and for each of the seven regions. This scenario averaged the
prices paid over the first 7 months of 1990; if prices were reported over a
range, the midpoint of the range was used in the average.
Because of concerns with increased supply driving down prices, a lower
scrap price scenario was developed. Under this scenario the lower end of
price ranges reported in January through July 1990 was averaged. Clearly,
2-18
-------
prices could well fall below this range, but without additional research into
the price response of the various scrap markets to increasing supply, it was
not possible to estimate a lower bound. In addition, exogenous factors such
as energy prices and prices of substitute materials (e.g., virgin material
inputs) could dramatically affect scrap prices, making projections extremely
uncertain.
2.2.2 Price Data
Table 2-4 summarizes the scrap price data used in the analysis.
Regional price differences are apparent for every material; for corrugated
and newsprint, prices are negative in the Northeast and Mid-Atlantic at least
in some market scenarios. Revenues per ton are the highest, by far, for
aluminum and even through aluminum only accounts for a small share of
municipal waste disposed, aluminum revenue is still the largest single
contributor to total revenue. Again, these prices represent relative price
levels between materials and regions. Many other factors affect the prices
actually paid and the trend in prices in the future. As a result, the scrap
price estimates should be evaluated cautiously.
2.3 MODEL STRUCTURE
To conclude this chapter, the integration of each of the components of
the analysis described above is summarized. Most of the components of the
modeling simply contribute additional data to the database. The raw data
included only MWC location and size. From this point:
Waste quantities were adjusted to reflect larger service areas for
existing MWC's;
The quantity of each recyclable material generated within each
MWC's service area was computed based on the Franklin data* on
waste composition;
The potentially recoverable quantity and the actual quantity
diverted for each recyclable material in each MWC service area was
computed using the potentially recoverable fractions and
participation rates, respectively;
Scrap values for each recyclable material (given current and lower
scenarios and prices paid by both processors and end users) were
added. Prices were matched to individual MWC's based on the state
in which they were located; and
2-19
-------
TABLE
2-4.
SUMMARY
OF SCRAP PRICES USED
(dollars per ton)
IN THE
MWC
ANALYSIS
Current Scrap Price Scenario
Old Newsprint
Corrugated
Class Containers
Steel
Cans
A l umi nun Cans
PET
HDPfc
Region
Processor
<*)
End-
User
(S>
Processor
(*>
End-
User
<*>
' Processor
<ป)
End-
User
Processor
(*)
End-
User Processor
<$) <*>
End-
User
{ป>
Processor
<*>
End-
User
(*)
Processor
<*)
End
Usei
(*)
South
8
16
15
38
26
61
58
58
733
911
136
136
97
97
South-Centrat
7
18
21
41
9
60
67
67
753
939
100
100
86
86
Uest
9
32
27
60
53
64
56
56
832
1,255
121
121
60
60
Northeast
-10
18
3
27
25
33
58
58
641
1,011
156
156
92
92
East-Central
7 -
14
16
29
28
54
61
61
732
989
156
156
100
100
Uest-Central
8
21
16
28
27
55
54
54
610
963
'10
90
60
60
Hld-Atlantic
-5
12
9
29
17
53
58
58
677
951
1"V
179
105
105
Lower Scrap Price Scenario
Old Newsprint
Corrugated
Glass Containers
Steel
Cans
Aluninun Cans
PET
HDPE
Region
Processor
End-
User
(S>
Processor
(S)
End-
User
Processor
<*>
End-
User
<ซ>
Processor
<*>
End-
User Processor
<*> (*>
End-
User
(S)
Processor
(ป>
End-
User
<*>
Processor
(*>
End-
Use i
South
3
4
10
35
20
53
45
45
649
857
80
80
97
97
South-CentraI
1
16
14
36
6
50
63
63
667
883
100
100
86
86
West
4
30
17
56
19
64
49
49
480
845
81
81
60
60
Northeast
-25
10
14
21
11
18
54
54
543
993
127
127
80
80
East-Central
2
8
9
24
20
49
57
57
672
810
120
120
80
80
West-Central
1
20
6
20
17
54
50
50
544
861
80
80
60
60
Mid-Atlantic
-14
7
4
24
11
50
46
46
593
901
169
169
80
80
-------
The unit costs for recycling program components (yard waste
collection and processing, curbside collection of other
recyclables, and a MRF) were computed given the MWC-specific
quantities of recyclables (since costs depend only on throughput).
This database provides most of the information needed to select and assign
recycling programs to MWC's and then compute waste diversion and net costs.
The next modeling step was to identify the recycling program that would
minimize costs and meet the 25 percent diversion target for each MWC. For
this analysis, only one mix of programs to meet the target was focused on: a
combination of curbside yard waste collection and curbside collection of
other recyclables. The only variation available was the decision to build
and operate an MRF or not. As explained earlier, the model assumed that use
of an MRF would enable the community to earn higher prices for the scrap
material. For each MWC, the model tested whether the higher revenue would be
sufficient to cover the cost of the MRF. If so, an MRF was selected;
otherwise, no costs were incurred for an MRF.
Once the recycling program was selected, the final step was the
calculation of costs or savings for individual MWC's given all of the cost
components and cost offsets described in this chapter. Total costs included:
Recycling program costs;
Costs for scales to measure compliance with the diversion target;
and
Administrative costs for local governments.
Cost offsets included*:
Avoided landfilling costs for existing MWC's;
Downsizing savings for planned MWC's;
Avoided trash collection costs; and
Scrap revenues.
The total costs offsets were computed by multiplying the unit cost savings
or revenue by the number of tons of recyclables collected and actually sold
As a result, the tonnage estimates for recyclables sold and tons diverted
from landfills were adjusted downward to take account of waste in with the
materials.
2-21
-------
Net costs/savings were modeled only for the 171 MWC's that were assumed
to incur incremental costs as a result of the federal requirement. The
results of the modeling are summarized in the next chapter.
2-22
-------
3. RESULTS
This discussion focuses on the 171 MWC's (out of a total population of
280 MWC's) that were assumed to face incremental costs from the federal
requirement. The MWC's that were excluded were those located in states with
current or impending 25 percent recycling requirements or those with
capacities of less than 40 tons per day.
3.1 WASTE DIVERTED
For the 171 MWC's with incremental costs, 11.7 mi 11 on tons are collected
through the various programs. This accounts for 28.6 percent of the total
number of tons of waste generated in the service areas around these MWC's
(40.9 million tons of total generation). This total generation estimate
includes waste generated by residents within the expanded service areas
assumed for existing MWC's. Out of the 11.7 million tons collected,
2.7 million tons is diverted from planned MWC's and 9.0 million tons is
diverted from existing units.
While this estimate equals the tonnage collected in recycling programs,
it overstates the total quantity of recyclables actually marketed because of
contaminants in the recyclables. Scrap revenues and avoided disposal costs
were computed after correcting for contamination, since the contaminants are
ultimately discarded and not sold.
One additional consideration in-the diversion estimates is the
10 percent cap on yard wastes' contribution toward the 25 percent target. We
assumed that yard waste programs would collect as much material as possible,
even though only a portion of it would count toward the target. As a result,
some of the material diverted provides trash collection and disposal cost
credit, even though it does not count toward the 25 percent target. The
total waste diverted that counts toward the target is only 10 million tons.
3.2 NET COSTS
The net costs (collection and administrative costs minus scrap revenue
and other savings) were computed under two different scenarios given varying
assumptions about scrap prices and the avoided trash collection and disposal
costs. The results summarized in Table 3-1 represent the highest and lowest
3-1
-------
TABLE 3-1. COST ESTIMATES FOR MUNICIPAL WASTE COMBUSTOR SOURCE
SEPARATION RULE3
Low estimate
(Scenario 1)
($)
High estimate
(Scenario 2)
($)
Separation costsb
668
669
Avoided Disposal
Landfilling (Existing (MWC's)
Downsizing Planned MWC's
Trash Collection
(356)
(96)
(298)
(174)
(96)
(121)
Scrap Revenues
(263)
(220)
Net Cost (Savings)
(345)
58
Cost (Savings) per ton diverted
(29)
5
Cost (Savings) per ton combustedc
(12)
2
aBased on 171 MWC's.
^Includes collection and processing of matrials and administrative costs.
cTotal cost divided by tons of MWC throughput after the regulation.
3-2
-------
net costs scenarios; Scenario 1 reflects the lowest cost scenarios (because
the higher costs offsets are assumed for scrap value, trash collection, and
landfill.avoided costs) while Scenario 2 reflects the highest cost scenario
(because it applies the lower cost offsets).
Separation cost estimates are dominated by collection and processing
cost for recyclables, which account for 87 percent of the total. Costs for
administration of the program ($84 million annualized) and measuring
compliance ($0.6 million annualized) account for the remaining 13 percent of
the total. The slight difference between the separation costs in the two
scenarios results from the different scrap value assumptions. Changes in
scrap revenue may change in the selection of the lower-cost curbside program
(with or without an MRF), so the different separation costs reflect a
slightly different mix of collection programs in the two scenarios.
The cost savings are distributed across all of the categories shown in
Table 3-1. Under Scenario 1 avoided landfilling'costs account for the
largest share of savings ($356 million annualized or 35 percent), followed by
avoided trash collection costs (29 percent), scrap revenues (26 percent), and
downsizing credits (10 percent). Total offsets under Scenario 1 are
$1,013 million. For Scenario 2, total offsets drop to $611 million and are
dominated by scrap revenues (36 percent), avoided landfilling costs
(28 percent), avoided trash collection (20 percent), and downsizing credits
(16 percent).
The two modeling scenarios demonstrate the sensitivity of the results to
the offset.assumptions. Scrap revenues do not vary dramatically; the
difference between the two scrap scenarios is about $43 million per year.
The disposal/collection assumptions do make a substantial difference,
however. The alternative assumptions account for a $182 million difference
in avoided disposal costs for existing MWC's and a $177 million difference in
avoided trash collection costs. Downsizing savings are the same in all
scenarios.
The net costs of the tworscenarios vary over a large range, from an
annualized net cost of $58 million to a savings of $345 million. This range
implies that the net effect of the regulation is somewhat uncertain; the cost
and savings categories that appear relatively constant across scenarios are
also subject to variation since recycling program costs can vary dramatically
and scrap prices (as with any commodity prices) can fluctuate wildly. Taking
3-3
-------
these estimates as an indication of national costs and savings, however, they
provide a reasonable bound on the cost consequences of the regulation.
Table 3-1 also shows the net costs (savings) per ton of waste diverted
through the program and per ton of waste combusted in the 171 MWC's after the
regulation. Results per ton diverted range from a savings of $29 per ton to
a cost of $5. The incremental cost/savings per ton of waste burned after the
regulation ranges from a savings of $12 per ton combusted to a cost of $2 per
ton.
3-4
-------
4. LIMITATIONS OF THE ANALYSIS
As with any modeling effort that simulates regulatory responses, the
cost results are quite sensitive to certain assumptions. Modeling
site-specific responses to the federal source separation requirements is
particularly difficult because of the range of local issues that will
influence the ultimate reaction of the MWC's. Given a large enough
population, however, the estimates of national level cost should be
reasonable predictors of the actual outcome. In this analysis, alternative
scenarios were introduced to capture some of the uncertainty inherent in
computing the costs of source separation programs. Other factors that were
not explicitly considered in defining the scenarios may also affect costs;
all of these limitations should be considered when interpreting the model
results.
4.1 WASTE DATA
The results are sensitive to the assumption that existing MWC's
would comply with the separation requirement by expanding their
service areas and diverting the same quantity of waste that they
add to their service area. Because of this assumption, revenues
from the sale of power are unaffected by the requirement and
savings are included for avoided landfilling costs and trash
collection costs for the jurisdiction from which the waste is
transferred.
Waste characterization data are assumed to be constant across the
United States. No regional differences 1n waste composition are
included in the analysis, which could affect the ability of some
communities to reach the 25 percent target and could affect the
revenue earned from recycling.
Data on participation rates and recoverable fractions are based on
limited empirical experience. Many factors contribute to
participation rates,; ranging from climate to socioeconomic and
educational levels of the participants. Changes in either of these
fractions would affect the ability of communities to reach the
25 percent target and could affect revenue earned from recycling.
4-1
-------
The model results reflect collection costs and savings based on
more than the 25 percent requirement. Because of the 10 percent
cap on credit provided for yard waste, the actual tons diverted,
the collection program throughput, and the cost offsets are based
on a higher diversion rate.
The cost estimates are probably the most sensitive to the first
limitation. Because tonnage from existing MWC's dominates the diversion
estimates, changes in the compliance assumptions for those units could change
the cost estimates substantially.
4.2 RECYCLING PROGRAMS AND COSTS
The data on costs for recycling programs are extremely limited.
Empirical data from secondary sources are difficult to verify and
may be biased in an unknown direction. Several different issues
arise in this context:
Programs were chose based on data availability and adequacy,
not random selection. As a result, the costs may not be
representative of existing programs.
The number of cases was limited for all of the programs, so
that statistically significant associations with independent
variables were difficult to establish. A number of variables
probably affect curbside costs, for example, but a significant
association is only shown between size and cost.
Based on experience with municipal landfill costs,
self-reported costs in the literature are probably below
actual costs. For community-operated programs in particular,
data on full costs are not always available and capital
components often go under-reported or unreported.
The selection of alternative recycling programs is limited in the
model. Only a few different styles of programs were examined,
while different permutations of these programs and many other
program types could provide a more efficient means of collecting
and processing recyclables.
Administrative costs were significant, yet do not reflect any
economies of scale. As a result, these costs may be overstated.
Overall, the total cost of the recycling programs is probably
understated because of systematic under-reporting of actual program costs.
Two other considerations tend to overstate costs, however. First, a more
diverse set of recycling program options would allow for some lower-cost
response to the regulation, thus decreasing actual compliance costs. Second,
administrative costs are overstated, so actual costs would be less. Despite
4-2
-------
these two corrections, the under-reporting of costs has a more significant
effect on the total cos-t estimate.
4.3 AVOIDED COSTS
4.3.1 Avoided Landfillinq Costs
The disposal cost savings for all existing MWC's is assumed to
result from avoided landfilling costs; this implies that additional
waste is drawn from landfills near the MWC in order to maintain the
current throughput of the MWC. If expansion of the service area
were infeasible, waste would have to be diverted from the MWC
itself.
Key areas of uncertainty are the size of landfills from which waste
is diverted and the fraction of the landfills' costs that is
avoidable.
4.3.2 Downsizing Credits
Planned MWC's were assumed to downsize in anticipation of the
source separation requirement and would avoid the appropriate
capital and operating costs associated with a larger unit. If
planned MWC's responded as the existing units were modeled (i.e.,
diverting waste from neighboring jurisdictions), the avoided costs
would be different.
4.3.2 Avoided Trash Collection Costs
Current trash collection costs were assumed constant for all
jurisdictions; in fact, differences in demographics and
type-of-service across communities would affect current collection
charges.
The avoided collection costs fractions were based on assumptions
arbout landfilling costs. While the magnitude of the estimates
appears reasonable, further research into the actual share of costs
that could be avoided is necessary.
4.4 SCRAP PRICES
Scrap prices may move outside of the range defined for a number of
reasons. First, in some regions, sudden increases in the quantity
of collected scrap may overwhelm demand for the material, driving
down prices at least in the short run. Prices may also increase or
decrea.se for several 'exogenous reasons. Markets for many of these
materials are volatile and respond to changes in the underlying
market conditions for the primary materials that they replace. As
a result, prices for these commodities could change dramatically,
independent of the future supply of recyclables. .
Prices quoted are FOB the purchaser and may not reflect actual
revenues if the seller must incur significant transportation costs.
4-3
-------
Material collected and processed is assumed to be of sufficient
quality and quantity to command the prices quoted. Prices for
small programs or for contaminated materials could be well below
these estimates or even negative.
4.5 SUMMARY
Because of a lack of data, the analysis 1s based substantially on
national averages for waste data, costs, and most of the cost savings
components. While this will introduce errors into the net costs estimates
for individual MWC's, on a national basis, the estimates should be indicative
iv
of aggregate results once the rule is in place.
On balance, the analysis probably understates the cost of the regulation
because reported costs for recycling programs are probably underestimated and
avoided disposal costs and are probably overstated. Other issues, such as
the magnitude of avoided trash collection costs, would have an uncertain
effect on the net cost. The range of costs/savings would remain large, but
would shift upward, indicating a higher aggregate cost at one extreme and a
smaller aggregate savings at the other extreme.
4-4
-------
5. REFERENCES
1. "Characterization of Municipal Solid Waste in the United States,"
1990 Update, Franklin Associates Ltd., for U.S. EPA, EPA/530-SW-90-042A.
2. Several studies are also summarized in Facing America's Trash: What
Next for Municipal Solid Waste?. Congress of the United States, Office
of Technology Assessment, October 1989.
3. 1990-91 Materials Recovery & Recycling Yearbook. Governmental Advisory
Associates, Inc., New York.
4. Jim Glenn, "The State of Garbage in America", Biocycle. March 1990,
p. 50.
5. .1990-91 Materials Recovery & Recycling Yearbook (database), Governmental
Advisory Associates, New York.
6. Costs were compiled from personal communications with Shirley Smith,
DPRA, Incorporated, April 27, 1990; "Measuring Source Reduction and
Recycling by Weighing MSW at Landfills," Memorandum to Doreen Sterling,
QSW, from Carolyn Kaplan and Randy Freed, ICF, Incorporated,
May 7, 1990.
7. Cost data were extracted from the "Draft Final Regulatory Impact
Analysis of Revisions to Subtitle 0 Criteria for Municipal Solid Waste
Landfills," prepared for the Economic Analysis Staff, Office of Solid
Waste, July 31, 1990 draft.
8. Draft Economic Analysis, pp. 3-18 to 3-19.
9. "Survey of Solid Waste Charges," prepared for the City of Worcester,
Massachusetts, February 1990.
5-1
------- |