Skip Navigation | |
January 31, 2003
Dear Mr. Chairman and Dear Congressman: The Congressional Budget Office's (CBO's) recent analysis titled Future Investment in Drinking Water and Wastewater Infrastructure confirms that current funding from all levels of government and current revenues generated from ratepayers will not be sufficient to meet the nation's future demand for water infrastructure. That analysis highlights the major uncertainties involved in projecting the level of future spending by providing both low-end and high-end estimates of average annual costs of investment and operations and maintenance from 2000 to 2019. The Environmental Protection Agency (EPA) also recently published low-end and high-end estimates of the future costs of water infrastructure in its report titled The Clean Water and Drinking Water Infrastructure Gap Analysis. Following your request of October 2002, CBO has analyzed how its estimates compare with those published by EPA. The report attached to this letter highlights key similarities and explains important differences in the estimates offered by the two agencies. Natalie Tawil of CBO's Microeconomic and Financial Studies Division wrote the report under the supervision of Perry Beider, David Moore, and Roger Hitchner. Christine Bogusz edited the report, Leah Mazade proofread it, and Kathryn Winstead prepared it for publication. Annette Kalicki produced the electronic versions for CBO's Web site. I hope this report will be useful to you. Please call me if you have any questions, or have your staff contact the author directly. Sincerely, Barry B. Anderson
Attachment
SummaryBoth the Congressional Budget Office (CBO) and the Environmental Protection Agency (EPA) conclude that projections of the future costs associated with drinking water and wastewater infrastructure are highly uncertain. Because available data are limited, the agencies must use many assumptions to develop their projections, and the 20-year projection window provides ample opportunity for unforeseen developments to influence costs. Data limitations make it impossible for the agencies to know even baseline investment costs with certainty. Given that level of uncertainty, the differences between EPA's and CBO's projections of total investment costs are not especially significant; both EPA's $46.5 billion high-end estimate and CBO's $41.0 billion high-end estimate reflect a near doubling of baseline investment costs through 2019. The agencies' low-end estimates ($25.0 billion and $24.6 billion, respectively) reflect less than a 15 percent increase in costs through 2019 for investment in drinking water and wastewater.(1) Despite the similarity between EPA's and CBO's estimates of future investment costs, a few distinctions are worth noting. On the high end, if EPA had incorporated investment efficiency savings and made a more plausible assumption about the extent to which drinking water systems will use debt financing for investments from 2000 to 2019, its estimate for investment in drinking water infrastructure ($23.8 billion) would be closer to CBO's ($20.1 billion). On the low end, a case might be made for EPA's lower drinking water estimate ($8.9 billion) and CBO's lower wastewater estimate ($13.0 billion). EPA's lower estimate for drinking water merits attention because the agency's relatively low $52 billion estimate of 20-year pipe costs is the only one based on actual data on the ages of pipes from multiple systems. On the wastewater side, CBO's lower estimate reflects savings from increased efficiency in investment based on domestic and international evidence. EPA does not include such efficiency effects. The agencies' high-end estimates of operations and maintenance (O&M)
costs represent the most important point of divergence in their projections
(see Figure 1). That difference stems from
the investment side where, in CBO's view, EPA adopts the unrealistic assumption
that drinking water infrastructure is replaced in large quantities early
in the 20-year period rather than being replaced more evenly throughout
the span. That assumption, together with an estimating methodology that
predicts future O&M spending on the basis of the historical relationship
between O&M expenditures and the capital stock, yields a high-end O&M
estimate that exceeds CBO's by $25 billion on an average annual basis,
or $500 billion over 20 years. Consequently, EPA's high-end estimate of
the increase in future investment and O&M spending from 2000 to 2019
($59.4 billion on an average annual basis or $1.2 trillion over the 20-year
period) is 58 percent higher than CBO's ($37.5 billion on an average annual
basis or $750 billion over the 20-year period). In contrast, EPA's high-end
estimate of the increase in future investment costs alone ($23.1 billion
on an average annual basis or $462 billion over 20 years) is only 19 percent
higher than CBO's ($19.4 billion on an average annual basis or $388 billion
through 2019).
Comparing CBO's and EPA's Estimates of Future Spending on Water InfrastructureThe nation's drinking water and wastewater systems, and their ratepayers, face a major challenge in funding the replacement and maintenance of aging infrastructure over the next several decades. Assessing the amount and timing of future costs is difficult because information about existing water infrastructure on a national scale is scarce. For example, there is no accessible inventory of the age and condition of pipes for even the relatively few large systems that serve most of the country's households. Furthermore, there are many uncertainties inherent in long-range projections. The Congressional Budget Office and the Environmental Protection Agency have each published estimates of projected costs of investment and operations and maintenance for the nation's water system infrastructure during the two decades from 2000 to 2019.(2) Both CBO and EPA believe that to best convey the challenge facing water systems and their ratepayers at a given time, investments should be measured in terms of costs as financed. Costs as financed conveys the full costs of investments made out of funds on hand (on a pay-as-you-go basis) during the period analyzed and the debt service (principal and interest) paid in those years on new and prior investments that were financed through borrowing.(3) CBO's and EPA's low-end and high-end estimates of costs as financed reflect the significant uncertainty underlying the projections.(4) Compared with estimates of baseline expenditures (annual expenditures before 2000), EPA's projections imply an average annual increase in costs over the 2000-2019 period that ranges from 2.8 percent to 85.8 percent for investment and operations and maintenance combined (see Table 1). The implied increase in investment costs ranges from 6.8 percent to 98.7 percent; the implied increase in O&M spending ranges from 0.7 percent to 79.3 percent. EPA's low-end projection of total investment and O&M expenditures
($71.1 billion) does not differ greatly from CBO's ($70.7 billion); for
investment and O&M separately, EPA's low-end projections basically
match those of CBO. In contrast, at $128.4 billion, EPA's high-end projection
of investment and O&M expenditures exceeds CBO's by $30.4 billion--a
cumulative difference of $608 billion over 20 years. The two agencies'
high-end estimates of investment costs do not differ dramatically; each
estimate represents nearly a doubling of (admittedly uncertain) baseline
expenditures through 2019.(5)
But EPA's projected O&M expenditures exceed CBO's by $25 billion on
the high end--a cumulative difference of $500 billion over the 20-year
period. To evaluate the relative importance of the differences between
EPA's and CBO's estimates, this report first assesses the key factors that
contribute to those differences.
Investment Costs as Financed from 2000 to 2019Compared with CBO's projections and measured against a relatively similar
baseline estimate, EPA's low-end and high-end projections of future increases
in investment costs for water infrastructure portray a greater range of
uncertainty (see the top panel of Figure 2).
EPA's low-end estimate for total investment from 2000 to 2019 essentially
matches CBO's; the agencies' slightly different estimates of baseline expenditures
result in projected low-end increases of $1.6 billion ($25.0 billion minus
$23.4 billion) for EPA and $3.0 billion ($24.6 billion minus $21.6 billion)
for CBO. The agencies' high-end projections differ from each other to a
greater extent. EPA's $23.1 billion estimate ($46.5 billion minus $23.4
billion) of the increase in costs exceeds CBO's $19.4 billion estimate
($41.0 billion minus $21.6 billion) by $3.7 billion, resulting in a total
difference of $74 billion over the 20-year span.
In generating their estimates, EPA and CBO make assumptions about the values of many factors that influence future costs. For most factors, a strong case cannot be made for the choices of one agency over the choices of the other; one can, however, point to and evaluate a few choices that contribute significantly to the differences between EPA's and CBO's estimates. Drinking Water
Four key factors drive the wider range of costs that EPA projects relative to CBO's estimated costs. First, on the low end, EPA's estimate of investment in drinking water pipes is 71 percent below CBO's. EPA's estimates of 20-year pipe costs range from $52 billion to $352 billion; CBO's range from $178 billion to $331 billion. The discrepancy between the agencies' low-end estimates is attributable to different time profiles for the replacement of existing pipes. EPA's estimate reflects an annual rate of pipe replacement (determined by the projected aging of transmission and distribution lines) based on data on the ages of pipes from 20 systems.(7) Relative to the approaches of estimating the cost of pipe replacement taken by CBO and by EPA in its other scenarios, EPA's low-end estimate has more investment in pipes occurring after 2019.(8) Its figure of $52 billion is less than half of the next-lowest estimate of $120 billion, an alternate EPA estimate of pipe replacement costs that reflects survey results that were adjusted to compensate for documented underreporting of 20-year needs by medium-sized and large drinking water systems.(9) CBO's approach, which is based on information from Stratus Consulting, essentially applies assumptions about the average annual rate of pipe replacement (0.6 percent at the low end and 1.0 percent at the high end) to an estimate of the national pipe inventory at a cost per foot based on pipe size.(10) EPA also uses that basic approach to generate some of its other estimates, making alternate assumptions about the average annual rate of pipe replacement and using different estimates of the national pipe inventory and the cost per foot. Second, EPA's high-end estimate of investment costs for drinking water exceeds CBO's partly because EPA's estimate of 20-year nonpipe costs and the costs of recently promulgated and proposed regulations ($97.6 billion combined) is 25 percent higher than CBO's high-end estimate ($78.1 billion).(11) Both agencies based their estimates on figures from EPA's 1999 Drinking Water Infrastructure Needs Survey. EPA adjusted those figures upward to compensate for the documented underreporting of 20-year nonpipe costs by medium-sized and large drinking water systems.(12) CBO agrees that some adjustment to the survey data is appropriate given the limitations of the reported figures; many respondents, relying on planning documents covering one to 10 years, may have been unable to adequately document costs associated with investments slated for later in the 20-year period. The lack of such an adjustment in CBO's analysis lowers its estimate relative to EPA's. Third, EPA does not allow for savings from increased efficiency in investment, but CBO does. CBO's high-end estimate reflects 5 percent savings attributable to factors such as asset management and innovative contracting for construction. Evidence of the potential impact of efficiency savings on investment costs comes from systems abroad as well as from domestic systems.(13) Fourth, on the high end, EPA assumes that drinking water systems will use debt to finance only 35 percent of their investments from 2000 to 2019; in contrast, CBO assumes that systems will do more to defer 20-year costs, using debt to finance 70 percent of their investments.(14) As investment programs grow in size, systems aiming to keep rates low can reduce their use of pay-as-you-go financing in favor of borrowed funds; at present, systems undertaking large amounts of investment generally use debt financing for all but a small share (often just a few percent) of the total.(15) Although EPA's $52 billion estimate of 20-year pipe costs is an outlier compared with other estimates, it is the only one based on actual data on the ages of pipes from multiple systems. Even though those data are drawn from a rather small sample of 20 systems, EPA asserts that the sample "likely represent[s] a broad range of systems of various ages from across the country." Accordingly, EPA's low-end estimate of drinking water systems' investment costs warrants serious consideration. Conversely, EPA's high-end estimate of those costs may be too high. Adjusting CBO's high-end estimate for documented underreporting of nonpipe investment costs in the Drinking Water Infrastructure Needs Survey would boost it to $21.0 billion--still $2.8 billion lower than EPA's estimate ($56 billion lower over the 20-year projection period).(16) If EPA had incorporated efficiency savings and made an assumption that was more consistent with current practices about the extent to which drinking water systems would use debt financing for investments from 2000 to 2019, its high-end estimate would more closely match CBO's. Wastewater
EPA's and CBO's baseline estimates--the former's reflecting costs as financed in 1996 and the latter's reflecting costs as financed in 1999--differ primarily because EPA assumes that wastewater systems used much more debt financing in the years preceding 1996.(17) EPA assumes that 90 percent of nongrant expenditures from 1977 to 1996 were financed through debt. CBO assumes that 50 percent of nongrant expenditures in 1980 were debt financed, and it increases that figure by 1 percent each year so that debt financing accounts for 69 percent of nongrant expenditures in 1999.(18) Consequently, EPA's estimate of baseline expenditures reflects a much higher level of debt service associated with the investments undertaken in prior years. EPA asserts that wastewater systems have historically used debt to finance 90 percent of their capital stock. CBO's debt-financing assumptions are common to both its drinking water and wastewater analyses and build on two small surveys of water systems.(19) Given the dearth of data, it is difficult to determine the relative accuracy of the two approaches. For its high-end projections, EPA's higher estimate of investment costs as financed is also driven by financing assumptions. Measured in terms of average annual investment resource costs rather than costs as financed, EPA's estimate ($22.5 billion) essentially matches CBO's ($22.3 billion).(20) When estimating costs as financed, however, EPA assumes higher debt service associated with pre-2000 investments (more than $6 billion annually from 2000 to 2019 compared with CBO's $4.3 billion) and a shorter repayment period (20 years compared with CBO's 25 years).(21) On the low end, differences in financing assumptions are not as important in explaining EPA's higher estimate of investment costs as financed, for two reasons. First, both agencies assume repayment periods of 30 years. Second, the effect of EPA's higher debt-service payments associated with pre-2000 investments is countered by its lower pay-as-you-go costs associated with investments from 2000 to 2019. Instead, the primary reason that EPA's estimate is higher than CBO's is that it does not allow for savings from increased efficiency in investment. CBO's estimate reflects efficiency savings of 15 percent.(22) Ultimately, EPA's estimates of the increase in investment costs for
wastewater infrastructure from 2000 to 2019 are neither better nor worse
than CBO's. EPA's estimates of baseline investment and the associated future
cost of debt service on pre-2000 investments cannot be deemed more or less
credible than CBO's on the basis of information currently available. If
CBO adopted EPA's assumptions for those factors, its high-end estimate
of the increase would fall to a level similar to EPA's and its low-end
estimate of the increase would fall below the current estimates because
of the effect of efficiency savings.(23)
Operations and Maintenance Costs from 2000 to 2019EPA's and CBO's high-end projections of total O&M costs differ dramatically
(see Figure 3).(24) EPA's $36.2 billion estimate ($82 billion minus $45.8 billion) of the increase in costs doubles CBO's $18.1 billion estimate ($57 billion minus $38.9 billion), resulting in a $362 billion difference over the 20-year period. Almost all of that difference--88 percent--is found on the drinking water
side.
CBO's and EPA's baseline drinking water estimates vary because the two agencies make different adjustments to compensate for the fact that the underlying data do not reflect spending by private drinking water systems. EPA scales the underlying data up by 33.3 percent, reflecting information from the November 1998 "Regulatory Impacts Analysis of the Stage 1 Disinfection By-Products Rulemaking," a peer-reviewed document. CBO uses a figure of 15 percent, in keeping with conventional wisdom and data from the responses recorded in EPA's Community Water System Survey.(25) The dramatic gap between the agencies' high-end projections of drinking
water O&M is attributable to EPA's assumption in this scenario that
the investment profile from 2000 to 2019 is front-loaded (rather than distributed
more evenly over the 20-year period). Using the distinction between current
and future needs identified in the 1999 Drinking Water Infrastructure
Needs Survey, EPA chose to assign 25 percent of current pipe and nonpipe
needs to each year in the 2000-2003 period, 20 percent of the costs of
complying with recently promulgated or proposed regulations to each year
in the 2000-2004 period, and 6.25 percent of future pipe and nonpipe needs
to each year in the 2004-2019 period. Thus, most of the new investment
in drinking water occurs from 2000 to 2004 under that scenario. Together
with EPA's use of the historical relationship between O&M spending
and the capital stock as the basis for projecting future costs, the agency
forecasts high O&M costs throughout the period as a by-product of the
early increase in the capital stock. Although it is plausible that the
amount of O&M is related to the size of the capital stock, EPA's use
of the historical relationship between the two, in combination with a large
amount of early investment, sharply increases its O&M estimate. CBO's
approach (a simple linear extrapolation of estimated real O&M spending
from 1980 to 1998, which explains 99 percent of the variation from the
means) might understate the uncertainty surrounding O&M costs by failing
to capture some ways in which the future could differ from the past (tighter
effluent standards could raise O&M costs more quickly, for example).
Nevertheless, EPA's high-end estimate is probably too high.
|