2.1 The affordability doctrine meets arsenic

The affordability doctrine was tested in 2001 when the Agency promulgated a revised NPDWR for arsenic (USEPA, 2001c). The Agency determined that aggregate estimated annualized benefits of $170 million justified aggregate estimated annualized costs of $210 million, with unquantified benefits making up the difference. USEPA’s benefit estimates were contested; for example, Burnett and Hahn (Reference Burnett and Hahn2001a, Reference Burnett and Hahnb) re-estimated them to be only $20 million.

The Agency’s cost estimate also was controversial (USEPA, 2001b, pp. 7038–7041). For a 10 μg/L MCL, Frey et al. (Reference Frey, Owen, Chowdhury, Raucher and Edwards1998, table 5) estimated annualized national compliance costs at $708 million, later revised downward to $495 million (Frey et al., Reference Frey, Chwirka, Narasimhan, Kommineni and Chowdhury2000, table 4.8). Gurian et al. (Reference Gurian, Small, Lockwood and Schervish2001) estimated annualized national compliance cost at $294 million, with 90th per cent confidence intervals ($177–$495 million) spanning the USEPA and Frey et al. (Reference Frey, Chwirka, Narasimhan, Kommineni and Chowdhury2000) estimates. Estimates of the national annualized incremental compliance cost for 10 μg/L instead of 20 μg/L include USEPA’s figure of $129 million (USEPA, 2001b, table III.E-1), $174 million by Gurian et al. (Reference Gurian, Small, Lockwood and Schervish2001, table 4), $360 million by Frey et al. (Reference Frey, Chwirka, Narasimhan, Kommineni and Chowdhury2000, table 4.8), and $550 million by Frost et al. (Reference Frost, Tollestrup, Craun, Raucher, Stomp and Chwirka2002, table 1). This disagreement did not subside after promulgation of the final rule (Tiemann, Reference Tiemann2006; Hilkert Colby et al., Reference Hilkert Colby, Young, Green and Darby2010; Gingerich et al., Reference Gingerich, Sengupta and Barnett2017).

Congress reacted swiftly to stakeholder concerns about economic feasibility, noting that the revised standard would impose “significant financial costs on many small communities, and many of these communities may find it impossible, because of the financial burden, to be in compliance by 2006 as the rule requires.” Members on the appropriations conference committee were “concerned that, because of their complexity, the current waiver and exemption provisions … may not provide sufficient flexibility for the small communities to receive additional time to reach compliance.” They also were “very concerned that … very small communities may abandon their municipal systems in favor of untreated and unregulated private wells which could create significant other health risks for these communities.” They directed USEPA to, by March 1, 2002, “review the Agency’s affordability criteria and how small system variance and exemption programs should be implemented for arsenic,” taking account of the “undue economic hardship” faced by communities served by small public water systems.” But the conferees also hedged their bets, averring that they “do not intend to create loopholes in the Safe Drinking Water Act for compliance to a national arsenic standard” (U.S. House of Representatives, 2001, pp. 174–175).

2.2 USEPA’s proposed revision to the affordability doctrine

In its report to Congress (USEPA, 2002a), USEPA defended the affordability doctrine but did not include the review Congress had directed be conducted. Instead, the Agency briefly summarized public comments received on the 2000 proposed arsenic rule and promised an “ongoing affordability review” that would include soliciting advice from the Science Advisory Board and obtaining “input from stakeholders” (p. 11). Two USEPA advisory committees weighed in with conflicting recommendations (U.S. EPA Science Advisory Board, 2002; U.S. EPA National Drinking Water Advisory Council 2003 [including a minority report]).

In 2006, USEPA proposed a revision to the affordability doctrine (USEPA, 2006d). The proposed changes would have had mixed or ambiguous effects, most notably the relaxation of the 2.5 % MHI regulatory budget. Doctrinally, the proposal reiterated the Agency’s longstanding view that economic feasibility is limited to “what may reasonably be afforded by large metropolitan or regional public water systems” (p. 10682), thus leaving unaddressed the economic inefficiency and inequity discussed in Section 3. The proposed revision was widely criticized by public commenters (USEPA, 2006c) and subsequently by others (Rubin et al., Reference Rubin, Raucher and Harrod2007; Crawford-Brown et al., Reference Crawford-Brown , Raucher, Rubin and Lawson2009; Raucher et al., Reference Raucher, Rubin, Crawford-Brown and Lawson2011; U.S. Conference of Mayors et al., 2013a; U.S. Conference of Mayors, 2014). No revised guidance has been finalized.

2.3 Affordability v. economic feasibility in practice

USEPA has promulgated a number of drinking water rules since SDWA 1996, but in no case has the Agency clearly defined economic feasibility.Footnote ² Some inferences can be made from the Stage 2 Disinfectants and Disinfection Byproducts Rule (Stage 2 DBPR), in which the Agency stated that “it may not be economically feasible for some small systems to install and operate an on-site [granular activated carbon] reactivation facility” (USEPA, 2006a, p. 413). USEPA estimated the 50th, 90th, and 95th percentile costs for small systems at $18, $169, and $198 per household served (USEPA, 2006a, p. 459, table VI.E-1). Thus, it might be inferred that USEPA believes the latter two average household-level costs are not economically feasible. However, because the Agency did not report benefit estimates by system size, apparent statements about economic feasibility actually may be about affordability.

Inferences also cannot be readily derived from pre-SDWA 1996 rulemakings. For example, in 1994 USEPA re-proposed a NPDWR for sulfates and said that four options “would be economically feasible” despite the financial difficulties presented to small systems (USEPA, 1994, p. 65597). With respect to an option costing on average $287 to $811 per household, depending on system size, the Agency acknowledged “concerns that this option would not be economically feasible for small systems” (table 9). As before, because benefits are absent from the discussion, these statements may be about affordability, not economic feasibility. Further, any inference about USEPA’s intentions must be tempered by the knowledge that the reproposed standard was not promulgated.

Nothing can be inferred from USEPA statements about feasibility without a modifier. The Agency interprets any unmodified use “in terms of technological limits of concentration removal (i.e., treatment) and analytical methods (i.e., sampling and measurement)” (Raucher & Cromwell, Reference Raucher and Cromwell2004, p. 2) – i.e., without economic content. In short, it appears that a general principle of economic feasibility is ground that USEPA has never plowed in the context of drinking water because it was supplanted by the affordability doctrine.

But the affordability doctrine appears to have reached its practical limit, as the cost of drinking water now approaches (and may even exceed) the 2.5 % MHI regulatory budget. Raucher and Cromwell (Reference Raucher and Cromwell2004) relied on USEPA RIAs to build their best estimates of aggregate compliance costs of SDWA 1996 regulations ($1.8 billion per year) and seven regulations promulgated under SDWA 1986 ($4.8 billion per year) (both at 2004 $; 7 %). Household-level equivalents are generally not available because they are absent from the RIAs. Nonetheless, household-level estimates may breach the affordability threshold. In the case of the Arsenic NPDWR, for example, Raucher and Cromwell conclude that “average households in the smallest system size category will pay at least 14 times as much (on average) as households served by utilities in the largest size categories,” with annual compliance cost per household exceeding $700 (pp. 20–21).

Other analyses have obtained similar results. In a recent California study, the median (SD) total cost of drinking water for surveyed communities was estimated at $1,172 ($488) per household, with costs in many jurisdictions exceeding the 2.5 % MHI regulatory budget (U.S. Conference of Mayors, 2014, pp. 4, 12, 13, table C). At some point, it may not be possible for USEPA to promulgate new NDPWSs without unambiguously violating its affordability doctrine.

3.1 Inefficiency

NPDWRs are economically inefficient at least four ways. First, as noted earlier, USEPA sets standards based on what it determines to be economically feasible for very large public water systems (USEPA, 2006d, p. 10682). Because economies of scale in drinking water supply and treatment are so strong, standards that are economically feasible for very large systems are virtually certain to be economically infeasible for small systems.

Second, neither variability across systems within a system-size category nor household variability within each system is considered (USEPA, 2006d, p. 10673). Thus, even if a NPDWR were economically feasible for the median system in a system-size category, it would be economically infeasible for as many as half of the systems in that category, and as many as half of the households in the median system.

Third, USEPA estimates compliance expenditures, not opportunity costs. At the household level, this means not counting the benefits foregone from reduced expenditures on food, shelter, education, health care, transportation, and expenditures on goods and services that improve health (Keeney, Reference Keeney1990, Reference Keeney1994; Lutter & Morrall III, Reference Lutter and Morrall1994; Lutter et al., Reference Lutter, Morrall and Viscusi1999; Cory & Taylor, Reference Cory and Taylor2017). At the system level, expenditures to comply with economically infeasible NPDWRs reduce resources available for infrastructure investments, the need for which has been estimated to exceed $1 trillion over 25 years (AWWA, 2012). This includes extensive replacement of drinking water service lines to reduce or eliminate lead exposure, ironically to comply with a different and notoriously expensive NPDWR. When net social benefits are estimated, they are upwardly biased because expenditures, not opportunity costs, are subtracted from estimated benefits.

Fourth, the affordability doctrine does not take account of benefits, and even if benefits were counted, USEPA’s estimates tend to be upwardly biased by the Agency’s longstanding practice of “policy framing” risk estimates at the “upper end” (U.S. EPA Office of the Science Advisor, 2004). Thus, even if opportunity costs were properly estimated and subtracted from benefits, estimates of net social benefits would be upwardly biased.

Evidence of economic inefficiency extends back to early SDWA 1974 regulations, when several economic inefficiencies were identified (Council on Wage and Price Stability, 1978) in USEPA’s 1978 proposed trihalomethane NPDWR (USEPA, 1977). The standard was estimated to cost on average $3.9–6.3 million ($1978) per excess cancer case prevented, depending on stringency, but exceed $12 million per case for water systems serving 75,000 or fewer persons. These inferences were confirmed and extended to other NPDWRs in subsequent analyses (Morrall III, Reference Morrall1986; Office of Management and Budget, 1991; Raucher et al., Reference Raucher, Dixon, Trabka and Drago1994).

3.2 Inequity

Households served by small systems are assured of being treated inequitably multiple ways. First, the affordability doctrine imposes disproportionate burdens on lower-income households (U.S. EPA Science Advisory Board, 2002, Berahzer, Reference Berahzer2012; Teodoro, Reference Teodoro2018). Households with income below the MHI must pay more than 2.5 % of their income for drinking water.

Second, USEPA ignores regional and State differences in MHI. Median households in regions and States with lower MHIs therefore must pay more than 2.5 % of their income for drinking water. National MHI in 2016 was $58,820 (90 % CIs: ± $102), but ranged from $40,528 (90 % CIs: ± $258) in Mississippi to $72,935 (90 % CIs: ± $1,164) in the District of Columbia – a range of 1.8x. About 60 % of Mississippi households earned $50,000 or less compared to 38 % of District households (U.S. Census Bureau, 2016).

Third, communities with below-average incomes must bear costs that the Agency regards as unaffordable by households served by large metropolitan systems. MHI in all 82 Mississippi counties was below the national MHI in 2016. In 20 % of Mississippi counties, MHI was less than half of the national MHI (U.S. Census Bureau, 2016). The affordability doctrine requires that they commit 5 % of income, or more, to drinking water. This is equivalent to a peculiar regressive tax that, instead of producing government revenue, transfers wealth from the poor to firms in the water treatment industry.

3.3 Arbitrariness

The affordability doctrine relies on other arbitrary elements besides the 2.5 % national MHI threshold, and thus results in outcomes that are inherently arbitrary on other margins as well (U.S. EPA Science Advisory Board, 2002; Irvin, Reference Irvin2017). Each of the 151,000 water systems in the USA (USEPA, 2017) could be defined as its own domain or aggregated into a single domain for standard-setting. In practice, water systems are aggregated into arbitrarily defined system-size categories in the expectation that variability is captured by economies of scale. But as Gingerich et al. (Reference Gingerich, Sengupta and Barnett2017) showed for small systems participating in an Agency-sponsored demonstration project exploring arsenic treatment alternatives, direct costs can be highly variable for reasons other than system size. The median annual treatment cost among systems in this demonstration project was $147 per household, but ranged from $17 to $662 per household. The median increase in total water bill was 55 %, but ranged from 6 to 274 %.Footnote ³ These ranges suggest that the arsenic MCL may not be technologically feasible for small systems “under field conditions,” as required by SDWA 1996 § 1412(b)(4)(D).

Similarly, there is nothing special about MHI as an affordability metric, nor does the particular MHI statistic obtained from the Census Bureau lack critics (see, e.g., Eskaf, Reference Eskaf2013; Irvin, Reference Irvin2017). Several alternative approaches have been suggested to make the affordability doctrine sensitive to the disproportionately high opportunity costs experienced by low-income households (Raucher et al., Reference Raucher, Rubin, Crawford-Brown and Lawson2011; U.S. Conference of Mayors et al., 2013a, b; UNC Environmental Finance Center, 2017; Teodoro, Reference Teodoro2018, Reference Teodoro2019; Czerwinski et al., Reference Czerwinski, Fretwell, Fosler, Lindsey and Pagano2018; Raucher et al., Reference Raucher, Clements, Rothstein, Mastracchio and Green2019). USEPA did not propose to adopt any of the earlier recommendations but may be considering the more recent ones.

3.4 USEPA’s policy of ensuring equal ex post health risk from drinking water imposes highly unequal ex post risk elsewhere

Equal protection can be defined as ensuring that people receive the same quantity at different prices or pay the same price and receive different quantities. Quantity-based equal protection is reasonable and appropriate for the provision of constitutional rights (e.g., free speech, protection from unreasonable search and seizure, trial by jury) and public goods supplied by government and funded by general taxation (e.g., national security). However, drinking water is a private good even where it is provided by a public entity. Consumption is rivalrous, and sellers can deny access to those who refuse to pay.

Drinking water is more like other private goods and services, such as electricity, natural gas and sewage treatment that often are supplied by regulated public or private monopolies because of high fixed costs. Where risk reduction is an attribute of a private good or service, consumers typically pay the same price but obtain different quantities of risk reduction depending on other factors, such as their propensity for risk-taking. The result is variability in ex post risk outcomes, not ex post price differences. Indeed, in some jurisdictions, public utilities are statutorily required to charge all customers the same rate (Berahzer, Reference Berahzer2012; U.S. EPA Environmental Financial Advisory Board, 2016; UNC Environmental Finance Center, 2017), a practice that is consistent with price-based equal protection.

A few examples from outside the world of public utilities provide additional insight. For decades, automobile manufacturers marketed as options various safety features, such as seat belts, air bags, and antilock brakes, charging the same price to all similarly situated customers. But customers varied greatly in the amount of risk reduction they obtained, largely due to fixed differences in baseline risk and their propensity to engage in risky behavior. The same pattern applies to newly invented automotive safety features, such as lane-departure, forward-collision, and blind-spot warning technologies. The unit price of these technologies is fixed, at least in broad categories, but the quantity of risk reduction obtained varies across consumers. Persons who are relatively risk-averse, or have reason to believe that they are riskier than average, are more likely to purchase these safety innovations. Some consumers also are more likely to adapt their behavior in ways that reduce risk-reduction benefits (Peltzman, Reference Peltzman1975). Regardless of their baseline risk, preference for risk-taking, or propensity to engage in adaptive responses that reduce benefits, they still pay the same price.

There is a vibrant consumer market for inherently risky power tools. Operating risk varies greatly because consumers differ in baseline conditions (e.g., experience, technical skill, intensity of use) and their propensity to engage in risky behavior (e.g., read and follow directions, wear protective clothing and equipment). Still, all consumers pay the same price for each risk-reduction technology built into these products. What varies across consumers is the quantity of risk reduction obtained.

This pattern applies even to speculative benefits. The establishment of the National Organic Program by the Agricultural Marketing Service of the U.S. Department of Agriculture gave food producers a government-sanctioned way to appeal to consumers who believe that foods certified as organic are safer than conventional foods. Consumers who purchase certified organic foods pay the same, higher unit prices regardless of how much psychological benefit they receive.

It is difficult to find examples of private goods with risk-reducing attributes that deliver the same ex post risk but, like drinking water under EPA’s quantity-based equal protection policy, cost consumers widely different amounts. The typical consumer experience is one of variable ex post risk resulting from differences in baseline risk, risk preferences, and the net quantity of risk reduced – not differences in the price of the risk-reducing attribute or feature.

Since 1974, USEPA has implemented SDWA to ensure the same level of ex post protection from drinking water contaminants (Schnare, Reference Schnare1998), not the same price for risk reduction. The practice appears to have been first codified as policy during the National Performance Review (NPR), a management effort to streamline federal government operations to improve performance and reduce costs (Gore Jr. Reference Gore1993, Reference Gore1994). Although official NPR reports are silent on equal protection, USEPA appears to have leveraged the NPR to make quantity-based equal protection as Agency policy (USEPA, 1996).Footnote ⁴

By insisting on quantity-based equal protection for potential health risks, USEPA imposes highly unequal protection from actual financial risks that inevitably result from highly unequal prices (Raucher, Reference Raucher and Pontius2003).Footnote ⁵ Households served by small systems cannot obtain the same quantity of ex post health risk except by bearing disproportionately large financial risks. Similarly troubling is the highly unequal protection from financial risk imposed on nonmetropolitan areas, low-income communities, and low-income households generally, that inevitably results from compulsory quantity-based equal protection from potential health risks. Environmental justice advocates have challenged differential health protection in drinking water as a violation of Title VI of the Civil Rights Act, but these efforts have not succeeded for lack of discriminatory intent (Lado, Reference Lado2017). But the case for intentional discrimination is stronger with respect to regulatory policies that, like USEPA’s practice in SDWA, knowingly impose disproportionate financial risks on the poor.

4.1 Pre-SDWA 1996: Deadweight Losses and Inequity through Selective Economic Feasibility

For all but the largest metropolitan water systems, NDPWRs have been economically infeasible by design (USEPA, 2006d, p. 10673). Variances and exemptions were theoretically available for small water systems (Schnare, Reference Schnare1998), but relief if granted was temporary, and systems were generally required to comply with standards that were less stringent than NPDWRs but still economically infeasible. That is, systems for which NPDWRs were economically infeasible had no option except to comply by making their customers poorer.

This is illustrated in Figure 1. Assume six alternative MCLs ranging from MCL_a (the most stringent) to MCL_f (the least stringent), and three different sizes of public water systems (large [L], small [S] and very small [VS]), which for simplicity in exposition are all presumed to have the same baseline level of a contaminant. Economies of scale in water treatment result in three different marginal cost schedules (MC_L, MC_S, and MC_VS) and the common marginal benefit schedule MB. All six alternative MCLs are technologically feasible for all three system sizes, so USEPA would have chosen MCL_b as the NPDWR. MCL_b is economically feasible for large systems because the marginal benefit of treatment equals the marginal cost (MB = MC_L). Households served by large systems would pay a price of P_L.

Figure 1. Traditional standard-setting under the Safe Drinking Water Act.

Though it is technologically feasible for small systems to achieve MCL_b, they can only do so at the higher household price P_S. Households served by small systems therefore are unambiguously worse off. At the margin, they pay P_S to obtain benefits worth only P _L. The difference in expenditure, P_S – P_L, multiplied by the quantity of drinking water consumed, is transferred from their family budgets to firms that design, manufacture, install, or operate water treatment technology. They suffer a deadweight loss equal to area A.

The situation is worse for households served by very small systems. They must pay P_VS for benefits worth only P_L. They must transfer even more income to firms in the water treatment industry. They suffer the deadweight loss of areas A plus B.

These inefficiencies have been defended on equity grounds by defining equity in terms of a quantity-based definition of equal protection. While it is true that all households gain the same ex post quantity of health risk, households served by small and very small public water systems obtain this only by being made permanently poorer. USEPA achieves quantity-based equal protection from health risk by reducing the welfare of those it purports to benefit.

4.2 Achieving efficiency and reducing inequity through economic feasibility

Interpreting SDWA 1996 as a problem of risk minimization subject to technological and economic constraints allows economic inefficiency and price-based unequal protection to be substantially reduced. This can be accomplished by setting NPDWRs that are both technologically and economically feasible for the smallest size systems subject to regulation. That is, USEPA would interpret SDWA 1996 §§ 1412(b)(4)(B)-(C) as applying consistently to all water systems subject to federal standards, not just the handful of metropolitan systems large enough to spread compliance costs over hundreds of thousands or even millions of households. The structure of the amended statute arguably is more consistent with this approach, which makes national standards a flexible floor instead of a de facto ceiling. States can choose to set more stringent standards, but they would have to transparently acknowledge, and take responsibility for, the resulting deadweight losses, reverse Robin Hood income transfers to the drinking water treatment industry, and the inequities of quantity-based equal protection.

Figure 2 shows how this would work. very small systems would be exempt. The MCL would be set at MCL_d, the most stringent standard that is economically feasible for small systems. Households served by large and small systems would pay prices P_L* and P_S*, respectively. The deadweight loss in Figure 1 would be avoided.

Figure 2. Economically feasible standard-setting under the Safe Drinking Water Act.

Initially, large and very small systems would suffer the deadweight losses in areas C and D, respectively, because in both cases there would be uncaptured net benefits from more stringent controls. But these deadweight losses likely would be temporary. Households served by large systems would be willing to purchase treatment technology sufficient to achieve MCL_b, the most stringent standard that is economically feasible for them. These households would pay $ {P}_L^{"} $ instead of the lower price $ {P}_L^{\ast } $ , but the marginal value of additional health benefits would exceed the price difference. Similarly, even though very small systems would be exempt, their customers would be willing to purchase treatment technology sufficient to achieve MCL_f, the most stringent standard that is economically feasible for them. Households served by very small systems would be willing to pay $ {P}_{VS}^{"} $ to gain the value of additional health benefits, which exceed marginal cost for every unit of contaminant greater than MCL_f. Footnote ⁶

These choices are rational because both groups of households would be better off than under MCL_d. No regulatory coercion is needed to motivate large or very small systems to voluntarily achieve standards more stringent than MCL_d. Firms in the drinking water treatment industry would gain new incentives to invest in research and development that reduces the cost of serving the needs of very small systems instead of lobbying for federal standards that are profitable for them but financially damaging for very small system customers.

This solves the allocative efficiency problem because NPDWRs would not impose an economically infeasible drinking water treatment technology on anyone. Price-based unequal protection would remain, however, because households would pay prices inversely related to the size of the system serving them. But these price differences would reflect differences in cost of service, which generally are not perceived as a discriminatory practice in drinking water supply (UNC Environmental Finance Center, 2017, pp. 17–18).

The only other alternative that solves the allocative efficiency problem and achieves similarly equitable results requires EPA to set multiple MCLs within a single NPDWR.Footnote ⁷ This might be compatible with SDWA 1996, but it would have other significant drawbacks. First, it would require the Agency to obtain much more extensive knowledge across tens of thousands of water systems in the USA. This would be expensive and undoubtedly require a significant expansion in the ranks of its technical staff. Second, the task would be duplicative. The only logical sources for this information are local water systems. It is suboptimal to recycle local information through USEPA and return it as a regulation imposed on the same systems from which the data were obtained. A suboptimal federal information recycling loop might make sense if systems lacked sufficient expertise or faced perverse incentives to analyze their situations incorrectly. But systems have access to a wealth of technical expertise elsewhere (e.g., the American Water Works Association and the National Rural Water Association, among others), and the incentives they face are compatible with allocative efficiency, minimizing price-based unequal treatment, and responding to customer concerns.

5.1 Surface water treatment regulations

A suite of surface water treatment rules targeting Cryptosporidium (USEPA, 1998d, 2001a, 2002b, 2006b) depended on benefit estimates obtained by risk modeling. These rules were determined to be economically feasible, but modeled illnesses and deaths averted substantially exceeded the number of illnesses and deaths reported from all pathways by the Centers for Disease Control and Prevention (CDC). Because the model was not validated and the substantial gap between modeled and reported illnesses and (especially) deaths was not adequately explained, it is reasonable to be skeptical about economic feasibility claims. It is possible that some or all of these rules were economically feasible for very large systems, where economies of scale yield low marginal costs. However, the gap between modeled and reported illnesses and (especially) deaths is so great that even this inference is doubtful. Moreover, it is highly unlikely that any of these rules were economically feasible for smaller systems.

5.2 Disinfection byproduct regulations

Two regulations were promulgated to manage risks from disinfection byproducts (DBP) (USEPA, 1998c, 2005). Any determination of economic feasibility depends on the assumption that mixed, weak, and unconfirmed epidemiological associations between DBP and cancer are causal. USEPA acknowledged that estimated cancer risks could be zero but assumed causality anyway. Without this assumption, neither regulation would have been economically feasible even for very large systems.

5.3 Specific contaminant NPDWRs

For radionuclides, USEPA retained then-existing MCLs for combined Ra²²⁶ and Ra²²⁸ (though with intensified monitoring requirements), beta particle and photon emitters, and gross alpha activity. It declined to set a more stringent MCL.

For uranium, USEPA’s estimates of aggregate net benefits were negative at every analyzed alternative. Thus, even the least stringent alternative analyzed (80 μg/L) was too stringent to be an economically feasible uniform NPDWR. According to USEPA’s economic analysis, the 30 μg/L uranium MCL was estimated to yield about $50 million in negative net benefits, with a marginal cost of $68 million per cancer case averted. Promulgating 80 μg/L instead of 30 μg/L as the MCL would have saved $36.4 million while reducing the number of cancer cases averted by 0.35, a marginal cost of $190 million per cancer case. It is possible that 80 μg/L (or 30 μg/L, the MCL promulgated) would have been economically feasible for very large systems. However, uranium is not a significant constituent in source waters used by very large systems, so it is probable that the rule was economically feasible for no water system.

Similarly, the arsenic MCL (10 μg/L) was estimated to yield negative aggregate net benefits at every MCL examined. The MCL was economically feasible for systems serving ≥ 1 million persons under fairly broad conditions, however. Under several restrictive conditions (3 % discount rate, upper-bound arsenic exposure, average household size ≥ 5), the MCL may have been economically feasible for the 10,001–50,000 person system size category. These restrictions are so significant – especially the minimum average household size – that no systems in this size category probably qualifies.

The 1991 lead and copper rule (LCR) was estimated by USEPA to cost systems $460–$750 million per year and States $40 million per year (USEPA, 1991, table 20). USEPA did not report benefit and cost estimates for the 2000 LCR revision (USEPA, 2000a), so no inferences can be gleaned concerning the specific effect of the economic feasibility principle had it been adopted.Footnote ⁹ Abernethy et al. (Reference Abernethy, Chojnacki, Farahi, Schwartz and Webb2018) addresses the high cost of simply determining where lead service lines exist and developing a cost-effective strategy for line replacement. The USEPA Science Advisory Board (2011) has raised concerns that service line replacement may increase rather than reduce risk. This has obvious relevance under the economic feasibility principle, but it is immaterial if technological feasibility alone controls decision-making. USEPA’s forthcoming Long-Term LCR (USEPA, 2018a), which purports to overcome limitations that made the 2000 regulation ineffective, has been estimated to cost another $44–274 million per year (Slabaugh et al., Reference Slabaugh, Arnold, Chaparro and Hill2015, table 6), with annual costs rising exponentially as system size declines (Slabaugh et al., Reference Slabaugh, Arnold, Chaparro and Hill2015, fig. 5).

5.4 Summary

If USEPA had adopted the economic feasibility principle after SDWA 1996, the entire portfolio of SDWA 1996 regulations would have looked very different. USEPA would have set MCLs such that they were economically feasible for the smallest system size subject to regulation. The Agency would have made tradeoffs between stringency and regulatory scope. These tradeoffs would have been more striking if the Agency had also complied with the relevant SDWA 1996 provisions governing objective risk assessment and benefit cost analysis.

Allocative inefficiency would have been reduced or avoided to an extent dependent on the objectivity of USEPA's risk and economic analyses. Cost-savings would have totaled hundreds of millions of dollars annually. These savings would have enabled systems to better fund key infrastructure improvements, and households would have had greater disposable income for more highly valued uses.

Post-1996 drinking water regulations, like those that preceded SDWA 1996, imposed gross inequities on nonmetropolitan areas, low-income communities, and low-income households. The economic feasibility principle would have substantially attenuated these inequities. The poor would not have borne disproportionately high and rising costs for drinking water, and drinking water regulation would not have made them poorer with each successive regulation action.