Last Monday we began a multi-part series on Susan Dudley, the Bush administration’s nominee to head OMB’s Office of Information and Regulatory Affairs. Her confirmation hearing is scheduled for November 13. A campaign is being led by Public Citizen and OMB Watch to scuttle her confirmation. Our series analyzes the merits of theiropposition report. A quick review:
- Monday, October 30: Background on OIRA
- Tuesday, October 31: Public Citizen and OMB Watch, the Authors of the Opposition Report
- Wednesday, November 1: An Introduction to Opposition Report
- Thursday, November 2 The Case of Airbags
- Friday, November 3: Dudley’s Impossible Requirements
When we began we promised three posts this week.
- Monday, November 6. Dudley’s Views on Consumer Sovereignty, Nonuse Value, and Lifesaving
- Tuesday, November 7: Dudley’s Radical Ideas
- Wednesday, November 8: Summarizing the Opposition Report
Today’s post is reduced in scope because last Friday’s post included an extended discussion of market failure. It turned out that this was necessary because Dudley’s views on this technical subject were intimately tied into the complaints about her claims made in the Opposition Report.
The Opposition Report says that Dudley holds unconventional views on three important concepts in economics besides market failure. We address each of these today.
The Opposition Report says Dudley is “inconsistent and contradictory” when she professes allegiance to the principle that consumers are fully capable to make their own decisions, and thus do not need governmental help through regulation, except when significant market failure exists (pp. 25-28). The examples cited in support of this point are her public interest comments on National Highway Traffic Safety Administration’s advanced airbag rule, EPA’s ozone standard, and a proposed expansion of EPA’s Toxic Release Inventory, and publicly expressed views on NHTSA’scorporate average fuel economy standard for light trucks as reported in a Washington Post article (cited incorrectly in the Report). We’ve previously discussed both airbagsand ozone at length. Both CAFE and TRI are new, and we address them separately below
Corporate Average Fuel Economy Standards (CAFE) for light trucks
The criticism against Dudley is that she judged this rule the “worst of 2003” even though it had positive net social benefits. Thus, the Opposition Report alleges, Dudley opposes regulation even when it passes a normative benefit-cost test:
Dudley dismissed NHTSA’s finding of net benefits from the increase, writing, “We know this must be false, because any regulatory constraint that forces consumers away from their preferred choices must have negative net benefits (i.e.,
make Americans worse off)” (p. 26).
The Opposition Report quotes Dudley correctly but deletes a crucial sentence, which weboldface for emphasis:
“Worst rule of 2003: The National Highway Traffic Safety Administration corporate average fuel economy (CAFE) standards for light trucks. NHTSA continues to force vehicle manufacturers to achieve higher miles per gallon than the market would offer, or consumers would choose, in the absence of the regulation. Absurdly, its economic model shows large net benefits to consumers even if markets are assumed to operate perfectly, i.e., without counting any externalities. We know this must be false, because any regulatory constraint that forces consumers away from their preferred choices must have negative net benefits (i.e., make Americans worse off).”
If it’s true that NHTSA claimed net social benefits in the absence of market failure, then Dudley’s opinion is easily defended. Net social benefits are technically infeasible without market failure. If markets work perfectly, then consumers begin from a position of maximum satisfaction. It’s mathematically impossible for government (or anyone else) to improve consumers’ level of satisfaction beyond the maximum. By failing to include this crucial sentence in their quote, the authors of the Opposition Report unambiguously misled readers concerning Dudley’s views.
Toxic Release Inventory
The Opposition Report criticizes Dudley for opposing certain expansions of EPA’s TRI program, “claiming that the general public was too ill-informed to make good use of such information” (p. 26). Dudley’s 1999 public interest comment does not say that. Rather, she said the TRI database was riddled with errors and did not communicate to citizens meaningful measures of human health risk. The problem of error has been addressed by OMB’s information quality guidelines, which postdate Dudley’s comment by three years. The problem of potential mismatches between what risk-based information citizens might desire to have, and TRI’s mass-based reporting regime, remain. An important threshold question is whether these mismatches are defects of the TRI program or completely intentional departures from the social net benefit maximization framework. We’re inclined to think they are intentional.
James T. Hamilton has studied TRI extensively, and documented the nature of the informational market failure posed by chemical releases and the inability of citizens to overcome transactions costs to obtain the information they wanted directly. He says the law enacting TRI
shifted the property rights to [information about] pollution from private information held by a firm (if it collected the data at all) to public information assembled and distributed by the government (p. 246).
Hamilton points out that TRI data also have an instrumental value in that they “are a means to an end” — in this case, they enable interest groups to pressure firms to reduce their releases of covered chemicals. For this instrumental purpose, the actual risk posed by a chemical release is at best a secondary matter, and citizens’ confusion about the difference between mass and risk may have been an asset. Dudley’s public interest comment seems to naively ignore these instrumental purposes and assume that TRI’s statutory purpose was to maximize social welfare. As Hamilton notes, Congress did not craft the law that authorized TRI or debate it from an economic perspective. The law was frankly (if not transparently) redistributional. Therefore, it should not be surprising that EPA’s economic analysis suffers from internal confusion and contradiction, such as identifying as a social benefit the capacity TRI provides to interest groups to generate social costs.
Taking at face value the economic framework that guides Dudley’s public interest comment, there are a number of problems with TRI that go unmentioned. For example, because only releases of covered chemicals must be estimated and reported, TRI creates an incentive for firms to reformulate using chemicals that are not covered. The incentive for substitution is at least potentially perverse because uncovered chemicals could be riskier upon occupational or environmental release or if they are incorporated into products, which is not reportable. Another perverse incentive, though this time affecting EPA rather than firms, concerns reporting units. Risk perceptions can be magnified by requiring firms to report releases in very small units of mass. A release reportable as two tonnes does not appear nearly as large as a release reportable as 2,000 kilograms. A useful rule of thumb is that reporting units which are smaller than measurement or estimation error bounds don’t have much value for data quality purposes.
At the end of the day, Dudley recommended that EPA examine the value of TRI data; consider less frequent reporting, especially for small businesses; examine the accuracy of TRI data; and otherwise streamline the system to reduce burden. She also made several other recommendations specific to the new chemicals EPA proposed to add to the required list. None of these recommendations was revolutionary or extreme, but all seemed to begin and end with the presumption that TRI could be made consistent with the economic framework set forth in Executive Order 12866. Hamilton’s analysis of the TRI program, which was published six years after Dudley’s public comment, suggests that there may be no way for TRI to be compatible with this framework.
Use and Nonuse Values
The Opposition Report has a short discussion of a public interest comment co-authored by Dudley on a proposed EPA rule regulating fish kills from power plant cooling water intakes. The proposed rule was supplemented by a notice of data availability (NODA), and the public comment dealt with the issues presented in the NODA. Quoting from the introduction:
The proposed requirements, implemented through National Pollutant Discharge Elimination System (NPDES) permits, reflected the best technology available for minimizing adverse environmental impact from the cooling water intake structure based on water body type, and the amount of water withdrawn by a facility. More stringent requirements were proposed for more sensitive or biologically productive the waterbodies.
Though EPA provided estimates of the benefits of the proposal (based on a decrease in expected mortality or injury to aquatic organisms that would otherwise be subject to entrainment into cooling water systems or impingement against screens or other devices at the entrance of cooling water intake structures) on March 19, 2003 it supplemented these data with a notice of data availability (NODA).
The NODA suggests that in the North Atlantic Region alone the rule will provide annual benefits of $80,000 a year for commercial fishermen, another $880,000 a year for recreational fishermen, and between $14 million and $27 million a year in nonuse value benefits. In other words, the nonuse values attributed to the fish saved through the proposal are 175 to 335 times greater than the commercial value, and 17 to 32 times greater than the total use (commercial and recreational) value. The implications of these results are implausible (pp. 1-2, footnotes omitted)
The dispute raised by Dudley and co-author Daniel Simmons is not about the provisions of the proposed regulation. It’s about the economic analysis EPA prepared in support of the regulation, and a subsequent contingent value survey intended to estimate the “nonuse value” of fish. Nonuse value is the term economists use for the value people place on something that they have no intention of using — in this case, fish they do not intend to eat. Simmons and Dudley wrote that EPA significantly overstated the social benefits of the proposed rule — that it was implausible that a rule to protect fish from cooling water intakes would offer trivial benefits to commercial and recreational fishermen but millions in gains to people who don’t fish.
What methods did EPA employ to estimate nonuse value? In its 2002 proposed rule, EPA assumed that the nonuse value of fish killed in cooling water intakes was 50% as great as their use value to fishermen. Simmons and Dudley imply that EPA received a great deal of critical public comment on this assumption. In the NODA, EPA used a “benefits transfer” model to extrapolate existing nonuse value estimates from another context. Quoting Simmons and Dudley:
In this case, the existing study EPA used is a contingent valuation (CV) survey conducted of the value of eelgrass and wetlands in the Peconic Estuary on the East End of Long Island (p. 2).
Leaving aside the merits of the Long Island survey, Simmons and Dudley said estimates of the value of habitat are fundamentally different than estimates of the value of the fish living therein.
The EPA’s approach estimates the amount of wetland that could hypothetically produce the habitat services necessary for the fish hypothetically impinged or entrained, and then uses information from people’s hypothetical willingness to pay for the fish production services of that habitat. Each hypothetical estimate further detaches the final estimate from any mooring connected with actual values. Each estimate, assumption, and hypothetical weakens the explanatory power of the final valuation.
In the NODA, EPA “solicits comments on whether [this] benefits transfer approach provides a more comprehensive value that address all impingement and entrainment losses.” Due to the number of assumptions and hypotheticals involved in this approach, there is little reason to believe that the approach provides more or less of a comprehensive value of impingement and entrainment losses than the arbitrary 50 percent method. The real question is if the benefit transfer approach as applied here provides any information at all (p. 4).
Simmons and Dudley say EPA’s estimate of nonuse value is implausible: If the value to fishermen of fish that otherwise would be killed is only $282,339 per year, the value ofnot killing them for any purpose cannot be between $76 million and $140 million a year. Fish that are worth $1.12 per pound in the seafood market cannot be worth between $61 and $113 per pound left in the ocean.
This gigantic discrepancy between the estimated nonuse value of the fish and the commercial (or consumption) value begs the question, “why do we still have commercial fishing?” If Americans really value knowing that fish are swimming free so much more than they value eating fish, why do we pay commercial fishermen to catch them for our consumption? If the values EPA produced are truly people’s “willingness to pay” for the nonuse value they place on the fish, then why don’t people organize, raise money, and buy out the fishermen? Obviously there are some organizing costs to such an endeavor, but the possible societal benefits are enormous. In fact, the societal benefits are so enormous that EPA’s estimate of nonuse value could be overstated by an entire order of magnitude, and nonuse values would still dwarf use values. If nonuse values were anywhere near the estimate the EPA provides, we have to assume that environmental groups would organize to collect money and buy out commercial fishing.
To further put the EPA’s estimate in perspective, according to the National Marine Fisheries service, in 2001, commercial fishermen landed 9.5 billion pounds of fish. The value of these fish is $3.3 billion. Applying the same benefits transfer approach EPA used here to all fish taken by commercial fisherman, the nonuse value of the 9.5 billion pounds of fish landed may be worth between $580 billion and $1 trillion. Therefore, according to the EPA’s logic and estimates, commercial fishing costs the nation between $500 billion and $1 trillion a year – almost 5 to 10 percent of GDP! (p. 6).
In short, Simmons and Dudley say EPA’s estimate of nonuse value was beyond belief. Therefore, they say, EPA should not use it in its estimate of the benefits of its proposed cooling water intake rule.
The Opposition Report does not dispute the Simmons and Dudley analysis of EPA’s nonuse value estimate. Rather, the Report’s authors say it’s “sophistic” to reduce the value of fish to monetary terms (p. 21-22). That is, they object to the economics paradigm that EPA attempted to follow and which Simmons and Dudley said EPA did not properly apply.
After receiving public comments from Simmons and Dudley, and many others, EPA apparently decided to perform a contingent valuation survey to estimate nonuse values without the need for controversial benefits transfer methods. The Opposition Report criticizes a public interest comment submitted by Dudley and Simmons on this survey (pp. 21-22). The Dudley and Simmons comment itself deals with EPA’s responses to an earlier comment provided by Dudley and Simmons, apparently in response to a 60-day notice, to which Dudley and Simmons clearly believe that EPA’s response was inadequate. While we have not reviewed this initial public comment, the issues presented in this series of documents appear to be technical matters related to the practical utility of EPA’s proposed contingent valuation survey. What’s clear from the exchange is (1) EPA needs nonuse value benefits to justify the proposed rule because conventionally estimated benefits are well below regulatory costs; (2) EPA has tried three different approaches to generate the benefits it needs — assumption, benefits transfer, and now contingent valuation; and (3) Dudley and Simmons believe that each of EPA’s approaches has been technically defective. The technical issues are far too abstruse for the audience of the Opposition Report to follow, and perhaps too complex for the authors of the Report.
The Opposition Report criticizes Dudley for supporting what they call the “senior death discount” (pp. 39-40). This is a dispute over whether regulatory analysts should count as a social benefit the number of statistical lives saved by a regulation or the number of statistical life-years saved. Public Citizen and OMB Watch imply that they prefer to value lives over life-years, but they cite as their authorities Frank Ackerman and Lisa Heinzerling. In their 2004 book Priceless, Ackerman and Heinzerling objected to both.
Yet there is an inescapable mathematical fact that Ackerman and Heinzerling mention but the Opposition Report does not. Using life-years places greater weight on regulatory alternatives that prevent the premature death of children and young adults. Ackerman and Heinzerling say this “lifeboat principle” mimics what people would actually choose if they had to make a choice:
Many parents, faced with the “who gets in the lifeboat?” type of disaster scenario, would put their children ahead of themselves (p. 74).
The objections Ackerman and Heinzerling raise to the use of life-years involve cases where lifesaving trade-offs don’t exist and moral convictions are easier to locate with confidence. For their prime example they say “[t]he legal penalties for killing someone do not depend on the victim’s age” (pp. 74-75), and this presumably argues in favor of treating all premature deaths averted by regulation the same. This claim has both empirical and conceptual problems.
Empirically, there’s ample evidence suggesting that this claim is factually untrue. There are many examples in which criminal law treats murderers differently depending on the characteristics of their victims. The killing of a law enforcement officer is widely (if not universally) regarded as more serious. Many jurisdictions have established so-called “hate crimes” that make bigotry or bias as a motive for murder a more serious offense than, say, jealously, envy, or utilitarianism.
And there also is age-based discrimination in how murder is penalized. Americans consider child murderers especially detestable, so it seems quite plausible that they are more likely to face life imprisonment or capital punishment than are murderers in general. Historically, a majority of state laws (but not federal law) sanctioned the killing of an unborn child as murder. This is a clear example of an age-based discontinuity in legal penalties that Ackerman and Heinzerling deny exists. But their claim of absent age discrimination may be more true now than when they made it. The Unborn Victims of Violence Act of 2004 extended to unborn children a series of existing federal murder laws and thereby erased the previous age-based discontinuity at the federal level. It’s ironic that “Laci and Connor’s Law” would help Ackerman and Heinzerling make their case given its politics. In the Senate, the law was opposed by36 Democrats (80% of the caucus) and 2 Republicans (4% of the caucus).
Conceptually, cases without trade-offs seem highly inapt for use as metaphors for regulation. Even if it is proved that society punishes the murder of children more severely than all others, there’s no evidence that society would want murderers to choose their victims based on maximizing net benefits. Nor are there identifiable economists or other policy analysts clamoring for murder penalties to be meted out as a rising function of the estimated number of life-years the victim had left. The comparison made by Ackerman and Heinzerling is transparently invidious.
In contrast, regulatory decision-making always involves trade-offs among competing alternatives that offer different marginal benefits and marginal costs. Graduate schools of public policy teach these tools to enable future government officials to know how to maximize the value to society of irreversible commitments of societal treasure. It is commonplace to observe that changes in regulatory design alter the distribution of benefits and costs. When benefits consist of preventing premature death, changes in their distribution mean that more members of one subpopulation are made better off at the expense of more members of another. The authors of the Opposition Reportobject even to analyzing these distributional effects.
Still, the choice of lives or life-years as the unit for benefit measurement has important implications for regulatory policy. When statistical lives are used, lives are assumed to be persons of average age with about 35 life-years of life expectancy remaining. This doesn’t matter if lifesaving benefits are distributed uniformly by age; the choice of metric won’t make any difference. But if benefits are distributed asymmetrically, using life-years will significantly alter the benefit estimate.
The Opposition Report correctly points out the fact that using life-years significantly reduces benefits from regulations that prevent late-in-life health effects and premature mortality among the elderly and infirm. But the Report omits (and OMB Watch is silent about) the fact that using life-years significantly increases benefits from regulations that prevent early-in-life health effects and premature mortality among children.
The table below sets forth five different scenarios that span the range of ways lifesaving benefits could be distributed by age and shows how benefits change when statistical life-years are used instead of statistical lives. Using life-years reduces benefits when it’s the elderly who would avoid the health effect, but it increases benefits when it’s children who experience the health effects. For this reason we can predict which lifesaving metric various stakeholders will prefer. Advocates for the elderly will prefer lives, and advocates for children will prefer life-years. In the relevant discussion within the Opposition Report, the focus is solely on examples in which regulatory benefits accrue to the elderly (e.g., fine particulate matter, cancer). Examples in which regulatory benefits accrue to children (e.g., ozone) are not presented.
HOW BENEFIT ESTIMATES ARE AFFECTED BY
Age Distribution of Lifesaving Benefits
Change in Benefits When Statistical Life-years Are Used Instead
|Distributed uniformly by age
|Concentrated among children||
Normalized benefits much greater than 100
|Concentrated among young adults||
Normalized benefits greater than 100
|Concentrated among middle-aged adults||
Normalized benefits less than 100
|Concentrated among the elderly||
Normalized benefits much less than 100
In the spirit of Ackerman and Heinzerling, who appear to have been first to use the phrase, OMB Watch accuses the Bush administration (and now Dudley) of “devaluing the elderly.” They illustrate the point with the following stock photo. OMB Watch says counting life-years “systematically bias[es] the process against regulation aimed at preventing cancer or other diseases of old age, which have long latency periods.” The irony is that counting lives instead systematically biases the process again st her grandchildren.
Pub L. 108-199, Sec. 419 [118 STAT. 416].
OMB Watch led the charge for (and succeeded in obtaining) an appropriations rider prohibiting agencies from spending funds to apply different monetary values for statistical lives (VSLs) for adults of different ages in regulatory impact analyses during FY 2004:
None of the funds provided in this Act may be expended to apply, in a numerical estimate of the benefits of an agency action prepared pursuant to Executive Order No. 12866 or section 312 of the Clean Air Act (42 U.S.C. 7612), monetary values for adult premature mortality that differ based on the age of the adult.
The Opposition Report cites this accomplishment (p. 40, footnote 97) but does not mention that it expired over a year ago. Ironically, it’s not clear that this language prohibited agencies from estimating and using values for statistical life-years (VSLYs). By definition, a VSLY is invariant by age, which places it outside the domain of prohibited metrics.
“60-day notices” are required under the Paperwork Reduction Act. Agencies are required to respond to comments they receive in response to 60-day notices in their Information Collection Request they submit to OIRA.