2.1. Discount rate

The proposed social rate of time preference of 1.7% on page 76, line 3,525, reflects more fine-tuning than is warranted. For that rate, I would suggest 2% rather than 1.7%. Regardless of what number the OMB selects as the focal rate, it is essential to provide a strong economic justification for that number to avoid regulatory impact assessments (RIAs) being overturned by the courts because of their choice of the discount rate. I review these cases relating to how agency discount rates have fared in the courts in my peer-reviewed article, “The Social Rate of Discount: Legal and Philosophical Underpinnings.” Annual Review of Financial Economics 15: 127–145 (2023), also available as SSRN Working Paper 4083202. Simply citing OMB guidance is not sufficient for agencies unless there is compelling economic evidence in support of the selected rate. This issue is likely to be particularly prominent for analyses that adopt long-term discount rates different than the base social rate of discount.

I believe that OMB should continue to advocate for the reporting of benefits and costs using multiple discount rates. I advocate 3% as an additional discount rate of interest for two reasons. First, if the current low interest rates are the result of a Federal Reserve policy that will not continue, the 3% rate will be more appropriate than 2%. Second, the use of a 3% rate is consistent with past Circular A-4 guidance and, consequently, it will make possible a comparison of RIA results with policies analyzed using the previous version of Circular A-4. It is noteworthy that 3% is in line with the 30-year average of the 10-year real Treasury interest rate, 1985–2014, as indicated on page 20 of the OMB Preamble document. Due to the long future time horizon of policy analyses, such as those pertaining to climate change, interest rate performance during such recent periods is not irrelevant. Given the effect of Federal Reserve policies in depressing interest rates in recent decades, which is a monetary policy that is unlikely to continue indefinitely, a 3% rate remains a reasonable rate.

There is also support in various submissions for reporting the results using multiple discount rates. The public submissions by Susan Dudley (Reference Dudley and Viscusi2023), OMB-2022-0014-0129, and Art Fraas et al. (Reference Fraas, Graham, Krutilla, Lutter, Shogren and Thunström2023) OMB-2022-0014-3917 include two previous Office of Information and Regulatory Affairs (OIRA) administrators in Republican administrations, Susan Dudley and John Graham. The longevity of the revised A-4 document will be enhanced if administrations with a different perspective are able to use the RIA results. Dudley does not advocate a specific alternative discount rate, but suggests that agencies present results using multiple discount rates (see Susan Dudley, Public Submission, pp. 12–13). Fraas et al. (Reference Fraas, Graham, Krutilla, Lutter, Shogren and Thunström2023) advocate 3% and 5% as the two discount rates (see Fraas et al., (Reference Fraas, Graham, Krutilla, Lutter, Shogren and Thunström2023) Public Submission, p. 3). Their submission provides details regarding their rationale for this approach and extensive references. The pertinent factors include the influence of the Federal Reserve Bank policies on market rates of interest as well as references suggesting that these policies are not likely to continue over the long term.

2.2. Distributional analysis

In my view, this section should be substantially reworked, strengthening the guidance for providing distributional impact information, but with much of the discussion of weighting eliminated. In particular, I would eliminate the “Section E: Weights and Benefit–Cost Analysis” that begins on page 65.

The most promising component of this section is the advocacy that RIAs report the distributional effects of their policies. This is a very worthwhile advance. However, the draft Circular A-4 provides agencies with too much leeway regarding the dimensions on which to report the distributional effects. As Kniesner and I document in Thomas J. Kniesner and W. Kip Viscusi. Reference Kniesner and Viscusi2023b. “Promoting Equity through Equitable Risk Tradeoffs,” Journal of Benefit-Cost Analysis, 14, no. 1: 8–34, in their implementation of the Biden Administration’s Justice40 equity effort, agencies have used a myriad of different profiles to determine who is disadvantaged. These dimensions include rural residence, being over age 65 years, living near a highway, and demographic characteristics, such as being Black or Hispanic. Consistent with our Journal of Benefit–Cost Analysis article, I propose that OIRA establish standardized income-based categories for reporting distributional effects so that the effects across agencies can be compared. Thus, my proposal is that OMB specify a specific, standardized set of income-based categories, which would be a concrete implementation of the government’s identification of categories for the distributional effects on page 62, line 2,872, and the importance of having consistency in these analyses, as noted on page 62, line 2,879. Such a structure will facilitate comparisons of distributional effects across different agencies. Of course, agencies could supplement these distributional breakdowns with other breakdowns of distributional effects. For example, Hamilton and I found that an analysis of risk levels by minority status was also useful in our analysis of over 200 Superfund sites in James T. Hamilton and W. Kip Viscusi. Reference Hamilton and Viscusi1999. Calculating Risks? The Spatial and Political Dimensions of Hazardous Waste Policies. Cambridge: MIT Press.

The discussion of distributional weights ignores the substantial implicit redistribution that takes place by using average VSL levels and average unit benefit levels rather than population-specific values. This notion is implicit in the discussion on page 66, line 3,044, but not examined in detail. My 2023 Journal of Benefit–Cost Analysis article with Kniesner (Thomas J. Kniesner and W. Kip Viscusi. Reference Kniesner and Viscusi2023b. “Promoting Equity through Equitable Risk Tradeoffs.” Journal of Benefit-Cost Analysis, 14, no. 1: 8–34) documents the extent to which there is redistribution as a result of using average population-wide values for the VSL. The magnitude of the discrepancy is greater with large income elasticities of the VSL. Using a VSL income elasticity of 1.0, which agencies such as Department of Transportation (DOT) now use, the article reports that the ratio of the average VSL to the VSL for the target population is 1.9 for the 10th percentile of the worker distribution, 1.4 for the 25th percentile of the worker income distribution, 1.2 for the median Black worker, and 1.3 for the median Hispanic worker. In effect, these disadvantaged groups already receive a premium as a result of using average benefit values for that group. The income elasticity discussion of distributional weights in Table 1 of OMB’s Preamble is based on three articles that I coauthored. The Preamble uses these results as part of its advocacy for distributional weights. However, the main implication of this research has the opposite message, which is that there is substantial redistribution that is already taking place by using an average VSL for all. In effect, the baseline for conceptualizing what redistribution is needed to achieve equity would be quite different if the population-specific estimates of the VSL were used rather than the population average. Similar concerns are raised by Scott Farrow, Society for Benefit–Cost Analysis blog, “On Balance: When All Lives Matter Equally: Equity Weights for BCA by Combining the Economics of VSL and US Policy,” on 16 March Reference Farrow2021.

The draft document proposes an explicit distributional weighting scheme for which there is no consensus in the economics literature or as a policy matter. A principal underlying assumption pertains to the diminishing marginal utility of income. For any individual, there is generally diminishing marginal utility of income. However, empirical evidence making comparisons across different individuals is less clear, as individual utility functions are only estimated up to a positive linear transformation, such as a + bu( x ) (see my articles: W. Kip Viscusi and William Evans. Reference Viscusi and Evans1990. “Utility Functions that Depend on Health Status: Estimates and Economic Implications.” American Economic Review, 80, no. 3: 353–74, and W. Kip Viscusi. Reference Viscusi2019. “Utility Functions for Mild and Severe Health Risks.” Journal of Risk and Uncertainty, 58, no. 2/3: 143–66, among others). More generally, see John Pratt, Howard Raiffa, and Robert Schlaifer. 2001. Introduction to Statistical Decision Theory. Cambridge: MIT Press.

In justifying distributional weights, the Preamble, page 14, line 615, alludes to the role of happiness surveys. However, self-reported happiness measures are inconsistent with the lifetime trajectory of the VSL. In particular, the happiness measure trajectory over the life cycle is opposite that of the VLS. Happiness scores over the lifetime display a U shape, whereas the VSL has an inverted U-shaped relation. There are other reasons why happiness scores should not supplant market-based WTP measures of mortality risk reduction values (see W. Kip Viscusi. Reference Viscusi2020. “Wellbeing Measures of Mortality Risks: Life-Cycle Contradictions and Ordinal Index Challenges.” Behavioural Public Policy, 4, no. 2: 245–253).

2.3. Scope of the analysis including geographic scope

The discussion of intertemporal scope issues was on point. However, I disagree with the discussion of geographical scope. Agencies should be required to report the benefits to the United States, including benefits to US citizens and military personnel who are abroad. The benefits to the United States should serve as the primary analysis rather than possibly using global benefits as indicated on page 10, line 393. Instead of saying that it is “generally appropriate” to present impacts on the United States, I believe that it is always appropriate. This information is important for three reasons. First, statutory guidance frequently specifies that the objective of the statute is to provide benefits to the “Nation,” not the world. While there has been an effort to emphasize benefits to the world with respect to climate change policies, the stated objective of the US policies is not to promote worldwide social welfare but to reflect the preferences of the US citizenry. Second, knowing the benefits to the United States is essential to better understand the equity implications for the United States. The concerns with respect to equity expressed in Biden’s executive order cannot be addressed without this knowledge. Third, in its 2022 analysis of the social cost of carbon (SCC), the Environmental Protection Agency (EPA) uses the US VSL to value the US mortality costs, but uses an income-adjusted value for other countries. In the absence of information that distinguishes the United States from the rest of the world, it is impossible for independent analysts to assess the global mortality costs based on different benefit transfer assumptions.

When appropriate, as in the case of global warming policies, I also support the reporting of the global benefits. This information is valuable for two principal reasons. First, the global benefit value is what is pertinent to determining the economically efficient global climate change policy (see W. Kip Viscusi. Reference Viscusi2023. “The Social Rate of Discount: Legal and Philosophical Underpinnings.” Annual Review of Financial Economics, 15: 127–145, also available as SSRN Working Paper 4083202). Knowing this value enables the United States and other countries to set globally efficient targets for climate change policies. Second, even from the standpoint of the domestic benefits, the value of any reciprocity resulting from other countries undertaking effective climate change policies because of the US actions is a domestic benefit that should be recognized. In addition, if there is altruism among the US citizens for the well-being of those in other countries, that altruism also should be recognized as a US benefit. My article with Ted Gayer (Ted Gayer and W. Kip Viscusi. Reference Gayer and Viscusi2016. “Determining the Proper Scope of Climate Change Policy Benefits in U.S. Regulatory Analyses: Domestic versus Global Approaches.” Review of Environmental Economics and Policy, 10, no. 2: 245–63) recognizes each of these influences. However, ultimately, the benefit number that should be used for the SCC should be conceptualized as the benefits that can be traced to the benefit derived by the United States either directly or indirectly.

2.4. Development of analytic baselines

My main suggestion is that the status quo serve as the baseline unless the RIA provides empirical evidence, specific evidence of future policy changes, or other regulatory guidance that provides a credible basis to assume a different temporal pattern for benefits and costs. Perhaps, the proposed Circular A-4 could strengthen the role of empirical evidence noted on page 13, line 551. Our research on EPA analyses of Superfund sites (see James T. Hamilton and W. Kip Viscusi. Reference Hamilton and Viscusi1999. Calculating Risks? The Spatial and Political Dimensions of Hazardous Waste Policies. Cambridge: MIT Press) found that the EPA often asserted hypothetical changes in land use that affected assessed benefits, even though there was no sound empirical basis for this assessment. In effect, I am proposing a stronger language in A-4.

2.5. Unquantified impacts

This section seemed fine and is of potentially broad significance. For example, many non-monetized effects involve health impacts for which it is difficult to assign quantitative measures of the impact, such as the number of people exposed to chemical levels above the reference safe dose.

2.6. Uncertainty

This section included a surprising reference to non-expected utility frameworks. Expected utility theory is generally accepted as the normative reference point. Departures from this theory are sometimes labeled as a form of irrationality and often serve as an indicator of potential market failure. To adopt this approach to dealing with uncertainty is inconsistent with Circular A-4’s treatment of behavioral market failures.

Risk neutrality is desirable when the losses are spread broadly across the population. When individuals incur substantial losses, their personal risk aversion does come into play and is a legitimate concern when monetizing benefits for financial losses.

It is important for OMB to emphasize that risk assessments should be guided by the mean risk levels, not the upper bound of the risk. There could be an elaboration of this issue around page 69, line 3,218. Any presentation of confidence intervals in regulatory analyses that reports the upper bound of the risk should also report the counterpart lower bound value. This procedure is not the norm in analyses by agencies such as the EPA and the FDA, but it should be (see James T. Hamilton and W. Kip Viscusi. Reference Hamilton and Viscusi1999. Calculating Risks?: The Spatial and Political Dimensions of Hazardous Waste Policy. Cambridge: MIT Press; W. Kip Viscusi, James T. Hamilton, and P. Christen Dockins. Reference Viscusi, Hamilton and Dockins1997. “Conservative Versus Mean Risk Assessments: Implications for Superfund Policies.” Journal of Environmental Economics and Management, 34, no. 3: 187–206; and W. Kip Viscusi, Joel Huber, and Jason Bell. Reference Viscusi, Huber and Bell2019. “Responsible Precautions for Uncertain Environmental Risks.” Journal of Benefit–Cost Analysis, 10, no. 2: 296–315).

2.7. Behavioral economics and nudges

The proposed Circular A-4 and the Preamble recognize behavioral biases as a form of market failure, and they also suggest that nudges can serve as an effective policy instrument. Each of these aspects of the document requires that the underlying behavioral economics rationale is sound.

Consider first the role of behavioral economic factors generally. The existence of behavioral biases in narrowly defined experimental contexts should not serve as a sufficient rationale for government regulation. A case in point is that of the EPA-DOT fuel economy standards analyzed in Ted Gayer and W. Kip Viscusi. Reference Gayer and Viscusi2013. “Overriding Consumer Preferences with Energy Regulations.” Journal of Regulatory Economics, 43, no. 3: 248–264. The RIA was based on the assertion that consumers completely ignore the long-term fuel economy gains, and that government regulation was required to address this problem. While it is not impossible that there is some market shortcoming, the extent and prevalence of the intertemporal irrationality were simply asserted, not documented.

To address this problem of not properly documenting the behavioral failure, Ted Gayer and I have proposed that OIRA adopt a behavioral transfer test to serve much the same function for behavioral economics findings as does the more conventional benefit transfer test. The criteria we recommend for the behavioral transfer test include the following:

• – Does the evidence reflect the stakes, characteristics of decision-makers, opportunities for learning, and frequency of decisions comparable to the market context of interest?

• – Is the sample in the study reflective of the beliefs and preferences of the policy’s target population group?

• – Do respondents in the study understand what is being valued?

• – Is there sufficient evidence of consistency of responses and attention to the experimental task to make us confident of the findings?

• – Are the decisions comparable to market-based decisions with respect to the commodity in terms of the financial stakes and consumer attributes?

• – Was the experiment incentivized and published in a peer review outlet?

For additional support and elaboration of these issues, see W. Kip Viscusi and Ted Gayer. Reference Viscusi and Gayer2016. “Rational Benefit Assessment for an Irrational World: Toward a Behavioral Transfer Test.” Journal of Benefit–Cost Analysis, 7, no. 1: 69–91. Furthermore, see Ted Gayer (Reference Gayer2023), Public Submission, OMB-2022-0014-0127. Similar concerns, along with appropriate citations, are raised in Susan Dudley (Reference Dudley and Viscusi2023), Public Submission, OMB-2022-0014-0129. The proposed Circular A-4 alludes to the importance of recognizing behavioral distortions (p. 15, line 654) and behavioral biases (p. 15, line 662, and Section iv on pp. 18–19). The behavioral transfer tests that Gayer and I propose are applicable to all such matters.

The evidence in support of nudge policies is often similar in that it may be based on experimental contexts and small samples. Behavioral transfer tests are also applicable to studies of nudges. Among the most prominent nudge policies are informational policies, such as hazard warnings. These and other types of nudges are potentially effective policy instruments. Often, the choice is between nudges and policies involving financial incentives. Proper comparison of the performance of nudges and financial incentives requires that payments in the form of transfers, which typically constitute the main cost of financial incentive policies, be treated appropriately, consistent with the guidance in the earlier Circular A-4. Previous comparisons in the literature, other than in my work, have not recognized the proper treatment of transfers (see W. Kip Viscusi. Reference Viscusi2022. “Efficiency Criteria for Nudges and Norms.” Public Choice, 191, no. 3–4: 465–82, and W. Kip Viscusi. Reference Viscusi, Sunstein and Reich2023a. “Nudges Versus Financial Incentives.” In Cass R. Sunstein and Lucia A. Reich (Eds) Research Handbook on Nudges and Society, pp. 189–208. Cheltenham: Elgar Publishing, also available as SSRN Working Paper 4422704). Together, these articles provide a comprehensive review of the most prominent nudge and financial incentive policies in four policy areas of interest.

2.8. Fatality risks and the value of a statistical life

In a series of publications, I have advocated a concept of promoting risk equity through the use of the same VSL across the population, despite evidence of heterogeneity of the VSL (see W. Kip Viscusi. Reference Viscusi2000. “Risk Equity.” Journal of Legal Studies, 29, no. 2: Part 2, 843–71; W. Kip Viscusi. Reference Viscusi2018b. Pricing Lives: Guideposts for a Safer Society. Princeton: Princeton University Press; and Thomas J. Kniesner and W. Kip Viscusi. Reference Kniesner and Viscusi2023b. “Promoting Equity through Equitable Risk Tradeoffs.” Journal of Benefit–Cost Analysis, 14, no. 1: 8–34). From an equity standpoint, doing so treats all lives equally, irrespective of income, minority status, or age. The proposed guidance is consistent with this approach but does not require it. On page 51, line 2,348, the proposed guidance establishes a floor on the VSL for children, for whom a lower VSL is not permitted, but does not prevent agencies from adopting a child premium, as some agencies may do to justify their regulation. See the critique of Consumer Product Safety Commission’s (CPSC’s) proposed doubling of the VSL for children in Thomas J. Kniesner and W. Kip Viscusi. Reference Kniesner and Viscusi2023a. “Is a Child’s Life Twice as Valuable as an Adult’s?” Regulation, Summer, 11–12.

The proposed A-4 does not endorse the use of Quality-Adjusted Life Years (QALYs) for valuing mortality risks but does support the use of QALYs for non-fatal risks in many places, including caveats regarding QALYs. The revised A-4 should disavow the use of QALYs even more. The underlying assumption of QALYs is that the number of life years is the driving concern, where these years become quality-adjusted. This focus on counting the number of life years is inconsistent with the application of the VSL by agencies, which do not scale the VSL in terms of the number of remaining life years but instead generally use a standard VSL across the population. For non-fatal risks, OMB should continue to rely on WTP measures.

For fatal risks for which evidence based on occupational hazards is appropriate from a benefit transfer standpoint, agencies should use revealed preference data based on labor market studies. The applicability of general VSL estimates to traumatic risks such as those in transportation contexts is documented in W. Kip Viscusi and Elissa Philip Gentry. Reference Viscusi and Gentry2015. “The Value of a Statistical Life for Transportation Regulations: A Test of the Benefits Transfer Methodology.” Journal of Risk and Uncertainty, 51, no. 1: 53–77. The discussion in the proposed Circular A-4 of benefit transfer methods on page 37, line 1,670, is directly pertinent, since in the case of transportation accidents, it is feasible to establish the comparability of the VSL of transportation-related fatalities and the VSL from occupational risks more generally.

The best evidence based on labor market studies uses the Bureau of Labor Statistics (BLS) Census of Fatal Occupational Injuries (CFOI) mortality data. To the best of my knowledge, DOT is the only agency that relies exclusively on CFOI-based labor market studies of the VSL. They all should do so. Earlier mortality risk measures entail much greater measurement error. Moreover, studies based on these earlier risk variables have been shown to be subject to substantial publication selection effects, possibly overstating the VSL by 70–80%. Estimates based on the CFOI measures are less susceptible to such biases. There are additional publication selection biases that arise from using the “best estimate” from different studies rather than all estimates (see W. Kip Viscusi. Reference Viscusi2018a. “Best Estimate Selection Bias in the Value of a Statistical Life.” Journal of Benefit–Cost Analysis, 9, no. 2: 205–46; W. Kip Viscusi. Reference Viscusi2015b. “The Role of Publication Selection Bias in Estimates of the Value of a Statistical Live.” American Journal of Health Economics, 1, no. 1: 27–52; and W. Kip Viscusi. Reference Viscusi2018b. Pricing Lives: Guideposts for a Safer Society. Princeton: Princeton University Press.

Evidence from stated preference studies with respect to mortality risk valuation is subject to rampant publication selection effects and does not provide a sound basis for benefit assessment (see Clayton J. Masterman and W. Kip Viscusi. Reference Masterman and Viscusi2020. “Publication Selection Biases in Stated Preference Estimates of the Value of a Statistical Life.” Journal of Benefit–Cost Analysis, 11, no. 3: 357–79.

For very short life extensions, agencies can use the VSL year (VSLY). The underlying theory for this measure first appeared in Michael J. Moore and W. Kip Viscusi. Reference Moore and Viscusi1988. “The Quantity-Adjusted Value of Life.” Economic Inquiry, 26, no. 3: 369–88. The best measures of the VSLY are in Joseph E. Aldy and W. Kip Viscusi. Reference Aldy and Viscusi2008. “Adjusting the Value of a Statistical Life for Age and Cohort Effects.” Review of Economics and Statistics, 90, no. 3: 573–81. It is noteworthy that we find that the VSLY varies over the life cycle, which I have also found in subsequent studies. The result is inconsistent with a constant QALY. In addition, note that the proposed Circular A-4 document’s discussion regarding the heterogeneity of the VSL on page 50, including FN 85, is very dated, as it is based on a 2003 article by John Graham. My 2008 RESTAT article with Aldy finds that there is a pronounced inverted U-shaped pattern to the VSL, which I have found in other articles as well.

One benefit area for which I believe stated preference studies are desirable is for valuing illnesses that might not be comparable to traumatic injuries in terms of their morbidity effects. However, all such stated preference evidence should meet the criteria specified by the OMB for such studies and also include a detailed description of the health impacts involved. For example, many studies simply elicit a WTP for cancer risks generally, but the morbidity effects are not identical for all cancers (see W. Kip Viscusi, Joel Huber, and Jason Bell. Reference Viscusi, Huber and Bell2014. “Assessing Whether There Is a Cancer Premium for the Value of a Statistical Life.” Health Economics, 23, no. 4: 384–96, in which we find a 20% premium for bladder cancer, but we find a larger premium for blood cancer in Wesley A. Magat, W. Kip Viscusi, and Joel Huber. 1996. “A Reference Lottery Metric for Valuing Health.” Management Science, 42, no. 8: 1118–130.

Due to increases in income levels over time, agencies have been updating the VSL for income using a positive income elasticity of the VSL, often about 1.0. Estimates of the income elasticity of the VSL vary, and for the United States, they are in the range of 0.6 in W. Kip Viscusi and Joseph E. Aldy. Reference Viscusi and Aldy2003. “The Value of a Statistical Life: A Critical Review of Market Estimates throughout the World.” Journal of Risk and Uncertainty, 27, no. 1: 5–76; 0.25–0.63 in Hristos Doucouliagos, T.D. Stanley, and W. Kip Viscusi. Reference Doucouliagos, Stanley and Viscusi2014. “Publication Selection and the Income Elasticity of the Value of a Statistical Life.” Journal of Health Economics, 33: 67–75; 1.4 in Thomas J. Kniesner, W. Kip Viscusi, and James P. Ziliak. Reference Kniesner, Viscusi and Ziliak2010. “Policy Relevant Heterogeneity in the Value of Statistical Life: New Evidence from Panel Data Quantile Regressions.” Journal of Risk and Uncertainty, 40, no. 1: 15–31; a US Income Elasticity of 0.5 in W. Kip Viscusi and Clayton J. Masterman. Reference Viscusi and Masterman2017. “Income Elasticities and Global Values of a Statistical Life.” Journal of Benefit-Cost Analysis, 8, no. 2: 226–50; and 0.55 for affluent nations in Clayton J. Masterman and W. Kip Viscusi. Reference Masterman and Viscusi2018. “The Income Elasticity of Global Values of a Statistical Life: Stated Preference Evidence.” Journal of Benefit-Cost Analysis, 9, no. 3: 407–34. An income elasticity of 1.0 makes explaining income elasticity updates very straightforward. I have used that number in a couple of articles, and agencies often use that value as well. However, for purposes of an RIA, a lower income elasticity for the United States may have a stronger empirical justification.

2.9. WTA and WTP

The proposed guidance frequently suggests that policy analyses can use either WTP or WTA measures. This type of comment appears repeatedly as WTP and WTA are treated as being equally valid. See, for example, page 28, lines 1,257 and 1,262, which oddly refer to WTA and WTP as “similar.” While they are similar theoretically, they are often quite different empirically, as is noted on page 29, line 1,280. Empirical estimates of the WTA estimates are inconsistent with any effort to reconcile the WTA premium with rational economic behavior or attempts to explain the WTA–WTP discrepancy based on influences such as income effects. OIRA should insist on the use of WTP values, unless agencies can document the validity of WTA measures (see W. Kip Viscusi, Joel Huber, and Jason Bell. Reference Viscusi, Huber and Bell2012. “Reference-Dependent Valuations of Risk: Why Willingness-to-Accept Exceeds Willingness-to-Pay.” Journal of Risk and Uncertainty, 44, no. 1: 19–44; W. Kip Viscusi. Reference Viscusi2015a. “Reference-Dependence Effects in Benefit Assessment: Beyond the WTA-WTP Dichotomy and WTA-WTP Ratios.” Journal of Benefit-Cost Analysis, 6, no. 1: 187–206; and W. Kip Viscusi. Reference Viscusi2018b. Pricing Lives: Guideposts for a Safer Society. Princeton: Princeton University Press.

1. 2. Where the guidance reflects assumptions, are they supported by the theoretical and empirical peer-reviewed academic literature in economics or other relevant disciplines? If unsupported assumptions are identified, are there alternatives you would recommend? Please provide supporting references for both parts of the response—concerns about assumptions, if any, and suggested alternatives.

2.10. Discount rate

As noted above, basing the future discount rate on the recent period assumes that the monetary policies of the Federal Reserve will continue in a similar manner in the future, which, many economists do not believe, will be the case. See the references in the public submission by Art Fraas et al., (Reference Fraas, Graham, Krutilla, Lutter, Shogren and Thunström2023) OMB-2022-0014-3917.

2.11. Distributional analysis

My discussion in the Distributional Analysis heading for Question 1 identified two main areas where the assumptions were not warranted. The proposed Circular A-4 imposes a particular social welfare function incorporating distributional weights that is not consistent with standard WTP measures. There is no widespread support for this arbitrary social welfare function. Moreover, it neglects the extent of redistribution that will already occur, based on the application of average WTP values for all those affected by a policy. I provided the reference to my 2023 Journal of Benefit–Cost Analysis article with Kniesner on that issue. The other area of disagreement with the assumptions in this section was the claim that happiness scores are a useful measure of benefit values. I provided my article reference above.

2.12. Scope of the analysis

The discussion of the geographic scope of the analysis in the proposed Circular A-4 ignored the statutory guidance with respect to whether benefits to the Nation are consequential. I propose that domestic benefits should always be presented, even if global benefits are calculated. Doing so is also essential to understanding the domestic distributional impact of policies.

2.13. Fatality risks and the value of a statistical life

The proposed Circular A-4 guidance illustrated the use of VSLY, indicating on page 50, line 2,288, that for someone with a 40-year remaining life expectancy, the calculated VSL would equal 40× VSLY. While the logic is correct, someone with a 40-year life expectancy would be in their 40s around the peak of their lifetime VSL trajectory, so there is no need to resort to the use of VSLY in this instance. A typical mean age in a VSL study is about 40 years. For example, the mean age is 41 years in Thomas J. Kniesner, W. Kip Viscusi, Christopher Woock, and James P. Ziliak. Reference Kniesner, Viscusi, Woock and Ziliak2012. “The Value of Statistical Life: Evidence from Panel Data.” Review of Economics and Statistics, 94, no. 1: 74–87.

1. 3. Does the guidance appropriately recognize and account for potential challenges for implementation (e.g., technical feasibility or constraints on data availability or other resources)?

2.14. Distributional analysis

By failing to establish OMB guidance for the categories for which information regarding distributional impacts is required, OMB may be in a position where it is difficult to make distributional judgments across policies. With respect to implementation, let me also emphasize that OMB should not readily accept any claims that it is not feasible to do a distributional analysis. Officials at the EPA and in various District of Columbia organizations told us that analyzing the distributional effects of Superfund sites was not feasible, but we did it for over 200 sites using information at the block group level, which is more refined than the analyses being undertaken under Justice40 (see James T. Hamilton and W. Kip Viscusi. Reference Hamilton and Viscusi1999. Calculating Risks? The Spatial and Political Dimensions of Hazardous Waste Policies. Cambridge: MIT Press.

1. 4. Do you have any other suggestions for improving the completeness, objectivity, and/or transparency of agency regulatory analyses? If so, how might these be incorporated into guidance?

2.15. Scope of the analysis

The most important suggestion for improving the completeness and transparency of the analysis is to report domestic impacts as well as global impacts for policies with international impacts, such as climate change policies.

2.16. Uncertainty

The most important way to improve objectivity in the analysis is to use mean risk estimates and to report lower bounds in any situation in which upper bounds are also reported.

1. 5. What practices might be identified in the guidance to encourage accounting for non-monetized (possibly also non-quantified) effects?

This section of the guidance was very good as written.

1. 6. Do you have suggestions that would improve the clarity and logical presentation of the guidance and/or ease the execution of analyses?

The guidance was clear and presented logically. However, the guidance document is very long, in part because there is a detailed effort to justify all the components of the guidance. Substantial trimming of the background discussion would make the document more accessible.

1. 7. Should the guidance include suggestions of broadly useful data sets? If so, which data sets, and how should this information be presented in the guidance? How should the guidance reflect best practices related to data quality (including timeliness of data)?

The guidance is likely to serve as a reference for RIAs for many years. During that time, data set availability will change, as will best practices. Should OMB be motivated to provide such information, this can be done apart from this guidance document.

1. 8. Do you have any additional recommendations for ensuring that the guidance and associated methodologies are supported by the theoretical and empirical peer-reviewed academic literature in economics or other relevant disciplines? If so, please provide them here.

In my responses to Question 1, I provided comments on the main areas of OMB interest as well as topics that were not explicitly indicated as matters for which comments were requested, such as behavioral economics and nudges, fatality risks and the VSL, and the use of WTP and WTA values.

3.1. Discount rate

Agencies should show multiple discount rates. In addition to the administration’s proposal, which I would change to 2%, I would also recommend that agencies provide results based on a 3% rate. Some other groups also advocate showing results using higher discount rates, such as 5%.

3.2. Distributional analysis

The proposed presentation of distributional impacts is likely a long-term improvement in policy analyses, particularly if done correctly. My proposal that the OMB and agencies be cognizant of the implicit distributional consequences of applying uniform unit benefit measures to all should be incorporated in any subsequent welfare analysis. The proposed social welfare function weights in Circular A-4 are arbitrary, are not generally accepted, and will undermine the constructive progress that Circular A-4 will be able to achieve with respect to distributional impacts.

3.3. Geographical scope

Economic analysts and presidential administrations differ in the weight that they would place on domestic and global impacts. Requiring agencies to present both sets of information in situations in which global benefits are consequential will enable administrators to weigh these impacts in their preferred manner.

3.4. Uncertainty

Reporting the mean risk assessments and symmetric presentation of upper and lower bounds of the risk will lead to the greatest expected health improvements from government policies.

3.5. Behavioral economics and nudges

Adopting the behavioral transfer tests as suggested by Ted Gayer and me will restrict the introduction of behavioral economic concerns to the situations where they are most appropriate.

3.6. Fatality risks and the value of a statistical life

Agency estimates of the VSL and the VSLY should be based on revealed preference data from the labor market using mortality risk data from the BLS CFOI. There is no need to augment this information with stated preference evidence, except for health impacts, for which there is no comparable labor market data, and for which morbidity effects are greater than for traumatic accidents, as in the case of cancer.

Throughout their analyses, agencies should rely on WTP measures, not WTA. Restricting the focus to WTP will also eliminate the applicability of QALYs.

3.7. Final comment

Overall, the proposed A-4 document is very well done and reflects a tremendous amount of work, as well as a careful examination of the pertinent issues.