Inequality is a burning issue in our society but plays only a limited role in the design of regulations. This article defends two features of the existing system that do promote equality: the controversial practice of using uniform valuations of life and health, regardless of income, and the use of disparate-impact analysis in rulemaking. Rather than relying on arguments for using regulation as a possible form of income redistribution or remedy for existing racial disparities, it argues that justice requires devoting equal resources to prevent equal harms. The reality is that low-income communities and communities of color often suffer the greatest harms (and not just by chance). By using much more granular approaches to determining who is exposed to risks and their vulnerability to harm, agencies could do far more to implement this principle, identify the needs of disadvantaged communities, and effectively address them.

Since 1981, the dominant method of assessing new regulations has been cost–benefit analysis.1 Cost–benefit analysis is based on willingness to pay,2 and the rich can pay more for safety and good health than the poor. Therefore, the government’s cost–benefit analyses3 should place a higher value on the lives of the rich than on the lives of the poor.4 Similarly, the time of the affluent is economically more valuable than the time of the poor, as measured by the amount of compensation they receive for working. Accordingly, the government favors projects that reduce commuting times for the affluent rather than the poor.5 And in other settings, the time of some people is considered to be an order of magnitude more valuable than the time of others.6 The obvious conclusion seems to be that regulation is part of the inequality problem, not its solution.7

This article focuses on ways to make regulations more responsive to the needs of the poor and people of color.8 The most obvious approach is making equality an independent goal in designing rules. Agencies do already take some heed of income inequality at this stage. They elevate the interests of the poor by deviating from theory and treating the value of life as constant regardless of wealth. They take race into account by applying disparate-impact analysis to proposed regulations. I support both policies. But giving equality more weight as an independent goal turns out to be difficult for a combination of legal and practical reasons.

Surprisingly, the most promising way to respond to inequality begins at an earlier stage, when agencies are mapping (sometimes literally) the problem they are trying to address. Race and economic inequality help drive who is at risk (risk exposure) and how much they will be harmed (vulnerability). Drilling deeply into individual variations in exposure and vulnerability would automatically reflect the racial and economic disparities that those individuals face. In turn, mapping risk and exposure would shape regulatory solutions without the need for explicit consideration of inequality in regulatory design.

For instance, as discussed later, disadvantaged communities are often subject to the most pollution and suffer greater harm from any given level of pollution. When an agency discovers this to be true, it has several options. It can make the regulation as a whole more rigorous, providing those communities and others with greater protection. It can add features to the regulation providing special restrictions on sources that impact areas with elevated pollution levels. It can also address this impact in various other ways: by requiring more rigorous monitoring of those sources, making enforcement actions against them a priority, or launching (or working with other agencies to launch) additional rulemakings aimed at pollution hotspots. Compared with other options for addressing inequality, an intensified focus on risk exposure and harm is administrable and legally unobjectionable. And it could do a great deal to lift the burdens on disadvantaged communities.

Making regulation responsive to inequality is important because inequality is important. Regulation should not be complicit in perpetuating inequality. While ignored in cost–benefit analysis, inequality has become a central issue in American public discourse. And no wonder. Inequality is a stark and growing reality along multiple, interrelated dimensions: race, income, and geography.

  • In terms of race, the Black infant mortality rate is twice as high as the rate for whites.9 Black median household income is only sixty percent that of white households.10 The wealth gap is much bigger than the income gap, with the typical Black household having only a tenth the wealth of its white counterpart.11

  • Income inequality in America is equally stark. By 2018, the top twenty percent of the American income distribution made as much as the entire rest of the population combined.12 As of 2019, the top ten percent of households owned three-quarters of all wealth, while one percent was owned by the bottom half of the population.13

  • And the life chances of Americans depend greatly on where they live. Metropolitan areas with populations over a million have contributed nearly all the recent population growth and have far outpaced smaller cities in growth in employment and economic output.14 As a recent article put it, “[s]uperstar cities … continue to attract talent and grow, while the economies of other cities and rural areas are left behind.”15

In a final example of particular relevance to this article, exposure to pollution is another form of inequality. Low-income people, Blacks, and Hispanics are consistently exposed to higher levels of dangerous PM2.5 pollution,16 resulting in more deaths among those groups.17 Increased mortality is especially pronounced in areas where the Black and Hispanic populations are most concentrated.18 While PM2.5 levels declined nationally during the last decade, the rate of decrease was highest in white areas.19 Thus, although pollution levels in both predominantly white and predominantly Black or Hispanic areas have declined, the relative disparity between them has increased.20 Perhaps most strikingly, one study found a clear connection between redlining decades ago and current levels of air pollution in U.S. cities.21 The environmental consequences of these inequalities call out for attention.

In considering what role, if any, the regulatory system should play in responding to such inequalities, this article makes two major contributions.22 First, it shows that equality norms are already embedded in the government’s version of cost–benefit analysis. To reflect differences in how much different groups are willing to pay to avoid risks, economic efficiency would require placing a lower value on the lives and health of the poor versus the lives of the rich.23 But as practiced by the government, cost–benefit analysis doesn’t work that way. Other scholars have recognized this fact but viewed it as a pragmatic adjustment to uninformed public attitudes. I will argue instead that it embodies a distinctive vision of equality. Under that concept of equality, individuals have equal entitlements to protection against harm, regardless of wealth or other personal characteristics. Society as a whole as well as actors creating risks have a duty to devote equal resources to preventing equal harm.24

A second contribution involves racial justice. Efforts to aim regulations at overcoming the disadvantages suffered by people of color are likely to run into legal and political obstacles. Before cost–benefit analysis takes place, however, regulators need to assess the severity of the risks they seek to regulate. This article is the first to show how race, income, and other factors, such as education level, subtly enter into risk assessment and result in greater protection for the disadvantaged. That approach could be deepened considerably with greater attention and better data, including much more granular information on exposures and demographics.25 Therefore, I propose, regulators should focus more intently on issues of unequal exposure and vulnerability to remedy the plight of disadvantaged communities.

Because vulnerability and exposure determine the extent of harm, this approach is consistent with the principle that justice requires society to invest equal resources to prevent equal harms, regardless of the identities of the victim. That focus has the advantages of a strong theoretical basis in public health and of wider acceptability, as opposed to reorganizing regulation analysis around racial and economic inequality. It also provides an indirect way of addressing economic and racial inequalities, because they play important roles in determining exposure and vulnerability.

I will focus on environmental regulation because it looms so large in the regulatory state, accounting for a large majority of the most expensive regulations. We can best address problems by knowing exactly who is most harmed, and that will frequently turn out to track racial and economic divisions. The presumption should be that equal harms deserve equal resource commitments, which may require allocating more resources to areas where exposure levels or vulnerabilities are high.

As background to the analysis, Part II will begin with a description of how the practice of regulatory impact analysis has evolved in the past four decades. It also explains how the environmental justice movement has influenced (or more often failed to influence) regulatory practice.26 This movement has stressed the failure of the regulatory system to take into account the disadvantages experienced by low-income communities and communities of color. Part II will also describe unsuccessful efforts by past presidential administrations to address inequality in regulatory decision-making and the more intense efforts now underway in the Biden administration.

Part III addresses economic equality and its role in regulatory policy, the issue that has gotten the most attention from scholars. Economic purists argue that net regulatory benefits should be the sole focus of regulation, while issues of income distribution should be addressed through government tax and spending programs. I side with other critics in rejecting this argument. The regulatory system’s capacity to directly address income inequality is limited, however.

Part IV discusses racial inequality, an issue raised most forcefully by the environmental justice movement. There are clearly legal limits on how much the regulators can take race into account, though some of those limits are better defined than others. In my view, use of disparate-impact analysis in crafting regulations should be legally unproblematic. I conclude, however, that regulatory agencies probably lack the legal authority to adopt the goal of remedying historical or structural injustices. There is also a non-negligible constitutional objection to doing so, though not one that I think should succeed under current law.

Before the brief concluding section of the article, Part IV turns to a topic that has escaped notice in the scholarly literature: the ways in which risk modeling has sometimes, almost by accident, helped mitigate issues of inequality. Exposure is an obvious factor in risk assessment, and exposure models have become increasingly granular geographically. Risk assessment also sometimes takes into account the greater vulnerability of certain populations to harm from the same level of exposure, particularly the poor and people of color. It is little wonder that this has previously escaped notice because it is buried deep in the technical details of risk assessment.

With better data and analysis, much more could—and should—be done to address differential exposure and vulnerability. Doing so involves few of the conceptual, legal, or institutional problems that plague explicit attention to economic and racial inequality in regulatory analysis. Consequently, in practical terms, a closer focus on the distributions of risk and vulnerability may offer the best path to making regulatory decisions serve social justice.

At its heart, my argument is simple. If pollution and its impacts were equally distributed across the whole population, it is unclear whether anyone would propose designing environmental regulation as a redistributive mechanism to provide more benefits to the poor or people of color than to others. What makes environmental justice a compelling claim is that pollution exposure and harms are far from being equally distributed. If we attend to who is exposed and who is most vulnerable, we will by the same token address economic and racial inequality. A similar argument holds for other types of regulation.

The basic story is simple: For the past forty years, in Democratic administrations as well as Republican ones, cost–benefit analysis has been the dominant methodology for deciding regulatory issues. There is little reason to expect this to change anytime soon. In the meantime, issues of inequality, including environmental justice, have gotten short shrift.

While the outlines of the story are simple, it is important to unpack the history. Section A demonstrates the durability of cost–benefit analysis despite shifting partisan control and increasing political polarization. Although increasing polarization could ultimately result in sidelining cost–benefit analysis, no president in the four decades from Reagan to Biden has rejected it. As we will see, there have been some efforts to expand regulatory analysis beyond economic efficiency, but they have had little lasting effect. In particular, as Part B shows, efforts to reorient regulatory decision-making around environmental justice have had little traction.

A. Cost–Benefit Analysis and the Regulatory State

Soon after taking office, President Ronald Reagan signed Executive Order 12,291, which was aimed at improving the economic efficiency of government regulations.27 This executive order not only required cost–benefit analysis in rulemaking but also made it the dominant consideration. Section 2 instructed agencies that a major regulation could be issued only if a regulation’s potential benefits to society outweighed potential costs and net benefits were at a maximum. Review of agency cost–benefit analyses was conducted by the Office of Information and Regulatory Affairs (OIRA), a White House agency inside the Office of Management and Budget (OMB).28

Because the Reagan order was associated with conservative resistance to regulation, one might have expected a sharp change of direction after Democrats retook the White House in 1992. In 1993, however, President Bill Clinton issued an executive order maintaining Reagan’s basic framework though softening it in several respects, such as emphasizing the need to consider regulatory benefits even if they could not be quantified.29 Clinton’s rule was also intended to reduce the number of regulations needing OIRA approval and to make OIRA’s review more flexible.30 Still, the fundamental reliance on cost–benefit analysis that began with Reagan remained intact.

Clinton’s successor, George W. Bush, attempted to restore the stringency of regulatory review. In a 2007 executive order, he required agencies to “identify in writing the specific market failure (such as externalities, market power, lack of information) or other specific problem” warranting agency action.31 The order also expanded the scope of OIRA review to include guidance documents as well as legally binding regulations.32 Economic methodology remained central. Administrative guidance issued under Bush did, however, reiterate the possible relevance of nonquantifiable values in making regulatory decisions.33

Power changed hands again in 2008, with the Democrats retaking the White House. Once again, there was more continuity than change in regulatory review. President Barack Obama rescinded President Bush’s executive order on cost–benefit analysis,34 and the government then reverted to the Clinton-era version of the cost–benefit mandate.35 To the dismay of progressives, the Obama administration turned out to be a staunch supporter of cost–benefit analysis, requiring rigorous documentation in support of proposed regulations.36

President Trump layered additional restrictions on top of cost–benefit analysis. Although he reaffirmed the Clinton version of cost–benefit analysis, he also imposed two important new limitations on government rulemaking. One limitation was the so-called two-for-one rule. Section 2(a) of Executive Order 13,771 provided that “whenever an executive department or agency (agency) publicly proposes for notice and comment or otherwise promulgates a new regulation, it shall identify at least two existing regulations to be repealed.”37 Section 2(b) of the order directed all agencies that “the total incremental cost of all new regulations, including repealed regulations, to be finalized this year shall be no greater than zero.”38 Notably, however, if a regulation reduced expenses for consumers or business, the fact that a rollback increased those expenses would not be counted as one of a rollback’s costs.39

The theory behind cost–benefit analysis is that government should try to maximize the net benefits of regulations rather than simply avoid regulatory costs as such. Trump’s directive, however, prioritized cost reduction over maximizing net benefits.40 Perhaps this shift in emphasis signaled deeper dissatisfaction with cost–benefit analysis, which may curtail its use by future conservative presidents. Perhaps this overlay on cost–benefit analysis reflected libertarian values at odds with the basically utilitarian outlook of cost–benefit analysis. As Michael Livermore and Richard Revesz painstakingly documented in a recent book, the Trump administration’s actual cost–benefit analyses were often so poorly executed as to raise questions about either official competence or good faith.41 Yet the Trump administration did leave the basic framework of cost–benefit analysis intact and, at the very least, felt compelled to go through the motions of applying it.

Not surprisingly, Trump’s Democratic successor abandoned this anti-regulatory overlay on the rulemaking process. Upon taking office, President Joseph Biden immediately signed an executive order repealing Trump’s executive orders relating to regulatory review.42

Biden also issued a memorandum to federal agencies on “modernizing regulatory review.”43 The memorandum instructed the Director of OMB to provide recommendations on how to improve the regulatory process. Like his Democratic predecessors, Biden called for changes in regulatory review to allow greater consideration of values other than economic efficiency. OMB’s recommendations, the memorandum elaborated, “should provide concrete suggestions on how the regulatory review process can promote … racial justice” as well as “proposals that would ensure that regulatory review serves as a tool to affirmatively promote regulations that advance these values.”44

Biden also issued an executive order about cost–benefit analysis for rules relating to climate change.45 Section 5 of the order addressed the social cost of carbon, which economists define as the amount of harm caused by the addition of a single ton of carbon dioxide to the atmosphere.46 As the executive order explained, “[a]n accurate social cost is essential for agencies to accurately determine the social benefits of reducing greenhouse gas emissions when conducting cost–benefit analyses of regulatory and other actions.” In an effort to encourage consideration of normative issues as well as economics efficiency, section 5(E) directs modifications to existing estimates “to the extent that current methodologies do not adequately take account of climate risk, environmental justice, and intergenerational equity.”

As we have seen, regulatory agencies like the EPA have been required to use cost–benefit analysis for more than four decades. There has been a great deal of continuity, at least in terms of the general approach, as shown by the enduring vitality of Clinton’s Executive Order 12,866. Democratic administrations have tried to make cost–benefit analysis less rigid by encouraging consideration of values other than economic efficiency, while Republican administrations have hewed more closely to Reagan’s approach, with the exception of Trump’s effort to hinder the issuance of even regulations with positive net benefits for society.

Cost–benefit analysis has become entrenched despite a continuing, sometimes strident debate about its legitimacy. Although many economists and some legal scholars favor the use of cost–benefit analysis for government regulation,47 environmentalists and other progressives are often sharply opposed.48 For instance, two leading environmentalist critics49 have contended that “cost–benefit analysis promotes a deregulatory agenda under the cover of scientific objectivity.”50 Unlike these critics, some writers who are sympathetic to environmental protection embrace cost–benefit analysis.51 Although they view it as just one input into the ultimate regulatory decision,52 Richard Revesz and Michael Livermore have argued that cost–benefit analysis is needed to determine when further spending on risk reduction is no longer worthwhile.53 They argue that just as environmental impacts of actions must be taken into account, so should the economics of regulation.54

Despite the consistent use of cost–benefit analysis by regulators for forty years, its future is not completely assured. While some hail the arrival of the cost–benefit state, others doubt that the courts have gone so far as to require use of this technique.55 Conservative support for cost–benefit analysis is wavering, and conservatives, worried about broad delegations of power, may deem cost–benefit analysis’s claim to encompass all relevant considerations as unacceptably open-ended.56 On the left, criticism remains endemic,57 and if anything the criticisms are harsher now.58 Yet for now, cost–benefit analysis is firmly embedded in regulatory practice despite its critics.

B. Equity and Regulatory Review: A Brief History

The executive directives discussed in the preceding section focused almost entirely on economic efficiency; that is, on maximizing the net benefits of actions without regard to how benefits or costs are distributed. However, they did make some efforts to incorporate general considerations of equity and environmental justice in particular. In comparison to the series of directives on cost–benefit analysis, presidential orders relating to equity have been vague and seem to have received little more than lip service.

1. Inclusion of Equity Factors Within Cost-Benefit Analysis

Clinton’s executive order on cost–benefit analysis instructed agencies to attend to the nonquantifiable benefits of regulatory actions.59 Under Bush, OIRA issued Circular A-4, which provides detailed direction to agencies on how to conduct cost–benefit analysis.60 The circular flags the relevance of distributional effects.61 Like Clinton, Obama made an effort to soften cost–benefit analysis. An Obama executive order indicated that agencies “may consider (and discuss qualitatively) values that are difficult or impossible to quantify, including equity, human dignity, fairness, and distributive impacts.”62 Despite this gesture toward deemphasizing the need for quantitative analysis, the Obama administration turned out to be a staunch supporter of cost–benefit analysis, requiring rigorous documentation from agencies in support of proposed regulations.

Although attention to distributional issues was allowed under Clinton’s executive order on cost–benefit analysis and was reaffirmed by Obama, it seems to have been rare in practice. A study of major environmental and safety rules from 2010 to 2013 found few that described how the risks being regulated were distributed across the population.63 And only three indicated whether the costs of the rule would fall on consumers, workers, and investors.64 A 2018 study found that agencies rarely identified the recipients of regulatory benefits or the bearers of regulatory costs,65 and another 2018 study found that agencies generally either failed to take distributional concerns into account or gave them only cursory attention.66

More recently, Biden’s memorandum on modernizing regulatory review instructed OMB to “propose procedures that take into account the distributional consequences of regulations, including as part of any quantitative or qualitative analysis of the costs and benefits of regulations, to ensure that regulatory initiatives appropriately benefit and do not inappropriately burden disadvantaged, vulnerable, or marginalized communities.”67 Whether Biden’s directive will be implemented any more fully than his predecessors’ remains to be seen.

2. Environmental Justice

The foundational executive action on environmental justice was a 1994 executive order by President Clinton.68 Section 1-101 of the order directed each agency to address “disproportionately high and adverse human health or environmental effects of its programs, policies, and activities on minority populations and low-income populations.”69 Section 6-102 of the order said that it was intended to “supplement but not supersede” an earlier Carter order enforcing civil rights laws.70 Much of the Clinton order was devoted to implementation measures that are beyond the scope of this article.

The Clinton order remains in effect today. Even the Trump administration was careful to certify that regulatory actions were unlikely to have “disproportionately high and adverse human health or environmental effects on minority populations, low-income populations and/or indigenous peoples as specified in Executive Order 12898.”71

Under Obama, the EPA issued Plan EJ 2014,72 which attempted to integrate environmental justice into rulemaking and more generally emphasize it in EPA activities.73 It defined environmental justice as “the fair treatment and meaningful involvement of all people regardless of race, color, national origin or income.”74 “Fair treatment” was defined as avoiding disparate impacts.75 The guidance document noted that “risk factors may be intrinsic in nature, based on age, sex, genetics, race or ethnicity, or acquired (such as chronic medical conditions, or smoking status); as well as extrinsic, non-biological factors such as those related to socioeconomic status, reduced access to health-care, health-care, nutrition, fitness and/or exposures related [sic] factors.”76

Two other White House documents addressed the role of environmental justice in assessing a project’s environmental impacts77 and the connection between environmental justice and Title VI of the 1964 Civil Rights Act.78 Several times, the Obama administration projected disproportionate regulatory impacts on minority communities and took action to reduce those impacts.79

These were exceptions, however. Out of four thousand EPA rules issued during the Obama administration, only seven seriously considered environmental justice.80 Only a few regulatory impact analyses covered unquantifiable benefits.81 In short, Obama’s efforts to promote consideration of distributional issues and other “soft” factors seemingly had little effect on agency behavior.82

President Biden has taken several steps to emphasize the importance of environmental justice. A “Day 1” executive order emphasized the administration’s determination “to hold polluters accountable, including those who disproportionately harm communities of color and low-income communities.”83 A summary of environmental justice actions at the end of Biden’s first year of office recounted his commitment to “ensur[ing] that federal agencies deliver 40 percent of the overall benefits of climate, clean energy, affordable and sustainable housing, clean water, and other investments to underserved communities.”84 Other actions included making a climate and economic justice screening tool a top priority; establishing a White House Environmental Justice Advisory Council, and making efforts “to reduce pollution burdens and exposures, including lead exposure and asthma disparities in children of color.”85

A Biden executive order on climate change also emphasized environmental justice. It directed agencies to “make achieving environmental justice part of their missions by developing programs, policies, and activities to address the disproportionately high and adverse human health, environmental, climate-related and other cumulative impacts on disadvantaged communities, as well as the accompanying economic challenges of such impacts.”86 What is not yet clear is how these commitments will cash out in concrete action by agencies.

Economic inequality is an increasing concern in the United States. Part III explores whether, and how, regulators might engage in efforts to combat income inequality. In theory, cost–benefit analysis is resolutely blind to the issue of income inequality. Its methodology is designed to express values in terms of the current wealth distribution rather than challenge it. Part A explains that stance, the arguments given in its favor, and the reasons—which I find compelling—for abandoning it. Part B shows how, at least in the American setting, cost–benefit analysis is practiced in a way that contains powerful egalitarian elements. Part C then turns to equity weighting, a methodology that attempts to more rigorously and consistently incorporate distributional considerations into the assessment of regulations. Like a number of other recent scholars, I conclude that the possible benefits of adopting this approach are outweighed by practical difficulties.

A. Cost–Benefit Analysis and Wealth (Theory)

To understand the relationship between economic inequality and cost–benefit analysis, it is helpful to begin with an example. One major valuation issue in cost–benefit analysis involves putting a monetary value on reductions in mortality risks. We could assign monetary values to different levels of mortality risk by looking at how much consumers are willing to pay for safer products, or how much income workers are willing to give up for safer jobs, or how much travel time people are willing to sacrifice for the safety benefits of driving more slowly. All of these would be different ways of determining the market value of safety.

If people demand $10,000 in return for accepting a one-in-a-thousand risk of death, they are implicitly valuing their lives at $10 million, since $10,000 is one-thousandth of ten million. To assign a value of $10 million per statistical life is the same as saying that people would demand $10 in return for running a one-in-a-million chance of death.

How does valuation on the basis of willingness to pay relate to economic inequality? The answer is obvious. When it comes to measuring the seriousness of risks by how much people are willing to pay to avoid them, the rich are clearly willing to pay more than the poor to avoid identical risks.87 For this reason, the value of a statistical life is linked to the wealth of the affected population. Thus, cost–benefit analysis should vary the value of a statistical life depending on the economic status of the people exposed to a risk. Or more bluntly, in terms of cost–benefit analysis, the lives of the poor should simply be worth less than those of the rich. For instance, all else being equal, we would care more about improving air quality in affluent areas than in impoverished ones.

Applied rigorously, cost–benefit analysis mirrors the current distribution of wealth in society, assiduously avoiding any consideration of how regulations might change that distribution. This does not necessarily mean that cost–benefit advocates endorse the current distribution of wealth, although some undoubtedly do. Rather, they think this issue should be addressed outside the regulatory sphere, for instance through progressive income taxes.88

A more sophisticated version of this argument, often associated with Louis Kaplow and Steven Shavell, is that the government should strictly separate two different tasks.89 One task is to use regulations and legal rules to enhance economic efficiency, thereby making the economic pie as large as possible. The other task is to ensure a fair distribution of income, which the government should do solely through taxes and spending.

The rationale for separating these functions is that using regulations of conduct to redistribute income causes two economic losses: it sacrifices economic gains that could have been obtained through a more efficient regulation, and it disincentivizes work and investment by transferring money away from the better-off. In contrast, when the government uses taxes or spending to redistribute income, only the second loss results. This is commonly known as the double-distortion argument, since both the inefficiency of the regulation and its redistributive effects are thought to distort economic choices.

To take a simple example, we might adjust the damage rules in tort law to provide more generous damages when the victim is poorer than the defendant. Since we would then have moved away from the economically efficient rule, there would necessarily be a direct efficiency cost to the rule change, in this case because we would be encouraging people to exercise an increased level of caution toward the poor beyond what is cost-justified. But there is also a second distortion. Just as someone considering an additional hour of work might be deterred by an increase in the marginal tax rate, so they might be deterred by knowing they are going to be subject to disadvantageous tort rules.90 Thus, in addition to the direct distortionary effects of the redistributive rule, it has the same distortionary effect on labor supply as an income tax. Hence, Kaplow and Shavell have maintained, we can always transfer the same amount of funds more efficiently just by adjusting the tax rate.91

Some critics challenge the view that redistributive regulations cause the same distortions as redistributive fiscal measures. Kaplow and Shavell’s claim is based on the assumption that the major inefficiencies of the tax system are its impact on incentives for work and savings. There is disagreement about the magnitude of those incentive effects. The evidence relating to the strength of the impact of taxes on work and saving is mixed, and the more important effect of taxes may be on socially wasteful tax-avoidance strategies.92 Clearly, the tax system causes inefficiencies unrelated to these incentive effects, as money is shifted from more socially productive enterprises into the hands of tax lawyers and accountants and is wasted on complicated efforts to game the tax laws. If these tax-avoidance methods represent much or even a major part of the inefficiencies of tax and spending, use of regulations to redistribute wealth may be warranted.93

Other arguments against Kaplow and Shavell have been made. Distortions can be kept small by spreading redistributive efforts across many different regulations, so that each regulation imposes small efficiency costs to achieve redistribution.94 Also, the redistributive effects of regulations may be less apparent than taxes or anti-poverty measures and hence less likely to cause reductions in work or savings. It is a little hard to believe that the wealthy would work harder or save more if the government provided them cleaner air than it provided to people in nearby poorer areas. If anything, they might stay home from work to enjoy the cleaner air.

A more obvious question is whether we can rely on the tax and spending system to adequately deal with massive income inequality. One issue is congressional gridlock,95 although presidents have arguably had some success in implementing tax changes administratively or inducing Congress to pass major tax and spending legislation.96 Also, the rich have disproportionate political influence via campaign spending. A more fundamental issue may be that redistribution through fiscal measures encounters more voter resistance than other redistributive policies.97 Voters tend to believe that people deserve the income they earn and that using taxes to redistribute income is therefore suspect. They also notoriously worry that giving money to the poor encourages wasteful spending and opportunistic behavior. Other forms of redistribution, such as providing the poor with food stamps or subsidizing medical care for poor children, encounter less resistance. One might expect that if the poor are particularly exposed to pollution, cleaning up the pollution may be similarly acceptable as an in-kind form of redistribution.

Zachary Liscow argued in a recent article that even if this is considered to be a psychological defect on the part of the public, it does mean that the public would give greater support to forms of redistribution other than tax and spending measures. As a result, he argued, fiscal measures will not fully reflect society’s view of the fair distribution of income.98

It may be a mistake, however, to write off the rationality of the public’s views. It may be true that if the goal is to improve the utility levels of the poor, transfers of cash from the rich to the poor are the way to go. Maybe the public is simply too uninformed or dim to grasp that fact. But it is also possible that the public does not think of equality in the same terms as welfare economists. Rather than thinking that society has a duty to maximize the overall level of welfare, the public may think that society has the duty to ensure that everyone has equal opportunity to access necessities such as food or medical care. This may make little sense if your ethical theory is that maximizing social welfare is the be all and end all of morality.99 But welfarism itself has problems, not least the lack of any clear definition of individual utility.100

Given the level of disagreement among philosophers over moral theory, it is by no means clear that we should dismiss the public’s view of economic equality as irrational. It seems particularly questionable whether, in a democracy, the government should disregard the values widely held by the public when considering issues of social policy.

In any event, as a practical matter, exclusive reliance on taxing and spending to deal with economic inequality involves a degree of wishful thinking. To be blunt, the optimal tax theory on which Kaplow and Shavell rely has zero policy traction.101 Moreover, as Richard Revesz has pointed out, there are particular reasons to doubt that tax policy can be responsive to the disproportionate impacts of environmental harms.102 Those impacts, and our knowledge about them, can shift fairly quickly, which would require unrealistically nimble responses in tax legislation.

B. Cost–Benefit Analysis and Wealth (in Practice)

As discussed earlier, given the willingness-to-pay metric of cost–benefit analysis, measures of regulatory benefits such as reduced mortality should take into account the wealth of the individuals affected.103 Failure to do so has major redistributive consequences.104

There are other examples of how regulatory analysis in practice has contained equity-enhancing provisions. For instance, in assessing the years of life saved by a health measure, Circular A-4 advises using the general population’s life expectancy rather than adjusting for disability, income, or membership in a “particular demographic or income group.”105 This treats all individuals of a given age alike regardless of their other characteristics.

In addition, in estimating the harm from illnesses such as asthma attacks or nonfatal heart attacks, the EPA uses fixed dollar amounts rather than adjusting for the wealth of the affected individual or, in the case of children, for the wealth of their parents. Indeed, it is not clear how valuation of children’s lives based on willingness to pay could work at all.106

These wealth-neutral valuations are hard to justify in terms of maximizing net benefits—the governing standard of cost–benefit analysis—except perhaps as a public relations necessity. The government’s one-time effort to place a lower value of life on the elderly was met by a tsunami of resistance from the aged, and the government had to beat a hasty retreat.107 More than public relations may be in play, however. The public’s commonsense view may be normatively correct. The arguments in its favor involve the idea that the appropriate amount to pay depends on who is paying and how their actions relate to the risk.

First, we might consider that society as a whole, represented by the government, has a duty to give equal respect to the lives of all individuals regardless of wealth or other personal characteristics.108 In a democracy, the government should be impartial toward its citizens rather than valuing them based on wealth, a norm reflected in the principle of one person, one vote. In the case of government projects intended to save lives, for instance, it is appealing to say that average taxpayers should be willing to pay as much to protect others as themselves—a form of the Golden Rule.

There is no evident moral justification for equating the amount society is willing to sacrifice to protect the life or health of an individual with the amount that a particular individual would have been willing to spend themselves.109 This is particularly clear in the case of children, who generally lack control of significant funds and therefore have no meaningful “willingness to pay.” Their parents may have varying amounts of resources and therefore varying willingness to pay, but parental resources do not seem relevant moral facts in considering how much society as a whole would be willing to pay to reduce a risk to the child.

It is not clear why wealth becomes a more relevant factor when the issue is society’s willingness to pay when the individual is an adult. The public’s failure to adopt that position could be seen as simply a public relations or political problem.110 Or it could be that the public is right about the stance the government should take toward its citizens.

Rather than focusing on the government’s moral duties, we might focus on the moral obligations of a regulated party, such as a polluter. This idea reflects the commonsense assumption that causing a harm is morally different from merely allowing the same harm to happen rather than intervening. That is not the mindset of welfare economics, which is purely consequentialist. From that point of view, it is the end result, not the actor’s responsibility, that matters. For instance, it has been argued that our obligation to pay to prevent harm from our carbon emissions to other countries is no different than our obligation to pay to prevent them from being harmed by a meteor.111

This may be a defensible position, but it is at odds with strong moral intuitions. The idea that we have a special obligation to avoid harming others, as opposed to our possible duty to help them, is deeply embedded in our moral thinking. If we do have such an obligation, it is not apparent why the extent of our duty to avoid harming others should depend on their wealth. We would not think well of a company that justified its failure to protect the lives of its neighbors on the ground that they were poor. Or of a state or country whose tort law system reduced the required level of reasonable care when the potential victims are poor. “I was only endangering poor people” is not a defense to a reckless driving charge.

Under either of these accounts, the amount that particular people are willing to pay to reduce risks to themselves does not define how much of a cost a regulation should impose to reduce their risks. Note that these arguments are not really about “redistribution.” They are not premised on the idea that the world would be better with more equal incomes, although I think that is also true. Instead, the arguments relate to the duty of collective or individual actors to treat everyone in society with the same respect and concern regardless of wealth or other personal characteristics.112

The operative principle here could be called harm egalitarianism. It posits that equivalent harms should be treated the same by regulators, regardless of the identity of the victim. It also resists judging equivalency by the victim’s ability to pay to avoid similar risks. The use of standard figures for physical harms, unadjusted for income, is one way to operationalize this principle.113

The equal-harm principle is consistent with some uses of individual characteristics by regulators but not others. Some individual characteristics may be relevant to an assessment of which individuals are exposed to risks or are more vulnerable to harmful effects, both of which are important for an understanding of how individuals are affected by the risk. The equal-harm principle is consistent with consideration of these effects. What it does not allow, however, is weighing the loss of life or health of an individual differently depending on extraneous considerations such as the individual’s wealth, gender, or race.

C. The Status Quo Versus Equity Weighting

Equity weighting is a well-understood, though not often used, economic methodology for taking wealth differences into account in regulation. The first step is to convert regulatory benefits and costs into monetary terms, scrupulously basing this conversion on willingness to pay. The second step then translates these monetary sums into utility units.114 The same dollar of cost or benefit will translate into more utility units (“utiles”) in the case of a poor person than in the case of a rich person. The reason is that money has declining marginal utility: the value of an extra dollar to a rich person is much less than its value to a poor person, who has less to begin with. The final step is to combine all these individual measurements into a measure of social welfare, either by simply adding them up or by using what is called a social welfare function, which might further increase the weight given to the interests of the poor.115 The upshot is that the interests of the poor get heavier weight than the interests of the rich.

Equity weighting for regulatory benefits may not change regulatory outcomes much from the current system of uniform dollar values for risks to lives and health. To use equity weighting properly, we should abandon the use of fixed-dollar values on life. Instead, we should use valuations based on the willingness to pay of various income groups. That would reduce the dollar value of preventing illness or death among the poor while increasing the dollar value for the rich. When we convert these dollar amounts into utility units, however, each dollar of harm to the poor is translated into a larger amount of utility than is a dollar of harm to the rich. Thus, we would be shrinking the measure of harm for the poor in step one and then expanding it in step two. The two changes from current cost–benefit practice are partly offsetting and would be entirely offsetting under some assumptions.116

Equity weighting of costs could have a greater impact on outcomes than equity weighting of benefits. Regulatory costs falling on the poor would count for much more than they do currently. Daniel Hemel has suggested that some environmental regulations of the energy sector might fail under equity weighting, though some (including some of the most important) would survive, either because the benefits massively outweigh the costs117 or because the regulations so disproportionately benefit the poor.118 The reason equity weighting might disfavor regulation in this setting is that energy costs are regressive because the poor spend a greater proportion of their income directly on energy or on goods whose production requires energy use.119 Judging from Hemel’s sources, however, the empirical evidence on the distributive effects of air pollution regulation is limited and dated.120

Hemel’s argument highlights the biggest practical question about equity weighting—the difficulty of knowing who ultimately bears the costs of a regulation.121 The energy context he discussed is an apt example and a particularly important one because of its relevance to climate change. Although energy use may be inelastic in the short term, allowing producers to pass along energy prices to consumers, this may not be true in the long term. A recent European study found that the long-term elasticity of residential electricity use is about -0.5, meaning that a two percent increase in price will lead to a one percent decrease in electricity use.122 Industrial electricity use was even more elastic.123 The ability of producers to pass along price increases would be limited, since doing so would reduce demand. In addition, in deregulated markets, regulations that apply only to some generators (such as coal-fired power plants) cannot be readily passed on to consumers due to competition from other generation sources. Moreover, electricity prices are regulated in many states, creating the potential to transfer some of the burden to higher-income consumers.124

Furthermore, although using tax and spending powers to offset environmental harm to the poor seems fairly unlikely, more plausible options exist for offsetting the increased cost of electricity price increases. To begin with, transfer programs linked to the cost of living, like social security, may automatically adjust for widespread increases in energy prices. Moreover, some states have instituted carbon-trading systems, which can generate considerable revenue.125 California has spent half the revenue from selling carbon allowances on programs benefiting disadvantaged communities (over $5 billion to date).126 Biden’s pledge to devote forty percent of certain funds to benefit disadvantaged communities could also in effect compensate for higher energy costs due to tighter regulations.127

Electricity price increases can also be tempered or offset by state and federal programs that provide support for energy costs that fall on the poor. The primary federal program is the Low Income Home Energy Assistance Program (LIHEAP).128 Though underfunded, it provides assistance to about a fifth of poor families.129 A variety of state level programs protect the poor against electricity rate increases. California provides roughly a one-third discount on electricity rates to low-income customers, and eight other states also have rate discount programs.130 Some other states have more complex discount programs for the poor.131 The New York Public Service Commission began a thorough inspection of existing programs in 2015 that may serve as a model for other states.132 These programs do not offset all cost increases for the poor, but they do provide a cushion that varies across states.

The difficulties are even greater when regulations require infrastructure investments that may be passed along to consumers over long time periods, given the problems inherent in predicting changes in markets or in policies governing redistribution over time. The long-term incidence of compliance costs seems too uncertain to incorporate systematically into regulatory analysis, as equity weighting requires. Given the uncertainties, an effort to do so might have trouble surviving arbitrary-and-capricious review in court, at least in the “hard look” version of review.

The difficulty of determining the incidence of regulatory costs does not mean that the issue is unimportant. When that information is available, the government should be prepared to intervene if regulatory costs have a disparate financial impact, either by modifying the regulations (perhaps by targeting greater protection for the poor) or by considering possible compensatory mechanisms. As Richard Revesz has argued, if we are really concerned about the distributional impacts of regulations, such compensatory actions could be managed most effectively at the White House level.133

Another important issue relates to climate change. Incorporating distributional effects within the United States into regulatory policy may make little ethical sense if we then ignore the potentially much larger distributional implications globally. Because poor countries will suffer the worst harm from climate change, mitigation measures in the United States have important distributional effects. As a result, global equity weighting would dramatically increase estimates of the benefits of controlling climate change.134 Since the poorest countries make little contribution to greenhouse gas levels, applying equity weighting at the global level results in much more aggressive measures by developed countries to limit emissions.

In contrast, using domestic equity weighting as a justification to weaken climate policy would have perverse consequences. Increased cost burdens on the American poor from climate policy may be dwarfed by harms to the global poor. The poverty level in the United States in 2021 was $12,880 for a single individual,135 while per capita income in the Congo is $544.136 If we use utility weighting for domestic impacts but not for foreign impacts, we could reduce efforts to cut emissions, sacrificing the welfare of the desperately poor to assist the relatively less poor.

A further problem with the use of equity weighting by regulators relates to the drastic mismatch between this methodology and the degree of redistribution that the U.S. political system is prepared to explicitly embrace. The methodology carries implications for income redistribution that are far outside the range of views now considered politically plausible. For instance, utility-based policies would suggest what amounts to a universal basic income of around $12,000 per person, an idea that has no political traction in the United States.137 Such policies would also suggest much higher taxes on top incomes,138 limited only by the possibility that raising rates any further would decrease tax revenue by discouraging work.139 Most tellingly, as Liscow has observed, tax legislation reflects values that are far removed from equity weighting, implicitly valuing a dollar for someone in the tenth income percentile as 1.5 times more than for someone in the ninetieth, whereas the difference reflected in personal behavior is over eight times higher.140

Given that fiscal policies are an imperfect gauge of the public’s belief in redistribution, regulators need not slavishly mirror the progressivity of the fiscal system. But we should not completely write off the relevance of fiscal policy as an indicator of public values regarding redistribution.141 As Liscow has documented, the public believes in more redistribution than the fiscal system is designed to achieve. Still, the fiscal system does prove a significant signal of public attitudes, which perhaps should not be discounted entirely.142 In contrast, the principle of harm inequality discussed earlier seems to hew more closely to public values.

D. Whether (and How) to Consider Equity Effects

I have argued in favor of using constant valuations for the value of a statistical life and for illness, regardless of income. Compared with the economically “correct” approach of using lower values for the poor, this amounts to a degree of redistribution in their favor. Unlike some other writers, I believe that this approach is justified by more than simplicity, public acceptability, and the benefits of sticking with a well-established practice.143 It is also morally justified by the principle of devoting equal resources to prevent equal harms.144 Like other authors, I reject equity weighting as an alternative, partly because of feasibility issues. I also have doubts about the legitimacy of regulators adopting an approach like equity weighting that deviates so dramatically from the redistributive policies embodied in tax policy.

The principle of resource equity has implications for the role of equity that go beyond the environmental regulations that I have focused on. The core principle that I have advocated is one of equal protection from equal harms. Regulators should treat a death as a death, whether the victim is rich or poor, and a child’s asthma attack as a child’s asthma attack, whether the child’s parents are billionaires or homeless. This should be as much true in healthcare policy as in environmental law. This stance is not redistributive in the sense of being designed to remedy unfairness in the distribution of economic resources. Nevertheless, it does embody an important form of egalitarianism, one that is echoed in statutes.145

A remaining question is how economic analysis fits into this picture. To begin with, resource egalitarianism does not speak to whether some form of cost–benefit analysis should be used in regulatory decisions. To say that the government and individuals creating risks should be willing to spend the same amount to protect the lives of the poor and the rich does not tell us what that amount should be, nor does it tell us whether specific regulations impose costs beyond that amount. Cost–benefit analysis (with constant values for life and health) is one way of addressing those questions, whether or not it is the best way.146

The principle of equal protection against equal harms takes us only so far. There are inevitably going to be cases where considering the extent of harm and the costs of harm reduction does not lead to a clear answer. In those cases, it is appropriate to turn to other considerations, such as redistributive concerns, as a “soft factor” or a tiebreaker. Moreover, the argument here is that this principle should play a central role in regulation. That does not necessarily preclude the possibility that in some cases it is outweighed by other considerations. In short, in appropriate cases, income redistribution might still play a role as a “soft” factor in decision-making.147

The principle of equal protection against equal harms implies that we should be especially attentive to possible inequalities in the incidence and degree of harm. Part V of this article will show that current regulatory efforts do attend to the possibility that the poor may be more exposed to environmental harms than others. However, much more could be done to pinpoint the incidence of harm, and I will argue that the government has an obligation to do so.

Economic inequality is not the only dimension our society cares about. Before turning to possible solutions, we must first consider when and how regulators should consider racial inequality in their decision-making. That will set the stage for a discussion of how regulatory analysis of regulatory vulnerability could help address both economic and racial inequality.

In his presidential campaign, President Biden emphasized both the racial and economic dimensions of environmental justice. Stressing the need to eliminate “the systemic racism in our laws, policies, institutions, and hearts,” he said that energy and environmental policy must “recognize that communities of color and low-income communities have faced disproportionate harm from climate change and environmental contaminants for decades.”148 He added that the government “must also hold corporate polluters responsible for rampant pollution that creates the types of underlying conditions that are contributing to the disproportionate rates of illness, hospitalization, and death from COVID-19 among Black, Latino, and Native Americans.”149

Part IV begins by reviewing the environmental justice arguments that underlie Biden’s stance. It then considers two approaches to implementing this vision. The more limited approach rejects regulations that would have a disparate impact on communities of color and poor communities. The more sweeping approach seeks to use regulation as a tool to reduce existing health disparities between those communities and the population as a whole (including disparities unrelated to pollution). I share Biden’s belief in the importance of correcting the burdens that pollution imposes on poor communities and communities of color—and of more generally addressing the disturbingly worse health outcomes in those communities. My focus, however, will be on the extent of regulators’ legal authority to pursue these goals rather than on defending the goals themselves.150

Given the newly formed conservative supermajority on the Supreme Court, a looming question is how far the Court will go in enforcing its vision of colorblindness. Predicting the Court’s future trajectory is obviously perilous. However, in considering the likelihood that the Court will hold a practice unconstitutional, two factors do seem relevant. The first is the amount of fire the practice has drawn from conservative Justices in the past. The number of Justices who have condemned a practice, and the number of cases in which they’ve done so, both seem indicative of the likelihood that the practice will someday be declared unconstitutional. A prime example is the history of fervent anti-abortion dissents by conservative Justices, which culminated in the elimination of constitutional protection for abortion. Second, in terms of racial issues, the primary target of conservative ire has been affirmative action, which the Court views as providing racial preferences in employment and education. The more distinctions that exist between a practice and affirmative action, the lower the likelihood the practice will be invalidated.

It seems plausible that these factors are also related to how soon we might expect the Court to take up an issue. Thus, although we cannot say with assurance how the Court might eventually rule, it is possible to gain some sense of the immediacy and scale of the risk of an adverse ruling.

I will begin with some history of efforts to address the racial dimension of environmental problems. We will then turn to the most direct ways of doing so: prohibiting regulations that have a disparate racial impact and making communities of color a top priority as regulatory beneficiaries.

A. The Environmental Justice Movement

From the beginning, the prime movers behind environmental justice have been activists rather than academics.151 Even in the 1970s, the time when the strongest national consensus supported environmental protection, there were dissenting voices from some Black leaders. They declared environmentalism to be either irrelevant to the lives of people of color, an effort to perpetuate white lifestyles at their expense,152 or a diversion of public attention from their problems.153 The criticisms seemed especially aimed at efforts to preserve nature while ignoring many of the problems facing the urban poor.154 These criticisms received increasing attention after a major national conference in 1993.155

Much more recently, sociologist Robert Bullard made a similar charge: “Before the environmental justice movement burst onto the national scene, it was commonplace and a generally accepted norm by society, government, and industry that steering pollution to poor and people-of-color communities and away from affluent and white communities was no big deal.”156

The term environmental racism was apparently coined by the Rev. Dr. Benjamin Chavis while he was preparing to present a report to the national press on the disproportionate siting of toxic waste sites in Black communities.157 In trying to decide how to communicate the findings, he said, “It came to me—environmental racism. To me, that’s what it is.”158 Chavis’s charge of racism built on longstanding complaints about the disproportionate clustering of pollution sources and hazardous waste sites in communities of color.159 The added implication was that current environmental protection measures either do nothing to address the problem or make it worse.160 Another influential early study of the distribution of environmental burdens was found in a book by Robert Bullard,161 which made two key claims: that race is the predominant factor in the unequal distribution of environmental harm and that people of color had been frozen out of the relevant decision-making.162

There is obvious truth to Dr. Bullard’s more recent observation that “the Environmental Justice Movement is much stronger in 2021 because of new and invigorated rallying calls for racial justice with the rise of Black Lives Matter, after the police killings of George Floyd, Breonna Taylor, and countless other Black people, and the intergenerational protests during the Summer of 2020.”163 Even before then, lead contamination in the water supply of Flint, Michigan, had brought headlines to official disregard for the welfare of Black communities.164 As Bullard went on to say, and as we saw in Part II in terms of President Biden’s actions since taking office, “[e]nvironmental justice has been given a new sense of urgency and visibility at the highest level of government.”165

Environmental justice has also found purchase in judicial decisions reviewing individual projects. Friends of Buckingham v. State Air Pollution Control Board166 involved the approval of a gas compressor station that was planned as part of a new gas pipeline. The court reversed the approval because of a failure to consider the possibility of disproportionate health impacts on a predominantly Black community.167 The court emphasized that “environmental justice is not merely a box to be checked.”168Vecinos para el Bienestar de la Comunidad Costera v. FERC169 was another case involving a gas pipeline project. The D.C. Circuit reversed the Commission’s approval of the project, in part on environmental justice grounds.170 The agency had failed to justify limiting its environmental justice analysis to census tracts within two miles of the project.171

As Friends of Buckingham makes clear, delineating the boundaries of environmental justice communities is a crucial first step in any assessment. This has been controversial in the design of the government’s environmental justice screening tool.172 The initial design does not include race as a factor in designating environmental justice communities, instead relying on socioeconomic and other indicators. Although this choice was harshly criticized, the administration defended it due to litigation risks.173 This is a reasonable concern given the Supreme Court’s strongly conservative bent, but the legal risks may vary depending on the nature of the decisions made using the screening tool.174

The remainder of Part IV considers the possible legal problems arising from inclusion of race as a factor in regulatory decisions.175 The discussion is divided into two parts. Part B considers agency efforts to ensure that a proposed regulation does not harm communities of color. Part C considers a more contentious issue: can an agency tailor a regulation to reduce preexisting racial disparities?

It may be helpful to distinguish three categories of cases, depending on the remedy to be used in response to a racial disparity.176 The remedies that seems most easily defensible provide increased protection for all groups. Somewhat more problematic remedies may have little or no benefits for white communities but increase protection for communities of color. The most problematic remedies are those that increase protection for communities of color but decrease protection for white communities compared to available alternatives.

B. “First Do No Harm”: Prevention of Disparate Regulatory Impacts

Clinton’s executive order on environmental justice imposes a two-part duty on agencies. First, an agency is required to identify any “disproportionately high and adverse human health or environmental effects of its programs, policies, and activities on minority populations.”177 Second, “to the greatest extent practicable and permitted by law,” it must address “as appropriate” those effects.178 There is an obvious parallel between these duties and the disparate-impact component of civil rights laws.179

The important point to note is that the status quo is the baseline in determining whether a proposed regulatory action has a disparate impact.180 Environmental regulations generally decrease existing risks across a broad population, including people of color as well as whites.181 Thus, they improve on the status quo for all groups (though some regulatory actions may be exceptions).182 When such a negative impact does occur, Clinton’s executive order on environmental justice comes into play. However, the executive order is a prohibition of disproportionately harming certain groups, not a mandate to improve their standing relative to others.

As an illustration of the potential uses of this approach, consider a study of the first few years in which California’s carbon-trading system was in operation.183 The study found that there were increases in greenhouse gas emissions and emission of co-pollutants such as particulates in some areas. The study also found that these areas tended to have a higher percentage of residents of color and residents with less education.184 Assuming that this finding held up,185 an approach like the Clinton executive order would require an effort to mitigate this disparate effect, perhaps by imposing additional pollution limits on sources in those areas.186

Although environmental justice has received new prominence, so have criticisms of disparate-impact analysis. Judge James Ho put the conservative argument against disparate-impact theory in a nutshell in a Fifth Circuit opinion: “Prohibiting racial discrimination means we must be blind to race. Disparate impact theory requires the opposite[,] … advanc[ing] some people at the expense of others based on their race.”187 He added, “There’s a big difference between prohibiting racial discrimination and endorsing disparate impact theory. … It’s the difference between securing equality of opportunity regardless of race and guaranteeing equality of outcome based on race.”188 Thus, in Judge Ho’s view, “if we are to adopt disparate impact theory as a matter of national policy, it must be done by Congress.”189

Judge Ho’s comments point toward two separate arguments. The first is that consideration of disparate impact is barred by antidiscrimination law. In particular, he cites Title VI of the Civil Rights Act of 1964 and its prohibition of intentional racial discrimination,190 though a similar argument could be made under the Equal Protection Clause. The second argument is that disparate impact analysis must be explicitly authorized by Congress. Although he links that point with the first argument, he seems more generally worried by the prospect of “unelected agency officials usurping Congress’s authority when it comes to disparate impact theory.”191 I will discuss the antidiscrimination issue and statutory authority issues separately below.

1. Disparate Impact Analysis and Equal Protection

The Supreme Court has never directly addressed the constitutionality of disparate impact analysis. The closest it has come is Ricci v. DeStefano.192 The city of New Haven had used a standardized test combined with simulation exercises to determine which firefighters were eligible for promotion.193 When the entire eligible list turned out to be white, the city rejected the results on the basis of disparate impact.194

In the Court’s view, because the decision to reject the list was explicitly based on race, it violated Title VII unless the employer had an adequate justification.195 To prevail, the city would have had to show that it had “a strong basis in evidence that, had it not taken the action, it would have been liable under the disparate-impact statute.”196 The Court left unresolved whether compliance with the “strong basis in evidence” standard would satisfy the Equal Protection Clause, but at least it satisfied Title VII.197

Use of disparate-impact analysis by an agency designing a facially race-neutral regulation raises few of the concerns that worried the Court in Ricci. The agency has not created reliance interests in white members of the population which it has then dashed, as was the case in Ricci.198 Indeed, in the environmental setting, the agency is not making any decisions at all about specific individuals, an important point in light of the Supreme Court’s individualistic perspective on discrimination law.

Justice Scalia’s concurring opinion raised a more fundamental issue. He considered Title VII’s prohibition on disparate impact to be constitutionally suspect because it “place[s] a racial thumb on the scales, often requiring employees to evaluate the racial outcomes of their policies, and to make decisions based on (because of) those racial outcomes.”199 He conceded that a limited protection against disparate impacts might be justified as a prophylactic measure against intentional discrimination, but the statute went too far because it did not provide a defense for nonracially motivated actions.200

Despite the issues raised by Justice Scalia in Ricci, the Court reaffirmed the use of disparate-impact analysis in a 2015 fair housing case, Texas Department of Housing and Community Affairs v. Inclusive Communities Project, Inc.201 Justice Thomas’s dissent questioned the legitimacy of disparate-impact analysis. He argued that “the reason for this wholesale inversion of our law’s usual approach is the unstated—and unsubstantiated—assumption that, in the absence of discrimination, an institution’s racial makeup would mirror that of society.” Justice Thomas’s dissent was not couched in constitutional terms, however. Moreover, not even Justice Scalia joined the dissent.202

The question at this point is whether a majority of the Court would be willing to follow the path suggested by the late Justice Scalia and hold disparate-impact analysis unconstitutional.203 Although predictions are always hazardous, this does not seem a likely prospect at present.204 While conservatives may be unhappy with disparate-impact analysis, criticism by conservative Justices has been sporadic.

Though not a constitutional case, Ricci suggests that concerns about disparate-impact analysis are at their height where an action distinguishes between identifiable individuals. As Judge Ho has said, “opponents of disparate impact theory worry that it will only exacerbate, rather than alleviate, racial tension—by pressuring defendants to adopt policy changes for the explicit purpose of taking from some and giving to others based on their race.”205 Use of disparate-impact analysis in the regulatory context has the potential to increase protections for people of color without diminishing protections for whites and could even result in greater protection for the entire population. Moreover, the effects of regulatory choices are felt across broad, diffuse populations, muting any implication that the government values some individuals more than others.

The upshot is that a successful constitutional attack on the Clinton executive order seems unlikely to succeed anytime soon. Such an attack would in effect require the Court to classify disparate-impact analysis as a form of affirmative action. Only one current member of the Court (Justice Thomas) has endorsed a broad-based condemnation of disparate-impact analysis, and even he did not couch his attack in constitutional terms. If the Court were to hold disparate-impact analysis unlawful in situations like employment involving selecting individuals for benefits, it would be a further step to extend that condemnation to facially neutral regulations where one individual’s gain does not necessarily mean another’s loss.206

2. Statutory Authority to Consider Disparate Impacts

In questioning an agency’s power to adopt a disparate-impact test in the context of federal funding, Judge Ho argued that disparate impact involved issues too fraught to be decided by agencies without clear congressional authority. Though he did not use the term, the same concern is central to the major question doctrine. That doctrine requires a clear grant of agency authority when a question involves a matter of major social and economic importance.207

Unlike other cases in which the Supreme Court has applied this doctrine, however, the disparate-impact test is not a recent innovation and does not seem to feature heavily in political debate, whether in media like Fox News or in speeches by major politicians.208 Clinton’s executive order adopting this test has been in effect since 1994, nearly thirty years. There are no obvious examples of the test adding or reducing regulatory costs by a billion or even hundreds of millions of dollars. Nor does it require agencies to issue regulations that are unprecedented in some way or that would otherwise be outside their accustomed regulatory domains.

Putting aside the major-question doctrine, agency authority to consider disparate impacts in rulemaking depends on the statute under which the agency acts. At least in the environmental area, consideration of disparate impact as a factor seems generally within statutory authority. Most environmental statutes contain cost as a factor. In Michigan v. EPA, Justice Scalia explained that the term cost “includes more than the expense of complying with regulations; any disadvantage could be termed a cost,” including “harms that regulation might do to human health or the environment.”209 Increasing the level of harm to communities that already suffer disproportionately from environmental harms could reasonably be said to qualify as a “disadvantage.” Other regulations are keyed to public health, and there seems little reason why that term could not include consideration of the fact that an increased risk affects a population that is already suffering disproportionately from health hazards or poor health.210

C. Closing the Gap: Remedying Racial Disparities

The message of disparate-impact analysis is like that of the Hippocratic Oath: first, do no harm. But avoiding harm to communities of color might not be considered sufficient. Instead, improving the welfare of those communities may be the goal. This approach might make risk reductions in communities of color a priority. For instance, this approach might recommend a more costly or more stringent regulation to achieve deeper pollution reductions in communities of color.211

Along these lines, the White House Environmental Justice Advisory Council defines “just treatment” to include “elimination of systematic racism and other structural barriers” that lead to disparate health or environmental risks.212 For this reason the group called for creation of a mapping tool that would contain demographic data, including regarding race and ethnicity, as well as “data on redlining and other indicia of structural racism and other inequities.”213 The Council put forward a draft executive order building on this theme. It observed that health disparities have resulted from “inequitable and discriminatory treatment of communities of color … including the legacy of de jure segregation and other forms of discrimination.”214 It maintained that “[t]he Federal Government must be committed to taking decisive action … to dismantle the institutions and practices that inequitably place disproportionately human health, environmental, climate-related and other cumulative burdens on already disadvantaged communities.”215 On this basis, the draft order would require agencies to adopt strategies to “reduce, prevent, and eliminate pollution, legacy pollution, and cumulative impacts in environmental justice communities.”216

The question is whether an agency can adopt a regulation with the goal of reducing racial disparities—or, put another way, because it benefits communities of color more than white communities. Because of its “leveling up” effect, such a regulation might be considered a form of affirmative action subject to strict scrutiny.217 Applying that test would require that an agency show that the regulation was narrowly tailored to a compelling government interest. Moreover, the Court does not consider that eliminating the effects of societal racism, past or present, counts as a compelling interest for these purposes.

There are several differences between a regulation seeking to reduce racial disparities and the classic affirmative action problem posed by college admissions. First, regulators would be taking into account the racial or ethnic makeup of communities rather than individuals. It seems unlikely that the Court would give much weight to this distinction. Discrimination on the basis of community racial composition is involved in other contexts, such as redistricting, where the Court has applied strict scrutiny.218

A second distinction is that college admission is a zero-sum game given the limited number of slots, whereas increased regulatory protection against risks for one group does not necessarily mean decreased protection for another. For instance, the government might decide to strengthen requirements regarding lead paint removal partly because children of color are disproportionately exposed to lead paint.219 Doing so would not expose white children to greater health risks; in fact, the program would equally benefit numerous white children who might otherwise be exposed to lead paint.

A third distinction is that, unlike affirmative action in a selective admissions program, the policy adopted by the government is race neutral, even though the government’s purpose was partially based on race. Normally, a discriminatory purpose is enough to taint a race-neutral action and require strict scrutiny. It seems clear that strict scrutiny would apply to a government program that distinguished between individual beneficiaries on the basis of race. In the regulatory context, however, the more likely scenario is that one reason for adopting a regulation that is itself race-neutral could be to improve health outcomes specifically in communities of color.

The Supreme Court has not squarely decided whether race-neutral means adopted to redress the underrepresentation of Black or Hispanic Americans are subject to strict scrutiny. In applying strict scrutiny to affirmative action programs, it has taken into account the existence of race-neutral means of achieving the same end.220 A case working its way through the lower courts may shed light on this issue. A district court struck down a race-blind magnet-school admissions policy because its adopters wanted to increase Black and Hispanic representation.221 The Fourth Circuit stayed the ruling, and the Supreme Court declined to intervene.222 In the Supreme Court’s consideration of interim relief in the case, only Alito, Thomas, and Gorsuch voted to overturn the Fourth Circuit’s stay.223 This might suggest a possible majority in favor of upholding the program on the merits, but we have no way of knowing the thinking of the Justices in the majority.

Are regulations designed to reduce racial disparities in exposure to environmental and health risks subject to strict scrutiny? At present, the best one can say is that there is a plausible constitutional argument for applying strict scrutiny but also some plausible arguments the other way. Moreover, use of race-neutral programs to pursue diversity or address racial disparities is undoubtedly widespread, so illegalizing them would have daunting repercussions on current practices. On balance, the better bet may be that the Court would find such programs constitutional.

Another open question is what causation standard the Court would apply to a racially neutral law designed to reduce racial disparities. Would it be enough to show that the desire to reduce racial disparities is one factor behind adoption of the regulation,224 or must this motive be a “but for” cause of the agency’s action225 or the dominant motive?226 None of this is crystal clear.

Taking all this together, regulations designed at least in part to reduce racial disparities in health outcomes seem to face a limited risk of invalidation on constitutional grounds. If the Court does decide to subject race-neutral programs of all kinds to strict scrutiny whenever the government considered the program’s ability to reduce racial disparities, the effect on health or safety regulations would be the least of the fallout.

While the constitutional risks may be manageable, there may remain statutory issues. Those issues depend, of course, on the specific statute. Environmental statutes contain a variety of standards. Some are keyed to public health. For instance, section 109 of the Clean Air Act requires the EPA to set air-quality standards that protect public health with an adequate margin of safety.227 In making this determination, could the EPA use the desirability of reducing racial health disparities as an argument for selecting a more rigorous standard?

Given this standard, the EPA should be able to make regulations that reduce pollution in communities that have especially poor health higher priorities, which might disproportionately benefit communities of color. But it seems doubtful that the statute authorizes the EPA to give more weight to risk reductions in communities of color than to similar reductions in white communities simply on the theory that the risk levels or health disparities in communities of color are the result of structural racism or historic discrimination.

Pollution standards based on the best available pollution control technology are also common in environmental law.228 The question is whether, in deciding what systems are best for particular industries, the EPA could give more weight to risk reductions in communities of color than in white communities because of structural racism. The word best is elastic, but this may be too much of a stretch when it comes to a statute that does not include any reference to race. And in particular, it may be too much of a stretch for the current conservative majority on the Supreme Court.

Overall, the general goals of combatting structural racism or the effects of intentional discrimination seem like poor fits with health and safety regulations.229 This does not mean that the circumstances of people of color, individually or on a community basis, are irrelevant to environmental regulation. Part IV explains how regulators can properly take into account the effects of structural racism and past discrimination in an indirect way, and how in fact risk-assessment methodology sometimes already does so.

Exposure to many environmental harms varies with location. So do race and income. Consequently, geography presents a potential mediating framework in which the life circumstances of poor communities and people of color can enter into regulatory decisions.

It is no surprise that the poor are often located in areas with high pollution levels. But they also have greater vulnerability to the resulting health risks. As Hemel points out, the empirical evidence shows that any given pollution level causes more illness and death among the poor.230 Thus, the disadvantaged are more vulnerable to ill effects even controlling for exposure, perhaps because of poor medical care for conditions caused by pollution, because existing health conditions are exacerbated by the pollution, or because the poor are more likely to be exposed to multiple pollutants that have synergistic effects.231

It seems undeniably appropriate for regulators to consider not only exposure levels but differences in vulnerability.232 Such a focus on vulnerability is commonplace in related settings such as protection against natural disasters, where certain groups (racial and ethnic minorities, women, the elderly, children, and the disabled) disproportionately suffer from the disaster impacts.233 In my view, a heightened focus on differences in exposure and vulnerability offers the most promising path forward for environmental justice to expand protection for the goals of environmental justice. It is also consistent with the norm that I defended earlier: devoting equal protection to the prevention of equivalent harms.

A. Vulnerability as a Factor in Risk Assessment

Current methods of risk evaluation sometimes incorporate the idea of vulnerability, but in a subtle way that has not previously received attention from scholars. One methodology used by the EPA, after estimating the extent to which a pollutant raises the risk of mortality or illness, then uses this estimate as a multiplier of the existing risk level in a particular locale. In other words, it treats pollution as a risk multiplier rather than simply an add-on to existing health risks. The greater the existing health risk in a locale, the greater the impact of the multiplier. So, if discrimination and structural racism are harming health in a particular locale, the formula automatically takes that effect into account in determining the resulting harm.234

The diagram in Figure 1 unpacks this concept:

Figure 1.

Modeling Vulnerability and Exposure Effects

Figure 1.

Modeling Vulnerability and Exposure Effects

Close modal

Such a model is used in the regulatory impact analysis accompanying a proposed rule dealing with pollution from heavy trucks. The model used by the EPA determined the total number of deaths caused by exposure to a pollutant (PM2.5) by (a) taking the background mortality for each county by age group, (b) estimating the increased mortality rate taking into account the county’s pollution exposure; and (c) summing the results across the country.235

Communities of color are likely to have higher background mortality rates due to greater health problems and poorer medical care. The EPA’s model amplifies preexisting differences in mortality rates because the calculation treats pollution as a risk multiplier. This makes sense because the same level of pollution is likely to cause more harm in places where people’s health is already bad and they receive worse medical care. Third, to the extent that those communities also tend to have higher levels of exposure to pollution, the formula also takes that into account, because exposure is another multiplier. In effect, the formula operationalizes the insight that disadvantaged communities suffer greater harm from pollution because their populations are more vulnerable and have higher exposure levels.

What is important about this formula is not the exact method by which background mortality rates and current exposure levels factor into the predicted individual risk of harm. Rather, it is the assumption that predicted harm is a function of these two factors, which essentially means that differences in background mortality rates fully capture differences in individual vulnerability of harm from the same levels of exposure. If that assumption is at least approximately valid, it means that the impacts of demographic characteristics on health or mortality do not need to be considered at the stage of determining individual risk. Instead, to the extent they need to be considered at all, it is only for modeling purposes where data gaps make it impossible to measure background mortality rates or exposures in a direct way or with sufficient detail. Once those gaps are filled in, however, those other individual characteristics become irrelevant to determining the benefits of regulations. In short, characteristics like race are significant as modeling inputs (and even that only in limited circumstances). However, these characteristics carry no weight in the normative analysis after the regulator has estimated the degree of health or mortality reduction that a regulation can achieve.

The EPA’s current modeling technique suggests a promising method of implementing environmental justice. However, the EPA’s ability to implement this approach falls short in two ways. The first issue involves geography.236 In the absence of individual-level data on exposure and health, estimates for specific locations must be derived from larger geographic areas. Counties are too big to map well onto differences in communities. Los Angeles County in California or Cook County in Illinois (the Chicago area) contain an array of communities with different levels of affluence and different racial compositions. Averaging across a whole county wipes out these crucial but more granular variations, and the same may be true for census tracts.237

Indeed, smaller units like zip codes might not be small enough. Studies indicate that people within a quarter mile or so of a major roadway have the highest rate of health effects from emissions, including greater incidence of respiratory ailments like asthma and higher cardiovascular death rates.238 Particulate emissions from diesel are a serious health risk that is concentrated on areas within 650 feet of a road.239 Almost one in twenty Americans lives closer than that to a major road; in cities, the number is much higher; and more than half that group are nonwhite.240 It is also plausible that those living closer to major roads are likely to be poorer than those in nearby quieter streets. Zip codes may lump these people together with people further from roadways.

Current monitoring practices are not sufficiently fine-grained to pick up on these micro-effects, particularly since states and cities may place monitors strategically to avoid being found in violation of air quality standards.241 Monitoring can be improved but perhaps not sufficiently to identify small-scale differences in exposure on a national basis. Thus, modeling of exposures and vulnerability may be needed to downscale risk analysis.

Beyond the need to downsize geographic units below the county level, a further problem is that we need a better grasp on exactly how background disadvantages combine to produce higher vulnerability. This matters because meaningful mortality and health data may not be available at the micro level and may need to be extrapolated from data about larger populations combined with local demographic data. Race and income seem to be factors. Other factors may correlate with them but also have an independent effect, such as level of education, socioeconomic status, or availability of medical care.242

Differences in exposure and vulnerability are especially important when statutes make protecting those at greatest risk a top priority. The EPA has deployed this argument in a proposed rule dealing with mercury emissions by coal-fired power plants.243 The statutory setting and history of this regulation are complex. In an earlier phase of the regulatory process, the Supreme Court held that the EPA had to make at least a qualitative comparison of the costs and benefits of reducing mercury emissions from burning coal.244 In assessing regulatory benefits, the EPA stressed that it had focused both on the overall risk reduction and the “distribution of those benefits.”245

The EPA also discerned a congressional intent to “protect even the most exposed member of the population” from the risk of exposure to hazardous air pollutants (HAPs).246 Congress had ordered a study to “determine the level of mercury that can be consumed by even sensitive populations without adverse effect on health.”247 A key route for mercury exposure is eating fish, which Congress knew posed a special risk to children and fetal development.248 Thus, exposure varies with fish consumption, and vulnerability is greatest for the very young.

The EPA observed that subsistence fishers and their families are at high risk, and that subsistence fishing activity “can be related to a number of factors including socio-economic status (poverty) and/or cultural practices, with ethnic minorities and tribal populations often displaying increased levels of self-caught fish consumption.”249 In particular, “[l]ow-income Black and white populations in the Southeast and tribal fishers near the Great Lakes had the potential for increased risk in 25 percent of the watersheds modeled.”250 In the EPA’s view, consideration of the exposure levels of these groups and of the vulnerability of children were enough to show that regulation of mercury emissions is “necessary and appropriate,” as the statute required. This focus on the needs of the most impacted part of the population was particularly important because only a small portion of the benefits of the regulation could be quantified.

Although I have focused the discussion of vulnerability in the setting of environmental regulations, it is an important factor in other settings. For instance, the stark wealth gap between Black and white populations deprives Black individuals of an important buffer against negative economic shocks. The absence of such a buffer may make Black people more vulnerable to a wide range of disruptive life events, from credit crunches and prolonged illness to job loss and natural disasters.251 Thus, taking into account wealth inequality would be relevant in many regulatory settings. Since wealth represents the accumulated product of past income and spending,252 considering wealth as a factor independent of current income would in effect respond to historical racial injustice. This use of wealth is consistent with the equal-harm principle discussed earlier in the paper. For instance, a person with wealth may be able to withstand an economic shock without having to sacrifice important aspects of well-being such as adequate housing. That may not be true of someone who had an equal income before the shock but little accumulated wealth. The measure of harm here is the loss of adequate housing, and wealth is merely a factor in determining vulnerability to the loss. But two people who both suffer a loss of adequate housing should be equally entitled to societal resources, even if one began with higher wealth than the other.

B. Use of Race in Modeling Vulnerability

The use of geographic mortality and exposure data is unproblematic, even if race is one factor driving mortality and exposure levels behind the scenes. A different problem arises, however, if race is explicitly used as part of a formula for modeling the current distribution of risk or how a regulation might reduce risk. This is an important issue in practical terms. Richard Revesz and Samantha Yi surveyed studies that disaggregated the distribution of risk253 and found that thirty-one of the thirty-seven studies used race as a factor.254 Moreover, studies of pollution risk have found that race has explanatory power independent of income,255 so including it in risk modeling would improve the accuracy of those models. Indeed, the imprint of historic redlining is still observable in the current distribution of pollution risks.256

The general use of race in various forms of statistical modeling is vigorously debated. Some scholars argue that the use of race as a variable should be considered a form of discrimination, or at least should be prohibited when the effect is to disadvantage people of color.257 A forthcoming article by Sonja Starr makes a powerful case for this view and analyzes several important examples of such statistical modeling.258 In a nutshell, Starr’s argument is that “explicit use of discriminatory factors in formal, quantitative algorithms or tests” violates the Supreme Court’s repeated holdings that “otherwise illegal discrimination cannot be justified based on statistical generalizations about groups, even if those generalizations are empirically supported.”259

Despite Starr’s forceful arguments, there are several reasons for doubting that use of race as a factor in modeling poses serious legal risks in the regulatory context. The first is that, as Starr laments, courts are very far from accepting her broad condemnation of statistical modeling.260 For instance, she argues that using race in life-expectancy determinations should never be allowed in assessing tort damages. Yet, she reports that courts have been almost universally untroubled by this practice.261

Second, in the situations Starr discusses, the ultimate use of the statistical models is to make individual-level decisions. The core argument is that it is wrong to disadvantage individuals based on statistical generalization about race or sex. In the context of regulation, however, statistical generalizations are used to create race-neutral rules that provide uniform protection to large numbers of people rather than to make individualized decisions.

Third, as Starr recognizes, it is not clear that courts would reject the use of statistical models to devise facially neutral remedies for racial disadvantage.262 Although the Court has increasingly embraced color blindness, its opinions thus far have seemed at least tentatively accepting of race-neutral measures that are motivated by a desire to remedy racial inequality.263

In short, my conclusion is that agencies ought to focus on the extent of the risk facing different people, taking into account both differences in exposures to risk and different vulnerabilities when a risk materializes. To the extent race or income drives either factor, I argue that it should be unproblematic to use them statistically in estimating individual vulnerability and exposure levels.264

In the long run, risk modeling may be able to move away from broad indicators of vulnerability based on community demographics. With full information about each individual, we would know directly a great deal about their vulnerability, including whatever imprint racial disparities, including comorbidities, health insurance status, access to health care, quality of the relevant healthcare providers, and exposures to other pollutants, have. At that point, consideration of race might be superfluous, since we can directly gauge all the health-relevant factors that race might have influenced. In a world of burgeoning data mining, Big Data analysis, and artificial intelligence—a world in which advertising can already be tailored to individuals based on minute knowledge of their behavior—we may not be as far from completely disaggregated models of risk and vulnerability as one might think. Combined with micro-level exposure data, we could best craft regulations to target those most at risk of harm—very often, the poor and people of color.

In this article, I have in effect adopted the perspective of a regulator who views inequality as a critical issue but is unsure how best to respond, particularly taking into account the prospect of judicial review. The article takes as a given that inequality is a serious problem in our society and that it is the business of the government to address it. Consequently, I do not engage libertarian arguments against redistribution. Nor do I address arguments that the government should be unconcerned about racial inequality apart from banning intentional discrimination.

This article has considered three dimensions of inequality: income, race, and risk levels. With regard to economic inequality, I reject the theoretical argument that redistribution should be solely the domain of fiscal policy (taxing and spending). Implementing income redistribution through regulation turns out to be difficult, however, because we so often do not know the ultimate economic incidence of regulations. Current regulatory analysis in the United States does pursue a form of equality by assigning the same monetary value to human life and health regardless of an individual’s wealth. I argue that this practice is justified by the principle of equal resources to prevent equal harms. Or to put it another way, wealth blindness in assessing harms instantiates the principle of equal respect for all individuals. In this view, society as a whole (and those creating risks in particular) have a duty to devote equal resources to avoiding equal harms, regardless of the victim’s wealth. This form of equality is distinct from any affirmative social duty to reduce inequality.

With regard to racial equality, my conclusions have been more nuanced. Using disparate-impact analysis to determine whether a proposed regulation unduly harms communities of color seems to pose relatively little legal risk. It seems questionable, however, whether an agency could legally justify prioritizing the goal of reducing relative health disparities over reducing the absolute level of risk. For example, as a legal matter, agencies might find it difficult to justify sacrificing an overall reduction in risk to focus risk reduction on communities.265

Finally, I highlight the great potential of improved risk assessment to address environmental justice. Much more could be done with better data and modeling. In particular, I argue that statistical models that incorporate race as a factor in estimating risk may be useful and should not be subject to legal challenge, at least absent a radical shift in current legal understandings.

The prescriptions offered in this article are not revolutionary. They are incremental improvements and therefore subject to the criticism that they do not address the need for transformative change. They would, however, do a great deal to improve the regulatory response to inequality. And they would do so in a way that fits naturally with current practice and existing legal doctrine. Those characteristics are especially important for assuring the durability of any policy in a time of deep partisan polarization and great political volatility.266


There is by now a voluminous literature on cost–benefit analysis and its possible virtues and vices. See, e.g., Richard L. Revesz & Michael A. Livermore: How Cost-Benefit Analysis Can Better Protect the Environment and Our Health (2008); Matthew D. Adler & Eric A. Posner, Cost-Benefit Analysis: Economic, Philosophical, and Legal Perspectives (2001); Zachary Liscow, Redistribution for Realists, 107 Iowa L. Rev. 495 (2022); Cass R. Sunstein, The Office of Information and Regulatory Affairs: Myths and Realities, 126 Harv. L. Rev. 1838 (2013); Cass R. Sunstein, Cost–Benefit Default Principles, 99 Mich. L. Rev. 1651 (2001); Douglas A. Kysar, It Might Have Been: Risk, Precaution, and Opportunity Costs, 2. J. Land Use 1 (2006).


See, e.g., Cass R. Sunstein, The Cost-Benefit Revolution 40 (2018). The centrality of this concept to cost–benefit analysis is apparent from the book’s index, where the listing for “willingness to pay” has fifty entries. Id. at 265–66.


As it turns out, however, at least in the context of federal regulation, willingness to pay does not control valuation of life.


Liscow, supra note 1, at 515; Daniel Hemel, Regulation and Redistribution with Lives in the Balance, 89 U. Chi. L. Rev. 640 (2002).


Liscow, supra note 1, at 497–98.


See Adam M. Samaha, Death and Paperwork Reduction, 65 Duke L. J. 279, 281 (2015).


Liscow, supra note 1, at 497–99.


This article focuses on regulatory substance, but there are also promising ways to ensure that communities of color have a greater voice in the regulatory process. See Brian D. Feinstein, Identity-Conscious Administrative Law: Lessons from Financial Regulators, 90 Geo. Wash. L. Rev. 1 (2022).


Off. of Minority Health, U.S. Dept. of Health & Hum. Servs.,


Katherine Schaeffer, 6 Facts About Economic Inequality in the U.S. (Feb. 7, 2020),


Kriston McIntosh et al., Examining the Black–White Wealth Gap (Feb. 27, 2020), In another telling statistic, “[b]y ages 30 and 35, respectively, the White working class has 4 and 7 times the wealth of Black working-class adults and double and triple the wealth of Black adults holding managerial positions.” Fenaba R. Addo, At the Intersection of Race, Occupational Status, and Middle-Class Attainment in Young Adulthood, AEA Papers & Proceedings 48, 51 (May 2022).


Schaeffer, supra note 10.


Ana Hernández Kent & Lowell R. Ricketts, Has Wealth Inequality in America Changed over Time? Here Are Key Statistics, Fed. Rsrv. Bank of St. Louis (Dec. 2, 2020),,total%20household%20wealth%20in%202019.


Mark Muro & Jacob Whiton, Geographic Gaps Are Widening While U.S. Economic Growth Increases, Brookings (Jan. 23, 2018), Geographic inequality is also reflected by differences in life expectancy, which vary sharply between locations (particularly for the less affluent). Tanyana Deryugina & David Molitor, The Causal Effects of Place on Health and Longevity, 35 J. Econ. Persp. 247 (Fall 2021).


Ganesh Sitaraman et al., Regulation and the Geography of Inequality, 70 Duke L.J. 1763, 1765 (2021). For discussion of the normative dimension of geographic inequality, see Ann M. Eisenberg, Distributive Justice and Rural America, 61 B.C. L. Rev. 189 (2020).


Abdulrahman Jbaily et al., Air Pollution Exposure Disparities Across US Population and Income Groups, 601 Nature 228, 228 (2022).


Id. PM2.5 consists of ultrafine particulates (under 2.5μm) commonly produced by oil or coal combustion. This distance is about one-seventieth of the width of a human hair. Among the most widespread pollutants, it is also the one with the most greatest impact. Particulate Matter (PM) Basics, U.S. Envt Prot. Agency


Jbaily et al. report that in 2016, the average PM2.5 concentration for Blacks was 13.7% higher than for whites, with a “steep incline” for zip codes whose population was more than 85% Black. Jbaily et al., supra note 16, at 229.


Id. at 230–31.


Id. at 231. The authors concluded that “targeted air pollution reduction … may be needed to cause a decrease in relative disparities.” Id. Another recent study focused on sources of exposure but reached similar conclusions. Christopher W. Tessum et al., PM2.5 Polluters Disproportionately and Systemically Affect People of Color in the United States, 7 Sci. Advances 3 (2021),


Haley M. Lane et al., Historical Redlining Is Associated with Present-Day Air Pollution Disparities in U.S. Cities, 9 Envt Sci. & Tech. Letters 345 (2022). The authors’ concluding observation is that “[p]resent-day disparities in U.S. urban pollution levels reflect a legacy of structural racism in federal policy-making—and resulting investment flows and land use decisions—apparent in maps drawn more than 80 years ago.” Id.


Notable contributions to that literature include Hemel, supra note 4; Richard L. Revesz & Samantha Yi, Distributional Consequences and Regulatory Analysis, 52 Envt L. 53 (2022); Liscow, supra note 1; Arden Rowell, Ethical and Trans-Economic Choices Inside Cost-Benefit Analysis, 52 Envt L. 53 (2022); Robert W. Hahn, Equity in Cost Benefit Analyses, 372 Science 439 (2021); Richard L. Revesz, Regulations and Distribution, 93 N.Y.U. L. Rev. 1489 (2018); Lisa A. Robinson et al., Attention to Distribution in U.S. Regulatory Analysis, 10 Rev. Econ. & Poly 308 (2016); David A. Weisbach, Distributionally Weighted Cost-Benefit Analysis: Welfare Economics Meets Organizational Design, 7 J. Legal Analysis 151 (2014); Matthew D. Adler, Risk Equity: A New Proposal, 32 Harv. Envt L. Rev. 1 (2008).


See Hemel, supra note 4, at 658 (stating that uniform value of lives saved cannot be justified on efficiency grounds). Cass Sunstein has explained the reasons for these differences and their implications as follows:

[W]ith full individuation, overall WTP [willingness to pay] would be lower for poor people than for wealthy people, for African Americans than for whites, and (possibly) for men than for women. But … government would not discriminate against groups; for example, it would neither decide on high VSLs [values of a statistical life] for programs predominantly benefiting whites nor decide on low VSLs for programs predominantly benefiting African Americans. The difference would be a product of aggregation of fully individual VSLs—aggregation of the kind that most conventional markets, including those for automobiles and consumer goods, now provide.

Cass R. Sunstein, Valuing Life: A Plea for Disaggregation, 54 Duke L. J. 385, 417 (2004). Sunstein maintained, however, that the economic logic of considering race when determining valuations is trumped by the antidiscrimination principles adopted through the deliberative democratic process. Id. As he said, “Our constitutional system is a deliberative democracy, not a maximization machine, and many social judgments should be made by citizens engaged in deliberative discussion with one another rather than by aggregating the individual choices of consumers.” Id. at 429–30. I take the same argument one step further by arguing that regulators should treat the loss of life identically regardless of income.


Much of the scholarly literature concerns whether regulators should attempt to redistribute income. There is an argument to be made for doing so, but in practice it may be difficult to make much headway because the long-term incidence of regulatory costs is difficult to ascertain. Using uniform values for intangible values like human life or human dignity, however, is more easily implemented and invokes ethical standards outside welfare economics, even though it may also have distributional consequences. See Part III(B) for further discussion.


See Hahn, supra note 22, at 439.


For a comprehensive introduction of environmental justice, see Clifford Villa et al., Environmental Justice: Law, Policy and Regulation (3d ed. 2020). Chapter 3 is particularly relevant to this article. For an introduction in the special context of climate change, see Elizabeth Cripps, What Climate Justice Means and Why We Should Care (2022).


Exec. Order No. 12,291, 3 C.F.R. 127 (1981).


For a description of the development of OIRA’s role in regulatory oversight, see Daniel H. Cole, “Best Practice” Standards for Regulatory Benefit-Cost Analysis, 23 Rsch. in Law & Econ. 1 (2007). Finally, for an insider’s view of OIRA’s operation in the Obama administration, see Cass R. Sunstein, The Office of Information and Regulatory Affairs: Myths and Realities, 126 Harv. L. Rev. 1838 (2013).


Exec. Order No. 12,866, 3 C.F.R. 638, 638–39 (1994).


See Ellen Siegler, Executive Order 12866: An Analysis of the New Executive Order on Regulatory Planning and Review, 24 Envt L. Rep. 10070 (1994).


Exec. Order No. 13,422, 72 Fed. Reg. 2763 (2007).


Exec. Order No. 13,422 § 9, 72 Fed. Reg. 2763, 2764 (2007).


OIRA’s instruction on the preparation of regulatory analysis by agencies provides as follows:

You should exercise professional judgment in identifying the importance of non-quantified factors and assess as best you can how they might change the ranking of alternatives based on estimated net benefits. … This discussion should also include a clear explanation that support designating these non-quantified factors as important. In this case, you should also consider conducting a threshold analysis to help decision makers and other users of the analysis to understand the potential significance of these factors to the overall analysis.

Circular A-4, Off. of Mgmt. & Budget, Bush White House (Sept. 17, 2003).


Exec. Order No. 13,497, 74 Fed. Reg. 6113 (2009).


Exec. Order No. 13,563, 3 C.F.R. 215, 215 (2012).


This led to sharp complaints from progressives. See OIRA’s Role in Regulation: Helping or Hurting, Ctr. for Progressive Reform, According to this Center, “OIRA … has continued to serve as a conduit for industry attacks on environmental, health and safety regulations, often substituting the scientific judgment of its inexpert staff for that of the agencies themselves, and it has persisted in imposing a deeply flawed cost–benefit analysis, all while slowing down an already glacial regulatory process.” Id.


Exec. Order No. 13,771, 83 Fed. Reg. 9339 (Jan. 30, 2017).


Robinson et al., supra note 22, at 314.


Dominic J. Mancini, Acting Administrator Office of Information and Regulatory Affairs, Guidance Implementing Executive Order 13771, Titled “Reducing Regulation and Controlling Regulatory Costs,” at 9 (April 5, 2017), available at Id. at 9.


As Cass Sunstein said, “the right approach is not ‘one in, two out’ but a careful check on issuing new rules, with the help of cost–benefit analysis—accompanied by an insistence on issuing those rules if the benefits justify the costs and an ambitious program to scrutinize rules on the books to see if they should be scrapped.” Cass R. Sunstein, The Cost–Benefit Revolution 211 (2018). Sunstein called this a “decisive objection” to both the regulatory budget and the two-for-one requirement. Id. at 20–21.


Michael A. Livermore & Richard L. Revesz, Reviving Rationality: Saving Cost-Benefit Analysis for the Sake of the Environment and Our Health (2020).


Exec. Order No. 13,992, 86 Fed. Reg. 7049 (2021). Section 1 of the executive order explains its purpose. It also provides a succinct statement of Biden’s regulatory philosophy, leaving no doubt as to the contrast with Trump: it calls “racial justice” one of the “urgent challenges facing the Nation.” Id.


Memorandum, 86 Fed. Reg. 7223 (2021). In the arcane language of Washington insiders, labeling this document a memorandum apparently gave it somewhat less prestige, although perhaps no lesser binding effect on agencies. See Gregory Korte, Presidential Memoranda vs. Executive Orders. What’s the Difference?, USA Today (Jan. 1, 2017), It is as if the term executive order added “and I mean it!” at the end of the document. Like the seating plan at presidential dinners, the difference apparently matters to insiders.


Memorandum, supra note 43.


Exec. Order No. 13,990, 86 Fed. Reg. 7037 (2021).


Id. at 7040.


For an introduction to opposing views of cost–benefit analysis, see Adler & Posner, supra note 1.


For citations to some of the key critical works, see Douglas A. Kysar, It Might Have Been: Risk, Precaution, and Opportunity Costs, 2. J. Land Use 1, 6 n.23 (2006).


They make it clear, however, that their quarrel is with the economic methodology of cost–benefit analysis, not with taking costs and benefits into account. Frank Ackerman & Lisa Heinzerling, Priceless: On Knowing the Price of Everything and the Value of Nothing 211 (2004).


Id. at 8–9; see also id. at 12 (“Cloaked in the language of scientific objectivity, economic arguments have repeatedly played a partisan role.”).


For references to the writings of some of the leading academic supporters of cost–benefit analysis, see Kysar, supra note 48, at 5.


Id. at 15.


Revesz & Livermore, supra note 1, at 12 (2008). Revesz was recently selected by President Biden to head the OIRA.


Id. at 13. They also view cost–benefit analysis as a way of disciplining the wide discretion given to administrative agencies, thereby ensuring that decisions are made on the basis of reasoned analysis and uniform criteria. Id.


See Amy Sinden, “A Cost-Benefit State”? Reports of Its Birth Have Been Greatly Exaggerated, 46 Envt L. Rep. 10933 (Oct. 2016).


See Caroline Cecot, Congress and Cost-Benefit Analysis, 73 Admin. L. Rev. 787, 793–94 (2021).


Heinzerling recently wrote, “Willfully blind to the maldistribution of society’s resources and the historical injustices that underwrite this maldistribution, cost–benefit analysis whistles past two of the central moral concerns with today’s regulatory system.” Lisa Heinzerling, Climate Change, Racial Justice, and Cost-Benefit Analysis, LPE Project (Sept. 28, 2021),


Consider James Goodwin, Cost-Benefit Analysis Is Racist, Ctr. for Progressive Reform (Oct. 9, 2020), The core of Goodwin’s critique is that cost–benefit analysis is “a powerful weapon for attacking regulatory safeguards and undercutting landmark laws” and “the results that cost–benefit analysis produces are typically unmoored from reality and completely devoid of the credibility and legitimacy its supporters have sought to cultivate for it.” Id.


Exec. Order No. 12,866, 3 C.F.R. 638, 638–39 (1994).


Circular A-4, Off. of Mgmt. & Budget, Bush White House (Sept. 17, 2003). As this article goes to press, the Biden administration has released a proposed major revision of Circular A-4. The proposed revisions place more emphasis on distributional issues and are largely consistent with the positions taken this article.


Circular A-4 explains that “the effects of various regulatory alternatives should be described quantitatively to the extent possible, including the magnitude, likelihood, and severity of impacts on particular groups. You should be alert for situations in which regulatory alternatives result in significant changes in treatment or outcomes for different groups.” Id.


Exec. Order No. 13,563 § 1(c), 76 Fed. Reg. 3821 (2011).


Robinson et al., supra note 22, at 313.


Id. at 314.


Jerry Ellig, Why and How Independent Agencies Should Conduct Regulatory Impact Analysis, 38 Cornell J.L. & Pub. Poly 1, 27 (2018).


Revesz, supra note 22, at 1540.


Memorandum §2(b)(1), 86 Fed. Reg. 7223 (2021).


Exec. Order No. 12898, 50 Fed. Reg. 7629 (1994).


Interestingly, this requirement is limited to impacts in the “United States and its territories and possessions, the District of Columbia, the Commonwealth of Puerto Rico, and the Commonwealth of the Mariana Islands.” Thus, impacts on the global poor are not covered by the order.


Section 6-103 adds that the order does not supersede an earlier order encouraging regulatory flexibility so that states, local governments, and tribes have “more flexibility to design solutions to the problems faced by citizens in this country without excessive micromanagement and unnecessary regulation from the Federal Government.” Exec. Order No. 12875, 58 Fed. Reg. 58093 (1993).


See, e.g., 84 Fed. Reg. 32520, 32574 (2019).


U.S. Envt Prot. Agency, Plan EJ 2014 (Sept. 2011),


U.S. Envt Prot. Agency, Guidance on Considering Environmental Justice During the Development of Regulatory Actions (May 2015),


Id. at 4. The definition of “low income” groups takes into account a number of factors other than current income, allowing agencies to consider that “educational attainment, baseline health status and health insurance coverage may also be useful for identifying, characterizing and developing strategies for assessing and engaging low-income populations in the context of specific regulatory actions.” Id. at 6.




Id. at 15.


Council on Environmental Quality, Environmental Justice: Guidance Under the National Environmental Policy Act (1997),


Memorandum for the Heads of All Departments and Agencies re Executive Order on Federal Actions to Address Environmental Justice in Minority Populations and Low-Income Populations (Feb. 11, 1994),


Revesz, supra note 22, at 1539–40.


Id. at 1540.




Id. at 1541.


Exec. Order 13,990, 86 Fed. Reg. 7037 (2021).


Fact Sheet: A Year Advancing Environmental Justice, White House (Jan. 26, 2022),




Exec. Order No. 14,008 § 219, 86 Fed. Reg. 7619, 7629 (2021).


See Liscow, supra note at 1, at 515.


There is also crude force to the argument that the essence of poverty is a lack of money, and the best way to remedy that is to give people more money rather than messing around with indirect benefits like improving their air quality. A number of assumptions must hold for this argument to hold, including perfect rationality on the part of recipients and complete altruism toward family members.


The theory is summarized in Louis Kaplow & Steven Shavell, Should Legal Rules Favor the Poor? Clarifying the Role of Legal Rules and the Income Tax in Redistributing Income, 29 J. Legal Stud. 821 (2000). On the strong influence of this perspective within the legal academy, see Revesz, supra note 22, at 1510–11.


If government benefits phase out as income increases, this operates as a kind of tax on increased income and diminishes the incentive to seek that additional income. If the poor are given augmented tort damages but the augmentation phases out with higher income, there is a similar reduction in the incentive to obtain higher income.


Even on its own terms, this argument is subject to limitations, as Kaplow and Shavell have conceded. Id. at 827–32. If individual utility functions differ, then at any given income level, the marginal utility of income will be different for some people than for others. We can improve a redistribution scheme by adding a feature that sorts out people’s marginal utility for money more accurately than income alone can do. See Chris William Sanchirico, Deconstructing the New Efficiency Rationale, 86 Cornell L. Rev. 1003 (2001); Chris William Sanchirico, Taxes Versus Legal Rules as Instruments for Equity: A More Equitable View, 29 J. Legal Stud. 797 (2000). It is also possible that a legal rule may have the effect of reducing the value of leisure, thereby incentivizing work, or making it less desirable to increase current consumption, thereby incentivizing saving. These effects would reduce the distortions caused by the tax system, thereby improving efficiency. For a particularly clear explanation of these points, see Weisbach, supra note 22, at 262–65.


See David Gamage, How Should Governments Promote Distributional Justice? A Framework for Analyzing the Optimal Choice of Tax Instruments, 68 Tax. L. Rev. 1 (2014).


An additional argument may be that the ability of the tax system to fund redistribution may be limited by the ability of the rich to avoid taxation by moving assets and activities between national tax systems. See Tsilly Dagan, The Global Market for Tax and Legal Rules, 21 Fla. Tax Rev. 148 (2017).


Liscow, supra note 1, at 535–36.


See Revesz, supra note 22, at 1510–11.


Hemel, supra note 4, at 725–29. This claim does not respond to Revesz’s argument that Congress cannot respond fluidly as environmental problems change and as their impacts are revealed. Revesz, supra note 22, at 1519–20.


See Liscow, supra note 1, at 516–18.


Id. at 514–18.


Kaplow and Shavell, for instance, have argued that government decisions should consider only social welfare and should give no weight to values like human dignity, racial equality, or human rights. Louis Kaplow & Steven Shavell, Fairness Versus Welfare (2006). For a critique of their argument, see Daniel A. Farber, What (If Anything) Can Economics Say About Equity, 101 Mich. L. Rev. 1791 (2003) (review essay). For a more recent critique of welfarism by a leading law and economics scholar, see Eric Posner, The Boundaries of Normative Law and Economics, 38 Yale J. on Regul. 657 (2021). Posner finds the economic definition of utility questionable, id. at 660–61, argues that non-welfarist moral values are important, id. at 664–66, and views economic analysis as only one component of normative analysis, id. at 670–71.


On the issues involved in defining utility, see Mathew D. Adler, Measuring Social Welfare: An Introduction 41–76 (2019); see also Posner, supra note 99, at 660–61 (questioning the economic definition of utility.) Cass Sunstein, who has probably been the foremost advocate of cost–benefit analysis in the legal academy, has recently touched on the need for a broad (but therefore difficult to measure) understanding of welfare. See Cass R. Sunstein, Some Costs & Benefits of Cost-Benefit Analysis, 150 Daedalus 208, 213–15 (2021).


Liscow, supra note 1, at 518–23.


Revesz, supra note 22, at 1513–18.


Liscow, supra note 1, at 515. For other relevant discussions, see Howard S. Friedman, Ultimate Price: The Value We Place on Life 57–59 (2020); Weisbach, supra note 22; Adam M. Samaha, Death and Paperwork Reduction, 65 Duke L. J. 279 (2015). In an interesting variant of this argument, David Weisbach argued that agencies like EPA should apply cost–benefit analysis and an income-adjusted value of life because their legally assigned task is to correct market failures, not to fight poverty. Weisbach, supra, at 167–69. But to say that an agency’s task is addressing externalities is to provide an economist’s interpretation, not necessarily a description of the actual job of the agency as incorporated in statute. It is highly unlikely, to say the least, that the 1970 Congress thought of the Clean Air Act in terms of incomplete markets. Indeed, the central mandate of the Clean Air Act is to achieve national air quality standards, which EPA must set with no consideration of costs whatsoever, regardless of who will bear the costs. See Whitman v. Am. Trucking Ass’ns, Inc., 531 U.S. 457 (2001).


Rowell, supra note 22, at 8.


Circular A-4, supra note 60.


As OIRA conceded in Circular A-4, “The valuation of health outcomes for children and infants poses special challenges. It is rarely feasible to measure a child’s willingness to pay for health improvement and an adult’s concern for his or her own health is not necessarily relevant to valuation of child health. For example, the wage premiums demanded by workers to accept hazardous jobs are not readily transferred to rules that accomplish health gains for children.”


Hemel, supra note 4, at 717–19.


At least one leading advocate of cost–benefit analysis seems to agree:

Of course it is offensive and wrong to suggest that in principle, poor people are “worth less” than rich people. If poor people are subject to a risk of 1/10,000, they do not have less of a claim to public attention than wealthy people who are subject to the same risk; in fact they may have a greater claim, if only because they lack the resources to reduce that risk on their own.

Sunstein, supra note 23, at 394. Sunstein went on to say, “But the topic here is regulation rather than subsidy, and the two ought not to be confused.” Id. The remainder of the paragraph continued on the assumption that the full cost of any regulation is borne by its beneficiaries, though Sunstein then conceded that the cost for using an income-based value of life is weaker when the beneficiaries do not bear all of the costs. Id. at 395. As discussed in the next section, however, it is hard to be sure of the long-term incidence of pollution control requirements.


In contrast, equity-weighted cost–benefit analysis, which is discussed in the next section, gives effect to the fact that the poor make different (lower) trade-offs between life and money than the rich. But it is not clear why those individual trade-offs should determine the trade-offs that others such as polluters or society as a whole should make. If the costs of a regulation fall on a polluter, wealthier consumers, or the taxpayers, the regulation could be welfare-enhancing even if it costs more than the beneficiaries of the regulation would be willing to pay.


See Hemel, supra note 4, at 718–19.


See David Weisbach & Eric A. Posner, Climate Change Justice (2010). For a critique of this view, see Daniel A. Farber, Climate Justice, 110 Mich. L. Rev. 985 (2012).


To the extent that the protection for individuals is coming out of their own pockets, which may be the case when specific contract terms are regulated, government may want to take into account those individuals’ willingness to pay for risk reduction. In this situation, indviduals are receiving a private good, and their willingness to pay for this good is directly relevant.

However, this argument may not apply when beneficiaries are receiving a collective good such as pollution reduction. We could analyze a pollution regulation as a mandate for the company to improve its pollution control, combined with a “tax” to compensate the company in the form of higher consumer prices. The problem is that the “tax” may be regressive. Thus, the situation is analogous to using a regressive tax to finance a public good such as national security. That problem may arise in some situations and does deserve consideration, but weakening the standard is not the only way of countering the regressive effect.

Another way of looking at this is to use individual welfare as a metric. Health and safety risks inflict the same loss of welfare on people regardless of their income. The real distinction is on the payment side: because the marginal value of money is greater for the poor, the same dollar payment represents a greater welfare loss for them. If a regulation has this kind of regressive effect on the poor—something that is unclear in the case of some of the most important regulations—we should consider whether shifting the financial burden is feasible.


By using the term principle here, I mean that devoting equal resources to equal harms is a strong but not always controlling consideration. For instance, in some situations we might deviate from strict equality of resources by devoting greater resources to harms experienced by the poor in the interest of redistribution.


One potential problem is that there may be disagreement over the shape of the relevant utility curves. See Revesz, supra note 22, at 36.


For a comprehensive discussion of the issues involved in determining and combining individual utilities into a measure of social welfare, see Mathew D. Adler, Measuring Social Welfare: An Introduction 7–20, 41–76, 115–58 (2019).


Hemel pointed out that if the income elasticity of the value of a statistical life is one and welfare is a logarithmic function of income, all lives end up with the same value in units of welfare. Hemel, supra note 4, at 681. Thus, the two effects of income-sensitive valuation plus conversion to welfare units would cancel out. One wonders whether this is plausible. The upshot would be that the poor have lower utility in every time period than the rich (because the marginal utility of money is always positive), but the value of their future lives—the combination of all these time periods—is equal.


Id. at 708.


Hemel notes this disproportionate benefit. Id. at 707.


Id. at 705.


Id. at 706–07.


Another complication relates to whether lifetime or current income should be the basis of the weighting, a distinction that matters because average income changes over the life cycle. See Daniel A. Farber, Pollution Markets and Social Equity: Analyzing the Fairness of Cap and Trade, 39 Ecology L.Q. 1, 49–50 (2012).


Zsuzsanna Csereklye, Price and Income Elasticities of Residential and Industrial Electricity Demand in the European Union, 137 Energy Pol’y 111079 (Feb. 2020),




See Adrienne L. Thompson, Protecting Low-Income Ratepayers As the Electricity System Evolves, 37 Energy L.J. 265, 274–80 (2016).


The distributive impacts of these programs are heavily dependent on program design. More specifically, as I have discussed in a previous article, auctioning allowances is equivalent to imposing a carbon tax, which is likely to fall most heavily in proportional terms on poor households, but use of the proceeds can cancel out this regressive effect or even increase redistribution to the poor. Farber, supra note 121, at 51.


See Cal. Climate Invs. (data dashboard) (May 31, 2022),


See text accompanying note 84, supra.


Thompson, supra note 124, at 271. For additional information about community support programs, see Low Income Home Energy Assistance Program (LIHEAP), Admin. for Child. & Fams., U.S. Dept of Health & Hum. Servs.,


Thompson, supra note 128 at 271–72.


Id. at 276.


Id. at 276–78.


Id. at 278–79.


Revesz, supra note 22, at 1566–77. Revesz’s proposal relies on ad hoc spending programs to redress inequalities created by environmental regulations. In the absence of possible mitigation measures, the regulator would proceed with distributionally regressive regulations, leaving it to other parts of the executive branch to offset the regressive effect.


One recent study found, for instance, that equity weighting would increase the social cost of methane (the harm caused by emitting a ton of methane) by almost an order of magnitude, from $973 to $8,290. Frank C. Erickson et al., Equity Is More Important for the Social Cost of Carbon Than Climate Uncertainty, 592 Nature 564 (2021).


U.S. Federal Poverty Guidelines Used to Determine Financial Eligibility for Certain Federal Programs, Off. of the Assistant Secy for Plan. & Evaluation, U.S. Dept of Health & Hum. Servs. (Feb. 1, 2021),


GDP per Capita (Current US$) - Congo, Dem. Rep., World Bank, (2020 data).


Liscow, supra note 1, at 519. Taxes would also look quite different, with regressive effects that would be offset by demogrants (equal payments to members of demographic groups, such as all 18 year olds). Id.


Perhaps as high as 75%. See Hemel, supra note 4, at 665.


Liscow, supra note 1, at 523.




As Liscow observed, “[a]cross the law, there is no normatively neutral place from which to set distributional goals. All policies have distributional consequences. Making some choice is unavoidable. And it is not at all clear that either Congress or the public wishes for no ‘redistribution’ outside of Congress.” Id. at 555.


This argument does not weigh against Liscow’s proposal of making small redistributive adjustments to regulations, which would hold if the public’s true redistributive preferences were only somewhat different from the tax law’s redistributive effect.


See Revesz & Yi, supra note 22, at 37 (avoiding wholesale changes in the regulatory system); Hemel, supra note 4, at 47, 53–57.


I should note that, under certain assumptions, equity weighting also implies spending roughly equal amounts to save lives regardless of the beneficiaries’ income. Although rich individuals are willing to pay more to reduce risks, this is roughly canceled out by the fact that poorer individuals place a higher value on each dollar. See Hemel, supra note 4, at 681.


For instance, the OSHA general duty clause provides: “(a) Each employer – (1) shall furnish to each of his employees employment and a place of employment which are free from recognized hazards that are causing or are likely to cause death or serious physical harm to his employees.” 29 USC § 654(a).


One might argue that the willingness of the average person to pay to avoid a risk is a good gauge of community values, given that the average person is considered representative of the community in a democracy.


This proposal is made in Revesz and Yi, supra note 22, at 38. This role could also be consistent with Liscow’s adage, “Redistribute in a lot of cases, but modestly.” Liscow, supra note 1, at 535.


The Biden Plan to Secure Environmental Justice and Equitable Economic Opportunity, Joe Biden, Accessed June 6, 2023.


Id. Similarly, the revised executive order proposed by the Council calls for “not only repairing past and current harm and preventing future injustices, but also rooting out and dismantling systemic racism and other forms of institutionalized bias in our laws, policies, and practices.”


Libertarian readers and those embracing complete governmental color blindness may disagree with these goals. Engaging their views is outside the scope of this article, but those readers should find the legal analysis relevant regardless of this normative disagreement.


As Gerald Torres explained, the environmental justice movement is “rooted in local activism,” with scholars and lawyers mostly struggling to keep up, leading “a critique of environmental decision-making based upon observed results rather than allegations of specific procedural defects.” Gerald Torres, Environmental Justice: The Legal Meaning of a Social Movement, 15 J.L. & Com. 597, 601 (1996).


Richard J. Lazarus, Pursuing “Environmental Justice”: The Distributional Effects of Environmental Protection, 87 Nw. U. L. Rev. 787, 788 (1993).


Id. at 789.




Id. at 790 n.21.


Robert D. Bullard, Introduction: Environmental Justice – Once A Footnote, Now A Headline, 45 Harv. Envt L. Rev. 243 (2021).


Richard J. Lazarus, “Environmental Racism! That’s What It Is,” 2000 U. Ill. L. Rev. 255, 257 (2000).


Id. Another version of the story places his first use of the term a few years earlier, but there does not seem to be much dispute about his origination of the term. Id. at 257 n. 17.


Id. at 260.




Robert D. Bullard, Dumping in Dixie: Race, Class, and Environmental Quality (1990).


Torres, supra note 151, at 602.


Bullard, supra note 156, at 248.


Barry E. Hill, Human Rights, Environmental Justice, and Climate Change: Flint, Michigan, 46 Hum. Rts. 4, 16–17 (2021).




Friends of Buckingham v. St. Air Pollution Control Bd., 947 F.3d 68 (4th Cir. 2020).


Id. at 71.


Id. at 92. One of the main points in dispute was whether the immediate neighborhood qualified as a “minority EJ community.” Id. at 87–88.


Vecinos para el Bienestar de la Comunidad Costera v. FERC, 6 F.4th 1321 (D.C. Cir. 2021).


Id. at 1325.


Id. at 1330–31.


Jean Chemnick, Experts to White House: EJ Screening Tool Should Consider Race, Climatewire (June 1, 2022, 6:40 AM EDT),


Id. To the same effect, see Lisa Friedman, White House Takes Aim at Environmental Racism, but Won’t Mention Race, N.Y. Times (Feb. 15, 2022), An earlier effort as part of a financial assistance program to include race as a factor in a program in order to aid disadvantaged farmers got swept up in litigation. See Miller v. Vilsack, No. 21-11271, 2022 WL 851782 (5th Cir. 2022).


As discussed below, generally applicable regulations are less likely to run into barriers regarding color blindness than decisions that select individuals or firms on the basis of criteria that appear to be connected to race. Simply put, a decision to tighten air quality standards (even if intended to improve air quality in a Black community) may appear less akin to affirmative action than a decision to designate funding for communities based on their racial composition.


Environmental justice may also cover other factors such as poverty. Since income inequality was the subject of the preceding section, this section will focus on the issue of race.


A further issue is the extent to which the agency should disaggregate effects in terms of multiple racial groups, perhaps with gender as an additional factor. Attempting to do so would involve significantly greater analytic efforts by agencies whose resources are already stretched. Moreover, such fine-grained analysis would undoubtedly increase the divisiveness of regulatory decisions.


Exec. Order No. 12,898 § 101, 50 Fed. Reg. 7629 (1994).




The foundational civil rights case is Griggs v. Duke Power Co., 401 U.S. 424 (1971), in which the Court held that Title VII “proscribes not only overt discrimination but also practices that are fair in form, but discriminatory in operation, unless an employer can prove business necessity.”


In the context of employment, where a fixed number of people are being hired or promoted or a fixed pool of money is available for salaries, any improvement in the percentage of benefits to whites means a decrease for other groups and thus has a disparate impact. The same is not true in the regulatory context: a disproportionate decrease in pollution experienced by whites does not necessarily translate into a worsening of the pollution experienced by other groups.


Disparate impact analysis is likely to have less traction in the case of regulations than in other situations such as permitting projects, where the project may cause net environmental harms and the only question is how they are distributed.


For example, a regulatory rollback might increase risk across the board, with a disproportionate impact on communities of color. Alternatively, a regulation increasing protection against one risk might indirectly result in another risk that could disproportionately impact communities of color or create hotspots impacting those communities.


Lara Cushing et al., Carbon Trading, Co-Pollutants, and Environmental Equity: Evidence from California’s Cap-and-Trade Program (2011–2015), PLOS Med. (July 10, 2018),


Id. at 14.


A later study found that this effect was only temporary. Danae Hernandez-Cortes & Kyle C. Meng, Do Environmental Markets Cause Environmental Injustice? Evidence from California’s Carbon Market (Nat’l Bureau of Econ. Rsch., Working Paper No. 27205, 2020).


The California Air Resources Board (CARB) recounts efforts to control large pollution sources and pollution from trucks in certain locations in response to these concerns. Environmental Justice Communities and Local Air Pollution, Cal. Air Res. Bd.,


Rollerson v. Brazos River Harbor Navigation Dist. of Brazoria Cnty., Tex., 6 F.4th 633, 648 (5th Cir. 2021) (Ho, J., concurring in part and concurring in the judgment). The case involved a boat terminal expansion that was part of a harbor improvement effort, other parts of which had received federal funding. The court remanded on the question of whether this federal funding subjected the terminal expansion to Title VI of the Civil Rights Act.


Id. at 648. In addition, Judge Ho remarked, “opponents of disparate impact theory worry that it will only exacerbate, rather than alleviate, racial tension—by pressuring defendants to adopt policy changes for the explicit purpose of taking from some and giving to others based on their race.” Id. They fear that disparate impact theory means not only presuming discrimination, but requiring it. Id. at 649.


Id. at 650.


See 42 U.S.C. § 2000d; Alexander v. Sandoval, 532 U.S. 275, 280–81 (2001) (stating that the statute “prohibits only intentional discrimination,” not “activities that have a disparate impact on racial groups”).


Rollerson, 6 F.4th at 648.


Ricci v. DeStefano, 557 U.S. 557 (2009).


Id. at 553–56.


Id. 562.


Id. at 580.


Id. As originally enacted, Title VII did not explicitly address disparate impact, though the Supreme Court interpreted the statute as allowing the imposition of liability on the basis of disparate impacts under some circumstances. See Griggs v. Duke Power Co., 401 U.S. 424 (1971). Title VII was later amended to add what is now 42 USC § 200e-2(k), which codifies the Griggs doctrine in employment discrimination cases.


Ricci, 557 U.S. at 584.




Ricci, 557 U.S. at 594 (Scalia, J., concurring).


Id. at 595.


Tex. Dep’t of Hous. & Cmty. Affs. v. Inclusive Cmtys. Project, Inc., 576 U.S. 519 (2015). The Court held that “[a]ntidiscrimination laws must be construed to encompass disparate-impact claims when their text refers to the consequences of actions and not just to the mindset of actors, and where that interpretation is consistent with statutory purpose.” Id. at 533.


Justices Scalia, Thomas, and two others joined another dissent arguing that disparate-impact analysis was a poor fit for fair housing problems. Id. at 557 (Alito, J., dissenting).


For a thoughtful exploration of the constitutional considerations in play, see Richard A. Primus, Equal Protection and Disparate Impact: Round Three, 117 Harv. L. Rev. 493 (2003).


Others have come to the same conclusion. See, e.g., Deborah Hellman, Measuring Algorithmic Fairness, 106 Va. L. Rev. 811, 862–64 (2020).


Rollerson v. Brazos River Harbor Navigation Dist. of Brazoria Cnty., Tex., 6 F.4th 633, 648 (5th Cir. 2021).


Because regulation is not a zero-sum game among regulatory beneficiaries, it could also be difficult for whites to show that disparate-impact analysis resulted in any harm to them, a requirement for standing. There is also an interesting question regarding standing to raise the issue. The most likely challengers of a regulation—regulated firms—might not have third-party standing to raise the equal protection claim.


See, e.g., West Virginia v. EPA, 142 S. Ct. 2587 (2022); FDA v. Brown & Williamson Tobacco Corp., 529 U.S. 129, 159–60 (2000); Util. Air Regul. Grp. v. EPA, 573 U.S. 302, 324 (2014); King v, Burwell, 576 U.S. 473, 484–86 (2015). There is already an extensive literature about this doctrine, which is sure to grow after West Virginia. See, e.g., Cass R. Sunstein, There Are Two “Major Questions” Doctrines, 73 Admin. L. Rev. 475 (2021); Blake Emerson, Administrative Answers to Major Questions: On the Democratic Legitimacy of Agency Statutory Interpretation, 102 Minn. L. Rev. 2019 (2018); Jonas Monast, Major Questions About the Major Questions Doctrine, 68 Admin. L. Rev. 445 (2016).


West Virginia v. EPA, 142 S. Ct. 2587 (2022).


Michigan v. EPA, 576 U.S. 743, 752 (2015).


Agencies may also be empowered to issue disparate impact regulations by Title VI of the Civil Rights Act, or at least the Supreme Court has been willing to entertain this view without expressly ruling on it. See Alexander v. Sandoval, 532 U.S. 275 (2001).


In practice, the situation might be more complex. For instance, a 2018 study found that a California emissions trading system known as RECLAIM reduced risks in all segments of the population, with greater reductions for Blacks than whites but smaller reductions for Latinos than whites. See Cobett Grainger and Thanicha Ruangmas, Who Wins from Emissions Trading? Evidence from California, 71 Env. & Res. Econ. 703 (2018). A previous study had found no evidence that changes in emissions were related to race. See Meredith Fowlie et al., What Do Emissions Markets Deliver and to Whom? Evidence from Southern California’s NOx Trading Program, 102 Amer. Econ. Rev. 965 (2012). But still earlier, there was evidence that trading produced greatly uneven impacts due to the purchase of allowances by oil refineries; the allowances were generated by retiring polluting cars, a dispersed pollution source, whereas the refineries were geographically concentrated, which allowed them to avoid installing expensive pollution control equipment. See Richard Toshiyuki Drury et al., Pollution Trading and Environmental Injustice: Los Angeles’ Failed Experiment in Air Quality Policy, 9 Duke Env. L. & Poly F. 231, 242–79 (1999). Because the program was modified over the course of this time period, it is difficult to directly compare these studies.


White House Environmental Justice Advisory Council, Final Recommendations: Justice40 Climate and Economic Justice Screening Tool & Executive Order 12898 Revisions (May 21, 2021).


Id. at 89.


Id. at 77–78.




Id. at 85.


See Fisher v. Univ. of Tex. at Austin, 136 S. Ct. 2109 (2016); City of Richmond v. J. A. Croson Co., 488 U.S. 469 (1989).


Cooper v. Harris, 137 S. Ct. 1455, 581 U.S. ___ (2017) (holding that race cannot be used as a proxy for political affiliation in redistricting).


See, e.g., Deniz Yeter et al., Disparity in Risk Factor Severity for Early Childhood Blood Lead Among Predominantly African-American Black Children: The 1999 to 2010 US NHANES, 3 Intl J. Envt Rsch. Pub. Health 1552 (2020), doi:10.3390/ijerph17051552.


See, e.g., Parents Involved in Cmty. Schs. v. Seattle Sch. Dist. No. 1, 551 U.S. 701 (2007) (school district failed to show that race-neutral alternative was unavailable); Fisher v. Univ. of Tex. at Austin, 570 U.S. 297 (2013) (university must show that no race-neutral alternative was available); Wygant v. Jackson Bd. of Educ., 476 U.S. 267, n.6 (1986); Adarand Constructors, Inc. v. Peña, 515 U.S. 200, 238 (1995); City of Richmond v. JA Croson Co., 488 U.S. 469, 507–10 (1989). Arguably, Fisher is distinguishable because the Court had already found a compelling interest (diversity) and the only issue was whether the means were narrowly tailored.


The district court struck down the policy. Coalition for TJ v. Fairfax Cnty. Sch. Bd., No. 1:21cv296, 2022 WL 579809 (E.D. Va. 2022).


Coalition for TJ, 2022 WL 986994 (Mar. 31, 2022). A concurring judge expressed “grave doubts about the district court’s conclusions regarding both disparate impact and discriminatory purpose” as well as a reservation about the burden of the injunction on the school. Id. at *4. The dissenting judge maintained that “[r]acial balancing is no less pernicious if, instead of using a facial quota, the government uses a facially neutral proxy motivated by discriminatory intent.” Id. at *7.


Coalition for TJ v. Fairfax Cnty. Sch. Bd., 142 S. Ct. 2672 (2022). The dissenting Justices did not provide explanations for their votes.


See Vill. of Arlington Heights v. Metro. House Dev. Corp., 419 U.S. 252 (1977) (adopting this standard when the motive in question is a desire to discriminate against Blacks in housing).


For use of this test in the discrimination context, see Univ. of Tex., Sw. Med. Ctr. v. Nassar, 570 U.S. 338 (2013); Gross v. FBL Fin., 557 U.S. 167 (2009).


See Miller v. Johnson, 515 U.S. 900 (1995).


42 U.S.C. § 7409(b)(1).


For instance, the Clean Air Act requires new sources to meet a standard of performance, defined as “the degree of emission limitation achievable through the application of the best system of emission reduction which (taking into account the cost of achieving such reductions and any non-air quality health and environmental impact and energy requirements) the Administrator determines has been adequately demonstrated.” 42 U.S.C. §§ 7411(a)(1), (b).


In a recent immigration case, the Fifth Circuit made a similar argument:

For example, [an official memorandum] provides that the guidelines “are essential to advancing this Administration’s stated commitment to advancing equity for all, including people of color and others who have been historically underserved, marginalized, and adversely affected by persistent poverty and inequality.” DHS’s replacement of Congress’s statutory mandates with concerns of equity and race is extralegal, considering that such policy concerns are plainly outside the bounds of the power conferred by the INA [Immigration and Nationality Act].

Texas v. United States, 40 F.4th 205 (2022). While this argument may have been misplaced in that particular case, the issue of statutory fit is bound to arise in other settings.


Hemel, supra note 4, at 707.


Some of these issues are discussed in Robert E. Kuehn, The Environmental Justice Implications of Quantitative Risk Assessment, 1996 U. Ill. L. Rev. 103, 116–25 (1996).


See Clifford J. Villa, Remaking Environmental Justice, 66 Loy. L. Rev. 469, 515 (2020).


See chapter 5 of Daniel A. Farber et al., Disaster Law and Policy (3d ed., 2015).


This approach has a bipartisan pedigree. My example is from the Biden administration, but the Trump administration used it in analyzing the costs and benefits of its own replacement for the Obama rule. EPA, Regulatory Impact Analysis for the Repeal of the Clean Power Plan, and the Emission Guidelines for Greenhouse Gas Emissions from Existing Electric Utility Generating Units 4–19 (June 2019),


EPA explains the calculation as follows:

We estimate counts of PM2.5-related total deaths (yij) during each year i (i = 1, …, I where I is the total number of years analyzed) among adults aged 30 and older (a) in each county in the contiguous U.S. j (j = 1, …, J where J is the total number of counties) as
where moija is the baseline all-cause mortality rate for adults aged a = 30–99 in county j in year i stratified in 10-year age groups, β is the risk coefficient for all-cause mortality for adults associated with annual average PM2.5 exposure, Cij is the annual mean PM2.5 concentration in county j in year i, and Pija is the number of county adult residents aged a = 30–99 in county j in year i stratified into 5-year age groups.
EPA, Control of Air Pollution from New Motor Vehicles: Heavy-Duty Engine and Vehicle Standards Draft: Regulatory Impact Analysis 385 (March 2022), I have corrected the typography in equation 1, where an exponential is clearly intended. The correct typography can be seen in EPA’s risk analysis for a regulation the prior year. See EPA, Regulatory Impact Analysis for the Final Revised Cross-State Air Pollution Rule (CSAPR) Update for the 2008 Ozone NAAQS (March 2021), The same formula was used by the Trump administration. See EPA, Regulatory Impact Analysis for the Repeal of the Clean Power Plan, and the Emission Guidelines for Greenhouse Gas Emissions from Existing Electric Utility Generating Units 4–19 (June 2019),


For a discussion of the difficulties in choosing the appropriate geographical unit for analysis, see Revesz & Yi, supra note 22, at 70–73.


Urban census tracts average two square miles See Ann E. Carlson, The Clean Air Act’s Blind Spot: Microclimates and Hotspot Pollution, 65 UCLA L. Rev. 1036, 1059 n.92 (2018). Depending on the size and shape of the tract, there could be considerable variation in distance from a roadway.


Id. at 1056–57 (in the original, 300-500 meters); I have converted to units more readily understood by American readers.


Id. at 1057 (200 meters).


Id. at 1057–58 (150 meters).


Id. at 1061–64 (the evidence about strategic behavior is cited at 1061 n.100). Congress has attempted to provide greater monitoring for environmental justice communities in two sections of the Inflation Reduction Act (P.L. 117–169). Section 60105 provides over $150 million to fund additional monitoring, and the $2.8 billion environmental justice block grant program in section 60201 (which will become Clean Air Act section 138) includes improved monitoring as a grant purpose.


In addition, this kind of multiplicative model might not be appropriate in all circumstances. Although many pollutants may amplify problems for those who already are in poor health, this kind of interaction between existing risks and pollution effects conceivably might not be true for all pollutants.


National Emission Standards for Hazardous Air Pollutants: Coal- and Oil-Fired Electric Utility Steam Generating Units-Revocation of the 2020 Reconsideration, and Affirmation of the Appropriate and Necessary Supplemental Finding; Notice of Proposed Rulemaking, 87 Fed. Reg. 7624 (Feb. 9, 2022).


Michigan v. EPA, 576 U.S. 743 (2015).


87 Fed. Reg. 7624, 7636.


Id. at 7627.


Id. at 7633.


Id. at 7634.


Id. at 7635 n.29.


Id. at 7647.


Black individuals may also be disadvantaged because they may be less likely to belong to informal networks including others with accumulated wealth through which they can gain financially. See Addo, supra note 12, at 49.


Notably, Black people are less likely to receive inheritances than whites, and when they do, the average amount is lower. Id. at 49.


Revesz & Yi, supra note 22, at 69–93.


Id. at 73.


See text accompanying notes 16–21, supra.


RAND has created an interactive tool that may be helpful in applying disparate impact analysis. As its creators describe it:

Researchers from RAND have developed a free online tool that shows the results [of redlining]. It maps those historic red lines against more than a dozen environmental hazards, from air pollution to toxic waste sites to smothering summer heat. It shows, for more than 200 metro areas, where one racist policy continues to shape lives even decades after it was revoked.

Doug Irving, Essay: Environmental Racism: How Historic Redlining Continues to Affect Communities, RAND Corp. (June 27, 2022),


See Sonja B. Starr, Race-Norming and Statistical Discrimination: Beyond the NFL, Yale L. J. (forthcoming), (May 2022); Ronen Avraham & Kyle D. Logue, Understanding Insurance Antidiscrimination Laws, 87 S. Cal. L. Rev. 195 (2014); Kimberly A. Yuracko & Ronen Avrham, Valuing Black Lives: A Constitutional Challenge to the Use of Race-Based Tables in Calculating Tort Damages, 106 Cal. L. Rev. 325 (2018); Deborah Helman, Measuring Algorithmic Fairness, 106 Va. L. Rev. 811 (2020).


Starr, supra note 257.


Id. at [3].


Starr writes that “[t]hese binding precedents are squarely applicable to the above-described practices. Yet with almost no exceptions, the courts have not weighed in, and the practices persist.” Id. at [3].


Id. at [23–24].


Id. at [74–77].


Id. at [76].


If risk caused by pollution is a linear function of exposure across the entire population, then only the average exposure level of the entire population is relevant to assessing total harm. However, if the same exposure can cause different harms to different groups, perhaps due to differences in comorbidities or health care, it becomes crucial to connect exposures with demography to calculate harm. Thus, taking race and health into account can improve the accuracy of a model for estimating harm.


For example, suppose for the sake of simplicity that the population is 30% Black and 70% white. Assume Alternative A reduces risk levels for Black communities by 30% and white communities by 20%, for an overall risk reduction of 23%. Alternative B reduces risk levels for Black communities by 40% and risk levels for white communities by 10%, for an overall risk reduction of 19%. The question is whether a regulator could adopt Alternative A on the ground that health improvements for Blacks are a higher priority given current and historic inequities. Even apart from any potential constitutional issue, this could be hard to justify on the basis of a statutory mandate to protect public health or one to select the “best” method of reducing risks. I certainly would not want to make the contrary argument to the current Supreme Court.


As Revesz has pointed out, it is also important to avoid wholesale changes in methodologies that courts are familiar with. See Revesz, supra note 22, at 37.

Author notes


Sho Sato Professor of Law and Faculty Director of the Center for Law, Energy, and the Environment at the University of California, Berkeley. I would like to thank Hanoch Dagan, Catherine Fisk, Jon Gould, David Hausman, Sharon Jacobs, Mark Gergen, Manisha Padi, Jon Simon, Jennifer Urban, and Emily Zhang for their generous and helpful comments. I also thank the editors for inviting a stellar group of commentators to respond to this article. Rather than making revisions in the response to the comments or providing a rushed reaction in this issue of the journal, I plan to provide a more considered response in a follow-up article.

This is an open-access article distributed under the terms of the Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License, which permits you to copy and redistribute in any medium or format, for non-commercial use only, provided that the original work is not remixed, transformed, or built upon, and that appropriate credit to the original source is given. For a full description of the license, please visit