This post is part of a symposium on the future of cost-benefit analysis. Read the rest of the symposium here.
US federal regulatory cost-benefit analysis has long focused on “efficiency.” Though longstanding language in executive orders nods in the direction of considering distributional impacts, the reality is that distributional concerns are missing in regulatory analysis.
Consider how current cost-benefit analysis at the federal Department of Transportation is used to allocate spending. The most important benefit in the analysis of potential projects is the value of time saved. However, not everyone’s time is valued equally. Current policy values an hour of transit time saved by building an airport, which primarily serves the rich, at $63 per hour; an hour saved on a bus line, which primarily serves the poor, is only $25 per hour. This is efficient: rich persons can earn more in that hour saved. But since the time of the rich is valued at such a higher rate, this policy pushes funding toward the rich instead of the poor, making it harder for low-income people to access jobs.
This inattentiveness to distribution—and, worse yet, exacerbation of inequality—is wrong. The standard economic rationale for “efficient” cost-benefit analysis is that all nontax policies should be indifferent to distributional impacts, and that any necessary redistribution should take place through taxes and transfers. But as I describe in my new paper Redistribution for Realists, the idea that taxes would be sufficient to address all redistributive needs is unrealistic because the real-world public resists this degree of redistribution through taxation. One reason for this is that most people tie taxes to “desert”: they think the money people earn is in some meaningful sense “theirs” and are thus reluctant to adopt—even hostile to—heavily redistributionist taxation. The Article estimates that, partly as a result of this resistance, the tax code is only about one ninth as redistributive as reasonable baselines suggest that it would need to be to maximize welfare. This is not because the public doesn’t want greater equality—it does. The public just doesn’t want it achieved exclusively through taxation.
If we cannot achieve a fair distribution through taxation alone, we’ll need to do it elsewhere. We should adopt what I call a “thousand points of equity” approach: throughout each of many areas of policy, we should redistribute modestly, since there are typically at least some low-cost opportunities to redistribute in each policy area. This approach does the most good at the least cost. For example, the Department of Transportation could spend more on rapid bus lines in dense, high-benefit neighborhoods of low-income people to help them get to work.
More generally, regulatory cost-benefit analysis is one such area to address distribution. And, indeed, President Biden has directed his administration to come up with a method to consider distributional impacts in regulatory cost-benefit analysis. But, practically, what process should the Administration use to do that?
Broadly, there are three possibilities:
- Measure distributional impacts. The most basic response is to simply require measurement of distributional impacts, broken down by income. That way, if a proposed regulation hurts the poor or disproportionately benefits the rich, regulators would be made aware of this fact and can alter the policy in response. Their goal would no longer be to solely maximize efficiency. Instead, on a case-by-case basis, they would be maximizing some combination of efficiency and distributional goals.
- “Cleansing” measures of costs and benefits. An alternative would be to not only measure the distributional impacts, but also “cleanse” those distributional impacts across different incomes. So, for example, when measuring the effects of transportation spending, both rich and poor would be assigned the same value for their time. That would systematically tilt spending back toward the poor. (The government already does something like this with a uniform “value of statistical life.”) The same idea could apply more generally. When measuring the costs and benefits, regulators could counteract the fact that policies benefitting the poor—such as cleaning up polluted areas, which tend to be in poor neighborhoods—will tend to be given less weight because of the poor’s lower willingness (or rather, ability) to pay for those benefits.
- Distributional weights. From the perspective of economics, if maximizing well-being is the goal, the most natural approach would be to use explicit “distributional weights,” requiring by executive order that a dollar of benefits is worth more to the poor than the rich. Under this approach, the recommended regulation would depart from the efficient one if enough benefits or few enough costs are concentrated on the poor.
The UK has long had such weights as part of its formal regulatory regime. They assign weights to regulatory costs and benefits by the inverse of each person’s income. If your income were $10, your weight would be 1/10. If your income were $100, your weight would be 1/100.
This weighting used by the UK has a fairly strong economic underpinning. It is supported by both individuals’ own behavior—in how much more they value $1 when they are poor versus rich—and cross-country studies on how happiness varies with income. Ultimately, how heavily to weight by income is a normative choice (though, of course, no less a normative choice than refraining from weighting). The Administration could choose whatever weights it finds appropriate.
Distributional weights would come closest to maximizing social welfare (at least as conventionally defined). For one, it directly targets that goal: If the increase in welfare of $1 spent on the poor is ten times the increase in welfare of $1 spent on the rich, then accounting for that precise tradeoff will maximize welfare. Measuring the distributional impacts and “cleansing” disparate cost and benefit calculations would help, but these two approaches would not target welfare maximization as precisely as distributional weights. Simply measuring distributional impacts does not require any specific, systematic welfare tradeoff, and “cleansing” ultimately treats rich and poor the same, rather than redistributing to the poor.
Which of these distributional strategies to adopt ultimately boils down to a political choice. Imposing distributional weights might be controversial politically, not to mention legally. It might be that the public finds explicit weighting quite problematic. We have no direct evidence on this, but the US tends to have strong legal and social norms of formal equality. And explicitly weighting the benefits conferred to some individuals more than others could violate those norms. On the other hand, it could be that the public would pay almost no attention to such procedures of government bureaucrats—or that they would like it if they did pay attention. Ultimately, this choice brings up deep questions of democratic theory: How, if at all, does it matter normatively what the public thinks? Or, alternatively, is this just a question of political feasibility, which depends on the difficult-to-predict reception of a new policy?
This is an exciting time for those interested in achieving greater equity in cost-effective ways through administrative cost-benefit analysis. The Administration has at least three good options for doing this. It is about time that the regulators take equity seriously.