Our focus here is on whether the administration’s view that there are noncompetitive labor market influences also leads to downward biases in the value of a statistical life (VSL) used to monetize mortality risks in regulatory impact analysis. The VSL suggested for use by the Office of Management and Budget in the latest proposed draft of Circular A-4 is in the $10 million to $12 million range and is to be adjusted regularly for inflation and real income growth. The new issue raised by concern with noncompetitive forces is whether there is a need for additional correction for possible employer monopsony power. Based on our examinations of the VSL in different labor market contexts, we do not find that there is any current rationale for increasing the VSL to account for possible noncompetitive forces.
VSL and labor market noncompetition / The U.S. Treasury has produced a lengthy report on estimates of the amount of labor market noncompetition in the United States. The report does not document any inequity and inefficiency consequences from this noncompetition, but instead jumps straight to policy recommendations that go so far as to have the Department of Justice pursue criminal cases against employers. Most recently the general counsel of the National Labor Relations Board issued a memo stating her opinion that noncompete provisions in employment contracts and severance agreements most typically violate the National Labor Relations Act. Our concern here follows a different thread of the labor market effects of noncompetition, which is whether it biases downward the estimates of the VSL, which in turn reduces the assessed benefits of risk regulations.
It is useful to review how labor markets work. Jobs have many different attributes, including the wage rate and the fatality rate. Workers considering jobs will face a variety of employment possibilities. The best wage available at different health risk levels will be greater for higher fatality risk jobs. More dangerous jobs tend to pay more. Workers select the wage and fatality risk combination they most prefer. The slope of the market wage offers at the risk level selected by the worker reveals the VSL. In particular, it indicates how much extra pay the worker demands for an increase in the risk.
Perfectly competitive labor markets are not required for riskier jobs to have to pay more. However, if the labor market is not competitive, the locus of the best available jobs at any given risk level—known as the hedonic wage equation—will be flatter and lower. That is, the lack of maximally competitive buyers will both lower the wage level and reduce the marginal compensation for additional fatality risk. As a result, noncompetitive labor markets should produce a lower VSL. It is important to note that monopsony’s qualitative effects as just described are similar to lower estimated VSLs for disadvantaged groups such as blacks and Mexican immigrants who do not speak English.
Estimates of VSL controlling for noncompetitive forces / Emerging research cited in the Treasury Department report experiments with how to measure monopsony power and how to infer its wage consequences. A straightforward approach has been to append variables connected to the degree of monopsony to an estimated hedonic wage equation. The set of variables believed to be connected to the degree of employer wage-setting power has included measures of local labor market concentration, which researchers have questioned regarding their use for anti-monopsony policy.
For the topic of possible employer monopsony power, we continue to embrace the philosophy that simpler is better, or KISS (Keep It Sophisticatedly Simple). As we have already noted, discussions of monopsony power focus on noncompetitive aspects of industries or occupations, such as licensing laws and noncompete agreements. Our strategy here is to examine how the wage equations used to estimate the VSL are affected by controlling for detailed occupation and industry information. If monopsony power is important, then adding industry and occupation controls should boost the implied VSL by conditioning out all economically relevant industry and occupation characteristics. The difference between a VSL estimated by wage regression equations with and without labor market employer concentration controls via the worker’s detailed industry-occupation cell gives an indication of monopsony power (among other things associated with the local labor market). It also serves as a check on the interpretation of the difference in the height and slope of the hedonic wage equation as possible monopsony power estimates.
In a previous article using panel data with extensive model specification searches, we found that the estimated VSLs that include the consequences of monopsony power were one-third to one-half lower than when monopsony power is controlled for with detailed industry and occupation indicators. The VSL as currently estimated typically controls for the negative effects of monopsony power (employer concentration) on the slope and level of the hedonic wage equation so that adjustments need not be made to VSL estimates currently in play for the degree of labor market competition.
Summary and conclusion / Another way of thinking of what we are saying is that we have examined VSL estimates from less ambitious but more robust econometric models that bypass any attempt to measure the wage effects of differences in monopsony power, in favor of removing it along with other forces of industry and occupation on wage outcomes and VSL. The results are in essence the opposite of what might be feared by policymakers. The controls for detailed industry and occupation of employment should also help to purify estimates of effects of the situation where employer monopsony power is accompanied by product market monopoly power such that both forms of noncompetition impinge on the wage setting process.
The procedure we have explained here provides a good indication of whether there is a large-scale problem in the application of VSL estimates as currently produced by researchers. There is no such monopsony bias problem in VSL estimates. The approach for using VSL as described in the draft of Circular A-4, and as typically applied in benefit–cost analysis in general, should stand unchanged.
Readings
Readings- “Compensating Differential for Occupational Health and Safety Risks: Implications of Recent Evidence,” by Thomas J. Kniesner and W. Kip Viscusi. Research in Labor Economics 50: 83–116 (2023).
- Pricing Lives: Guideposts for a Safer Society, by W. Kip Viscusi. Princeton University Press, 2018.
- “The Value of a Statistical Life: Evidence from Panel Data,” by Thomas J. Kniesner, W. Kip Viscusi, Christopher Woock, and James P. Ziliak. Review of Economics and Statistics 94(1): 74–87 (2012).
