Pricing Lives

Guideposts for a Safer Society

W. Kip Viscusi

• Editorial: Princeton University Press
• Año: 2018
• ISBN:
  9781400889587
• Categoría: Business
• Categoría: Government & Business

How society's undervaluing of life puts all of us at risk—and the groundbreaking economic measure that can fix it Like it or not, sometimes we need to put a monetary value on people's lives. In the past, government agencies used the financial "cost of death" to monetize the mortality risks of regulatory policies, but this method vastly undervalued life. Pricing Lives tells the story of how the government came to adopt an altogether different approach--the value of a statistical life, or VSL—and persuasively shows how its more widespread use could create a safer and more equitable society for everyone.In the 1980s, W. Kip Viscusi used the method to demonstrate that the benefits of requiring businesses to label hazardous chemicals immensely outweighed the costs. VSL is the risk-reward trade-off that people make about their health when considering risky job choices. With it, Viscusi calculated how much more money workers would demand to take on hazardous jobs, boosting calculated benefits by an order of magnitude. His current estimate of the value of a statistical life is $10 million. In this book, Viscusi provides a comprehensive look at all aspects of economic and policy efforts to price lives, including controversial topics such as whether older people's lives are worth less and richer people's lives are worth more. He explains why corporations need to abandon the misguided cost-of-death approach, how the courts can profit from increased application of VSL in assessing liability and setting damages, and how other countries consistently undervalue risks to life. Pricing Lives proposes sensible economic guideposts to foster more protective policies and greater levels of safety in the United States and throughout the world.

1

How Pricing Lives Saves Lives

The Challenge of Valuing Mortality Risks

“We can’t do that. That’s immoral.” This was the reaction I got in 1980 when I suggested to a prominent Occupational Safety and Health Administration (OSHA) official that the agency monetize the reduced risks of death from job safety regulations using labor market estimates of workers’ valuation of fatality risks. The values I advocated were based on the extra amounts workers are paid for each expected workplace death. Early studies often referred to these figures as the value of life, but the terminology used to describe this approach has evolved to be the “value of a statistical life,” which is both more accurate and somewhat less inflammatory.

The idea of monetizing the benefit of reduced worker fatality risks was not controversial. OSHA and other agencies had routinely attached dollar values to the expected lives that would be saved by regulations. In doing so, they followed the general approach patterned after that used by the courts in wrongful death cases, in which they equated the benefit of reduced risks of death to the value of lost earnings and medical expenses. This formulation enabled agencies to generate a mortality risk reduction benefit number and to be able to point to the use of a similar approach by the courts, thus providing some evidence of its reasonableness. However, there is a fundamental disconnect between these values and the core principle underlying benefit assessment, which is that benefit values for government policies should reflect society’s willingness to pay for the benefit. An instructive way to ascertain these values is to examine the revealed preferences based on risk-taking behavior. The estimates of the value of a statistical life make such a connection by using the value that the workers themselves place on risks of death.

The benefit valuation amounts per expected fatality prevented that I suggested to OSHA were quite substantial, far in excess of the figures OSHA was using to value reductions in mortality risks. So the net effect of adopting my numbers would be to make lives considerably more valuable than the agency’s practice at that time, a consequence that hardly seemed “immoral.” I was also personally disappointed in their negative reaction, having just published the book edition of my doctoral dissertation on job safety,¹ as well as related articles advocating the value of a statistical life approach. But in addition to my personal stake in the methodology, I viewed valuing risks of death based on how workers themselves valued these risks as the only correct economic procedure for producing sensible policy valuations.

The idea of conceptualizing the valuation task in terms of statistical lives had a firm economic basis and was due to Nobel laureate Thomas Schelling.² However, given the state of empirical methods at the time of his original analysis, he was not optimistic about the prospects of assessing this value based on either surveys or empirical work: “The main problem is that people have difficulty knowing what it is worth to themselves, cannot easily answer questions about it, and may object to being asked. Market evidence is unlikely to reveal much.” Fortunately, advances over the past half-century in available data, statistical methods, and survey procedures have enabled us to develop meaningful estimates of the value of a statistical life.

At the time of my original suggestion that OSHA adopt this methodology, I was on leave from my position at the Northwestern University economics department to the government, where I served as the Deputy Director of the President’s Council on Wage and Price Stability. This agency managed the pay-price guidelines and was responsible for oversight of all proposed federal regulations for the Carter administration. Under the Carter administration’s Executive Order No. 12044, agencies were required to assess the benefits and costs of all major rules and to show that the chosen regulatory approach was the most cost-effective alternative. Although agencies were required to analyze the benefits and costs of policies and ideally should monetize these benefits and costs, the executive order did not require agencies to demonstrate formally that the benefits of the proposed regulations exceeded the costs. That requirement arrived in 1981 with the Reagan administration, which moved the regulatory oversight staff to the US Office of Management and Budget (OMB), Office of Information and Regulatory Affairs. Nevertheless, even without a formal benefit-cost test, there were political pressures within the Carter administration to strike a sensible balance between calculated benefits and costs, in part because of the perceived potential inflationary impact of costly but ineffective regulations.

The shift of the regulatory oversight group to OMB by President Reagan in 1981 was accompanied by other policy changes. As part of its economic reform agenda, the Reagan administration had ramped up the emphasis on bringing regulatory costs under control. The Carter administration had initiated a variety of deregulation initiatives with respect to airlines, trucking, and banking. Although the economic rationales for risk and environmental regulation are quite different than for economic regulations, as is the potential justification for deregulation as well, the Reagan administration sought to extend the deregulation concept to these newer social regulations. As part of this effort, the Reagan administration’s Executive Order No. 12291 imposed a benefit-cost test for proposed major regulations. Agencies must show that the benefits exceeded the costs before being permitted to issue the regulation. This benefit-cost analysis requirement has remained in place through all subsequent administrations. The Reagan administration also strengthened the regulatory oversight process. The requirement that agencies obtain OMB approval before launching major regulatory initiatives replaced the advisory White House reviews under Carter. The Reagan administration also targeted many regulations for elimination, such as a host of recent safety regulations pertaining to the auto industry.

The Triumph of the Value of a Statistical Life Approach:
The Hazard Communication Policy Debate

In this era of deregulation, agencies nevertheless continued to develop new regulatory proposals.³ The most expensive major new initiative proposed in the early years of President Reagan’s first term was the OSHA hazard communication regulation, which OSHA proposed in 1982. If this regulation was enacted, for the first time there would be regulatory requirements that firms label dangerous chemicals used in the workplace. Since these chemicals are often considerably more potent than household chemicals, the absence of any such labeling regulation more than a decade after OSHA’s establishment was surprising. After having read these proposed chemical labels, workers would be aware of their chemical risk exposures and be able to take appropriate precautions, possibly including the decision to quit and seek safer employment. In addition, the regulation would require that firms maintain material safety data sheets so that if workers were exposed to a dangerous chemical, the medical personnel would be aware of the consequences of the exposure and know how to treat the worker.

OSHA’s regulatory impact analysis for the hazard communication standards tallied the prospective costs and expected improvements in worker health and attached a dollar benefit to these health effects. The dominant benefit component for this regulation, as it is for most other health, safety, and environmental regulations, was the value of the mortality risk reduction. However, instead of using the value of a statistical life to monetize these effects, at that time OSHA and other agencies used the value of medical costs and lost earnings, or what they termed the “cost of death.” After completing the analysis based on the costs of the deaths prevented by the regulation, OSHA submitted the proposed regulation to the OMB regulatory oversight group for the approval that was required before the regulation could be issued. Although OSHA’s evaluation concluded that the regulation was desirable on balance, its economic assessment was flawed in several respects. Based on the critique of the regulatory impact analysis by the OMB economists, if OSHA had done a proper analysis, the result would have been that costs exceeded the benefits so that the regulation failed a benefit-cost test. As a result, OMB rejected OSHA’s regulatory proposal.

OSHA nevertheless wished to pursue the possibility of issuing the regulation. The procedure that the Reagan administration had established for agency appeals was that in the event of a dispute, the regulatory agency could appeal the decision to then Vice President George H. W. Bush. The vice president characterized the disagreement as a technical economics dispute and suggested that an outside expert assess the merits of the competing arguments. I was asked to resolve the dispute after being approved by the Secretary of Labor and OMB. By that time, I had left the government and was then at Duke University, where my research continued to focus on risk regulation issues. The OMB regulatory oversight staff, most of whom were my former colleagues at the Council on Wage and Price Stability, raised a host of criticisms of OSHA’s benefit estimates. All of these critiques were well founded. The result of adopting the OMB corrections to the analysis was that the calculated costs of the regulation exceeded the estimated benefits in terms of improved worker safety. In my assessment of the competing agency arguments, I accepted all of OMB’s critiques as being sound.

Where my approach differed from that taken by both OSHA and OMB was with respect to how the expected lives saved would be valued. Both OSHA and OMB valued the lives saved based on what OSHA termed the “cost of death,” or the present value of the medical costs and lost earnings that would be saved by preventing workers from being killed by chemical exposures. Under this approach, lives would have a value of several hundred thousand dollars, which was not a trivial amount, but was an arbitrary accounting measure that bore little relationship to how workers themselves valued risks of death. The approach I suggested utilized my labor market estimates of how much compensation workers required to face small risks of death. My estimate of the value of a statistical life at that time was $3 million per expected fatality prevented, or about $7.4 million adjusted for inflation. More recent estimates of the value of a statistical life generally place its value at between $9 million and $11 million. Using this estimate in the regulatory benefits analysis instead of the cost-of-death approach boosted benefits by an order of magnitude. Following a similar approach, I also attached values to the prevention of nonfatal worker injuries, but the driving force in the benefit assessment was the value of the fatalities prevented by the proposed regulation. The result of abandoning the cost-of-death approach was that benefits now exceeded the costs so that the hazard communication regulation would now pass OMB’s economic test. President Reagan approved the regulation almost immediately after my report in support of the regulation reached the White House.

The Genesis of Estimates of the Value of a Statistical Life

Where did this $3 million figure come from? The average annual worker fatality risk at that time was 1/10,000, which is more than double the current level of dangerousness. In return for bearing this risk, workers received an annual wage premium of $300, where this amount was estimated statistically controlling for other aspects of the job and worker characteristics. The result is that for a group of 10,000 workers, on average one of them would be killed on the job in the coming year. The amount of compensation that this group of 10,000 workers would receive for the one expected death is 10,000 × $300, or $3 million. Thus, the value of a statistical life is simply the total amount of compensation required per expected workplace death. The value of a statistical life reflects the values that the workers themselves believe that bearing these risks is worth rather than an accounting measure or an arbitrary number assigned by a government analyst.

But why should any finite value be applied to the expected lives saved? One could treat each expected life saved as having an infinite value. In that case, it would be desirable to expend the entire federal budget on safety measures that would eliminate a small chance of even one expected death. Given the multiple risks that we face and the limits on our financial resources, this uncompromising approach is infeasible. Consider data for 2014, the most recent year for which comprehensive accident data are available. There were 136,053 accidental deaths in the United States in 2014.⁴ If the entire gross domestic product of $17.4 trillion in 2014 were allocated to preventing accidents, it would only be possible to spend an average of $128 million per death to prevent these accidents, leaving nothing left to prevent illnesses or to provide for daily living expenses.

To motivate the reasonableness of using workers’ wage-risk tradeoffs as the guide, it is useful to ask people to conceptualize scenarios in which the reader makes similar decisions that do not reflect an unbounded commitment to safety, whether it involves living in a riskier but less expensive neighborhood or driving a car that doesn’t have all possible safety enhancements. Most transportation choices and dietary decisions entail at least some risk. We do not plan our lives to minimize all possible risks. Through daily risk-taking decisions, people reveal that they place a finite value on reduced risks to their life. People’s unwillingness to display an unbounded commitment to safety is consistent with a myriad of other risk-taking decisions that we make. The labor market estimates undertake a similar comparison in which the tradeoff is based on the extra wages that workers are paid for the additional risks posed by their jobs.

Critiques of My Initial Estimates
of the Value of a Statistical Life

My estimates of the value of a statistical life on the order of $3 million came under attack from both extremes. Some critiques suggested that the numbers were too big, while others thought they were too small. The critiques suggesting that they were too small involved appeals to value lives at an infinite amount and were not grounded on any actual empirical estimates. Those suggesting that my estimates were too large were more empirically based. My figure exceeded the present value of lifetime earnings, so how could it be feasible for people to value their lives so highly? Surely people could not afford to pay more for their lives than their resources permitted. The raw numbers presumably suggested that the valuations were excessive. However, what workers are valuing is not the certainty of death but rather a very small risk of death of 1/10,000 annually. It should also be noted that workers are being compensated for these small risks. They are not paying for greater safety so the budgetary constraint issues are less salient. Moreover, for small risks of death, budgetary issues are not influential whether we are talking about the amount that workers are paid to incur a risk or the amount that they would be willing to pay to reduce their level of risk. The amounts that workers require to face slightly greater risks will equal the amount that they are willing to pay to make their jobs safer to that same degree. Even if workers were paying for the risk reduction, it is not unrealistic to assume that workers would be willing to pay $300 to reduce their annual fatality risk by 1/10,000. This is a quite different matter than assuming that workers have $3 million to buy out of the risk of certain death.

A more sophisticated critique was based on the economics literature at that time. Some other competing estimates of the value of a statistical life pegged the value at well below $1 million. Didn’t that lower value imply that my numbers were too high? The disparity in the estimates also gave the false impression that estimates of the value of a statistical life differed widely and were too unreliable to be used for policy analysis. However, the observed differences were quite plausible. Whereas workers in my studies faced risks that were comparable to the US average of 1/10,000 annually, the estimates of under $1 million were based on workers in very high-risk jobs with annual fatality rates of 1/1,000 per year.⁵ These divergent estimates are not incompatible. The value of a statistical life is not a universal constant. Rather, it reflects the average rate of tradeoff between wages and risk for particular samples of workers. Those who place a comparatively low value on risk to their lives will tend to gravitate to higher-risk jobs and reveal through their choices a lower wage requirement per unit of risk, which will imply a lower value of a statistical life.

Notwithstanding such critiques and the political sensitivity of the task of valuing risks to life, other agencies also adopted the value of a statistical life, or the VSL. While it may be the case that agencies were swayed by the compelling economic logic of the VSL approach, it is also likely that the fact that the VSL enabled agencies to boost the calculated economic benefit estimates by a factor of ten contributes to its attractiveness. The mortality risk reduction benefit valuation approach using the VSL is now the standard practice throughout the US government as well as in many other countries. The largest benefit component of all US federal regulations is the monetized value of the statistical lives that will be saved by the regulation.⁶ The VSL has become the most important parameter driving the attractiveness of government regulations generally as it plays the central role in the evaluation of health, safety, and environmental regulations.⁷

Why Monetizing the Effects Matters

The monetization of the reduced mortality risks through application of the value of a statistical life is instrumental in the assessment of the benefits by the government, as it enables these effects to be put in the same terms as other economic impacts. The benefit-cost analysis procedure that lies at the heart of regulatory analyses involves a comparison of the benefits and the costs and a judgment that the benefits exceed the costs. To make such a comparison, at some point all effects must be put in comparable units, at least implicitly. Cost figures are dollar amounts that appear to be real economic consequences. Indeed, the regulatory oversight efforts in the Ford and Carter administrations were motivated primarily by a concern with the economic burdens arising from the inflationary effects of regulatory costs, not a concern with b...