An Expressed Preference Determination of College Students' Valuation of Statistical Lives: Methods and Implications

Beginning of article

I wish to thank the USU Office of the Provost and the Department of Economics for providing financial support. I would also like to thank Cris Lewis and Paul Jakus for reading earlier drafts of the manuscript and providing invaluable suggestions. Lastly, I am most grateful to my wife for her unfailing support.

   Government agencies typically apply a general value of statistical
   life (VSL) estimate when performing cost-benefit analysis (CBA).
   However, theory suggests that college students attach a value to
   statistical lives that differs from society's VSL; therefore, CBA
   may lead to inefficient levels of risk reduction among students. A
   contingent valuation survey of 137 students was conducted to
   calculate Utah State University students' VSL; the sample's
   representativeness enables statistical inferencing. Regression
   models indicate that the students possessed an adequate
   understanding of mortality risk reduction. The students' inferred
   VSL of $2.96 million is significantly lower than the $5.4 million
   societal estimate computed by Kochi et al. (2006). This suggests
   that students are required to pay more for risk reduction than they
   are willing.

It is customary and often mandatory for government agencies to perform cost-benefit analysis (CBA) before initiating specific programs. Premature death prevention is a conspicuous benefit stemming from many government actions. For example, the CPSC's regulation of space heaters saves an estimated 63 lives annually, the EPA's management of asbestos saves about 10 lives, and OSHA's guidelines concerning hazard communications save around 200 lives (Morall, 2003). However, each of these programs has accompanying costs. Because mortality risk reduction benefits are subject to the law of decreasing marginal utility, there exists a specific level of supplied risk reduction such that benefits per dollar are maximized. Also, because risk reduction is often a non-rival and nonexclusive good, government intervention is frequently required to augment supply. Economists seek to calculate an efficient supply of mortality risk reduction through the determination of the value of a statistical life, or simply, VSL.

VSL is a statement of how much a given population is willing to pay to reduce the total amount of expected premature deaths by one.

To state that a population's VSL is equal to x is to assert that the population values statistical lives at x. This is different from stating that the value of statistical lives among the population is x; the actual amount may be higher or lower. It is important to note that just as individuals within free market determine the price of goods through sovereignty, consumers determine the value of a statistical life through implicitly or explicitly expressed risk-benefit tradeoffs.

U.S. government agencies typically apply a general VSL estimate when analyzing the costs and benefits of specific programs. However, the costs of government programs are frequently borne by homogenous subpopulations. The VSL of these subpopulations may be significantly higher or lower than the societal VSL. It has been shown that a given subpopulation's VSL is typically lower if the members of the subpopulation are younger--under the age of 30--and have a lower than average income (Viscusi, 1992). This suggests that college students' VSL is lower than society's. However, this is impossible to know without performing statistical tests because other factors in addition to age and income can influence willingness to pay (WTP) for mortality risk reduction.

For this article, an expressed preference contingent valuation study was conducted to calculate the VSL of students at Utah State University. Survey results were gathered from 137 students. Statistical inference is possible because (a) these students are shown to be representative of the general USU student body, and (b) the respondents demonstrate an adequate understanding of mortality risk reduction. Regression modeling reveals the students grasped small changes in probability. Also, the model's income and age coefficients are significant and meet a priori expectations. The students' VSL of $2.96 million is found to be significantly lower than the $5.4 million estimate of U.S. citizens' VSL computed by Kochi et al. (2006). Because this study uses a single WTP elicitation technique to compute the VSL of college students at one university only, the results do not prove that the VSL for students at all colleges is lower; nonetheless, it does provide evidence. Further research is encouraged in order to determine a definitive estimate of college students' VSL, thereby allowing the supply of an efficient level of mortality risk among students. The article's organization follows this general outline: literature review, implications and theory, survey methods, results and analysis, tests for disparity between the VSL of college students and society, and summary.

Literature Review

Before the advent of VSL, the value of life assumed for CBA was calculated using the present value of individuals' expected future earnings (Rice and Cooper, 1967; Mishan, 1971). This methodology is still used in court cases involving wrongful death (Bowles and Lewis, 2003). However, because it was found individuals were willing to pay more to reduce the risk of lost lives than the sum of expected future earnings, this method was determined to be inappropriate for CBA.

In a recent study published by the EPA, 90 percent of benefits derived from the Clean Air Act are quantified as saved statistical lives (U.S. Environmental Protection Agency, 1999). Because VSL plays such an important role in government planning, it is important to find the VSL that truly reflects society's WTP for risk reductions. Hundreds of studies have been preformed to estimate the societal mean WTP for reduced mortality risks (Viscusi, 1992; Miller, 1990; Kochi et al., 2006).

Because no structured risk reduction market exists (at least in the same sense as the markets for guns or butter), statistical techniques must be used to estimate WTP. There are two popular methods for estimating WTP: the revealed preference (RP) method and the expressed preference (EP) method (Young, 2005). Examples of RP estimations of WTP include hedonic wage-risk differential calculations (Viscusi, 1992) and demand estimates for risk-reducing goods (Miller, 1990). Speed limits have also been used to calculate VSL …