Doctoral Dissertation Research in DRMS: A Comparison of Value of Statistical Life Estimates Derived from Revealed and Stated Preferences

This award is funded in whole or in part under the American Rescue Plan Act of 2021 (Public Law 117-2).

Every day, people consider tradeoffs they are willing to make to reduce risks. The Value of a Statistical Life (VSL) describes the tradeoffs in aggregate. The VSL is used to evaluate safety regulations across many areas of life.

Accurate measurement is critical, but VSL estimates vary widely. This dissertation advances the national health, prosperity, and welfare by investigating three sources of variability in the VSL. First, it compares the methods of discovering people’s risk preferences.

This is important because preferences are a key input into the VSL metric. Second, it examines whether people’s risk preferences differ across contexts when the probability of death does not. This is important because VSL estimates derived from labor market data are used to guide regulation in unrelated areas.

Finally, it explores people’s subjective experiences of objective risks. This is important because use of the VSL requires that people accurately comprehend risk. The project contributes to the advancement of the decision sciences and economics by testing prevailing assumptions about the nature and measurement of risk preferences.

Findings promote a more complete theory of how perceptions of risk guide decision making. Findings may also reshape how regulatory agencies approach cost-benefit analyses of safety regulations. This is of particular importance in promoting the health, prosperity, and welfare of vulnerable populations, who are more susceptible to fatality risks and health hazards.

The Value of a Statistical Life (VSL) captures the trade-offs people are willing to make to reduce the probability of death. The VSL is used widely in safety regulation. Questions remain, however, regarding how well the VSL captures risk preferences.

First, VSL estimates vary considerably by measurement approach (revealed vs. stated preferences), reflecting different assumptions about how people evaluate risk. Second, VSL estimates from labor contexts are leveraged for non-labor regulations, which may not be justifiable if people value risk mitigation differently across contexts. Finally, use of the VSL relies on the assumption that people accurately comprehend risk; this ignores differences in subjective experiences of risk.

To address the first two issues, subjects complete a revealed preferences (RP) survey about the labor market and three discrete choice experiments (DCE) eliciting stated preferences (SP) to reduce risks in other contexts. To assess the VSL’s criterion validity, RP responses are predicted from the labor SP DCE. To test the stability of preferences across contexts, comparisons of people’s wealth/risk trade-offs across the DCEs are examined with machine learning techniques.

To address the third question, baseline risk preferences for each subject and their perceptions of risk in their jobs, and in each DCE are measured and compared. The data are used to develop and test a model of risk preferences in contexts significant to VSL estimation. The work advances understanding of how decision making is shaped by context and how subjectivity in experiencing risk affects the VSL over and above the objective risky features of one’s environment.

Ultimately these findings have the potential to inform how governments allocate public resources to increase safety.

This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.