The Age-Adjusted Value of a Statistical Life: Evidence from Vehicle Choice

In this paper I estimate an age-adjusted value of statistical life (VSL) by examining consumer automobile purchases. Many public policies target the health and safety of older people, but most VSL estimates come from hedonic wage regressions with few older workers and no retirees. Although underrepresented in the labor force, senior citizens account for a sizeable share of vehicle acquisitions (around 12 percent). Using automobile purchase decisions I am able to extend the range of revealed preference evidence to include individuals up to 85 years old. Combining information on vehicle holdings and use, household attributes, used vehicle prices, and yearly fatal accidents for each make, model, and vintage automobile, I use a multinomial choice framework to model the vehicle purchase decision. For each age group I estimate a separate willingness to pay for reduced mortality by comparing the relative importance of fatality risk and annualized cost. I also employ a control function approach using vehicle crash test results to account for potential bias from individuals sorting into certain vehicle types. On average I find that seniors and non-seniors display similar VSLs, but that these averages mask a significant inverted-U-shaped VSL function that ranges from $1.6 to $17.1 million. Overall the results of this study corroborate the growing consensus on the shape and scale of the age-VSL relationship and highlight the importance of considering the specific age of affected individuals when evaluating the benefits of reduced mortality risk. Abstract word count: 237 JEL classifications: J17, K32, I12