Values of a Statistical Life Among the Medicare Population

We develop a revealed-preference approach to estimating the values of statistical life (VSL) among the US Medicare population. We build a life cycle framework in which retirees choose between medical expenditure and other consumption while facing uncertainty about their health and survival. We estimate the model using data from nearly 40,000 person-years of Medicare claims and Medicare Current Beneficiary Survey data. These are individual-level panel data that provide detailed information on total and out-of-pocket medical spending, self-reported health, limitations to activities of daily living, smoking, education, income, demographics, the presence of many chronic illnesses, geographic location, and the timing of death. Using geographic variations in Medicare spending identified by those who move between regions to instrument for medical spending, we estimate that each additional $1,000 in medical spending reduces annual mortality by 0.4 percentage points. Using this result we find that the average VSL for a 66-year old with no observed chronic illnesses is estimated to be $1.1 million, while it is $510,000 for a 66-year old of average health. Age-specific VSL measures are systematically lower for people diagnosed with major chronic illnesses; for those with limitations on activities of daily living; and for those with lower levels of self-reported health. Further, females and non-smokers have significantly higher conditional VSL. Overall, our VSL estimates are an order of magnitude smaller than estimates derived from hedonic wage models from younger adults in the labor market. This difference is due to an inverted-U VSL life cycle profile in which the VSL declines with age among the Medicare population. We use these estimates to assess the value of marginal increases in Medicare spending and air quality. Our results imply that the value of fatalities avoided due to the Clean Air Act Amendment was $400 billion for 2010 compared to the $1.4 trillion reported by the Environmental Protection Agency. At the same time, our findings point toward the importance of understanding how complementarity between the quantity and quality of life affects the way that the VSL is used in policy evaluations. The standard approach to valuing mortality and morbidity reductions attributed to regulating air pollution can be improved by modeling how reductions in morbidities increase welfare endogenously through the VSL.