Working Paper: NBER ID: w14364
Authors: Thomas Deleire; Christopher Timmins
Abstract: Wage-hedonics is used to recover the value of a statistical life by exploiting the fact that workers choosing riskier occupations will be compensated with a higher wage. However, Roy (1951) suggests that observed wage distributions will be distorted if individuals select into jobs according to idiosyncratic returns. We illustrate how this type of sorting may bias wage-hedonic VSL estimates and describe two estimators that correct for it. Using data from the CPS, we recover VSL estimates that are three to four times larger than those based on the traditional techniques, statistically significant, and robust to a wide array of specifications.
Keywords: Valuation of a Statistical Life; Wage Hedonics; Roy Model; Worker Sorting
JEL Codes: J17; J31; Q5; Q51
Edges that are evidenced by causal inference methods are in orange, and the rest are in light blue.
| Cause | Effect |
|---|---|
| worker sorting based on idiosyncratic returns (J29) | biased VSL estimates (J17) |
| traditional wage-hedonic methods (J33) | biased VSL estimates (J17) |
| worker sorting based on non-pecuniary attributes (J29) | biased VSL estimates (J17) |
| individuals with higher wage potential in safer jobs (J28) | opt for riskier occupations (J28) |
| opt for riskier occupations (J28) | underestimation of compensating wage differential associated with fatality risk (J17) |
| corrected VSL estimates (J17) | greater willingness to pay for reductions in fatality risk (J17) |
| corrected VSL estimates (J17) | less sensitivity to specification changes (L15) |
| corrected VSL estimates (J17) | strong relationship with individual characteristics (C91) |