The Full Returns to the Choice of Occupation and Education

Measuring Job Risks When Hedonic Wage Models Do Not Do the Job

“…Hundreds of hedonic wage studies and dozens of metanalyses (including a metanalysis of metanalyses by Banzhaf 2022), report a positive correlation between on-the-job mortality risks and wages. The VSLs estimated in individual studies are however, heterogeneous, and despite a publication bias that makes small positive and negative VSL values less likely to be reported (Ashenfelter andGreenstone 2004, Doucouliagos et al 2012), there is mounting evidence that lower wage workers tend to experience worse workplace conditions and greater injury rates than would be expected if compensating differentials held (Dorman and Hägstrom 1998, Böckerman and Ilmakunnas 2006, Bonhomme and Jolivet 2009, Maestas et al 2018, Park et al 2021, Dorman and Boden 2021, Clark et al 2022.…”

“…Hedonic wage regression models rely on strong assumptions of perfectly competitive labor markets, where workers are perfectly informed, rational, and can switch jobs freely. Even when these assumptions hold, as noted by Hamermesh (1999), to the extent that workplace safety is a normal good, increases in earnings inequality like the ones observed in the 1980s and 1990s (and that have continued in the last two decades, very markedly in the US), have led to increases in the inequality in the burden of job disamenities and of workplace injury risks in particular, so that wage-inequality understates the inequality in the total returns to work (Clark et al 2022).…”