“…Other reasons for why VSL and VSI estimates differ so widely across studies include the role of unobservables and sorting into occupations (Hamermesh and Wolfe, ; Black and Kniesner, ; Lalive, ; Leeth and Ruser, ; Ashenfelter, ; Kniesner et al., ; Kochi and Taylor, ; Bommier and Villeneuve, ; DeLeire et al., ; Lavetti and Schmutte, ; Lavetti, ). For example, Shogren and Stamland () argue that VSL estimates would be likely upward biased if they did not account for worker heterogeneity in both risk preferences and skill at reducing risk.…”