The Wear and Tear on Health: What is the Role of Occupation?

Average health and life expectancy differ substantially across occupational groups and while it seems evident that occupations affect health, effect estimates are scarce. In this study, we estimate a fixed effects model that controls for lagged health in order to estimate the effect of exposure to occupational stressors in the previous year on current health. From a theoretical model of occupation and health over the life cycle, we derive health-related selection mechanisms into occupation, show how our econometric estimators relate to the structural parameters, and explicitly formulate conditions under which our estimates have a causal interpretation. We argue that, in the absence of exogenous variation in the regressor(s) of interest, but with panel data spanning 26 years, our model offers the most trustworthy causal estimates. Our alternative formulation of the Grossman model illustrates its implicit assumptions on the decaying effect of past shocks and health investment on current health. These insights are informative for anyone trying to provide theoretical foundations for an econometric dynamic panel data model.

We argue that manual occupations are not only more physically demanding, but are simultaneously often characterized by low psychosocial workload. In previous research, most existing studies have characterized occupation with a binary indicator of manual versus nonmanual occupation, or have focused only on the manual aspects of occupation. This made it impossible to disentangle the contributions of the different ergonomic and psychosocial stressors, which meant that the resulting policy implications were unclear. In the present study, we link detailed Finnish data on occupational stressors to individual-level German longitudinal data. This provides a level of detail that was not available in earlier studies. The US DOT, for example, lacks information on psychosocial workplace conditions and exclusively includes information on physical requirements and environmental conditions.

Our findings suggest that about 50 percent of the association between physical demands at work and self-reported health is due to the causal effect of physical demands. Selection accounts for the remaining 50 percent. The average immediate effect of a one standard deviation increase in the degree of manually handling heavy burdens (e.g. from a wholesale worker to a plumber or from a mail sorter to a bricklayer) is comparable to the effect of aging five months and the effect increases with age. A lower degree of control over daily activities at work (e.g. kitchen assistant instead of cook or nurse instead of physiotherapist) is harmful to health at older, but not at younger ages. Under the assumption that the coefficient of lagged health captures the decay rate of past choices and shocks, we estimate that exposure to a one standard deviation increase in handling heavy burdens between age 60 to 64 leads to a health deterioration that is comparable to aging 16 months. The estimated effect of exposure to low job control between age 60 to 64 is comparable to aging 23 months.