Match Effects

We present an empirical model of earnings that controls for observable and unobservable characteristics of workers (person effects), unmeasured characteristics of their employers (firm effects), and unmeasured characteristics of worker-firm matches (match effects). The distinction between these components is important, because they have different implications for the persistence of individual earnings and the returns to employment mobility. We find that match effects, which have been ignored in previous work, are an important determinant of log earnings. They explain about 16 percent of observed variation, and much of the change in earnings when workers change employer. Specifications that omit match effects over-estimate the returns to experience by as much as 30 percent, attribute too much variation to person effects and little to firm effects, and underestimate the correlation between person and firm effects. Overall, our results suggest that some of the returns previously attributed to general human capital actually reflect the returns to sorting into higher-paying firms and better worker-firm matches.