Drawing on administrative data from the Social Security Administration, we find that individuals that go through a long period of non-employment suffer large and long-term earnings losses (around 35-40 percent) compared to individuals with similar age and previous earnings histories. Importantly, these differences depend on past earnings, and are largest at the bottom and top of the earnings distribution. Focusing on workers that are employed 10 years after a period of long-term non-employment, we find much smaller earnings losses (8-10 percent). Furthermore, the large earnings losses of low-income individuals are almost entirely due to employment effects.

We investigate the importance of job-to-job (JJ) transitions for cyclical wage dynamics. By exploiting cross-state variation, we find that wage growth is tightly linked to variation in the JJ transition probability, and conditional on this, the job finding probability of the unemployed has no explanatory power. We investigate the robustness of our results to several caveats and find the result to hold. Finally, we discuss the implications of our findings for competing theories of wage dynamics.