We use data from the Longitudinal Employer‐Household Dynamics program to study the causal effects of location on earnings. A model with fixed effects for workers and for the interaction of industry and commuting zone (CZ) provides a good approximation to observed earnings changes as people move between CZs and/or industries, though it takes several quarters to adjust to new labor markets. Additional returns to previous work experience in the largest U.S. cities are small. We find some indication of match effects between places and industries, and of local agglomeration effects, but these are small and explain only a small fraction of CZ‐average wage differences. In contrast, worker skills, captured by the individual fixed effects in our model, vary widely between locations and explain about 2/3 of the observed earnings differences across CZs. Fitting separate models for college and non‐college workers we find very similar local wage premiums. The degree of assortative matching across CZs is much larger for college‐educated workers, however, leading to a positive correlation between CZ size and the college‐high school wage gap that is driven by sorting within the college workforce. Differences in local housing costs more than offset the corresponding earnings premiums, suggesting that workers who move to larger CZs have lower net‐of‐housing consumption.

Attendance by invitation only.