Estimating the education-earnings equation using geographic variation

•We estimate the causal impact of postsecondary education on earnings.•We use a richer set of location-based measures as instruments for years of education.•Alternative measures give differing estimates of the impact of college on earnings.•Results suggest 9.7% increase in yearly earnings for each additional year of college.•Larger impact found for women with no measurable impact found for men.

We expand on the literature on the causal impact of postsecondary education on earnings by introducing a richer set of location-based measures as instruments for years of education. Utilizing data from the National Longitudinal Study of Youth, 1997, we implement six different sets of instruments based on geographic variation: presence of a four-year or two-year college in the county, inverse log distance to in-state two-year colleges, distance-weighted tuition and distance-weighted enrollment at in-state two-year colleges, and inverse log distance to all colleges. We find that these alternative measures yield differing estimates of the impact of educational attainment on earnings. Using our preferred measure of geographic variation, one additional year of postsecondary attainment results in a 9.7% increase in yearly earnings. We find a larger impact of postsecondary attainment for women, and no measurable impact of postsecondary attainment for men.