• I20;
  • J24;
  • J31
  • College quality;
  • wages;
  • selection on observables

Abstract. We estimate the effects of the quality of the college a student attends on their later earnings using data from a cohort of US college students from the late 1970s and early 1980s. We rely on a linear selection on observables identification strategy, which is justified in our context by a very rich set of conditioning variables. We find economically important earnings effects of college quality for men and women, as well as effects on educational attainment, spousal earnings and other demographic variables. These effects remain roughly constant over time and result primarily from effects on wages, rather than from effects on hours or labor force participation. We find that, over the lower part of the range of college quality, increases in college quality (which entail higher expenditures per student) pass a simple social cost–benefit test.