This paper examines one type of assortative matching in college markets: students with high socioeconomic status (SES) are more likely to attend high quality colleges. Assortativity matters if SES and college quality are complementary educational inputs. I develop an econometric framework that provides tests for the existence and sign of this complementarity. I implement these tests by exploiting a 2000 reform of the national college admission exam in Colombia, which caused a market-wide reduction in assortative matching in some regions of the country. I find that the reform lowered average graduation rates and post-college earnings in affected regions, consistent with a positive complementarity between SES and college quality. I also find evidence of mismatch: part of these negative effects came from the low SES students who were shifted into higher quality colleges. However, both the market-wide and mismatch effects die out several cohorts after the exam reform, which suggests that complementarity may evolve with large-scale changes in assortativity.