Identifying the Effect of Ability and Schooling on Wages:Going beyond the NLSY

This paper estimates returns to education using US data. Using the NLS and NLSY79 (dataset) average wages for workers with different ability and educational levels can be estimated. Because of the high correlation between schooling and ability it is not possible to estimate across the entire ability-schooling support. The PUMS dataset (which includes wage and education data, but excludes ability) from the U.S. Bureau of the Census contains information that can be exploited to improve the precision of the NLSY79 estimates. The source of the improved precision is the non-parametric bounding technique described in Cross and Manski (2002). By incorporating the PUMS dataset, estimated returns to education at different ability levels are substantially sharpened. Results show a positive wage gap that does not increase over time for the most able during the 80’s, and between 1980 and 2000