Alternative approaches to evaluation in empirical microeconomics

This paper reviews some of the most popular policy evaluation methods in empirical microeconomics: social experiments, natural experiments, matching, instrumental variables, discontinuity design, and control functions. It discusses identification of traditionally used average parameters and more complex distributional parameters. The adequacy, assumptions, and data requirements of each approach are discussed drawing on empirical evidence from the education and employment policy evaluation literature. A workhorse simulation model of education returns is used throughout the paper to discuss and illustrate each approach. The full set of STATA datasets and do-files are available free online and can be used to reproduce all estimation and simulation results.