A COMPANION TO Theoretical Econometrics
Self-Selection
This paper provides some account on econometric models and analysis of sample selection problems. The paper is divided into three parts. The first part considers possible selection-bias issues in econometric data. Selection biases can occur as the observed outcomes are results of individual's self-selection. The second part points out some development on econometric models with sample selection. This part concentrates on the specification, estimation, and test problems for parametric models. The third part lists some of the development of semiparametric estimation of sample selection models. Related surveys on the earlier development on sample selection models are Maddala (1983) and Amemiya (1984), which concern mainly parametric specification and estimation. Recent developments of the subject concern semiparametric estimation methods. Surveys on the latter are Powell (1994), Vella (1998), and M.-J. Lee (1997). Powell's survey covers broad areas of semiparametric estimation methods for microeconometric models in addition to sample selection models. A large part of the survey in Vella (1998) concerns the recent development on sample selection panel models. Treatment effect models as compared with sample selection models have been discussed in M.-J. Lee (1997). Because of their coverage and many developments on panel and treatment effect models being currently in their development stages in working paper format and because of space limitation, we skip these topics in this survey.
