摘要：This paper presents a new generalized maximum entropy (GME) approach to estimation of sample-selection models with small data sets, such as are found in many empirical agricultural economic analysis. For small samples, the GME approach produces more stable estimates and has smaller mean square error measures than other well-known estimators such as ordinary least squares, Heckman's two-step method, full-information maximum likelihood, and Ahn and Powell's method. The technique is used to analyze whether hired agricultural workers will work in piece-rate or time-rate jobs and to compare female-male wage differentials for both types of jobs.
关键词：agricultural labor;agricultural wages;data analysis;entropy;monte carlo analysis;piece rates
