Abstract：The paper uses the propensity score-matching method to estimate the effects on training on rural labor's earnings based on micro data of Jilin Province in 2006.The estimating results show that both pre-job and on-the-job training have positive effects on rural labors, and the former is higher than the latter.Cross-sectional estimates and OLS estimates also demonstrate the conclusion. But comparing between treatment group and control group in different methods, we find that individual characters are closer between them in propensity score-matching method than others.That is, the estimates of propensity score-matching method are closer to that of random experiments.

Keywords: training; earnings effects; non-experimental data; propensity score-matching method