本文根据2008年云南省红河州的农户调查数据,在考虑个体异质性和村庄异质性的基础上,运用多水平框架下的异质性工具变量模型估计劳动力培训的效应,得出以下结论,可观测因素对农户收入和培训决策的影响随着村庄的不同而变化,农户劳动力培训的效应不仅受个体异质性的影响而且受村庄异质性的影响;目前的培训政策发挥了政策引导作用,使参加培训的农户多数是参与意愿高的农户;劳动力培训能够提高农户的收入水平,但效果并不显著,农户从培训中获得的收益仅为2.5%,说明目前的培训效果有待加强。 Based on the survey data of farmer households in Honghe Prefecture of Yunnan Province in 2008, based on the heterogeneity of individual and village heterogeneity, this paper uses the heterogeneous instrumental variable model under multi-level framework to estimate the effect of labor training and draws the following conclusions , The impact of observable factors on farmer's income and training decisions varies with the village. The effect of farmer labor training is not only affected by individual heterogeneity but also affected by village heterogeneity. The current training policy has played a policy-oriented role The result is that most of the farmers participating in the training are farmers with high willingness to participate. Labor training can improve the income level of farmers, but the effect is not significant. The income gained from training is only 2.5%. This shows that the current training effect needs to be strengthened.