Abstract: This paper is a theoretical and empirical study to compare different policy evaluation models. Using sloping land conversion program (SLCP) in China as an example, the authors apply a matching regression model, a DID model and a matching DID model to estimate policy impact on households’ incomes. The authors also explain the bias reduction from the approach of combining DID and matching strategies and proves the matching DID model a more efficient regression method. The matching DID is also useful to be adopted to evaluate policies in other fields.

Key words: matching regression, difference in difference model, matching DID model, slopping land conversion program (SLCP)

摘要： 通过理论和实证研究，比较几类政策评估计量经济模型的回归效果。采用匹配法、倍差法和匹配倍差法3种模型对退耕还林政策对农户收入影响进行案例研究。结果表明，匹配倍差模型显著修正了估计偏差，提高了评估精确性。该模型对国内其他经济政策效果评估方法的改进亦有借鉴意义。

关键词: 匹配回归, 倍差回归, 匹配倍差回归, 退耕还林政策