LAUSR

Abstract Past impact evaluations of China's largest ecological restoration program have assumed the absence of self-selection (endogeneity) in the likelihood and extent of participation. Using appropriate testing procedures and a panel dataset of > 1000 households over 11 years in two primary provinces, we found evidence of self-selection in household behavior of generating off-farm income. But the hypothesis was rejected that there was a significant self-selection component in households' decision to participate in the program and generate farming income. Evaluations ignoring the self-section for off-farm labor were found to be biased and overly positive on program income impact. Self-selection should thus be explicitly included, unless there is counter evidence, in any study of this kind.