By the widely used difference-in-difference method, the Southwest
China Poverty Reduction Project had little impact on the proportion of people in
beneficiary villages consuming less than $1/day — despite a public outlay of
$400 million. Is that right, or is the true impact being hidden somehow? We find
that impact estimates are quite sensitive to the choice of outcome indicator, the
poverty line and the matching method. There are larger poverty impacts at lower
poverty lines. And there are much larger impacts on incomes than consumptions.
Uncertainty about the project’s impact probably made it hard for participants to
infer the gain in permanent income, so they saved a high proportion of the shortterm
gain.