China has long fought geographic poverty traps. These arise when poor geographic endowments–lousy roads, poor schools and health care, remoteness–constrain prospects of escaping poverty even when a household’s own endowments are adequate. Comparing two equally poor households, one living in a poor area one not, the latter sees faster progress over time.

In a paper with Jyotsna Jalan, “Geographic Poverty Traps,” we found that this was indeed a serious problem in rural China. We used panel data to model the microeconomics of the growth process, allowing for both individual and geographic initial conditions. Yes, there were clear signs of the existence of geographic poverty traps in the region of rural Southwest China that we studied.

How can policy best respond? Should the government try to move capital into poor areas or labor out? Is lack of capital the problem, or a low productivity of capital? These are difficult questions, but careful study of poor-area development programs offers some clues. And China has relied heavily on these programs. As an evaluation problem the challenge is to deal with the fact that poor-area programs are deliberately placed in areas with poor initial conditions.

In a subsequent paper to the work on geographic poverty traps, “Lasting impacts of aid to poor areas,” Shaohua Chen, Ren Mu and myself studied the a major initiative by the Government of China, with substantial financial and technical support from the World Bank:. the Southwest China Poverty Reduction Project. This tried to permanently break the geographic poverty traps in Southwest China. We studied the impacts the project over a 10 years period, including four years after the World Bank’s disbursements finished. (This is a much longer period than most impact evaluations.)

Alas, we did not find much overall long-term impact of this major poor-area development project. The project’s aim was to eliminate poverty in the targeted villages. But the impact fell far short of that aim when judged against the progress of the comparison villages over this 10 year period.

There is a policy lesson from this disappointing performance. An important reason lies in the local political economy of these policies, related to efficiency-equity trade offs that are faced locally. In a nutshell, success requires a combination of both better external funding (including for local infrastructure) and success at assuring that the immediate direct beneficiaries of the extra resources made available have the human capital to take up the new opportunities.

Alas, given the deep-rooted inequalities in human capital, giving everyone an equal share can greatly dilute the impact. The way to assure greater impact is to target those among the poor who have the education needed to benefit from the injection of extra capital. This will assure greater impact on poverty, but it can seem inequitable at local level. Over the longer term, eliminating the inequalities in human capital will eliminate this troubling trade-off.