Did China's Poor-County Designation Move Counties Across a Development Threshold?

Abstract

Area-based antipoverty policy rests on the premise that a transfer large enough can change a poor place's trajectory and not merely its level. This proposal tests that premise using China's national poor-county programme, which still covered 832 counties when it was abolished in 2020. Existing evaluations measure its effects while transfers were still flowing, and so cannot say whether the programme temporarily enlarged the capital stock of economies that will return to their original path or pushed them past an unstable threshold in an S-shaped accumulation function; the two separate only once support ends. Identification therefore comes from exit rather than entry: counties graduated in cohorts between 2016 and 2020, and outcomes are analysed in event time around each county's own graduation year. Three hypotheses—continued divergence after graduation (threshold crossed), a gap that decays toward zero (level effect only), and no gap at any horizon—are tested with a staggered event study using the Callaway and Sant'Anna (2021) estimator, a border discontinuity across adjacent designated and non-designated counties, and a secondary regression discontinuity at the 2012 income cutoff. Development is measured by two outcomes with different error sources: nighttime luminosity from DMSP-OLS and VIIRS, produced entirely outside the Chinese administrative system, and reported county income per capita, which enters as an outcome read against luminosity and never as a control, so that agreement is informative and divergence around certification is diagnostic of reporting behaviour. Persistence after graduation is read as a threshold only if it survives checks for successor support, migration, spillovers and endogenous exit timing, and is accompanied by non-linearity in the cumulative push and bimodality across counties. The answer determines whether a one-off push can permanently change a place, or whether area- based support must be permanent to have a permanent effect.