A new approach for detecting multiple-equilibria poverty traps

Despite a large literature, empirical evidence on multiple-equilibria poverty traps remains elusive. Current empirical methods estimate average welfare-dynamics, which masks inherent heterogeneity across households. We develop an alternative approach that allows welfare-dynamics to differ across the household wealth distribution. Drawing on the literature on regression discontinuity designs, we propose a new statistical test for detecting multiple-equilibria and poverty traps. We demonstrate the approach using panel data on over 6000 rural households in India. Our application contributes to the economic development literature by examining poverty within caste, the main source of social stratification in India.