In earlier research we identified the start of the growth turnaround in the late 1980s. This is consistent with the pattern of (particularly trade) policy liberalisation at the time. Since then there has been a remarkable improvement in per capita incomes. But a puzzle remains. The change in policy should have had a symmetric effect across India. Yet the participation of different states in the turnaround has been very uneven. In this paper we examine whether the relative size of shifts in growth across states could have been predicted from data on state characteristics, measured before the turnaround. We use the "robustness" techniques first proposed by Sala-i-Martin. As might be expected, higher initial literacy, urbanisation and access to ports all predicted stronger growth. But we also find that relatively high shares of both agriculture and registered manufacturing predicted weaker growth across all sectors of a given state, suggesting negative externalities. We guess, along with some other evidence, that this reflects the negative impact of state intervention.
