UC Berkeley

Does the sectoral composition of aggregate economic growth affect poverty? We ask whether agricultural growth in developing countries increases the expenditures of poorer households more than growth in other sectors. While some reduced form analyses have tackled this question using either country-level time series data, regional panel data for one country, or cross-sectional country data, this paper is unusual in using panel data for many countries. We improve on much of the existing literature by devising an instrumental variables strategy to correct for the endogeneity of sectoral GDP growth, involving averaging over sectoral income growth rates for neighboring countries. Our principal finding from our instrumental variable estimator is that the estimated elasticities associated with growth in agricultural income are significantly greater than for non-agricultural income for all but the extreme top and bottom deciles. In the middle range of the income distribution the effect of a given GDP growth due to agriculture is 3–4 times larger than if it was due to non-agricultural activities. Having established that on average growth in GDP originating in agriculture is more beneficial for poorer deciles, we finally explore whether this is a pattern which holds across different groupings of countries. A second important finding is that there is heterogeneity across some groupings. Most particularly, we find that it is the poorest people in the poorest countries for whom agricultural income growth is the most beneficial.