Structural Change and Cross-Country Growth Empirics

Abstract

One of the most striking features of
\n economic growth is the process of structural change whereby
\n the share of agriculture in GDP decreases as countries
\n develop. The cross-country growth literature typically
\n estimates an aggregate homogeneous production function or
\n convergence regression model that abstracts from this
\n process of structural change. This paper investigates the
\n extent to which assumptions about aggregation and
\n homogeneity matter for inferences regarding the nature of
\n technology differences across countries. Using a unique
\n World Bank dataset, it estimates production functions for
\n agriculture and manufacturing in a panel of 40 developing
\n and developed countries for the period from 1963 to 1992. It
\n empirically models dimensions of heterogeneity across
\n countries, allowing for different choices of technology
\n within both sectors. The paper argues that heterogeneity is
\n important within sectors across countries implying that an
\n analysis of aggregate data will not produce useful measures
\n of the nature of the technology or productivity. It shows
\n that many of the puzzling elements in aggregate
\n cross-country empirics can be explained by inappropriate
\n aggregation across heterogeneous sectors.