In an earlier post, I presented some preliminary results on what I called “growth-reducing structural change.” This refers to the highly anomalous and puzzling phenomenon whereby labor moves from high- to low-productivity activities.
In joint work with Maggie McMillan (Tufts and IFPRI) we have now updated and expanded our dataset and have more complete results. Here is a preview:
This shows a decomposition of aggregate labor productivity growth during 1990-2005 between two components: (i) labor productivity growth within sectors (“within”); and (ii) the economy-wide productivity effects of labor reallocation across sectors (“structural”). The economy is disaggregated into nine sectors.
Remarkably, the African countries in our sample (Ethiopia, Ghana, Kenya, Malawi, Mauritius, Nigeria, Senegal, South Africa, and Zambia) have experienced even more growth-reducing structural change than the Latin American countries. This runs totally against our expectations for a set of countries at such low levels of development (for the most part), where a Lewis-style dual economy growth dynamic ought to be in place.
The chart makes clear that the bulk of the difference in economic performance since 1990 between Asia, on the one hand, and Latin America and Africa, on the other, is accounted for by differences in patterns of structural change. Asian countries have, on average, experienced growth-enhancing structural change; not so the others.
Note that structural change has played a negligible role in the high-income countries (bottom bars, denoted by “HI”). This is as expected since inter-sectoral productivity gaps tend to be small in rich countries.
What explains these differences? We discuss several hypotheses in the paper, but one stands out: the pattern of comparative advantage. Countries with a comparative advantage in primary products tend to have a much worse performance in structural change than countries than ones without.
This is shown in the next chart, which looks at the (partial) correlation between the share of raw materials in exports (horizontal axis) and the contribution of structural change to overall productivity growth.
Note that I have regional dummies here (along with the initial share of agricultural employment) as controls, so we are picking up the effects differences within regions (and not broad regional differences).
The paper that discusses these and related findings is here.