Heterogeneous livelihoods, welfare dynamics, and structural transformation in Tanzania

Whether for the purpose of national policy setting or for localized targeting, understanding the welfare dynamics of a population is essential for the efficient allocation of limited resources as well as to avert unintended consequences. However, the literature on welfare dynamics largely focuses on population means and two-technology economies, either by restricting analyses to single asset economies or by relying on theoretical models and simulations with a number of restrictive assumptions. This focus may overlook the plights or prospects of heterogeneous subgroups in complex economies. Using a theoretically grounded, data driven approach to identifying a livelihood strategy choice set in a flexible manner, I estimate livelihood-conditioned returns to assets and associated welfare dynamics using a long panel dataset from Kagera, Tanzania. I find that, between 1991 and 2004, a subset of households moves from the dominant, farm-based, livelihood to a livelihood that allocates more assets to off-farm wage and entrepreneurial activities. In estimating marginal returns to assets across livelihoods, I find significant differences in returns to business, labor, and human capital assets by livelihood strategy, suggesting that households would realize locally increasing returns if they could switch livelihoods. Analysis of welfare dynamics across livelihoods suggests conditional convergence and uncovers heterogeneous welfare dynamics that would be masked in an analysis of population mean dynamics alone. Although beginning with a flexible framework and employing a data driven strategy, the analysis applied here confirms many of the stylized facts of the structural transformation literature, in particular the emergence of two sectors, sector-differentiated returns to labor and other factors, and catch up in the low return sector.