The self-employed constitute a large proportion of the workforce in developing countries and the sector is growing. Different accounts exist as to the causes of this development, with pull factors such as high returns to capital contrasted with push factors such as barriers to more desirable salaried jobs. Using data from Ghana, we investigate the changing structure of earnings in self-employment relative to salaried work. We decompose earnings in a two-sector labour market allowing for flexible patterns of sorting on unobservables by means of a correlated random coefficient model estimated by IV-GMM. A unique panel dataset provides us with suitable instruments to tackle the endogeneity of sector choice and capital accumulation. We show that returns to productive characteristics in SE have increased significantly over the period 2004-11 and the sector has attracted workers with higher skills. We conclude that pull factors have significantly strengthened, pointing against the grim view of self-employment as an occupation of last resort.
self-employment semiparametric models comparative advantage segmentation African labour markets