Abstract This paper addresses the questions as to the size and causes of earnings differentials in two urban African labor markets, those of Ghana and Tanzania. We have panel data so we can ask how far time invariant unobservables, market ability for short, matters in the determination of earnings. We also have information on whether the individual is own self-employed, self-employed with employees, a private or public sector wage employee and the size of the enterprise in which wage employees work or which the self-employed owns. We find, mirroring work on developed economies, that unobserved individual market ability is by far the most important factor explaining the variance of earnings. With controls for such ability the gap between private wage employment and civil servants is about 50%. With controls for enterprise size we find that wage employees earn the same as the self-employed in both Ghana and Tanzania. Enterprise size matters. At most half of the OLS effect of size on earnings can be explained by unobservable ability. Workers in the largest firms are the high earners with wage rates which exceed those of civil servants. These results all assume exogenous movement. We find evidence that endogeneity bias may be serious and may be understating the extent of both the size effect and the private sector wage (negative) premium. The implications of our results for understanding the nature of formal and informal employment in Africa are discussed.