Public compensation policies, occupational choice, and aggregate productivity

This paper argues that the observed skill-bias toward the public sector in many developing countries cannot be accounted for by returns to schooling differentials between the public and private sectors, as evidenced by classical Mincerian wage regressions. The bias may instead reflect unobservable individual characteristics. We formulate and estimate a simple occupational choice model in which heterogeneous risk-averse workers choose between a guaranteed public sector employment with a relatively flat-wage profile, and a private sector job featuring a more skill-sensitive wage and unemployment risk. The main implication of the model is that misallocation of skills may occur in the absence of unemployment insurance, as risk aversion induces some high-skilled workers to select themselves into the public sector, notwithstanding the productivity edge that they would enjoy in the private sector. We estimate the model structurally using both maximum likelihood method and semi-parametric estimation techniques—with micro data from the 2006 Ghanaian Households Living Standard Survey. We also use reduced form estimation results to assess the empirical relevance of the model predictions for the Ghanaian data. The estimated model is then used in counterfactual policy simulations to evaluate the impact of public wage policies and unemployment benefits on occupational choice and aggregate productivity. We find that limiting the private sector unemployment risk would reallocate some high-skill workers to the private sector and generate important aggregate productivity gains. In contrast, widening the public-private wage differential or adopting a flat unemployment benefits scheme would have perverse effects on the allocation of skills and aggregate productivity.