In search of true productivity differences

Recent work on production functions estimation revealed that substantial biases can be introduced into the estimates when the assumption of perfect competition and price exogeneity is not satisfied in the data itself. As Klette and Griliches (1996) show applying traditional econometrics in differentiated good markets will negatively bias the scale estimates of the production function. In fact, when deflated sales are used as a proxy for output in case of differentiated good industries scale economies (and subsequently productivity) cannot be estimated independently of markups. We extend this basic framework to show that, if exporting markups are smaller than those attainable in the domestic market, the Klette-Griliches estimation procedure will tend to overestimate exporting firm markups and underestimate their productivity. In addition, we provide an estimation algorithm based on the Olley-Pakes (1996) framework that could serve to ensure unbiased estimates of exporter productivity.