The Sensitivity of Productivity Estimates

Researchers interested in estimating productivity can choose from an array of methodologies, each with its strengths and weaknesses. This study compares productivity estimates and evaluates the extent to which the conclusions of three important productivity debates in the economic development literature are sensitive to the choice of estimation method. Five widely used techniques are considered, two nonparametric and three parametric: index numbers, data envelopment analysis, instrumental variables estimation, stochastic frontiers, and semiparametric estimation. Using data on manufacturing firms in two developing countries, Colombia and Zimbabwe, we find that the different methods produce surprisingly similar productivity estimates when the measures are compared directly, even though the estimated input elasticities vary widely. Furthermore, the methods reach the same conclusions on two of the debates, supporting endogenous growth effects and showing that firm-level productivity changes are an important contributor to aggregate productivity growth. On the third debate, only with the parametric productivity measures is there evidence of learning by exporting.