TFP Estimates and Frontier-Generated Production Function: A New Dataset for Political Science

Abstract

There have been numerous studies concerning productivity, representing two general approaches toward measuring the concept. Parametric approaches specify the actual form of the production function based on theoretical assumptions, while non-parametric approaches use empirical best-practice cases as a benchmark for productivity measures. We propose a new approach to obtain cross-country total factor productivity estimates and a method to derive the production function, which builds on empirical data and does not require a priori assumptions about its functional specification. The approach is based on radial model of data envelopment analysis. The obtained TFP estimates are validated through comparing to PWT and UNIDO datasets. Some preliminary analysis is also provided concerning application of the TFP estimates to cross-country divergence/convergence in productivity. We also demonstrate the potential utility of the estimates for political science research.