Total factor productivity estimation for Polish manufacturing industry: A comparison of alternative methods

The concept of total factor productivity (TFP) and its measurement have been of interest to researchers for more than half a century, and are intensively discussed topics in debate on economic growth1. This chapter discusses the different problems related to methodologies for estimating TFP at the establishment (sector and firm) levels. These include simultaneity and selection bias, deflated input and output values, and endogeneity of product choice. It then describes existing techniques for overcoming these methodological problems - specifically, it is shown that these are addressed at sector level by computing TFP using a semi parametric technique at the establishment level. A manufacturing sector data is used at three levels of aggregation for Poland for the period 1995 to 2007; the results indicate significant TFP growth and also intensive dynamic changes to productivity over time. The results from three different techniques - index measures (non-parametric), parametric production function estimation (General Method of Moment - GMM) and production function – to account for endogeniety (semi-parametric) provide consistent results. This suggests that the estimates are sensitive to the technique and definitions of variables used, and indicate the biases related to traditional TFP estimations.sensitive to fat tails than climate policy based on CRRA utility.