TFP in Germany and the USA (2000–2017): Classical and Bayesian Approaches to Trend Analysis

Abstract

This study analyses the trends in total factor productivity (TFP) in Germany and the USA during the period 2000–2017 using a combination of classical and Bayesian methods. First, TFP was calculated with the Cobb–Douglas production function and the Solow residual, and trends were estimated using linear and interaction models. Parametric estimates confirmed statistically significant higher TFP growth in the USA, while nonparametric permutation tests did not confirm these differences, highlighting the limitations of short time series. The Bayesian panel model made it possible to incorporate prior knowledge about the production function and provided more stable estimates, while posterior tests confirmed good model fit and robustness of the results. Despite the short time span and limited sample, the study offers a meaningful methodological framework for comparative productivity analysis that combines frequentist and Bayesian approaches. Such an approach can also be applied in studies of other countries or over longer time periods.