Time stability of the beta convergence among EU countries: Bayesian model averaging perspective

This paper analyzes the time stability of the GDP beta convergence in two subsamples: EU27 countries during 1993-2010 and EU15 during 1972-2010. Additionally, the article checks for the strength and stability of influence of particular economic growth factors. In order to address the problem of variables' selection, Bayesian model averaging (BMA) is used while choosing the appropriate variables for the regression. In order to loose the assumption of stability overtime in the regression, interaction terms of particular regressors are introduced with time dummies (the whole sample is divided into time intervals and is allowed for structural breaks). Finally, in order to address the problem of potential inconsistency of “typically used” estimators, the study employs Blundell and Bond's GMM system estimator. The main findings are the following. (1) The EU27 countries converged at the rate of about 5% per annum while the EU15 countries-at 3% p.a., which is an enormous difference as compared with the widely cited 2% rate of convergence. (2) The pure mechanism of conditional convergence of the countries under study was rather constant over time: there were periods of more rapid or slower convergence but the differences were not as huge as one could expect. (3) The considered economic growth determinants exhibited very mixed and differentiated impact on economic growth in various subperiods.

► Bayesian model averaging (BMA) is applied to the analysis of β convergence. ► The EU27 countries converged at the rate of about 5% per annum. ► The EU15 countries converged at the rate of about 3% per annum. ► The pure mechanism of conditional convergence was rather constant over time. ► The results are robust to different model specifications.