Nonlinear Stochastic Convergence Analysis of Regional Unemployment Rates in Poland

This paper analyses convergence of unemployment rates in Poland at NUTS4 level by testing nonlinear convergence, applying the modified KSS-CHLL for each pair of territorial units. The results suggest that actually the convergence is a rare phenomenon and occurs only in 1916 cases out of potential over 70 000 combinations. This paper inquires what systematic reasons contribute to this phenomenon. There are some circumstances under which unemployment convergence should be more expected than in others. These include sharing a higher level territorial authority, experiencing similar labour market hardship or sharing the same structural characteristics. For each of these three criteria we analyse the frequency of the differential nonstationarity within groups (as evidence of convergence) and across groups (as evidence of "catching up").