Growth convergence clubs: Evidence from Markov-switching models using panel data

In this paper we model the rate of growth of per capita output as a two-state Markov-switching process. We consider panel data sets for OECD countries, USA states and two wider samples of countries. For each sample, we attempt to characterize each regime's first and second moments, the transition probabilities as well as the unconditional probabilities of being in each regime. We also make inference on the regime that is most likely to have generated each observed growth rate in the panels. We find that the low growth regimes exhibit high volatility but are not very persistent while the high growth regimes are substantially less volatile and very persistent. Also we find that, at the world level, only a few countries seem to have remained in the high growth regime over the entire sample period. On the other hand, all OECD countries and most USA states have remained in the high growth regime over the last decade of their corresponding sample periods. In all cases, no economy seems to have remained in the low growth regime over the different time spans considered in this study.