Acknowledging the fact that the growth experience of countries is seldom well described by the average growth rate, this paper aims at identifying countries that are similar in terms of their growth process, thus emphasizing the dynamics of growth rates. To that end, the growth experience of countries is interpreted as a Markov switching process with countries switching between four distinct growth regimes: crisis, stagnation, stable growth, and miracle growth. In the model, different growth patterns arise because countries switch between the growth regimes with different frequencies. In order to account for the distinct dynamics, the traditional Markov switching model is extended by a classification mechanism that endogenously assigns countries exhibiting similar dynamics into the same, and countries exhibiting distinct dynamics into different clusters. Three distinct growth clusters are obtained: the first cluster consists of countries that have achieved relatively fast and steady growth mainly by spending time in the stable and the miracle growth regime. Countries in the second cluster have achieved only moderate growth and often found themselves in stagnation for longer periods. The third cluster might be referred as a growth failure cluster because the countries associated with this cluster have suffered from small growth rates and frequent crises. It appears that developing countries can avoid falling into the growth failure cluster by securing a minimum amount of human capital. In contrast to that, the most distinguishing feature of the countries in the successful growth cluster is their reasonable quality of institutions.
