Fitting a fuzzy consensus partition to a set of partitions to analyze the modern economic growth across countries

Summary

In this paper the economic development of a set of countries from 1975 to 1995 is estimated by considering different variables, reflecting the degree of the Modem Economic Growth, MEG, that is the economic progress of nations as a whole (Kuznets, 1966). In each year of the analysis (h=1,…r) the MEG is investigated by a fuzzy partition of then countries measured by four macroeconomic indicators. Since the objective of this paper is to catch the different aspects of the MEG not only in each year but also over the entire period, a new model is developed. In particular, according to the least-squares fitting criterion, aconsensus fuzzy partition is introduced for fitting the best fuzzy partition toP. beingP the set ofr fuzzy partitions of the same set of then countries. The results show that the empirical MEG is well approximated by two groups in 1975, 1980 and 1985, representing two well separated clusters of underdeveloped and developed countries. In 1990, these two clusters tend to split into three groups and the third group includes the countries characterised by a marked acceleration in the rate of output growth. The optimal number of clusters is determined by generalising the empirical test proposed by Calinski and Harabasz (1974) to detect the optimal number of classes of a crisp partition.