Análise de convergência de renda local entre países

Abstract

Income convergence analysis has undergone in the last twenty years several theoretical and empirical developments. However, the variety of models, ideas and results put forward in literature are a compelling evidence of the relevance of this kind of research and its notorious evolution. Within this context, this study investigated the conditional convergence hypothesis in a local perspective, thereby contributing to the current literature estimating not only global results but providing information about each country coefficient. The growth rate of GDP per worker is the dependent variable for 148 countries over a five-year period between 1985 and 2005. In addition to the “basic” explanatory variables related to convergence literature (initial GDP, population growth and physical capital), other explanatory variables were included, such as: human capital, life expectancy, government spending, international markets opening degree and market distortions measurement. According to WEO/IMF classification, there is a clear division among advanced economies and other emerging markets and developing countries. This specification suggests the existence of different spatial regimes. The extreme spatial heterogeneity was handled by Geographically Weighted Regressions estimation (GWR). The results obtained by Exploratory Spatial Data Analysis indicated spatial dependence for global growth rate of GDP per worker across 148 countries in the form of a concentration pattern. The econometric results indicate unobserved effects suggesting a better fit when fixed effects are considered in the estimation. The best global spatial model is the spatial lag model confirming the convergence hypothesis and providing an estimate of 2.07% speed per year. As expected, investment in physical capital and human capital and the international market insertion act positively on economic growth. On the contrary, government spending and purchasing power parity measurement impact negatively. Moreover, the neighbors’ economic growth rate is an important factor in growth of a particular country. As to spatial regimes, the best model for advanced economies is the error spatial model. Once again convergence hypothesis was confirmed, reaching a 1.92% speed per year. Private investment influences substantially the development attained by the advanced ones, as opposed to the insignificant role played by the government. Addressing the extreme spatial heterogeneity by means of GWR approach, significant differences were verified when considering local relationships, specific to each country. Mapping results of local speed of convergence and local half-life demonstrated considerable geographic concentration. Besides, international evidences validate the convergence proposed by neoclassical economists. The speed of convergence around 2% per year seems to be “a ubiquitous constant”.