Convergencia Regional en Chile: Nuevos Tests, Viejos Resultados

The traditional tests of convergence implicitly test the unit root hypothesis for per capita income series. Despite that under the null the statistics do not have standard asymptotic distributions, most papers on this topic use standard distributions to evaluate this hypothesis, with the corresponding wrong inference. The goal of this paper is to determine the existence of convergence in per capita GDP and per capita income across the regions of Chile using not only the traditional cross-section and panel tests, but also the recently developed test for panel data that allows making the correct inference. The tests also include confidence intervals for the variance of per capita GDP, to check convergence in standard deviation, and a multimodality test, to analyze the existence of convergence clubs. The main conclusions are: (1) there is evidence of absolute ß convergence in per capita GDP and per capita income; (2) the convergence rate for income is higher than in GDP; however, both are low for international standards; (3) the convergence rate increases if the mining production share is included in the regression; (4) despite that the data seems to show the existence of convergence clubs, they are not statistically significant; (5) there is evidence of s convergence in GDP over the 1960-2000 period, but not in per capita income for the period 1987-2000.