Clubes de convergência de renda na América: uma abordagem através de painel dinâmico não-linear com variável limiar

Abstract

The main objectives of this work are to test empirically the hypothesis of income convergence process among American countries, to classify this convergence process as either absolute or conditional and to determine if this process happens in either a linear or non-linear manner. Estimations were made through both TAR (threshold autoregressive) panel and linear autoregressive panel and the results were compared to each other. The sample of countries are composed by Argentina, Bolivia, Brazil, Canada, Chile, Colombia, Costa Rica, Dominican Republic, Ecuador, El Salvador, Guatemala, Honduras, Jamaica, Mexico, Nicaragua, Peru, Trinidad and Tobago, Uruguay, USA and Venezuela. The sample period is 1953-2003 and data are in annual basis. Results held for this sample show no evidence of convergence in both TAR and linear models. Additional estimations were made in sub-samples of countries that compose three American free trade agreements. The porpoise was testing two different hypotheses. The first one is that convergence process occurs in clubs. The second one is the theoretical hypothesis that foreign trade leads to convergence among countries involved in it. The three free trade agreements widened were Nafta, CAN and Mercosul. Results held for Nafta also show no evidence of convergence in both models. CAN’s results show empirical evidence of convergence, as TAR model concludes for absolute convergence in one of two regimes. Results held for Mercosul sample show stronger evidence of convergence process. Both linear and TAR models conclude for absolute convergence, the former in both regimes. Even using a different methodology than conventional β-convergence and σ-convergence, results are in consonance with those found in the literature.