O modelo de três equações is-pc-mr e violação da condição de marshall-lerner. Uma análise empírica via barvx para américa latina entre os anos de 2000 a 2024

Abstract

This paper analyzes the dynamics between the real exchange rate, exports, imports, domestic output, interest rates, and external income for four Latin American countries—Brazil, Chile, Mexico, and Colombia—within the framework of testing the dynamic version of the Marshall- Lerner condition. The study combines five domestic variables—export volumes, import volumes, the real effective exchange rate, the monetary policy interest rate, and real GDP— with one external variable, serving as a proxy for global income growth. In order to reconcile measures of external competitiveness and international demand while avoiding dimensionality issues, the study estimates, for each country in the sample, two Bayesian Vector Autoregressive models with an exogenous variable (BVARX). The results indicate significant heterogeneity across the economies analyzed in response to real exchange rate appreciation shocks. While Brazil and Chile exhibit results broadly consistent with the Marshall-Lerner condition, particularly through the behavior of imports, Colombia and Mexico display deviations from the model’s theoretical predictions. In general, both GDP and interest rates do not conform to the expected dynamics under the validity of the Marshall-Lerner condition, highlighting limitations in the explanatory power of the three-equation model. Finally, the variance decomposition analysis reveals that, for Brazil, Chile, and Colombia, exports are predominantly explained by their own shocks, whereas in Mexico they are more sensitive to domestic income. Imports exhibit a similar pattern across countries, with their dynamics largely driven by their own lags and showing low sensitivity to the other variables included in the model.