Ensaios sobre o comércio exterior brasileiro: elasticidades dinâmicas e exercícios regionais e setoriais

Abstract

The three essays in this thesis aim to deepen the analysis of the impacts of real exchange rate fluctuations on Brazilian foreign trade, considering dynamic elasticities and regional and sectoral exercises. The first chapter seeks to examine the elasticities of the Brazilian trade balance from a dynamic, or non-linear, perspective, considering the classification by major economic categories (BEC); that is, capital goods, durable consumer goods, semi-durable and non-durable consumer goods, intermediate goods, and fuels and lubricants. The national and international economic changes that have ccurred in recent decades, as well as the shocks to which emerging economies like Brazil are subject, provide a good opportunity for investigations of this nature. To do this, monthly data from January 2000 to December 2022 are used, along with the Time-Varying Cointegration Vector Model (VECM-TVC) proposed by Bierens and Martins (2010). The results indicate positive dynamic elasticities of the real exchange rate in all sectors, except for fuels and lubricants. The evidence for the impacts of external and domestic income variables also shows the expected direction by the literature for most of the sample, except in the model for semi-durable and non-durable consumer goods, whose elasticities showed inverse impacts for much of the analyzed period. In the second chapter, the short-term and long-term impacts of real exchange rate devaluations on the trade balance of Brazilian states are investigated, considering the classification of products by major economic categories (CGCE) to infer about the J-curve hypothesis and the validity of the Marshall-Lerner condition (MLC). To achieve this, autoregressive models with distributed lags in panel data (PARDL) were applied, using information for the 26 Brazilian states between 2000 and 2020. This was followed by the non-parametric estimator for the covariance matrix by Driscoll and Kraay (1998), which is robust to general forms of cross-sectional (spatial) or temporal dependence, commonly present in panel data of this nature. The results indicate that real exchange rate devaluations have negative impacts on the trade balance in all sectors considered in the short term, followed by a positive and elastic long-term effect in most sectors of the CGCE. Therefore, it is consistent with the J-curve phenomenon and the validity of the Marshall-Lerner condition (MLC). Finally, the last chapter of the thesis investigates the asymmetric impact of the exchange rate on the trade balance of Brazilian states, considering sectors of the manufacturing industry classified by technological intensity (SIIT); high technology, medium-high technology, medium-low technology, and low technology. The analysis was based on the Pooled Mean Group Panel Nonlinear Autoregressive Distributed Lag (PNARDL-PMG) estimator, containing annual information for Brazilian states between 2000 and 2020. The evidence found shows that the long-term response of the trade balance of states by SIIT is asymmetric in the face of long-term exchange rate variations. Additionally, while the impact of positive changes (real devaluations) in the exchange rate benefits the trade balance of Brazilian states, regardless of the technological intensity classification considered, the impact of negative changes (real appreciations) in the exchange rate benefits the balance for the high and low technological intensity sectors.