Estimation of Non-Linear Dependence between Exchange Rate and Stock Market through Different Time Scales: Latin Case

Abstract

The aim of this paper was to investigate a relationship between stock returns and exchange rate in Brazilian stock market over different time scales. Therefore we used daily quotations of Ibovespa and ratio between the U.S. Dollar and Brazilian Real, from April, 2003 to August, 2011. The initial analysis evidenced that Brazilian stock market is more volatile than Real/Dollar exchange rate. With the time scales coefficients we estimated families of copulas for the relationship between daily log-returns of Dollar/Real exchange rate and Ibovespa. The results of estimated copulas evidenced that dependence between the variables is negative in all time scales, thus confirming the balanced portfolio theory. The coarse scales exhibit great negative dependence, which is reduced in the middle scales and reaches the lower level at the finest scales. This confirms that investors, holding portfolios composed by Brazilian stocks for long time, are more sensitive to Dollar/Real exchange rate. Regarding the shape of joint probability of the variables, there was a predominance of Student’s t copula. This result emphasized that there is more dependence in the tails between exchange rate and Brazilian stock markets than the normally expected. It highlights the necessity of risk management in portfolios composed by stocks of Brazilian market in situations of great variation in Dollar/Real exchange rate. No less important is the caution of the exchange rate policy of the Brazilian government to huge variations and turbulence periods in the stock market. DOI: http://dx.doi.org/10.5755/j01.ee.25.3.3710