Forecasting volatility in the brazilian foreign exchange market

Abstract

This paper presents a comparative study of the predictive capacity of the models EWMA, GARCH (1,1) EGARCH (1,1) and TARCH (1,1) when applied to forecast the volatility of the exchange rates in the Brazilian inter-bank market. The sample consists of the daily closing quotations of the exchange rate real/US dollar obtained in the period from August 20, 2001 to September 30, 2003. The results showed that the TARCH (1,1) model, provided the most accurate forecast performance for this period, followed closely by the EGARCH (1,1) model, then the GARCH (1,1) model, and finally the EWMA model. There was also evidence that all the models revealed a tendency to overestimate the future volatility. It was confirmed as well that the Brazilian Exchange Clearinghouse operates in an extremely traditional and subjective way concerning the definition of the exchange rate variation indexes guaranteed under contract and, consequently, in collateral requirements.