Risco de câmbio no mercado interbancário brasileiro: um estudo comparativo entre modelos de predição de volatilidade

Abstract

Volatility is an important variable of the modern financial theory, present in financial decision-making and asset valuation processes, as well as in risk management methodologies. However, choosing the appropriate model for forecasting volatility adjusted for each application, asset and market - is not a trivial task, and as such has been the focus of attentionof many researches over the last twenty years. The studies demonstrated that the performance of volatility forecast models differ depending on the analyzed asset, the specified sample, and on the existence of characteristics that usually observed in financial time series, such asheteroscedasticity, volatility clustering and the leverage effect. Thus, this work presents a comparative study of the forecast capability of the EWMA (adopted by the BM&F Foreign Exchange Clearinghouse), GARCH (1,1), EGARCH (1,1) and TARCH (1,1) models - employed to predict the volatility of the exchange rate in the Brazilian interbank market. Thesample consists of daily exchange (closing) rates of Real/US-dollar between August 20, 2001 and September 30, 2003. Study results demonstrated that the TARCH (1,1) model presents the best forecasting performance, closely followed by the EGARCH (1,1), followed by theGARCH (1,1) model and, at last, by the EWMA model. Moreover, all the studied models have shown a propensity to overestimate future volatility. Finally, due to the overly conservative position and progressive detachment amid the values set for the contractually guaranteed exchange rate fluctuation index and the index based on an econometric model, the eventual substitution of the EWMA model by the TARCH (1,1) model would not imply in alterations in the current operational context of the Brazilian Clearinghouse.