The objective of this research is to study the statistical properties of the time series of the CLP / USD nominal exchange rate, for the period of recent free float, specifically to model the returns and volatility, to finally evaluate the short-term predictive efficiency out of the sample and the ability of the models to generate abnormal returns through simulation of buying and selling strategies. ARIMA and GARCH models are applied to the daily data series of the interbank exchange rate, January 1999-December 2001; The predictive efficiency outside of sample uses the year 2002. The results are rather even in the short term, so the predictive power of a martingale is almost as good as the estimated models. This would lead us to think that the exchange market in Chile is efficient at least in its weak form. The final simulations indicate that when applying rolling, all the models improve profitability compared to the recursive technique and regarding the profitability of the static strategy, but it is not conclusive.
Gregoire C., J., & Obilinovic P., F. (2004). Exchange rate in Chile, floating period 1999-2002: return and volatility models. Estudios De Administración, 11(2), 27–65. https://doi.org/10.5354/0719-0816.2004.56800
