Volatility Forecasting and the Efficiency of the Finnish Option Index Market
Jussila, Henna (2005)
Kuvaus
Kokotekstiversiota ei ole saatavissa.
Tiivistelmä
The purpose of this thesis is to compare the predictive power of different volatility forecasting models and to study whether the Finnish option index market is informationally efficient. Volatility forecasting plays an important role in financial decision making, option pricing and risk management. On efficient markets the volatility implied by option prices reflects all available information. Hence, volatility forecasts based on past returns should not include any information beyond implied volatility. The data used in the empirical study consists of daily closing prices of the Finnish option index from March 1994 to December 1999. As a measure of implied volatility the publicly reported volatility measure provided by Helsinki Stock Exchange is used.
The empirical study is carried out by making one-week-ahead volatility forecasts with several different techniques including: simple historical model random walk, more sophisticated statistical model GARCH(1,1), volatility implied by option prices and combination model GARCH(1,1)-implied volatility. Simple average and regression-based approaches are tested in combining GARCH(1,1) and implied volatility. Forecasting performance of different models is evaluated with four error measures: the root-mean-square error, the mean absolute error, the mean absolute percentage error and the Theil–U.
Hypotheses drawn in this thesis are following: (1) the more sophisticated models outperform the simple model in forecasting of actual volatility, (2) the option-implied volatility gives better forecasts than the history-based models and (3) the combination of history-based forecast and implied volatility forecast do not offer more accurate predictions than the pure implied volatility. No statistically significant differences between the models could be found. The ranking of models vary depending on which error measure is considered. The fact that the combination models could not significantly improve the forecasts made based only on the implied volatilities, support the efficient market hypotheses for the Finnish option index market.
The empirical study is carried out by making one-week-ahead volatility forecasts with several different techniques including: simple historical model random walk, more sophisticated statistical model GARCH(1,1), volatility implied by option prices and combination model GARCH(1,1)-implied volatility. Simple average and regression-based approaches are tested in combining GARCH(1,1) and implied volatility. Forecasting performance of different models is evaluated with four error measures: the root-mean-square error, the mean absolute error, the mean absolute percentage error and the Theil–U.
Hypotheses drawn in this thesis are following: (1) the more sophisticated models outperform the simple model in forecasting of actual volatility, (2) the option-implied volatility gives better forecasts than the history-based models and (3) the combination of history-based forecast and implied volatility forecast do not offer more accurate predictions than the pure implied volatility. No statistically significant differences between the models could be found. The ranking of models vary depending on which error measure is considered. The fact that the combination models could not significantly improve the forecasts made based only on the implied volatilities, support the efficient market hypotheses for the Finnish option index market.