ABSTRACT
This work presents a novel method to determine structural breakpoints in volatility of a time series. The proposed method, called the breakpoint-search algorithm, utilizes the log-likelihood function value derived from the fitted Generalized Auto-Regressive Conditional Heteroskedasticity (GARCH) model of the given data. In many time series models, volatility is approximated to be constant over a period of time or moving from one regime to another. In such cases, it is important to determine when shifts in volatility patterns occur, in order to capture the behavior of the data very well. The proposed model, together with the GARCH model, are used to identify periods of high and low volatility, and to assess the long-term volatility of the time series. To test the efficacy of the proposed algorithm, the method is applied to a real-world time series data, such as the Philippine Peso-US Dollar currency exchange rate. Empirical results agree with the events in Philippine history indicating periods of economic instability.
