Volatility on Oslo Stock Exchange : structural breakpoints in volatility using the ICSS algorithm

Abstract

This thesis uses the Iterative Cumulative Sum of Squares (ICSS) algorithm (Inclán and Tiao, 1994) to search for structural breakpoints in volatility in the Norwegian stock market. I analyze data both with the original method, and with a revised version by Bacmann and Dubois (2001). I also analyze the stock markets in the other Scandinavian countries, a European index, the US stock market, the oil market and exchange rates. I then state tentative explanations for the breakpoints. I present a problem with the algorithm; the number of, and dating of breakpoints in a period is dependent on the choice of time period. I also show that the algorithm has problems discovering breaks occurring in the beginning or end of the time series. Through simulations, I show that the algorithm almost always fails to discover the correct number of breakpoints when there are several breaks quite close in time.