Change Point Estimation in Volatility of a Time series using a Kolmogorov Smirnov Type Test Statistic

Abstract

Detection of structural change in volatility of a time series is very important for understanding volatility dynamics and the stylized facts observed in financial time series. By applying the Nadaraya Watson kernel estimator of the mean function, estimated residuals are obtained. In this work, a Kolmogorov Smirnov type test statistic for change point estimation is developed and applied to conditional variances obtained from the squared residuals. The consistency of the change point estimator is shown through simulations. The developed estimator is then applied to KES/USD exchange rate data set to estimate a single change point.