Piecewise Constant Modeling and Tracking of Systematic Risk in Financial Market

The objective of this thesis is to study the time-varying systematic risk in capital market represented by beta. By using statistical hypothesis testing, we show that beta changes in a piecewise constant pattern in which the changes are governed by triggering economic events. This pattern of beta is different from previously modeled time-varying patterns in literature, such as random walk and mean-reverting models and is consistent with the efficient market hypothesis. We also present a new modeling technique based on Poisson process to represent piecewise constant beta. We develop a new tracking algorithm based on Kalman Filter in which Bayes' selection criteria is incorporated to track piecewise constant beta. Our simulation results show that our proposed tracking method outperforms the traditional random walk and mean reverting model based Kalman Filter tracking. Our empirical case studies also show that our method is efficient in capturing the significant risk changes which are attributed to economic events.