Title: Change point detection in the distribution of errors in dynamic linear models
Authors: Lajos Horvath - University of Utah (USA)
Zhenya Liu - Renmin University of China (China)
Shixuan Wang - University of Reading (United Kingdom) [presenting]
Yaosong Zhan - China Capital Market Institute (China)
Abstract: A new test procedure is developed for detecting changes in the distribution of errors in dynamic linear models. Under the null hypothesis, the distribution of errors remains the same, while there are multiple changes in the distribution of errors under the alternative. Our procedure is based on the cumulative sum (CUSUM) process that compares the empirical distribution functions of the residuals in the first part observations and the whole sample. We derive the asymptotic properties of the proposed test statistics. Monte Carlo simulations show that the proposed test has good size control and high power. We provide two empirical applications for GDP and inflation forecasting.