An investigation into using news analytics data in GARCH type volatility models

Abstract

In the work we study different dynamic volatility models. We consider the family of ARCH and GARCH models to compare the performance of the models using both unconditional coverage Kupiec’s test and the test of conditional coverage proposed by Christoffersen. In-sample estimation procedure and out-of-sample evaluation will be based on General Electric stock market closing daily prices (January 2, 2008 - December 31, 2010). We consider different volatility models augmented with news analytics data to examine the impact of news intensity on stock volatility. First we consider two types of GARCH models: augmented with volume and augmented with news intensity. Based on empirical evidences for some of FTSE100 companies it will be shown that the GARCH(1,1) model augmented with volume does remove GARCH and ARCH effects for the most of the companies, while the GARCH(1,1) model augmented with news intensity has difficulties in removing the impact of log return on volatility. Then we compare GARCH model with jumps and GARCH–Jumps model augmented with news intensity using likelihood ratio test. The study shows that the problem of examining the impact of news intensity on volatility is far more sophisticated than it might seem at first sight. Some hypothesists and suggestions for future work are proposed in the final chapter.