Tampere University of Technology

TUTCRIS Research Portal

News Arrivals, Jumps and Variance in Stock Markets

Research output: Book/ReportDoctoral thesisMonograph

Details

Original languageEnglish
PublisherTampere University of Technology
Number of pages116
ISBN (Electronic)978-952-15-4045-5
ISBN (Print)978-952-15-4038-7
Publication statusPublished - 20 Nov 2017
Publication typeG4 Doctoral dissertation (monograph)

Publication series

NameTampere University of Technology. Publication
Volume1508
ISSN (Print)1459-2045

Abstract

News containing important financial and economic information plays a crucial role in the process of investment and trading in financial markets. Sudden large changes and strong fluctuations observed in asset prices are normally related to the arrival of certain important news. However, the relationship between market reaction and news flow is complex and ambiguous. This thesis focuses on two classes of important news—firm-specific and macroeconomic announcements—and the impact of firm-specific announcements and macroeconomic announcements on jumps and variance of stock prices. Jumps, as abnormally large returns, and variance, measuring market fluctuations, are the two most important financial risk variables. A clear investigation into the impact of these two classes of news on jumps and variance will substantially contribute to financial risk management.

The first part of this thesis concentrates on the impact of news on jumps. First, a non-parametric statistical framework is introduced to examine the association between news arrivals and jumps in stock prices. To uncover the market reaction to news alerts, I focus on the time distances (waiting time) between news arrivals and the nearest detected jumps. For a given news item, both backward and forward waiting times are calculated with the jumps detected before and after the news arrival. In particular, backward waiting times may reflect possible information leakage. To examine whether observed jumps are associated with real news, a set of timestamps of general reference news is simulated considering intraday seasonality. Applying non-parametric tests, we are able to extract the statistical profiles of the empirical waiting times and their simulated references. As a result, the association between news and jumps is quantitatively demonstrated.

Taking advantage of the developed statistical framework, a thorough empirical analysis is implemented using Nordic and U.S. data to show the impacts of Nordic firm-specific and U.S. macroeconomic announcements on stock prices in both Nordic and U.S. markets. Specifically, the impact of scheduled and non-scheduled firm-specific announcements on Nordic stock prices is tested. I also investigate the sizes of jumps related to Nordic scheduled and non-scheduled firm-specific announcements following the same non-parametric methodology. In order to feature the importance of certain types of firm-level news, such as acquisitions, five important firm-specific announcements are selected to test their impact on Nordic stock prices in term of jumps. Regarding U.S. economic news, I provide empirical results for the impact of U.S. macroeconomic announcements on the U.S. stock index. In addition, U.S. macroeconomic announcements are grouped by announcing time. Their impacts on Nordic stock prices are studied to examine the importance of announcing clock time and the global influence of U.S. economic releases.

The second part in this research relates to the impact of macroeconomic news on equity variance modeling and the related option valuation performance with GARCH models. Impact variables of macroeconomic news are constructed using both the arrival timings of U.S. macroeconomic announcements and realized variance, and are incorporated into classical GARCH models. The impact variables of macroeconomic news slightly improve the joint likelihood of returns and VIX for all models compared with standard GARCH models. Regarding option valuation, an affine GARCH model with news event data consistently outperforms a pure affine GARCH model. However, there is no such consistent result for NGARCH and GJR models, implying that the explicit use of macroeconomic news events data does not improve the performance of variance modeling and option pricing with non-affine GARCH models.

Field of science, Statistics Finland

Downloads statistics

No data available