Use this URL to cite or link to this record in EThOS:
Title: Three essays on information, volatility, and crises in equity markets
Author: Clark, Shane K.
ISNI:       0000 0004 5921 6025
Awarding Body: University of Strathclyde
Current Institution: University of Strathclyde
Date of Award: 2015
Availability of Full Text:
Access from EThOS:
Access from Institution:
This thesis consists of three essays, which examine the behaviour of stock market indices in light of the recent crises, in relation to volatility, information and sentiment. Essay 1 focuses on the link between the flow of public information and stock market volatility in the FTSE, DJIA and the S&P 500 indices. This essay builds on existing literature in several ways. First, a new proxy for the daily public information flow is created to encompass a wider range of publication than previous contributions. It is also disaggregated by media type. This proxy is tested as an explanatory variable in index return volatility. Through the use of augmented GARCH models, the essay explores (a) whether information flow is a significant explanatory variable in volatility persistence, (b) whether the type of media the information flow is carried in impacts the volatility-information relation, (c) whether information backlog is incorporated into volatility when market re-open, and (d) whether there is a lag in the way information is incorporated into volatility (including by media type). Results show conclusive evidence of a strong relation between information and volatility. Further, the media types which deliver the most current information, i.e. Wire-Feeds, tend to have the largest impact on returns volatility. The evidence on the role of backlog and information lags in volatility is mixed, with significance only in the S&P 500 for Backlog, and only in S&P 500 and DJIA for lags. Essay 2 further examines the stock market volatility and public information relation in light of daily market anomalies. Building on the literature on the Day of Week Effect, this essay investigates whether a Day of the Week effect is present in US and UK indices both before and during the latest financial crisis. Augmented GARCH models are constructed to test for a Day of the Week Effect in either returns and/or volatility, and are further explored in light of the information arrival proxy created in Essay 1. Results show weak evidence of Day of the Week effect in the data, except for a Friday Effect in the Russell 2000, both before and during the crisis; which confirm that small size stocks are more likely to exhibit Day of the Week effects, as suggested by previous uses of equal-weighted indices. Through the interaction variables, results show that information flow can satisfactorily explain Day of the Week effects. Essay 3 investigates the relation between proxies for investor sentiment and stock market crises and recoveries on international indices. Using an Early-Warning-System (EWS) model, the essay examines whether investor sentiment is a useful predictor for the occurrence of stock market crises and early signs of recovery. Three alternative proxies are used to measure investor sentiment, including previously cited measures of stock market riskiness, investors’ risk aversion and investors’ optimism about stock markets. The results show that investor sentiment is overall a significant predictor of the occurrence of crises within a one year period, and that the addition of sentiment into early warning signal models of stock market crises can improve the predictive performance of the model (increases in investor sentiment increase the probability of occurrence of a crisis, which is in line with previous contributions finding a negative lead-lag relation between sentiment and stock returns). The extension of the model to early signs of recoveries also shows that sentiment is a reliable predictor. The measure of stock market riskiness (Baker and Wurgler, 2006) is found to be a better predictor than the Volatility Index (VIX) and the Put-to-Call Ratio (PCR). The cross-country comparison results confirms the literature findings that the link between sentiment and stock market returns varies across indices and cultures, as the predictive power of the variable appears strongest in the French and U.S. indices.
Supervisor: Not available Sponsor: Not available
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral