Empirical Aspects of Financial Stability

This thesis discusses the empirical aspects of financial stability and presents evidence that suggests that stock market bubbles and volatility are related, and that financial crises are also triggered by events related to non-financial sectors. Financial crises are predominantly related to boom episodes and asset price bubbles, which can seriously impact the financial system when they burst. This thesis draws upon the findings of previous papers and argues that the risk of financial instability (systemic risk) is formed during the boom phase and materialises on the eruption of crisis. In so doing, this study considers stock market bubbles as a potential source of risk for financial stability.

The severe impact on the economy in the wake of the recent financial crisis has not only demonstrated the way in which trouble in a relatively small market can escalate into a serious crisis exerting economy-wide effects, but is also an example of the important role financial stability plays in the functioning of modern economies. Chapter 1 addresses factors that contribute to financial crises and policy tools to mitigate their effects. The Global Financial Stability Map (Map), summarising and graphically presenting underlying factors that may lead to a systemic threat, shows the complex interactions among different factors that affect each other and, in combination, are relevant to financial stability. In this connection, the importance of countercyclicality is addressed and the weaknesses of the Value at Risk (VaR) measure are discussed.

Chapter 2 examines whether longer periods of low volatility influence the formation of bubbles, which are defined as the difference between current prices and an adaptive moving average of an alternate history of asset prices, and whether stock market bubbles increase the likelihood of stock market crashes. The regression analysis employed confirms that longer episodes of low realised volatility exerts a significant influence on the formation of stock market bubbles, which, in turn, significantly increase the likelihood of stock market crises. This relationship is referred to as volatility paradox. Furthermore, the bubble is incorporated to inflate Value at Risk, in order to generate a countercyclical capital buffer for extreme events. It is shown that the inflated VaR covers the majority of the extreme negative returns. This leads to the conclusion that the information content of bubbles should be taken into account while measuring risk in stock markets.

The recent financial crisis revealed that even relatively small markets of the economy are capable of jeopardising financial stability, and the objective of Chapter 3 is to assess the contribution of both financial and non-financial sectors of an economy to systemic risk. For this purpose, the marginal systemic risk contribution, measured by ?CoVaR of 10 sectors, is estimated for the US, the UK, and Germany, through the employment of quantile regressions. The estimated ?CoVaRs are tested for significance and dominance in order to classify sectors as systemically relevant and to obtain a formal ranking of the sectors in terms of contribution to systemic risk. The outcomes reveal a weak link between VaR and ?CoVaR and significant externalities of sectors. Chapter 3 discusses the role of low volatility in excessive risk-taking and lending behaviour during booms as well as the deleveraging behaviour during bust episodes. It argues that countercyclical tools, which reduce such behaviour, can successfully mitigate financial crises. This line of argumentation is related to the countercyclical capital buffer discussed in Chapter 2, which is aimed at dampening the upward movements of asset prices. Taking into account the finding that real economy sectors also have significant effects on systemic risk, Chapter 3 proposes the application of macroprudential policy tools individually to those sectors in which bubbles emerge.

Chapter 4 compares the realised volatility levels between international stock markets. Although the volatility patterns are fairly similar, the pairwise t-test reveals a significant difference between the volatility levels of national stock markets. A two component GARCH-MIDAS model is applied to decompose conditional volatility into a short-run and a long-run volatility component and to link macroeconomic variables directly with stock market volatility. The outcomes of the GARCH-MIDAS model indicate that stock market volatility is associated with macroeconomic variables, and that stock market volatility depends upon different variables in different countries. Realised volatility is found to explain a considerable proportion of conditional volatility. The Granger causality test reveals no significant causal relationship of volatility with illiquidity or with sentiment.