As events during the 2008 financial crisis and its aftermath have shown, extreme movements in financial time series can have wide socio-economic impact. The immediate effects involve huge swings in wealth, which in turn can lead to financial insecurity, institutional insolvency, misallocation of economic resources, and global consequences that threaten the credibility of major economic institutions. In addition, economic globalization has led to strong financial linkages that can increase risk exposure in the event of a large common shock. Understanding this phenomenon, exploring its causes, mapping its evolution over financial markets, and studying its effects on the real economy all present major challenges to the economics profession.
This project aims to develop quantitative methods that will contribute to our understanding of some of these issues. Beyond the immediate concerns that have confronted policy makers during the recent financial crisis lie questions that relate to the emergence and detection of the bubble phenomena and the evolutionary course of the resulting crisis through the financial and economic systems. These issues form the focus of the primary program of proposed research, which seeks to design new econometric methodology to enable early detection of bubble behaviour and to provide empirical date stamping technology that quantifies dynamic timing and helps to map the evolutionary course of a financial crisis.
The project will provide a rigorous econometric approach to dating financial bubbles, deriving a limit theory for empirical estimates of the origination and collapse dates of bubble episodes, and using a new model of financial bubble activity that is based on the successive conjunction of ?efficient market? models and ?mildly explosive? models that can capture episodes of financial exuberance. An extensive empirical implementation of this technology to recent financial crisis data is envisaged, yielding empirical information on the temporal extent, the magnitude, and the course of the various bubble phenomena that have involved financial and commodity markets, exchange rates and real economic activity.
In addition to this primary research program, two secondary projects will be pursued. The first deals with co-movement in economic and financial data where there is some uncertainty about the persistence characteristics in the data, as is common in empirical research. The second program of research will explore new econometric methodology for dynamic panel modelling in the presence of individual effects. The importance of these fields is reflected in the vast empirical literatures in the social sciences that utilize cointegrating methods to study long run linkages among time series and dynamic panel regression to study the effects of individual decision making over time.
The dot.com bubble of the 1990s and the global financial turmoil over 2007-2009 have focussed attention on financial asset price bubbles and their potential global consequences. There is now widespread recognition among policy makers as well as economists that changes in the global economy over the last two decades, far from decoupling economic activity as was earlier believed, have led to powerful latent financial linkages that have increased risks in the event of a large common shock. As many commentators have emphasized, a substantial percentage of the world's accrued wealth was destroyed within 18 months of the subprime crisis, with manifold effects ranging from the collapse of major financial institutions to the near bankruptcy of national economies. These phenomena have led to general recognition in the economics profession that new empirical methods are needed to improve understanding of speculative phenomena and to provide early warning diagnostics of financial bubbles. This background of financial exuberance and collapse with the concatening effects of the crisis across markets and nations is the central motivation for the research that was conducted on this project. Beyond immediate policy issues lie fundamental questions that relate to the emergence and detection of bubble phenomena and the evolutionary course of crisis events through financial markets and the economic system. The program of research on this project sought to address these issues. Bubbles are a form of nonstationarity, where there is explosive or mildly explosive instability in the system. Looking back on the 1990s Nasdaq bubble, a fascinating empirical question that underpins the present research progam is whether there was evidence for a bubble in stock markets when the Fed Chairman Alan Greenspan famously spoke of "irrational exuberance" in an after dinner speech in December 1996. The recursive diagnostic tools developed in the project successfully address this question. Our findings confirm that there was clear empirical evidence of a bubble in the Nasdaq for at least 15 months prior to Greenspan's comment. Importantly, Greenspan's comment, like that of similar comments by subsequent Fed chairs and Central Bank governors, was phrased as a question: "how do we know when there is irrational exuberance in financial markets". The findings of this research project provide tools that enable us to give an evidence-based answer to this question. The object has been to devise a real-time detector, a form of warning alert system that can be used by banks, central banks, regulators, and even individual investors to determine whether bubble conditions apply in a financial market. The primary outcome of the project is the econometric methodology that serves this purpose. Some of the methods are explicitly designed to deliver early warning diagnostics that enable detection of bubble behavior prior to market collapse. Associated with these diagnostics are empirical date stamping technology and testing procedures that help to map out the course of a financial crisis such as the recent general financial crisis (GFC). The PI's work on the project has provided a rigorous econometric approach to dating bubble phenomena that involves the derivation of a limit theory for empirical estimates of the origination and collapse dates of bubble episodes. This limit theory forms the basis of an econometrically sound approach to inference about bubble phenomena. The project has successfully applied these new techniques to financial market data and real estate data in the USA, mapping out the progess of the crisis and showing the dates at which evidence confirmed the existence of market exuberance. The methods have also been applied to data from many other countries and are now being used by central banks. Looking forward, we should have some accumulated evidence in a few years time about how successful these methods are in identifying crisis situations in many different markets and countries.