CIEPS Project on Financial Crises and New Perspectives on Financial Behavior
Missing Views and Why Many Failed to Predict the Financial Crisis
Our studies of the global financial crisis show that the belief in false theories or mental models by many officials and private sector participants played a major role in generating the crisis. There was too much confidence in the new technologies of risk management models, policy views based on rational expectations and market efficiency. There was also too much faith that competition among financial institutions would enforce self discipline and protect against excessive risk taking.
While the crisis has clearly shown errors in these views there appears to be a serious danger that many in both the public and private sectors have failed to fully recognize the role of these defective beliefs of theories in the crisis and thus are advocating policies that are likely to lead to repeats in the future, as well as generating new sources of crises. A key unresolved issue is the inability to effectively measure the systemic risk of a bank or other organization. A major case in point is that Basel III, the major international response so far to putting in place a regime to make financial systems safer still relies heavily on the types of risk analysis that allowed major financial institutions to game the system and substantially reduce their capital requirements and hence increase their leverage. Likewise the EU Commissioner for competition policy recently argued that Europe needed bigger banks, despite the evidence that the proliferation of huge financial institutions that are too big to fail have created incentives to engage in excessive risk taking. When things go bad their very size creates enormous drains on the public finances and can have devastating effects on the real economy.
Nor can we safely rely on competition among the banks and financial market participants to provide sufficient discipline of financial behavior. Indeed to the contrary, our research finds that because of short time horizons and the difficulties of accurately assessing risks in complex institutions and financial instruments, competitive pressures have often encouraged institutions to take on more risks in order to keep up their stock prices and market share. The financial markets failed to give strong early warning signals of the growing problems related to subprime lending in the United States. They also failed to give early warnings of the fiscal excesses, loss of competitiveness, and real estate bubble problems that were growing in the eurozone.
The Need for New Paradigms on Financial Market Behavior
Such developments clearly demonstrate that we need to redesign applicability range of the dominant paradigms in economics and finance of financial markets that almost always operate on the basis of far sighted rational expectations based on good information, always liquid markets, continuous trading and normally distributed disturbances.
What types of analysis and models should replace or complement this paradigm is far from clear, however. There is no longer any question that economists need to pay careful attention to the once out of the mainstream views such as Minsky's credit cycle analysis, Kindleberger's analysis of bubbles and busts, behavioral finance models, Soros' view of reflexivity, and developments in mathematical and computer modeling of complexity. All of these give reasons why markets may display their own internal dynamics as well as responding to news as seen in the rational expectations models.
Over the second half of the twentieth century, macroeconomics theory and policy discussions came to be dominated by general equilibrium models in which identical, rational, economic agents armed with rational expectations and perfect information transact in markets that are in equilibrium at all times. In spite of their elegant structures, these models are of no use in anticipating, understanding, or resolving the periodic booms, busts, recessions, depressions, and financial crises that present the most interesting questions to economists.
Perhaps because of these failures, groups of economists have broken away from the general equilibrium tradition to carry out several avenues of research. Behavioral economics, experimental economics, and neuro-economics have challenged the notion that economic agents are unemotional, fully informed constrained optimization machines. Agent-based modeling has challenged the notion of a representative economic agent. And institutional economics, evolutionary economics, and complex adaptive systems have challenged the notion that the economy is in equilibrium at all times. Each of these new research directions has contributed toward a new framework that helps us understand the anomalies that characterize real-world economic behavior.
Over the past forty years, scientists have made dramatic advances in physics and chemistry by examining the behavior of systems that are far from equilibrium. In general, they have found that far from equilibrium systems behave in qualitatively different ways from equilibrium systems. In particular, far from equilibrium systems exhibit abrupt, discontinuous change that is reminiscent of the booms, busts and financial crises we find in economics.
In this view, an economy is a dissipative structure—an open thermodynamic system, far from equilibrium, which transforms energy into work and entropy through a continuous, irreversible process of thermodynamic change. We are particularly interested in the fact that such systems are subject to episodes of abrupt, discontinuous change known in the network theory literature as cascading system failures. In economics we know them as booms, busts, recessions, depressions, credit crises and financial panics.
The economics literature refers to these ideas as evolutionary economics, developed by Prigogine, Haken, Simons, Dopfer, Foster, Witt, and Metcalfe in the tradition of non-equilibrium thermodynamics, complex adaptive systems, and network theory.
Developments in behavioral and neuro economics and finance give us strong micro foundations for why human behavior may lead financial markets to behave in such manners. There are many different ways in which markets may behave imperfectly, however, and the various non traditional approaches often give different predictions, or there are several competing explanations/behavioral biases for the same market phenomena. This has made it easier for strong believers in efficient markets to dismiss such explanations. For example, some behavioral theories predict that markets will underreact while others predict overreaction. Some research is beginning to attempt to reconcile such divergent predictions, but in our view the most important conclusion that we should take from this is not that the behavioral approach is incoherent but that markets are not always correct and thus we should not always take them as giving good pictures of underlying vulnerabilities. There is debate over whether a focus on behavioral and other biases can be used to systematically beat the market, but even if markets are generally efficient in the sense that its very difficult to earn above average risk adjusted returns, , this need not imply, as has been frequently assumed, that markets are always giving correct, full equilibrium prices. This means that there is a strong need for reforming financial market regulations and for public oversight of substantial portions of our financial systems. Ultimate goal of our research is to help develop these specific policy recommendations.
A Focus on Policy Implications
Poor regulation as well as far from perfect market behavior played a major role in stimulating the crisis, but this is no reason to not try to have better regulation in the future. Undoubtedly this would require major reforms in incentive structures and more focus on the financial and economic implications of behavior rather than an almost exclusively legal approach, kind that led to credit default swaps not being treated as the insurance contracts that they were.
It will also require focus on simpler rules which may not be ideal in some circumstances, but work reasonably well in a wide variety of cases and are relatively simple to understand and enforce. This is a major conclusion from complexity economics. The risk weighting systems which the large banks were allowed to use under the Basel regulatory rules were in one sense quite mathematically sophisticated, but they relied on several sets of assumptions that worked fairly well during normal periods but failed to hold in times of crises. Their complexity also allowed the banks to game the system to understate risk and substantially lower their capital requirements. As with the failures of the ratings agencies, which used similar types of models it is not clear to what degree these underestimates of risk were due to cynical gaming and how much to genuine excessive faith in these risk models. Both causes seem to have been important. These failures illustrate the importance of more study of the generation and use of information. A key aspect of such analysis must focus on developing a better understanding of the incentives to generate both information and disinformation and how these are influenced by institutional structures in both the public and private sectors. It is clear that many public officials were subject to some of the same types of incentive problems as were private sector actors. Thus when focusing on financial regulation and supervision in the public sector, much more attention needs to be focused on the political economy of enforcement. While a lack of relevant information certainly played a role in the regulatory failures leading up to the crisis, and if not more important were failures of regulation to take needed actions based on the information that was available. An area in which political economy analysis will be especially important is in the design of mechanisms to facilitate the resolution procedures which deal with the distress of large financial institutions.
Our Project: A Focus on the interactions of multiple factors
Our project focuses on trying to improve the types of analysis which can be used to identify emerging vulnerabilities in financial systems and thus help authorities and financial institutions to take preventive steps before it is too late. The standard approaches to estimating risk promulgated under the Basel rules are essentially backwards looking. Thus they are of limited value in diagnosing emerging problems as opposed to assessing relative risks while the system is operating well. The key to providing better frameworks for identifying emerging dangers we believe is to recognize that financial systems do not always operate in the same manner. Our standard economic models based on equilibrium concepts work well under some circumstances but we must recognize as in the views of Kindleberger, Minsky and Soros that markets at times develop far from equilibrium behavior. While their insights were quite valuable in suggesting that such perverse behavior could develop, their analyses were primarily illustrative of these possibilities with reference to historical examples, but they presented no formal empirical analysis of how the early stages of emerging non equilibrium behavior could be detected. One of our major goals is to develop better methods of identifying these transition phases from equilibrium to non equilibrium behavior, i.e., from normal to bubble phases. Such analysis can be especially useful in developing strategies for stress testing risk models. Again, however, such technical analysis needs to be complemented by analysis of the incentive structures that influence how such stress testing is undertaken in both the private and public sector. For example, it is a widespread view that the stress tests for banks recently required by the European Banking officials were far too lenient in terms of the types of potential problems that the banks were required to analyze. For example, the possibility of sovereign defaults was not included. The official EU position that sovereign defaults were unthinkable undoubtedly played an important role in this omission from stress tests. The results however was a substantial decline in the credibility of the exercise. Of course there already exists a substantial literature on early warning systems for currency and banking crises and institutions such as the International Monetary Fund are engaging in further research on this topic. Apart from the basic value of multiple studies on such an important topic a major advantages of our project are that we will explicitly adopt the perspective of looking for factors that generate and/or are associated with shifts from equilibrium to non equilibrium or bubble behavior, that we take the role of institutions into account in our analysis, and that we focus on the interactions among multiple factors.
The importance of such interactions has been missed in much of the empirical work to date on early warning systems for currency and financial crises has focused on the predictive power of a large number of economic and financial variables taken independently. This misses that the interrelationships among variables can often give important clues about emerging disequilibrium.
We demonstrated this in our research on the Asian crisis of 1997-98. There was considerable controversy about the importance of trade deficits and appreciated currencies and the initial empirical studies found conflicting results. Economic theory suggests that there can be a number of reasons for both trade deficits and currency appreciation and in some cases these developments will be benign and in others they can cause serious problems. We used the combination of large trade deficits and substantial appreciations as a method of discriminating between these types of situations and found that it had considerable explanatory power.
Credit Booms and Asset Price Bubbles: Our Recent Agenda
Other research we have undertaken has investigated how political variables interact with exchange rate regimes and capital controls to affect the probabilities of currency crises. In the next phase of our research we will focus on the roles of credit growth, international financial flows, and asset prices in providing warning signals.
Recent preliminary research has found that the degree of financial liberalization and quality of financial regulation and supervision both have important effects both on rates of credit growth and on the likelihoods that there will be financial crises.
It is clear that excessive rates of credit growth are associated with financial crises but so far it has been difficult to find a reliable metrics for identifying what rates of credit growth are excessive. In part this reflects many different methods of measuring credit booms that have appeared in the literature. Also important we believe is that whether a given rate of credit growth is excessive for a particular economy at a particular time will be dependent on a number of other factors. Many studies include the growth of the economy as a factor but we believe that a broader range of considerations need to be taken into account including the quality of financial regulation, the behavior of international financial flows, and the relationships between monetary policy and credit growth.
International Capital Flows
Of particular controversy here is the role of international capital flows. Many have argued that these have been important contributors to a large number of crises such as the Asian crisis in 1997-98, the US subprime crisis and the current euro zone crisis. This is a plausible hypothesis but there has been little systematic quantitative research to test the importance of this factor. This will be one of the major facets of our project. We will draw on our previous research that shows that some types of capital flows are more likely to have sudden stops than others and that surges of financial capital are particularly likely to be followed by crises.
Another type of recent research has attempted to identify bubbles in asset prices before they burst. One particularly interesting branch of this research draws explicitly on developments in the analysis of fractals and complexity analysis pioneered by Mandelbrot. There is considerable evidence that in their macro level behavior stock markets follow more closely power laws than normal distributions, in other words, that they have more large ups and downs than would be predicted by normal distributions. This is a signature of power law behavior which can be generated by fractals.
Fractals in nature tend to display the same patterns at different scales. If this applies to stock markets than the micro dynamics of market behavior can be used to identify the emergence of bubbles at the macro level, i.e. at the scale at which their bursting would have major effects. Recent research by one of our PhD students, Dhari Al-Abdulhadi, suggests that further exploration of neural network analysis to identify inefficient financial market behavior may be quite valuable. The recent research in this area has had mixed results so we would like to explore this approach further. We would also explore what additional information we can derive from looking for interrelationships among asset price behavior and other variables such as credit growth and capital flows.
The Use of New Political Economy and Financial Policy Data Sets and Case Studies
Our project will draw heavily on the use of recently developed data sets on institutional variables such as measures of capital controls, the quality of financial regulation and supervision, and political capacity. We are now completing a major project that evaluates the quality of many of these measures. This project developed from our use of case studies to complement our formal econometric work. In the process we found that the quality of the large sets of institution measures matter a great deal. Thus we plan to continue to combine our quantitative work with careful case studies. This not only helps us gain a better idea of how well these measures capture the actual economic phenomena. The case studies have also often helped us gain a better idea of the types of interrelationships that are most important for us to test empirically. This is especially important for the study of non traditional views of financial behavior.
While these approaches suggest the possibilities of nonlinear dynamic behavior that can lead to transitions to far from equilibrium behavior and collectively they point to a number of factors to be considered, they do not give us uniform expectations about the specific types of dynamics that we would expect to find. Simply testing a bunch of apriori mathematical functional forms is extremely unlikely to be the best way to uncover the types of dynamics and interrelationships among variables that will give us the best warning systems.
Episodes to be Investigated
In looking for such patterns we will examine many crises beyond the recent global financial crisis since it is clear that financial crises can come in more than one form.
We have already done a good deal of work on the Asian crisis and on the recent global crisis and on larger scale empirical studies that use data from both crisis and non crisis experiences. We will also draw heavily on other studies of crises such as the important recent book by Reinhart and Rogoff. While this gives us a strong background to start from, we have not yet extended our research to the larger scale study of the interrelationships we have discussed above. This combined with the study in detail of additional crises will be the major thrust of our project.
In carrying out this research we are fortunate that in addition to our group of senior researchers we are able to make use of a substantial number of current and former PhD students from a wide range of countries many of whom are associated with their national central banks or finance ministries. This gives us valuable access to a great deal of country expertise.
Products of the project
We expect the initial products of the project to be a series of research studies that would be published in economics and finance journals. As the research proceeds we would then produce papers for policy oriented journals and disseminate our findings in conferences and workshops and meetings with international organizations such as the IMF and national governments and central banks.
Ultimately we would also plan to pull together our findings into a book manuscript.