Three observations on market discipline
Speech delivered by Malcolm D Knight, General Manager of the BIS, at the BIS-Federal Reserve Bank of Chicago Conference on 'Market Discipline', Chicago, 30 October 2003.
We certainly all agree that market discipline is an indispensable element of a well functioning economy. When market discipline is strong, it serves to guide funds to their most productive uses with due regard to risk. That is because market forces can raise the cost or restrict the volume of funding for those activities that have unattractive risk/return trade-offs. By threatening the demise of enterprises with poor risk/return profiles, market discipline can deter the wasteful use of scarce resources and excessive risk taking.
Market discipline is clearly a necessary condition for financial stability, even if it is not a sufficient condition. In the repressed financial systems that existed in many countries well into the 1980s, where there were many types of interest rate and capital controls, financial system instability could remain hidden for years. But these repressed financial environments created ponderous inefficiencies, stifled incentives to price and manage risks, and resulted in "covert" insolvencies that only became exposed, sometimes with severe financial consequences, once these systems were opened to the healthy rigours of competition. There is by now a solid consensus that strengthening market discipline within a clear prudential framework can help to achieve a better balance between financial system stability and economic efficiency.
In my remarks today, I want to take a forward-looking perspective and stress three observations about market discipline on which there is no such consensus. In my view, these three issues merit greater attention from the academic and policy-making communities.
The first observation identifies a limitation of market discipline; the next two observations draw out some policy implications of this limitation. In particular, they highlight two existing gaps that will need to be addressed in the strategies aimed at boosting market discipline if these strategies are to be effective. Filling these gaps will require improving the information available to exercise discipline, and sharpening market participants' incentives to enforce discipline.
My first observation: Market discipline is less effective in restraining a generalised, overextension of risk taking than it is in restraining excessive risk taking by individual firms. And it is generalised overextension that lies behind the forms of financial instability that have the greatest macroeconomic costs.
My second observation: As regards policies designed to improve market information, we need to understand better what types of information are best suited to identifying the build up of aggregate risk, as opposed to firm-specific risk, in the financial system. Call this, if you will, the "information gap". The disclosure of such information is an essential complement to improving the markets' ability to identify the risk profiles of individual institutions.
My third observation: As regards policies that target incentives to control risks, a key challenge is to find ways of narrowing the gap between individually rational behaviour and courses of action that are collectively desirable. This "incentives gap" in part reflects the inevitable "endogeneity" of aggregate risk with respect to collective behaviour.
Let me expand on each of these observations.
I - My first observation: market discipline's limitations are in the large, not so much in the small
Experience indicates that markets are at their best when identifying a risky institution in an otherwise healthy financial system - the black sheep in the flock, as it were. As long as sufficient information about the financial condition of the firm exists, and as long as there is little prospect of external assistance that might insulate claimants from losses, market prices and funding conditions will tend to reflect the firm's relative riskiness. In fact that's why it is not uncommon for supervisors to rely partly on market intelligence and market reactions to spot early signs of distress in individual institutions.
At the same time, experience also indicates that markets find it harder to identify and limit excessive exposure to risk in financial systems as a whole. Evidence for this view comes from the many banking crises that have punctuated economic history and that have become more frequent since the financial liberalisation that has taken place in many countries over the past quarter century. The root cause of these crises typically was not the increased fragility of a few institutions whose subsequent failure spread to others through contagion. Rather, it was the build-up of widely shared exposures across the financial system to macroeconomic risks, which were in turn closely associated with business fluctuations.
I think, there are two reasons why markets are less good at identifying and mitigating systemic risk than the risks run by individual institutions.
The first has to do with perceptions of risk. It is no doubt easier to measure the cross-sectional dimension of risk than its time dimension, especially that of system-wide risk. The benchmark to assess cross-sectional risk, namely average behaviour at a point in time, is more easily identifiable. In particular, it is less dependent on our limited knowledge of the causes of business fluctuations and on competing and controversial paradigms about the interaction between the financial and the real economy. And, above all, there is less of a temptation to justify outliers. In the cross-sectional dimension, all that is required is for one institution to be "wrong" [ie: an outlier]. In the system-wide time dimension, somehow most institutions must be "wrong" at the same time. The burden of proof is higher; discounting past experience is easier. Paradoxically, the reluctance to question average behaviour makes average behaviour more questionable.
The second reason has to do with incentives to take on risk. Actions that may appear reasonable for individual economic agents may collectively result in undesirable outcomes. This is true both for the managers of firms and for those who provide the external financial capital to firms. For instance, during booms fear of loss of tangible short-term profit opportunities can numb risk aversion and encourage overextension. It can do so by boosting credit expansion, asset prices, and real expenditures beyond sustainable levels in a self-reinforcing process. And during downswings or in response to signs of financial strains, rational individual retrenchment can result in a "rush for the exits" that creates generalised financial distress. This fundamental "endogeneity" of risk is very difficult to price into individual financial instruments.
I think that this line of reasoning helps to explain why risk perceptions, and the willingness to take on risk, move in such a highly procyclical way, thereby tending to amplify business fluctuations. It helps to clarify why markets may fail to prevent the build-up of aggregate risks sufficiently promptly. And it underscores the need to address the "macroprudential" dimension of financial stability in order to strengthen market discipline. By this I mean the dimension that highlights the role of shared system wide exposures, the intimate relationship between the macroeconomy and the financial system, and the endogeneity of aggregate risk with respect to collective decisions.
II -My second observation: better information: both in the large and in the small
By now, there is a well-established consensus on the need to improve the disclosure of the risk profile of individual financial institutions. The steps taken to strengthen Pillar 3 in the new Basel Capital Accord are just the latest example of this. And yet, vital as they are, these steps are not sufficient to address generalised, as opposed to institution-specific, overextension. To this end, we need to close the corresponding "information gap". In particular, we need to sharpen market participants' ability to identify the risk profile of the financial system as a whole, and we need to explore how best to disclose that information to the financial markets so as to strengthen their disciplinary role. Such disclosure, if sufficiently timely, can make market participants more cautious.
Admittedly, this task is far from straightforward. Indeed, it is complicated by the same factors that make it so difficult for markets to identify generalised overextension in the financial system. At the same time, one can make some informed guesses about its contours. First, in order to identify such vulnerabilities with sufficient lead-time, the assessment should pay particular attention to conditions during economic booms, the phase in which risk builds up. Second, it would have to consider both the likelihood and the severity of potential stress. Consequently, one might expect indicators of asset price misalignments and of excessive leverage to figure prominently. Third, it would have to be based on some form of aggregate information, with the degree and form of aggregation depending on the type of vulnerability considered.
Some encouraging progress has been made in recent years. Research by my colleagues at the BIS and elsewhere suggests that it may be possible to develop simple indicators of generalised overextension that are capable of foreseeing fairly well, and with a reasonable lead time, the types of banking crises that have struck a number of countries since the 1980s. Much effort has also been devoted to developing system-wide macro-stress tests. In particular, the IMF's Financial Sector Assessment Programs (FSAPs) have confirmed the heuristic value of these exercises. And broadly based evaluations of financial vulnerabilities now feature regularly in the activities of international policy forums such as the Committee on the Global Financial System (CGFS) and the Financial Stability Forum (FSF).
Even so, this work is only in its infancy, and there are many challenges ahead. Much more research is needed to develop the necessary analytical frameworks and to elaborate and assess indicators of vulnerability. For instance, little has been done beyond the analysis of traditional banking crises: financial market distress, as illustrated in the autumn 1998 turbulence, remains largely unaddressed. The fact that episodes of serious financial instability are, by their very nature, rare, makes it hard to reach conclusions with the necessary confidence. And the onward rush of structural change in the financial system further adds to the need for caution. On the positive side, the continuous advances in risk management and disclosure at the level of individual institutions promise to improve the raw material on which system-wide risk assessments could be made.
III - My third observation: the need for better incentives: not just safety nets
It is not much of an exaggeration to say that for a long time the prescriptions to improve the incentives to exercise market discipline were a short-hand for "reducing the scope of financial safety nets". No doubt ill-designed safety nets can numb the incentives to behave prudently. After all, insulating creditors and investors from losses weakens the ultimate disciplinary sanction of a market economy. And since insulation is more likely in the event of generalised distress, as opposed to institution specific distress, this may also be a factor behind the comparative difficulty that markets face in limiting generalised overextension. Policymakers have rightly devoted much attention to this problem in recent years. Prominent examples include the implementation of "structured early intervention" mechanisms in a number of countries, and the search for a more balanced burden sharing with the private sector in managing sovereign debt restructuring problems.
And yet, if the main source of distortions in private incentives is the gap between individually rational and collectively desirable courses of action, the problem is more endemic and, consequently, the right answers are harder to identify.
The problem is more endemic because the conditions that give rise to this "incentives gap" are quite common. As noted, the gap arises whenever two conditions hold: an individual action comes to be seen as more reasonable as the number of economic agents doing the same thing increases, and systemic risk is an increasing function of that number. In particular, the gap is not the result of the moral hazard induced by safety nets, but is much more "primitive". In fact, safety nets are sometimes seen as a second-best response to this gap. The obvious example is liquidity-driven bank runs.
The right answers are harder to identify because there need be nothing "irrational" about the corresponding forms of behaviour. For instance, the widespread existence of short time horizons is not an irrational feature of economic life. Rather, it often reflects well-grounded contractual arrangements that try to address the uneven distribution of information between the parties in an economic relationship. For example, the need to monitor closely the performance of asset managers or corporate managers may encourage an "excessive" reliance on short-term performance indicators, such as quarterly portfolio returns or share returns.
This is a largely unexplored area, and I have hardly any specific suggestions to make. Generally, however, in order to narrow the "incentives gap", we need to give more thought to what forms of non-intrusive public intervention can help agents internalise some of the unintended consequences of their collective behaviour. We also need to recognise that there are inevitable limits to the worthy search for "incentive compatible" prudential standards, because the perspectives of private agents and public authorities necessarily diverge. Finally, the effort should focus more on prevention rather than cure. Hence the desirability of encouraging the build up of cushions in good times, as risk builds up, so as to be able to run them down, up to a point, in bad times, as risks materialise.
To conclude, there is considerable scope to strengthen market discipline, particularly in the time dimension, as the economic cycle unfolds. We all agree on the goal. However, there are still many open questions about the best means to achieve it. I have tried to suggest that the questions are deeper and broader than sometimes thought. I hope that my remarks can encourage us to move a bit further towards identifying the right questions to ask in order to identify the source and dynamics of systemic risk, as a first step towards providing the answers for reinforcing market discipline.