Gary H. Stern
Federal Reserve Bank of Minneapolis
Global Credit & Operational Risk Forum 2007
November 19, 2007
The overarching theme of my remarks this morning is that regulators of the financial services industry are likely, perhaps inevitably, to confront difficult tradeoffs in designing policies responsive to the recent tumult in financial markets. As a general statement, this may seem self-evident, but specific cases illustrate the challenging nature of such tradeoffs and strongly suggest that the correct course of action, when put to a cost/benefit test, may be far from obvious. Policymakers have faced such tradeoffs before; indeed, the transition from Basel I to Basel II can be viewed as a move from a relatively simple, low cost regime with limited ability to link bank capital to institutional risk-taking to a more complex framework with the potential to establish a stronger connection between capital and risk.
In these remarks, I plan to consider recent events in credit markets and to explore lessons we might take from them and policy choices we might make. Let me emphasize at the outset that, as always, I am speaking only for myself and not for others in the Federal Reserve. Further, let me remind you that it is exceptionally early to begin a retrospective of recent experience: certainly some situations are far from resolved, and thus identification of principal lessons learned from the disruption will necessarily be incomplete and probably prioritized inadequately as well. Nevertheless, I think we can say something meaningful about potential reforms and about the tradeoffs inherent in their adoption. These are important matters; if I am right about tradeoffs, then some reforms might impose significant costs and contribute to outcomes we would prefer to avoid. Ultimately, policymakers could find themselves relearning old lessons rather than improving social welfare.
These comments are organized along the following lines. I will begin by identifying an issue raised by the recent financial turbulence and consider a potential policy response, highlighting the tradeoffs embodied in the response. I will consider four such issues. These discussions are at a fairly high level, but I would expect the tradeoffs to hold at the micro-level as well.
Policymakers will certainly find opportunities to improve current regulations and practices; the status quo will need to change in some areas. But, as we will see, and as foreshadowed previously, tradeoffs suggest that policymakers will want to be extraordinarily careful in addressing perceived inadequacies in the current environment.
Higher than expected losses on subprime mortgages were the proximate catalyst for the recent turmoil in credit markets. And many observers have attributed poor performance of these mortgages to the "originate to distribute" model commonplace in the mortgage market. With originate to distribute, the steps underlying the production of a mortgage are divided into distinct activities, frequently with different entities carrying out each. For example, mortgage brokers might market various mortgage products to customers, another firm might purchase mortgages and pool them, and still other investors might ultimately fund them, with another firm servicing the mortgages after origination.
Because of the many hand-offs in the process—and the terms of the contracts between at least some of the firms—a number of the firms involved in the process did not have a clear stake in the longer-run performance of the mortgage. The incentives in this model, then, may have encouraged large-scale production of low-quality mortgages. Moreover, this system distributes mortgage risk to a wide range of investors, many of whom do not have to reveal in detail (or at all) their holdings. Diffuse distribution of risk follows in this context because the original funders of the mortgages likely have modified their exposures through sales of structured financial products, some of which, by the way, may be difficult to understand and to evaluate.
Wide distribution, and the complexity of structured products, may obscure where risks truly reside. In these circumstances, should ultimate investors decide that they want to reduce their exposure to subprime mortgages, they may be unsure about which assets to sell or from which firms to pull funding, leading them as a defensive measure to curtail exposures to a wide range of assets and firms, some of which may in fact have little to do with the subprime market.
By implication, this litany of issues and concerns may commend the "vertically integrated" approach to lending, wherein a bank takes the loan and its risk on its balance sheet and earns all the revenue and pays all the expenses associated with it. In this world, the residence of the risk would be relatively clear.
However, having a single institution carry out all the activities associated with mortgage lending, or similarly with other loans, does not eliminate principal-agent problems. We all know that, even within firms, it is essential to get the incentives and contracts right so that all work to achieve the same goal.
And the alternative—the originate to distribute model—has a core and fundamental economic advantage propelling it: specialization. Over time, firms have developed that specialize in the distinct steps of the lending process, from originating the loan to funding it. Such specialization contributes importantly to cost efficiencies, innovation, and a broadening of access to financial capital. Another advantage of the model is diversification; the originate to distribute process allows a firm to significantly diversity the asset side of its balance sheet.
In short, these benefits are valuable in that they facilitate effective use of resources. Actions to limit specialization and diversification would therefore likely to be costly, with adverse consequences for economic performance and living standards over time. Policymakers would need to consider such costs in assessing reforms proposed to constrain the originate to distribute model.
Some observers have identified reliance on the work of credit rating agencies as a dramatic case of the downside of specialization—of separating out aspects of the credit extension process. In particular, it is asserted that investors gave too much credence to the views of the rating agencies and failed to do enough independent credit analysis. While not an expert in this area, I think it important that we think critically about what constitutes excessive reliance on external credit analysis in these circumstances.
If critics mean that there is some systematic failing in the existing rating process and that this is not apparent to those who use the ratings, then reforms in this area may be fully appropriate. But if instead observers are concerned that the rating agencies simply got expected outcomes wrong in some sense (as they have before), then reforms that might subvert the agencies and the information they provide may simultaneously impose significant costs.
To be specific, it could be exceptionally costly for each investor to build the infrastructure required to conduct serious credit analysis, and these costs need to be weighed against the losses suffered by investors in the current regime. Moreover, were the agencies unique in underestimating the losses in, say, the subprime mortgage market? It is not obvious that a different infrastructure will produce better results.
More positively, the rating agencies represent one way of economizing on the production of information on credit instruments. And by charging issuers, they also try to address the public nature of this information for, once the information is produced, there is almost no cost to distributing it and hence it is otherwise difficult to get paid. Absent these charges, there could be too little credit information produced. Overall then, reforms that might compromise the viability of the agencies or discourage use of ratings present the tradeoff of potentially raising costs and ultimately requiring another solution to the issues the agencies help to address.
Prior to this summer's experience, some commentators had noted that both the volume of subprime lending and the returns required on risky assets had moved into the tails of historic experience (or even beyond historic experience). Some interpreted this observation to suggest that the level of risk taking, both by investors and home buyers, was not sustainable. Moreover, some observers attributed outsized risk taking to provision of "excessive liquidity," an evocative but highly imprecise term, on the part of central banks. Others attributed the situation to the so-called savings glut.
In either case, some have asserted that preemptive action on the part of central banks—and particularly the Federal Reserve—could have cooled the credit markets and at least damped the excesses and losses suffered by subprime mortgage borrowers and investors. This argument deserves serious consideration, but it should be recognized that such preemptive action is not without costs.
To wit, we cannot pretend that actions which effectively curtail subprime lending won't have a negative effect on homeownership since, after all, many subprime borrowers continue to meet their financial obligations and reside in their homes. As advertised in previous comments, there is a tradeoff here that requires careful calibration as policies are considered; foreclosures might well have been avoided if subprime lending were reined in, but in all likelihood homeownership would have been lower than otherwise as well.
More broadly, most of the tools which central banks have at their disposal are blunt instruments which do not permit policymakers to narrowly target their effects. An increase in interest rates designed, say, to address excessive liquidity will likely have restrictive effects on a wide range of households and businesses. Again, we shouldn't pretend that these consequences don't exist; they must be part of the cost-benefit calculation of policy alternatives.
Interestingly, the excesses in asset prices perceived in recent years seem related, at least casually, to innovation. Consider the run-up in prices of technology stocks in the late 1990's and this year's turbulence linked to pricing of structured financial products and subprime mortgages. It may be costly to try to address these situations ex ante if, in fact, such actions would inhibit the underlying innovation. Common to all of these concerns is the difficulty of appropriately valuing financial assets. It is quite plausible that, in pursuing preemptive action, the unintended consequences rival or exceed the desired outcomes.
A classic tradeoff facing policymakers concerns financial stability and market discipline. Greater reliance on market disciple runs the risk of less stability. And stability can seemingly be assured through government support which overrides the occasionally harsh dictates of the market.
But as we know, government support involves substantial costs of its own. Elimination of market discipline encourages excessive risk taking on the part of financial institutions which, in turn, makes financial disruptions both potentially more likely and more costly. It would seem, then, that an appropriate balance has to be struck here, and that policymakers have to determine their tolerance for the tradeoff between market discipline and instability.
While true in general, I do think that the market discipline—instability tradeoff has been exaggerated. In fact, by taking steps to reduce the threat that the failure of a large bank, or decline in asset values in one market, will spillover to other institutions or markets, policymakers can actually increase market discipline and simultaneously achieve greater financial stability.
Let me expand on how we might approach this happy state of affairs. Right now, uninsured credits or large banks likely expect government support in the event of problems at such institutions. This is the "too big to fail" problem which arises because creditors know full well that policymakers worry about so-called contagion effects—the possibility that weakness at one bank will spillover to other banks and other markets and perhaps ultimately to the real economy. Protection of uninsured creditors forestalls or at least limits the costs of such spillovers.
But if policymakers act in advance to reduce the probability and/or magnitude of spillovers, they also reduce the incentive to protect creditors. And if creditors accurately perceive the change in circumstances, they then have greater incentive to appropriately apply market discipline.
What kind of proposals might accomplish these ends? I have co-authored a book which addresses these matters in detail; suffice it to say here that constructive proposals range from limiting the size of losses in the first place (a form of effective prompt corrective action) to reducing the degree to which exposures on payments systems can act as the conduit for spillovers. One specific option we discuss in the book is scenario planning, which might usefully simulate, for example, a failing bank situation. Policymakers can, in the simulation exercise, consider if they have or can get adequate information, if they have adequate powers to address the situation, and what if any changes may be advisable prior to the development of a real problem.
The collapse of Northern Rock—a large mortgage lender in the United Kingdom—provides evidence of the potential importance of such simulations. There are ongoing discussions in the U.K. about the adequacy of stress tests and contingency planning in light of Northern Rock. Without claiming to know the preparatory steps the British authorities took, one might imagine that recent experiences might contribute to potential improvements in regulators' simulations.
If and as authorities take steps to reduce potential contagion effects and thereby lay the foundation for enhanced market discipline, timely communication of such activities becomes important. Policymakers need to assure that creditors understand that prospects for protection are diminishing, so that they in turn can act accordingly. It may be difficult to communicate effectively in the midst of turmoil, as the Northern Rock case suggests. The Bank of England hoped that its announced support of Northern Rock would stem the run on the bank but the initial communication appeared to have the opposite effect. A lesson here is that it is important to communicate credible intentions and plans prior to the onset of a crisis.
My comments this morning are not intended to defend the regulatory and financial status quo, although I can understand that some may interpret them in that way. Rather, they are meant to suggest that there is likely little, if any, "low hanging fruit" to harvest and that, specifically, reforms may well impose inefficiencies and other costs of their own. The message is thus one of caution but not of inaction: we should take considerable care in drawing conclusions about the origins of the recent bout of financial turbulence and about the implications for public policy.