Fault Lines: How Hidden Fractures Still Threaten the World Economy

by Raghuram G. Rajan

  More generally, in an ever-changing global economy, stasis is often the greatest source of instability, for it means the system does not adapt to change. Competition and innovation, by contrast, help the system adapt, and if properly channeled, are key to ensuring variety, resilience, and therefore dynamic stability. Critics will be quick to point out that competition and innovation lead to greater instability during the run-up to a crisis: after all, securities like the CDO squared and the CDO cubed were so devilishly difficult to price that they had no market once mortgages started defaulting. However, it is not competition per se, but rather the distorted banker incentives and the distorted price of risk that led to the creation of these instruments.

  Reduce Incentive and Price Distortions:

  Manage Expectations of Government Intervention

  As I argue in chapter 7, some of the incentives to take excessive risks may have resulted from a breakdown of internal governance within banks, and some from a breakdown of external governance. These mechanisms need to be fixed, and I discuss some options below. But the primary reason for a systemic breakdown is invariably the underpricing of risk. One reason for underpricing is irrational exuberance: initial, moderate underestimates of risk feed on each other until they become a frenzy. Usually, though, such bubbles rarely occur out of the blue. The underpricing of risk in the period leading up to the recent crisis stemmed, in part, from anticipated government or central bank intervention in markets. And certainly, since the crisis hit, the Treasury and the Fed have intervened massively in markets, thus verifying the expectations. We need to find a way to dispel the notion that the government or its agencies will prop up a market, whether the market for housing or the market for liquidity.

  End Government Subsidies and Privileges to Financial Institutions

  As damaging as government intervention to help specific markets is government intervention to prop up specific financial institutions. The essence of free-enterprise capitalism is the freedom to fail as well as to succeed. The market tends to favor institutions that are protected by the government from failing, asks too few questions of them, and gives them too many resources for their own good. Moreover, such protection distorts the competitive landscape. We have to aim for a system in which no private institution has implicit or explicit protections from the government, one in which every private institution that makes serious mistakes knows it will have to bear the full costs of those mistakes.

  Enact Cycle-Proof Regulation

  As we emerge from the panic, righteous politicians feel the need to do something. Regulators, with their backbones stiffened by public disapproval of past laxity, will enforce almost any restrictions, while bankers, whose frail balance sheets and vivid memories make them eschew any risk, will be more accepting of such restrictions. But we tend to reform under the delusion that the regulated institutions and the markets they operate in are static and passive, and that the regulatory environment will not vary with the cycle. Ironically, faith in draconian regulation is strongest at the bottom of the cycle, when there is little need for participants to be regulated. By contrast, the misconception that markets will govern themselves is most widespread at the top of the cycle, at the point of maximum danger to the system. We need to acknowledge these differences and enact cycle-proof regulation, for a regulation set against the cycle will not stand. To have a better chance of creating stability throughout the cycle—of being cycle-proof—new regulations should be comprehensive, nondiscretionary, contingent, and cost-effective.

  Regulations that apply comprehensively to all levered financial institutions are less likely to encourage the drift of activities from heavily regulated to lightly regulated institutions over the boom. Such a drift has been a source of instability, because its damaging consequences come back to hit the heavily regulated institutions in the bust, through channels that no one foresees. For example, the asset-backed securities that Citigroup had placed in thinly capitalized off-balance sheet vehicles came back onto its balance sheet when the commercial paper market dried up, contributing immensely to Citigroup’s troubles. We have to recognize that because all areas of the financial sector are intimately interconnected, it is extremely hard to create absolutely safe areas and imprudent to ignore what happens outside supposedly safe areas.

  Regulations that are nondiscretionary and transparently enforced have a greater chance of being adhered to, even in times of great optimism. Compliance can be more easily monitored by interested members of the media and the public, thus offering some check on the regulator. The regulated also have a good sense of how regulations will be implemented. One example of such regulation is the FDIC Improvement Act of 1991, which mandates that regulators take a series of ever more stringent actions against a bank as its regulatory capital drops below specified levels. The weakness in the act is that regulatory capital is hard for the public and the press to measure in real time, so it is difficult to gauge whether regulators are following through on the requirements of the law. Nevertheless, the act suggests a possible direction for new regulation.

  Wherever possible, regulations should come into force only when strictly necessary: they should be triggered by adverse events rather than be required all the time. Such contingent regulations have two effects. First, because they kick in only some of the time, they are less cumbersome than regulation that is not contingent, and banks will not invest as much ingenuity in trying to evade them. Second, the level of the regulatory requirement—such as the required level of capital—can then be increased if necessary, so that it has the desired effect when really needed.

  Finally, regulations that are not particularly costly for the regulated to implement or for the regulator to enforce are less likely to be evaded or ignored. Moreover, they are likely to have more staying power as memories of past problems fade.

  Reducing the Search for Tail Risk

  Let us apply all these principles to a concrete question: how do we prevent the systematic tail risk taking that nearly destroyed the financial system? I have focused on two related examples of tail risks—the risk of default embedded in senior asset-backed securities and the liquidity risk inherent in financing potentially illiquid assets with very short-term debt. I examine detailed proposals for reducing such risk, because measures that seem reasonable at first glance often raise more concerns on closer examination.5

  As I argue in chapter 7, there are huge incentives at every level in the financial system to take on these tail risks if they can be concealed from those assessing performance. Those giving up the tail risk are willing to pay a premium to do so, while those taking it on and downplaying the eventual risk of payout can treat the premium as pure profit, the product of their natural brilliance rather than merely a compensation for risk. The premium paid by those selling the risk increases in proportion to the anticipated loss if the risk actually hits (they pay more if they think earthquakes will do more damage); and the higher the premium, the more of the risk the sellers are willing to take on (because they do not anticipate being around when their firms have to make good the loss). Too many firms will be eager to take on risks that have extremely costly consequences, and they will compete to take the risk at too low a price. The net result is that too much of the risk will be created.

  Matters grow worse if, because of the extent of risk taking, everyone anticipates that when the risk actually materializes, the government or the Federal Reserve will be drawn in to support the markets underlying the tail risk, or the institutions taking it on. In that case, those taking on the tail risk will find it worthwhile even if it is common knowledge that they are doing so, because the government backstop will make it profitable. Few financial institutions will want to be left out of the orgy of risk taking. So what starts off as an attempt to take on hidden risk and fool the market will, if enough firms get in on the act and induce an anticipation of government intervention, become an overt and profitable play that is rewarded by the market.

  If tail risk is knowingly take
n, or knowingly encouraged by securities markets, the way to deal with it is to alter incentives throughout the corporate hierarchy, the financial firm’s liability structure, and its regulatory structure. One seemingly obvious way to reduce the perverse incentive to take tail risk before risk taking becomes systemic is to do away with all pay tied to performance: to pay bankers like bureaucrats. Of course, firms can offer other rewards for performance, such as status and promotions, that are harder to eliminate. And even if it were possible to prevent evasion of the rules, there would be another obvious downside; bankers would have no incentive to work hard or take calculated risks. In today’s competitive, fast-moving economy, bureaucratic bankers would not be an improvement over the status quo. What we need from bankers is competent risk management, not complete risk avoidance. So we have to find ways to reduce the incentive to take tail risk even while rewarding bankers for performance so that they continue to offer innovative products that meet customer needs and lend to the risky but potentially very successful start-up.

  This means that wherever possible, the risk taker should suffer targeted penalties if the risk materializes, so that society does not feel the need to absolve them because the innocent, the connected, or the politically vocal will suffer alongside. Of course, extremely high penalties will deter even ordinary risk taking, so the penalties will have to be appropriately calibrated. It may also well be that no one, including markets, can anticipate some risks. To deal with such possibilities, it is necessary to build some private-sector buffers to absorb shocks, as well as make the system resilient to them. I will now be more specific.

  Altering Incentives Generated by Compensation

  The most obvious form of tail risk taking is conducted by individual traders or company units, as was the case at AIG’s Financial Products Division, whose staff benefited from the extraordinary profits and associated huge bonuses on the way up and were retained with high bonuses to clear up the mess they created on the way down. One way to make units internalize small-probability tail risks that senior management or risk managers may not see or understand is to hold a significant part of a unit’s bonuses in any year in escrow, subject to clawbacks based on the unit’s performance in subsequent years—a suggestion I made before banker bonuses became a political football.6 Simply put, if a trader makes a hefty bonus this year, only a fraction should be paid out to her this year, with the remaining amount held back by the bank to be paid out over time on condition that her positions do not lose all the money earned this year in subsequent years. This will give the trader a longer horizon, creating some uncertainty about whether any tail risk she takes could actually hit before her bonus is paid, thus giving her greater incentive to avoid it.

  Of course, such a compensation structure will be effective only if a trader knowingly takes tail risks, not if she is unintentionally guided into taking them. Crude limits on the positions individual traders or units are allowed to take, and mandatory diversification requirements, may also be necessary, not so much to prevent tail risk taking but to minimize the loss if it does occur.

  Incentives at the Top

  The proposals above to manage tail risks presume that top management wants to curb such risk taking. I have discussed a number of reasons why CEOs might not want to do so, such as their desire to match the profits shown by other risk-taking managements and their insouciance about the downside because they believe that the government will intervene. Some old-time bankers reminisce fondly about the time when investment banks were partnerships, and partners had their entire wealth at stake. Given the size of banks today and their international sprawl, it would be difficult to convert them back into closely held partnerships in which mutual monitoring by partners inhibits excessive risk taking. And forcing banks to shrink might not make them safer. We have also seen that simply giving management an equity stake may not be enough—they realize enormous gains if the risk taking pays off but have limited liability if it does not.

  Instead it may be useful to consider ways to place some of the burden of risk on top management, without necessarily having entire classes of claims being subject to that risk. For instance, the Squam Lake Working Group (a nonpartisan group convened by Professor Ken French of Dartmouth College after the recent crisis to propose reforms) has suggested not only holding back some portion of top management bonuses and reducing them if there are future losses—much like the clawbacks I discussed earlier—but also writing these “holdbacks” down if the firm has to be bailed out in any way. Thus the holdbacks would serve as junior equity and give strong incentives to management to take precautions to avoid a bailout.

  The financial firm’s board is meant to be another check on mismanagement. But even when tail risk taking is not in the shareholders’ interests, the bank’s board of directors may not be an effective source of deterrence. Board members are generally poorly informed when they are truly independent, and excessively cozy with management when they are not. Lehman’s board, for instance, consisted of very respectable independent directors. But at the time it filed for bankruptcy in 2008, of the ten-member board, nine were retired, with four over the age of 75.7 One was a theater producer, another a former Navy admiral. Only two had direct experience in the financial services industry. Although advanced age is no disqualification, it typically suggests some remove from the cut-and-thrust of modern finance. Such a board, whose risk committee met only twice a year, had only limited ability to monitor Lehman’s risk taking.

  Boards can be strengthened by requiring more financial services expertise of directors, as well as by drawing them from outside Wall Street. Furthermore, a board can obtain better information if the risk managers in the firm are required to report directly to it on a regular basis. The board’s risk committee should also have regular meetings to discuss firmwide risk with unit heads across the firm, without top management present, so that they have a sense of what is going on from those who are closest to the action. Of course, if competent boards are propelled by the same risk-taking incentives as top management, we will have to look elsewhere for ways to discipline risk taking.

  Could bank supervisors play a role in monitoring risk taking by top management, as a second line of defense beyond the board, so to speak? Some commentators, including, famously, Alan Greenspan, were skeptical that supervisors would be able to do so. Supervisors are typically less well paid than private-sector executives, though they have more job security. Except for those really motivated by public service, supervisors tend to be either less talented or extremely risk averse, neither of which is a particularly helpful attribute in modulating private sector risk taking. Nevertheless, supervisors have two strengths that can make them useful checks on private risk taking. First, they have different incentives: they focus more on the small-probability risks of disaster than does the private sector. Second, they can demand data from firms across the industry and obtain a very good picture of where risk concentrations are building up. Because the tail risks that matter most are those that the whole system is exposed to, well-informed supervisors can monitor aggregate financial-sector exposures and warn firms that are taking too much risk to cease and desist.8 The key is to be neither so intrusive that supervisors constantly substitute their judgment for that of the private sector nor to be so laissez-faire that they ignore systemic risk buildup.

  For such a system to work, we need better information. Currently, far too few financial institutions know on a daily basis what their risk exposures are: what might happen if interest rates move up by 25 basis points, if Italian government bonds are downgraded, or if a bank in Ohio is seized by regulators. As a consequence, the regulators and supervisors do not know, either. That AIG was in deep trouble seemed to be news to the regulators who were attempting to deal with Lehman’s impending failure—in part because AIG had found itself a weak regulator by buying a small savings and loan and ensuring its banking activities were regulated by the Office of Thrift Supervision.9 In this day and age, for a regulator to be uninf
ormed is unconscionable. Much of the process of acquiring and analyzing information can be automated. Regulators can require that this information be shared with them on a continuous basis, offering standard procedures and models with which to calculate the exposures. Standardization would be especially useful in the case of illiquid securities, assets, or positions, for which each institution currently calculates values and exposures in its own way, so that their reports are not comparable.

  Of course, supervisors rarely find the right balance between intrusion and laissez-faire, and political pressures tend to make them excessively lax in booms and conservative in downturns, just the opposite of what their behavior ought to be. If the gathered information were made public, however, it could offer a measure of public oversight over supervision. For instance, once the information about aggregate exposures and individual financial-firm exposures was put together, much of it could be shared with the markets, after some delay, in a way that could be easily digested. Supervisors would be forced to explain their actions (or inaction) if exposures were seen to be excessive.

  Regulatory authorities are often unwilling to reveal too much detail about exposures for fear that this may trigger the very panic they seek to avoid. Clearly, the wrong time to start revealing exposures is when the public is anxious about bank health. The right moment to start is in normal times, when no one is likely to panic. After that, if data on exposures are made public on a regular basis, the public can exert steady and healthy pressure on the financial firms.