CHAPTER SEVEN
ROUGHLY 60 PERCENT of all asset-backed securities were rated AAA during the lending boom, whereas typically less than 1 percent of all corporate bonds are rated AAA. How could this be, especially when the underlying assets against which the securities were issued were subprime mortgage-backed securities? Was this a sham perpetrated by the rating agencies?
Theory suggests it did not have to be a sham. In certain circumstances, a significant percentage of the securities issued against a package of low-quality loans can be highly rated.1 An example and some simple probability analysis can make the point. Suppose two mortgages, each with a face value of $1 and a 10 percent chance of total default, are packaged together. Suppose further that the investment bank structuring the deal issues two securities against the package— a junior security with face value of $1 that bears the brunt of losses until they exceed $1, and a senior security that bears losses after that.
The senior security suffers losses only if both mortgages default. If mortgage defaults occur independently (that is, they are uncorrelated), then the senior security defaults only 1 percent of the time. This is the magic of combining diversification with tranching the liabilities—that is, creating securities of different seniority. Put a sufficient number of subprime mortgages together from different parts of the country and from different originators, issue different tranches of securities against them, and it is indeed possible to convert a substantial quantity of the subprime frogs into AAA-rated princes, provided the correlation between mortgage defaults is low.
In normal times, the correlation between residential mortgage defaults is low, because people default only because of personal circumstances such as ill health or because they lose their jobs (for cause, rather than as part of a general layoff). No one really knew what that correlation would be in bad times, when many people might lose jobs because of the poor economy and house prices might fall across the country, making refinancing hard. If the correlation was still low, then the ratings were appropriate. If the correlation was high, then all bets were off— if, for example, the correlation was 1, then the senior securities would default as often as the junior securities, that is 10 percent of the time.
The AAA-rated tranches of mortgage-backed securities looked very attractive because they offered a higher return than similarly rated corporate securities. But some should have paid a far higher return because they were in fact very risky. Default correlations were much higher than the rating agencies or investors anticipated. First, the quality of the originated mortgages was low, and many borrowers relied on refinancing as house prices rose to make their payments, so a fall in house prices and the drying up of refinancing almost ensured default for many. Second, far too many packages were poorly diversified across areas: too many mortgages came from the same suspect, aggressive broker from the same subdivision in California.
Indeed, the fact that so many banks were exposed to the same diversified pools increased the likely default correlations, for banks across the country would simultaneously cut back on mortgage lending and refinancing if there was a problem in the market. This collective response would ensure that the problem spread across the country. Of course, the good times gave no inkling of the size of the problem, because in an atmosphere of rising prices and easy refinancing, no one defaulted. Much like a financial Venus flytrap, though, AAA mortgage-backed securities masked their risk with their ratings, and their attractive returns drew in many an investor innocent about finance and many more who should have known better.
Among the firms that should have understood the risk better was the American International Group (AIG). Its now-infamous financial products unit (AIGFP) sold insurance through credit-default swaps on billions of dollars of asset-backed securities, including senior (AAA-rated) tranches of the mortgage-backed securities described above. It promised buyers of the swaps that if the insured securities defaulted, AIGFP would make good on them. The unit was thus betting that defaults would be far rarer even than the market anticipated. Privately, AIGFP executives said the swaps contracts were like selling insurance for catastrophic events that would never happen: they brought in money for nothing! As was widely reported in the media, AIG recognized billions of dollars of profits over this period, and AIGFP's head, Joseph Cassano, pocketed over $200 million in compensation.2
However, in 2007 and 2008, the asset-backed bonds that AIGFP insured plummeted in value as the economy slid into recession, mortgage-default correlations proved larger than anticipated, and defaults became more likely. Even though few bonds actually defaulted, AIGFP's liability on the swaps it had written increased steadily as it became more likely that AIGFP would have to pay out. As late as 2007, Cassano maintained confidently that “it is hard for us, without being flippant, to even see a scenario within any kind of realm of reason that would see us losing $1 in any of those transactions,” even while AIGFP was losing billions of dollars as it had to mark its portfolio down.3 Eventually, the losses became too heavy to ignore, and Cassano was let go. But he wasn't fired: he “retired,” with a contract paying him $1 million a month for nine months and protecting his right to further bonus payments. AIG's counterparties started demanding collateral to ensure that AIG would make good on its swap liabilities. In September 2008 AIG started the process of becoming the recipient of the largest monetary bailout in U.S. history, receiving more than $150 billion from the U.S. government.
Although it is not surprising that risky mortgage-backed securities were created, it is surprising that seemingly sophisticated financial institutions, including those who originated these securities, held on to significant portions of them. These were typically AAA-rated securities, which had some default risk associated with them. Financial firms also took on other kinds of default risk, such as the securities issued by the collateral loan obligations where they had parked the loans made to finance acquisitions and buyouts. To top it all, many of these investments were financed with extremely short-term debt, ensuring that if problems emerged with the asset-backed securities, the financial firms would have immense problems rolling over their debt.
Why did financial firms take on both the default risk associated with highly rated asset-backed securities and the liquidity risk associated with funding long-term assets with short-term borrowing? As I explain in this chapter, the particular way these risks were constructed made them especially worth taking for large banks—indeed, perverse as it may seem, it made sense for banks to combine both risks.
There was something special about the nature of these risks. Clearly, banks felt that default was unlikely. Not only were the securities issued against a diversified pool of mortgages or loans, but also the securities the banks held (or that AIG guaranteed) were senior, highly rated ones, so that defaults on the mortgages or loans had to be numerous to trigger off default on the securities. Similarly, the chances that financing would dry up were also deemed small. These risks were, then, what are known as tail risks, because they occur in the tail of the probability distribution—that is, very rarely.
A second feature of these risks, though, was that systemwide adverse events would be necessary to trigger them: to cause the senior securities to default, mortgages across the country would have to default, suggesting widespread household distress. Similarly, funding would dry up for well-diversified, large banks only if there was a systemwide scare. A third feature, perhaps the most important one from society's perspective, is that these risks are very costly when they are realized, so they should not be ignored despite their low probability.
Unfortunately, these very features of systemic tail risks ensure that they are ignored by both financial firms and markets. Ironically, this also increases the probability that they will occur. When bankers attribute their problems to an unlikely event akin to a one-in-ten-thousand-year flood (thereby implicitly absolving themselves, for who could anticipate such a rare event?), they neglect to mention that their actions have increased the probability of such an event—to something like one in every ten years, approximately the periodicity with which Citibank has gotten itself into trouble in the past three decades.
I describe here how the structure of incentives in the modern financial system leads financiers to take this kind of risk. I next discuss why the corporate governance system did not stop such risk taking, and why various markets, especially markets for bank debt, were also unperturbed.
To understand the structure of incentives in the financial sector, we have to understand the relationship between risk and return. The central tenet in modern finance is that investors are naturally risk averse, so in exchange for taking on more risk, especially risk that may hit them when they are already in dire straits, they demand a higher return. Therefore, riskier assets tend to have lower prices (per dollar of future expected dividend or interest that they pay) and thus produce higher expected returns: stocks typically return more than Treasury bills. There is therefore an easy way for a banker or fund manager to make higher average returns for his investors; all he has to do is take on more risk by buying stocks instead of Treasury bills. This means the relative performance of a fund manager cannot be judged by returns alone: they must be adjusted downward in proportion to the risk being taken.
The bread-and-butter work of financial economists is to build careful econometric models describing the “appropriate” or market-determined level of return for taking on a certain level of risk. Financial managers are deemed to outperform the market only if they beat this benchmark return. The lay investor's version of such benchmarking is to compare the manager's return with a return on a benchmark portfolio consisting of similar securities: for example, the returns generated by a fund manager investing in large U.S. firms will be compared with the return on the S&P 500 index of large U.S. stocks. Such benchmarking is logical, because the investor can easily achieve the returns on the S&P 500 index by buying a low-cost index fund, and a manager should not earn anything for merely matching this return. Instead, investors will reward a manager handsomely only if the manager consistently generates excess returns, that is, returns exceeding those of the risk-appropriate benchmark. In the jargon, such excess returns are known as “alpha.”
Why should a manager care about generating alpha? If she wants to attract substantial new inflows of money, which is the key to being paid large amounts, she has to give the appearance of superior performance. The most direct way is to fudge returns. In recent times, some fund managers, like Bernard Madoff, simply made up their numbers, while others who held complex, rarely traded securities attributed excessively high prices to them based on models that had only a nodding acquaintance with reality. But it is easy to track and audit the returns most financial managers generate, so fudging is usually not an option, even for those with consciences untroubled by committing fraud. What, then is a financial manager to do if she is an ordinary mortal—neither an extraordinary investor nor a great financial entrepreneur—and has no bright ideas on new securities or schemes to sell?
The answer for many is to take on tail risk. An example should make the point clear. Suppose a financial manager decides to write earthquake insurance policies but does not tell her investors. As she writes policies and collects premiums, she will increase her firm's earnings. Moreover, because earthquakes occur rarely, no claims will be made for a long while. If the manager does not set aside reserves for the eventual payouts that will be needed (for earthquakes, though rare, eventually do occur), she will be feted as the new Warren Buffett: all the premiums she collects will be seen as pure returns, given that there is no apparent risk. The money can all be paid out as bonuses or dividends.
Of course, one day the earthquake will occur, and she will have to pay insurance claims. Because she has set aside no reserves, she will likely default on the claims, and her strategy will be revealed for the sham it is. But before that, she will have enjoyed the adulation of the investing masses and may have salted away enough in bonuses to retire comfortably to a beach house in the Bahamas. With luck, if the earthquake occurs in the midst of a larger cataclysm, she can attribute her disastrous performance to a one-in-ten-thousand-year event and be back in another job soon. Failing in a herd rarely has adverse consequences.
More generally, at times when financing is plentiful, so that there is immense competition among bankers and fund managers, the need to create alpha pushes many of them inexorably toward taking on tail risk. For tail risk occurs so rarely that it can be well hidden for a long time: a manager may not even be aware he is taking it. But the returns are high, because people are willing to pay a lot to avoid being hit by cataclysmic losses in bad times. So if the manager produces the returns but his investors do not (at least for a while) account for the additional risk the manager is taking with their money, the manager will look like a genius and be rewarded handsomely. He may well come to believe that he is one. In other words, it is the very willingness of the modern financial market to offer powerful rewards for the rare producer of alpha that also generates strong incentives to deceive investors.
Because these incentives are present throughout the financial firm, there is little reason to expect that top management will curb the practice. Indeed, the checks and balances at each level of the corporate hierarchy broke down. What is particularly pernicious about tail risk is that when taken in large doses, it generates an incentive to take yet more of it. A seemingly irrational frenzy may be a product of all-too-rational calculations by financial firms.
A well-managed financial firm takes calculated and limited risks, risks that will make money for the firm if they pay off but will not destroy the firm if they do not. Firms like AIG, Bear Stearns, Citigroup, and Lehman Brothers took risks that were virtually unbounded, albeit low in probability. The most obvious factor driving this behavior seems to be the compensation system, which typically paid hefty bonuses when employees made profits but did not penalize them significantly when they incurred losses. The profitable one-sided bet this offered employees was known variously as the Acapulco Play, IBG (I’ll be gone if it doesn't work), and, in Chicago, the O’Hare Option (buy a ticket departing from O’Hare International Airport: if the strategy fails, use it; if the strategy succeeds, tear up the ticket and return to the office). That such strategies were common enough in the industry as to have names suggests that not all traders were oblivious to the risks they were taking.
The Swiss banking giant UBS ran into trouble because its investment banking unit became entranced by the profit it was making from borrowing at the AA-rated bank's low cost of funding and investing the funds in higher-return, high-rated asset-backed securities.4 The regulatory requirements for bank capital to be set aside to back such a strategy were minimal because the underlying investments were highly rated. The resulting interest spread was small, but multiplied by the $50 billion the unit invested in the strategy, it made a tidy profit for the bank while the going was good and resulted in large bonuses for the unit. Needless to say, this practice of picking up pennies in front of a steamroller was successful only until the subprime catastrophe rolled all over UBS's profits.
Some smart traders in a number of banks understood and grew increasingly concerned by the risks that were being taken by the units creating and holding asset-backed securities. At Lehman, for example, fixed-income traders started selling these securities short, even while the real estate and mortgage unit loaded up on them.5 Clearly, any unit that is focused on creating and holding a certain kind of asset is naturally reluctant to declare an end to the boom it has ridden. The unit's size, power, and reputation become too closely related to the asset class, and its head becomes an interested booster. For Lehman's mortgage unit to declare an end to the mortgage boom would have been to sign its own death warrant. But knowing that those close to the action may become unreliable in assessing the associated risks, a firm's risk managers should step in to curtail further investment. In many firms they did not, and it is important to understand why.
Risk managers should adjust every unit's returns down for the risk it takes, reducing perverse incentives to take risk. The kinds of risks that were taken in the recent crisis—default or credit risk and liquidity risk—were not difficult for a trained risk manager to recognize, so long as she could see the unit's books. For risk managers to become concerned, however, and for top management to share their worry can be two very different things.
In many of the aggressive firms that got into trouble, risk management was used primarily for regulatory compliance rather than as an instrument of management control. At Citigroup, for example, risk managers sometimes reported to operational heads who were responsible for revenue, putting the fox in charge of the chicken coop.6 Reflecting the typical firm's view of their importance, risk management positions were paid significantly less than positions in operational units, thereby ensuring that they attracted less talented people who commanded less respect: not surprisingly, studies show that firms where risk managers were not independent of the operational units and were underpaid relative to other managers performed poorly in the crisis.7 Their weakness was compounded as the boom continued. When a CEO adjudicated a dispute between his star trader, who had produced $50 million in profits every quarter for the past ten quarters, and his risk manager, who had opposed the trader's risk taking all along, the natural impulse would be to side with the trader. The risk manager was often portrayed as the old has-been who did not understand the new paradigm—and the risk takers had the track record to prove it.
I remember a meeting between risk managers of major banks and academics in the spring of 2007 at which we academics were surprised that the managers were not more worried about the risks stemming from the plunging housing market. After our questions elicited few satisfactory replies, one astute veteran risk manager took me aside during a break and said: “You must understand, anyone who was worried was fired long ago and is not in this room.” Top management had removed all those who could have restrained the risk taking precisely at the point of maximum danger. But if that were the case, then the blame for encouraging the bet-the-firm tail risk taking that was going on must lie with top management.
What was management thinking? An obvious answer is that they, like their traders, were taking one-way bets. However, an intriguing study suggests that bank CEOs in some of the worst-hit banks did not lack for incentives to manage their banks well.8 Richard Fuld at Lehman owned about $1 billion worth of Lehman stock at the end of fiscal year 2006, and James Cayne of Bear Stearns owned $953 million. These CEOs lost tremendous amounts when their firms were brought down by what were effectively modern-day bank runs. Indeed, the study shows that banks in which CEOs owned the most stock typically performed the worst during the crisis. These CEOs had substantial amounts to lose if their bets did not play out well (no matter how rich they otherwise were). Unlike those of some of their traders, their bets were not one-way.
One explanation is the CEOs were out of touch. An unflattering portrayal of Fuld has him holed up in his office on the 31st floor of Lehman's headquarters with little knowledge of what was going on in the rest of the building.9 Indeed, in a tongue-in-cheek op-ed piece in the New York Times, Calvin Trillin argued that Wall Street's problem was that it had undergone a revolutionary change in the quality of personnel over generations.10 In Trillin's time in college, only those in the bottom third of their university class used to go on to Wall Street careers, which were boring and only moderately remunerative. But even while the dullards ascended to the top positions at the banks, Wall Street became a more exciting and challenging place, paying people beyond their wildest dreams. It started attracting and recruiting the smartest students in class, people who thought they could price CDO squared and CDO cubed (particularly egregious forms of securitization involving collateralized debt obligations) and manage their risks. As Trillin writes: “When the smart guys started this business of securitizing things that didn't even exist in the first place, who was running the firms they worked for? Our guys! The lower third of the class! Guys who didn't have the foggiest notion of what a credit default swap was. All our guys knew was that they were getting disgustingly rich, and they had gotten to like that.”11
The suggestion that bosses, recruited in a staid and regulated era, were of lower caliber than the employees they had recruited from the top of the class in a deregulated and high-paying era is not completely without foundation. An intriguing study of the U.S. financial sector indicates that the earnings of corporate employees in the financial sector relative to employees in other sectors started climbing around 1980, as the sector was deregulated.12 Moreover, jobs became more complex in the financial sector, requiring significantly more mathematical aptitude. Indeed, although there is little divergence between the wages of financiers and engineers at the college level, there is significant divergence among postgraduates (with postgraduate financiers increasingly earning more than postgraduate engineers). MBAs and PhDs began to fill the ranks of analysts and managers in financial firms. Therefore, not only was the financial sector demanding more highly educated people, but it was also paying them more and therefore probably attracting better talent than it had in the past— consistent with my observations that many of the smarter students in my MBA classes gravitated to finance. Clearly, deregulation and the subsequent surge in competition and innovation increased the demand for, and hence returns on, skills in the financial sector.
Although it is tempting to conclude that some of the CEOs were both untalented and clueless relative to their subordinates, the corporate hierarchy is inherently a tough climb and weeds out a lot of incompetents, especially in the unforgiving and fiercely competitive financial sector. It is hard to imagine that the majority of top management in the early 2000s, most of whom had probably joined in the already exciting 1980s and survived a number of ups and downs, were not highly capable and intelligent individuals. Sheer incompetence among the top management does not explain the crisis.
A better explanation is that CEOs were vying among themselves for prestige by making more profits in the short term or by heading league tables for underwriting or lending, regardless of the longer-term risk involved. I wrote a paper describing such incentives following bank troubles in the early 1990s, and I think the phenomenon is more widespread.13 Stan O’Neal, the CEO of Merrill Lynch, pushed his firm into the seemingly highly profitable asset-backed securities business in an attempt to keep up with rivals like Goldman Sachs. He monitored Goldman's quarterly numbers closely and often questioned colleagues on the companies’ relative performance.14 Merrill's lack of experience in the area eventually resulted in enormous losses and a shotgun marriage with Bank of America.
The pressures on the CEO may have come not just from shareholders or personal egos but also from aggressive subordinates. Citigroup CEO Chuck Prince's comment in July 2007, only a month before markets started freezing up, has become emblematic of CEOs’ role in the current crisis. Replying to a journalist who asked why his bank continued to make loans on easy terms to fund takeovers, he said: “When the music stops, in terms of liquidity, things will be complicated. But as long as the music is playing, you’ve got to get up and dance. We’re still dancing.”15
This comment was commonly interpreted as reflecting the cavalier attitude of bankers toward risk and the mad chase for immediate profits. Months later, I met Prince at a conference where we were on a panel together, and I asked him what he had meant. He explained that even though he knew there were risks, as the first sentence of the quote suggests, he simply could not shut down lending, which was critical to securing investment banking deals: the moment he did so, he would have lost many key employees to other rivals who were still “dancing.” So the decision to continue lending was not so much an attempt to make shortrun profits as an attempt to preserve Citigroup's franchise in investment banking and its capabilities for the future. Of course, in making the kind of loans they did, his employees jeopardized not only their unit's franchise but the entire bank. Hindsight suggests that Prince and Citigroup would have been better off if they had sat out a few dances.
A few CEOs appear to have stood up to their employees. The CEO of JP Morgan, Jamie Dimon, played a key role in preventing his bank from taking a bigger position in highly rated mortgage-backed securities, and in unwinding its existing positions, beginning in 2006.16 As he often emphasized to his staff, “We have got to have a fortress balance sheet!…No one has the right to not assume that the business cycle will turn! Every five years or so, you have got to assume that something bad will happen.”17 He also beefed up pay for risk managers, so that these positions attracted knowledgeable traders. He tried to ensure that they had clout. And although he had a much deeper understanding of derivatives than many of his fellow CEOs, he also had a rule: if he did not understand how a business made money, he would not participate in it. Not taking risks one doesn't understand is often the best form of risk management. Firms with less confident or respected CEOs simply followed the herd over the cliff, pushed by the ambitions of their employees.
But before we attribute too much or too little foresight to CEOs, let us consider the findings of another sobering recent study, which looks at total top management pay across financial institutions before the crisis and its relationship with subsequent performance.18 The study finds that some firms tended to pay their top management a lot more aggressively in the period 1998–2000, correcting for obvious factors like the size of the bank (big banks pay more because they tend to attract, and need, better talent). Aggressive payers included the usual suspects like Bear Stearns, Lehman, Citigroup, and AIG, whereas more conservative paymasters included firms like JP Morgan. The study finds that those who paid the most aggressively before the crisis were also those who had the worst stock-return performance during the period 2001–2008, the highest stock-return volatility, the highest exposure to subprime mortgages, and, by some counts, the highest leverage. Aggressive pay practices seem to have gone together with aggressive risk taking and subsequent poor performance during the crisis, much as my earlier discussion suggests.
Interestingly, though, the researchers repeated the exercise over a different time frame, looking at how those who compensated aggressively during the 1992–94 period fared between 1995 and 2000. Over this period, the same firms were aggressive payers, but they did phenomenally better than the conservative payers. Their stock returns were much higher, though measures of risk, such as stock-return volatility, were also high. The authors conclude that performance did not depend on the astuteness or incompetence of particular CEOs: rather, some banks had a culture of risk taking and of compensating very heavily over the short term, which attracted like-minded traders, investors (aggressive banks had more short-term institutional investors holding their shares), and even CEOs. When these banks did well during boom times, their CEOs were lionized as heroes; but when they did extremely poorly during the credit crisis, their (usually former) CEOs became villains. The CEOs were probably neither. They were just loading up on risk, including tail risk; but this time it just did not pay off.
Past experience may even have led CEOs to overestimate their ability to deal with tail risk. A passage from a New York magazine article about Lehman is revealing:
But given the importance of real estate to Lehman's bottom line, that wasn't what Fuld wanted to hear. Fuld had seen his share of cyclical downturns. “We’ve been through this before and always come out stronger,” was his attitude. “You’re too conservative,” Fuld told Gelband.
“We’ve been lifted by the rising tide,” Gelband insisted.
Fuld, though, wondered if the problem was with Gelband, not the market. “You don't want to take risk,” he said—a deep insult in the trader's vernacular.19
Soon Gelband was fired, and Lehman continued piling up risk. In its last days, it brought back Gelband to try to save the bank, but it was too late, and Lehman was bankrupted by the panic of 2008. More generally, aggressive banks’ risk taking had paid off in the past, which is why their richly compensated CEOs were sitting on enormous amounts of equity. They did not seem to realize that it was risk, not capabilities, that had brought them their past returns. And this time when they rolled the dice, what turned up was very different.
Although it would be too strong to say that CEOs had little influence—Stan O’Neal converted staid Merrill into an aggressive risk taker—perhaps the most important thing any CEO did was to arrive in the right CEO suite. Somewhat tellingly, Jamie Dimon parted ways with Citigroup and, by way of Bank One, joined the conservative JP Morgan. He tightened processes considerably at JP Morgan, perhaps partly because his admonitions fell on receptive ears. It is unclear whether he would have had the same influence at Citigroup. A New York Times columnist, Andrew Ross Sorkin, reports a conversation between Bob Willumstad, the CEO of AIG, and Robert Gender, its treasurer, as AIG was running out of money: “It was then that Willumstad accepted the fact that JP Morgan might not be willing to provide any further funds. AIG's treasurer, Robert Gender, had already warned him that that might be the case, but Willumstad hadn't fully believed him. ‘JP Morgan's always tough,’ he reminded Gender. ‘Citi will do anything you ask them to do; they just say yes.’ But the prudent Gender only acidly replied, ‘Quite frankly, we can use some of the discipline that JP Morgan is pushing on us.’ ”20
In sum, the pattern of tail risk taking in some aggressive banks paid off for a considerable time. The management of these banks does not appear to have realized how much their performance depended on luck, or how their own collective actions precipitated the events they should have feared.
One question arises immediately: If indeed the aggressive banks were clearly identifiable, why did the market not punish them before the crisis? Banks that ranked in the worst quartile of performance during the crisis had much higher stock returns in the year before the crisis, 2006, than banks that ranked in the best quartile.21 So the market seemed to support the behavior of the risk takers by boosting their stock price before the crisis.
Those who believe that markets are grossly inefficient would quickly construe this outcome as evidence that the stock market typically gets it wrong, and that theories that markets efficiently aggregate all public information into prices —versions of the “efficient markets hypothesis”—are hopelessly misguided. Yet nothing in the theory says the market should be spot on all the time. The market may not have full information—after all, even regulators were later surprised by the quantities of asset-backed securities the banks carried both on and off their balance sheets. Moreover, even if it assigns appropriate probabilities to all possible events, only one of those events will be realized. Viewed with the benefit of hindsight, especially if an extreme event occurs, the market will seem as if it got matters wrong, and indeed it will have done so. But this is not to say that anyone could have consistently done better. In the jargon of economists, that the market is believed to have rational expectations about events does not mean that it has perfect foresight.
More generally, there is a danger in judging risk taking while looking back from the depths of a crisis, especially one as severe as the recent one. From that perspective, any risk taking beforehand seems irresponsible, redolent of mismanagement. Conservatism seems prescient and astute—indeed, it seems so much in tune with the times that it becomes the strategy of choice after the crisis, when in fact more risk taking would be appropriate. However, the right way to judge actions taken before the crisis is whether the risk taking was expected to be profitable.
And it may well have been that shareholders, protected by limited liability from bearing the extreme losses induced by tail risk (because shareholders can simply abandon their shares when their value hits zero, whereas partners in an unlimited-liability partnership must repay the money owed to debt holders or forfeit their wealth), deemed the expected profits from taking on tail risk worthwhile—they took the gains while the debt holders and the taxpayer absorbed the tremendous losses. Put differently, Jimmy Cayne (of Bear Stearns), Dick Fuld (of Lehman), and Chuck Prince (of Citigroup) might still be feted as giants of the financial industry had events followed the most probable course. This is not to say that the risks they took on were good for society, only that they may have been reasonable bets for shareholders to take.
The actions of corporate boards, which are the representatives of shareholders, might give us a sense of where shareholder interests lay. Not all boards were equally competent, but Citigroup's board, with stalwarts like Robert Rubin, the former treasury secretary and CEO of Goldman Sachs, might give us an inkling as to what knowledgeable shareholders might have opted for. There is evidence that this board pushed Citigroup into taking more of the risk that brought it to its knees.22 Although we cannot tell whether the board was independent or in management's pocket, it apparently did not restrain the bank's risk taking.
Finally, equity markets were not entirely unaware of the risks. From the second quarter of 2005 to the second quarter of 2007, the two-year implied volatility of S&P 500 options prices—the market's expectations of the volatility of share prices two years ahead—was 30 to 40 percent higher than the short-term one month volatility.23 This figure suggests that the market expected the seeming calm would end, even though the high level of the market indicated it did not place a high probability on events turning out badly for shareholders. But this is precisely how we would expect the market to behave if it believed the banks were taking on subsidized tail risk.
Thus far, as we have moved through the corporate hierarchy, from trader to risk manager to CEO to corporate board to shareholder, we have found little concern anywhere about the tail risks that were building up, especially in the aggressive banks. Many of the actors—traders, management, and shareholders— typically focused on the advantages of taking the tail risk. In insurance parlance, they would get a share of the premiums that flowed in while the going was good, and they would be protected by limited liability from having to make massive payouts if the extreme risk hit. Who then would absorb the losses?
Typically, the answer ought to be the bank's debt holders. In the case of commercial banks, some were FDIC-insured deposit holders, who would not bear losses in any case, while others protected themselves by lending short term and demanding security to back their lending. If defaults on asset-backed securities mounted, the short-term lenders thought they would be able to withdraw ahead of the collapse. But not all of them could hope to escape without taking a hit. Why were they not more worried?
Similarly, why were holders of long-term, unsecured debt not extremely fearful, especially given the higher future expected volatility reflected in share options? Bank debt holders typically hate volatility, as they get none of the upside gains and bear all the downside risk. Bank debt spreads, a measure of a bank's anticipated risk of default, remained very moderate until just before the crisis.
It is hard to argue that debt holders were ignorant of the risks, especially when equity options markets seemed to be signaling possible trouble. The obvious explanation for their continued exposure to risk is that debt holders did not think they would need to bear losses because the government would step in. There were two possible reasons for this complacency—reasons that were in fact borne out by events. First, unsecured bondholders worried less than they should have because of the prospect of direct government intervention in housing and credit markets if matters took a turn for the worse. Second, the institutions that took the most risk were those that were thought to be too systemic to be allowed by the government to fail. The bank's debt holders would not have had to face any risk of default if the government bailed the firm out. Not only would confidence in a bailout have kept debt costs from rising in proportion to bank risk taking, but also, with little concern expressed by debt holders, management had an even broader license to take on leverage to boost returns for equity.
Consider the nature of the tail risks. Unlike ordinary loans or individual mortgages, where defaults occur in isolation, highly rated, diversified mortgage-backed securities were likely to risk default only if mortgages across the country defaulted. If such an improbable eventuality were to occur, the government would likely be drawn in to supporting the market for housing and for housing finance: it could not possibly sit idly by as millions of homeowners defaulted. Similarly, when such a systemic event occurred, not only would large banks find refinancing difficult, but corporations ranging from the large to the tiny would also face significantly greater financial constraints. Again, it was unlikely the Fed would stand by idly if liquidity vanished, especially given the promises Chairman Greenspan had made. So the systemic nature of tail risks ensured that banks would be collectively in trouble if a crisis occurred, and that government support would be forthcoming. This mitigated the costs of those risks.
And that support has indeed been forthcoming. Specifically, in order to support the housing market, the federal government has introduced tax measures that encourage home ownership, including the first-time home buyer's tax credit. Since September 2008, Fannie Mae, Freddie Mac, and the Federal Housing Authority have lent hundreds of billions of dollars to low-income borrowers in an attempt to keep house prices from collapsing. The financial website Bloomberg.com estimates that in 2009, the Fed and the Treasury together purchased $1.3 trillion worth of agency-issued mortgage-backed securities, 76 percent of the gross issuance (including refinancings), and more than three times the net increase in the size of the market. Estimates suggest that these purchases lowered mortgage rates by about 75 basis points.24 The special inspector general's report to Congress on the Troubled Asset Relief Program (TARP) in January 2010 states: “The government has done more than simply support the mortgage market. In many ways it has become the mortgage market with the taxpayer shouldering the risk that had once been borne by the private investor.”25 The Financial Times reports Neil Barofsky, the special inspector general, as saying: “All of the things that were broken in the housing market and the different roles that different private players have played, some of which we recognize now…actually contributed to the bubble and to the ensuing crisis, are really being replicated by government actors.”26
Even outside the housing market, the Fed pulled out all stops, especially after the collapse of Lehman. The Fed cut interest rates to rock bottom and created a variety of innovative programs to lend to the private sector, while the Treasury recapitalized firms through TARP. The Federal Deposit Insurance Corporation (FDIC) also chipped in by temporarily insuring all bank debt in 2008 and upping the quantum of deposit insurance. Banks that remembered the Fed riding to the rescue in 2001 with rock-bottom interest rates, and Alan Greenspan's subsequent dictum that the Fed would “mitigate the fallout when it occurs and, hopefully, ease the transition to the next expansion,” were not wrong in anticipating that they would be helped out of a tight corner yet again. The banks that had taken on liquidity risk and survived were rewarded with substantial profits— but some did not live to see happy days return.
Perhaps an equally important driver of bank bond-holder complacency was the knowledge that the large banks they had lent to were likely to be deemed too systemic to fail by the authorities. Herd behavior put the issue largely beyond doubt: if many large banks took the same tail risks, they would all be weak at the same time, and the chances that the government would risk a panic by letting any one of them go would be proportionately diminished. Henry Paulson, secretary of the treasury and thus the leader of the rescue in the Bush administration, writes about the FDIC-led resolution of Washington Mutual in September 2008:
Unfortunately, the WaMu solution was not perfect, although it was handled smoothly using the normal FDIC process. JP Morgan's purchase cost taxpayers nothing, and no depositors lost money, but the deal gave senior WaMu debt holders about 55 cents on the dollar, roughly equal to what the securities had been trading for. In retrospect, I see that in the middle of a panic, this was a mistake. WaMu, the sixth-biggest bank in the country, was systemically important. Crushing the owners of preferred and subordinated debt and clipping senior debt holders only unsettled the debt holders in other institutions, adding to the market's uncertainty about government action.27
Thus even though shortly before WaMu's resolution the market fully expected debt holders would not be paid in full, and even though the FDIC had the full legal authority to impose losses on the bondholders, the secretary of the treasury expressed remorse over the action. Given that the only time a large, well-diversified bank can get into trouble is in the midst of a nationwide downturn and possible panic, the secretary's logic would protect bank bondholders from ever suffering losses and remove an important source of market discipline on banker actions.
In sum, if bank management had fully understood the risks they were taking, their decisions could have evolved as follows. In the early stages of taking on a tail risk, such as the default risk in mortgage-backed securities or the liquidity risk in borrowing short term to fund long-term assets, they would have proceeded cautiously and even surreptitiously. After all, the profits from such activities would look a lot healthier if no one knew the risks they were taking. Accordingly, Citibank's off–balance sheet conduits, holding an enormous quantity of asset-backed securities funded with short-term debt, were hidden from all but the most careful analysts.
But as enough banks imitated the innovators and took on similar risks, and as it became common wisdom among market participants that the market would be supported in the event of a crisis, there would have been strong incentives to load up on the tail risks, even if such activity became visible. The market would have focused on the profit potential of such risk taking, knowing that most of the losses, in the remote eventuality that they occurred, would be passed on to the government and the taxpayer. Indeed, an entity financed with short-term borrowing, knowing that it would be unable to repay its debts if liquidity dried up, had a powerful incentive to double up on its bet that liquidity would always be available —by buying assets that would fall in value if liquidity dried up and by leveraging even more. The simple reason is that limited liability protected its shareholders against losing more if its bets went wrong and liquidity did dry up; and if it guessed right, it would make a ton of money.28 No wonder exposures to both mortgage-backed securities and short-term leverage increased steadily before the crisis.
Anticipation of government interventions would have made it even harder for any bank to justify a conservative stance during the run-up to the crisis. For instance, one of the rewards of maintaining a very liquid balance sheet is that when liquidity dries up in markets, the bank can purchase assets at bargain basement prices from those who have been too aggressive. But if the Fed intervenes to lend freely at such times, the discount on assets is far less than it would otherwise be. Many troubled banks held assets through the crisis that they should have been forced to sell, reducing the punishment they suffered for maintaining illiquid balance sheets and reducing the potential gain to bottom-feeding conservative banks.
Finally, all this behavior increased the likelihood of the tail risk's materializing substantially. When few banks maintain liquid reserves even while leveraging their balance sheets to the hilt, the slightest adverse shock can tip the system over into a full-fledged panic. Similarly, as purchases of mortgage-backed securities increase without much attention being paid to default risk, mortgage lending expands and the quality of lending deteriorates, making widespread default more likely if house prices start dropping.
In all this, one cannot ignore the actions of the regulator. One of the factors propelling banks into mortgage-backed securities was their low capital requirements relative to direct lending. The market, however, priced these securities as if they were riskier than the regulators believed them to be (as indeed they were). Banks thus collected the higher return on these securities while maintaining little capital, thereby obtaining a seemingly healthy return on capital. In a sense, therefore, regulators inadvertently pointed banks toward these securities. In some ways practices like this are an unavoidable consequence of regulation. If banks have an incentive to take risk, they will always look for opportunities to get the greatest bang for the regulatory buck. But the regulatory mistake of requiring too little capital for certain activities is then compounded because in taking advantage of regulatory mistakes, banks build up exposure to the same risks. The dynamic associated with systemic risk exposures then kicks in: if everyone is exposed to the same risk in a big way, the authorities have no option but to intervene to support banks and the market if the risk materializes—in which case a bank maximizes profits by increasing exposure to the risk.
The problem of tail risk taking is particularly acute in the modern financial system, where bankers are under tremendous pressure to produce risk-adjusted performance. Few can deliver superior performance on a regular basis, but precisely for this reason, the rewards for those who can are enormous. The pressure on the second-rate to take tail risk, thus allowing them to masquerade as superstars for a while, is intense.
The market should theoretically encourage good risk management and penalize excessive risk taking. But tail risks are difficult to control for two reasons. First, they are hard to recognize before the fact, even for those who are taking them. But second, once enough risk is taken, the incentive for the authorities to intervene to mitigate the fallout is strong. By intervening, the authorities reduce market discipline, indeed inducing markets to support such behavior. Bankers may in fact have been guided into taking tail risks as markets anticipated government intervention in the housing market and liquidity and lending support from the Fed and the FDIC.
This argument is not meant to absolve bankers. Some understood the risk they were taking and ignored it; many did not recognize it but should have. What is particularly alarming is that the risk taking may well have been in the best ex ante interests of their shareholders. One should judge the Citigroup's board's competence not only by the fact that its share price sank below $1 in the midst of the crisis but also by the fact that the price reached the mid-$50s just before the crisis in the spring of 2007. The stock market is not an anonymous, distant entity: it is us, and collectively we feted activities that eventually proved highly detrimental to society. Indeed, bank CEOs who remained conservative were doing the right thing by society but quite possibly not by their shareholders. Certainly, this seems to have been the market's view before the crisis hit.
Put differently, solutions are fairly easy if we think the bankers violated traffic signals: we should hand them stiff tickets or put them in jail. But what if we built an elaborate set of traffic signals that pointed them in the wrong direction? We could argue that they should have used their moral compass, and some did; but, as I indicated in the previous chapter, the industry's entire system of values uses money as the measure of all things. Solutions are much more difficult if it turns out that the signals are broken, at least from the perspective of our collective societal interests.
Moreover, I do not mean to suggest, by attributing some of the crisis to bankers’ and market confidence in a government bailout, that the authorities should sit back and watch the economy collapse. Rather, I want to emphasize that the combination of incentives for high-powered performance that are inherent in the modern financial system and the unwillingness of a civilized government to let failure in the financial sector drag down ordinary citizens generates the potential for tail risk taking and periodic, costly meltdowns. Even as I write, the enormous amounts of taxpayer money being directed at the housing market and the banks are creating new expectations about government and Fed behavior in the next crisis. Our central focus in any reforms should be on dispelling such expectations, and that is the topic I turn to now.