Manias, Panics and Crashes

The bankruptcy of Lehman Brothers Holdings, the fourth largest US investment bank, in mid-September 2008 triggered the most severe financial panic in a century. Lehman had been aggressive buyer of mortgage-related securities using money obtained from short-term loans; its assets were more than thirty times its capital. Lehman was very profitable during the bubble years because of its high leverage and its aggressive posture toward risks. The rates of return to investors – to individual homeowners, property developers, and financial firms – that are extensively leveraged are very high as long as real estate prices increase. The paradox is that the leverage of some of these investors may decline even as their indebtedness increases because their net worth increases even more rapidly.

The earliest manias reviewed in this book occurred nearly four hundred years ago, and many of the manias and panic took place more than one hundred years ago. The 2008 crisis occurred in ‘real time’ – as the sixth edition of this book was under preparation.

When a bubble implodes, the market participants that are most vulnerable are those that have expanded most rapidly and have the greatest leverage. As real estate prices declined in the summer of 2008, Lehman was subject to two runs – investors and hedge funds sold their shares, both those they owned and those they could borrow, and owners of Lehman’s IOUs were reluctant to renew their loans to the firm as they matured. Both groups were concerned that Lehman seemed slow in raising capital to counter the decline in its net worth from the fall in the value of its mortgage-related securities.

When Lehman failed, the credit markets ‘froze’ – liquidity disappeared. Trading volumes declined sharply. There was the traditional rush to cash that occurs when the financial environment suddenly becomes highly uncertain; the firms that were in a position to extend credit were extremely reluctant to do so because they wanted to increase their liquidity and their holdings of cash and the firms like Lehman that relied extensively on borrowed short-term funds to finance their assets found it exceedingly difficult – and extraordinarily expensive – to re-finance their maturing IOUs. The risk premium embedded in the structure of interest rates soared; the interest rates in the interbank market increased by a factor of 30 to 40 relative to the interest rates on US Treasury bills, and the interest rates on high-risk bonds surged relative to those on US Treasury bonds.

The policy problem for the US government was the familiar one – should new measures be adopted to restore liquidity or instead should the government continue with a hands-off approach, even though asset prices would decline further, more financial institutions would fail, and the recession would be steep and perhaps prolonged. Eventually markets would stabilize; the presumption is that the next crisis would be less severe because the lenders would be much more cautious. In contrast, if the government intervened to stabilize the credit markets, the economic slowdown would be moderated but the next crisis would be more severe because lenders and investors would conclude that they would be ‘bailed out’ – whatever that means.

Bubbles result from the growth of credit that proves too high to be sustainable; when the credit growth slows, some asset prices begin to decline. US house prices peaked at the end of 2006. Between 2002 and 2006, the market value of the US residential real estate had increased from $16,000 billion to $23,000 billion, or from 110 percent of US GDP to 150 percent of US GDP – which was 15 percent higher in 2006 than in 2002. The sharp increase in property prices resulted from the ready availability of mortgage credit and led to a boom in the construction of new homes and condominium apartments; housing starts were one-third larger than the number associated with the growth of the population and the replacement of the homes lost to fires, storms, and highway construction.

The increase in the credit available for mortgages was facilitated by securitization, which had begun in the 1970s and initially involved depositing mortgages with similar attributes in terms of maturity in a trust, essentially its own corporation, and then issuing new securities (collateralized mortgage obligations or CMOs) that were claims on the interest income of the trust. Securitization accelerated after 2000; the second round involved ‘slicing and dicing’ groups of CMOs into four or five different tranches that differed in terms of whether they had the first, last or an intermediate priority to the interest income of the CMOs. The tranches with the first priority to the interest income of the CMOs were placed in one trust, the tranches with the second priority to the interest income were placed in a second trust, and so on. Conservative investors could buy the CDOs (collateralized debt obligations) that had the first claim to the interest income because they were the least risky, while investors that wanted higher yields could buy those that had lower priority to the interest income of the CMOs – and hence were riskier.

A priori it would seem that the CDOs that had the first claim to the income of the CMOs would be less risky than the CMOs, while the CDOs that had the last claim to the their income would be riskier. Somehow the credit-rating agencies concluded from their models that even those CDOs that had the last claim to the income of the CDOs were nevertheless less risky than the CMOs. This outcome seemed counter-intuitive, and after the collapse of the mortgage market, it appeared that the credit-rating agencies had relied on information about the riskiness of various CDOs that was supplied by the investment banks that were eager to satisfy the investors’ demand for securities that had high anticipated rates of return and yet were deemed not especially risky. The demand for free lunches is unending.

One consequence of securitization was that the amount of money available for mortgages increased because these mortgage-related securities were more liquid than individual mortgages, and the interest rates on mortgages accordingly were lower. The increase in the supply of money available for mortgages meant that some borrowers with less attractive credit histories qualified for mortgages because ‘the money was there’. Moreover there was substantial innovation about both the minimal down payments required of home buyers and the interest payments on mortgages – there were adjustable-rate mortgages or ARMs, and interest-only mortgages, and negative amortization mortgages (part of the interest payment that the borrower was scheduled to make to the lender was added to the principal of the mortgage). Moreover many of these mortgages had ‘teaser interest rates’ – the interest rates that the borrowers were scheduled to make in the first two or three years were exceptionally low and then would increase by two or three percentage points.

These innovations in the terms of mortgages were motivated because the demand for mortgage-related securities was significantly larger than the supply of loans from creditworthy borrowers – typical of the last several years of an expansive credit cycle.

Another consequence of securitization was ‘financial deepening’, the supply of mortgage-related securities increased more than twice as rapidly as the supply of mortgages. During this deepening process the investment banks carried large inventories of mortgages and mortgage-related securities in their ‘production and distribution pipelines’ awaiting sale to the investors; some of these pipelines were six months long. The investment banks carried the credit risk attached to these mortgages as long as they were in the pipelines, although some may have hedged this risk. Some of the investment banks ‘sold’ some of these mortgage-related securities to their captive affiliates if they could not be readily sold to non-related investors.

Securitization of mortgages also led to an anonymous relationship between the home-owner borrowers and the mortgage lenders that replaced the traditional one where the lenders knew the borrowers because they shared the same neighborhood. The new relationship invited fraud in the mortgage-origination process; some of the fraud was on the part of borrowers, who provided incomplete or misleading information about their incomes, assets and credit histories (which were known as ‘liar’s loans’). Some of the fraud was by the mortgage brokers, who encouraged home owners to take on indebtedness that was extremely high relative to their incomes. Some mortgage brokers were not concerned about the ability of the borrowers to repay because the credit risk would be transferred to others in the credit distribution pipeline. The eagerness of the credit rating agencies to be asked to provide their ranking of individual mortgage-related securities ‘clouded’ their objectivity. Moreover this anonymity complicated the ‘workout arrangements’ when borrowers had problems making their monthly payments.

At some stage it was inevitable that the booms in real estate in the United States, Britain, Ireland, Spain, and other countries would end, property prices would decline, new home construction would fall, and probably sharply, and recessions would follow. There was significant disagreement about the impacts of the recession and the slowdown in construction on house prices and whether the upward momentum would stall or whether instead real estate prices would decline. The view of many lenders was that property prices would not decline because, well, ‘US real estate prices had never before declined on a national basis’. (A factually incorrect statement since these prices had declined in the Great Depression.) Most of those involved in the real estate industry believed that there would not be a significant reduction in home prices. A second much smaller group – members of the ‘reversion-to-the-mean school’ – believed that prices would decline until the traditional ratio of the market value of US real estate to US GDP, plus or minus 10 percent, had been restored. A third view was that house prices would decline 10 to 15 percent below this traditional ratio because of the large excess supply of houses that had developed during the bubble years; these declines below the trend line would be larger in those regions with the largest excess supply.

Housing starts began to decline toward the end of 2006, which contributed to the slowdown in the US GDP growth rate. Still, the decline in US GDP in 2007 and the first eight months of 2008 was modest. The number of people employed peaked in December 2006, and thereafter the number declined continuously but modestly. During the first half of 2008 it appeared as if the United States would have a gradual slowdown that would be prolonged until the excess supply of homes and apartments was nearly absorbed – and that the slowdown would not be especially severe. Then the bankruptcy of Lehman on 15 September triggered the most severe crisis since the Great Depression.

The Japanese experience in the 1990s suggests that declines in asset prices need not led to a crisis or panic, even though the metric of the increase in the ratio of household net worth to GDP in Japan was more than ten times larger than that in the United States – both stock prices and real estate prices in Japan had increased by a factor of five to six while US real estate prices had increased by 40 percent. Although the banks and most of the other financial institutions in Japan had negative net worth in any meaningful mark-to-market test, there never was a run on a major Japanese bank, despite very modest levels of deposit insurance – at least in a formal sense.

Japan was different because it was ‘understood’ that bank depositors would not incur losses even if a major financial institution were closed. The tradition in Japan is that many losses are socialized that would fall on individuals in the United States and in other western industrial countries – the losses are made good by the government and ultimately by the taxpayers. In effect Japan had 100 percent deposit insurance and the Japanese institutions were ‘too big to fail’ even though the shareholders in many of these banks eventually would lose most or all of their money.

The first group impacted by the decline in US real estate prices was those who had purchased homes in 2005 and 2006, often first-time homeowners with spotty credit histories and modest if any down payments. If these borrowers had difficulties in making their monthly payments, they would sell their properties and repay their mortgage indebtedness – as long as property prices continued to increase.

The share of subprime mortgages in total mortgages increased from 6 or 7 percent of total mortgages in 2003 and 2004 to 20 percent in 2005 and 2006. Many of these subprime mortgages required very low interest payments in the first several years. Once property prices began to decline, many individuals that had purchased properties with minimal down payments would have ‘upside-down mortgages’ – their loan indebtedness would be greater than the market value of their homes. These homeowners then would have three choices – they could continue to make the monthly mortgage payments, they could engage in a short sale, that is sell the property and hope that the lender would ‘forgive’ the excess of the indebtedness over their sales proceeds or they could get involved in ‘jingle mail’ – return the keys to the lenders and walk, leaving the lenders with the costs and burdens of foreclosure. The larger the decline in house prices, the greater the decline in the net worth of home owners, and the higher the number of mortgage loans that would become non-performing.

The second group impacted by the decline in property prices were the brokers that bought the mortgage loans from the homeowners with the intent to sell them to the investment banks, who had twelve to eighteen months to put these mortgages back to the brokers if the borrowers stopped making their monthly payments. A few mortgage brokers that had extended loans to subprime borrowers failed in 2005, many more failed in the last few months of 2006. The failure of these brokers meant that the investment banks were stuck with non-performing loans that they had recently acquired and would have returned to the brokers if they had not gone bankrupt.

The investment banks were the third group affected by the defaults of the homeowners, and in several additional ways beyond being stuck with the non-performing mortgages that they had recently acquired. Once real estate prices began to decline, it became much more difficult to sell mortgage-related securities because they were much harder to value. As mortgage defaults increased, the investment banks incurred losses on their holdings of tens of billions of mortgages and mortgage-related securities, including some CMOs and CDOs that they had been unable to sell.

Once an investment bank was a distress seller of mortgage-related securities, their prices declined sharply. The ‘mark to market’ convention was that firms that owned identical securities had to value them at the recent market price. There was an immense anomaly in that the market prices of these securities were much lower than the prices based on their current and prospective cash flows.

Subsequently the commercial banks that had acquired significant mortgages in a few states failed because of large losses. The Federal Deposit Insurance Corporation closed those banks that had exceptionally large losses relative to their capital; their deposits were sold to banks that had a secure capital position. Nearly two hundred commercial banks had failed by the end of 2009. The money that had been collected by the FDIC from the insurance premium paid by the banks was used to ensure that no depositor incurred a loss. The reserves that they had accumulated from these premium payments were being depleted at a rapid rate.

The FDIC had a $500 billion credit line at the US Treasury; it had rarely drawn on this line. If this line were exhausted, the Congress almost certainly would vote more funds to the FDIC. In anticipation that the losses would exceed its premium payments the FDIC wanted banks to make advance payments of the premiums.

Several of the hedge funds managed by Bear Stearns failed in March 2007; these funds owned mortgage-related securities and were extensively leveraged. Then in mid-August Countrywide Financial, the largest mortgage lender in the United States, had a liquidity problem; the firm had sought to rapidly increase its share of the national market for mortgages, and had relied on money borrowed in the commercial paper market to finance about one-third of the mortgages that it had acquired. Countrywide was unable to renew its commercial paper loans as they matured, and initially it drew on the credit lines that it had with commercial banks. Then Countrywide sold a new issue of preferred shares to Bank of America, and a few weeks later Bank of America purchased Countrywide. About the same time Northern Rock, the largest mortgage lender in Britain, began to borrow from the Bank of England because it also was not able to re-finance its IOUs in the commercial paper as they matured.

The Bagehot doctrine and insolvent banks

The Bagehot doctrine discussed in Chapter 11 was that a central bank should lend freely – but at a penalty interest rate – to forestall the likelihood that a liquidity crisis would cascade into a solvency crisis because a commercial bank would sell assets to get the cash to meet the demands of their depositors that wanted their money. The prices of these assets would decline, and some of the banks that previously had been well-capitalized might then have too little capital as the value of the assets in their portfolios fell.

The extension of credit from the central bank to a bank that has a shortage of liquidity is a ‘bail-out’ of its shareowners; otherwise this bank might be forced to close and its shareholders would lose all of their money.

Bagehot probably would have advised against the extension of financial assistance to an insolvent bank. Assume, however, that Alpha Bank – or Lehman or the Royal Bank of Scotland – is insolvent. Should the government provide financial assistance to ensure that Alpha Bank can continue to operate even if it has a negative net worth – as the US Treasury did for Fannie Mae and Freddie Mac? If Alpha Bank goes bankrupt, the legal procedure is that its assets would be sold to get the money to satisfy its creditors, the prices of these assets would decline, and some previously solvent institutions then might have too little capital.

The logic that Bagehot used to justify why the central bank should provide credit to an illiquid bank that is solvent also justifies why the government should provide financial assistance to an insolvent bank. Both arguments centered on the impacts of the failure of one financial institution on the solvency of other financial firms.

Early in 2008 Bear Stearns, the sixth largest US investment bank, was subject to two runs, one on its share price by investors who thought the firm might go bankrupt and another by the holders of its short-term IOUs who were reluctant to renew their loans to Bear. Bear was acquired by JPMorganChase, but only after the Federal Reserve had agreed to take up to $29 billion of losses on some of the ‘toxic waste’ that had been in Bear’s portfolio. Then in the first week of September Fannie Mae and Freddie Mac were taken over by the US Treasury in what was announced as a conservatorship; it was estimated that it would cost the US Treasury at least $100 billion and perhaps $200 billion to re-capitalize each of these firms.

In September 2008, Washington Mutual (WaMu) which was one of the largest and fastest growing mortgage lenders, was subject to a run; the FDIC intervened and detached the bank from its holding company; the bank was sold to JPMorganChase without cost to the FDIC.

The financial position of Lehman Brothers was well known, which made it difficult for the firm to obtain more capital; potential investors had no way to value many of its assets and hence how much capital the firm needed. The US government tried to facilitate the sale of Lehman to other financial institutions – but the statement was that no US government money would be involved. The Secretary of the Treasury subsequently said that the arrangement to sell Lehman to Barclay’s Bank had been prevented by the financial authorities in London, who were concerned that the British government might have to bail out Barclay’s if the losses of Lehman and its need for capital had been underestimated.

One day after the failure of Lehman, the US government intervened to prevent the collapse of AIG, then the world’s largest insurance company. AIG had sold several hundred billion dollars worth of credit default swaps (CDSs), essentially an insurance policy to the buyers of bonds that they would not incur losses if the firms that had issued bonds defaulted. Many of these CDSs were purchased by the firms that owned the bonds. Some were purchased by third parities, who had no direct insurable interest; from their point of view, the premium required to buy a credit default swap was low relative to the payoff if the borrower defaulted.

Financial history would have been different if Lehman Bros had not failed. The US government officials said that they did not have the authority to prevent the failure of Lehman – that they could not support an institution that was bankrupt. Perhaps, but they might have been able to come up with a temporary arrangement until they had the authority. And many in the government were concerned that preventing the failure of Lehman would have meant that there would be many more Bear Stearns and Lehmans and Countrywides that would be on the ropes in the next financial crisis in 2015 or 2020.

The implosion of every real estate bubble leads to the failure of a Lehman or a Northern Rock, usually one of the firms that had expanded most rapidly during the bubble years.

The traditional debate is between those who believe that the government should not intervene to save firms that are financially challenged; those who advance this view believe that if asset prices decline sufficiently sharply, the markets eventually will stabilize and investors who had been waiting on the sidelines will then buy the institutions that had lost most of their capital. Those who support this view subscribe to the moral hazard position that the next financial crisis will be more severe because the lenders would believe that they will be bailed out.

The challenge to this view is that each of these crises is different, and that the economic costs of not stabilizing the system are severe. Moreover, the moral hazard view fails to distinguish between government intervention to stabilize a single failing firm, and intervention to stabilize a group of firms as a real estate bubble implodes. Is the failure or near failure a result of faulty management decisions at a time when the economic environment is stable; in this case, the failure is idiosyncratic and reflects the problems of management. Alternatively, if a large number of financial institutions are on the ropes at more or less the same time, the failures are systemic, and result from a dramatic change in the financial environment, and what might charitably be called the mismanagement of the economy by the monetary authorities.

There are two responses to the view that public money should not be used to stabilize the financial system. One is that the costs in terms of unemployment and business failures and the increase in the government’s indebtedness are high. A second is eventually the government will change its policy in the effort to stabilize the system, and it would have been less costly if the government had intervened earlier in the crisis.

If the government decides to intervene to stabilize the financial system before property prices have declined to their long-term equilibrium levels, then there may be need for a series of injections of government money as the decline in the property prices leads to continuing shrinkage in bank capital. If in contrast the intervention is long delayed, then the recession in the economy will lead the banks to incur significant loan losses. The timing of intervention is a bit of an art form.

Banks traditionally have borrowed at short term and lent at long term – one of the risks that they acquired was that short-term interest rates might increase significantly relative to long-term interest rates and another was the liquidity risk that depositors might withdraw their money and that their creditors might be reluctant to purchase more of their IOUs when they matured. The prospect was that the depositors and the creditors would discipline the lending practices of the banks.

The Federal Deposit Insurance Corporation was established in 1934 to reduce runs on banks by owners of small deposits who were concerned that they might lose their money if their bank failed. In effect these depositors were provided by a US government guarantee against loss if the banks failed. Initially the ceiling on insured deposits was $2,500, which was soon raised to $5,000, and then to $10,000 in 1950, $20,000 in 1969, and $100,000 in 1980. Individuals could readily circumvent this ceiling by setting up accounts in the same bank under the names of different family members and different combinations of family members, and by opening accounts in different banks. In effect deposit insurance was unlimited to anyone who wanted to break up a large amount of money into $100,000 units.

The institutional innovation of deposit insurance eliminates the ‘discipline’ that small depositors might impose on banks, based on the threat that they might rush to get their money, which would mean that the managers of the banks would be super-cautious; the belief was that these small depositors are unsophisticated and would be trigger-happy. There was no need for insurance of larger depositors because their owners were believed to be sophisticated, and much less likely to trigger a run.

The practical issue involved in the decision whether to ‘save Lehman’ was which of the stakeholders in the firm would avoid losses if government assistance had enabled the firm to avoid bankruptcy. Lehman was significantly larger than Bear. Lehman was not a commercial bank, and hence it was regulated by the Securities and Exchange Commission rather than by the Federal Reserve, although after the rescue of Bear the Fed had opened its discount window to investment banks.

One interpretation of ‘too big to fail’ is that a faltering bank is so large that there is no other institution has sufficient excess capital to acquire the firm, much as Bear had been purchased by Morgan; hence the government must provide financial assistance to any of the five to ten very large firms because its demise would have costly consequences both for its competitors and the economy.

One of the concerns implicit in the ‘too big to fail’ cliché is that the various stakeholders in the bank benefit from the proverbial ‘free lunch’ because they are protected against losses that their counterparts in smaller firms would have incurred when they failed. The largest bank failure prior to the 2008 crisis was that of Continental Illinois Bank in 1984, then the seventh largest US bank. Continental Illinois had incurred large loan losses when the oil price declined sharply. Continental differed from most other banks in that it had a relatively small domestic deposit base. There had been a run on Continental Illinois in 1982, but that had been stanched; the run in 1984 started after a report of very weak earnings. The run started on its offshore branches. Initially the bank borrowed from the Federal Reserve, and then it was rescued by the FDIC.

The owners of the common stock of Continental Illinois lost all their money, and their shares became virtually worthless. The top executives and senior managers lost their positions and a large part of their personal wealth that was invested in the bank’s stocks – and they lost a significant part of their professional reputations. The careers of many managers as bankers were shortened. The directors of Continental Illinois were publicly embarrassed – and most lost money since they owned shares in the firm.

Who, then, was ‘bailed out’? The owners of the insured deposits did not incur any losses because of the FDIC guarantee – but these depositors had been charged premiums for deposit insurance. The owners of the uninsured deposits also avoided losses because they withdrew virtually all of their money before Continental Illinois was taken over by the FDIC. The owners of the subordinated debentures which had been issued by the bank holding company for Continental Illinois were bailed out. (There is general agreement that the FDIC made a mistake by providing money so that the owners of the bonds issued by the bank’s holding company did not incur losses.)

For all practical purposes, Continental Illinois failed even though it remained in business. Presumably the only way to ensure that the uninsured depositors took a ‘haircut’ was to have had the bank closed while it was still liquid.

Compare the costs and the benefits of ‘saving Lehman’ with the costs and benefits of ‘letting Lehman go’. In August and September 2008 the Secretary of the Treasury encouraged other financial firms to acquire Lehman, to invest in Lehman, or to develop a cooperative arrangement that would acquire some of the dodgy assets that Lehman owned – but he was adamantly against providing any US government financial assistance to Lehman because he did not want to be known as ‘Mr Bailout’.

There were several different models of how Lehman might have been saved. One was Morgan’s purchase of Bear; a buyer would have been found for Lehman after the Federal Reserve agreed to acquire $40 or $50 or $60 billion of its mortgage-related securities that were believed most sensitive to default. The owners of Lehman shares would receive a token $1 or $2 a share from the acquiring firm. Some of the senior executives of Lehman would be given positions in the acquiring firm, most would not. The owners of Lehman’s short-term IOUs would be made whole, and perhaps the owners of its bonds would be made whole like those of Fannie and Freddie; alternatively these bondholders might have been given haircuts. The Lehman name would disappear from the market place. The financial cost to the Federal Reserve might have been $50 billion or $60 billion or $70 billion that it would have used to buy the toxic securities – but these securities would have had some value after the economy stabilized.

A second model is provided by the US Treasury’s investment of Troubled Assets Relief Program (TARP) money in Citibank and Bank of America; the US government would buy 60 or 80 million new shares in Lehman at $1 a share – the amount needed to re-capitalize the firm. The immediate financial cost would be the amount that the Treasury would have to lay out to acquire these shares.

A third model is provided by the US Treasury’s takeover of Fannie Mae and Freddie Mac; Lehman also would be placed in a conservatorship and the US Treasury would become its owner. The US Treasury would lend money to Lehman. At some stage this loan would be converted to newly issued shares, and then after the crisis, these shares would be sold to the public. The amount that the Treasury would need to invest would be determined by the size of the ‘hole’ in Lehman’s capital structure.

Estimating the cost to the US government of saving Lehman is conjectural and involves the valuation of its mortgages and mortgage-related securities in the post-crisis period. (During the crisis, the market values of these securities would be highly uncertain, and almost certainly far below their intrinsic value based on the cash flows in a stable environment.). Before its collapse, Lehman had a net worth of $30 billion, and a leverage of thirty, so its assets were about $900 billion and its liabilities $870 billion. If its loan losses were three times its capital – then the cost to the US Treasury would have been $60 billion.

As Lehman approached bankruptcy, it was ‘two firms’; one firm continued to generate operating profits of $10 billion to $15 billion a year while the other firm had negative capital as a result of the massive decline in the value of the mortgages and mortgage-related securities that it owned. One advantage of the use of government funds to re-capitalize the firm is that the good bank would continue to generate $10 billion or $15 billion of annual profits from its trading and investment activities, while the bad bank would contain most or all of the low-grade mortgage-related paper. The advantage of the second and the third approaches is that the US government would benefit from the operating income of Lehman, which would offset some of the losses.

The ‘additional cost’ – always noted by those concerned with the moral hazard argument – is that the next crisis will be more severe because Lehman would have been ‘bailed out’. Regardless of the approach taken to prevent Lehman’s bankruptcy, the Lehman shareholders would have been wiped out, its top executives would have been replaced and lost much of their personal wealth, and its directors would have been humiliated by their failure to protect the shareholders. Two groups of stakeholders would have benefited if Lehman had not failed – one was employees, perhaps some of those with bonuses, and the other was the counterparties and the owners of Lehman bonds.

AIG – one of the world’s largest insurance companies – also was experiencing a run as efforts were made to avoid the bankruptcy of Lehman. AIG had written tens of billions of credit default swaps. In effect AIG acquired the credit risk attached to these bonds in exchange for the premium income; the buyers of the credit default swaps believed that premiums that they paid were modest relative to the credit risk. The run on AIG was like the run on Lehman once removed, since AIG would have to reimburse those who had bought these swaps to protect themselves against losses on their holdings of Lehman bonds when the firms went bankrupt.

The counterfactual is to determine how the crisis would have evolved if Lehman had been saved; there is some likelihood that the run might have been deflected to other firms, perhaps Merrill Lynch or Morgan Stanley. One possibility is that that run on AIG would have continued, perhaps less rapidly. Another possibility is that US government would have made it known at the time that Lehman was saved that it stood ready to assist large firms that had liquidity problems, or had depleted capital – which was essentially the policy that was adopted when AIG was bailed out. The terms of the ‘bail-outs’ would have included that the common and preferred shareholders would have lost all of their money and the managements and the directors replaced.

If AIG had not been saved then it is highly likely that other financial firms would have been subject to runs. Asset prices would have declined, and other financial firms would have failed. Lenders would be extremely cautious in making new loans, and in renewing maturing credits. The US recession would have become much more intense, and the US economy might have fallen into the debt-deflation trap.

Thus the amounts needed to save Lehman – the counterfactual – would have been the $50 billion or $100 billion that would have been the financial outlay by the US Treasury minus any recovery and the incremental cost of financial crisis in the future because the lenders would have realized that there was an increased likelihood that they would be bailed out if they followed a more risky investment strategy, The additional cost would be the cost of saving other large financial institutions that had invested too aggressively.

This cost must be adjusted for the additional outlays that the US Treasury and the Federal Reserve incurred when they intervened to save AIG, Citibank, Bank of America, and twenty or thirty other firms. Estimating this amount is extremely difficult – the runs on individual institutions would continue until it was clear beyond the shadow of a doubt that no institution would fail because of a lack of liquidity. The US Treasury invested more than $400 billion in other financial institutions, and the data suggest that virtually all of this money will be recovered. (The investment of TARP money in GM and Chrysler is another issue.)

The immediate impact of the decision not to save Lehman was a panic and crash. Some of the costs of not saving Lehman can be estimated in terms of the increase in unemployment, the decline in US GDP, and the increase in the US government’s fiscal deficit. In the first eight months of 2008, employment had declined by an average of forty thousand a month, in the six months after the failure of Lehman, employment declined by six hundred thousand a month.

The striking phenomenon was the surge in interest rates on risky assets – on LIBOR relative to the US Treasury bills. Those banks with a relatively strong cash position no longer had trust in the credit standing of the banks that had incurred large loan losses, which then had to sell assets and shrink; they hoarded cash. Those banks in a weak cash position were then obliged to raise cash in other ways and to shrink their assets.

The primary costs of not saving Lehman are those associated with the incremental decline in US GDP because of the surge in interest rates. In August 2008, the non-partisan Congressional Budget Office projected that the US fiscal deficit would be $450 billion in fiscal 2009; six months later, the projection was $1500 billion, which was ten or fifteen times larger than the highest estimate of the costs of saving Lehman. Most of the increase in the projected fiscal deficit was due to the sharp decline in economic activity because of the dramatic reduction in the willingness to lend, and because households and firms hunkered down – as employment declined sharply, spending declined – individuals were less confident about their jobs.

A significant number of large financial institutions had failed prior to the run on Lehman; the US government had intervened with both Bear Stearns and Fannie Mae and Freddie Mac. Why was Lehman different?

Banks, mortgage brokers, insurance companies and other financial institutions can be placed in one of four groups in terms of how they fared in the 2008 crisis. One group consists of those firms that went out of business because they lost all or virtually all of their capital; the shareowners in these firms lost all of their money and these firms have disappeared from the marketplace. This group includes Indy Mac, Washington Mutual, and Lehman Brothers. Nearly three hundred mortgage banking firms failed. Nearly two hundred commercial banks were closed by the Federal Deposit Insurance Corporation. CIT is in this group. The top executives lost their positions, and a great deal of their personal wealth; the careers of many of these individuals were shortened. The boards of directors of these firms were disgraced.

A second group comprises those firms that that incurred such large losses that they were acquired by other, better capitalized firms, either with or without financial assistance of the US government. Countrywide Financial was in this group, as was Wachovia Bank and Bear Stearns. Merrill Lynch was in this group. These firms disappeared from the marketplace and their shareholders lost most but not all of their money. Similarly most of the top executives in these firms lost their positions, and their careers were truncated. In some cases, however, the top officials still walked away with tens of millions of dollars.

The third group includes the firms that probably would have been forced to close their doors if they had not had assistance from the US Treasury or the Federal Reserve. AIG, Citibank, and Bank of America are in this group. So are some of the regional banks. The prices of the shares of these firms declined by eighty to ninety percent, although these prices subsequently increased once the US government became a significant shareholder and the likelihood that the firms would fail diminished sharply.

The fourth group consists of the firms that did not take any funds from the US Treasury, or else took these funds because they were instructed to do so by Secretary of the Treasury Paulson even though they did not want the US government’s money. JP Morgan Chase is in this group, as is Goldman Sachs, both in what had been the top tier of US investment banks. Several of the firms in this group were able to raise capital in the private markets; Morgan Stanley sold new shares to Mitsubishi Bank.

Why did the failure of Lehman lead to the unprecedented panic in the credit markets? One view is that the initiatives of the US government to prevent the bankruptcy of Bear Stearns and subsequently of Freddie and Fannie followed by the unwillingness to commit government money to save Lehman meant that there appeared to be no coherent government policy toward providing financial assistance to weakened large institutions. If two points determine a line, then measures adopted by the US government to deal with the Bear problem and the Freddie and Fannie problem set the policy. But then the ‘policy’ was reversed by not saving Lehman. And while the initiatives toward AIG represented a reversal of the reversal, market participants were shaken by the inconstancy of the approach of the US government.

Haircuts for all?

The principal stakeholders that have benefited from the use of government money to stabilize the banks and other financial institutions have been the bondholders and other counterparties, the uninsured creditors, and the senior managers. If these firms had gone bankrupt like Lehman, the bondholders would have taken a haircut, and depending on the severity of the losses, they might have lost 30 or 50 percent of their money. The employment contracts of the senior managers would no longer be binding if the firms failed, and those who had large amounts due because of their success as traders would find that they were just another group of unsecured creditors.

If the bondholders are subject to loss if the firm closes, then many would sell their bonds in anticipation of these losses. Interest rates on these bonds would increase sharply, and the banks would not be able to sell new bonds when the existing bonds mature. The banks would be forced to shrink their assets. Similarly the counterparties would be reluctant to re-finance existing positions. The likelihood of losses by these groups could trigger runs.

One position is that bond instruments should be designed so that the bondholders would incur losses if the firms failed; then they would exercise more discipline on the senior managers so that they would be more cautious in undertaking risky investments. Perhaps. The credit-rating agencies already evaluate the riskiness of the bonds, and it is not obvious that the prospect of haircuts might induce the investors to be more cautious in buying the bonds of the banks.

The hidden cost of too much discipline is that bankers would be extremely cautious lenders, and many worthwhile projects would find it difficult to obtain financing.

The design challenge is to ensure that large firms are allowed to fail when the causes of failure are idiosyncratic to one or two institutions in a stable financial environment. When the failures are systemic and involve a large number of institutions, the policy problem differs; failure has many more costly externalities. The capital of financial institutions as a group declines, and there may be too little capital relative to the liabilities of these institutions. Either asset values will decline until the ratio of bank assets to bank capital is at a satisfactory level, or the government must provide financial investments to the banks. The nub of the problem is that the value of bank capital is highly sensitive to changes in asset prices.

Assume that there had been no rescue program – no TARP money, no Fed investments, and so on. Then other banks that had been dependent on borrowed short-term money to carry long-term assets would have failed. Other banks that had been conservative in their lending policies would have encountered large loan losses because the recession would have been much more severe.

Fast forward to 2020 or 2025, the approximate time of the next major period of rapid credit expansion. How many of the top executives of the major US financial firms at that time would look back at the government intervention in the 2008 crisis and conclude they can afford to take greater risks in the allocation of credit because so many large financial institutions had been ‘bailed out’.

These executives would have observed that the stock prices and market value of financial firms as a group had declined by ninety percent or more. They would have observed that the senior executives of most of these institutions were replaced, and that the directors also lost their positions. They would have noticed that many managers lost their careers.

The US investment banking industry had been decimated, with one dominant firm as a survivor; another of the major firms had to sell about twenty percent of its shares to a Japanese firm.

‘Too big to fail’ is a catchy cliché. But the slogan obscures more than it reveals. And because of the reluctance of government officials to provide financial assistance to Lehman so that the firm could have avoided bankruptcy even as its shareholders lost all their money, a market adjustment cascaded into the most severe crisis in one hundred years.