(i) The NFA Is Built on a Very Weak Theoretical Foundation:

The NFA is based on light regulation of commercial banks, even lighter regulation of investment banks, and little if any regulation of the “shadow banking system” – hedge and private equity funds and the bank-created Special Investment Vehicles (SIVs) discussed below. Support for lax regulation embedded in the NFA was founded on the central claim of neoclassical financial economics that capital markets price securities correctly with respect to their risk and return. Given the accurate risk-return information provided by capital markets, buyers and sellers of financial securities can make optimal decisions that lead to risk being held only by those capable of managing it. The celebratory narrative associated with the NFA states that relatively free financial markets minimise the possibility of financial crises and the need for government bailouts. (See Volcker 2008 for a summary of this narrative.)

Economic & Political Weekly

EPW

STRUCTURAL CAUSES

Crotty 2008 explains that this theoretical cornerstone of the NFA is based on patently unrealistic assumptions and has no convincing empirical support. Thus, the “scientific” foundation of the NFA is shockingly weak and its celebratory narrative is a fairy tale.

(ii) The NFA Has Widespread Perverse Incentives That Create Excessive Risk, Exacerbate Booms and Generate Crises: The current financial system is riddled with perverse incentives that induce key personnel in virtually all important financial institutions – commercial and investment banks, hedge and private e quity funds, insurance companies and mutual and pension funds

– to take excessive risk when financial markets are buoyant.1 For example, the growth of securitisation generated fee income throughout the system – to mortgage brokers who sold the loans, investment bankers who packaged the loans into securities, banks and specialist institutions who serviced the securities, and rating agencies who gave them their seal of approval. Since these fees do not have to be returned if the securities later suffer large losses, everyone involved had strong incentives to maxi mise the flow of loans through the system – whether or not they were sound. This is the hidden downside of the new “originate and distribute” model of banking. Total fees from home sales and mortgage securitisation from 2003 to 2008 have been estimated at $2 trillion (Financial Times, “The Fatal Banker’s Fall”, 1 October 2008).

Investment banks played a key role in generating the crisis. Top investment bank traders and executives receive giant b onuses in years in which risk-taking behaviour generates high revenue and profits. Of course, profits and bonuses are maximised in the boom by maximising leverage, which in turn maximises risk. In 2006, Goldman Sachs’ bonus pool totalled $16 billion – an average bonus of $650,000 very unequally distributed across Goldman’s 25,000 employees. Wall Street’s top traders received b onuses up to $50 million that year. In spite of the investment bank disasters of the second half of 2007 which saw Wall Street investment banks lose over $11 billion, the average bonus fell only 4.7%. In 2008, losses skyrocketed and all five of the largest independent investment banks lost their independence: two failed, one was taken over by a conglomerate, and two became bank holding companies to qualify them for bailout money. Yet, Wall Street bonuses were over $18 billion (DiNapoli 2009). At Merril Lynch 697 employees received bonuses in excess of $1 million last year from a total bonus pool of $3.6 billion in spite of the fact that the firm lost $ 27 billion in 2008. The top four recipients alone received a total of $121 million while the top 14 got $249 million (Wall Street Journal, “Merrill Gave $1 Million Each to 700 of Its Staff”, 12 February 2009). It is therefore rational for investment bank “rainmakers” to take excessive risk in the bubble even if they understand that their decisions are likely to precipitate a crash in the intermediate future. Since they do not have to return their bubble-year bonuses when the inevitable crisis brought on by their excessive risk-taking occurs and since they continue to receive bonuses even in the crisis, they have a powerful incentive to pursue high-risk, high-leverage strategies.

Credit rating agencies that play a crucial role in the NFA were also infected by perverse incentives. Under Basel I rules, banks were required to hold 8% of core or tier one capital against their total risk-weighted assets.2 Since rating agencies determined the risk weights on many assets, they strongly influenced bank capital requirements. High ratings mean less required capital. They therefore facilitate higher leverage, higher profit and higher b onuses, but create higher risk as well. Moreover, important financial institutions are not permitted to hold assets with less than an AAA rating from one of the major rating companies. There is thus a strong demand for high ratings. Rating agencies are paid by the investment banks whose products they rate. Their profits therefore depend on whether they keep these banks happy. In 2005, more than 40% of Moody’s revenue came from rating securitised debt such as mortgage backed securities (MBSs) and collateralised debt obligations (CDOs). If one agency were to give realistic assessments of the high risk associated with mortgagebacked securities while others did not, that firm would see its profit plummet. Thus, it made sense for investment banks to shop their MBSs around, looking for the agency that would give them the highest ratings, and it made sense for agencies to provide excessively optimistic ratings. Re-regulation of financial markets cannot be effective unless it substantially reduces the perverse incentives that pervade the system.

(iii) Innovation Created Important Financial Products So Complex and Opaque They Could Not Be Priced Correctly and Therefore Lost Liquidity When the Boom Ended: Financial innovation has proceeded to the point where important structured financial products such as MBSs and CDOs are so complex and so opaque that they are inherently non-transparent. They cannot be priced correctly, are not sold on markets, and are inherently i lliquid. According to the Securities Industry and Financial M arkets Association (SIFMA), there was $7.4 trillion worth of MBSs outstanding in the first quarter of 2008, more than double the amount outstanding in 2001. Over $500 billion in CDOs were issued in both 2006 and 2007, up from $157 billion as recently as 2004 (SIFMA web site). The explosion of complex MBSs and related derivative products created large profits at giant banks and other financial institutions, but also destroyed the transparency necessary for any semblance of market efficiency.3 Eighty per cent of the world’s $680 trillion worth of derivatives are sold over-the-counter (OTC) in deals negotiated between an investment bank and one or more customers. Indeed, the value of securities not sold on markets may exceed the value of securities that are. Thus, the claim that competitive capital markets price risk optimally, which is the foundation of the NFA, does not apply even in principle to these securities.

A mortgage-backed CDO is a complex security that converts the cash flows from the mortgages in its domain into tranches or slices that have different risk characteristics. Banks sell the tranches to investors. Several thousand mortgages may go into a single MBS and as many as 150 MBSs can be packaged into a single CDO. A CDO squared is a CDO created by using other CDO tranches as collateral. Higher power CDOs are particularly difficult vehicles because the many mortgages appear in more than one of the underlying CDOs. Synthetic CDOs use credit default swaps (or other credit instruments) as their collateral. In the early 2000s, synthetic CDOs became the dominant form of CDO issuance, possibly because the investor demand for CDOs was rising faster than

March 28, 2009 vol xliv no 13

the supply of mortgages. A textbook on credit derivatives explains a part of the price calculation process as follows.

The expected default payment is calculated by looking at the number of default references and loss at default for each tranche, given a certain [historical] correlation between the underlying reference assets. The correlation is mathematically handled through a Cholesky composition and the actual loss distribution is arrived at by running a number [10,000 to 100,000] of so-called Monte Carlo simulations, which generate a random number of defaults (as defined by the Merton model) in the underlying assets (Chacko, Sjoman, Motohashi and D essain 2006: 227).

The authors caution that:

Even with a mathematical approach to handling correlation, the complexity of calculating the expected default payment, which is what is needed to arrive at a CDO price, grows exponentially with an increasing number of reference assets [the original mortgages] … As it turns out, it is hard to derive a generalised model or formula that handles this complex calculation while still being practical to use (226).

The relation between the value of a CDO and the value of its mortgages is complex and non-linear. Significant changes in the value of underlying mortgages induce large and unpredictable movements in CDO values. Rating agencies and the investment banks that create these securities rely on extremely complex simulation models to price CDOs. It can take a powerful computer several days to determine the price of a CDO. These models are unreliable and easily manipulated. Given the ubiquity of perverse incentives, it is not surprising that market insiders refer to the process through which investment banks and rating agencies price or mark CDOs as marking to “magic” or to “myth”. New York University’s Nouriel Roubini considered the CDO market to be profoundly inefficient:

[S]omehow greedy and clueless investors searching for yield bought tranches of instruments – CDO or CDO cubed – that were new, exotic, complex, illiquid, marked-to-model rather than marked-to-market and misrated by the rating agencies. Who could then ever be able to correctly price or value a CDO cubed? And for all the talk about the benefits of financial innovation what was the social value of a CDO cubed? There was indeed zero social value in this type of financial innovation that is closer to a con game than to a financial product of any use (Roubini 2008).

Demand for non-transparent CDOs was strong in the boom b ecause buyers could borrow money cheaply to buy these products, returns were high, and the products carried top ratings. But when the housing boom ended and defaults increased, the fact that no one knew what these securities were worth caused d emand and therefore liquidity to evaporate and prices to plummet. CDOs could be sold, if at all, only at an enormous loss. It is estimated that by February 2009, almost half of all the CDOs ever issued had defaulted (Financial Times, “Half of All CDOs of ABS Failed”, 10 February 2009). Defaults led to a 32% drop in the value of triple A rated CDOs composed of super-safe senior tranches and a 95% loss on triple A rated CDOs composed of mezzanine tranches (Financial Times, “Time to Expose Those CDOs”, 26 February 2007). Facing a wave of criticism for having led investors astray with overly optimistic rating in the boom, rating agencies belatedly slashed ratings for these products in the crisis, exacerbating the problem. Since these securities were often purchased with borrowed funds, losses triggered margin calls, which forced the sale of safer assets

Economic & Political Weekly

EPW

March 28, 2009 vol xliv no 13

because they were the only ones the market would accept. This spread the crisis across market segments. Since no one knew how much these assets were worth or who held them, credit dried up everywhere. Note that since banks are obligated to value their securities at their estimated current value, bank capital evaporated along with the price of these securities. In response, banks tightly restricted credit, first to non-bank investors and then to each other in the inter-bank market.

(iv) The Claim That Commercial Banks Distributed Almost All Risky Assets to Capital Markets and Hedged Whatever Risk Remained Was False: The conventional view was that banks were not risky because in contrast to the previous era when they held the loans they made, they now sold these loans to capital markets through securitisation. Moreover, it was believed that banks hedged whatever risk remained through credit default swaps. Both these propositions turned out to be myths.

Banks kept risky products such as MBSs and CDOs for five reasons, none of which were considered in the NFA narrative about efficient capital markets. First, to convince potential investors that these securities were safe, banks often retained the riskiest part – the so-called “toxic waste”. Second, given banks’ incentive to generate high profits and bonuses through high risk, they purposely kept some of the riskiest products they created to maximise bankers’ compensation by maximising short-term profits.

Third, the rate of flow of these securities through banks was so great and the time lapse between the bank’s receipt of a mortgage and the sale of the MBS or CDO of which it was a part was sufficiently long that at every point in time banks held or “warehoused” substantial quantities of these securities. When demand for MBSs and CDOs literally collapsed in the crisis, banks were left holding huge amounts of mortgages and mortgage-backed products they could not sell. According to SIFMA, global CDO issuance fell from $186 billion in the first quarter of 2007 to $11 billion one year later, a collapse of 94% (SIFMA web site). The Bank of England estimates that the global issuance of all asset-backed securities fell by about 85% from the second quarter of 2007 to the first quarter of 2008 (2008, 6). Plummeting prices for these products were the main source of the bank losses that are the centre of the crisis.

Consider Merrill Lynch and Citigroup, the two largest underwriters of CDOs in the three years leading up to the crisis. The Wall Street Journal estimated that in November 2007 Citigroup had $12 billion in warehoused mortgage related securities, and $43 billion in super-senior tranches of CDOs (“Why Citi Struggles to Tally Losses”, 5 November 2007). Merrill Lynch had only $3.4 billion in CDO origination in 2003, but in 2006 it posted $44 billion in CDO deals. Merrill’s rainmakers became addicted to the fees that flowed from financing CDOs, which reached $700 million in 2006.

Fortune described the end game of this process.

Merrill apparently made a pivotal – and reckless – decision. It bought big swaths of the AAA paper itself, loading the debt onto its own books ... The amounts were staggering. By the end of June, Merrill held $41 billion in subprime CDOs and subprime mortgage bonds. Since the a verage deal is between $1 billion and $1.5 billion, and the AAA debt is around 80% of each deal, Merrill must have been buying nearly all the top-rated debt from dozens of CDOs … Merrill’s $41 billion exposure to subprime paper was more than its entire shareholders’ equity of $38

STRUCTURAL CAUSES

cost of hedging can occur quickly. The cost of insuring countrywide debt rose from $75,000 in early July 2007 to $230,000 a month later; it then jumped to $350,000 just one day later (Wall Street Journal, “Reversal in Fortune”, 11 August 2007). By January 2008, the cost of insuring countrywide debt was $3 million up front and $500,000 annually.

Moreover, hedging often involves sophisticated, complex, d ynamic derivative trading strategies. This form of hedging relies on the existence of liquid continuous markets with low to moderate transactions costs. A typical dynamic hedge involves shorting the risky asset held and investing in a risk-free asset. The hedge adjusts whenever the asset price or interest rate or volatility changes in the market, which they do continuously. Every time the asset price declines or volatility increases, the risky asset must be sold; this is what makes the hedge “dynamic”. When problems hit a risky asset market, price falls and volatility rises. Institutions with dynamic hedges must sell their risky assets, which accelerates the rate of price decrease, which in turn forces more hedged-asset sales. If many investors have made similar d ynamic hedges and are selling, liquidity dries up and prices can free fall.

Third, the narrative insists that derivatives unbundle risk, d ividing it into simpler segments. But, in fact, sophisticated derivatives such as CDOs rebundle risk in the most complicated and non-transparent ways: this is what financial engineering and structured derivative products do.5 These derivatives also add substantial “embedded” leverage to the underlying or primitive products to enhance their profits. Das explains how the layers of unseen leverage investment bankers added to derivative products sold to Orange County California caused catastrophic losses. “Greenspan had been right – risk had truly been unbundled. We had packaged it right back up and shoved it down the eager throats of the wealthy taxpayers of Orange County” (Das 2006, 50).

Fourth, the celebratory NFA narrative applauded globalisation of financial markets because it created channels of risk dispersion. But securitisation and funding via tightly integrated global capital markets simultaneously created channels of contagion in which a crisis that originated in one product in one location (US subprime mortgages) quickly spread to other products (US prime mortgages, MBSs, CDOs, home equity loans, loans to residential construction companies, credit cards, auto loans, monoline insurance and auction rate securities) and throughout the world. The complexity of the networks linking markets together created immense fragility in the system.

Complexity adds to the danger that any one part of the hyper-financial system can bring down the whole. Monoline insurers exemplify this kind of reef under the water. Their capitalisation is tiny – a few tens of billion dollars in equity – yet because they act as guarantors to h undreds of billion dollars in bonds, their failure would cause losses at almost every bank and insurance company in the world (Financial Times, “Editorial: The Start of the Great Unwinding”, 25 January 2008).

The quantum jump in the potential for contagion created in the NFA raised systemic risk and, ultimately, triggered a severe crisis that affected much of the global financial system.

(vi) Regulators Allowed Banks to Hold Assets Off Balance Sheet with No Capital Requirement: In the NFA banks were allowed to hold risky securities off their balance sheets in SIVs and conduits with no capital required to support them. Since capital had to be held against on-balance-sheet assets, the regulatory system induced banks to move as much of their assets off-balance sheet as possible. The concept of the SIV emerged in the late 1990s, “when large global banks … realised they could exploit loopholes in the Basel capital-adequacy rules if they placed assets such as mortgage-backed securities, in off-balance-sheet vehicles” (Financial Times, “A Ray of Light for Shadow Banking”, 18 June 2008). When the demand for risky structured financial products cooled off in mid-2007, bank-created off-balance-sheet SIVs and conduits became the buyer of last resort for the ocean of new MBSs and CDOs emanating from investment banks. The SIVs were supposed to be standalone institutions that paid service fees to the originating banks, but to which the originating banks had no obligations or commitments. At the end of 2007, JP Morgan Chase & Co and Citigroup Inc each had nearly $1 trillion in assets held off their books in special securitisation vehicles. For Citigroup this represented about half the bank’s overall assets (Wall Street Journal, “FASB Signals Stricter Rules for Banks’ Loan Vehicles”, 2 May 2008).

SIVs borrowed short-term in the commercial paper market and used this money to buy long-term, illiquid but highly profitable securities such as CDOs – a very dangerous game. To enable this commercial paper to receive AAA ratings, and thus low interest rates, the originating banks had to make commitments to provide their SIVs with guaranteed lines of credit. This made the banks vulnerable to problems experienced by their supposedly independent SIVs. Noriel Roubini noted that when

this garbage of CDOs (or CDOs cubed) was not fully distributed to clueless and greedy investors banks created off-balance sheet SIVs and conduits that would buy the leftover trash that no investor wanted to touch and repackaged it into structures that were financed with the most short-term ABCP; these SIVs were then blessed with credit e nhancement and guarantees of liquidity lines from the banks that made them de facto on balance sheet items even if they were de jure off balance sheet; but there were extra fat fees to be made from managing this toxic SIVs and conduits and thus the fee generation machine kept on rolling (Roubini 2008).

And then the machine stopped rolling. When problems in the housing market triggered a wave of subprime defaults, the value of MBSs and CDOs collapsed, revealing to everyone the house of cards on which the SIVs were built. This naturally triggered a mass exodus from the asset-backed commercial paper market. US asset-backed commercial paper outstanding fell from $1.2 trillion in July 2007 to $840 billion by year’s end. With the disappearance of their major source of funding, banks were forced to move these damaged assets to their balance sheets. In late July 2008 analysts at Citigroup forecast that up to $5 trillion worth of assets might be forced back on to bank balance sheets, exacerbating their shortage of capital and further restricting their ability to lend to consumers and companies (Financial Times, “Rule Changes Could Hit Balance Sheets”, 21 July 2008). The combi nation of bank write downs on assets held on-balance sheet combined with devalued SIV assets that had to be moved back onto balance sheets severely eroded bank

March 28, 2009 vol xliv no 13

c apital. This, in turn, forced banks to try to lower their risk by reducing lending and raising interest rates, both to other financial institutions and to households and non-financial businesses. The resulting credit squeeze exacerbated the fall in home prices and mortgage d efaults and slowed economic growth, which led to additional bank losses. These forced more capital write-downs and spread panic.

(vii) Regulators Allowed Giant Banks to Measure Their Own Risk and Set Their Own Capital Requirements, Which Naturally Led to Excessive Risk-Taking: Deregulation allowed giant financial conglomerates to become so large and complex that n either insiders nor outsiders could accurately evaluate their risk. Conceding that outsider regulators could not do the job, the Bank for International Settlement allowed banks to evaluate their own risk – and thus set their own capital requirements – through a statistical exercise based on historical data called Value at Risk (VAR). Government officials thus ceded to banks, as they had to rating agencies, crucial aspects of regulatory power. VAR is an estimate of the highest possible loss in the value of a portfolio of financial assets and liabilities over a fixed time interval with a speci fic statistical confidence level. The standard exercise calculates VAR under negative conditions likely to occur only 5% of the time.

There are three fundamental flaws in this mode of risk assessment. First, there is no time period in which historical data can generate a reliable estimate of current risk. If firms use data from the past year, as is commonplace, then during booms periods like 2003 to mid-2007 a VAR exercise will conclude that risk is minimal because defaults and capital losses on securities are low. This means that banks will need to set aside only a small amount of capital as insurance against estimated risk. Given the high maximum asset-to-capital ratio that the under-estimation of risk permitted, banks aggressively expanded leverage, which left them vulnerable when the crisis appeared. The chairman of the UK’s Financial Services Authority said that VAR “fails to allow for the fact that historically low volatility may actually be an indication of irrationally low risk-aversion and therefore increased systemic risk” (Financial Times, “Insight: Risk Needs a Human Touch but Models Hold the Whip Hand”, 22 January 2009). On the other hand, if data from past decades are used in the VAR calculation, the existence of past crises will raise estimated risk, but financial markets will have undergone such fundamental change over the long period that these estimates are unlikely to be reliable p redictors of current risk.

Second, VAR models assume that financial security prices are generated by a normal distribution in which the likelihood that an

o bservation several standard deviations beyond a 95% confidence interval will occur is infinitesimal. In fact, security prices follow a distribution in which the preponderance of observations are “normal,” but upon occasion observations far from the mean appear – the well-known “fat tail” phenomenon. Tail events – such as the precipitous drop in stock prices that took place in A ugust 1987 or the global crisis brought on by the collapse of the giant hedge fund Long Term Capital Management (LTCM) or the recent global stock market crash – inevitably occur from time to time. This means that banks were permitted to evaluate risk on the assumption that very large losses are impossible, which left banks extremely vulnerable each time a crisis episode erupted. In August 2007, two large

Economic & Political Weekly

EPW

hedge funds managed by Goldman Sachs collapsed, forcing Goldman to inject $3 billion dollars into the funds. To explain why Goldman should not be held responsible for their collapse, CFO David Viniar said, “We were seeing things that were 25 standard deviation moves, several days in a row” (Financial Times, “Goldman Pays the Price of Being Big”, 14 August 2007, 25). The Director for Financial Stability of the Bank of England noted that under a normal distribution, “a 25-sigma event would be expected to occur once every 6 x 10124 lives of the universe”, adding, tongue in cheek, that when he tried to calculate the probability of such an event occurring several days in a row, “the lights visibly dimmed over London and … the computer said ‘No’.” Even a 7.3 standard deviation event should occur only once every 13 billion years (Haldane 2009, 2).

Third, the asset-price correlation matrix is a key determinant of measured VAR. The lower the correlation among security prices, the lower the portfolio’s risk. VAR models assume that f uture asset price correlations will be similar to those of the r ecent past. However, in crises the historical correlation matrix loses all relation to actual asset price dynamics. Most prices fall together as investors run for liquidity and safety and correlations invariably head toward one. Again, actual risk is much higher than risk estimates from VAR exercise

By allowing banks run by people with perverse incentives to assess their own risk and set their own capital requirement through VAR, regulators created a system in which banks severely underestimated risk in the boom. Reliance on VAR helped create the current crisis and left banks with woefully inadequate capital reserves when it broke out. Nassim Nicholas Taleb, the ubiquitous critic of over-reliance on mathematical risk models, called VAR “a fraud”. A Financial Times editorial observed that “risk management models and compensation practices were catastrophic” (“The Year the God of Finance Failed”, 26 December 2008) V AR-determined capital requirement is just one of many possible e xamples of totally ineffective regulatory processes within the NFA. Financial markets were not just lightly regulated, such regulation as did exist was often “phantom” regulation ineffective by design.6

The flaws of managing risk through VAR were apparent to everyone who understood the process; you do not need specialist knowledge to see them. Yet only a few important financial o bservers warned against the futility of standard risk management practices before the crisis broke out, perhaps because VAR-based risk assessment maximised rainmaker bonuses and no one wanted to kill the goose that was laying golden eggs. Frank Partnoy did call attention

7

to the flaws in VAR.

VAR was dangerous. It gave firms a false sense of complacency, because it ignored certain risks and relied heavily on past price movements. In some markets, VAR actually increased risk, because every trader assessed risk in the same flawed way. In other markets, traders [using different VAR models] calculated VAR measures that varied ‘by 14 times or more’ … LTMC’s VAR models had predicted that the fund’s maximum daily loss would be in the tens of millions of dollars, and that it would not have collapsed in the lifetime of several billion universes (Partnoy 2003, 263).

(viii) The NFA Facilitated the Growth of Dangerously High S ystem-Wide Leverage: As noted, the structural flaws in the NFA c reated dangerous leverage throughout the financial system. The l everage problem was made evident with the 1998 collapse of LTCM, a hedge fund that had managed to control $125 trillion in derivative

STRUCTURAL CAUSES

products on a capital base of less than $5 billion. Regulators said they had to rescue LTCM to avoid a global financial meltdown. Yet the l esson of the 1998 crisis that high leverage is dangerous seemed to be quickly forgotten. Borrowing by US financial institutions rose from 62% of gross domestic product (GDP) in 1997 to 114% of GDP at the end of 2007. Just prior to the crisis, large investment banks had asset to equity ratios in the mid-1930s; this represented much higher leverage than in earlier periods. In 2007 Goldman Sachs used only about $40 billion of equity as the foundation for $1.1 trillion of a ssets. At Merrill Lynch, $1 trillion of assets was backed by a paltry $30 billion of equity. Moreover, at least half of their borrowing was in the form of overnight repos, money that could flee at the first hint of trouble. Half of the spectacular rise in investment bank’s return on equity in the four years leading up to the crisis was generated by higher leverage rather than smart investing, efficient innovation, or boom-induced capital gains on trading assets. Commercial banks a ppeared to be adequately capitalised, but only because they overestimated the value of on-balance sheet assets while they kept a high percentage of their most vulnerable assets hidden off-balance-sheet. In fact, they were excessively leveraged, as the crisis revealed. By the end of 2008 many large banks had seen their equity position evaporate to the brink of insolvency and beyond.

According to one recent estimate, the total leverage ratios (on- and offbook assets and exposure divided by tangible equity) for the two biggest US banks were 88:1 for Citibank and 134.1 for Bank of America. The bursting of the property bubble causes such ratios, which were already too high on the eve of the crisis, to explode as off-balance-sheet commitments and pre- arranged credit lines came home to roost (Ferguson 2008).

Rising leverage in the recent boom was facilitated in part by the easy money policies of the Federal Reserve. To avoid a deep financial crisis following the collapse of the late 1990s stock market and internet booms, the Fed began to cut short-term interest rates in late 2000 and continued to hold them at record lows through mid-2004. F inancial firms were thus able to borrow cheaply which, u nder different circumstances, might have fuelled a boom in productive capital investment. However, given the structural flaws in financial markets and a sluggish real economy kept alive only by the impact of rising debt and financial wealth on aggregate demand, the additional funds were used instead for speculative financial investment that fed the high financial sector profits of the boom.

The rise in leverage helped push the size of financial markets to unsustainable heights relative to the real economy. By 2007 the global financial system had become, to use Hyman Minsky’s famous phrase, “financially fragile”. The term is usually applied to a cycle phase, but in this case the condition was secular. Any serious deterioration in the cash flows required to sustain security prices could have triggered a dangerous de-leveraging pro cess. Falling housing prices and rising defaults provided that trigger. By January 2009, housing prices had declined by almost 28% according to the Case-Shiller i ndex and default rates rose steadily. Asset price declines were large. One reason was that: “The leverage used to put [CDOs] together can amplify losses [in the downturn]. For example, a 4% loss in a mortgage backed security held by collateralised debt obligations can turn into almost a 40% loss to the holder of the CDO itself” (New York Times, “How Reverse Leverage in Structured Financial Instruments Works”, 29 July 2007). Margin calls forced borrowers to sell s ecurities, which accelerated their price decline. The de-leveraging process froze credit markets. Since modern non-financial business and household sectors run on credit, the shrinking availability and rising cost of borrowing led to a slowdown in economic growth that in turn worsened the global financial crisis. The NFA finally brought the global economy to the edge of the abyss.