by Adam Tooze
Chapter 6
“THE WORST FINANCIAL CRISIS IN GLOBAL HISTORY”
America’s real estate prices peaked in the summer of 2006 and began, slowly at first, to ease. In Ireland the turning point came in March 2007. By the summer construction sites began to close in Spain. October 2007 saw the first dip in UK house prices. Tens of millions of home owners now felt the force of asset valuation go into reverse gear.1 As house prices fell, equity dwindled, and the hardest hit slid into negative equity. Families scrambled to slash spending and pay down credit card and other short-term debt. The result was a smothering recession in consumer demand. Whatever else happened, a large part of the North Atlantic economy was headed into a downturn. On the face of it, this was precisely the sort of contingency that financial engineering was supposed to deal with. Through securitization, risks were supposed to have been spread so that even severe losses would be absorbed across the broad base of the economy. That was the theory. By the late summer of 2007 it was evident that the reality was different. Though mortgage-backed securities had indeed been sold far and wide, lethal pockets of risk were concentrated in some of the most vulnerable nodes of the shadow banking system.
The first mortgage issuers to die were in the bottom tier.2 The aptly named Ownit Mortgage Solutions, one of the feeders for Merrill Lynch’s securitization pipeline, was the first to go, on January 3, 2007. On February 8, 2007, the crisis moved up the food chain when HSBC, whose offices spanned Hong Kong, Shanghai and London, announced it was making a $10.6 billion provision for losses on mortgage investments. On March 7, Ben Bernanke was still in a sanguine mood, declaring that he thought the subprime problem was contained. But the bad news kept coming. In April New Century Financial, the largest stand-alone subprime lender, folded. May saw Swiss megabank UBS announce that it was closing its Dillon Read Capital Management hedge fund.3 On June 22, Bear Stearns was forced to bail out two funds that had made heavy losses on MBS. By then it was clear that the foundations were shaking. It was in the late summer that the full scale of the financial fallout began to become clear. On July 29, 2007, the small German lender IKB had to be bailed out by a consortium of banks with public backing.4 On August 8, 2007, another of Germany’s overextended regional banks, WestLB, announced outsized losses on a real estate fund and stopped payouts. Within days it was followed by Sachsen LB. But the really decisive break in market confidence came on the morning of August 9, 2007, when BNP Paribas, France’s most prominent bank, announced that it was freezing three of its funds.5 The explanation Paribas offered marked a decisive moment in the opening of the crisis: “The complete evaporation of liquidity in certain market segments of the U.S. securitisation market has made it impossible to value certain assets fairly regardless of their quality or credit rating.”6 Without valuation the assets could not be used as collateral. Without collateral there was no funding. And if there was no funding all the banks were in trouble, no matter how large their exposure to real estate. In a general liquidity freeze, the equivalent of a giant bank run, no bank was safe. As the implications of the announcement from Paris sank in, around noon Central European Time on August 9, 2007, the cost of borrowing on European interbank markets surged.7 As one senior bank executive commented, the event was disorientating: “It was something none of us had experienced. It was as if your entire life you had turned the spigot and water came out. And now there was no water.”8
The ECB did not at this point have data on the subprime exposure of Europe’s banks. But the stress in the interbank lending market was all too obvious. In response Jean-Claude Trichet and his colleagues opened the liquidity tap, offering funds at attractive rates in unlimited quantities. By the end of the day on August 9, Europe’s banks had taken 94.8 billion euros, and they took another 50 billion on August 10. It was the scale and urgency of this action that finally brought home to Ben Bernanke and Hank Paulson the true severity of the situation. As Larry Elliott, economics editor of the Guardian, commented: “As far as the financial markets are concerned, August 9 2007 has all the resonance of August 4 1914. It marks the cut-off point between ‘an Edwardian summer’ of prosperity and tranquility and the trench warfare of the credit crunch—the failed banks, the petrified markets, the property markets blown to pieces by a shortage of credit.”9 Quite how bad things were soon going to get was suggested three weeks later, when on September 14, Northern Rock, one of Britain’s largest mortgage lenders, failed. On TV screens, the Northern Rock panic looked like a classic bank run. Anxious depositors queued up outside beleaguered bank branches to retrieve their funds. News photographers and camera crews had a field day. But off camera something even worse was happening. The trillion-dollar global funding market was shutting down.10
I
Northern Rock was a product of Britain’s overheated housing bubble. Formed in the 1960s through amalgamation of two nineteenth-century building societies (thrifts) and headquartered in gritty Newcastle, it had by the 1990s acquired fifty-three competitors across the north of England. To create the platform for further growth in October 1997, it converted from thrift status to a public limited company and floated on the London Stock Exchange. Then, between 1998 and 2007, in a gigantic surge of growth, it quintupled its balance sheet. As house prices faltered, some of its more marginal loans were primed to go bad. It was no surprise that “the Rock” got into trouble. But the obviousness of this connection is deceptive. What triggered the collapse in 2007 were not the loans on its balance sheet but the mechanism of their funding. Northern Rock was the model of a modern highly leveraged bank: 80 percent of its funding was sourced not from deposits but wholesale, at the lowest rates global money markets would offer. The bank’s 2006 annual report gives an idea of this far-flung funding operation:
“During the year, we raised £3.2 billion medium term wholesale funds from a variety of globally spread sources, with specific emphasis on the US, Europe, Asia and Australia. This included two transactions sold to domestic US investors totalling US$3.5 billion. In January 2007, we raised a further US$2.0 billion under our US MTN [medium-term notes] programme. Key developments during 2006 included the establishment of an Australian debt programme, raising A$1.2 billion from our inaugural issue. This transaction was the largest debut deal in that market for a single A rated financial institution targeted at both domestic Australian investors and the Far East.”11
Northern Rock had minimal exposure to US subprime. But that didn’t matter, because it sourced its funding from markets heavily used by banks that did. The bad news from Paribas on August 9 was enough to shut down the interbank lending markets and the market for asset-backed commercial paper. It was the seizure in the funding market that poleaxed the entire securitization business and in particular the European side, which had been most actively involved in the issuance of ABCP. Given Northern Rock’s extreme dependence on wholesale funding, it took only two working days after the markets dried up for the bank to notify the Financial Services Authority of an impending crisis.12 But the Bank of England was in no mood to help. Governor Mervyn King took the view that the overextended mortgage lender should suffer the consequences of its irresponsible expansion. By the end of August Northern Rock’s liquidity problems had become life threatening. But it wasn’t until September 13, after the BBC reported the story and the government acted to address the crisis by announcing a guarantee, that the retail depositors panicked. After that, the main damage to Northern Rock’s balance sheet was done by online withdrawals. The elderly savers queuing in the streets made for alarming TV footage. But it was not their panic that was bringing down the bank. It was a bank run operating on an altogether different scale at the speed of computer terminals in money markets across the world. It was a bank run without deposit withdrawals. There had been no deposits. There was nothing to withdraw. For banks to find themselves a trillion dollars short, all that needed to happen was for major providers of funding to withdraw from the money markets.
The Rise and F
all of Commercial Paper and Repo Financing, 2004–2014
Source: Tobias Adrian, Daniel Covitz and Nellie Liang, “Financial Stability Monitoring,” Annual Review of Financial Economics 7 (2015): 357–395, chart 15.
ABCP was always the weakest link in the shadow banking chain. Repo, as fully collateralized lending, was supposed to be safe. Initially, that expectation was borne out. Bear Stearns, the smallest of the US investment banks, reported the first loss in the firm’s history in the first quarter of 2007.13 As was common knowledge, it was heavily involved in mortgage securitization. That was enough to restrict its access to commercial paper markets. The bank’s ABCP issuance plunged from $21 billion at the end of 2006 to $4 billion a year later. Initially, Bear was able to make up for this shortfall by increasing its repo funding from $69 billion to $102 billion. To back this up, as late as Monday, March 10, 2008, Bear still held an $18 billion “pool” of ultraliquid, high-quality securities. But then collateralized borrowing began to fail too.
Unlike the implosion of ABCP, the “run on repo” was a surprise.14 Under British and American law, the holder of repo collateral is entitled to seize it ahead of any other claimant in the bankruptcy queue. So even allowing for Bear’s large portfolio of toxic mortgage-backed securities, its repo ought to have been good. A Treasury security is a Treasury security. Unfortunately for Bear, given that there were plenty of other counterparties to engage in repo trades with, no one wanted to take the risk of having to seize collateral from a failing bank, even if the collateral was as highly rated and as liquid as US Treasurys. When news of a new round of mortgage failures hit the markets in March 2008 and hedge funds began emptying their prime brokerage accounts, quite suddenly the haircuts Bear Stearns faced in the bilateral repo market steepened and access to trilateral repo funding was shut off. A bank that in early March had easily been able to raise $100 billion overnight in exchange for good collateral could no longer fund itself. On Thursday, March 13, with its liquidity reserve down to only $2 billion, Bear’s directors were told that $14 billion in repos would not “roll” the next day and that they were at imminent risk of running out of cash. This was a modern bank failure. There were no queuing depositors. Bear did not cater to pensioners. It died, because doubts about its business led it to be cut out of wholesale funding markets.
Then something even worse began to happen. The uncertainty spread from individual weak banks to the entire system. First in the spring of 2008 and then in June, the haircuts on bilateral repo took a severe step up across the board, for all asset classes, for all parties.15 This meant that the amount of capital that was required to hold the outstanding stock of bonds leaped upward, across the entire banking system. Repo in US Treasurys and GSE-backed mortgage-backed securities was the least badly affected. As top-quality collateral they were reserved mainly for use in triparty repo overseen by JPMorgan Chase and Bank of New York Mellon. As long as a counterparty remained in good standing and had top-quality collateral, that repo market remained open and stable. But in the interbank bilateral repo market where private label ABS was used as collateral, funding terms were getting stiffer and stiffer.16
Haircuts on Repo Agreements (%)
Securities
April ’07
August ’08
US Treasurys
0.25
3
Investment-grade bonds
0–3
8–12
High-yield bonds
10–15
25–40
Equities
15
20
Senior-leveraged loans
10–12
15–20
Mezzanine-leveraged loans
18–25
35+
Prime MBS
2–4
10–20
ABS
3–5
50–60
Source: Tobias Adrian and Hyun Song Shin, “The Shadow Banking System: Implications for Financial Regulation,” Federal Reserve Bank of New York Staff Reports 382, July 2009, table 9. Based on IMF Global Financial Stability Report, October 2008.
The step up in haircuts would put huge pressure on the investment banks that relied most heavily on short-term funding markets. And it was clear which of those, after Bear, was most vulnerable. The warning signs at Lehman were unmistakable.17 Like Bear, it was known to have taken huge risks on real estate in the hope of catapulting up the Wall Street league table. It had fully integrated its business with the mortgage securitization pipeline. Since the beginning of 2008, the bank’s stock had lost 73 percent of its value. As at Bear, commercial paper issuance by Lehman fell from $8 billion in 2007 to $4 billion in 2008. Nevertheless, on May 31, 2008, its liquidity pool, which was intended to cover cash outflows over a twelve-month period, was as high as $45 billion.18 In June 2008 investors were sufficiently confident to commit $6 billion in new share capital. What pushed Lehman over the edge were collateral calls by anxious lenders. Given the falling value of its stock, J.P. Morgan demanded large postings of collateral to back up daytime triparty repo risks. By Tuesday, September 9, allowing for liens on its assets, Lehman’s liquidity pool was down to $22 billion. Two days later, on Thursday, September 11, Lehman was still posting $150 billion as collateral in the repo market.19 But then confidence broke. S&P, Fitch and Moody’s all downgraded Lehman. Its share price fell and with that went its standing in the repo markets; $20 billion in repo did not roll and J.P. Morgan demanded $5 billion in collateral to sustain even the most essential part of Lehman’s triparty repo business. Within a matter of hours on Friday, September 12, the Lehman liquidity pool was down to $1.4 billion and it was clear that, barring a weekend rescue, it would be forced to file for bankruptcy.
On Monday, September 15, as Lehman’s staff around the world stumbled dazed out onto the pavement, the question was who might be next. Bear and Lehman were badly run. Under intense competitive pressure they made high-risk bets on some of the worst parts of the mortgage securitization business. But they were not exceptional. Merrill Lynch too had huge real estate exposure, and it was funding $194 billion of its balance sheet on a short-term basis in the summer of 2008.20 In total, prior to the Lehman bankruptcy, $2.5 trillion in collateral was posted in the triparty segment of the repo market alone on a daily basis. This gigantic pile of claims and counterclaims could become destabilized in a matter of hours. Market analysts recognized the bimodal quality of this experience. It was a “massive game theory,” one commented.21 In trilateral repo, given the unimpeachable quality of the collateral used, there was effectively no price adjustment mechanism. One day the investment banks, dealers and those they borrowed from and lent securities to all functioned as a gigantic trillion-dollar machine based on confidence and widely acceptable collateral. The next day even a very large player in the system could be shut out.
Quantites of Assets of Various Classes Financed by Lehman Brothers Through Tripart Repos
Source: Adam Copeland, Antoine Martin and Michael Walker, “Repo Runs: Evidence from the Tri-Party Repo Market,” Federal Reserve Bank of New York Staff Reports 506, July 2011 (revised August 2014).
After Lehman, the next link in the shadow banking chain to come under acute pressure was AIG, the insurer. In a dramatic burst of expansion from the 1990s onward, the Financial Products divisio
n of AIG had developed into a major player in the derivatives markets. In total in 2007 it had a book of $2.7 trillion in derivatives contracts.22 Of this total, credit default swaps accounted for $527 billion. Of these, $70 billion were on mortgage-backed securities, and of those, $55 billion had exposure to dangerous subprime. Given its inside knowledge of the property market, AIG had stopped writing new CDS already in 2005. But given the relatively small size of the portfolio and the AAA rating of the assets it had written CDS on, it had not thought it necessary to insulate itself against losses. It was a fatal mistake. Out of a total of 44,000 derivatives contracts on the books of AIGFP, there were, it turned out, a cluster of 125 CDS on mortgage-backed securities that were about to go bad in a spectacular way. Those 125 contracts would inflict book value losses on AIG of $11.5 billion, twice what the ill-fated AIGFP unit had earned between 1994 and 2006. This was a heavy blow, but given its enormous global business, AIG could absorb portfolio losses on this scale. In due course the market would bounce back. Nor was AIG facing demands to pay out on MBS that had actually defaulted. As at Bear and Lehman, it was not the slow-moving crisis in real estate markets that threatened AIG. An avalanche of defaults and foreclosures would in due course grind its way through the system. But that would take years. The first credit default event on which AIG had to pay out did not occur until December 2008. The problem was the anticipatory reaction of financial markets and the fast-moving revaluation of securitized mortgages and the derivatives based on them. In the case of AIG, as it lost its top-tier credit rating, this triggered immediate margin calls from the counterparties to AIG’s insurance contracts. They wanted collateral to prove that AIG could meet its obligations if the mortgages did go bad. It was these collateral calls, running into tens of billions, that threatened to tip AIG over the edge.