Authors: Gillian Tett
The Treasury, however, was persistent. “We
really
want you to take part!” the message came back. Winters continued to resist. If Citi needed the scheme so badly, he asked, why couldn’t Citi organize it all? Why lean on a bank with no shadow banks of its own? The US officials let it be known that
that
was the whole point: the Treasury wanted JPMorgan to help lead the scheme precisely because it had relatively clean hands. Robert Steel, the undersecretary of the US Treasury, called senior JPMorgan officials to press the point. Steel argued that the financial system had become infected with the banking equivalent of “mad cow disease”; there was now so much consumer anxiety that everyone had to fight back. Then Henry Paulson spoke to Jamie Dimon directly, asking him to get on board. “We are being drafted!” Kodweis grimly joked to his colleagues.
Over in London, Winters realized he was fighting a losing battle. Dimon was a savvy operator, and he could see which way the political
wind was blowing in Washington. “We need to do the right thing! This is about being a good citizen,” Dimon declared. In the end, JPMorgan told the Treasury that it would take part. Winters was determined to at least avoid exposing the bank to risk, and he insisted that JPMorgan would not provide any credit lines to the superfund unless it could control the design of the scheme. The JPMorgan bankers did not entirely trust the Citi financiers who had drawn up the initial plan. They were led by a banker who was a general expert in restructuring, rather than a specialist on CDOs; his previous job had been to reorganize Iraqi debt.
After days of wrangling, Citi and JPMorgan issued a joint press release on October 14 announcing the launch of a Master Liquidity Enhancement Conduit, or MLEC, the technical term for the superfund. Bank of America put its name on the press release, too; it did not have a particularly large presence in the structured credit world, but it did have a network of money-market funds exposed to SIV failure. Treasury officials had decided at the last minute that it would be better to get a third bank on board, to give the impression that it was an industrywide scheme, and Bank of America was a good pick.
Paulson and Steel pronounced their backing for the plan, fervently denying that it was any form of “bailout.” “I don’t know how anyone could characterize it as a bailout,” Steel testily responded to reporters’ questions. “There’s no federal money and there’s no federal organization. We were the original convener, and now the market participants are off developing a strategy themselves…. We didn’t bring the people together with an idea; we brought them together to get their ideas. I think that is an important distinction.” Irrespective of whether the government had formally organized the scheme, a great deal was riding on it. “We really don’t have a lot of time to get this sorted,” confessed Steel.
If any evidence were needed that time was running out, the ratings agencies were ready to supply it. On October 11, just as Citi and JPMorgan were fine-tuning their MLEC announcement, Moody’s cut its ratings on some $32 billion worth of mortgage-backed bonds. Those had largely been issued in 2006 and had previously carried medium-risk ratings of
around single-A or BB. The agency also warned that it might downgrade more than $20 billion of mortgage-backed bonds that carried the triple-A stamp and downgrade CDOs composed of those bonds, too. All told, the cuts affected a hefty $50 billion of securities.
That was alarming for investors. Worse still, Moody’s seemed unsure how much further the downgrades might go. “The performance, particularly in the US housing (and) mortgage sector, [has been] deteriorating more quickly and more deeply than the ratings agencies or most other participants in the market anticipated,” Raymond McDaniel, CEO of Moody’s, told the
Financial Times
on October 12. “We [have been] adjusting standards, making our credit criteria more stringent, but the incremental moves…[have been] rather overwhelmed by the pace and magnitude of the deterioration in the US housing market.” That was a polite way of saying that the subprime mortgage market was not behaving as the models had predicted. The “class of 2005 and 2006” borrowers were defaulting at a dramatically faster pace than households that had taken out mortgages before those dates. There were also variations even among mortgage lenders. By September 2007, more than 30 percent of the subprime mortgage loans issued in late 2005 by Fremont, a California mortgage lender, were in default. At Countrywide and Wells Fargo, the default rate for the same vintage was “only” 15 percent.
A particularly pernicious aspect of the defaults was that when this new breed of subprime borrowers walked away from their homes, they often left them in such a bad state that it was hard for lenders to realize any value from the repossessed properties. Until the autumn of 2007, Moody’s had assumed, on the basis of past housing cycles, that lenders could recoup 70 percent of their loans in case of default. By October 2007, it had slashed that projection to just 40 percent.
To add to the confusion, by the autumn of 2007 it seemed that in some neighborhoods of the US, feedback loops were developing that threw the ratings agencies’ models off even further. As house prices fell, defaults were rising to such a degree that they were blighting entire neighborhoods. That was pushing house prices lower still, sparking yet
more
defaults. This vicious cycle had never been witnessed in the world
of corporate defaults; nor did it fit the logic of the “bell curve” technique central to the risk assessment systems so pervasive inside banks and ratings agencies.
Predicting default rates was proving baffling enough, but it was even harder to guess how defaults would impact CDOs at the other end of the chain. The basic concept behind the tranches in CDOs was that when mortgage borrowers paid back their loans, the cash from those repayments would flow into the different tranches like a river pouring down a waterfall into several stacked buckets. The “senior” tranche would be paid first, and when those note holders had received their due, the cash would spill down to the mezzanine tranche, and so on, to the junior level. If the flows ran a little dry because mortgage borrowers were skipping repayments, there would not be enough to fill the bottom bucket. But as long as
some
water flowed, the senior note holders were safe.
In real life, though, it was proving difficult to work out precisely how and when changes in cash flows would affect the various buckets under the waterfall. For one thing, there was a time lag problem. When households went into default, the delinquency process could last for months, a sort of limbo for cash flow. When losses reached a certain predetermined level, the trustees of a CDO were supposed to declare an “event of default” and repay assets to note holders. But that process could also take months. On top of that, different CDOs had subtly different rules about how their “waterfalls” worked, in terms of who received what when. Even worse was the complexity of trying to determine how cash should flow within the CDOs of CDOs.
When the J.P. Morgan team created BISTRO, it had bundled up loans of a well-diversified pool of companies specifically to minimize the chance of widespread defaults. In making mortgage CDOs, bankers had tried to similarly diversify, by including loans from numerous regions of the US. The common assumption was that even if one region suffered a housing bust, the property market would never collapse across the country as a whole. But by the autumn of 2007, it had become clear that
this diversification theory wasn’t working in the subprime mortgage world. Defaults were rising in all regions.
The problem of cash flow was particularly vexing for those managing what had become an especially popular type of CDO during 2005 and 2006, those known as “mezzanine CDO of ABS.” These were made up out of
only
mezzanine notes—or those rated around BBB. Bankers liked to claim that there was still a high level of diversification in these structures because the mezzanine notes were linked to the loans of a vast pool of different households. In practice, though, because they were all in the mezzanine tranche, they would
all
be hit by default losses at once. Any “diversification” was an illusion.
For all of these reasons, the ratings agencies felt forced to continue slashing ratings. A few days after Moody’s cut its ratings, Standard & Poor’s put 590 CDO tranches on review for a downgrade. It also cut the rating on $3.7 billion of CDOs. Then Fitch, the third smallest rating agency, warned it might cut the ratings on $37 billion of CDOs. Most chilling of all, the agencies warned that these reviews did not affect just the junior tranches of CDO debt. The senior, or supposedly safe triple-A, CDO debt was at risk for downgrades as well. “Whereas in a CLO [a CDO built of corporate loans] it is unlikely that all the loans default simultaneously, the additional layer of securitization…in a CDO of ABS has led to an all-or-nothing scenario,” pointed out Matt King, an analyst at Citi, in a research note. “It is this all-or-nothing which is now causing significant downgrades to CDO of ABS tranches, in particular to the most senior ones.”
It was shocking news for investors. After all, the entire structured credit edifice had been built on the assumption that AAA was ultrasafe and AA almost rock solid, too. Now that pillar of faith was crumbling. It was impossible for anyone to know exactly how the downgrades might affect the value of particular CDOs. By the autumn of 2007, these were still not being widely traded, partly because bankers and policy makers had made such frantic efforts in previous months to prevent any public fire sales of CDO assets. But the behavior of the ABX index was readily visible, and by mid-October, the BBB component of this index had tumbled to 30 percent of its face value, down from 95 percent at the start of
the year. Most ominously of all, the AAA tranche of the ABX was trading at around 90 percent of face value, while the AA was falling towards 80 percent. To a casual observer, such dips might not have looked extreme. However, bankers and investors had always assumed the prices of AAA or AA assets would never move at all. They were utterly unprepared for the damage that such price falls might inflict.
On November 4, Bill Demchak dialed into a conference call arranged by the mighty Citigroup. From his perch at PNC in Pittsburgh, Demchak had spent the autumn following the events in the CDO world with mounting alarm. Back in 2006, when his team at PNC had made the ballsy decision to start cutting the bank’s credit risk, Demchak had done so largely because he reckoned that conditions in the
corporate
loan market looked dangerous. He assumed that when the credit bubble went “pop,” it would be that area where the pain was felt first. By the autumn of 2007, it was becoming clear that Demchak’s prediction had been only half right. Conditions in the corporate loan market had indeed turned worse, because investors had become reluctant to purchase CDOs built out of risky corporate loans. The banks were left with some $400 billion worth of unsold corporate loans on their books, stuck in the securitization pipeline. What was worse, the price of those loans had fallen sharply (meaning that the cost of borrowing money for risky companies had risen, since the two move in opposite directions).
Demchak had expected as much. But by November 2007, he could see that other problems were brewing. In the middle of October, Citigroup had revealed that its net income had slumped from $6.2 billion in the second quarter of the year to $2.1 billion. The reason, it reported, was that it had been forced to write down the value of its corporate and mortgage assets by $5.7 billion. That number looked very large, but most analysts were not too surprised. Citi had a vast loan book, and other banks were announcing similar losses. On November 4, though, Citi suddenly warned that it would report more losses of between $8 billion and $11 billion. That was such shocking news that Citi CEO Chuck Prince announced that he would resign. It was also baffling to analysts. Citi was
supposed to be expert at measuring credit risk. So how had the bank managed to misjudge its losses so badly? And why was it still so uncertain about the total bill?
Demchak dialed into the conference call eager to find out. “The issue is super-senior,” one of the senior executives explained. The problem, he added, was that the bank held on its books $43 billion worth of super-senior risk linked to CDOs backed by mortgage debt. Citi had previously assumed that the value of those assets was 100 percent of face value; now the price was falling.
Super-senior?
Demchak could hardly believe what he had heard. Almost a full decade had passed since Varikooty, Demchak, and the rest of the group had invented the term. Back then, the term had seemed like a geeky in-joke; a concept so quirky and obscure that only a few technical experts knew what it meant.
Now the Citi executives had casually tossed the word
super-senior
into a conference call with hundreds of mainstream investors, analysts, and financiers. Demchak didn’t know whether to laugh, cry, or just shake his head in wonder. In other circumstances, he might have felt almost proud that his team’s once-obscure brainchild had suddenly burst into the limelight. In reality, though, he was overwhelmed with horror. The way Citi told the story, super-senior had turned into a scourge that had created most of its unexpected losses.
“How could this happen?” Demchak wondered. Back in the days when he had been chasing his credit derivatives dreams at J.P. Morgan, his team had considered super-senior to be so safe that it was “more than triple-A.” Even though Demchak himself had gone to great lengths to sell J.P. Morgan’s super-senior risk to AIG and other insurance groups, he had never imagined for a second that super-senior debt could pose more than a moderate level of risk. Nor had he guessed that Citi was holding so much on its own balance sheet. Citi had never discussed the issue on conference calls before or highlighted it in previous corporate reports. “How did this happen?” Demchak asked himself again and again. As he listened to the rest of the call, he got the distinct impression that the Citi managers were almost as baffled as he was.