Despite extensive credit crisis postmortems, many of the widely accepted explanations of what happened are at odds with facts on the ground. These superficial explanations are hard to dislodge because they tally with widely held beliefs about how the real estate and securitization market operate. The waters have been muddied even more by self-serving PR from various market participants.
The consensus reality of the credit crisis appears to be: it was the result of a complex combination of factors, no one can be blamed all that much (save maybe greedy borrowers and complicit rating agencies) and almost no one saw it coming.
We've argued that many of the arguments that support that view are myths. In particular, the more we have dug into the CDO market, the more we are convinced that it was central to the crisis. Furthermore, we believe that this market did not operate on an arm's length basis, that many of the practices that were widespread in the industry amounted to collusion.
Collusion and resulting price distortions serve as the most likely explanations for behaviors that are consistently glossed over in the consensus accounts of the crisis. By early 2006, many mortgage market participants felt that the housing market was overheated and unsustainable. Many felt that mortgage rates should be higher, but despite interest rate tightening by the Fed, mortgage rates were not increasing. Even more distressing, credit spreads remained narrow despite widespread concerns that mortgage risk was increasing and deals were weakening.
Many economists and academics described this as a conundrum at the time and tried to come up with theories to explain it, none of which were terribly satisfying. None of them looked at a more likely culprit - the securitization market and, specifically, the CDO market.
CDOs distorted the mortgage market because they undermined the normal processes for pricing risky assets. For subprime debt, demand for the lower rated tranches had served to constrain market growth. If investors started to shun the BBB to AA rated tranches of subprime mortgage bonds, dealers were not willing to retain them, no new deals would be sold, and the market would need to find better quality mortgages or grind to a halt. But CDOs were the dumping ground for these tranches. A 1990s version of mortgage-related CDOs proved ultimately to be a Ponzi scheme (unsold risky CDO tranches were rolled into new CDOs), but even then, that CDO market imploded early enough that the damage was comparatively minor.
This time, the CDO market distortions were more significant and wide-ranging. In particular:
1. Demand for CDOs came not from long investors, who would be concerned about credit losses, but primarily from (a) short investors who wanted to bet aggressively against the housing market and needed a tool to allow them to do so without disclosing their real intentions (b) investment banks who created the CDOs so they could generate fees and bonuses by putting the CDO bonds in their trading portfolios (negative basis trades) and off balance sheet vehicles (SIVs) without regard to risk and (c) correlation traders who were indifferent to credit risk
2. The normal mechanisms for pricing risk were upended because of manipulation of the demand for mortgage and CDO bonds by a consortium of banks and CDO managers who masked the real appetite for the bonds and fabricated pricing for the bonds
3. By creating the illusion of demand for the mortgage and CDO bonds, the CDO managers and arranging banks operated under a well disguised conspiracy that allowed a massive housing bubble to be created which only exploded when the shorts became impatient for realizing their gains.
If traditional cash investors and insurers were avoiding the mortgage securities market, who was driving the yields and spreads lower? Many industry participants agreed that the "CDO bid" was distorting the market.
The mechanism was the CDO managers, who assembled the assets for cash or hybrid deals (ones like Magnetar's that used a combination of mortgage bond tranches and credit default swaps). They were effectively extensions of investment banks, dependent on substantial credit lines from them. Perhaps more important, it appears that many of the larger CDO managers bought much, perhaps all, of the AA to BBB tranches of entire subprime mortgage bond issues to be placed into CDOs. Having a single affiliated party take down the riskiest layers of subprime deals means that normal arm's length pricing was not operating, and the profit potential of CDO issuance, rather than investor demand, was driving the market.
Consider this series of interconnected transactions:
A "sponsor" indicates an interest in creating a CDO to an investment bank. In combination, the sponsor and the bank would select the CDO manager who would buy the mortgage bonds for the CDO at start up and oversee the portfolio after closing. The sponsor would typically provide the CDO manager with an investment objective and find a manager that could achieve these aims.
Since the CDO deals were typically over a billion dollars, the CDO manager didn't usually have the capital to purchase the mortgage bonds. As a result, the investment bank for the deal would offer the manager a line of credit to use to purchase the bonds that the manager selected. When the CDO closed, the CDO would repay the line of credit.
The bank for the CDO would not offer the line of credit to a thinly capitalized CDO manager casually. They were sure to get an attractive rate of interest plus a security interest in the bonds being financed to protect them in case the CDO manager ran into trouble. In addition, the CDO manager would work hard to find investors in the CDO to pay of the loan from the investment bank.
Many CDO managers were repeat issuers and many had a fairly systematic approach to how they covered the market. For instance, in a particular period, a CDO manager might be responsible for a mezzanine deal and a high grade deal or two. This would mean that the CDO manager had multiple lines of credit active
This execution strategy meant that the CDO manager had significant capital at its disposal for the purpose of buying mortgage bonds. Normally, the process of bidding on newly issued mortgage bonds while trying to meet the eligibility criteria of the proposed CDO transaction can be timing consuming and arduous for the CDO manager. The clever ones with more influence and access to generously termed lines of credit, could use their capital to tremendous advantage. Rather than face the competition of multiple bidders on a particular bond of a new mortgage deal, the CDO manager, armed with multiple upcoming deals and lines of credit, could offer to buy the entire stack of subordinated bonds that the issuer was bringing to market: BB all the way up to AA. This would be very attractive to the issuer, since it made it easier to get his deal sold. It was attractive to the CDO manager, since they could slot the bonds into both their mezzanine deal and their high grade deal at the same time, saving them a considerable amount of work. In addition, it could be very attractive for the bank on the mortgage transaction, particularly if they were the same bank that was issuing the CDO. A bank that knew it would be able to sell its mortgage deal and supply bonds to its CDO deal at the same time would take comfort that it was not terribly exposed to market risk.
One additional feature that some CDO managers might employ is to have a line of credit established for an upcoming CDO squared. A CDO-squared is made up of other CDO bonds, rather than MBS bonds. Putting aside how ridiculous the concept sounds now, this type of deal served a tremendous importance at back in 2006 and 2007. Since the riskier tranches of a CDO were more expensive to the issuers and harder to place, a CDO manager who knew that he had a home for these slices of his upcoming mezzanine or high grade CDO could certainly sleep easier. If managed properly, a CDO manager working with a friendly bank could pre-place the all of the sub bonds for a number of mortgage deals into their mezzanine and high grade deals and also pre-place all of the sub bonds from their mezzanine and high grade deals into a CDO squared transaction.
This example illustrates that pricing was often not based on market demand. Is there any real price discovery if one buyer (the CDO manager) is snapping up all of the risky tranches from a mortgage deal, and the bank on the mortgage deal is the same bank on the CDOs where the bonds will end up? Similarly, if all the risky tranches of a CDO were all pre-placed into another CDO, did anyone even bid on them? And since all of these pieces fit so nicely together, wouldn't getting competitive bids really have been rather inconvenient?
Consider the role that a company like TCW played in the market. TCW was the biggest CDO manager in the ABS CDO market. In 2006, TCW acted as manager on about $9.5 billion worth of CDOs over 7 transactions.
The deals have an interesting pattern - alternating between high grade ($5.5 billion) and mezzanine ($3.4 billion) and across four banks, Goldman, Merrill, Wachovia and Morgan Stanley. The high grade deals included not just A and AA MBS bonds but also similarly rated bonds from other CDOs, including potential the mezzanine and high grade deals managed by TCW during this period.
During this same time 2006, those four bankers owned or acquired subprime lenders who typically securitized most of their originated loans. By rotating among the lenders owned by these banks, TCW could achieve decent diversity in their CDOs without ever having to pursue other lenders for their bonds. While they certainly mixed the bonds of other lenders into the mix to achieve better diversity scores from Moody's (and lower rating agency cost of issuance, TCW may have offered to take all, or nearly all, of the mortgage bonds issued by the acquired lenders of Merrill Lynch, Wachovia, Morgan Stanley and Citigroup when they brought a subprime or Alt A mortgage deal or perhaps even the occasional deals where the banks had offered the bonds of third party mortgage lenders. If so, it's likely the offer was received well.
Consider the systemic impact. Lower costs on for the CDO translated into lower, more aggressive bids for the mortgage bonds, which translated into lower mortgage rates - all of which were potentially being set between just 4 or 5 traders
But the risky tranches represented only a relatively small portion of the mortgage or CDO transactions. What happened to the biggest portion of the transactions - the senior (AAA) bonds? The bond insurers insured a decent amount of the market in 2006 (about a third), but even the many of the insured bonds needed a buyer and the uninsured senior bonds still needed a home. As we learned last week when Citigroup testified at the FCIC, Citigroup were big buyers of their own CDOs. Just like with the mezzanine and high mortgage deal, it was probably much more convenient for bank who was selling the senior CDO bonds, to convince management to acquire the bonds themselves rather than try to sell them in a messy, time consuming bid process. Similarly, Yves discussed in ECONNED that Eurobanks frequently retained AAA tranches because Basel II rules gave them considerable latitude in how much (as in how little) capital to charge against them.
As a result, from the top of the structure - the senior bonds of a high grade or mezzanine CDO, all the way down through the mortgage bonds and into the price of the mortgage loans - third party assessments of the risk and rewards of the loan appear to have been limited to non-existent.
The result was that riskier and riskier loans were being originated at effectively lower costs for issuers with little outside feedback. In one big happy family among the mortgage issuers, CDO managers and CDO investors, there would have been little motivation to worry about increasing risk or wider spreads. They were all keen to keep the great fee machine rolling.
Finally, if you throw the shorts into the equation, you complete the picture. Hedge funds who wanted to short subprime were pushing for more and more CDS on MBS, which led to the creation of more CDOs, which in turn, bought more cash and synthetic MBS bonds, helping to keep spreads low. The tight spreads on the mortgage deals created a great buying opportunity for the shorts, who were getting to bid on what we now know were extremely risky loans at bargain basement prices. Once the risks in the mortgage loans began to emerge, spreads on the bonds finally started to widen, sometime in mid 2007. By then it was too late - the deals were already created. Since the bonds had never really been distributed very widely and sat with highly leveraged firms that could not take much in the way of losses, the result was systemic risk and financial crisis.