The Financial Abstraction Ladder: The Story of the Global Financial Crisis
“It is difficult to get a man to understand something, when his salary depends upon his not understanding it!” — Upton Sinclair
In this paper, I present the analogy of a financial abstraction ladder to capture the causes of the financial crisis. The ladder starts at the ground level, where you have a housing market consisting of buyers and sellers. Most buyers need a loan to cover the majority of the initial cost of a home, turning to their local bank who after assessing their creditworthiness puts together a mortgage contract.
This local bank is sitting on the first rung of the financial abstraction ladder, as we have gone from a buyer with some assets and a credit history, to a contract that involves complex legal language and multiple parties. From there, the bank sells off the mortgage to another institution sitting on rung 2 that packages that mortgage along with a lot of other mortgages into a single security through a process called securitization. Climbing to rung 3, that institution turns around and sells the securitized mortgages to an investment bank who further repackages and complicates the mortgage contracts and the process of financial abstraction continues.
Each rung up the ladder represents a step away from what at rung 1 is a simple mortgage agreement. By the time you reach the top of the ladder, information asymmetries have totally distorted the free-market. In other words, the buyers of the financial contracts at rung 4 have virtually no idea what they are actually buying, save what the seller of that contract is willing to disclose.
This is a ladder constructed from the bad incentives present at each rung and held up by the very institutions responsible for correcting these incentives (rating agencies). As a result, introducing regulation that corrects individual incentives fundamentally misaligned with the wellbeing of society is the key to preventing another financial crisis of this kind.
At a high level, the macro data shows a housing market that surged 121% between 1997 and 2006. Whereas 1987–1997 saw a paradigmatic decade-rise of 31 %. Simple supply and demand economics tells us that either a restriction in supply and/or a rise in demand for houses must have occurred starting in the late 1990s to account for this price increase. Supply vs. demand-side analysis is not the purpose of this essay, although the answer seems to fall heavily in favor of the demand side.
I mention supply and demand only to acknowledge the proximate cause of the rise in housing prices. Namely, a rapid increase in housing demand (with a lagged supply response) was fueled by low-interest rates, deregulation, securitization, and a red-hot global demand for high-yield, low-risk investments.
Additionally, these factors resulted in a shocking reduction in the requirements for receiving a mortgage, as evidenced by 20 % of loans originated in 2006 going to subprime borrowers But simply naming the proximate causes of the crisis does little to get to the underlying causes. For that, let’s begin our journey up the financial abstraction ladder (FAL).
FAL Rung #1: Originate-to-Distribute for Local Mortgages
Let’s start at the bottom of the ladder, with the people responsible for writing the mortgage contracts and assessing the credit quality of borrowers. The incentives placed before local banks originating mortgages offered no positive reinforcement for originating quality loans. In fact, loan originators were paid for just that, origination.
The loans once originated were not tied to individual originators and therefore whether they succeeded or failed in the long term to be paid back was of no concern to the originator. All that had to happen was for these originators to find a buyer to “distribute” these loans too.
This is the origin of the term “originate-to-distribute” to describe the model of operation of these local institutions. These buyers that originators distributed to represent the next rung on the FAL. On rung 2, we find the Government Sponsored Entities that bought these loans and packaged them into more abstract financial instruments called mortgage-backed securities.
FAL Rung #2: GSEs and Securitization in the Mortgage Market
The securitization of mortgages plays a crucial role in the non-intuitive incentive structures present at each rung of the financial abstraction ladder. Starting in February 1970, the US government began pooling residential mortgages together into a single security called a mortgaged-backed security (MBS).
This security was initially put together by a Government Sponsored Entity (or GSE). These GSEs were established in their modern form in 1968 by the US Department of Housing and Urban Development and consist of Ginnie Mae, Fannie Mae, and Freddie Mac. Their mission is to “bring global capital into the housing finance market — a system that runs through the heart of our nation’s economy — while minimizing risk to the taxpayer.”
The specific goal of GSEs is to improve the ability of the average American to get a loan to buy a house. The political incentives for this are clear and justifiable from both an economic and social well-fare standpoint. Homeownership grounds individuals in communities, and historically delivers consistent economic returns while serving as a defining characteristic of middle-class membership.
One of the only ways to expand homeownership in the short term is to open credit up to new borrowers. This was accomplished by the creation of a diversified security (i.e. one that pulls exposure from a large set of local housing markets within the US) and then selling this security to new buyers both domestically and internationally, increasing the supply of credit to the mortgage market. This was the incentive that drove the securitization model instantiated initially in GSE-backed MBSs.
In 1992, President George H.W. Bush signed the Housing and Community Development Act amending the charter of Fannie Mae and Freddie Mac to respond to a congressional view that the GSEs “have an affirmative obligation to facilitate the financing of affordable housing for low- and moderate-income families in a manner consistent with their overall public purposes, while maintaining a strong financial condition and a reasonable economic return”. As a result, the requirements for a mortgage to be purchased by a GSEs was lowered, setting the stage for the subprime lending frenzy that was to follow.
So what effect did securitization have on the mortgage market on the ground? One metric known as the “Combined-Loan-To-Value” (CLTV) ratio sheds light on this question. The CLTV captures the relative amount of debt vs. value in a home. Taking rates of the CLTV at the time of loan origination over time gives a sense of how easy access to credit expanded the level of leverage used to purchase homes.
Obviously, with cheap credit, a buyer who is cash-rich may still decide to use more debt to finance a purchase. But the degree to which the CLTV rose alongside the Housing Price Index (HPI) is quite startling. And when taken alongside the data on subprime loan rates, paints a picture of a lending market ripe with cheap and readily available credit even to those who had very poor credit histories.
At its peak in 2005, the US CLTV at origination hit nearly 90%, up from less than 78% in 1998. Meaning that at origination, the average home purchase was financed by nearly 90 % debt despite a 121 % rise in housing prices. The implication here is that an enormous amount of new debt entered the US economy during this decade, which is why many economists have described the 2008 financial crisis as a uniquely leveraged crisis, distinct from other crises such as the technology sector collapse of 2001.
Why did these GSEs not look more carefully at the mortgages they were buying and impose stricter standards? In part, they realized that they could simply sell the MBSs they were originating and take them off their balance sheets to further their mandate of encouraging new owners of US houses. Who was buying these MBSs? The answer takes us up to the next rung of the financial abstraction ladder on which stands the investment banks.
FAL Rung #3: Originate-to-Distribute of CDOs by Investment Banks
Investment banks are at least partly in the business of issuing creative and complex financial contracts and selling them to buyers. They get paid when they successfully execute a financial agreement between two outside parties. As a result, the same incentive misalignment of originate-to-distribute was at play here that was present at the local origination level.
Investment banks started creatively packaging these mortgages originated by local banks and then packaged into MBSs by GSEs into even more complex financial instruments they could then turn around and sell to hedge funds, pension funds, and other institutions interested in gaining exposure to relatively safe real estate investments. This behavior makes sense given investment banking incentives.
In fact, it is not inherently a bad thing to try and tailor financial exposure to clients who have different risk vs. reward needs. The problem emerges when investment banks recognize their ability to manufacture information asymmetries by taking already fairly complicated instruments (MBSs) and making them vastly more complicated through further tranching and securitization. In other words, these investment banks were taking a bunch of lemons, putting a coat of paint on them and selling them as peaches.
How Investment Banks used CDOs as a Polishing Agent for Lemons
Rung 3 of the financial abstraction ladder is a critical one because it is here that we first find evidence for the use of financial abstraction as a tool for market distortion. To continue with George Akerlof’s Nobel prize-winning verbiage on information asymmetries, the paint these investment banks used to coat their lemons was the Collateralized Debt Obligation (CDO). CDOs took relatively benign MBSs, divided them up into tranches according to relative default risk and then repackaged them into CDOs.
The lower the tranche in the CDO, the higher the relative rating of the tranche as an investment. A tranche in the bottom 10 % of the CDO meant that the holder would only not receive payment after the top 90 % of the CDO defaulted and the last 10 % started defaulting as well. The catch was that as opposed to an MBS which might contain 10 % subprime mortgages, with 90 % being high quality, a CDO could be entirely made up of portions of MBSs that were 100 % subprime. The financial wizardry at play here was turning a set of entirely high-risk assets and making a security that could be sold in tranches rated to AAA-investment grade security. The same rating as government bonds!
The total percentage of subprime, alternative and prime mortgage loan related assets which made up only about 15% of the total assets in CDO products in 2000 increased to over 80% by 2006. Real estate related assets became the main collateral in the CDO deals during the securitization boom. Clearly mortgage backed CDOs are spectacular lemon paint.
According to a paper published in 2011 by the St. Louis Fed, 727 publicly traded CDOs were issued between 1999 and 2007 with a total value of $641 billion at issuance. Of that $641 billion, $420 billion was eventually written down, or 65 % of the original issuance balance. Returning to incentives, it is no wonder that despite this horrendous performance investment bankers went to such lengths to put together CDOs given that top CDO bankers earned in the neighborhood of $3 million to $4 million annually on the CDOs they created and sold.
But who were these investment banks selling too? It’s time to step up to rung #4 of the financial abstraction ladder.
FAL Rung #4: The Magnetar Trade and CDSs
Before getting into the nitty gritty on rung 4, it’s important to note that while many CDOs were made up of entirely subprime instruments, only a small fraction of the tranches sold out these CDOs were below AAA-investment grade.
These upper tranches of CDOs were also the hardest to sell. For an investment bank, the binding constraint on issuing new CDOs was finding a buyer for the junk-grade portion of the security. It was to hedge funds that investment banks turned, longtime stalwarts of high risk financial trades. The mandate of hedge funds generally is to provide “alpha” to their limited partners (investors), meaning that they need to beat market indices by taking long and short bets on specific assets.
One hedge fund operating in the mid 2000s through the financial crisis was Magentar. Magnetar was in the business during the last couple years of the financial crisis of buying the highest risk tranches of CDOs long after most smart money on wall street had seen the warning signs and sworn them off.
It is important to reiterate that the binding constraint of all CDO origination was these high risk-tranches representing just a small fraction of the total deal value. This means that Magnetars persistent demand for these high risk CDO tranches indirectly fueled far more high risk mortgage origination than the value held on their balance sheet alone reflects. This is a startling fact, given that the paper value of Magentars positions were already enormous. Between the end of September and the middle of December 2006, Magnetar supported the origination of 15 CDOs, worth an estimated $23 billion.
How bad were the tranches Magnetar was buying? According to a propublica research piece, 96 percent of the Magnetar deals were in default by the end of 2008, compared with 68 percent for comparable CDOs. Not only that, but evidence suggests that Magnetar was actually involved in the origination procedures encouraging investment banks to put the riskiest assets possible in CDOs they promised to buy, implying that Magnetar was either filled with blind optimism for the US housing market, or they were in some way profiting off of the failure of the very CDOs they owned.
The truth lies obviously with the second of these two possibilities and is evidentially true given Magnetars enormous returns of 76 % in 2007, at a time when the housing market was in free-fall. The question then is how did Magentar accomplish this? Their incentives are to make money as creatively as possible, that is simple enough. And make money they did, with Magnetar founder Alex Litowitz taking home an astounding $280 million in 2007 alone as the rest of the global economy hemorrhaged.
The answer is found in Magnetars ability to bet against the very CDOs they owned, using something called a Credit Default Swap (CDS). CDSs are derivatives that typically pay out the face value of the loan they derive from if that loan enters default. That means that if you own the rights to a loan that has a 30 year maturity (as is the case for a typical CDO) and you receive interest for a couple of years and then that loan enters default, you keep the money from the interest payments as profit after collecting the face value of the loan from the CDS issuer.
Magnetar was even more aggressive than that however, packaging up of these highly risky tranches into separate CDOs (called CDO²s or even CDO³s) then selling them off directly prior to their defaulting, increasing their overall profit.
What does Magentar tell us about incentives?
In a financial system where extremely complex financial instruments are being bought and sold, information asymmetries will run rampant and the financial incentives of individuals need to be offset by regulation promoting transparency and risk disclosures in the market.
On a final note, everything Magnetar did was totally legal, with no fines or criminal action ever taken against the firm or its employees.
Holding up the FAL: Rating Agencies
A final incentive misalignment is that of the rating agencies responsible for the assessment and classification of securities according to their risk level. In particular S&P and Moody were the two primary ratings agencies responsible for the risk classification of MBS and CDO tranches.
The famous AAA-rating is given to only the lowest level of risk, such as US treasury bonds. During the height of the securitization boom in 2005–2006, over 70 % of US residential mortgage backed CDO tranches were rated AAA by either Moody’s or S&P. By May 2009, only 52% of the tranches rated AAA were still rated AAA, only 58% of the AA tranches were still rated AA and only 14% of A tranches were still rated A. In other words, it was Moody and S&P at the bottom of the financial abstraction ladder holding it up!
What caused this fundamental misalignment between ratings given at the time of origination and the reality that played out in the market just a couple of years after? The incentives acting as a blindfold to these supposedly objective rating agencies.
The first area of incentive misalignment stems from the free market mechanisms driving the ratings market. When an investment bank went to a rating agency in 2005 to get a CDO rated, they had the choice of at least two different top tier agencies, Moodys or S&P. If S&P did not give a favorable rating on one CDO, a large investment bank could threaten to take all their business to the other rating agency, who were paid on a per-rating basis. In other words, there was no incentive to get the rating “right”, the incentive was instead to attract the very institutions who were most incentivized to sway the ratings in a way that did not align with reality, and in fact supported their active distortions of the free-market.
An important question for regulators and social designers is determining the appropriate level of blame to place on the various parties involved in causing the global financial crisis. Are small-time mortgage originators as culpable as the CEOs of investment banks for the role they played in the crisis? Clearly, the importance of the actions of one small mortgage originator pales in comparison to the damage caused by the pedal to the meddle approach to leveraged speculation in the housing market by the CEOs of some major financial institutions.
However, in aggregate, local mortgage originators may have had as great or greater a role in promulgating the crisis as the CEOs of these multi-trillion dollar institutions. So the question really boils down to what broader behavioral incentives are the most important to correct in order to avoid a future crisis of the kind we saw in 2007–08?
Let me recap succinctly: The global financial crisis of 2007–08 can be characterized as a ladder of individual and institutional actors where each rung up the ladder represents a step up in the level of financial abstraction. Starting with the local originator creating a basic mortgage contract on rung 1 stepping up to a GSE who turned the basic mortgage contract into a tranched MBS at rung 2. At each rung, a layer of financial obfuscation is added. What starts out as a mortgage on rung 1 ends up as a Magentar manufactured CDO³ on rung 4 being sold to Japanese investors who are unaware of the true financial relationship Magnetar has to the instrument due to their CDS positions and other relevant information.
In short, the layers of financial abstraction were packed onto the underlying mortgages until the level of un-regulated information asymmetries reached such a level that complete junk (subprime mortgages) were being sold at investment grades prices to unwitting international buyers with little or no connection to the underlying instruments. The institutions responsible for correcting these asymmetries, the rating agencies, were private market actors whose own incentives structures drove them to not only not prevent this kind of market distortion, but to actively enable it by assigning unrealistic ratings. In other words, the rating agencies were the ones on the ground holding the ladder to prevent it from falling, until the weight became too great and gravity took over.
For policymakers, more transparency must be mandated (and has been) in the wake of the crisis to prevent this level of engineered information asymmetry that distorts the principles of free-market capitalism founded on the premise of free and open access to information for all market participants.