The Boy in the Bubble

Did widespread fraud really drive the housing crisis? This former Citigroup employee was inside the bubble when it popped.

In the financial crisis of 2008, I lost my job at Citigroup. Thousands of other people in the financial services industry, and millions outside it, also lost their jobs, but, in my case, unemployment might appear an appropriate punishment for my sins. From 2000 until my ejection from Citigroup, I wrote prospectuses and other documents for mortgage-backed securities (MBS) issuances by CitiMortgage, a Citigroup subsidiary. The received history of the financial crisis casts MBS as the major villain, which would make me a villain-enabler.

Renewed attention to the causes of the financial crisis has been sparked by the film The Big Short, adapted from Michael Lewis’s 2010 book The Big Short: Inside the Doomsday Machine, as well as by candidate Bernie Sanders inveighing against all things Wall Street. The book, the movie, and the candidate all assert or imply that the financial crisis was largely propelled by fraud.

In the book, the word “fraud” appears 19 times, “corrupt” and its cognates 11 times, and “criminal” four times. I haven’t been able to check the full movie, but the two-minute trailer manages to use “fraud” three times (and “shady” once), beginning with “When the banks committed the greatest fraud in U.S. history ….”

Many critics have bought the story. Thus, A.O. Scott reviewing the film in The New York Times: “I don’t condone mob violence and I’m supposed to keep my political opinions to myself, but as soon as I’m done writing this I’m going out to the garage to look for a pitchfork,” and then “There is no happy ending to this story, no punishment for the crime.”

The Big Short retails two standard allegations of systemic fraud. First, it argues that lenders issued mortgages to people whom they knew could not afford them. Second, The Big Short alleges that the various financial instruments structured from these mortgages—the MBS and “collateralized debt obligations” (CDOs)—were misrepresented to investors. A trio of economists for the Federal Reserve Banks of Boston and Atlanta have exploded most of these fraud-based theories, but it’s a mythology that fits in so well with popular predilections (the “Wall Street vs. Main Street” mantra) that it will probably never go away.

What nobody told the homebuyers, because hardly anyone believed it, was that the value of their homes would sink dramatically over the next few years. In the preceding years, there had been a huge run up in house prices, fueled by the mass belief that if you didn’t buy now, you’d be priced out in the future—a classic bubble-builder. Much of the 2008 crisis can be traced to this single misperception.

If everyone believed that housing prices would continue to rise, then most of the behavior of homebuyers and the mortgage industry makes sense. In this mindset, if a homebuyer couldn’t make a monthly payment, he or she could just take out a bigger mortgage on the increased value of the home: The rising tide would lift all boats. Both buyers and lenders somehow came to believe that low-income homebuyers were now being allowed into the great moneymaking machine of homeownership, a machine that had always been available to middle- and upper-income folk. In the prevailing optimism of the time—the years of “the Great Moderation”—few considered the possibility that home prices would go down.

The theory that there was a systemic fraud against homebuyers always depended on a major implausibility—that you can have a successful business model based on lending large amounts of money to people who can’t pay you back.

In the real world, however, offering mortgages to buyers with sketchy credit histories, who put down almost no money and who paid no interest (the foregone interest payments being simply added to the principal amount of their mortgages) led to a frenzy of speculation, personified by The Big Short’s Las Vegas stripper who owned five investment properties. A good deal of mortgage fraud involved buyers claiming on their mortgage applications that they intended to live in their houses when they just wanted to flip them for a higher price. When the bubble burst, I had to take calls from MBS investors—fortunately, only a few—claiming that they had been defrauded, and it was this kind of fraud these investors were talking about.

CitiMortgage, headquartered in O’Fallon, Missouri, originated mortgages, most of which it securitized or sold. It also securitized mortgages bought from other originators. One reason for the securitizing was that CitiMortgage serviced all the mortgages it securitized, for a fee of 0.25% per year of the principal amount of the MBS.

A servicer records how much each homeowner has paid, calculates how much he or she still owes, sends out the monthly bills, tracks homeowners who fall behind, and manages the foreclosure, if it comes to that. It’s a profitable business if you can exploit its economies of scale; the five largest servicers had about 60% of the market. On my only visit to O’Fallon, I was shown the computer room—aisle after aisle of metal shelving packed floor to ceiling with computers. There are thousands of rooms just like it throughout the world, but it was the first one I’d seen, and I was impressed.

While I toiled for CitiMortgage, Citigroup had another MBS program, CMLTI (pronounced “See Milty”; I don’t remember what the letters stood for), run out of the investment bank. Unlike CitiMortgage, CMLTI neither originated, serviced, nor owned mortgages. It was purely a securitization middleman, buying mortgage pools originated by others, packaging them, and selling them off to the public as MBS under a “shelf registration” statement filed with the Securities and Exchange Commission (SEC)—an arrangement known in the trade as a “rent-a-shelf.” I had little contact with CMLTI; its prospectuses and other documents were drafted by Thacher Proffitt & Wood, a 300+ lawyer securitization powerhouse that in late 2008 followed its clients into oblivion.

CitiMortgage sold most of its smaller mortgage loans to Fannie Mae or Freddie Mac, and packaged the larger fixed-rate “jumbo” loans as MBS. But CitiMortgage held on to most of its jumbo adjustable-rate mortgages (ARMs). ARMs seemed a safe investment, since they took away the interest-rate risk that had cratered so many savings and loan associations (S&Ls) in the 1980s. (When interest rates spiked in the ‘80s, many S&Ls found themselves paying 6% for deposits while only receiving 3% on their fixed-rate 30-year mortgages. You can’t do that for very long, and they didn’t.) It’s hard to explain this behavior, which eventually proved quite costly, if CitiMortgage believed the loans, most of which it had originated, would fail.

When house prices finally turned down, they took a large part of the mortgage industry with it. Countrywide, Washington Mutual, and many other banks and brokers had retained too many of the mortgages they originated. Obviously, they didn’t see much farther than the homebuyers. Again, they wouldn’t have held such large numbers of these mortgages if they thought the homebuyers would default.

The banks were betting against a mortgage meltdown, not only by holding mortgages and MBS, but also by selling credit default swaps (CDSs) practically to the day the market turned. (A CDS is like an insurance policy on a bond: You buy the CDS, and only get paid if the bond defaults. The heroes of The Big Short—money managers who, smarter than the rest of us, bet on a catastrophic downturn and won big—were major buyers in the CDS market.)

Contrary to popular belief, banks don’t like to foreclose. The best a bank can do on foreclosure is to break even by recovering its unpaid principal, missed interest payments, and foreclosure expenses; any remaining foreclosure proceeds go back to the homeowner. Where real estate values have tumbled, it’s almost always better for the bank to keep the homebuyer in the house making some payments. But this was difficult, because of an under-appreciated, and to some extent accidental, aspect of securitization.

In my last months at Citigroup I would field several calls a week from CitiMortgage’s servicing people who were trying to work out payment plans for delinquent homebuyers. What they—the servicer, the homebuyer, and if they had been consulted, most MBS investors—would have liked was to reduce the principal and interest on the mortgage to a reasonable level, one that would keep the homeowner in the house making at least some payments. But any reduction in principal or interest was barred by the securitization documents, which could only be amended by a vote of two-thirds to 100% of the bondholders, a procedure that was never attempted. The best we were ever able to do was to defer payments, which didn’t help much, and we weren’t always able to do that; it depended on how the documents, which had evolved over the years, read for the particular securitization.

The securitization documents were written by people who had never experienced the kind of real estate market that developed after 2006. If they had, the documents would have allowed more flexibility. (I believe that for mortgages still held by the banks—that is, not securitized—homeowners could often get reduced principal or interest rates, but I had no direct experience with that process.)

When the dam finally broke, many firms in the MBS world went under, or only survived because of federal loans (popularly, “bailouts”). Some of the firms’ grandees got out early and profitably, but others lost millions when their firms collapsed.

About those bailouts, which have been pilloried by both the right (the Tea Party) and the left (candidate Sanders): We often think of companies like people, which is handy in many situations, but not here. In most corporations, there are three main groups of “stakeholders”: the stockholders, who own the company; the creditors, who are owed money by the company; and the employees, who work for the company. Federal loans only bailed out the creditors. When the regulators saw the panic that followed Lehman Brothers’ collapse, they staved off total disaster by making huge loans to the big banks. This was good for the creditors, whose loans would now be repaid, but gave only modest relief to the stockholders. The stockholders didn’t lose everything, just most of everything. For example, as late as April 2007, a share of Citigroup traded in the low $50s and paid a $2.16 dividend. At the bottom, you could buy Citi for $1 a share (it’s still under $5, adjusted for share splits), and the dividend swiftly fell to less than a penny, only lately rebounding to two cents. (Like many Citigroup employees, part of my compensation had been in company stock, most of which I rode all the way down.) As for the employees, a good number, including me, lost their jobs.

The theory that there was a systemic fraud against homebuyers always depended on a major implausibility—that you can have a successful business model based on lending large amounts of money to people who can’t pay you back. To overcome this difficulty, the theory requires that the mortgage originators sell off their dud loans at face value to unsuspecting investors.

Were investors misled? Did I mislead them? Should I be keeping an eye out for A.O. Scott and his pitchfork?

Before 2000 I had written articles about making prospectuses more comprehensible, so when the SEC decreed that prospectuses should be written in “plain language,” I found myself conscripted into the MBS world, with a brief to rewrite CitiMortgage’s prospectuses. My initial collision with the basic MBS documents—the prospectus, which is the informational document given to investors, and the “pooling and servicing” agreement, which is the contract that sets the rules for the issuance—left me floundering. I couldn’t make sense of them at all. Over the next few months, and with lots of hand-holding by our expert outside counsel, I was able to figure out what was going on, and then gradually to simplify the documents so that a clever reader who was new to MBS could understand them (sort of).

MBS documents are like wiring diagrams: They don’t make any sense to most people, but they have plenty of critical information for trained readers. MBS investors were almost exclusively institutions such as pension funds and insurance companies, and those people knew how to read a wiring diagram. My diagrams may have been simpler and clearer than others, but professional investors understood the more complex diagrams as well.

Why were the documents so convoluted? Partly because they had grown by accretion over the years, with various modifications introduced piecemeal to satisfy investor and regulatory demands but without anyone rethinking the overall structure. Also, the lawyers insisted on saying everything in words rather than using equations or graphical aids. For months I tried to get approval to purchase a $200 graphics software package, but was stymied until that happy day when the SEC “suggested” that our prospectuses might benefit from graphical aids. Here’s the resulting graphic for our MBS flow of funds – i.e., how the payments made on the mortgages underlying the securities would be distributed to bondholders:


Don’t worry if you don’t understand the terminology. Just think how you might try to explain this using only words. It’s possible to say it precisely, but nearly impossible for a reader to follow. (You think the flowchart overly complicated? A lawyer friend once constructed a flowchart for a credit card securitization that covered five pages and contained 89 boxes.)

Some of my drive to simplify the documents was motivated by self-preservation. Our issuances closed at month’s end, when we often had two or three MBS offerings closing on consecutive days. Because the terms of the offering were constantly being adjusted to make the securities more attractive to investors, and because mortgages might get kicked out of the pool at the last minute, the numbers would change right up to the night before closing. The challenge was to make the documents—crammed with dozens of tables and thousands of numbers—easy to create, and then easy to revise the day (and night) before the closing, which was when a great deal of the work got done. (I used to joke that it was an easy job—I only worked two or three days a month.) Simplifying the documents meant that on the night before a closing I could think about going home before midnight.

Lewis’s book asserts that investors were unable to get information on the particular mortgages in the various mortgage pools. True, the prospectuses only gave average numbers for the mortgage pool as a whole—“37% of the mortgages are in California,” “the average down payment is 18.7%”—but not information for individual mortgages, so-called “loan-level detail.” But much of that information was, in fact, available.

MBS issuances are structured with different “tranches” (French for “slice”) that defined all the different ways you could slice and dice the flow of money coming from a pool of mortgages (as well as who got paid first). For example, there are tranches with different ratings (AAA, AA, and so on, going all the way down to below B), depending on the likelihood that there would still be money left in the pool after the higher-rated bonds were paid. But even within the AAA group there could be dozens of tranches, with different interest rates, expected maturities, and other factors. The investment banks that bought the mortgage pool structured the tranches in cooperation with investors who were looking for specific characteristics. These investors would often demand to see the loan level spreadsheets.

Federal privacy regulations required us to “scrub” the spreadsheets before we gave them to prospective investors, so that individual homebuyers could not be identified. This meant that the spreadsheets the investors saw did not contain homeowners’ names or exact street addresses, but the investors could see the ZIP codes, buyer credit ratings (“FICO scores”), and other information for individual mortgages. And once such information was given to any investor, it had to be posted on the SEC’s website as a “free-writing prospectus,” available to everyone.

In The Big Short film, the Danny Moses character character goes to a house in Florida to talk to a defaulting homebuyer, and discovers that the house is being rented and that the name on the mortgage is that of the owner’s dog. The tenant, told that mortgage payments are not being made, is scared that he’ll be evicted. Later, there’s a brief shot of him and his family in a car with what seems to be all of their possessions.

The scene isn’t in the book, presumably because it didn’t, and couldn’t, happen. There was no way for an investor to get the name and address of a homebuyer, at least not legally. Nor is it likely that a foreclosing lender would have evicted a paying tenant; more likely, it would have thanked its lucky stars to see some cash flow from an underwater property.

The Big Short leaves the impression that there was something intrinsically crooked about the banks taking MBS securities rated BBB (the lowest “non-junk” rating) and repackaging them into CDOs rated AAA. But no hanky-panky was involved: Those BBB securities were almost certain to generate some cash flow, and it’s not surprising that the rating agencies would give AAA ratings to the CDO tranches with first dibs on that flow; other less-favored tranches of the CDO would have lower ratings, some a lot lower than BBB. In the event, however, the rating agencies wildly overestimated the size of the AAA tranches for the CDOs.

In The Big Short film, the Danny Moses character character goes to a house in Florida to talk to a defaulting homebuyer, and discovers that the house is being rented and that the name on the mortgage is that of the owner’s dog. It’s a scene that couldn’t have happened in real life.

Which brings us to the rating agencies’ role in the debacle. A popular explanation has been that the agencies gave inflated rating to securities because they were paid by the issuers; if an agency declined to give a AAA rating, it wouldn’t be used and wouldn’t be paid. The Big Short film gives voice to this theory in a scene where a Standard & Poor’s (S&P) executive (played by Melissa Leo) admits that S&P is rating securities AAA because if it didn’t, the business would go to S&P’s competitors. (There is no such scene in the book; there, the rating agency personnel defend their ratings.)

I think (along with some, though not all, economists) that a better explanation is that the rating agencies were working from the same faulty assumption as everyone else—that a substantial decline in home prices was highly unlikely. They didn’t believe there was a bubble.

My own experience with the rating agencies all related to prime and alt-A MBS, so my observations may not hold for subprime MBS. (MBS come in three flavors: prime, alt-A, and subprime. Prime has the mortgages with the most creditworthy homebuyers, subprime the least creditworthy, with alt-A hovering in-between. CitiMortgage only issued a few subprime MBS, the documents for which were drafted by the nice lady who worked down the hall.)

Most MBS were legally required to be rated by at least two “nationally recognized statistical rating organizations.” There were only four such organizations: S&P, Moody’s, Fitch, and Duff & Phelps. S&P and Moody’s were (and remain) the big fish, while Fitch is a smaller fish, and Duff is a minnow. No one would try to peddle MBS rated only by Fitch and Duff. (Only one CitiMortgage issuance was rated by Duff, and that was as a third rating. I was told that someone was doing someone a favor.) So, practically speaking, every securitization had to have a rating from either S&P or Moody’s. What’s more, investors occasionally insisted on ratings from both the big guys, so that issuances rated by both S&P and Moody’s were a bit more saleable. Under these circumstances, an issuer’s power to negotiate for higher ratings was limited.

An MBS issuance consists of tranches with ratings from AAA to below B. In rating MBS, the rating agencies’ principal decision was the level of protection that would justify a rating. That comes down to deciding how much padding there should be under each tranche: Should 5% of the issuance be rated below AAA, or should it be 4.75%?

In my experience, CitiMortgage shopped for ratings, but that shopping was limited to how much of the issuance would be rated AAA. The decision was important because AAA tranches fetched higher prices. But as a percentage of the offering, the differences between the rating agencies was small, on the order of tenths of a percent. So in a $500 million offering, one rating agency might want a $24 million cushion of below-AAA tranches while another might insist on $26 million. A financial crisis is not made of such differences. In any case, the ratings on MBS AAA tranches didn’t turn out to be that far off. Even for subprime MBS, “investors lost money on less than 10 percent of … AAA-rated securities.” (While the ratings for MBS were fairly accurate, the ratings of CDOs, which were more central to the crisis, were deeply flawed, for reasons that are not altogether clear.)

It’s often alleged that investors were misled about the investigations of homebuyers made by mortgage originators (“underwriting,” in the biz). Underwriting standards declined over the course of the boom, but this was no secret. For example, my last prospectus, in 2008, disclosed that approximately 44% of the loans (by principal amount) used underwriting policies that “require[d] proof of income and liquid assets and telephone verification of employment,” 21% “require[d] proof of income and telephone verification of employment, but [did] not require proof of assets,” 18% “[did] not require proof of income as stated on the loan application but [did] require telephone verification of employment and proof of liquid assets,” and 16% “[did] not require proof of income as stated on the loan application or proof of assets, but [did] require telephone verification of employment,” with the remaining 1% being “refinanced mortgage loans originated using streamlined underwriting policies.” Other MBS issuers may have used more relaxed underwriting procedures, but the use of such procedures was always disclosed.

What drove the crisis was the mass delusion that warped the global real estate market in the first years of the 21st century. Like Dutch buyers in the tulip mania of the 17th century, most homebuyers, lenders, and investors believed the market could only go up. Only a very few saw how big the bubble was.

Of course, information on underwriting practices was provided by the originators, who could have been lying. Perhaps some of them were. But so many of the originators hung on to so many of the mortgages, and defaults had been manageable for so many years, that it’s hard to argue that deception about underwriting standards drove the crisis.

What did drive the crisis was the mass delusion that warped the global real estate market in the first years of the 21st century. Like Dutch buyers in the tulip mania of the 17th century, most homebuyers, lenders, and investors believed the market could only go up, while more cautious folk merely thought that it could not go down. Others saw it as a bubble, but did not grasp its size or consequence. Only a very few saw how big the bubble was, and how to profit from it.

Which brings us to one of the hardest things to understand about the crisis, and about most bubbles: Lots of people saw this one coming. Michael Burry (played by Christian Bale in the Big Short movie) did his first CDS trade in May 2005. Just one month later, The Economist published a cover story on the global housing boom. They didn’t mince words: The subheading read “The worldwide rise in house prices is the biggest bubble in history. Prepare for the economic pain when it pops.” The article nicely demolished most of the “No, it’s not a bubble” arguments. Others were thinking along the same lines: Google searches for “housing bubble” peaked in August 2005. (The Washington Monthly also warned of falling house prices as early as 2004.)

Shortly after reading The Economist article, I inserted a new paragraph in the “Risk Factors” section of our MBS prospectuses:

Housing price cycle

A number of commentators have recently suggested that home prices in the United States are at a cyclical high, and likely to fall substantially in the near future. A substantial fall in housing prices could cause an increase in defaults on the mortgage loans, and would reduce the amount that could be realized on foreclosure.

And yet it continued. Nationwide, housing prices did not top out for another year, though some of the hottest areas had begun to cool. In late 2006, I added another sentence to the risk factors paragraph: “There have been reports that housing prices in some areas are currently declining.”

It wasn’t just crazy homebuyers. AIG, the insurance company that wrote the largest number of CDSs, didn’t stop selling them until early 2006. Yet the financial system turned a blind eye to what was happening. As Lewis wrote:

A smaller number of people—more than ten, fewer than twenty—made a straightforward bet against the entire multi-trillion dollar subprime mortgage market and, by extension, the global financial system. In and of itself it was a remarkable fact: The catastrophe was foreseeable, yet only a handful noticed.

Most of these people bought their CDSs as late as 2006. Kyle Bass, for example, set up a hedge fund in mid-2006 and bought CDSs on subprime mortgage bonds soon after. It doesn’t appear that others in that small and soon-to-be-very-rich group, with the exception of Burry, got in much earlier.

I remember a meeting, almost at the moment the MBS market imploded, with investors who thought the six tranches of MBS rated below AAA (the “six-pack”) were actually priced too low. They thought the foreclosure process could be improved, providing more protection to the lower-rated tranches. They arranged to buy several six-packs at a good (for CitiMortgage) price, in exchange for the right to service delinquent loans. They bought a few six-packs and, in short order, expired.

While lots of people saw what was coming, few of those people were in the mortgage or MBS business. For good psychological reasons, people who act on a set of beliefs about the world that seem to work—and things had been going swimmingly for quite some time—will not give up those beliefs unless impelled by some near-mortal shock. They were making a lot of money, and managed to convince themselves that what was good for them was good for the country. For example, people in the MBS world would often tell me (probably correctly) that by making mortgages a tradable commodity they were reducing mortgage interest rates for homebuyers. They sincerely believed that by providing mortgages to the less affluent, they were benefiting society. As success followed success, more and more people began to believe in the permanence of the housing boom. It’s no surprise that our short sellers were almost all outsiders to the MBS business—Greg Lippmann (the “Jared Vennett” character played by Ryan Gosling in the film) being a notable exception.

According to the trio of Fed economists mentioned earlier, “top-rated bonds backed by mortgages did not turn out to be ‘toxic.’ Top-rated bonds in collateralized debt obligations (CDOs) did.” Abuses in the CDO market seem to have been at the heart of the crisis. The Big Short film has a brilliant little sequence explaining CDOs and “synthetic” CDOs—CDOs created from pooling other CDOs rather than MBS—in a way that shows just how dangerous they could be. A single MBS bond could work its way into a pyramid of CDOs, so that when the MBS bond went down it could take multiple CDOs with it. Almost no one considered the possibility that the various MBS bonds backing the CDOs could all go bad at the same time. Relatively controllable failures were thus magnified into a catastrophic meltdown. (I had no first-hand contact with Citi’s CDO business, which was largely run out of London.)

As for me, I thought there was a bubble, and that it would pop as soon as interest rates went up, but I didn’t foresee a major crisis—the amount of subprime mortgages was simply too small. I, like others, failed to comprehend what an enormous CDO elephant was balancing on the subprime MBS tortoise.

Citigroup’s day of reckoning with the Justice Department finally arrived in 2014 when Citigroup agreed to cough up seven billion dollars in penance for its MBS sins, real and imaginary. I was relieved that none of the five examples of Citigroup skullduggery recited in the Justice Department’s statement of facts involved CitiMortgage; four involved CMLTI and one involved Citigroup acting purely as an underwriter. Whatever went on at CitiMortgage, it apparently didn’t rate a marquee appearance. In any case, I was out of the game.

Howard Darmstadter

Howard Darmstadter is a retired lawyer and philosophy professor who lives in Stamford, Connecticut and writes on law, philosophy, and public policy.