During the 1990s, the market for mortgage origination fluctuated between $500 billion and $1.5 trillion and provided the raw material for MBSs.

The banks that entered multiple parts of these markets profited greatly. Banks that had experi­enced difficulties in producing growth in their other businesses found explosive growth opportunities in the mortgage securitization industry.

Obviously, the roots of the financial crisis have to be traced to this rapid ex­pansion and eventual contraction of mortgage origination. The low interest rates after 2000 provided fuel for a large surge in house purchases, house refinancing, and house price appreciation. But interest rates began to rise in 2003, and the level of origination stayed high anyway. This is because when this rocket fuel was spent, banks were hungry to keep the party going. The banks were driven by the need to originate mortgages in order to secure fees to feed the demand for MBSs which remained high in an investment environment where lots of investors who now had experience buying MBSs were looking for safe, fixed-interest bonds. The basic problem was to find new originations in order to keep the securitiza­tion business going. That meant finding households to give mortgages to who had not participated in the first wave of refinancing or purchasing new homes.

Households who had been historically excluded from home purchases (such as racial and ethnic minorities) and people with more impaired credit became the target of originators and securitizers.

It was not just the desire on the part of home loan originators to keep their loan machines going that drove the process. Without the huge demand for se­curities produced with those mortgages, the origination boom might have col­lapsed earlier. Instead, the demand for securities based on mortgages that had high bond ratings and paid relatively high rates of return meant that mortgages were not just necessary to keep originators going; they also provided the raw ma­terial for securities that were greatly in demand. For vertically integrated banks that were making money off both of these markets, it was imperative to continue to secure mortgages and to turn them into securities. There were two indicators of this. Over time, the demand for MBSs pushed securitizers to also use the mortgages they had more effectively. After 2000, the percentage of mortgages that were eventually turned into securities rose from about half to almost 90 percent. From 2005 to 2007, the demand for securities remained high, and this brought banks to repackage their hard-to-sell lower tranches into CDO-CDOs, where they could claim that the top tranches were AAA rated.

Not surprisingly, Countrywide Financial led the way to pivot toward the non- conventional mortgage market as a source for new mortgages when the mar­ket for refinancing existing homes and purchasing new homes was saturated. From 2003 to 2007, the entire industry increasingly became involved in pro­viding credit for the subprime, adjustable-rate, interest-only, home equity, and Alt-A markets. The fees associated with these loans were larger. The MBSs based on them paid higher rates of return because the mortgages they were based on charged higher interest rates for nonconventional mortgages. This pushed banks to increase their integration by buying up providers of those loans and increas­ing their output of MBSs.

But eventually, even this market began to dry up. Banks pursued mortgag­ors who had less and less ability to pay back their loans. Nearly everyone who could have refinanced their loans had done so, and most households who want­ed home equity lines had purchased them. By 2006, the quality of loans being made deteriorated. When house prices began to drop beginning in the summer of 2006, it made it more difficult for borrowers in distress to sell their homes. This process cascaded in places where prices had risen the fastest. As loans began to be foreclosed, the entire structure came under duress.

The banks didn't change their tactics when this happened, because their busi­ness model depended not just on originating mortgages but in turning those mortgages into securities that would be sold or held on their own accounts.

When the raw material for that throughput began to fail, the only real option was to double down and hope that they would be the last bank standing. Most banks had little choice but to keep going. Moreover, there was a huge worldwide demand for their end products: MBSs. The low interest rates of the early 2000s meant that investors were looking for higher rates of return on safer investments. Since most MBSs had AAA ratings and paid attractive returns, integrated banks could not get enough mortgages to produce securities to take advantage of the booming market for those securities. In the face of all of this, vertically inte­grated banks could do only one thing: try to buy and process more mortgages (Mortgage Banking, 2005; American Banker, 2007).

While it is difficult to prove, it is certainly plausible that part of what kept the housing price bubble going was the demand for MBSs. After the demand for re­financing lessened in 2004 and the Federal Reserve began to raise interest rates, originations should have dropped off much more. But because financial institu­tions needed mortgages for their MBS products, instead of following the cycle down (which is what has happened historically), they were pushed to find new mortgages to originate.

It is no surprise, then, that as the mortgages became harder to find, desperate financial institutions began to repackage tranches of MBSs with lower ratings into CDO-CDOs that could then be rated as AAA bonds. The proliferation of these products shows how large the demand for MBSs was. But the overall qual­ity of loans deteriorated as borrowers with worse and worse credit were given loans. The lower tranches of existing MBSs and CDOs were already based on loans with higher risk. Not surprisingly, it was the MBSs and CDO-CDOs that were created in 2006-2008 that had the worst financial performance. In essence, the vertically integrated model whereby banks needed to keep making securities in order to keep their vertically integrated businesses going pushed the banks to continue to scrape the bottom of the origination barrel. When that failed, they took to cannibalizing MBS tranches that they could not sell and turned them into CDO-CDOs that they could. The result was bonds that might be rated AAA but contained lots of mortgages that were guaranteed to fail, particularly as pric­es for houses stopped rising.

The final aspect of these markets was that all of the banks were running their integrated structures on borrowed money. This started out somewhat innocu­ously. Banks would borrow money to purchase mortgages in either the repo or asset-backed commercial paper (ABCP) markets. These markets are what we now call “shadow banking” (Pozsar et al., 2011). The main players in these mar­kets were other banks who would loan money at low interest rates for short peri­ods of time. The original idea was that once the loans were processed into secu­rities and sold, the lines of credit would be paid back, and the process would start again. But over time, banks used these markets to not only fund their short-term needs for capital but also borrow money to buy securities on their own accounts.

The problem here was that these loans were supposed to be short term, ninety days to a year. But banks thought that they could either roll the loans over or go back to the markets to borrow to continue to hold the bonds and, if necessary, sell them. Gorton (2010) has termed this tactic “borrowing short to go long.” This tactic proved quite lucrative. One made profit on the borrowed money with­out having to use one's own capital to fund the investment. This pushed banks to increase their indebtedness and explains why, on the eve of the crisis, banks were unable to come up with capital to cover their borrowings for mortgages and securities whose value was difficult to assess.

The vertically integrated structure of the largest banks meant that their busi­ness models were highly profitable when the origination market was deep and the demand for MBSs high. They lived off of borrowed money and used it to buy securities for their own accounts. But as the market for originations got more difficult and as the demand for MBSs began to dry up when mortgage defaults and foreclosures rose, these same banks found themselves having to come up with cash to either pay back the money they owed or increase the collateral on the money they had borrowed.

Beginning with Bear Stearns in the spring of 2007 and ending with Lehman Brothers in the fall of 2008, the vertically integrated model of mortgage securi­tization began to collapse. When banks had to cover what they had borrowed, their main assets were MBSs whose value was difficult to determine because of the defaults and foreclosures of the underlying mortgages. This created what was widely viewed as a liquidity crisis. Banks had assets, but they could not sell them fast enough to cover their short-term debts. But once banks started to teeter in the summer of 2008, it became difficult to price MBSs, and there was little mar­ket for them. This meant that many banks were unable to sell what they held at any price, and their liquidity crisis quickly meant they were insolvent.

In this chapter, I will document how this process unfolded. I begin with basic data on what happened from 2000 to 2008. Then, I consider how the industry operated to take advantage of low interest rates from 2000 to 2003. Next, I turn to how the banks shifted from conventional mortgages to unconventional mort­gages as their main source of MBSs after 2003. While this worked for a couple of years, it eventually broke down as the supply of these mortgages was limited. Banks were desperate to keep their securitization machines going, and, in es­sence, they kept going by trying harder to find such loans. I show that banks proliferated CDO-CDOs in order to meet the demands of their securities cus­tomers.