João A.C. Santos
There’s ample evidence that securitization led mortgage lenders to take more risk, thereby contributing to a large increase in mortgage delinquencies during the financial crisis. In this post, I discuss evidence from a recent research study I undertook with Vitaly Bord suggesting that securitization also led to riskier corporate lending. We show that during the boom years of securitization, corporate loans that banks securitized at loan origination underperformed similar, unsecuritized loans originated by the same banks. Additionally, we report evidence suggesting that the performance gap reflects looser underwriting standards applied by banks to loans they securitize.
Historically, banks kept on their books the loans they originated. However, over time they increasingly replaced this originate-to-hold model with the originate-to-distribute model, by syndicating the loans they originated or by selling them in the secondary loan market. The growth of securitization provided banks with yet another opportunity to expand the originate-to-distribute model of lending. The securitization of corporate loans grew spectacularly in the years leading up to the financial crisis. Prior to 2003, the annual volume of new collateralized loan obligations (CLOs) issued in the United States rarely surpassed $20 billion. Since then, this activity grew rapidly, eclipsing $180 billion in 2007.
Corporate loan securitization appealed to banks because it gave them an opportunity to sell loans off their balance sheets—particularly riskier loans, which have been traditionally more difficult to syndicate. By securitizing loans, banks could lower the risk on their balance sheets and free up capital for other business while continuing to earn origination fees. As with the securitization of other securities, the securitization of corporate loans, however, may lead to looser underwriting standards. For example, if banks anticipate that they won’t retain in their balance sheets the loans they originate, their incentives to screen loan applicants at origination will be reduced. Further, once a bank securitizes a loan, its incentives to monitor the borrower during the life of the loan will also be reduced.
To investigate whether securitization affected the riskiness of banks’ corporate lending, my paper with Bord compared the performance of corporate loans originated between 2004 and 2008 and securitized at the time of loan origination with other loans that banks originated but didn’t securitize. We found that the loans banks securitize are more than twice as likely to default or become nonaccrual in the three years after origination. While only 6 percent of the syndicated loans that banks don’t securitize default or become nonaccrual in those three years, 13 percent of the loans they do securitize wind up in default or nonaccrual. This difference in performance persists, even when we compared loans originated by the same bank and even when we compared loans that are “similar” and we controlled for loan- and borrower-specific variables that proxy for loan risk.
Our results suggest that banks use laxer standards to underwrite the loans they sell to CLOs. For example, we find that banks put less weight on the “hard” information on borrower risk when they set spreads on the loans sold to CLOs than on the loans they don’t securitize. We also find that banks retain less “skin in the game” when it comes to securitized loans, suggesting that they have less incentive to monitor these loans after origination. While on average banks retain 26 percent of each syndicated loan they originate but don’t securitize, they retain only 9 percent of each loan they do securitize. This difference in underwriting standards may help explain why banks’ securitized loans underperform unsecuritized loans.
Finally, we find evidence that all loan investors, including banks, expect that securitized loans will perform worse. Banks appear to do so because they charge significantly higher interest rates on these loans than on the loans they don’t securitize. Institutional investors, who together with the originating bank and CLOs acquire the loans that banks securitize, follow the loan originator and choose to acquire a smaller stake in securitized loans.
Our evidence that securitization led to riskier corporate lending is in line with similar findings unveiled by studies of the effects of securitization on mortgage lending. Taken together, these studies confirm an important downside of securitization.
The views expressed in this post are those of the authors and do not necessarily reflect the position of the Federal Reserve Bank of New York or the Federal Reserve System. Any errors or omissions are the responsibility of the authors.
João A.C. Santos is a vice president in the Research and Statistics Group of the Federal Reserve Bank of New York.
Ann, I think your understanding is wrong. If 16 years of data shows that the investments are relatively safe (even after surviving the worst financial crisis since the Great Depression), then the gaming has already occurred, as the regulatory motivations did not match the proper underwriting of the investment. Underwriting is a term that comes from the insurance industry and infers that the probability of default and the expected loss given default are accounted for in the pricing of a loan within a well diversified portfolio of loans. There is no expectation that no loan will go bad, only that, on a portfolio basis, the unanticipated losses will be covered by the premium within the pricing. This is what has happened in those CLO structures without junk bonds. In your example, Alt-A paper did outperform conforming mortgage paper. When demand increased, unscrupulous brokers and loan packagers (investment banks) were motivated to create new supply and the definition of Alt-A paper changed, without any repercussions. Perhaps this is where the regulators should have focused their efforts. When a regulator forces lenders to reserve against well-underwritten loans at levels that are significantly greater than the trading discount in the secondary market or the expected ultimate loss, they discourage well-underwritten investing (an opportunity cost) and create an arbitrage opportunity for smart investors outside of the regualtors’ control. This will always be the case.
When an NRSRO publishes research stating that historical defaults or losses on an asset class have been low, they create an incentive to game the historical statistic by diluting the underwriting standards without notification to investors. The classic example was in 2003, with Jay Siegal’s paper on Alt-A residential mortgages outperforming expectations in a rising real estate market. Right there, the incentive to delay recognition of falling real estate prices as long as possible was established.
Interesting read. I’m concerned your data is biased by using the SNC database. Clearly, syndicated loans are riskier, as defined by the regulators. However, the reason that syndications exploded in the early 2000’s was market investors felt the risk-return characteristics of leveraged loans were mis-priced and there was an arbitrage play, within the CLO structures. There is no feedback loop to tell regulators when they are being too conservative. What might be more meaningful would be an analysis of syndicated loans based on ultimate loss (or recovery) following a default. Despite being more likely to default, these loans trade at near par in the secondary market, indicating the market expects full recovery under absolute priority in a bankruptcy process (usually a “worst case” scenario). I believe that actual losses within syndicated loans since 2007 have been minimal within a well diversified portfolio, while still providing above market returns. In fact, Moody’s published a report on July 26, 2012 stating that “of the 719 broadly syndicated CLOs it has rated [since 1996], only 14 have suffered principal losses at maturity.” Furthermore, “all 14 deals closed between 1997 and 2001, and these deals had a large exposure to high yield bonds [not leveraged loans] which is said to have contributed significantly to their losses.” Therefore, it appears the market might be right, and regulators are not very good at understanding ultimate risk. I don’t see the downside.