A key reason that I was insufficiently worried in 2005 about bad mortgage loans being made at the time was that the people who funded the loans– most importantly, the packagers and buyers of private-label mortgage-backed securities– had more motivation and resources to evaluate the risks accurately than I did. That they made an incredibly costly mistake is now indisputable. But the question remains, why?
One interpretation emphasizes misaligned incentives. Money managers earned bonuses while others were left holding the bag. Institutions like Fannie, Freddie, and AIG profited handsomely during the run-up, but the government picked up the tab when things went bad. These kinds of explanations come very naturally to most economists, whose models are usually built on the assumption that economic decision-makers are responding in a rational way to the incentives they face.
But an alternative view is that the key players were simply mistaken as a group, lulled into a misunderstanding of what was going on through social and institutional feedback that sustained a misguided groupthink. This view is hard for many economists to embrace, though there is a good case to be made that this is an important part of the story of what we just went through.
A new research paper by Ing-Haw Cheng, Sahil Raina, and Wei Xiong has come up with an ingenious way to test which of those two explanations best describes what really happened. Quoting from the paper:
We sample a group of securitization investors and issuers from a publicly available list of conference attendees of the 2006 American Securitization Forum, the largest industry conference. These investors and issuers, whom we refer to collectively as securitization agents, comprise vice presidents, senior vice presidents, managing directors, and other non-executives who work at major investment houses and boutique firms. Using the Lexis-Nexis Public Records database, which aggregates information available from public records, such as deed transfers, property tax assessment records, and other public address records, we are able to collect the personal home transaction history of these securitization agents….
Our analysis shows little evidence of securitization agents’ awareness of a housing bubble and impending crash in their own home transactions. Securitization agents neither managed to time the market nor exhibited cautiousness in their home transactions. They increased, rather than decreased, their housing exposure during the boom period through second home purchases and swaps into more expensive homes. This difference is not explained by differences in financing terms such as interest rates, or refinancing activity, and is more pronounced in the relatively bubblier Southern California region compared to the New York metro region. Our securitization agents’ overall home portfolio performance was significantly worse than that of control groups. Agents working on the sell-side and for firms which had poor stock price performance through the crisis did particularly poorly themselves.
Our analysis presents evidence that is broadly inconsistent with systematic awareness of broad-based problems in housing among mid-level managers in securitized finance based on a revealed beliefs approach. However, a home purchase provides a consumption stream that may not be easily found in the rental market, and thus may reflect a consumption motive in addition to beliefs about the future path of asset prices. Despite this, our analysis can be interpreted as testing for whether agents believed income shocks from their jobs in mortgage securitization were permanent. In particular, it is difficult to rationalize why securitization agents endowed with income risk tied to housing would purchase additional second homes and swap into larger homes in 2005 if they simultaneously anticipated an imminent broad-based collapse in housing markets. We also find little evidence that securitization agents were conservative in the value-to-income ratios of their purchases, and that homes purchased in 2004-2006 were among those most aggressively sold in 2007-2009, relative to both control groups. This suggests that securitization agents overestimated the persistence of their incomes and that any consumption stream in these houses was short-lived.
Another way I sometimes pose this question is as follows. Suppose we gave an individual securitization agent perfect foresight of what was to come, that is, exact knowledge of the current and future path of their personal bonuses, stock options, and career path. If they had this information, would they have made the same decisions as they actually made in 2005-2006? If so, that would be confirmation that the basic problem was one of misaligned incentives.
But the data uncovered by Cheng, Raina, and Xiong suggest that, at least as far as their personal home purchases are concerned, the answer would be no.