The researchers were quite thorough in detailing the creation of these notoriously opaque, private vehicles. They also tackled the valuation of these deals in an effort to estimate foreseeable losses on the tranches issued. This is no mean feat, but rather an academic exercise which helps readers and researchers better appreciate the limitations suffered by outside parties, like regulators, who are trying to get a handle on this market. For example, the researchers were limited to “[working] with information trustees make public to potential investors (or researchers like us).”
The authors nevertheless honed in on several key aspects of the market, some of which haven’t been adequately addressed in prior publications.
In terms of the creation of these instruments, they note the multidimensional roles being played by market participants who constructed these deals. They remark that several “dealers also serve as collateral managers and consult with banks on valuations” and they comment on the curious role of rating agencies whose “primary motive is to generate business.”
The researchers were also alive to the fact that “early TruPS CDO investors were relying largely on rating agency ratings and surveillance from the dealers responsible for issuing TruPS CDOs.” (Oddly, they too fall back on Moody’s data as the sole point of comparison for validating their own model – seemingly indicative of the situation many are and were faced with, in which one is forced to rely heavily on the rating agencies given their heightened access to data, and the presumed advantage rendered by this informational asymmetry. Unfortunately, given the lack of predictive content of ratings in this realm, it is perhaps difficult to find comfort in the fact that their model's outputs are similar to those produced by Moody’s.)
The researchers were also critical of the rating agencies’ assumptions and methodological changes, especially on the correlation side. They point out that “the model used to justify the zero inter-regional correlation assumption, apparently critical to the development of the single industry TruPS CDO market, was based on a model developed for an unrelated class of securities.”
They also track the increasing correlation between underlying banks over time, and admit their concern that despite the realization and disclosure of the increased concentrations, “rating agencies made few, if any, adjustments for this fact nor did we find evidence that issuers or other analysts expressed any concerns until after the TruPS CDO market came undone.” We would have liked to have seen them analyze, more critically, the validity of changes made in recovery rate assumptions over time.
Analytically, their model puts forth a number of interesting data points, not the least of which is a deferral-to-default cure rate of 2.3%. We believe this is lower than the current market rate, but it certainly runs counter to some of the punitive assumptions being applied elsewhere when analyzing these CDOs, where deferrals are often assumed to always default, with no recovery. (See The Tripping Point for more on this.)
Importantly, the lack of accessibility to certain information hinders the quality of their projections – something the researchers appreciated and candidly disclosed. They recognize that “unfortunately, we do not have information to analyze the risk profile of small banks issuing TruPS into TruPS CDOs versus those that did not. A more thorough analysis of risks at these small banks will have to wait until more information is disclosed.”
They were also cognizant of the inherent difficulties on the data side, given the “[limited] historical performance for TruPS, particularly in a stress environment, [making] forecasting future [deferrals and defaults] more an art than a science.”
Their lack of access to the underlying names leaves them at a significant disadvantage to most investors and players in this market, who have direct access to these names. Unfortunately, they also comment that they were unable to see through to the pool level assets, leaving them unable to distinguish between deferring and defaulted assets. Their inability to look through to the asset level means they must treat all pools identically, based on their overall opinion of the future of banks. This approach leaves them open to significantly underestimating or overestimating the differences between the banks included in different pools. (The FDIC’s Supervisory Insights on winter 2010 was particularly forthcoming on the reasons why banks included in TruPS significantly underperformed other banks in this crisis. We graphed this dynamic in Adverse Selection? No Problem!)
Having advocated heavily for a Central Pricing Solution for TruPS CDOs, we warmed particularly to one part of their important conclusion, which proposed that:
"Banks should also all be disclosing their securities holdings in their investment portfolios to regulators each quarter. For these, bank regulators should follow the model adopted by the National Association of Insurance Commissioners (NAIC), which receives from members CUSIPs and other information on investment portfolios so that regulators can do a full evaluation of all holdings in insurers’ investment portfolios. Applying models like the one we developed to all banks’ TruPS CDO holdings would offer a consistent, independent assessment to compare with banks’ internal analyses. Exactly this type of exercise was conducted as part of the 2008 Supervisory Capital Assessment Program (SCAP), commonly referred to as the “stress tests,” and the 2011 Comprehensive Capital Analysis and Review (CCAR) exercise for the largest banks. With a simple NAIC-style schedule, this type of analysis could be extended to smaller banks’ investment portfolios, with enormous gains in information and the quality and consistency of regulatory supervision."