The Securities and Exchange Commission's recent roundtable on ratings shopping in structured finance was a reminder of how Sen. Al Frankens (D-Min.) efforts at reform are obstructed by a wall of denial. Standard & Poor's denied that any proposal for reform would be workable, and the SEC effectively denied that the issue requires action any time soon. Neither was as direct or as blunt as Moody's Investors Service CEO Raymond McDaniel, who has previously denied that any problem ever existed.
"That conflicts of interest led to inflated ratings at Moodys is a concept I categorically reject, he told the Financial Times last January. One way to try and answer this is to make the argument in the negative. Why would it affect only housing and why would it affect only mortgage securities? He pointed to other types of structured securitizations, involving credit card bills and car loans, which performed well during deep recession.
McDaniels question is illuminating, because the answer should be obvious to anyone who works at a rating agency. Heres what he seems to have overlooked:
Why housing finance?
First, no other major form of finance is so reliant on asset appreciation.
By way of contrast, newly purchased automobiles and credit card receivables depreciate rapidly. Commercial real estate is financed on cash flow projections. And with housing, asset bubbles conceal a multitude of sins. U.S. residential housing debt doubled over five-and-a-half years, from $4.9 trillion in January 2001 to $9.7 trillion in June 2006. Much of that increase came from $2.2 trillion in cash out refinancings and home equity loans. Historical data show that real estate is cyclical, that a cycle can last more than a decade and that the vast majority of real estate booms are followed by multi-year periods of price stagnation, or worse. The rating agencies simply failed to factor this cyclicality into their subprime ratings.
Second, fraud went viral under the originate-to-distribute model.
Entire libraries could be filled with government reports and legal filings documenting the pervasive illegal activities at every level of the mortgage distribution chain. Bank of America's $42 billion in legal settlements, so far, merely scratches the surface of an industrywide problem. Financial models cannot accurately measure the impact of widespread fraud.
Why residential mortgage securities?
Originators act differently when they have skin in the game.
The originate-to-distribute model encouraged sloppiness. According to the SEC, credit losses for private label mortgage securities are about four times larger than those suffered by Fannie Mae and Freddie Mac, which buy and hold their credit risk. Securitizations of credit card receivables and auto loans are structured so that the credit card issuers and automakers absorb the first losses in those debt portfolios. Consequently, originators were more careful about the credit they extended.
Additionally, the business model for rating structured finance deals is very different from that for rating corporates and sovereigns.
The structured finance sector is characterized by three obvious differences: First, almost everything in residential mortgage bonds was rated triple-A, so investors effectively suspended disbelief. Had they assumed an outcome 100 times worse than what was shown in Moodys expected loss tables, writeoffs would have been nominal.
Second, the ratings agencies' customer base is a fairly small number of banks and originators, who tend to act in lockstep, so the downside of displeasing a client can be pretty severe. Third, every structured finance deal is a one-time-only business opportunity. Either you get hired by the banks and you bring in revenues, or you bring in nothing. And if your credit enhancement standards are more conservative than two other rating agencies, the banks will likely shut you out of the business. That's a powerful incentive to bend to the will of "the market," which often acts like a cartel.
Finally, most of the triple-B tranches of residential mortgage-backed securities, which were deeply subordinated to about 20 times leverage and were infected with fraud, were bundled into collateralized debt obligations, with capital structures that were mostly rated triple-A. These CDOs had no valid analytic foundation. Former Moodys executive Gary Witt explained how Moodys arrived at estimates for default correlations. We made them up, he said. There wasn't anything to go on, there wasn't any history.
In fact, the collapse of the subprime market of the late 1990s showed what happened to originators that held on to residual interests, which were deeply subordinated tranches of different mortgage pools originated within a few months of one another. Most of those loan originators went out of business.
I dont know of a tool that would have improved our assumptions on housing, McDaniel told the Financial Crisis Inquiry Commission. His all-encompassing denial is perhaps the most eloquent argument for increased regulatory oversight.
David Fiderer has previously worked in energy banking for more than 20 years. He is currently working on a book about the rating agencies.