Excessive asset or liability concentrations have historically played a large role in bank failures. A recent proposal to revise the CAMELS ratings for banks appears to define asset concentrations more narrowly. The change is subtle and its motivation unclear. But it appears to rest on the underlying premise that concentrations of supposedly low risk assets are necessarily benign. But is that really the case? Auction rate preferred securities, taxicab medallion loans, and non-agency MBS all provide examples of assets that were perfectly safe … until they weren’t.
While some leading regulators like to portray themselves as champions of transparency, the CAMELS proposal is not especially transparent. It merely presents a proposed new approach with little in the way of comparisons with current policy or justification for proposed changes. Fortunately, Davis Polk has prepared a side-by-side comparison of the current and proposed new approaches. One little-noticed change relates to the consideration of asset concentrations when assessing overall asset quality. The current guidance includes “the existence of asset concentrations.” The proposed new guidance instead uses “financial risk arising from asset concentrations” (emphasis added). Regulators did not provide a rationale for this change, but it seems to suggest that some concentrations are benign.
It makes sense to focus on concentrations of historically risky assets than on concentrations of historically safe assets. The OCC’s examiner handbook on Concentrations of Credit notes that “not all concentrations pose the same level of risk,” with concentrations of highly cyclical assets presenting a particular concern. However, it also notes that “experience indicates that although a concentration has not proven problematic in the past, that does not preclude it from becoming a problem in the future.” Three specific examples come to mind: auction rate preferred stock, taxi medallion loans, and non-agency MBS.
Countries Don’t Go Broke – Or Do They?
Before delving into these examples, it’s useful to step back a little further to the 1980s and the Latin American Debt Crisis. Walter Wriston, Citibank’s CEO, brushed off the bank’s exposure to sovereign debt by declaring countries “never go bankrupt.”[1] In fact, countries do default on their sovereign debt, and it took an extraordinary intervention by the U.S. Treasury and the Internation Monetary Fund to stabilize the market. The “countries never go broke” idea reflected more wishful thinking than history. Sovereign defaults go back to Spain under Philip II, which defaulted four times in the 1500s. The cases described below presented better historical reasons for optimism but still proved disastrously wrong.
Auction Rate Preferred Securities (ARS)
ARS are a type of floating rate security with long (or even indefinite) contractual maturities, but where the rate resets based on a Dutch Auction. Under such a process, the rate is based on the lowest rate that clears the market. These auctions would occur at seven-, 14-, 28-, or 35-day intervals. ARS provided a relatively cheap form of financing for issuers while offering investors a safe and liquid asset. Emphasis on the past tense. Between 1984 and 2007, only 13 of more than 100,000 auctions (0.01%) failed.
That all changed in 2008. Many ARS issuers were municipalities, with specialized insurance companies back their debt. Unfortunately, these monoline insurers decided to diversify by also insuring securities backed by subprime mortgages, with predictable results. By February 2008, more than 80% of these auctions failed. This cheap source of financing was no longer available, and these liquid assets were no longer liquid. An article by Marc Ross for the CFA Institute provides an excellent overview of this “forgotten crisis.”
Taxicab Medallion Loans
Taxi medallion lending once looked like an attractive market niche. Taxi medallions are transferable permits that allow companies or individuals to operate a taxi. Municipalities strictly controlled supply, which caused values to increase significantly. A borrower could use medallions as collateral to obtain a loan. The rising price of medallions drew speculators, some of whom had no intention of operating a taxi. Some lenders focused on collateral values while ignoring more established underwriting considerations like cash flow analysis, debt service coverage, and secondary sources of repayment. When rideshare companies like Uber and Lyft disrupted the market, prices came crashing down, as shown below.
It wasn’t just poor underwriting, but also concentrations that caused trouble. A Material Loss Review by the Inspector General of the National Credit Union Administration (NCUA) focused on Melrose Credit Union, LOMTO Federal Credit Union, and Bay Ridge Federal Credit Union. These three CUs had combined total assets of $1.44 billion, but their failure cost the NCUSIF $765 million. As of December 2012, taxi medallion loans at Melrose, LOMTO, and Bay Ridge represented 81%, 97%, and 50% of total loans, respectively. Historically low delinquency rates also rose sharply. At Melrose, delinquent taxi medallion loans shot up from 0.22% in December 2014 to 28.64% just two years later.
Non-Agency MBS
No case illustrates the pitfalls of supposedly benign concentrations than that of non-agency mortgage-backed securities (MBS). Most think of mortgage securities in terms of issuances by Fannie Mae and Freddie Mac. But non-agency securities briefly dominated the market during the 2000s, accounting for more than 40% of mortgage originations in 2006.
Most of these securities were structured a CMOS. A CMO takes a pool of mortgages and rearranges the cash flows into separate classes, called tranches. For non-agency MBS, the credit support comes largely in the form of the tranche structure itself. Junior tranches absorb losses on the underlying collateral, protecting the more senior tranches. Senior tranches and even some mezzanine tranches typically carried AAA credit ratings. They also carried capital risk weights of only 20%, allowing these positions to be levered up to fifty-to-one.
AAA debt securities usually have virtually no credit risk. Between 1920 and 2008, Moody’s AAA corporate bonds had a cumulative one-year default rate of zero that only rose to 0.039% after ten years. In principle, a AAA non-agency MBS should be even safer since the thousands of underlying mortgages largely eliminate any idiosyncratic risk. As it turned out, however, they weren’t very safe at all. Roughly 80% of 2006 non-agency RMBS originally rated AAA by Moody’s were downgraded by the end of 2009. That figure rises to nearly 100% for those originated in the second half of 2007. Results were similar for S&P and Fitch.
The collapse of the non-agency MBS market resulted from a bursting housing bubble, an epic breakdown in mortgage underwriting standards, and flawed models that badly underestimated credit losses. Financial institutions (and some regulators) placed too much faith in the AAA ratings while ignoring concentration risk. This included some supposedly sophisticated investors. While the collapse of Fannie and Freddie is largely attributed to their core mortgage guarantee business, non-agency MBS in their investment portfolios also created a enormous problem with cumulative losses in the tens of billions. The nonagency portfolio at the Seattle FHLB went from 100% AAA to more than half junk over a two-year period. The Bank became undercapitalized and merged into the FHLB of Des Moines. A huge financial conglomerate required a state bail-out and largely exited the U.S. market for the same reason.
The Material Loss Review for Guaranty Bank provides an especially clear illustration of the problem. The bank, which had total assets of about $17 billion, held $3 billion in private label CMOs. There were also concentrations within this concentration, with 60% of the underlying mortgages originated in California and more than 80% option ARMs. The AAA rating caused both the bank and its regulators to disregard these concentrations. The report found “no instance where a board member voiced an objection about the thrift’s nonagency MBS investments or raised concerns about the return on assets and equity.” Nor was there “any significant discussion about the risks associated with the concentrations of the nonagency MBSs in California option ARMs.”
Regulators “continued to report in 2007 that risk with Guaranty’s nonagency MBS portfolio was minimal.” As Guaranty built up its nonagency portfolio from 2004 to 2007 the bank continued to receive satisfactory CAMELS ratings. In a July 2007 exam, the bank was rated 2 overall, including “2” ratings for both asset quality and management.
Concluding Thoughts
A recurring theme with these three asset classes is that their low loss history gave both bankers and their regulators a false sense of security. Missed signals occurred even while regulators had broadly defined asset concentrations. Leaving aside the proposed CAMELS change, examiners will likely become even more reluctant to call out these types of risks. Consider the Fed’s new supervisory operating principles. The issuance of an MRA or MRIA would require “significant probability of significant harm to the financial condition of the banking organization.” Supervisory staff will also need to show “sufficient evidence that a particular estimate of probability and severity is plausible.”
We might know today what went wrong with ARS, taxi medallions, and non-agency securities, but need to watch out for hindsight bias. Step back in time to 2007. Now try demonstrating a “significant probability of significant harm” if auctions had historic failure rates of only 0.01%, Uber didn’t exist yet, and a AAA rating really meant something. Good luck with that.
[1] Some sources quote Wriston as saying countries “never go broke.”
Leave a Reply