The Global Financial Crisis and Reinterpreting Lessons from History

Interest in CDO accounting is not mere historical curiosity. Since the near shutdown of the structured finance market after the financial crisis, segments of the market—in particular collateralized loan obligations and commercial mortgage-backed securities (CMBS)—are making a strong comeback as investors search for yield in a low interest rate environment. The International Organization of Securities Commissions noted that the 2013 issuance of CDOs was nearing pre-crisis levels and could be forming another asset bubble. Huw Jones, “Bugbear Securities of 2008 Crisis Make Big Comeback—IOSCO,” Reuters, Oct. 15, 2013. With these types of investments on the rise again, it is important to understand how they link to systemic risk (a key focus of banking regulators and policy makers following the financial crisis) and how a downturn in financial markets and wider macroeconomic conditions might affect investment value and the broader health of the financial sector.

Our recent research on whether banks valued certain types of CDO securities in 2007–2008 consistent with widely followed measures of supposed fair market value sheds light on these issues and provides insight on channels through which systemic risk can become elevated. Our findings run counter to the conventional wisdom that banks used wiggle room in accounting rules to avoid or delay marking assets to market. If banks did incur hundreds of billions of losses by marking CDO positions consistent with market values that may have been driven below fundamental long-term values by panic and the evaporation of liquidity, and these losses in turn drove further turmoil in financial markets, this provides an important lesson for the future. Hence, we believe that properly understanding the past will inform current policy debates, shed light on current areas of litigation risk, and may even help prevent another financial meltdown.

Background

CDOs are sometimes pejoratively referred to as financial sausages. In essence, CDOs are special-purpose entities that pool the cash flows of numerous underlying collateral securities and then divide the cash flows according to certain distribution rules to create new derivative securities, typically called “tranches.” Different tranches can have very different risk characteristics. Our research, described below, focuses on “super senior” tranches that rank above all other “junior” tranches in cash flow priority and are typically the last to suffer losses stemming from the performance of the underlying collateral assets. Prior to the financial crisis, super-senior tranches were considered relatively low risk and typically given the highest AAA/Aaa credit ratings by the major credit rating agencies.

A CDO’s collateral pool could contain a range of assets, including residential mortgage-backed securities, corporate loans, credit default swaps that mimic the cash flows of other securities (sometimes called “synthetic” collateral), and even other CDO tranches. A substantial subset of the CDO market is made up of “ABS” (asset-backed securities) CDOs, also known as “SF” (structured finance) CDOs. ABS CDOs are backed primarily by a class of structured finance securities that includes securitizations of subprime residential mortgages (sometimes called “home equity” or HEL ABS), commercial mortgage-backed securities (CMBS), consumer credit card debt, and auto loans.

Our analysis focuses on ABS CDOs because bank write-downs and losses related to their ABS CDO positions lie at the heart of the recent financial turmoil. Between 1999 and 2007, over $640 billion in ABS CDOs were issued, Larry Cordell et al., “Collateral Damage: Sizing and Assessing the Subprime CDO Crisis” (Federal Reserve Bank of Philadelphia, Working Paper No. 11-30/R, 2012). And at the end of the third quarter of 2007, Citigroup, UBS, and Merrill Lynch reported net exposure to super-senior ABS CDO tranches of $43, $20.2, and $14.2 billion, respectively. As losses on these opaque instruments reverberated through the economy and the financial crisis deepened, U.S. financial firm share prices dropped over 75 percent amid the worst bear market since the Great Depression of the 1930s. Meric et al., “Performance of U.S. Financial Sector and Bank Stocks: October 2002–August 2009,” Int’l Res. J. Fin. & Econ. No. 39 (2010). A wave of private securities litigation and regulatory inquiries ensued, challenging the integrity of the banks’ financial statements. Dickey et al., “Subprime-Related Securities Litigation: Where Do We Go from Here?,” 22 Insights (Apr. 2008). In many cases, plaintiffs alleged that CDO write-downs should have been recorded sooner and that bank financial statements misled investors about the risks of CDOs and other structured finance assets that caused harm to investors by artificially inflating bank stock prices. Some argued that accounting rules allowed banks to use models to mark to “myth” instead of to “market” and led to the problem of bank asset impairments being recognized “too little, too late.”

CDO tranches are complex securities and are challenging to value. CDO valuation was a much-debated topic, even before the financial crisis. According to standard financial principles, the value of an asset can be calculated as the present values of all its expected future cash flows. The expected cash flows of a CDO tranche are often difficult to estimate and depend on the expected cash flows of a portfolio that may include over 100 underlying securities, some of which may in turn also be complex structured finance securities. A CDO tranche valuation must also grapple with, among other factors, an estimation of the co-movement of default (i.e., the credit risk correlation) of the underlying securities that is also known to present difficult challenges from both a data and a modeling perspective.  Martin Scheicher, “How Has CDO Market Pricing Changed During the Turmoil? Evidence from CDS Index Tranches” (European Central Bank, Working Paper No. 910, June 2008).

In the case of frequently traded assets such as stocks, the value placed on the security by market participants is readily observable. However, unlike stocks, CDO tranches are “bespoke” instruments and do not trade on a daily basis. Even before the market came to standstill during the crisis, CDO tranches were generally illiquid and daily prices were typically unavailable. Further, even the CDOs’ underlying collateral securities typically did not trade regularly and were themselves difficult to price. As a result of the general lack of market-based observed inputs and asset complexity, banks frequently classified CDO tranches as “Level 3” assets, choosing to rely on proprietary models to arrive at values for their accounting statements.

Current Debate over the Role of Fair Value Accounting in the Financial Crisis

At the heart of post-crisis bank accounting debates was the use of “mark-to-market” accounting, also known as fair value accounting (FVA). Under FVA, assets are valued at the price the asset would theoretically sell for in the open market. In volatile markets, these values could deviate significantly from the original price paid. Critics argue that mark-to-market losses driven by market panic decreased banks’ regulatory capital, led to a downward spiral of forced asset sales that further depressed prices, and eventually threatened the soundness and functioning of the entire global financial system. Brad A. Badertscher et al., “A Convenient Scapegoat: Fair Value Accounting by Commercial Banks During the Financial Crisis,” 42 CFA Dig. (May 2012). Steve Forbes compared FVA to the “bubonic plague” and claimed it was the “principal reason that the financial disaster of 2007-09 threatened to destroy our financial system.” Steve Forbes, “Stop This Horror Before It Starts Again,” Forbes, June 28, 2010. Defenders of FVA argue that it played, at most, a minor role in a crisis that was the consequence of a complex nexus of factors and policy failures across the private and public spheres. Blaming FVA, they argue, is the equivalent of “shooting the messenger.”

The views of the FVA critics are in tension with another broad group of academics and policy makers who argue that banks exploited accounting rules (either properly or improperly) to effectively avoid or delay marking certain assets to true market prices.  Dushyantkumar Vyas, “The Timeliness of Accounting Write-Downs by U.S. Financial Institutions During the Financial Crisis of 2007–2008,” 49 J. Accounting Res. 823 (June 11, 2011). One widely cited academic paper concluded that overall “there is little evidence that banks’ reported fair values suffer from excessive writedowns or undervaluation in 2008” and that “[i]f anything the evidence points in the opposite direction, suggesting that banks used the discretion in accounting rules to keep asset value high relative to concurrent market prices and expectations.” Christian Laux & Christian Leuz, “Did Fair-Value Accounting Contribute to the Financial Crisis?” (Nat’l Bureau of Econ. Research, Working Paper No. 15515, Nov. 2009). If mark-to-market accounting rules allow sufficient discretion for banks to avoid losses, the logic goes, then how could such “excessive” losses have been responsible for contributing to distressed asset sales, depleting regulatory capital, and exacerbating the financial crisis?

Our Research

Our research attempts to shed light on these debates by examining whether the write-downs that banks took related to their super-senior ABS CDO positions during the financial crisis were timely, given publicly available contemporaneous market-based information. The timeliness of banks’ CDO write-downs has been previously examined in other academic studies.

However, unlike our approach, these prior studies fail to recognize that all ABS CDOs are not created equal and fail to properly account for the differences in quality of banks’ ABS CDO positions. Given differences in underlying collateral (among other factors), the values of super-senior tranches of different ABS CDOs can differ substantially.

We began our analysis by using banks’ public disclosures to construct a novel, hand-collected data set of 30 banks’ quarterly super-senior ABS CDO marks and writed-owns over the eight quarters in 2007–2008. While the precise collateral composition of each CDO in each bank’s portfolio is not public information, we accounted for differences in underlying collateral quality by using information in the public disclosures to divide these data into two categories: high grade and mezzanine ABS CDOs, a common distinction in the CDO industry. “High-grade ABS CDO” is a term used to refer to ABS CDOs with higher quality underlying collateral (often with average collateral ABS credit ratings of A or greater), and mezzanine ABS CDOs are those with lower quality collateral (often with average collateral ABS ratings of BBB or lower).  Laurie S. Goodman et al., Subprime Mortgage Credit Derivatives (June 2008).

After collecting write-down data reported in banks’ public disclosures, we next constructed a market value proxy to compare against the reported values. Because, as noted above, the details of the CDOs in banks’ portfolios were not known, we used the widely followed credit derivative indices created by a company called Markit to construct two market proxy values, one for high-grade and one for mezzanine ABS CDOs, based on the average collateral composition and tranche structure of a sample of over 100 ABS CDOs issued in 2006. Despite their widely discussed shortcomings as CDO or ABS valuation tools, this series of indices (e.g., the ABX/TABX/CMBX) was seen during the 2007–2008 period to be the only game in town for daily information about pricing in the ABS markets (particularly for subprime) and was closely followed by both market participants and regulators. Serena Ng et al., “A ‘Subprime’ Gauge, in Many Ways?,” Wall St. J., Dec. 12, 2007.

For each of the eight quarters of 2007–2008 that we analyzed, we calculated the difference between the aggregate CDO position value a particular bank had disclosed for that quarter end and its proxy market value calculated based on the relative notional value of the bank’s high-grade and mezzanine super-senior ABS CDO positions. We found that these differences were statistically insignificant (i.e., not distinguishable from zero with a reasonable margin of error) in all eight quarters of 2007–2008 across the sample of banks we analyzed. Therefore, contrary to prior academic findings and conventional wisdom, in aggregate banks do not appear to have delayed marking to market their super-senior ABS CDO positions during the crisis.

Our findings also address the debate about the role of FVA in the financial crisis. A number of industry and academic observers argue that the ABX index was an overly pessimistic, thinly traded index driven by hedge-fund short bets that became divorced from credit risk and “fundamental” asset values as liquidity dried up. Richard Stanton & Nancy Wallace, “The Bear’s Lair: Index Credit Default Swaps and the Subprime Mortgage Crisis,” 24 Rev. Fin. Stud. 3250 (2011). If so, then billions of bank CDO positions marked down in a manner consistent with the decline observed in these indices could have exacerbated the panic of the financial crisis by creating vicious cycles of downward price spirals and starting a run on the “shadow” banking sector akin to the classic U.S. bank panics of the pre-FDIC era. Many consider this run on the shadow banking sector (which broadly includes banking or bank-like entities outside the traditional regulated banking system such as investment banks, repurchase or “repo” markets, CDOs, and other structured finance entities) to have been the primary cause of the crisis. Our research indicates that banks’ FVA markdowns were in line with these potentially depressed and distorted market indices’ values that may have in turn added fuel to the fire of the panic.

Conclusion

In short, our research suggests that the models banks used to mark complex illiquid assets to market during the financial crisis were consistent with publicly available indicia of such value. This evidence runs counter to the conventional wisdom and should be considered in discussions about any future policy, accounting, or legal changes regarding the reporting of such mark-to-market values. The argument that FVA could not have exacerbated the financial crisis because banks exploited wiggle room in accounting rules to delay or avoid taking write-downs on Level 3 assets is not supported by our research. While some argue that banks should provide more detail about models and assumptions to help investors better understand how values were computed, this position must be balanced with the competitive cost of potentially revealing sensitive proprietary information.