Wednesday, February 20, 2013

A New Look at Truth in Numbers

New research finds support for valuing bank securities at current market value.

Research shows that fair-value calculations provide a far more accurate indicator of a bank's risk of failure than calculations based largely on historical cost.

It has been four years since Wall Street and the banking industry imploded under the weight of mortgage-backed securities, but the war over "fair value" accounting is still underway.

Put simply: Is the economic value of a security best reflected by what you paid for it ("adjusted historical cost") or by what you could sell it for today ("fair value")?

That question has been a subject of almost theological debate among accountants for decades. But the financial crisis of 2008 and 2009 elevated it to a pitched political battle between the banking industry and financial reformers.

Now comes a new empirical study, whose coauthors include a faculty member at Stanford's Graduate School of Business, which may add gasoline to the fire.

First, Some Background

During the financial crisis, when debt markets became frozen by panic, banks were caught with billions of dollars in securities that could only be sold for pennies on the dollar. Bank critics and many investors argued that many banks were effectively insolvent, and were hiding their problems by refusing to write down their holdings to "fair" or current-market value.

The banking industry fought back ferociously, arguing that current market values were irrelevant because banks hold many of their securities all the way to maturity. As long as a security was still paying as it should, they argued, the only thing that really mattered was its historical cost.

In 2010, the Financial Accounting Standards Board (FASB), which sets accounting rules for the United States, jumped into the fray. At the time (and still), U.S. Generally Accepted Accounting Principles (GAAP) required some financial instruments to be carried at fair value but allowed many others to be recorded at their adjusted historical cost.

The FASB's 2010 proposal was for institutions to value almost all of their financial instruments at fair value. Banks and their allies howled in protest. The FASB received 2,800 comments, most of them bitterly opposed to its proposal. Sympathetic members of Congress warned that the FASB would cripple the banking system and the economy. The FASB retreated somewhat, but it is still working on its proposals, and the fights still simmer.

Which Approach Comes Closest to the Truth?

The new study tries to answer that by analyzing the predictive power of fair-value calculations that banks are required to include as footnotes to their official financial statements.

The study was coauthored by Elizabeth Blankespoor, assistant professor of accounting at Stanford; Tom Linsmeier, a member of the FASB; Kathy Petroni, professor of accounting at Michigan State University; and Catherine Shakespeare, associate professor of accounting at the University of Michigan.

Their firm conclusion: Contrary to claims by the banking industry, the fair-value calculations provide a far more accurate indicator of a bank's risk of failure than the bank's official calculations based largely on historical cost.

The researchers began by examining how well a bank's fair-value calculation correlated with market assessments of a bank's risk, as measured by the yield spread on its bonds.

For the financial instruments not carried at fair value in the financial statements, these fair-value estimates don't currently affect a bank's income statement or balance sheet. They only tell investors what the bank might look like if it had used fair-value measurements for all its financial instruments.

But the researchers found that if the fair-value estimates were used, the bank's leverage based on those numbers correlated much more closely to a bank's bond yield spread — at least 25% more — than its official estimates under U.S. GAAP or under regulatory rules for "Tier One" capital.

Of course, market perceptions of a bank's fragility could be wrong. The researchers also looked, therefore, at how well the fair-value calculations correlated with actual bank failures. The team looked at 1,067 banks and 53 bank failures from 1997 through 2009. Once again, the fair-value estimates reflected reality far more accurately than the calculations under GAAP or for regulatory capital.

"Both two and three years prior to failure, fair-value leverage dominates the other two leverage measures in predicting failure," they wrote. "This demonstrates that leverage based on fair value provides the earliest signal of financial trouble."

Fair-Value Accounting May Have Drawbacks

Blankespoor cautions that the results don't settle the question of which system is best. Given that the markets already make their own assessments of bank risk, regardless of what a bank reports under GAAP, it's not clear whether fair-value accounting would improve market assessments.

There may also be practical policy concerns. Forcing banks to write down their holdings to current market value could set off a fire sale of assets, which could turn a modest market downturn into a full-fledged crisis.