Fair Value Accounting for Financial Instruments

The following post comes to us from Elizabeth Blankespoor of the Graduate School of Business at Stanford University; Thomas Linsmeier of the Financial Accounting Standards Board; Kathy Petroni, Professor of Accounting at Michigan State University; and Catherine Shakespeare of the Ross School of Business at the University of Michigan.

In our paper, Fair Value Accounting for Financial Instruments: Does it improve the Association between Bank Leverage and Credit Risk?, which was recently made publicly available on SSRN, we contribute to the debate on whether financial instruments should be measured at fair value in financial statements. Accounting standard setters have been deliberating the role of fair values for financial instruments for decades. A fair value is the price at which two willing parties would exchange an asset or settle a liability. Starting after the savings and loan crisis in the late 1980s, the Financial Accounting Standards Board (FASB) has increased the extent to which financial instruments are recognized at fair value (see Godwin, Petroni, and Wahlen 1998). In 2010, the FASB proposed to require that all financial instruments be recognized at fair value, with limited exceptions for receivables and payables and some companies’ own debt (FASB 2010). The proposal was controversial, with over 2,800 comment letters submitted, the vast majority of which objected to the fair value measurement of loans, deposits, and financial liabilities. The FASB is redeliberating this project and has tentatively decided that all financial instruments should be measured at fair value except certain debt financial assets and most financial liabilities (including deposits), which would be measured at amortized cost (FASB 2011).

To empirically provide insight on the controversy, we assess whether a fair value leverage ratio can explain measures of a bank’s credit risk better than a leverage ratio based on a mixture of fair values and historical costs consistent with the mixed-attribute model of US Generally Accepted Accounting Principles (GAAP) and a leverage ratio based on even fewer fair values than GAAP, which is consistent with regulatory Tier 1 capital. We focus on balance sheet leverage because it is very commonly used for assessing firm risk. We define a bank’s credit risk as the risk that the bank defaults on its obligations, and we focus on credit risk because understanding a bank’s credit risk is essential to understanding its financial condition.

We rely on the commonly used tangible common equity approach to measure leverage. We then vary how we measure the financial instruments by including financial instruments measured at 1) fair value; 2) GAAP; and 3) Tier 1 capital. The credit risk measures we consider are bank bond yield spread and bank failure because prior research has shown that the yield spread widens as the credit risk of the security increases and bank failure is the ultimate consequence of high credit risk.

For bank bond yield spreads, our tests on 80,393 observations for 46 banks with 1,861 bonds issued from 1998through 2010 and traded from 2002 through 2010 demonstrate that fair value leverage explains bank-specific bond yield spreads significantly better than GAAP and Tier 1 leverage in both univariate and multivariate analyses. The increase in explanatory power over GAAP and Tier 1 leverage is remarkable. For example, the univariate correlation between bond yield spreads and leverage is at least 25% greater using fair value leverage instead of using GAAP or Tier 1leverage. Further analysis demonstrates that the additional explanatory power of fair value derives primarily from fair valuing loans and deposits.

Our tests of bank failure on approximately 7,000 observations for 1,067 banks from 1997 through 2009, including 53 failed banks, demonstrate that in the year just prior to failure, our three leverage measures have similar predictive power. But both two years and three years before failure, fair value leverage dominates the two other leverage measures in predicting failure both univariately and in a multivariate analysis with controls for other financial statement variables known to be associated with bank failure. This demonstrates that leverage based on fair values provides the earliest signal of financial trouble. Overall, our results suggest that recognizing financial instruments at fair value, especially loans and deposits, results in a balance sheet that reflects the average bank’s credit risk better than a more historical-cost-based balance sheet.

Overall, our study provides evidence to standard setters that leverage ratios based on fair values of all financial instruments describe a bank’s credit risk better than GAAP or Tier 1 capital leverage ratios. This evidence suggests that banks’ financial statements with financial instruments measured at fair values, including loans, deposits, debt, and held-to-maturity securities, are more descriptive of the credit risk inherent in the business model of banks than the current GAAP financial statements.

The full paper is available for download here.

Both comments and trackbacks are currently closed.

One Comment

  1. Alan Collinge
    Posted Saturday, June 22, 2013 at 5:45 am | Permalink

    I have a question I hope you will answer:

    Currently, a push is underway by some to use fair value accounting for federal student loan instruments.

    One would assume that the first and most obvious way to arrive at a fair market value for these loans would be to look at historical data for the sale of student loan securities…a fairly robust, and well understood market. Given proper accounting for the unique characteristics of a given bundle based on historical data, I would think this would be the most straightforward and acceptable way to arrive at a fair value estimate.

    This is not what is being proposed, however. What is being proposed for student loans is to look, instead, at the private student loan market, and use the difference in interest rate (private loans having the higher rate) to calculate a loss for the federal loans (ie this is interest not made by the federal loans).

    Isn’t this completely different from what fair value accounting is meant to do? This method doesn’t determine a fair value for the lending instrument, it only determines an opportunity cost, at best. Do you disagree?

    Assuming I am wrong, and that this somehow does map into some exotic valuation model that I am not understanding, there still is a fundamental problem:

    The federal loan market is the driver of the private loan market. Students who take out private loans often do so because they could not get federal loans for one reason or another. The banks know this, and accordingly charge a higher interest rate.

    There are other obvious dependencies between the two lending markets that cry out, but generally, this looks like Exxon pointing to the guy selling biodiesel out of his garage for $8/gallon, and using this as justification for writing off $4 (or more) for every gallon of gas they sell!!

    I do appreciate your consideration and response to this question. There is a significant public interest, here.

    Regards,

    Alan Collinge