A number of respected people have blamed accounting rules for much of the current financial crisis. “Fair value” or “mark to market” accounting aims to present a more accurate picture of a bank’s condition and should not be abandoned. The application of fair value accounting, however, especially to assets like Mortgage Backed Securities (MBSs) that do not trade or trade in thin, distressed markets, is flawed and should be improved. The SEC’s recently revised guidance on valuing such assets is a potentially positive step in this direction. In addition, accounting rules should not be perverted to achieve other laudable objectives, such as moderating the pro-cyclical increases and decreases in bank capital that result from the pro-cyclical variations in the value of bank assets.
Accounting rules can be complicated but their purpose is simple, which is to provide as accurate a picture as possible of the financial performance and condition of an enterprise. Business and other decisions depend on such information. In the case of banks, an assessment of its soundness is critical to uninsured depositors and investors before entrusting their funds to the bank. A bank is insolvent if it does not have positive net worth (capital), the difference between the value of its assets and its deposit and other liabilities. One of the most important lessons of banking supervision of the last half century is that it is very unwise to allow an insolvent (but liquid) bank, i.e. a bank with negative capital, to continue to operate. Its losses almost always grow larger until it is finally closed.
Market participants will do their best to evaluate the soundness of banks before buying their debt or placing large (i.e. uninsured) deposits there. If they can rely on accounting statements to reflect soundness to the extent that it is possible to do so, they will do so. If accounting statements cannot be trusted or disguise the truth, investors will not rely on them, estimating the bank’s condition as best they can. But not having the best possible statement of condition increases market uncertainty and the associated risk premiums about what the true condition really is.
It has long been recognized that book values of loans or other assets do not generally reflect the actual market value of such assets. If a borrower is not making its payments on a loan (mortgage or otherwise) it is obvious that its book value (the contractual principle and interest payments first entered into the bank’s books) to the bank overstates and potentially greatly overstates its actual realizable value. It over states the resources the bank has with which to honor its deposit and other liabilities. The best possible statement of the loan’s value would be to record the present value of the expected income to be received from it (discounting the expected income with the prevailing market interest rate). This requires judgment and can’t be know perfectly, but it is not too difficult to arrive at a far more likely value than from using the loan’s original book value.
This is quite obvious and uncontroversial for non performing loans but for some reason the same truth is not so easily recognize for assets that will be fully repaid but with interest rates that are no longer attractive. If a bank buys a $1,000 treasury bond with a 6% dividend per year for thirty years and market interest rates go up to 12%, its real (market) value is reduced significantly even though its interest and principal will all be paid on time and in full. If the bank needed to liquidate the bond now it could only sell it for a bit more than $500. Holding it to maturity just doesn’t change the fact that it has lost value. Reporting it at face value (its purchase price) would be a fraud, a deliberate misstatement of the facts. The S&L crisis of the 1980s reflected exactly such a phenomenon. Savings and Loan banks lost money and became insolvent not because borrowers were defaulting on their mortgages more than usual (as they are now) but because market interest rates increased dramatically and these banks were stuck with otherwise good long term mortgages that yielded much less than the banks now had to pay to their depositors to keep their money there—the money that financed these mortgages. These good mortgages were producing huge loses that bankrupted almost 2,000 banks.
Rules for classifying and provisioning against non performing loans kept on a bank’s books are reasonably well defined. However, when loans are securitized and resold in the market, different rules apply. Basically, the prices received or prevailing in the market for similar loans or pools of loans provide the market’s assessment of current value and are generally expected to be used to value similar, potentially marketable loans still held by the bank.
The movement to mark to market accounting is an attempt to more correctly account for a bank’s performance and condition (net worth) by valuing its assets at their current market price. As such it is an important improvement over earlier book value accounting. Its implementation, however, is not without problems. Take mortgages and Mortgage Backed Securities (MBSs), for example. Mortgage defaults this year and last were much higher than had been expected, especially for Adjustable Rate Mortgages (ARMs) to Subprime and Alt-A borrowers. Banks and other owners of these MBSs have experienced unexpected losses and this should be reflected in lower valuations for these assets in their books. Existing loan classification rules for non traded loans (including mortgages) would require provisioning against expected losses on these loans (writing down the expected value) in light of recent experience even if they are not traded. Banks need to have the best estimate of the likely value of their loans (even if they are fully performing up till now). But for traded mortgages the secondary markets in which these assets trade are very thin and currently non existent (frozen) and thus market prices in this case might not be a good measure of the expected return from holding them. Furthermore, unlike trading government securities or corporate bonds, which are homogeneous within their class so that the market price of a 10 year treasury bond clearly should apply to an identical bond held by the bank, MBSs are heterogeneously and not fully comparable.
Criticisms of fair value or mark to market accounting fall into two main classes. The first is that the actual application of fair value accounting in some cases does not actually result in the best valuation of the asset. The second, which is misguided, is that even if it results in the best measure of actual value it should not be used because it contributes to undesirable pro-cyclical swings in bank capital. These are examined in turn.
Accounting rules formally define “fair value” as “the price that would be received to sell an asset… in an orderly transaction between market participants at the measurement date.”1 Peter J. Wallison, Chief Legal Council of the U.S. Treasury during the Reagan administration, criticized the rules for applying mark to market requirements to MBSs in a study for the American Enterprise Institute. “It seems an unavoidable conclusion, however, that the doubts about the financial stability of these institutions were sown by the drastic cuts in asset prices required by the mark-to-market valuations of fair value accounting, instead of a fair appraisal of the value of the cash flows their assets were producing,” he wrote.2 Wallison pointed to many legitimate problems with valuing MBSs when distressed sales and a lack of market liquidity are believed by many, including the U.S. Treasury, to have resulted in the undervaluation of these assets in the market. However, he persistently refers to the present value of “cash flow” as a preferred basis, when it is the “expected cash flow” that is relevant. On September 30 when the Securities and Exchange Commission, in conjunction with the Financial Accounting Standards Board, issued guidelines under “fair value” accounting rules for financial firms trying to value hard-to-trade assets on their balance sheets, they correctly referred to “expected cash flow” as a proper bases for valuing assets that are not trading or are trading in disorderly markets.3 The fact that the current (historical) cash flow from a mortgage pool is what it is does not change the fact that it is now generally expected to be lower in the future and thus the best estimate of the true value of the pool will be lower because of that.
Earlier SEC FASB interpretations of the rules for valuing MBSs leaned much more heavily toward current market prices as long as there were any at all, while the new clarification admits that current market conditions are not orderly and thus internal expected cash flow estimates may be appropriate. “Fannie has an underwriting and valuation shop with models for valuing mortgages that are up and running.”4 They forecast default rates for different assumptions about price declines and have a good track record. They might be deployed to establish valuations in lieu of reliable market prices until markets return to normal. This new SEC clarification is welcomed and should increase bank capital in much the way the Treasury’s new $700 billion TARP program was expected to.5 It should also do much to diminish the misguided call to abandon mark to market accounting.
The other criticism, also made by Wallison, is that “Procyclicality is obviously an unintended consequence of fair value accounting, but nonetheless an issue for policymakers…. The central purposes of fair value accounting were good—to make financial statements easier to compare and to bring asset values more in line with reality—but these goals, even if they had been achieved, are not as important as avoiding or reducing asset bubbles, producing steady growth in the economy, and encouraging stability in our financial institutions.”6
The pro-cyclical behavior of asset values is an economic reality. No good policy purpose would be served by attempting to hide the fact by corrupting accounting standards. Full transparency is desirable. However, an honest accounting of the pro-cyclical behavior of asset values does not prevent taking policy measures designed to moderate the effect of asset value swings on bank lending or other behavior. A far better approach to the tendency for raising asset values to encourage banks to expand lending or risk taking pro-cyclically would be to vary capital requirements over the business cycle such that banks are required to hold more capital relative to their liabilities during the up side and less during the down side of business cycles.
This issue is reminiscent of the slow evolution of central banks toward acceptance of International Accounting Standards (IAS) for themselves. Traditionally central banks rejected the use of IAS they required commercial banks to follow in order to hid their unrealized foreign exchange valuation gains and losses (“paper” gains and losses), which can be considerable for a typical central bank. The banks’ accounts are in their local currency and they necessarily have an exposure (open position) to foreign currency values through their reserves of foreign currencies. European central banks resisted the adoption of IAS with regard to the reporting of these gains and losses because their laws required them to surrender profits above some minimum to their Finance Ministries and during the first two or three decades after World War II they generally enjoyed valuation gains from their foreign exchange reserves because of the tendency for European currencies to depreciate against the U.S. dollar in which most of their reserves were held. They did not wish to report these unrealized gains because they did not wish to pay them to their Finance Ministries. Thus they hid them by not including them in income. However, in the 1980s the German mark and some other European currencies appreciated against the dollar for a number of years causing the famous Bundesbank to become insolvent. The Bundesbank was recapitalized by the German Government and there after led the field by amending its accounting standards to reflect unrealized as well as realized valuation gains and losses.
Drawing on the precedent set by the Bundesbank, I convinced the Bosnian authorities in 1997 (and the IMF’s legal advisor) to adopted IAS standards for reporting the Central Bank of Bosnia and Herzegovina’s income. As we all agreed that it would not be desirable as a matter of policy to remit to the government unrealized valuation gains (what some might call purely paper profits from changes in exchange rates when no transactions occurred), the rules for determining income subject to remittance were adjusted to exclude unrealized gains. This is far more transparent than the traditional central bank practice of hiding them from income in their financial statements.
While efforts to improve the implementation of fair market accounting are welcomed and should continue, the gains made in making financial statements a more accurate reflection of outcomes and conditions are helpful and important to the efficient functioning of markets.
1 Financial Accounting Standards Board, “Statement of Financial Accounting Standards No. 157.”Fair Value Measurements, September 2006”
2 Peter J. Wallison,”Fair Value Accounting A Critique”, AEI, July 2008
3 SEC Office of the Chief Accountant and FASB Staff “Clarifications on Fair Value Accounting” September 30, 2008
4 Susan E. Woodward, “Rescued by Fannie Mae” The Washington Post, October 14, 2008, Page A17
5 See Coats, “The Big Bailout—What Next?,” Cato Institute, Ocotber 3, 2008
6 Wallison, “op cit”