Arthur F. Burns Fellow
Peter J. Wallison
It has been almost two years since the financial crisis began, and debate about fair value accounting has only intensified. The banks and others contend that fair value accounting is responsible for their apparent weakness and instability, while accountants and investor advocates argue that the truth--the facts about the banks' assets--is what's ultimately causing their problems.
It seems to me that there are two fundamental questions that should be addressed in this debate, and neither has received sufficient attention: First, is fair value accounting, as it is currently structured, the appropriate way to present the financial reports of depository institutions such as commercial banks? I will argue that it is not, and therefore will also seek to answer the second question: How can the fair value accounting system be maintained where it still has value, while reducing the adverse effects that arise from unusual market movements?
It is impossible to do justice to something as complicated as fair value accounting in a few paragraphs, but I'll try to outline briefly the principal elements of the system that have caused most of the fuss. The foundational ideas of fair value accounting were adopted in 1993 by the U.S. Financial Accounting Standards Board (FASB) in Statement of Financial Accounting Standards 115 (FAS 115). The rule is applicable only to the valuation of securities, including mortgage-backed securities and other securities backed by assets. It covers debt securities of all kinds, but not whole loans.
FAS 115 divides financial assets into three categories--those held "to maturity," those held "for trading purposes," and those "available for sale." Each of these categories is treated slightly differently. Assets held to maturity are valued at amortized cost; assets held for trading are marked to market, with unrealized gains or losses included in earnings; and assets deemed available for sale are marked to market, with unrealized gains or losses excluded from earnings but included in shareholders' equity. This treatment allows unrealized gains or losses to affect the capital of banks. It is very difficult to categorize assets as "held to maturity" because they are subject in that case to severe restrictions on sale. As a result, most financial institutions, including banks, hold these assets either in "available for sale" or "trading" categories.
When assets are held in either of these two categories, they must be marked to market--if there is a functioning, liquid market. If there is no active market, then a variety of other methods can be used for valuation. Whether an active market is present or absent is a question of judgment, and accountants will say that any market price--even one derived from a distress or liquidation sale--is one of the elements that must be used in determining asset values. Banks and other financial intermediaries argue that accountants give too much weight to these distress prices, while accountants say they are only following the FASB rules. In a report at the end of December, the SEC staff largely sided with the accountants, concluding that fair value accounting had not caused the financial problems of the banks and others, and rejecting the idea of a wholesale revision of fair value accounting.
If you think financial accounting is simply a way of recording the results of business operations, think again. Financial accounting is a highly conceptual art in which many objectives and perspectives compete for priority. In this discussion, I will distinguish between the earnings perspective and the stability perspective. If we are interested in emphasizing earnings we would choose one system of valuing assets--marking assets to market--and if we want to focus on stability, we would choose another: amortized cost. The assets are the same, of course, only the way we look at them is different. It's like the uncertainty principle in physics, which posits that it is not possible to measure both the position and the momentum of a quantum particle at the same time; the properties are there in the particle, but can't be measured simultaneously.
Because accounting cannot give us all perspectives at the same time, we have to make a choice about which perspective we want to see. With the development of the global securities markets and widespread interest in equity investment, the potential of companies to produce earnings has become the fact that is of most interest to the most people, and thus the principal focus of financial accounting. Fair value accounting is a corollary of this significant development. What the market would pay for a company's assets at a given point in time is a better indicator of whether it is adding shareholder value than simply measuring the costs of those assets and the returns they are yielding. As the FASB noted in FAS 115: "Fair value portrays the market's estimate of the present value of the net future cash flows of those securities, discounted to reflect both the current interest rate and the market's estimate of the risk that the cash flows will not occur." (Paragraph 40; emphasis supplied). I will come back later to the question of whether this is actually what the market is doing in putting a value on asset-backed securities.
Since the focus of fair value accounting is on earnings, it is logical to ask whether a firm's earnings potential should be the only way to evaluate financial firms. For some business models, what they can earn may not be as important to know as their financial strength--their potential for stability or instability. Financial institutions of this kind--banks and insurance companies come to mind--might be better described by their financial reports if greater weight were placed on the elements of their makeup that signal stability or instability rather than their earning capacity. This raises the question why fair value accounting was adopted for all financial firms and not just those for which earnings--rather than stability--were important. Or perhaps why firms were not given a choice about how they would value their assets.
I can identify three fundamental goals of accounting that are likely to have influenced the choice of fair value accounting for all financial firms. One of these objectives is to minimize what is called management bias. Management has an obvious incentive to inflate the value of a company's assets, and many ways to do it. Marking a company's assets to market is an effective way to take this element of financial statement manipulation out of management's hands. The tight restrictions on moving assets into or out of the held-to-maturity category is intended to enforce this objective. In addition, there is a strong element in accounting theory that favors treating similar assets in similar ways. Financial intermediaries such as banks, securities firms, finance companies, hedge funds and insurance companies all hold similar assets, and accounting theory would say that insofar as possible these assets should be given the same values irrespective of the kind of financial institution that holds them. Finally, another of financial accounting's goals is comparability--the idea that investors should be able to compare the results of companies that are competing for capital. If comparability is possible, capital will be allocated more efficiently. If two firms hold the same assets, but value them differently, comparability is impaired. Comparability is of particular importance if the underlying goal of accounting today is to provide information about earnings potential. In that case, the differences between how companies earn their returns should be minimized.
These are worthy objectives, but they have costs--mostly not considered and unintended. One consequence that has been covered extensively in the media is the huge loss of asset values in financial firms that are holding asset-backed securities. This has afflicted all financial companies, but has been particularly troubling for banks, which of course have demand deposits and other very short term liabilities. Here is where emphasizing earnings over stability begins to have an effect. While equity investors in banks are justifiably interested in their earnings, depositors, lenders and counterparties are not. They are interested in whether the bank is solvent and likely to be financially stable over the long term. It is difficult to see why a depositor or a credit default swap counterparty would be interested in whether a bank could sell all its assets at a given point in time for a certain value. What the depositor or counterparty wants to know is whether the bank's return on assets is sufficient to allow it to meet its obligations as they fall due under most foreseeable circumstances. The way to know whether a bank is a stable going concern is to understand the sources and quality of its cash flows not the market value of those assets.
Fair value accounting assumes that the market can make this cash flow assessment accurately. In the statement I quoted earlier, FASB expressed the view that "Fair value portrays the market's estimate of the present value of the net future cash flows" on assets held by banks. How could the market know this? The answer is, it doesn't, and it can't. The market does not know what the cash flows are to specific portfolios of mortgage-backed securities held by banks. Market participants might know that the cash flows are not what they should be--that the losses are greater than expected--but they can't value these securities in any more sophisticated way.
In fact, in the current crisis it is the market's ignorance concerning the cash flows on these portfolios that has been the source of the crisis and the losses that the banks have registered. When it became clear that there were greater losses on mortgage-backed securities than their triple-A ratings implied, the market for these and other asset-backed securities basically collapsed, and has been closed for about 18 months. Liquidity in this market has disappeared. Under these circumstances, there are few buyers for these assets, not because they believe assets are "toxic"--that's the media's word--but because it is not clear that they will ultimately be able to resell the assets when necessary. What the market knows is not the value of the portfolio's discounted cash flows, but only that buying these assets creates a huge liquidity risk for the buyer.
When the banks mark to market, they follow two steps. First, they estimate the net realizable value of their portfolios of asset-backed securities. This involves discounting the cash flows on these assets. Then, under fair value accounting, they have to take a haircut on these values that takes into account the price at which they could sell the assets. When the market is not functioning, of course, this haircut is very large. This is important because what it suggests is that the huge decline in the value of bank assets is not due to a commensurate decline in the cash flows on the bank's portfolios of asset-backed securities--although some decline there has certainly occurred--but rather to the market's judgment about the risk of resale by a purchaser. It is this risk that--when combined with fair value accounting--has forced the writedowns in bank assets.
A hint of the true situation was contained in remark by Vikram Pandit, the CEO of Citibank, in testimony before the U.S. Congress in February. He noted that Citi marks to market and "those marks are reflected in the losses we've taken, as well as in our income statement and balance sheets." But he went on to point out that the bank has a duty to shareholders: "the duty is if it turns out [the assets] are marked so far below what our lifetime expected credit losses are"--i.e., their net realizable value on a discounted cash flow basis--"I can't sell [them]." In other words, the writedowns caused by fair value accounting may have driven some Citi assets below their value under discounted cash flow analysis. This will have made Citi look substantially weaker than the real value of its assets--their net realizable value--would suggest. In effect, fair value accounting has turned a liquidity problem in the asset-backed market into a solvency problem for the world's banks.
That is only one of the major unintended consequences of fair value accounting. Another may be putting banks or insurance companies--firms that are expected to be stable and prudential in their behavior--into earnings competition with securities firms and hedge funds. It is fashionable in Washington today to refer to securities firms, hedge funds, and other financial intermediaries as part of the world of "shadow banking." This is one of those phrases that obscures more than it reveals, but one of the things it reveals is that most commentators who use the term don't really see any material difference between banks and other financial intermediaries. This, as I suggested, is also the perspective of the accounting standards that are now applicable to banks.
But banks actually are different--a difference always recognized by government policy. Banks may hold assets that are similar to those of other intermediaries, but there the difference ends. Banks are generally backed directly by governments, through deposit insurance, lender of last resort facilities, and exclusive access to the payment system. Other enterprises have none of these advantages. Bank deposits can be withdrawn or transferred on demand, and by creating credit that draws on these facilities banks directly affect the money supply. The liabilities of other financial intermediaries do not have that unique characteristic. Because of the nature of their liabilities, banks cannot easily match the maturities of their assets and liabilities. In fact, one of their unique roles is converting short term liabilities into longer term assets, so that depositors can have the advantages of highly liquid assets, but also returns that are closer to the yields on longer term assets. Other intermediaries serve important purposes, but not these. In other words, banks have unique elements that seem to make their stability potential more important than their earnings potential.
Given these substantial differences, is it a sensible policy to ask banks to compete on the same financial playing field with securities firms and hedge funds? When we have created this competitive accounting environment, perhaps we should not be surprised that banks hired Wall Street traders and leveraged themselves to the hilt. There is such a thing, of course, as risk-adjusted earnings, in which companies' results are judged not by their absolute amount, but by the risks they took to earn these returns. Investors, in theory, should be happy with lower returns from companies that take fewer risks. Maybe this works from the standpoint of thoughtful and prudent investors, but what is the effect on banks' managements when securities firms are producing much higher returns, and when compensation depends on matching the other guy's earnings results? Are they satisfied to tell investors--and are investors satisfied when told--that although their bank's earnings are lower than other financial institutions they were produced by more conservative activity? Isn't there a temptation, since financial results are reported in the same way, to try to match those higher returns?
Banks had a head start on this goal when, as asset values climbed in the mid-2000s, fair value accounting allowed them to write up the value of their assets. The more assets they put in their trading accounts, the more risks they were taking--but the more unrealized gains from asset appreciation enhanced their bottom lines. This adds some context, and some bitter comedy, to the classic statement of Chuck Prince, the chairman of Citibank, who famously remarked, as the bubble began to slow, "as long as the music is playing, you've got to get up and dance. We're still dancing..."
So, my conclusion about fair value accounting is that it should not be applied without distinction to all financial institutions. While there is some value in uniformity of disclosure and achieving comparability among the financial reports of financial institutions, we lose more than we gain by doing so. Even more salient is the fact that what investors want to know about commercial banks is just not as important for the success of banks--and the economies that depend on them--as what depositors, lenders and counterparties need to know. Risk-takers such as securities firms and hedge funds should be judged by their returns, but banks are different and should be judged by their likelihood to remain stable in economic storms. This calls for valuing their assets in a way that focuses on their stability, not on their earnings potential.
Thus, except for assets held in trading accounts--that is, acquired or held for the purpose of sale--asset backed and debt securities held by banks should be valued on the basis of their discounted cash flows. An alternative would be to allow banks to choose how their assets will be valued, as long as they disclose the method they have chosen, and cannot move between the different methods without very good reason.
Assuming that banks are exempted--or at least have the option to choose--in what way should fair value accounting be modified in order to make it work better for those still bound by it?
First, accounting should reflect broader interests than the goals of investors and accountants. In other words, to paraphrase Clemenceau on war and generals, accounting is too important to be left to the accountants. Yes, accounting practitioners would like to make financial statements more comparable across financial industries, and this accords with the desires of equity investors. But a more important issue, as we now know, is making sure that the financial statements of financial institutions of all kinds are not distorted by unanticipated moves in market prices. The same issue arises in connection with another accounting objective, preventing management bias. This can be accomplished by insisting woodenly on market prices, but at too heavy a cost. Earnings management is an endemic problem throughout accounting--nonfinancial firms have always managed their earnings and still are--so a fix with much broader applicability is required.
Second, fair value accounting is highly pro-cyclical. We can now see how the mark-to-market effect of fair value accounting has caused a downward slide in asset values, and how this decline has evolved into a dangerous downward spiral. But it is important to note that rising asset prices have the opposite--and equally pro-cyclical--effect. As market values rise for homes, stocks, commodities, or any item that has a readily available price, more and more credit becomes available to carry these assets. As more credit is available, more money is chasing fewer assets; prices rise and risk premiums fall.
Under fair value principles, a rise in the value of assets is recognized in earnings if the assets are held for trading and recognized in the institution's capital or equity position if the assets are treated as available for sale. In both cases, the growing earnings and strengthening capital induces more borrowing and the acquisition of more assets, so the upward spiral—also known as a bubble--continues. Given the fact that we human beings are prone to irrational exuberance when values are rising and to irrational pessimism when they are falling, it would seem that using an accounting system that exacerbates those flaws in our nature would not be good policy. If anything, accounting--which has always been dominated by a principle of conservatism--should operate counter-cyclically, suppressing the effect on both balance sheets and income statements of rapid and substantial changes in asset values. There is nothing about fair value accounting that has this effect.
While banks, and probably insurance companies, should be exempted from fair value accounting, some way should be found to suppress the pro-cyclical effects of market prices on other financial institutions. As long as the focus on earnings is the dominant purpose of accounting, these risk-taking institutions should still be subject to fair value accounting. But some restrictions should be placed on its scope. The most fruitful way to approach this is to focus on the question of when there is actually a functioning market.
For example, we could specify that mark to market accounting for assets would be suspended if, during any three week period, it reflects less than 20 percent or more than 150 percent of the dollar value of trading that was the weekly average in the preceding year. Fair value accounting would then become applicable again when trading is again at least 80 percent or not more than 120 percent of that annual index figure. Obviously, there is no magic in these numbers, but they suggest one way that pro-cyclicality could be addressed for the financial institutions to which it is still applicable.
In any event, if we retain fair value accounting in its current form after the current crisis is behind us, we will always be living on the edge of another financial abyss.
Peter J. Wallison is the Arthur F. Burns Fellow in Financial Policy Studies at AEI.