As the U.S. savings and loan crisis that rocked the financial and real estate markets some 15 years ago came to an end, accountants strove to understand just what had gone wrong. A key problem, many experts suggested, lay with the methods used to value assets and liabilities on balance sheets.
At the time, the dominant valuation approach was historical-cost accounting. This method reports the original amount paid or received for assets and liabilities on a company’s balance sheet, as well as earnings generated by the assets and liabilities on the income statement. Banks and savings and loan institutions used historical-cost accounting to report the value of loans and core deposits on their balance sheets, and equity analysts valued these firms from their earnings.
But historical costs are not current values; if interest rates change, the value of deposits and loans also change, and this change forecasts the firm’s ability to generate earnings. Just as declines in the market value of mortgages in today’s subprime mortgage market indicate potential investment losses, so too the change in the value of loans relative to deposits of savings and loans 15 years ago would have indicated losses. While those losses eventually would have been reported in earnings, market values sound the distress signal a little earlier.
In light of the savings and loan failures, many experts argued for a switch to fair-value accounting — marking assets and liabilities on the balance sheet to their market value. Regulators, including the U.S. Financial Accounting Standards Board and the International Accounting Standards Board, are moving steadily in this direction. Fair value is defined by these bodies as the price at which a firm could sell an asset (or the price it would have to pay to be relieved of a liability).
However, many companies — including retail banks — may be more difficult to value if pushed to comply with the fair-value approach, say Professors Doron Nissim and Stephen Penman in a new white paper.
“We need to be careful about the application of fair value,” Penman cautions, noting that the accounting shift is arguably the most important and controversial issue facing regulators and accounting standard-setters today. “Fair value is appropriate for a bank’s trading book — the assets and liabilities like shares and bonds whose values depend directly on the change in their market price — but is not appropriate for valuing the banking book, the part of a bank that deals with customer transactions such as collecting deposits and making loans,” Penman says.
How effective a bank is at delivering its products to depositors and borrowers determines the value of the assets and liabilities in the banking book. Accordingly, the value of those assets and liabilities to the bank may be quite different from what they can be sold for on the market. Further, because active markets do not exist for most bank loans and deposits, accountants have to estimate their value. Estimates can introduce errors and, worse, invite abuse, Penman notes.
“The problems that arise in applying fair-value accounting to financial institutions are magnified when it comes to industrial and technology firms,” Penman says. “The value of a pile of coal to a steel producer, for example, comes from its use in making steel, not from what the firm can sell the coal for. Just as bank deposits derive their value as part of a business that makes money from customers, so too with the assets of industrial firms, but these firms have many more assets that do not lend themselves to fair-value accounting. Fair value is often informative, but it does not serve as a panacea for accounting.”
Nissim and Penman argue that value comes, not from the stand-alone market prices of individual assets, but from combining assets under a business model, like using coal, raw iron and smelters together to produce steel. Value is in the entrepreneurial idea, not the market value of the individual assets employed to execute the idea. Indeed, rather than seeing value in market prices, businesses arbitrage market prices to add value — selling products at prices higher than input prices — such as selling steel at a price that is higher than prices paid for coal and other inputs, and thus make a profit.
Accordingly, fair-value accounting that marks assets to market has its limitations for any business that gets value from trading with customers. For these companies, the notion of top-line revenues takes the fore; value is best determined by examining earnings that measure the value added from trading with those customers.
So when does it make sense to use fair-value accounting? In the white paper, Nissim and Penman lay out some principles about circumstances under which fair-value accounting makes sense.
Foremost is a one-to-one principle: fair value is appropriate when shareholder value varies directly with the market price. That is the case with shares and bonds held in a bank’s trading book: if the price of a share held by the bank goes up by a dollar, so does the value of the bank. That is also the case with a hedge fund that holds investments whose value changes with their market price.
Appropriate fair-value accounting also honors a matching principle: assets can be marked to market only if associated liabilities are also marked to market. If this principle is honored, the resulting gains and losses are matched with the offsetting gains and losses on the liabilities. So, the debt of a firm can be marked to market when its credit quality (and thus price) declines only if the associated assets (whose decline in value resulted in the deterioration of the debt) are also marked to market. In this way the gain from the change in the market price of the debt is offset by the loss from the drop in the value of the assets.
An estimate of market value would be required if market prices for those assets and liabilities were not available, as is often the case. Estimates are made using valuation models — thus the term marking to model rather than marking to market. So a no-arbitrage estimation principle says that the valuation models employed must be so-called no-arbitrage models. This means that the price estimated by the model must not imply that the firm makes an immediate profit against another observed price.
The Black-Scholes option pricing model, which assumes no arbitrage between the option price and the price of the underlying shares, has this feature. So, if the firm estimates the price of a stock option using the Black-Scholes model, there is no implication that the firm can sell the option at the estimated price and buy the stock to make a profit: the option price is a fair one given the price of the underlying stock. The same is not true for a discounted-cash-flow model, which involves arbitrage of current and future input and output prices — the firm is in the business of buying inputs at one price and selling products later at a higher price, thus arbitraging input and output prices to make a profit.
The no-arbitrage estimation principle disciplines the marking-to-model practice, which has been of particular concern to many who question fair-value accounting.
“Historical-cost accounting is far from perfect,” Penman concedes. “But we must be careful in moving to fair-value accounting as a remedy. Focus on better historical-cost accounting may be an alternative way for accounting standard-setters to go.”
Doron Nissim is professor of accounting and chair of the Accounting Division and Stephen Penman is the George O. May Professor of Accounting and codirector of the Center for Excellence in Accounting and Security Analysis (CEASA) at Columbia Business School.