A book-keeping error
Aug 30th 2007
From The Economist print edition
The accounting principle that is meant to capture fair value might end up distorting it
AS THE old joke goes, there are three types of accountant: those who can count and those who cannot. What and how they count is often contentious. A long-fermenting issue is how far “fair-value” accounting, which uses up-to-the-minute market information to price assets, should be pushed in banking. The bodies that set accountancy standards believe the more accurate disclosures are, the better. Regulators meanwhile have fretted that market-based accounting would increase fluctuations in banks' earnings and capital, which might increase risks to financial stability. And commercial banks are reluctant to expose the idiosyncrasies of their loan books to the glare of market scrutiny.
The attractions of fair-value accounts are straightforward. By basing values on recent prices (“marking to market”), they paint a truer picture of a firm's financial health than historical-cost measures. These gauge net worth from the arbitrary dates when assets and liabilities were first booked. In principle, fair-value accounting makes a firm's viability plainer and enables shareholders and regulators to spot financial trouble more quickly. Proponents say that market-based accounting would have limited the fallout from America's savings-and-loan crisis and stopped the rot from Japan's non-performing loans much earlier.
An arbitrary past versus a distorted present
New research suggests that the increasing reach of fair-value accounting might be a mixed blessing. A paper* by Guillaume Plantin of the London Business School, Haresh Sapra of the University of Chicago and Hyun Song Shin of Princeton University concludes that fair-value accounting could sometimes generate fluctuations in asset values that distort the very price information that it puts such store by.
The paper examines the incentives of a bank faced with a choice between selling a loan or keeping it on the balance sheet. Because the bank knows its borrower better than anyone else, it has the best idea of what the loan is really worth. Its managers are rewarded according to the accounting profit of the bank.
If loans are valued at historical cost and market values are rising, the loans are likely to be sold if this is the only way of realising profit, even if the market undervalues them. The banks' managers take a profit and get paid accordingly, although shareholders would be better off if the loans were kept. Fair-value accounting gets around this agency problem. Loans do not have to be sold to cash in on their rising value: marking the assets to their market value has the same beneficial effect on profits and on managers' pay.
However, in the wrong circumstances fair-value accounting could also induce wasteful sales—of long-term, illiquid loans. Left on the books and marked to market, a loan will be valued at the price at which others have managed to sell. But when there are only a few potential buyers, that may be especially low. So managers will be tempted to sell in the hope of a better price. Because all banks with similar assets face the same incentives, they will all sell, driving the price down. Their shareholders would have been better off had the loans been kept until they fell due. The temptation to sell is greater for longer-term loans.
In this way, a fair-value regime can itself distort the very prices that are supposed to reflect the true worth of assets. The prospect of lower prices can encourage selling which drives down prices further. The information derived from market prices becomes corrupted, and the result is a growing divergence between reported net worth and true value.
This theoretical model is a challenge to the ideal of fair-value accounting: that more information is always better. Although it is technically feasible to mark to market even idiosyncratic assets such as loans to small businesses, it might not be desirable. The authors point to a well-established principle in economics, that incremental moves towards perfect competition are not always good. Eliminating one market imperfection (such as poor information) need not bring the ideal of a frictionless economy closer, because this may magnify the effect of remaining distortions (such as managerial short-termism or illiquid markets).
The paper also underlines some lessons about market liquidity that have been painfully learned outside of academia in the recent market troubles. There is a fair chance that asset markets will stay liquid (in the sense that willing sellers are matched with willing buyers), as long as the actions of market participants are essentially random. But anything that co-ordinates the actions of sellers—in this case, the disclosure required by fair-value accounting—can easily lead to sharp movements in asset prices.
Is the model of self-defeating co-ordinated selling very realistic? Recently, for example, Bear Stearns, a Wall Street investment bank, held off from selling assets into an illiquid market because the transaction prices would have set a nasty benchmark for its other portfolios. So illiquidity prevented asset sales rather than induced them. Mr Shin replies that in instances like this, where there happens to be a dominant holder of assets, there is less chance of sales into a falling market.
Although more accurate disclosure of balance sheets is desirable, the work of Mr Shin and his colleagues is a reminder that there are always trade-offs to any policy change. These authors put their argument in stark terms: “The choice between these measurement regimes boils down to a dilemma between ignoring price signals, or relying on their degraded versions.” In their advocacy of fair-value accounting, accountants are rightly pursuing the interests of investors. But policymakers have to worry about wider issues. Accountancy may be too important to be left solely to accountants. Even the ones that can count.