Print
E-MailGuillaume Plantin is Professor at Sciences Po and Jean Tirole is Chairman at the University of Toulouse School of Economics. This is post is based on their recent article, forthcoming in the American Economic Review.
Accounting statements are the primary source of verified information that firms provide to their stakeholders, and therefore an important ingredient of corporate governance. Accounting measurements are in particular explicit inputs in executive compensation contracts, debt covenants, and regulations such as prudential rules for financial institutions. They also play a more implicit but pervasive role in the enforcement of stakeholders’ rights during events that are defining for corporations, such as takeovers, proxy contests, bankruptcy procedures, or rounds of venture-capital and private-equity financing.
Amidst a global debate that has been raging for years, accounting conventions have evolved from the use of historical costs towards “fair-value” measurements of assets and liabilities. The International Accounting Standard Board defines fair value as “the price that would be received to sell an asset or paid to transfer a liability in an orderly transaction between market participants at the measurement date”. This contrasts with historical-cost accounting whereby, broadly, balance-sheet items remain recorded at their entry value instead of reflecting all relevant data accruing from markets for similar items.
The goal of this article is to apply information economics to the study of optimal accounting measures. We focus in particular on the two-way interaction between the nature and quality of market data and the extent to which it is used for accounting purposes. Our starting point is a standard agency model of corporate finance in which the outside stakeholders of a firm need to provide inside stakeholders with incentives to figure out a value-maximizing strategy, which we simply model as optimally selecting an asset. Insiders’ rewards (prolonged employment, authorization to invest, managerial compensation…) must be decided before the asset pays off, and must therefore be based on measurements of this payoff. Two such measurements are available to outsiders. They can avail themselves of a costless but noisy public signal from a market for similar assets. They can also obtain a costly measure of asset value by reselling the asset to imperfectly competitive buyers. Uninformed and informed buyers make bids for the asset.
Given this informational environment, optimal corporate governance has the following simple structure. If the market signal for the asset is above a threshold, then insiders get rewarded. If the signal is below this threshold, insiders are allowed to resell the asset above a given reserve price, and get rewarded if the sale is actually executed above this price. This intuitive but abstract contract admits a realistic implementation, whereby insiders are rewarded if and only if a carefully designed accounting measurement of the project is above a threshold. The important feature of the measurement is its degree of conservatism. Under a more conservative regime, the accounting measure of the project (its “book value”) increases more gradually following positive market data. In this sense the recognition of latent capital gains is more conservative. Insiders are therefore induced to realize their latent gains, i.e. take their asset to the market, more often in order to get rewarded. In the limit of a most conservative regime, only realized capital gains are recognized, as is the case under pure historical-cost accounting in practice.
Interestingly, this (privately) optimal contract trades off costs that closely mirror some of those mentioned by each side in the policy debate on fair-value accounting. Advocates of fair-value accounting have argued that historical-cost accounting induces distortions such as costly and unnecessary realizations of latent capital gains by firms. This practice of taking to the market assets that are booked below their resale values for no other purpose than increasing a firm’s book value is referred to as “gains trading’’. The opponents of fair value point at the irrelevant noise that market data may add to corporate performance measure. Accordingly, our optimal contract (and accounting measure) minimizes the sum of two types of costs, those incurred when validating insiders’ claims with an ex-post inefficient asset resale (“gains trading’’) and the costs induced by the inefficient reward of insiders based on noisy market data absent such a resale (“reward for luck’’).
The article then determines the equilibrium quality of market data when each firm writes such an optimal contract taking market liquidity as given, and potential buyers of firms’ assets bid for the assets rationally anticipating such contracts. Our main insight is that laissez-faire generically leads to a socially insufficient degree of accounting conservatism. With such endogenous asset liquidity, the equilibrium exhibits a form of “bootstrapping”. Firms contract too much on transactions by other firms (excessive marking to market). They sell their own assets too rarely, and at deep discounts when they do so. Imposing a higher degree of conservatism benefits firms. This induces them to resale their assets more often, which spurs information acquisition by potential buyers thereby improving liquidity in the market for firms’ assets, where liquidity is defined both in terms of ease of trading and of the informativeness of price signals. Under such regulation, taking assets to market is more efficient because asset resales occur at higher prices. So is marking to market because resale prices are more informative.
The complete article is available here.
