The Real Costs of Disclosure

The following post comes to us from Alex Edmans, Professor of Finance at the London Business School; Mirko Heinle of the Department of Accounting at the University of Pennsylvania; and Chong Huang of the UC Irvine Paul Merage School of Business.

In our paper, The Real Costs of Disclosure, which was recently made publicly available on SSRN, we analyze the effect of a firm’s disclosure policy on real investment. An extensive literature highlights numerous benefits of disclosure. Diamond (1985) shows that disclosing information reduces the need for each individual shareholder to bear the cost of gathering it. In Diamond and Verrecchia (1991), disclosure reduces the cost of capital by lowering the information asymmetry that shareholders suffer if they subsequently need to sell due to a liquidity shock. Kanodia (1980) and Fishman and Hagerty (1989) show that disclosure increases price efficiency and thus the manager’s investment incentives.

However, the costs of disclosure have been more difficult to pin down. Standard models (e.g. Verrecchia (1983)) typically assume an exogenous cost of disclosure, justified by several motivations. First, the actual act of communicating information may be costly. While such costs were likely significant at the time of writing, when information had to be mailed to shareholders, nowadays these costs are likely much smaller due to electronic communication. Second, there may be costs of producing information. However, firms already produce copious information for internal or tax purposes. Third, the information may be proprietary (i.e., business sensitive) and disclosing it will benefit competitors (e.g., Verrecchia (1983) and Dye (1986)). However, while likely important for some types of disclosure (e.g., the stage of a patent application), proprietary considerations are unlikely to be for others (e.g., earnings). Perhaps motivated by the view that, nowadays, the costs of disclosure are small relative to the benefits, recent government policies have increased disclosure requirements, such as Sarbanes-Oxley, Regulation FD, and Dodd-Frank.

Our article reaches a different conclusion. We show that, even if the actual act of disclosure is costless, a high-disclosure policy can still be costly due to its effect on real investment. Central to our analysis is the idea that only some types of information (“hard”, i.e., quantitative and verifiable) can be credibly disclosed, but others (“soft”, i.e., non-verifiable) cannot be. For example, a firm can credibly communicate its earnings, but not the quality of its corporate culture. It may seem that this distinction does not matter: even if a firm cannot increase the amount of soft information, it can still disclose more hard information. The absolute amount of overall information will rise, reducing the cost of capital. However, the manager’s investment decision depends on the relative weighting between hard and soft information. If neither type of information is disclosed, the manager chooses the investment policy that maximizes firm value. In contrast, an increase in the absolute amount of hard information disclosed also augments the amount of hard information relative to soft. This in turn distorts the manager’s actions towards improving the hard signal at the expense of the soft signal—for example, cutting investment in corporate culture to increase current earnings.

Our model features a firm initially owned and run by a manager, who must raise funds from an outside investor. After funds are raised, the firm turns out to be either high or low quality, and this type is unknown to the investor. As in Diamond and Verrecchia (1991), the investor may subsequently suffer a liquidity shock which forces her to trade additional shares. Also present in the market is a speculator (such as a hedge fund) who has private information on firm value, and a market maker. The investor expects to lose to the speculator from her liquidity trading and thus demands a larger stake when contributing funds, augmenting the cost of capital.

The manager can reduce the investor’s informational disadvantage, and thus the cost of capital, by disclosing a hard signal (such as earnings) that is partially informative about firm value, just before the trading stage. We initially assume that the manager can commit to a disclosure policy when raising funds, as in the literature on mandatory disclosure. High disclosure indeed reduces the cost of capital, but has an important cost. A high-quality firm has the option to undertake an intangible investment that improves the firm’s long-run value, but this value cannot be disclosed as it is soft information. The investment also raises the probability of delivering low earnings, which lowers the short-term stock price since low-quality firms always generate low earnings. Thus, a manager concerned with the stock price will underinvest. While existing literature typically assumes that firm value is exogenous and studies the optimal level of information to disclose about this fixed value, here firm value is endogenous to the disclosure policy (even absent a competitor who can use the disclosed information).

The optimal level of disclosure is thus a trade-off between the benefits of disclosure (reduced cost of capital) and its costs (inefficient investment). The manager chooses either full disclosure to minimize the cost of capital, or partial disclosure to maximize investment. Thus, the model delivers predictions on how disclosure should vary cross-sectionally across firms. The effect of firm characteristics on disclosure is typically non-monotonic. Up to a point, increases in growth opportunities reduce disclosure: investment becomes sufficiently important that the firm is willing to sacrifice disclosure to pursue it. For example, at the time of its IPO, Google announced that it would not provide earnings guidance as such disclosure would induce short-termism. Their founders’ letter stated “[w]e recognize that our duty is to advance our shareholders’ interests, and we believe that artificially creating short term target numbers serves our shareholders poorly.” However, when investment opportunities are very strong, the manager will exploit them fully even when disclosure is high. Thus, disclosure is lowest for firms with intermediate growth opportunities, and high for firms with weak or strong growth opportunities. For similar reasons, disclosure is high either when information asymmetry (the difference in value between high- and low-quality firms), shareholders’ liquidity shocks, or signal imprecision (the risk that investment leads to a bad signal), are low, as the manager will invest fully even with high disclosure, or when these parameters are high, as the manager will invest fully even under high disclosure. For example, an increase in signal imprecision does not necessarily induce less disclosure of the signal. Such an increase makes the cost of capital relatively more important to investment, and so the manager may choose full disclosure to minimize the cost of capital.

We next consider the case in which the manager cannot commit to a disclosure policy, as in the literature on voluntary disclosure. If investment is important, the manager would like to announce a “low disclosure, high investment” policy. However, if the manager invests and gets lucky, i.e., still delivers high earnings, he will renege on the policy and disclose the high earnings anyway. Knowing that he will always disclose high earnings if realized, the manager will reduce investment, to maximize the probability that he realizes high earnings. Then, if the market does not receive any disclosure, it rationally infers that the signal must be low, else the manager would have released it—the “unraveling” result of Grossman (1981) and Milgrom (1981). The only dynamically consistent policy is full disclosure, and investment suffers as a result. In this case, government intervention can be desirable. By capping the feasible level of disclosure, it can allow the firm to implement the optimal policy. This conclusion contrasts earlier research which argues that regulation should increase disclosure due to externalities (Foster (1979), Coffee (1984), Dye (1990), Admati and Pfleiderer (2000), and Lambert, Leuz, and Verrecchia (2007)). If capping disclosure is difficult to implement, a milder implication of our model is that regulations to increase disclosure (e.g., Sarbanes-Oxley) may have real costs.

The full paper is available for download here.

Both comments and trackbacks are currently closed.
  • Subscribe or Follow

  • Supported By:

  • Program on Corporate Governance Advisory Board

  • Programs Faculty & Senior Fellows