An important and long-standing question in the economics of information is whether voluntary disclosure leads to full disclosure. A compelling and intuitive argument, often described as the “unraveling” argument (see Milgrom, 1981), suggests that the answer is, yes. In brief, the argument is that the firm, or more generally the “sender,” with the most favorable information will voluntarily disclose. So the audience for the disclosure—the “receiver”—will interpret non-disclosure as indicating that the firm does not have the most favorable information. But given this, the firm with the second most favorable information will disclose, and so on. All the firms thus disclose in the end.
Despite the force of the unraveling argument, the prediction of full disclosure appears overstated. There are many cases in which valuable information that is potentially disclosable is not disclosed. Firms do not voluntarily reveal all value-relevant information. In legal proceedings, defendants frequently opt for silence. Politicians do not always voluntarily reveal past tax returns. In such cases, potential disclosers believe that non-disclosure is in their best interests, even though audiences often interpret non-disclosure with skepticism.
In our study, we give a new yet simple explanation for non-disclosure, which captures the idea that non-disclosure is the safest option available. Our explanation has two key elements. The first is that the sender is unsure of the receiver’s preferences. For example, a firm selling financial securities may be unsure whether the buyers of its securities desire higher or lower cash flow variance. Even if the firm’s private information is about the level of cash flows rather than their variance, the firm’s disclosure preferences will differ depending on whether the “buyer” of the firm’s securities is an investor or a tax authority (or regulator or union). A politician who is considering disclosing past tax returns may be unsure whether voters wish to see high income (thereby indicating that he is rich and successful) or low income (thereby excusing the low taxes he is known to have paid).
By itself, however, sender uncertainty about receiver preferences is not enough to generate non-disclosure. The reason is that the expected payoff from disclosure can still be ordered from the highest to the lowest, so that one can still identify senders with the highest incentive to disclose, and the unraveling argument still applies. Thus, the second key element of our explanation is sender risk-aversion, so that senders care about the risk of payoffs induced by disclosure relative to non-disclosure, in addition to simply the expected payoff. We show that non-disclosure arises precisely when it is safer than disclosure, in the sense of leading to lower risk in payoffs.
Consider, for example, a firm with private information about its cash flow variance, which it can voluntarily disclose. The firm is selling securities. It knows that some investors would prefer high-variance cash flows, while other investors would prefer low-variance cash flows, but does not know which type of investor it is dealing with. Thus, a disclosing firm faces a lottery over different prices that it may receive for its securities, where the prices depend on the corresponding investor’s preferences over cash flow variance. Firms with extreme values of cash flow variance—i.e., either very high or very low variance—face a particularly high-risk lottery over prices, because in these cases the valuation of an investor who likes high-variance cash flows diverges widely from the valuation of an investor who dislikes cash flow variance.
In a non-disclosure equilibrium, firms with extreme information stay silent and do not disclose, while firms with intermediate information disclose. Investors correctly interpret non-disclosure as indicating extreme information—in the example above, either very low or very high cash flow variance. So the price they are willing to pay is based on an average of these extremes. In particular, this means that the price paid if investors want high variance is close to the price paid if investors want low variance. So non-disclosure generates a lower-risk lottery over prices for firms with extreme information, relative to the alternative of disclosing. In short, silence is safest.
Given the economic forces underlying equilibrium non-disclosure and the failure of unraveling, it is natural to conjecture that non-disclosure becomes more likely as sender risk-aversion increases. Our analysis makes this intuition precise. Similarly, non-disclosure exposes receivers to risk by reducing their ability to differentiate between different sender types. Consequently, non-disclosure becomes less likely as receiver risk-aversion increases.
Our theory has many applications in finance and economics. Before a potential disclosure, a start-up firm selling new securities may be unsure about the risk-return preferences of potential investors, a financial advisor may be unsure about clients’ preferences, a target firm may be unsure as to whether the bidding firms’ technology is a complement to or a substitute for its own technology, and a firm disclosing may be unsure as to whether its disclosure will be received by investors or by a regulator. In all these scenarios, non-disclosure may arise as an outcome due to our silence-is-safest argument. Our prediction that firms with extreme information stay silent and do not disclose, while firms with intermediate information disclose, is also supported by recent empirical findings. Luca and Smith (2015) find that top business schools are least likely to disclose their rankings, whereas mid-ranked schools are most likely to disclose. Similarly, Bederson, et. al. (2017) find that the highest-rated restaurants do not disclose their hygiene ratings, while relatively higher-rated restaurants disclose to stand out from other lower-rated ones.
Bederson, Benjamin B., Ginger Zhe Jin, Phillip Leslie, Alexander J. Quinn, and Ben Zou (2017), “Incomplete Disclosure: Evidence of Signaling and Countersignaling,” American Economic Journal: Microeconomics, forthcoming.
Luca, Michael and Jonathan Smith (2015), “Strategic Disclosure: the Case of Business School Rankings,” Journal of Economic Behavior & Organization, Vol. 112, pp. 17-25.
Milgrom, Paul R. (1981), “Good News and Bad News: Representation Theorems and Applications,” Bell Journal of Economics, Vol. 12, pp. 380-391.
This post comes to us from professors Philip Bond and Yao Zeng at the University of Washington’s Michael G. Foster School of Business. It is based on their recent paper, “Silence is Safest: Non-Disclosure When the Audience’s Preferences are Uncertain,” available here.