Review of Accounting Studies, 3, 261–287 (1998)
1998 Kluwer Academic Publishers, Boston. Manufactured in The Netherlands.
Investor Sophistication and Voluntary Disclosures
RONALD A. DYE
J. L. Kellogg Graduate School of Management, Department of Accounting and Information Systems,
Leverone Hall, Northwestern University, 2001 Sheridan Road, Evanston, Illinois 60208
about whether a ﬁrm has received information. The ﬁrm’s value is established via a ﬁrst price, sealed bid,
common value auction. The paper demonstrates that the threshold level determining whether the ﬁrm withholds
or discloses information uniformly declines in the probability investors are informed. The paper also shows that,
notwithstanding the risk-neutrality of investors, the expected selling price of the ﬁrm strictly decreases (increases)
in the probability individual investors are informed when that probability is small (large). These results follow
from “winner’s curse” effects.
This paper studies voluntary disclosures in a model in which some investors may be better
informed than others regarding when a ﬁrm receives private value-relevant information. An
“informed” investor knows when the ﬁrm has received information that it has not disclosed;
an “uninformed investor” cannot tell whether the ﬁrm’s lack of disclosure is due to its lack
of receipt of information or due to the ﬁrm having withheld information it has received.
Each investor’s information set is private to him: neither the ﬁrm nor any other investor
knows what any given investor knows. This is thus a model of trilateral information
asymmetry, with investors potentially ignorant of what the ﬁrm knows, the ﬁrm ignorant of
what investors know, and investors ignorant of what other investors know. It constitutes an
extension of previous models of voluntary disclosure of Dye (1985) and Jung and Kwon
(1988), in which the information asymmetry is unilateral: in those models, a ﬁrm sometimes
receives private information that it may or may not disclose; all investors are unsure when
the ﬁrm has received information, and they all interpret the ﬁrm’s disclosures or lack of
disclosures in the same way. The principal reason this extension is of interest is that it
may be regarded as the typical situation that ﬁrms and investors actually face: managers
cannot predict exactly the price reactions attending their disclosures or nondisclosures of
information to capital market participants, in part because they do not know the capital
market participants’ entire information set, and investors may not know either what rival
investors know or whether the ﬁrms they follow are withholding value-relevant information.
In this information environment, the paper examines the behavior of investors and an
expected value-maximizing ﬁrm in a model where the ﬁrm’s market value is established as
the outcome of a ﬁrst price, sealed bid, common value auction, with each investor having
the same chance f ∈ [0, 1] of becoming informed (in the sense above).
The models of Dye
(1985) and Jung and Kwon (1988) are special cases of this analysis in which f = 0. When
the ﬁrm does not disclose information, informed investors have different assessments of
the conditional expected value of the ﬁrm from uninformed investors, since only informed