Journal of Financial Economics 99 (2011) 289–307

Contents lists available at ScienceDirect

Journal of Financial Economics

journal homepage: www.elsevier.com/locate/jfec

Empty voting and the efficiency of corporate governance$

Alon Brav 1, Richmond D. Mathews 
Fuqua School of Business, Duke University, 1 Towerview Dr., Durham, NC 27708, USA

a r t i c l e in fo abstract

Article history: We model corporate voting outcomes when an informed trader, such as a hedge fund,
Received 11 March 2009 can establish separate positions in a firm’s shares and votes (empty voting). The
Received in revised form positions are separated by borrowing shares on the record date, hedging economic
3 February 2010
exposure, or trading between record and voting dates. We find that the trader’s
Accepted 2 March 2010
Available online 13 October 2010
presence can improve efficiency overall despite the fact that it sometimes ends up
selling to a net short position and then voting to decrease firm value. An efficiency
JEL classification: improvement is likely if other shareholders’ votes are not highly correlated with the
G34 correct decision or if it is relatively expensive to separate votes from shares on the
G38
record date. On the other hand, empty voting will tend to decrease efficiency if it is
relatively inexpensive to separate votes from shares and other shareholders are likely to
Keywords:
vote the right way.
Voting
Informed trading & 2010 Elsevier B.V. All rights reserved.
Hedge funds
Corporate governance

1. Introduction or other strategic traders can also significantly affect the

The impact of hedge funds on corporate governance
has received considerable attention recently as the rise in
popularity of hedge funds has coincided with an increased
focus on governance in general. Much of the attention has
been devoted to ‘‘activist’’ funds that take significant
stakes in firms and then advocate for change.2 However,
other more subtle strategies undertaken by hedge funds
$
We thank Daniel Ferreira (the referee), Alex Edmans, Ron Kaniel,
Samuel Lee, Adriano Rampini, David Robinson, Zacharias Sautner, Berk
Sensoy, S. Viswanathan, seminar/conference participants at Boston
College, Duke, NC State, the University of Washington, Virginia Tech,
the Financial Intermediaries and Markets at the Crossroads symposium,
the 2009 AFA Annual Meetings, and especially Simon Gervais for helpful
discussions and comments. All errors are our own.
 Corresponding author. Tel.: + 1 919 660 8026; fax: + 1 919 660 8038.
E-mail addresses: brav@duke.edu (A. Brav), rmathews@duke.edu
(R.D. Mathews).
1
The author is also at the NBER.
2 3
See Kahan and Rock (2007), Clifford (2008), Klein and Zur (2009),
Greenwood and Schor (2009), and Brav, Jiang, Partnoy, and Thomas
(2008).
efficiency of the corporate governance system.
In particular, recent work has shown that some funds
may use ‘‘empty voting’’—a practice whereby they
accumulate voting power in excess of their economic
share ownership—to manipulate shareholder vote out-
comes and generate trading gains. This practice is possible
even when one share, one vote is the explicit rule. It can
be accomplished, for example, by borrowing shares of
stock on the record date or hedging economic exposure in
the derivatives markets. Hu and Black (2006, 2007)
provide a number of examples where such behavior
seems to result in perverse voting incentives. In one case,
a hedge fund acquired votes by borrowing shares, then
voted against a buyout proposal and apparently profited
from a short position when the share price dropped
following the vote.3 These authors suggest that some
form of regulation, starting with additional disclosure
This incident involved a Hong Kong company named Henderson
Land, which wanted to buy out a 25% minority interest in its publicly
traded affiliate Henderson Investment.

0304-405X/$ - see front matter & 2010 Elsevier B.V. All rights reserved.
doi:10.1016/j.jfineco.2010.10.005
290 A. Brav, R.D. Mathews / Journal of Financial Economics 99 (2011) 289–307
requirements, may be necessary to curb the negative
effects of such activities.
Regulators have expressed significant concern over
empty voting, particularly given the boom in the hedge
fund industry and the increasing number and importance
of items requiring a shareholder vote. The Wall Street
Journal (January 26, 2007, p. A1) quotes Securities and
Exchange Commission chairman Christopher Cox as
saying that the practice of empty voting ‘‘is almost
certainly going to force further regulatory response to
ensure that investors’ interests are protected...This is
already a serious issue and it is showing all signs of
growing.’’ Many large institutional shareholders are
examining their share lending practices in response to
these concerns. In addition, some companies have
recently amended their bylaws to force additional
disclosure of complex transactions in their securities
due to concerns about corporate governance implications
(The Wall Street Journal, July 14, 2008, p. B4).
On the other hand, Christoffersen, Geczy, Musto,
and Reed (2007) argue that ‘‘vote trading’’ in the share
lending market can increase efficiency because informa-
tion about proposals can be costly to acquire. Uninformed
shareholders who are not willing to pay the cost to
become informed can sell their votes to informed parties
in order to increase the efficiency of the voting outcome.
Of course, this argument requires that the vote buyer
and vote seller have coincident interests, which often
seems to be violated in the examples cited by Hu and
Black (2006, 2007). To date, there is no agreement on
whether empty voting constitutes a significant problem
that should be regulated. Importantly, the literature does
not currently provide an integrated theoretical framework
to help assess the tradeoff between increased informa-
tion efficiency and the cost of possible manipulations via
empty voting.
In this paper, we develop a theoretical model to
explore this tradeoff. We derive the optimal share and
vote position of a strategic trader that has the ability to
acquire unique information about the value of a manage-
ment proposal and the ability to acquire votes separately
from shares. We show that while the trader may some-
times reduce efficiency by ultimately selling to a net short
position and then ‘‘voting the wrong way’’ (from a firm
value perspective), the cost of these possible manipula-
tions can be offset by a greater probability that the trader
will ‘‘do the right thing’’ and vote to maximize firm value.
In other words, in equilibrium both the presence of the
strategic trader and the ability to separate votes from
economic ownership can increase overall efficiency by
making the ‘‘right’’ outcome more likely. This occurs when
either the establishment of an empty voting stake on the
record date is relatively expensive or other shareholders’
votes are not very highly correlated with the true state.
However, we find that a negative efficiency effect is likely
when separating votes from shares is relatively inexpen-
sive and other shareholders are relatively likely to vote
the right way.
Our analysis deals with deviations from the one share,
one vote rule, on which there is a large existing literature
dating back to at least Manne (1964). Much of the modern
literature focuses on how the one share, one vote rule
affects the efficiency of the market for corporate control
(see, e.g., Harris and Raviv, 1988; Grossman and Hart,
1988; Burkart and Lee, 2008), or how disparities between
cash flow and voting rights held by insiders affects
efficiency (e.g., DeAngelo and DeAngelo, 1985; Gilson,
1987). These studies generally focus on long-term devia-
tions from one share, one vote that are codified in the
corporate charter. An important recent exception is Kalay
and Pant (2009), which shows that the ability to separate
economic and voting interests via derivatives markets can
increase efficiency by allowing shareholders to extract
more surplus in a control contest. Like Kalay and Pant
(2009), we examine short-term deviations arising from
activities in the derivatives or share lending markets.
However, we focus on how these deviations affect the
efficiency of voting by outsiders on regular proposals (as
opposed to control contests). We think of outsiders as
parties who do not make proposals themselves, but face
uncertainty over whether an insider’s proposal is value-
increasing or instead self-serving. There are many types of
proposals other than proxy contests or takeover bids for
control that can have important value implications for the
firm. Examples include proposals for the purchase of
another firm, a divestiture, or a change in the corporate
charter (often involving a takeover defense).
In our model, the firm’s management initially proposes
an action that requires shareholder approval. The pro-
posed action may be either good or bad (i.e., its approval
may either increase or decrease firm value), and its value
is not observable at this stage. All shares are initially held
by atomistic shareholders. After the proposal is
announced, a single strategic trader can buy or sell shares
in a transparent market prior to the record date (i.e., with
no noise trading) and can also acquire ‘‘extra’’ votes in
excess of its economic ownership by paying a convex cost.
This cost represents, for example, increasing difficulty in
finding shareholders from whom to borrow shares, or the
increasing cost of finding counterparties to hedge a large
economic interest.
On the record date, voting interests are set according
to share or vote ownership on that day. After the record
date, there is a significant time lag before the actual date
of the vote, so the strategic trader is able to further adjust
its economic ownership (but not its voting interest) as
well as learn about the value of the proposal.4 At this
intermediate trading stage, however, the market is not
completely transparent because there is noise trading by
atomistic investors. Finally, on the voting date the
strategic trader votes according to its economic incen-
tives, as determined by its net economic position in the
firm, while the voting of atomistic shareholders is
effectively random. We do not explicitly model the
atomistic holders’ voting decisions; the important feature
is that their behavior induces randomness in the final
voting outcome.
4
The timing of when the trader becomes informed does not
matter—it can occur either before or after the record date with no
significant change in the analysis.
 A. Brav, R.D. Mathews / Journal of Financial Economics 99 (2011) 289–307
Our assumptions are meant to reflect the realities of
corporate governance in the United States. Christoffersen,
Geczy, Musto, and Reed (2007) report that there is a
significant time lag between the record date and the
meeting date (a median of 54 calendar days in their
sample) as opposed to the relatively short time between
the announcement of the agenda and the record date.
Thus, it seems reasonable that there would be little ability
to trade strategically prior to the record date (which
corresponds to our assumption of a transparent market at
that stage), but a significant opportunity to gather
information and trade less transparently between the
record and voting dates.
It is important to note that we highlight two ways in
which empty voting can occur in the U.S. corporate
governance system. In addition to the lending and
derivatives markets, there is also the time lag between
record and voting dates. Even if voting and economic
interests have to be aligned on the record date, it is
possible to divorce the two prior to the voting date by
trading in the stock market during the intervening period.
Our model allows us to separate the two effects. As will
become clear, we find that the ease with which votes and
shares can be separated on the record date is of key
importance with respect to whether empty voting helps
or hurts efficiency.
In our model, we find that the strategic trader optimally
trades to a long economic position on the record date while
simultaneously acquiring ‘‘extra’’ votes, both of which set
the stage for the possibility of future trading gains. The
number of extra votes acquired depends on the cost of the
votes versus the value (in terms of larger expected trading
profits) of the increased ability to affect the vote outcome.
For the bulk of the analysis, we assume that the cost of
separating votes from ownership is high enough that the
trader will not acquire enough votes to single-handedly
determine the voting outcome.
The trader’s economic position on the record date is
driven by a separate tradeoff. On the positive side, a larger
economic stake increases the trader’s voting power. This
is valuable when extra votes are costly. On the negative
side, greater economic ownership reduces expected
trading gains for two reasons. First, the ‘‘future self’’ of
the trader will be concerned with protecting the value of
its stake in addition to maximizing trading gains. This
‘‘commitment effect’’ of owning an economic stake on the
record date thus reduces expected future trading profits.
Second, the strategic trader’s position reduces overall
market depth. In equilibrium, the extent of the long
position is determined by the expected amount of noise
trading between the record and voting dates, and the ease
with which votes can be acquired separately from shares
on the record date.
After the record date, the strategic trader plays a mixed
strategy; it either buys additional shares and then votes to
maximize firm value, or it sells to a net short position and
then votes to minimize firm value. We find in our
benchmark model that the presence of the strategic trader
is good for efficiency overall when the other shareholders’
votes are not highly correlated with the correct decision,
or when the ability to separate shares and votes on the
291
record date is highly restricted. Because of its long
position on the record date, the strategic trader tends to
‘‘vote the right way’’ more often than not, increasing the
probability of a correct decision in these situations. As
market depth increases, the optimal long position on the
record date increases, intensifying the positive effect.
Thus, we find that allowing for trading gains by a strategic
trader can increase efficiency even though the trader
sometimes engages in value-reducing strategies. Also,
since the strategic trader would not acquire any votes or
shares in the absence of possible trading gains, noise
trading improves voting efficiency.
The positive efficiency effect we document for these
cases is driven by the fact that the strategic trader has
unique value-relevant information that other share-
holders do not have. If the strategic trader brought no
new information to the model, there would be no
possibility of an efficiency improvement, but manipula-
tion could still be possible. As such, our model provides a
framework for determining whether and how an informed
trader’s unique information is ultimately reflected in the
final voting outcome and thus firm value.
On the other hand, we go on to show that efficiency
can be reduced by the strategic trader when other
shareholders’ votes are sufficiently biased toward the
correct decision and it is not too expensive to separate
votes from shares on the record date. Intuitively, when it
is easy to separate votes from shares on the record date,
the trader chooses a smaller long position and the
commitment effect is attenuated. In these cases, the
trader’s efficiency-reducing votes are relatively more
likely than its efficiency-enhancing votes to change the
actual decision if other shareholders are likely to vote the
right way on their own. Thus, the negative effects can
start to outweigh the positive ones. This occurs despite
the fact that the trader brings value-relevant information
to the table.
We also investigate how changes in the underlying
parameters affect the trader’s strategy as well as overall
efficiency. We find that making it easier to separate votes
from economic ownership on the record date tends to
decrease the trader’s economic ownership (in order to
maximize trading gains by avoiding the commitment
effect, as noted above). This increases the probability that
the trader will go net short and vote the wrong way later.
However, the net effect on efficiency can be positive when
votes are not too cheap, since the additional votes also
increase the trader’s ability to affect the voting outcome.
This increased ability to affect the outcome can outweigh
the higher probability of voting the wrong way. This result
can be reversed, however, if votes are cheap enough and
the correlation between others’ votes and the correct
decision is high enough.
Our results can provide some guidance on the efficacy
of proposed regulatory reforms designed to curb or
eliminate the negative effects of empty voting. For
example, Hu and Black (2006, 2007) advocate additional
disclosure requirements as a reasonable starting point. In
the framework of our model, disclosure of an ‘‘empty
voting’’ position on the record date would have no effect,
because we already assume that the market maker
292 A. Brav, R.D. Mathews / Journal of Financial Economics 99 (2011) 289–307
observes the strategic trader’s actions at that stage.
Disclosure of a change in economic position relative to
voting rights between the record and voting dates would
have the effect of reducing or eliminating any trading
profits the strategic trader could otherwise generate. This
would reduce the trader’s willingness to gather informa-
tion and vote. Thus, the efficiency effect of such a rule
would depend case by case on whether the model
predicted a positive or negative effect from the trader’s
presence. Overall, our results imply that regulators should
consider the possibility that curbs to empty voting
behavior could be costly in cases where there is significant
uncertainty about the value of a proposal.
One issue we do not formally analyze is how the
equilibrium would change if the strategic trader entered
the model with an ex ante long or short position in the
stock. The effect of this will be for the trader to have an
initial bias toward protecting firm value (if initially long)
or destroying firm value (if initially short). The trader’s
anticipation of the commitment effect discussed above
will then cause it to choose a record-date position that is
increasing in its initial ownership. If the trader arrives
long, the record-date position is increased and the
commitment effect is strengthened, leading to a higher
likelihood of a positive efficiency effect. However, if the
trader arrives short, the record-date position is decreased,
possibly even to a net short position, implying a higher
probability of a negative efficiency effect.
Another issue that is not explicitly considered in our
model is the possibility that the quality of managers’
project decisions could be positively correlated with
either their ability or the extent of agency problems in
the firm. In other words, bad project proposals may be
more likely when managers have low skill or in firms with
more severe agency problems. In this situation, the
strategic trader’s information about the quality of the
project translates into information about the manager or
firm’s overall quality, which may not be known by the
market. While we do not formally model this possibility, it
has interesting implications since it could result in some
positive efficiency effects from empty voting even when
the strategic trader votes the wrong way. In particular,
when the strategic trader votes in favor of a bad proposal
there will still be an efficiency cost of the proposal being
more likely to be accepted, but there could be a counter-
vailing positive effect of causing the manager’s type to be
revealed more quickly than it otherwise would. This could
increase efficiency if it led to greater price efficiency or an
earlier opportunity for managerial discipline or turnover.
Furthermore, this effect could be intensified if the
strategic trader were able to condition its pre-vote trading
on a good versus bad project (which we currently do not
consider). In such a setting, a correlation between
manager or firm type and project quality is likely to
induce some asymmetry in the strategic trader’s strategy
for good versus bad projects—it is likely to make the
trader go short and vote the wrong way more often with
bad proposals (and thus bad managers) than with good
ones because the fundamental value is lower in those
cases, making a short position even more attractive. We
believe this is an important avenue for further research.
1.1. Related literature
Our model obviously involves a form of stock price
manipulation, but where the manipulation is accom-
plished by affecting the firm’s real operations. Closely
related are Kyle and Vila (1991), Maug (1998), and Kahn
and Winton (1998). In all of these models, a strategic
trader can directly take an action that will affect firm
value, and its ability to trade in a noisy stock market
affects its incentives to do so. The main difference
between our analysis and theirs is that we endogenize
the trader’s ability to affect firm value by modeling the
voting game and deriving the optimal ex ante share and
vote position of the trader.
Another strand of literature studies incentives for
manipulation by traders when managers make invest-
ment decisions partially based on information gleaned
from stock prices. In particular, Khanna and Sonti (2004)
and Khanna and Marietta-Westberg (2005) show that
informed traders may trade against their private informa-
tion if that will send a valuable signal to managers about
investment prospects. On the other hand, Goldstein and
Guembel (2008) show that uninformed traders may take
advantage of feedback effects and sell shares to manip-
ulate prices negatively and generate trading gains. Attari,
Banerjee, and Noe (2006) show that informed investors
may have incentives to dump shares and move prices to
induce shareholder activism. Other models of manipula-
tion involving real activities include Bagnoli and Lipman
(1996) and Vila (1989), both of which study manipulation
involving direct actions such as a takeover bid. Manipula-
tion based on information alone has also been studied
widely, such as by Allen and Gale (1992) and Chakraborty
and Yilmaz (2004).
Many of these papers contribute to a more general
literature on how large shareholders affect corporate
governance. Other papers in this literature tend to focus
either on blockholders who exercise ‘‘voice’’ by directly
intervening in the firm’s activities (Shleifer and Vishny,
1986; Burkart, Gromb, and Panunzi, 1997), or those who
use informed trading, also called ‘‘exit,’’ to improve stock
price efficiency and encourage correct actions by man-
agers (Admati and Pfleiderer, 2009; Edmans, 2009;
Edmans and Manso, forthcoming). Several recent empiri-
cal papers specifically study activism by hedge funds
(Kahan and Rock, 2007; Clifford, 2008; Klein and Zur,
2009; Greenwood and Schor, 2009; Brav, Jiang, Partnoy,
and Thomas, 2008). Such strategies contrast sharply with
the trading strategy we study in this paper, where the
fund optimally hides its information about the value of a
proposal while strategically using it to manipulate real
outcomes and generate trading gains. Zachariadis and
Olaru (2010) study the governance implications of a
similarly subtle strategy, wherein hedge funds may own
both debt and equity in a distressed firm, which will affect
their voting on a reorganization plan.
Our analysis is also closely related to the small but
growing literature on vote buying. For example, Blair,
Golbe, and Gerard (1989) and Neeman and Orosel (2006)
show that allowing a contest for votes in addition to a
contest for shares can have efficiency advantages.
 A. Brav, R.D. Mathews / Journal of Financial Economics 99 (2011) 289–307
Fig. 1. Timeline of the model.
However, they do not model how stock trading interacts
with vote buying.
Other authors have modeled trading and voting
together, but without allowing for ‘‘empty voting.’’ For
example, Maug (1999) models a strategic voting game in
which voting and trading both help aggregate dispersed
information about the value of a proposal. Musto and
Yilmaz (2003) study how the operation of a financial
market affects political voting.
Our study is also related to papers studying the value
of corporate votes. For example, Zwiebel (1995) models
shareholders’ incentives to form blocks and participate in
voting coalitions. Barclay and Holderness (1989) study
block trades and find that there is a significant premium
paid for large minority blocks, which indicates that less
than majority voting control can be valuable.
The paper proceeds as follows. In Section 2 we describe
the model. In Section 3 we derive the equilibrium. In
Section 4 we illustrate the results with several numerical
examples. We discuss implications and conclude in
Section 5.
2. The model
The model focuses on a firm with an upcoming
shareholder vote. The firm has one perfectly divisible
share outstanding. The players, all of whom are risk
neutral, consist of the management of the firm, a strategic
trader (hereafter ‘‘H’’), a market maker, and atomistic
shareholders. Management sets the agenda for the vote,
but does not hold any shares and cannot vote. The market
maker also does not vote. The discount rate is zero. The
timeline of the game is illustrated in Fig. 1.
At the beginning of the game (at time 0), management
proposes an action that can be either good or bad. The
proposal’s ultimate approval status determines firm value,
which is either v or v. In particular, with a good proposal
firm value equals v if the proposal is defeated and
v ¼ v þ Dv if the proposal is approved. With a bad
proposal firm value is v if the proposal is defeated and v
if it is approved.5 We do not model the reason why
management may make a bad proposal. As an example, it
5
The assumption that firm value can take on only two values is
made for tractability. An alternative specification, where firm value
equals one if any proposal is defeated and is increased by Dv if a good
proposal is approved or decreased by Dv if a bad proposal is approved,
yields the same qualitative results but requires additional simplifying
assumptions.
293
could be caused by an agency problem or a lack of ability
or information on management’s part.
All players are initially uninformed about whether the
proposal is good or bad. We assume that H costlessly
becomes informed about whether the proposal is good or
bad at some point on or before the voting date (it makes
no difference whether this occurs before or after the
record date). The market maker and the atomistic share-
holders do not become informed.
At time 0, all shares are held by atomistic shareholders.
After the proposal is announced, but before the record
date (between times 0 and 1), H can submit a market
order to buy or sell shares in the firm. The order is filled by
the market maker at a price equal to the expected value of
the shares (the market maker and H have the same
information at this point, including any ‘‘extra’’ votes H
may acquire, and H’s trade is transparent to the market
maker). For simplicity, we assume that the market maker
holds no inventory at any stage of the model. So, for
example, if it sells shares to H it is immediately able to
purchase shares from atomistic holders at the same price.
This simplifies the analysis because it implies that all
shares will be held by H and the atomistic shareholders on
the record date. The important feature of the assumption
is that ownership of shares by H reduces ownership by
atomistic shareholders, who are the only other parties
allowed to vote, and who also determine later market
liquidity. We denote H’s final economic position on the
record date by aH .
The strategic trader is also able to acquire votes in
addition to those represented by its record-date share
ownership. It can do this at a cost, cðaX Þ, that is increasing
and convex in the number of ‘‘extra’’ votes, aX , as long as
aX r1aH (H cannot obtain more than 100% of the votes).
The cost for extra votes reflects any expense H incurs in
separating its voting interest from its economic owner-
ship. For example, the votes could be purchased on the
share lending market.6 When H approaches a given
atomistic shareholder to borrow its shares, H may have
all of the bargaining power, and thus be able to borrow
the shares at effectively zero cost; the shareholder does
not believe its vote will be pivotal, and thus is willing to
sell the vote for any nominal price. However, since the
6
We focus on buying votes through the share lending market for
expositional simplicity. The analysis is equivalent if the empty voting
position arises because H buys shares to gain votes and then hedges part
of the economic exposure. In that case, the cost of extra votes becomes
the cost of finding counterparties to hedge the economic exposure.
294 A. Brav, R.D. Mathews / Journal of Financial Economics 99 (2011) 289–307
lending market is decentralized H must first find the
shareholder. Our assumption then corresponds to a
convex search cost function (it becomes harder to find
the next shareholder the more you have already located).
For simplicity, we assume that the cost function takes the
form cðaX Þ ¼ max½0,ðaX aX ÞK, where aX Z0 and K is very
large. Thus, extra votes are free up to aX and then
prohibitively expensive beyond that.7 Consistent with
these assumptions, Christoffersen, Geczy, Musto, and
Reed (2007) find that the average vote sells for zero in
the share lending market. Furthermore, Kolasinksi, Reed,
and Ringgenberg (2008) show that the share loan supply
schedule is relatively flat at lower quantity levels but
becomes very steep at higher levels, and that the share
lending market exhibits features consistent with signifi-
cant search frictions.
On the record date (at time 1), voting rights are
determined according to the share and vote positions of H
and the atomistic shareholders. In particular, H’s voting
power is set at max½0, aH  þ aX votes while the atomistic
shareholders retain the remaining 1max½0, aH aX votes.
Next (between times 1 and 2), a random liquidity shock
may hit some of the atomistic shareholders. We assume
that with probability 12 the liquidity shock hits a propor-
tion aZ of the atomistic shareholders who held shares at
time 0. Any shareholders hit by the liquidity shock
immediately place market orders to sell all of their
holdings. If the liquidity shock does not hit, there are no
orders by atomistic shareholders. Since H owns aH shares
at time 1, this means that the total number of shares sold
by atomistic holders if the liquidity shock hits is
aZ ð1max½0, aH Þ.8 Thus, if H buys shares before the record
date, market liquidity is reduced because the pool of
potential liquidity sellers is smaller. If H sells shares before
the record date, liquidity is not affected because the new
atomistic holders are not subject to the liquidity shock.9 H
can also place a market order at this time to buy or sell
whatever quantity it wishes (without first observing
whether atomistic shareholders sell). The market maker
observes only the total net order flow, n, which equals the
sum of all orders by H and the atomistic shareholders, and
sets the price at the expected firm value (given its
information). There are no short sale constraints.
Finally, on the voting date (at time 2), H votes its
max½0, aH  þ aX votes according to its own economic
incentives and information. If at least 12 of the total votes
are cast in favor, the proposal passes. For the
1max½0, aH aX votes held by atomistic stockholders,
we assume that the proportion of ‘‘yes’’ votes cast, denoted
by Y, is distributed on [0,1] according to the distribution
function FG ðÞ with associated density fG ðÞ 2 C2 for a good
7
All of the qualitative results of the paper are unchanged if a
continuous, convex function is assumed for cðaX Þ, as long as the function
is sufficiently ‘‘steep.’’
8
This is equivalent to assuming that the atomistic traders buy
aZ ð1max½0, aH Þ=2 shares with probability 12 and sell aZ ð1max½0, aH Þ=2
shares with probability 12.
9
It seems reasonable to assume that a new shareholder buying just
prior to the record date is less likely to face a liquidity shock in the short
run than are pre-existing, long-term shareholders.
proposal, and FB ðÞ with associated density fB ðÞ 2 C2 for a
bad proposal. We assume different distributions for good
versus bad proposals to allow for the possibility of
correlation between atomistic stockholders’ votes and
the true state. For tractability, we assume that FG ðÞ and
FB ðÞ are always mirror images, that is, that for any
Z 2 ½0, 12, FB ð12 ZÞ ¼ 1FG ð12 þ ZÞ. This implies that if
FG ðÞ ¼ FB ðÞ (there is no correlation with the true state),
then the common distribution must be symmetric around
1
2, and thus the expected ex ante probability of approval if
H has no votes equals 12 in this case. Note that the
atomistic holders’ total number of ‘‘yes’’ votes equals
Yð1max½0, aH aX Þ. After the proposal passes or fails, the
resulting value of the firm is realized and immediately
reflected in the share price.
3. Equilibrium
We solve the model under the following parametric
assumptions:
 Assumption 1: aZ 2 ð0, 12Þ.
 Assumption 2: K is sufficiently large to deter any
aX 4 aX .
 Assumption 3: aX oð2aZ Þ=2ð2 þ aZ Þ.
For the range of aZ allowed by Assumption 1, the
maximum value of aX allowed by Assumption 3 varies
from just below 12 as aZ approaches zero to just above 10 3
as
1 10
aZ approaches 2. As noted previously, we also require
aX r1aH since H cannot have more than 100% voting
power. We derive a perfect Bayesian equilibrium (PBE),
using backward induction where possible.
Behavior at the voting date (time 2) is straightforward.
As noted above, a random proportion Y of the atomistic
shareholders vote in favor of the proposal according to
FG ðÞ or FB ðÞ. H’s vote depends on its economic position in
the shares. Since there is no information asymmetry after
the vote, the shares will trade at their true value of v or v.
Thus, if H is net long in the stock, it maximizes the value
of its stake by voting in the direction that makes the v
outcome more likely, i.e., in favor of a good proposal and
against a bad proposal. If it is net short, it maximizes the
value of its stake by voting in the opposite direction,
making the v outcome more likely.
We now derive the expected value of the firm
depending on whether H is net long or short on the
voting date. First consider a good proposal, and assume H
does not have effective voting control (that is,
max½0, aH  þ aX o 12Þ. If H votes in favor of the proposal,
the probability of acceptance is
 
1
Pr max½0, aH  þ aX þ Yð1max½0, aH aX Þ 4
2
" #
1
max½0, a H a X
¼ Pr Y 4 2
1max½0, aH aX
10
It turns out that this ensures that H will never optimally take full
control of the vote, i.e., will never set aH Z 12 aX .
 A. Brav, R.D. Mathews / Journal of Financial Economics 99 (2011) 289–307
!
1
a a
2 max½0, H  X
¼ 1FG : ð1Þ
1max½0, aH aX
Similarly, if H votes against, the probability of acceptance
is 1FG ð12 =ð1max½0, aH aX ÞÞ. Now note that if H has full
voting control, the probability of acceptance is one if H
votes in favor and zero if H votes against. To economize on
notation, we define g  min½12 =ð1max½0, aH aX Þ,1 and
d  max½ð12 max½0, aH aX Þ=ð1max½0, aH aX Þ,0. Now,
regardless of the value of aH , we can write the probability
of acceptance as 1FG ðdÞ if H votes in favor and 1FG ðgÞ if
H votes against. Expected firm value conditional on a good
proposal and H being net long is then
VL  v þ Dvð1FG ðdÞÞ: ð2Þ
Similarly, expected firm value conditional on a good
proposal and H being net short is
VS  v þ Dvð1FG ðgÞÞ: ð3Þ
With a bad proposal, the probability of approval given
that H is net long and votes against is 1FB ðgÞ, which
implies that the probability of achieving the value v is
FB ðgÞ. Since FG ðÞ and FB ðÞ are mirror images and g ¼ 1d,
we know that
FB ðgÞ ¼ 1FG ðdÞ, ð4Þ
which implies that expected firm value if H is net long
equals VL as derived above whether the proposal is good
or bad. Using a similar argument, it is easy to see that the
expected value if H is net short will always equal VS as
derived above.
Note that this implies that the ‘‘value wedge,’’ VL  VS,
that H can generate by varying the sign of its final position
on the voting date (and therefore its vote) can be
expressed as
VL VS ¼ DvðFG ðgÞFG ðdÞÞ Z 0, ð5Þ
where the inequality follows from g Z d.
We next derive the optimal trading strategy for H
between the record and voting dates (between times 1
and 2) given its economic position in the stock on the
record date, aH , and its quantity of ‘‘extra’’ votes, aX . H’s
continuation payoff at this stage equals the expected
value of its existing position plus any expected traded
profits. This implies that H will never trade a quantity at
this stage that leads to an expected trading loss. Any such
strategy is always dominated by not trading, letting its
stake remain at aH (trades at this date do not affect voting
power anyway), and voting as outlined above to maximize
the value of its stake. Thus, we consider only equilibria
with non-negative trading profits.
We assume H’s trading strategy cannot be conditioned
on a good versus bad proposal. Since the values VL and VS
are symmetric for good versus bad proposals, this
assumption is without loss of generality.11
11
It is easy to show that if H could condition on a good versus bad
proposal at this stage, the equilibrium we derive would still exist, but as
part of a family of equilibria where the total probability of H ending up
net long or short is the same, but the probability conditional on a good
versus bad proposal can be different. This is because of the symmetry of
295
We define H’s (mixed) trading strategy as a probability
distribution, sðjaH , aX , aZ ,v, DvÞ, over all possible trading
quantities t 2 ½1,1, conditioned on the fixed parameters at
this stage. The market maker’s strategy is to set a price
schedule, pðjaH , aX , aZ ,v, DvÞ, defined over all possible order
flows n. Since we are using the PBE concept, the price must
be consistent with the market maker’s belief about which
node in the game tree has been reached at any non-
singleton information sets. In particular, when the market
maker observes a given net order flow n, it has a non-
singleton information set because it is faced with two
possibilities (recall that the market maker observes only the
total net order flow, not individual orders): H could have
placed an order of n while atomistic holders did not trade, or
H could have placed an order for n þ aZ ð1max½0, aH Þ while
atomistic holders sold aZ ð1max½0, aH Þ.
In a PBE, the market maker must have a belief over
those two possibilities for every n and price the trades
accordingly, where the belief is determined by Bayes’ rule
and the equilibrium strategies where possible. Given the
analysis above, the important element of the market
maker’s belief is the probability with which it believes H’s
final position is net long (which would imply a correct
price of VL) versus net short (which would imply a correct
price of VS). Thus, for every possible order flow n we
summarize the market maker’s belief with the function
mðnjaH , aX , aZ ,v, DvÞ, defined as the perceived probability
given an observed net order flow n that H’s unobserved
trading quantity leads to a net short final position. For
example, if both trades that could possibly lead to a net
order flow of n leave H net long given a starting record-
date position of aH , we must have mðnÞ ¼ 0. In other words,
since only trades by H of n or n þ aZ ð1max½0, aH Þ could
lead to a net order flow of n, we have mðnÞ ¼ 0 if both
aH þ n and aH þ n þ aZ ð1max½0, aH Þ are positive. Similarly,
we have mðnÞ ¼ 1 if both trades that could possibly lead to
a net order flow of n leave H net short, that is, if both
aH þ n and aH þn þ aZ ð1max½0, aH Þ are negative. For all n
such that the two possible trades leave H with net
positions of opposite sign, we have mðnÞ 2 ½0,1. In a PBE,
this belief must be determined by the relative probabil-
ities of the two trades in equilibrium if either receives
positive probability in s, and can be assigned arbitrarily if
neither is played with positive probability. Note that we
henceforth drop the conditioning variables from the
functions defined above for notational simplicity.
First, consider possible equilibria with zero expected
trading profits for H. In this case, H’s continuation payoff
simply equals the expected value of its existing stake, so it
will optimally want to maximize the value of that
position. If aH 40, this is accomplished by ensuring that
H’s ‘‘future self’’ votes to maximize firm value (i.e., by
ensuring that the final position is still net long), while if
aH o 0 it is accomplished by ensuring a vote to minimize
firm value (i.e., by ensuring that the final position is still
net short). Thus, H will be indifferent between any trading
(footnote continued)
the situation—H’s expected payoff to buying to remain net long versus
selling to go net short are the same whether the proposal is good or bad.
296 A. Brav, R.D. Mathews / Journal of Financial Economics 99 (2011) 289–307
strategies s that have zero expected trading profits as
long as s gives zero probability to trades that change the
sign of H’s net position.
Next, consider possible equilibria with positive
expected trading profits. For any particular trading
quantity t to have a positive expected trading profit, the
market maker must sometimes be uncertain after obser-
ving one of the equilibrium n that could follow that trade,
i.e., t or taZ ð1max½0, aH Þ, whether H will end up net
long or net short. Otherwise, the price will always be
correct at those n and the expected trading profit will be
zero. Thus, for any t to have a positive expected profit in
equilibrium, H’s strategy s must also put positive
probability on another quantity that leads to the same
net order flow some of the time, and that also leads to a
different sign for H’s final net position.
This implies that we must consider only the universe
of strategies s that put positive probability on both
elements of one or more ‘‘quantity pairs’’ defined by the
fact that the market maker is sometimes unable to
distinguish between cases where atomistic holders sell
while H buys and cases where H sells while atomistic
holders do not trade, and by the fact that the two trades in
the pair lead to different signs for H’s final net position.
For any possible such quantity pair, we define the two
elements of the pair as one quantity that is bought, aB (i.e.,
a trading quantity of aB ), and one quantity that is sold, aS
(i.e., a trading quantity of aS ), where we specify that
aB 4 aS . We later show that aB , aS 4 0 is also required for
a pair to feasibly be part of an equilibrium, so these are
truly a buy and sell trade.
For a given quantity pair, Table 1 gives a matrix of
possible net order flows following trades by H and the
atomistic holders. Since we have specified aB 4 aS , the
only two of these four order flows that could ever be made
equal are aS and aB aZ ð1max½0, aH Þ. Thus, we have the
following constraint on the quantities within any given pair:
aB aZ ð1max½0, aH Þ ¼ aS : ð6Þ
For each possible quantity pair, the buy and sell trades
will result in the same net order flow only if the atomistic
holders sell when H trades aB or do not sell when H trades
aS . We denote the associated ‘‘hidden’’ net order flow,
after which the market maker cannot determine H’s trade,
as naB  aB aZ ð1max½0, aH Þ ¼ aS . Each quantity pair
can be indexed by any of these three quantities, naB , aB , or
aS , which then determines the other two according to (6)
and the definition of naB . In what follows, we generally
index them using naB .
Table 1
Shows possible net order flow combinations depending on the trades of
the strategic trader, H, who either buys aB or sells aS shares, and
atomistic holders of the stock, who either do not trade or sell an
aggregate quantity of aZ ð1max½0, aH Þ.
H Buys aB H Sells aS
Atomistics do not aB aS
trade:
Atomistics sell aB aZ ð1max½0, aH Þ aS aZ ð1max½0, aH Þ
aZ ð1max½0, aH Þ:
For any given quantity pair used in equilibrium, we let
qaB 2 ½0,1 denote the relative probability with which H’s
strategy s has it selling to end up short, so we have
qaB  sðaS Þ=ðsðaS Þ þ sðaB ÞÞ. Since the atomistic holders’
trades occur with equal and independent probability, this
means that for quantity pairs used in equilibrium, the
market maker’s belief at naB must be mðnaB Þ ¼ qaB , leading
to a required price pðnaB Þ ¼ qaB VS þ ð1qaB ÞVL .
Given this, H’s expected continuation payoff for buying
aB shares and voting to maximize firm value is
aB ðVL ð12 VL þ 12ðqaB VS þ ð1qaB ÞVL ÞÞÞþ aH VL ð7Þ
¼ 12aB qaB ðVL VS Þ þ aH VL : ð8Þ
The first term in (7) is H’s expected trading profit. The
expected price at which H buys shares,
ð12 VL þ 12 ðqaB VS þð1qaB ÞVL ÞÞ, reflects the fact that half of
the time, no atomistic holders sell and the net order flow
is aB , leading to a price pðaB Þ ¼ VL . This must be the price
since a net order flow of n ¼ aB could only be reached if H
bought aB or more, which means the market maker must
believe H’s position on the voting date will be net long,
implying mðaB Þ ¼ 0.12 The rest of the time, H’s trade
successfully hides among the noise, and the market maker
sets the price at the ‘‘uninformed’’ expected value,
pðnaB Þ ¼ qaB VS þð1qaB ÞVL . The second term in the equa-
tion reflects the value of H’s existing stake in the firm
given that H will ultimately vote to maximize firm value.
Similarly, H’s expected continuation payoff for selling
and voting to minimize firm value equals
ðaS ÞðVS ð12 VS þ 12ðqaB VS þ ð1qaB ÞVL ÞÞÞþ aH VS ð9Þ
¼ 12ðaZ ð1max½0, aH ÞaB Þð1qaB ÞðVL VS Þ þ aH VS , ð10Þ
where in (10) we use ðaB aZ ð1max½0, aH ÞÞ to replace
aS given the constraint in (6). In this case, note that
the price when atomistic holders also sell is
pðaS aZ ð1max½0, aH ÞÞ ¼ VS since that n could be
reached only by a sale of aS or more and therefore,
mðaS aZ ð1max½0, aH ÞÞ ¼ 1. Eq. (8) must equal (10) in
order for H to be indifferent and willing to put positive
probability on both elements of the pair.
Now, note that for the trading profit portion of (7) to be
positive, it must be the case that aB 4 0, and similarly for
the trading profit portion of (9) to be positive, it must be
the case that aB aZ ð1max½0, aH Þ ¼ aS be negative, i.e.,
that aS 4 0. As noted above, if expected trading profits for
any quantity are negative, H will be better off by not
trading. Thus, aB , aS 4 0 is another necessary condition for
a feasible quantity pair.
We next solve for the qaB that makes H indifferent
between the two quantities in a given quantity pair. For
any given quantity pair indexed by naB to feasibly be part
12
As noted previously, any net order flow n could result from two
possible trades by H, depending on the actions of the atomistic
shareholders. Thus, when a net order flow of n ¼ aB is observed, the
market maker’s beliefs could conceivably place positive weight on two
possible trades by H: aB and aB þ aZ ð1max½0, aH Þ. The latter trade is not
part of the quantity pair under consideration, but we have so far not
ruled out the use of other trades in s, so we note here that the belief will
be the same even if some probability is ascribed to this quantity.
 A. Brav, R.D. Mathews / Journal of Financial Economics 99 (2011) 289–307
of an equilibrium mixed strategy, it must be the case that
the associated qaB 2 ½0,1 makes H indifferent between the
two quantities in the pair. Setting (8) equal to (10) and
solving for qaB yields
1 aB
a a
2 B q ðVL VS Þ þ H VL
¼ 12ð Z ð1max½0, H Þ B Þð1qaB ÞðVL VS Þ þ
a a a aH VS ,
qaB ð12 aB þ 12ðaZ ð1max½0, aH ÞaB Þ
¼ 12ðaZ ð1max½0, aH ÞaB ÞaH ,
aB þ 2aH
qaB ¼ 1 : ð13Þ
aZ ð1max½0, aH Þ
We can also express this in terms of the corresponding
hidden net order flow, naB , by replacing aB according to
the identity aB ¼ naB þ aZ ð1max½0, aH Þ from above. This
yields qaB ¼ 1ðnaB þ aZ ð1max½0, aH Þ þ 2aH Þ=aZ ð1max
½0, aH Þ, or, simplifying,
naB þ 2a H
qaB ¼  : ð14Þ
aZ ð1max½0, aH Þ
This condition must be satisfied by any equilibrium
strategy s that uses the pair indexed by naB .
We now have three conditions that must be met for a
given quantity pair to feasibly be part of an equilibrium
mixed strategy for H: (a) it must be such that aB , aS 4 0
(for both to have any chance of yielding a trading profit);
(b) it must be such that H ends up net long after buying aB
shares and net short after selling aS shares (to induce
uncertainty in expected firm value); and (c) it must be
such that the associated qaB that makes H indifferent is in
the range [0,1] (for H to be indifferent between the two
trades).
Condition (a) implies that aB ¼ naB þ aZ ð1max½0, aH Þ
40 must hold, which translates to naB 4 aZ
ð1max½0, aH Þ. It also implies that aS ¼ naB 40 must
hold. That is, feasible pairs must have naB that fall into the
range ½aZ ð1max½0, aH Þ,0.
Turning to condition (b), for H to end up net long after
buying, it must be the case that aH þ aB ¼ aH þ naB
þ aZ ð1max½0, aH Þ 40, or, rearranging, naB 4aH aZ
ð1max½0, aH Þ. For H to end up net short after selling, it
must be the case that aH aS ¼ aH þnaB o 0, i.e., that
naB o aH . Thus, feasible pairs must also have naB that
fall into the range ½aH aZ ð1max½0, aH Þ,aH . Note that
this range coincides with that from condition (a) when
aH ¼ 0, but then diverges as aH rises or falls.
Finally, consider condition (c). Note from (14) that the qaB
required for indifference within pairs is monotonically
decreasing in naB . Thus, to find a feasible range for quantity
pairs we can set (14) equal to zero and one and solve for naB .
Setting (14) equal to one yields naB ¼ 2aH aZ
ð1max½0, aH Þ, while setting it equal to zero yields
naB ¼ 2aH . Thus, feasible pairs must also have naB that fall
into the range ½2aH aZ ð1max½0, aH Þ,2aH . As with
condition (b), this corresponds to the range given by
condition (a) when aH ¼ 0, but diverges as aH rises or falls
(but faster than the condition (b) range diverges).
For any pair to be feasible, its naB must fall within all
three ranges. Thus, to find the relevant feasible range of
quantity pairs for a given aH , we need consider simply the
297
union of the ranges provided by conditions (a) and (c)
(condition (b) becomes irrelevant since the range pro-
vided by condition (c) changes faster in aH Fi:e:, the union
of the other two ranges is always contained within the
range given by condition (b)). The union of the two ranges
is defined differently depending on the sign of aH . First,
ð11Þ consider aH 4 0. Note that both limits of the range given
by condition (c) are clearly decreasing in aH , so in this
case, the lower limit of the true feasible range will be
ð12Þ determined by condition (a) while the upper limit will be
determined by condition (c), i.e., the range of naB for
feasible pairs must be
naB 2 ½aZ ð1aH Þ,2aH : ð15Þ
By similar logic, the range of naB for feasible pairs given
aH o 0 must be
naB 2 ½2aH aZ ,0: ð16Þ
The range in (15) shrinks as aH rises above zero, so an
equilibrium with positive trading profits is feasible only
up to the point where it collapses to a single point. This is
when the limits are equal, or aZ ð1aH Þ ¼ 2aH
¼)aH ¼ aZ =ð2þ aZ Þ. Similarly, the range in (16) shrinks
as aH falls below zero, so an equilibrium with positive
trading profits is feasible only down to the point where
the limits are equal, or 2aH aZ ¼ 0¼)aH ¼ aZ =2. Thus,
equilibria with positive trading profits are feasible only
for stakes aH 2 ½aZ =2, aZ =ð2 þ aZ Þ. For all other values of
aH , the equilibrium must have H maximizing the value of
its existing stake by either not trading, or always trading
in a way that the sign of its position does not change, so
that its voting incentives are not affected by its trade (see
the proof of Lemma 1 in the Appendix for a verification of
the existence of such equilibria). Note that since we have
assumed aX o ð2aZ Þ=2ð2 þ aZ Þ (Assumption 3), the max-
imum voting power H can have in an equilibrium with
positive trading profits is less than ð2aZ Þ=2ð2þ aZ Þ
þ aZ =ð2 þ aZ Þ ¼ 12.
For aH within the specified range, an equilibrium with
positive trading profits is possible. Now we must
determine which feasible quantity pairs are used by H if
such an equilibrium exists. First, we find which quantity
pair, if used in equilibrium, gives the highest expected
payoff for H. To do this, we must simply maximize either
(8) or (10) with respect to aB using the qaB defined in
(13)—they will have the same payoff since this qaB was
chosen to make them equal. Using (13) in (8) yields
 
1 aB þ2aH
ðVL VS ÞaB 1 þ aH VS : ð17Þ
2 aZ ð1max½0, aH Þ
Taking the derivative of (17) with respect to aB yields
 
1 aB þ 2aH aB
ðVL VS Þ 1  ð18Þ
2 aZ ð1max½0, aH Þ aZ ð1max½0, aH Þ
 
1 2aB þ2aH
¼ ðVL VS Þ 1 : ð19Þ
2 aZ ð1max½0, aH Þ
The second-order condition is clearly satisfied, so the
maximal profit is found by solving the first-order condition
298 A. Brav, R.D. Mathews / Journal of Financial Economics 99 (2011) 289–307

(19)=0 for the optimal aB as follows: (8) where we set aB ¼ aB and qaB ¼ q , or
 
2aB þ 2aH q ð12ðaZ ð1max½0, aH ÞaB Þð1q ÞðVL VS Þ þ aH VS Þ
1 ¼ 0, ð20Þ
aZ ð1max½0, aH Þ þð1q Þð12aB q ðVL VS Þ þ aH VL Þ ð25Þ

2aH 2aB ¼ 12ðVL VS Þq ð1q ÞaZ ð1max½0, aH Þþ aH ðq VS þ ð1q ÞVL Þ:
1 ¼ , ð21Þ
aZ ð1max½0, aH Þ aZ ð1max½0, aH Þ ð26Þ

aZ ð1max½0, aH Þ Using (24) we have

aB ¼ aH : ð22Þ   
2 1 aH 1 aH
q ð1q Þ ¼  þ
The corresponding ‘‘hidden’’ net order flow is then 2 aZ ð1max½0, aH Þ 2 aZ ð1max½0, aH Þ
ð27Þ
aZ ð1max½0, aH Þ
n  naB ¼ aB aZ ð1max½0, aH Þ ¼ aH 

:  
2 aZ ð1max½0, aH Þ2aH aZ ð1max½0, aH Þ þ 2aH
ð23Þ ¼ ð28Þ
2aZ ð1max½0, aH Þ 2aZ ð1max½0, aH Þ
The corresponding probability of H selling aS ¼ aZ ð1max
½0, aH ÞaB shares as opposed to buying aB shares, i.e.,
a2Z ð1max½0, aH Þ2 4a2H
¼ : ð29Þ
q  qaB ¼ sðaS Þ=ðsðaS Þ þ sðaB ÞÞ, is given by (14) using the

4a2Z ð1max½0, aH Þ2
n* from (23), or
aZ ð1max½0, aH Þ Substituting this into the first term of (26), we can
aH  þ2aH 1 aH rewrite (26) as

q ¼ 2 ¼  :
aZ ð1max½0, aH Þ 2 aZ ð1max½0, aH Þ
ðVL VS Þða2Z ð1max½0, aH Þ2 4a2H Þ
ð24Þ þ aH ðq VS þð1q ÞVL Þ:
8aZ ð1max½0, aH Þ
ð30Þ
Now we have the question of when the pair indexed by
The first term represents H’s expected trading profit,
n* is feasible. First, note that n* is clearly within the
while the second is simply the expected value of its stake
feasible range when aH ¼ 0, and it is decreasing in aH . As
given its mixed strategy. For cases where trading profits
aH increases above zero, n* falls, but not as quickly as the
are zero (aH not in the range given by Lemma 1), the
upper limit of the relevant feasible range in (15). Also,
expected payoff is simply aH VL or aH VS depending on the
as aH increases, n* is clearly always greater than the
sign of aH .
lower limit of (15). The upper limit and n* converge
We now proceed to solve for H’s optimal share and
when 2aH ¼ aH aZ ð1aH Þ=2¼)aH ¼ aZ =ð2 þ aZ Þ. As aH
vote trading prior to the record date (between times 0
decreases below zero, n* increases, but not as fast as the
and 1). H’s expected profits viewed from this stage of the
lower limit of (16). As aH approaches aZ =2, the lower
game include its expected future trading profits plus the
limit of (16) approaches zero, as does n*. Thus, whenever
expected value of the stake it acquires today, less the price
any quantity pairs with positive trading profits are
it pays for the stake and the cost of any ‘‘excess’’ votes,
feasible, n* is one of them.
cðaX Þ. From above, H’s expected future trading profits are
In the Appendix, we show that since the pair indexed
ðVL VS Þða2Z ð1max½0, aH Þ2 4a2H Þ=8aZ ð1max½0, aH Þ if aH 2
by n* is always the most profitable feasible pair, it is the
½aZ =2, aZ =ð2 þ aZ Þ, and zero otherwise. Given our
only pair used by H in equilibrium when positive trading
assumption that the shares H buys at this stage are priced
profits are feasible. We also prove the existence of this
at their true expected value, the price of the stake exactly
equilibrium when aH 2 ½aZ =2, aZ =ð2þ aZ Þ, and prove the
offsets its expected value, so H chooses its economic and
existence of the zero trading profit equilibrium otherwise.
voting stakes solely to maximize expected future trading
To do so, we must consider all possible deviations by H
profits less the cost of extra votes. We thus have the
given out of equilibrium prices set by the market maker.
objective function
See the Appendix for the full details, which gives the
!
following result. ðVL VS Þða2Z ð1max½0, aH Þ2 4a2H Þ
max 1aH 2½aZ =2, aZ =ð2 þ aZ Þ
aH , aX 8aZ ð1max½0, aH Þ
Lemma 1. If aH 2 ½aZ =2, aZ =ð2 þ aZ Þ, then in the unique
perfect Bayesian equilibrium of the post-record-date sub- max½0,ðaX aX ÞK, ð31Þ
game, H plays a mixed strategy in which it sells aS shares
where 1aH 2½aZ =2, aZ =ð2 þ aZ Þ is an indicator function equaling
with probability q*, resulting in a net short position on the
one if aH is in the specified range, and VL  VS is given by
voting date, and buys aB shares with probability (1 q*),
(5) above.
resulting in a net long position on the voting date. Otherwise,
The basic tension in H’s stake purchase decision can be
H plays a strategy with zero expected trading profits, and its
seen in this expression. For a given value wedge (VL  VS),
final net position on the voting date has the same sign as aH .
trading profits are maximized by choosing a stake of
We can now characterize H’s expected continuation aH ¼ 0. This is reflected in the term ða2Z ð1max½0, aH Þ2
payoff at this stage of the game. For aH 2 ½aZ = 4a2H Þ=8aZ ð1max½0, aH Þ, which decreases as aH rises or
2, aZ =ð2 þ aZ Þ, H sells with probability q*, leading to a falls from zero. Intuitively, the larger the stake H holds
total expected payoff of q* times (10) plus (1 q*) times (in absolute value) at the later trading stage, the more it
 A. Brav, R.D. Mathews / Journal of Financial Economics 99 (2011) 289–307
will worry at that point about protecting the value of that
stake rather than generating trading gains. Furthermore,
positive economic ownership by H offsets ownership by
atomistic shareholders, and thus reduces market depth
and the potential for profitable trading. On the other hand,
the greater the (positive) stake, the more voting power H
has, so the larger is the value wedge it can generate. This
is reflected in the expression given in (5) for VL  VS, in
which g is increasing in aH and d is decreasing in aH over
the range for which positive trading profits are feasible.
The optimal stake trades off these effects. The optimal
amount of extra votes, aX , affects trading profits only
indirectly, through the increased value wedge. Thus, this
positive effect is weighed against the direct cost
max½0,ðaX aX ÞK.
Analyzing (31) yields the following result.
Proposition 1. The strategic trader’s optimal record-date
share and vote position is characterized by a long economic
position, 0 r aH o aZ =ð2þ aZ Þ, and an optimal number of
‘‘extra’’ votes aX ¼ aX .
The solution for H’s optimal quantity of extra votes,
aX ¼ aX , is trivial given the votes’ usefulness in generating
a value wedge together with the assumed cost function.
The economic stake, aH , is determined by the tradeoff
between voting power and trading gains. Going short at
this stage is never optimal since it confers no votes and
has no effect on liquidity, but reduces future trading gains
by biasing H toward value destruction. A long position
reduces trading gains both by reducing liquidity and by
biasing H toward value creation (the commitment effect).
Indeed, if the stake gets too large (larger than aZ =ð2 þ aZ Þ),
the desire to protect the value of the stake will completely
overcome any incentive to profit by trading (q* goes to
zero). Thus, when choosing an ex ante stake purchased at
its expected value, H will significantly limit the size of the
stake purchase. However, the stake is generally positive
since a long position increases H’s voting power and
enhances its ability to generate a value wedge. Note that
without further assumptions on the other shareholders’
vote distributions, we cannot show that there is a unique
optimal aH , just that it is weakly positive and within the
specified range. Also, the only instances where H does not
buy a strictly positive stake are when aX is sufficiently
large and/or the distribution of Y is sufficiently concen-
trated that the extra votes are essentially worthless.
Now that we have the solution to H’s share and vote
purchase strategy, it remains to determine how H’s
actions will affect efficiency. We measure efficiency by
the probability with which the correct value-maximizing
decision is made. H votes to minimize firm value with
probability q*, so the ex ante probability of the correct
decision is
q ð1FG ðgÞÞ þ ð1q Þð1FG ðdÞÞ:
Here, we again use the mirror image property of FG ðÞ and
FB ðÞ to express the probability in terms of FG ðÞ alone
regardless of the true state. Analyzing (32) yields the
following result.
299
Proposition 2. Whenever FG ðÞ ¼ FB ðÞ, the presence of the
strategic trader (weakly) increases the ex ante probability of
a correct decision.
Thus, despite the fact that H will seek to generate
trading profits by sometimes going net short, voting the
wrong way, and manipulating the firm’s decisions to
decrease value, its presence overall is actually beneficial
to the firm from an ex ante perspective whenever other
shareholders’ votes are uncorrelated with the true state.
(The only times this is not true are in the exceptional
cases discussed above where aH ¼ 0, in which case H’s
presence does not affect efficiency). This is because the
positive economic position H takes in order to increase
voting power and help generate a value wedge causes it to
vote the right way more often than not. Thus, the more
rare cases where H manipulates negatively can be seen as
the ‘‘price to be paid’’ for greater overall voting efficiency
when H’s information is particularly valuable. Note that if
there were no liquidity and thus no possibility of trading
gains ðaZ ¼ 0Þ, H would have no incentive to purchase
shares or votes, or to try to learn the value of the proposal.
It is the possibility of trading gains introduced by noise
trade that induces the information gathering and thereby
increases efficiency.
The key to the unambiguous nature of Proposition 2
is that the ‘‘noise’’ induced by the random votes is
centered around the threshold when FG ðÞ ¼ FB ðÞ. In
other words, the expected proportion of yes votes by
atomistic stockholders is the same as the acceptance
threshold, 12. When that is not the case, it is possible for
H’s presence to reduce overall efficiency. Thus, a correct
interpretation of Proposition 2 is that finding a negative
overall efficiency effect will require that atomistic
shareholders’ votes be correlated with the actual value
of the proposal. We now proceed to investigate these
issues in greater depth. In order to do so, we must
make further assumptions on the distributions FG ðÞ
and FB ðÞ.
For the remainder of this section, we assume that the
underlying probability density functions, fG ðÞ and fB ðÞ, are
linear on [0,1]. Further, we assume that fG ðÞ is (weakly)
positively sloped while fB ðÞ is (weakly) negatively sloped
by the same magnitude (to maintain the distributions’
mirror image quality). Thus, we have fG ðyÞ ¼ 1 þ aðy 12Þ,
where a 2 ½0,2 is the slope parameter, and fB ðyÞ ¼
1aðy 12Þ. When a= 0 the distributions are uniform, and
as a increases the probability of acceptance increases for a
good proposal and decreases for a bad proposal. Using
these properties, we derive the following comparative
statics for the optimal stake size.
Proposition 3. Assume fG ðÞ and fB ðÞ are linear on [0,1].
Then there is a unique optimal stake size aH 4 0. The optimal
stake size is increasing in aZ , decreasing in aX , and unaffected
by Dv and a.
ð32Þ
The greater the market depth ðaZ Þ, the greater is the
potential for trading profits, and the more equity H will
purchase, in general, in order to take advantage. On the
other hand, as votes become easier to purchase directly,
i.e., aX increases, control can be achieved by other means,
300 A. Brav, R.D. Mathews / Journal of Financial Economics 99 (2011) 289–307
so a smaller economic stake is taken to reduce the
commitment effect and increase future trading gains.
The importance of the proposal, Dv, does not affect H’s
stake purchase decision. The magnitude of Dv will
certainly affect the magnitude of trading profits H is
able to generate, but it does not affect the tradeoff
between trading profits and the ability to generate a value
wedge, which determines the optimal stake size. Finally,
the fact that changes in a, which measures the correlation
between other shareholders’ votes and the true state, do
not affect aH 40 is a function of the linear distributional
assumption.
Next we consider how efficiency is affected. First, we
derive results on how the probability of a correct decision
depends on H’s economic and voting positions. We derive
the first result without taking into account the equilibrium
choices of aH and aX (though we do assume that H does
not have full voting control, i.e., aH þ aX o 12, as is always
true in equilibrium).
Proposition 4. Assume fG ðÞ and fB ðÞ are linear on [0,1].
Then the probability of a correct decision is increasing in aH .
When a 40, the probability of a correct decision is increasing
in aX for sufficiently large aH , and decreasing in aX for
sufficiently small aH .
An increase in aH increases both H’s voting power and
the probability that H will vote to maximize firm value,
tending to reinforce a positive efficiency effect. On the
other hand, an increase in aX increases only H’s voting
power, so its effect on overall efficiency depends on H’s
propensity for voting the right way, measured by the
extent of its long position aH . These findings lead directly
to the following important result.
Corollary 1. Assume fG ðÞ and fB ðÞ are linear on [0,1]. Then if
aX is sufficiently low, H’s presence always (weakly) increases
efficiency.
When H’s voting power is closely tied to its economic
stake on the record date, we see that allowing H to ‘‘play
games’’ by sometimes selling to a net short position and
voting to destroy value always increases efficiency. In
other words, allowing for empty voting stakes generated
between the record and voting dates always improves
efficiency as long as the record-date stakes are close
enough to being equal, implying that the commitment
effect is large. Similar to Proposition 2, the negative
outcomes are the price to be paid for inducing H to
participate and contribute its information to the voting
process.
Next we consider how efficiency is affected as other
shareholders’ votes become more correlated with the true
state.
Proposition 5. Assume fG ðÞ and fB ðÞ are linear on [0,1].
Then the probability of a correct decision is increasing in a,
but the rate of increase decreases in both aH and aX .
As expected, as others’ votes become more ‘‘informed,’’
efficiency improves. However, this is attenuated when H
controls more votes. As shown in Proposition 3, changes
in a do not affect aH in equilibrium, so Proposition 5
provides the equilibrium result that efficiency will not rise
as quickly with a when H is present versus when there is
no strategic trader.
Putting Propositions 4 and 5 together with Corollary 1
leads to the following conclusion.
Corollary 2. Assume fG ðÞ and fB ðÞ are linear on [0,1]. Then
H’s presence is more likely to decrease efficiency the larger is
a, and a negative efficiency effect requires a sufficiently large
aX .
Not surprisingly, H is more likely to reduce efficiency
as other shareholders are more likely to arrive at the
correct decision on their own. Furthermore, as noted
above, a negative efficiency effect requires that a
significant empty voting stake be created on the
record date.
Finally, we relax the conditions that ensure H never
optimally takes full control of the vote by looking at the
extreme case where separating votes from ownership at
the record date is effectively unlimited ðaX Z 12Þ. The
equilibrium is as follows.
Proposition 6. Assume aX Z 12 and fG ðÞ and fB ðÞ are linear
on [0,1]. Then H will not trade in the stock prior to the record
date (i.e., aH ¼ 0), but will accumulate sufficient votes to
determine the election outcome (i.e., aX Z 12). The probability
of H selling short and voting to minimize firm value will be 12
in equilibrium, and H’s presence will not affect ex ante
efficiency if FG ðÞ ¼ FB ðÞ, but will decrease efficiency if
FG ðÞaFB ðÞ.
This result reflects the fact that H’s trading gains are
maximized when its stake is zero. Since buying enough
votes to swing the election maximizes the value wedge,
VL VS, there is no longer any reason for H to take a
position in the stock on the record date. It maximizes its
trading profits by buying/selling with equal probability. It
is interesting to note that this is an extreme version of the
model in which H’s ability to manipulate firm value is
maximized, yet the overall effect is neutral for firm value
in the absence of correlated voting by others.
4. Numerical examples
We now provide some numerical examples to illus-
trate the above results. Continuing with the assumption of
linear probability density functions, we consider the cases
a = 0 (uniform), a =1 (the probability of a correct decision
in the absence of H is 58), and a =2 (the probability of a
correct decision in the absence of H is 34). We also set
aZ ¼ 0:2.
We first consider how H’s optimal record-date eco-
nomic ownership, aH , varies as votes become easier to
acquire in the lending market, i.e., as aX increases. Fig. 2
graphs this relationship. As noted in Proposition 3, aH does
not depend on a so this graph has a single line.
As indicated in Proposition 3, the extent of H’s long
position on the record date is clearly decreasing in the
number of extra votes it can buy in the share lending
market. In this case, it falls from a maximum of about 25%
 A. Brav, R.D. Mathews / Journal of Financial Economics 99 (2011) 289–307
Fig. 2. Shows the strategic trader’s optimal record-date holding, aH , as a
function of the quantity of votes available in the lending market, a X .
Fig. 3. Shows the incremental probability of a correct decision as a result
of the strategic trader’s presence as a function of the quantity of votes
available in the lending market, a X , for varying levels of correlation
between atomistic holders’ votes and the correct decision, indexed by a
(larger a implies a greater positive correlation).
of aZ to about 8% of aZ as aX approaches its maxi-
mum value.
Next we consider how the ease of purchasing extra
votes affects efficiency. Fig. 3 graphs the relationship
between the number of extra votes available, aX , and the
difference in the probability of a correct decision caused
by H’s presence. In particular, the graph plots the
probability when H plays its optimal strategy minus
the probability when H is not present in the model. Here,
the results depend critically on the correlation between
others’ votes and the true state.
When atomistic holders’ votes are uniformly distributed
(a=0), making more votes available to H always improves
efficiency as long as aX o 12. This occurs despite the fact
that extra votes result in a smaller economic stake (as in
Proposition 3 and Fig. 2), and therefore a smaller commit-
ment effect and a greater probability that H will go net
short and vote to minimize firm value. The offsetting force
is the greater voting power H has in equilibrium despite his
reduced economic stake; i.e., the direct increase in voting
power via aX is greater in equilibrium than the decrease in
aH . This direct positive effect (with H still voting the right
301
way more often than not) outweighs the negative effect of
a greater probability of going short.
With the correlated distributions (a =1, 2), H’s presence
still improves efficiency when aX is sufficiently low, as
shown in Corollary 1. However, as indicated by Proposi-
tion 4, the extra votes at some point start to reduce
efficiency as aH is reduced and H’s increased voting power
together with the reduced commitment effect works
against the other shareholders, who are now fairly likely
to arrive at the correct decision on their own. Ultimately,
H’s presence reduces efficiency overall when aX is
sufficiently large, as indicated by Corollary 2.
5. Implications and conclusion
Our model is stylized, but it nevertheless provides a
number of useful empirical implications. Most obviously,
it implies that empty voting behavior should result in
both positive and negative outcomes from an efficiency
perspective. Thus, the negative anecdotes described by Hu
and Black (2006, 2007) should be only one side of the
story. However, it is likely that direct, large-scale evidence
of favorable (or unfavorable) empty voting behavior
would be difficult or impossible to gather.
Our model can also guide less direct empirical
investigations. For instance, with slight adjustments to
the model, we could predict which types of firms and
proposals are more likely to be targeted by strategic
empty voters. In particular, if we add an information cost
that the strategic trader must pay after the record date to
become informed about the value of the proposal, the
model would predict that such traders are likely to target
firms where the strategy is most profitable. This would
tend to be firms with:
(a) high liquidity (high aZ ), which could be measured by
trading volume or a statistic summarizing the disper-
sion of share ownership;
(b) potentially important pending proposals (high Dv),
which could be measured either directly by looking at
types of proposals, or indirectly using a proxy for the
quality of corporate governance;
(c) greater availability of ‘‘extra’’ votes (high aX ), which
could be measured by volume and specialness in the
lending market, dispersion of ownership, or the avail-
ability of derivatives to hedge economic exposure; and
(d) voting outcomes that are uncertain (i.e., the vote is fairly
likely to go either way) but where relatively few votes are
needed to swing the result (which would correspond to a
tighter distribution of non-strategic votes).
This last point could again be related to the quality of
corporate governance (outcomes are likely to be less
certain when bad proposals are more likely). Another
example could be a vote with a supermajority rule where
a high percentage is needed for approval of the proposal.
The distribution of share ownership between institutions
and individuals could also be important if different types
of shareholders have different information or voting
incentives.
302 A. Brav, R.D. Mathews / Journal of Financial Economics 99 (2011) 289–307
Our model would therefore predict that more empty
voting activity should occur in these types of settings.
Direct evidence could be sought by looking at share
lending volumes around record dates and/or share trading
by certain types of investors (such as hedge funds)
before and after record dates. The comparative statics in
Proposition 3 could also be tested in this context.
Furthermore, the model provides predictions about
when the actual decisions are likely to be more efficient
when empty voting occurs. Thus, an indirect test of the
model could focus on ex post measures of how voting
outcomes affect firm value.
As noted in the Introduction, our results may also
provide some guidance on the efficacy of proposed
regulatory reforms designed to curb or eliminate the
negative effects of empty voting. For example, Hu and
Black (2006, 2007) advocate additional disclosure require-
ments as a reasonable starting point. In the framework of
our model, disclosure of an empty voting position on the
record date would have no effect, because we already
assume that the market maker observes the strategic
trader’s actions at that stage. The effect of a rule requiring
disclosure of a change in economic position relative to
voting rights between the record and voting dates
depends on how the rule is implemented. If the rule
made it more difficult for the trader to hide its trades from
the market maker, this would have the effect of reducing
or eliminating any trading profits the strategic trader
could otherwise generate. Thus, the rule could reduce
efficiency if it causes the trader not to accumulate votes in
the first place (which is likely if there is a cost to gathering
information about the quality of the proposal) in cases
where the trader’s presence improves efficiency. As noted
previously, such cases are likely when either separating
shares from votes is not very expensive or other share-
holders’ votes are not too highly correlated with the
correct decision. Otherwise, such a rule is likely to
improve efficiency.
Extending the model to allow for initial holdings by the
strategic trader would lead to some additional implica-
tions (as noted in the Introduction). In particular, if the
strategic trader were an existing blockholder with a long
position, it would have an increased incentive to protect
the value of its initial stake, meaning that its record-date
stake would be increasing in the size of its initial position.
Thus, the commitment effect would be intensified and the
trader’s actions would be more likely to increase effi-
ciency. Indeed, if initial ownership were large enough, the
trader may actually find it optimal to forgo trading gains,
instead buying enough shares and/or votes to ensure that
the vote outcome always maximizes firm value.13 On the
other hand, if the trader arrived with an initial short
position, its record-date position would be reduced
(potentially to a net short position) due to the existing
incentive to decrease firm value, and the trader’s presence
would be more likely to decrease efficiency overall.
13
A full analysis of the case of arrival with a long position is
available from the authors upon request.
While our model provides a coherent framework for
addressing the efficiency consequences of empty voting,
there are a number of issues that remain unexplored. First
of all, it would be interesting to allow for some correlation
between managers’ project selection and their ability. As
noted in the Introduction, if project choices are correlated
with managerial ability or agency problems, empty voting
could have additional efficiency implications. Further-
more, it would be interesting to study how the results
would change if there were multiple strategic traders who
compete to generate trading gains. It would also be
interesting to study the interaction between a strategic
trader with no initial interest in the stock (such as in our
model) and an existing large shareholder who could also
act to influence the vote outcome. Finally, we would like
to more closely investigate specific mechanisms by which
shares and votes can be separated, such as the share
lending market. For example, if the uninformed share-
holders were not all atomistic, how would they analyze
the decision of whether to lend their shares on the record
date? How would this affect pricing in the lending
market? Our framework should provide a platform for
exploring these issues in the future.
Appendix
Proof of Lemma 1. First, we show that if n* is feasible
and aB and aS receive positive probability in H’s mixed
strategy s, then no other quantities can receive positive
probability. From the text, any feasible quantity pair that
receives positive probability in equilibrium must have
pðnaB Þ ¼ qaB VS þð1qaB ÞVL at its associated hidden order
flow, where qaB is determined by (14), while pðaB Þ ¼ VL
and PðaS aZ ð1max½0, aH ÞÞ ¼ VS . But at these prices, it
has been shown that trading profits are superior for the
quantities in the n* pair, so H cannot be indifferent
between the quantities in that pair and any quantities in
any other feasible pair. Furthermore, no quantity with
zero or lower expected trading profits can receive mixing
weight in an equilibrium where the n* pair is used. Any
such quantity will be dominated by one of the n*
quantities (either aB or aS will have the same value for
the existing stake as the no trading profit quantity, but
will also have positive expected trading profits).
Next, we show there cannot be an equilibrium with
positive trading profits when aH 2 ½aZ =2, aZ =ð2 þ aZ Þ that
places zero weight on the quantities in n*. First, note that
even if the order flow is not reached in equilibrium,
pðaB Þ ¼ VL must still hold since the market maker cannot
believe H is short with any probability, and similarly
pðaS aZ ð1max½0, aH ÞÞ ¼ VS . Clearly, then, if the off-
equilibrium price p(n*) were expected to be q*VS +(1 q*)VL,
then the n* quantities would always offer higher expected
profits than the quantities in any other feasible pair used in
the proposed equilibrium. Next, note that if p(n*) is expected
to be higher than q*VS +(1 q*)VL, a trade of aS will look
even more profitable, while if p(n*) is expected to be lower
than that, a trade of aB will look even more profitable. Thus,
no p(n*) exists that can prevent a deviation to one of the n*
 A. Brav, R.D. Mathews / Journal of Financial Economics 99 (2011) 289–307
quantities from the quantities in any feasible pair in any
proposed equilibrium.
Finally, to prove that the equilibrium given in Lemma 1
is unique if it exists, we must show that a no trading
profits equilibrium cannot exist when aH 2 ½aZ =2, aZ
=ð2þ aZ Þ. As noted in the text, any no trading profits
equilibrium must leave the sign of H’s stake unchanged
after all trades with positive probability in s. First, assume
aH 4 0. Now note p(n*) =VL is the only possible off-
equilibrium price that could prevent a deviation to aB
from any other trade that maintains a long position
(otherwise both trades would have a stake value of aH VL ,
and aB would additionally have positive expected trading
profits). Next, note that at this price, deviating to aS
yields expected trading profits of ðaS ÞðVS ð12 VS þ 12 VL ÞÞ
¼ 12 ðaZ ð1aH Þ=2 þ aH ÞðVL VS Þ while the cost of decreasing
the value of the existing stake is aH ðVL VS Þ. The former
exceeds the latter for all aH o aZ =ð2 þ aZ Þ, so the deviation
is profitable. Similarly, if aH o 0, a price p(n*)= VS is
required to prevent deviation to aS from any other
trade that maintains a short position, but this price makes
deviating to aB profitable.
Next we prove the existence of the equilibria given in
the result by characterizing the prices set by the market
maker on and off the equilibrium path and checking for
deviations.
First, consider the case with feasible positive trading
profits, aH 2 ½aZ =2, aZ =ð2þ aZ Þ. From the text, the equili-
brium has prices p(n*)= q*VS + (1 q*)VL, pðaB Þ ¼ VL , and
pðaS aZ ð1max½0, aH ÞÞ ¼ VS . These give the market
maker zero expected profit in equilibrium by construc-
tion, which is the market maker’s objective function, so
deviations by the market maker along the equilibrium
path are not a concern. Since the PBE concept puts no
constraints on off-equilibrium path beliefs, we also do not
have to worry about deviations by the market maker at
those nodes as long as the prices we specify are consistent
with the beliefs we specify.
We now rule out deviations by H to different trading
quantities. In order to test for such deviations, we must
specify the market maker’s beliefs and thus prices that
will be set at out-of-equilibrium net order flows. For all
feasible ‘‘hidden’’ order flows naB as defined in the text, we
assume mðnaB Þ ¼ qaB where qaB is as in (14). Thus,
pðnaB Þ ¼ qaB VS þ ð1qaB ÞVL . For all order flows n above the
range given by (15) if aH 4 0 or (16) if aH o0, we assume
mðnÞ ¼ 0, so that p(n)= VL, and for all those below the
relevant range, we assume mðnÞ ¼ 1 and thus, p(n)= VS. This
implies that for any feasible quantity pair, the price will
be VL if H buys aB and atomistic holders do not trade,
while the price will be VS if H sells aS and atomistic
holders also sell. Thus, the analysis in the text showing
that the quantities in the pair indexed by n* have superior
profits implies that H will not deviate to any quantity in
such an alternative feasible quantity pair.
303
Next we rule out deviations to any quantity not in a
feasible quantity pair. First, assume aH o0. The analysis in
the text shows that the upper end of the feasible range has
naB ¼ 0, and that for all pairs with higher naB , aS must be
negative, i.e., no sale trade can lead to a higher net order
flow. Finally, note that at the price specified above for
these order flows, p(n)= VL, H cannot expect a trading
profit, so such deviations have an expected payoff of aH VS
or aH VL or less, all of which are dominated by the
expected payoff to one of the quantities in the pair
indexed by n* given their positive expected trading
profits. The lower end of the feasible range with aH o 0
has naB ¼ 2aH aZ , at which point we have qaB ¼ 1 and
the price equals VS, since at this point H has no more
incentive to buy the corresponding quantity in the pair,
only to sell at the equilibrium prices. Now, consider
quantity pairs with hidden net order flows in the range
½aZ ,2aH aZ , where qaB 2 ½0,1 does not hold (condition
(c) fails) but aB , aS 4 0 still holds (condition (a) is satisfied).
As specified above, we keep the price at those order flows
at VS. To get to those lower order flows, H must either sell
more or buy less than the quantities in the pair with
naB ¼ 2aH aZ , but at the same expected price. This
leaves the expected sell profit the same (since VS is the
correct price from H’s perspective) while decreasing the
expected buy profit. Thus, since the quantities in the pair
indexed by naB ¼ 2aH aZ do not have higher expected
profit than the quantities indexed by n* (from above),
those indexed by these lower naB also cannot. Finally, for
quantity pairs with naB oaZ , the discussion in the text
shows that only sell orders can result in those order flows,
so a price of VS as specified prevents trading profits.
Now, assume aH 40. The analysis in the text shows that
the lower end of the feasible range has naB ¼ aZ ð1aH Þ, at
which point aB ¼ 0. Thus, any trade that leads to a lower
naB must be a sell trade, and H cannot profit given the
price of VS (as specified above). The upper end of the
feasible range has naB ¼ 2aH , at which point we have
qaB ¼ 0 and the price is VL, since at this point H has no
incentive to sell the corresponding quantity in the pair,
only to buy at the equilibrium prices. Now, consider
quantity pairs with naB 2 ½2aH ,0, where qaB 2 ½0,1 does
not hold but aB , aS 4 0 still holds. As specified above, we
keep the price at those order flows at VL. To get to these
higher order flows, H must either buy more or sell less
than the quantities in the pair indexed by naB ¼ 2aH , but
at the same expected price. This leaves the expected buy
profit the same (since VL is the correct price from H’s
perspective) while decreasing the expected sell profit.
Thus, since the quantities in the pair indexed by
naB ¼ 2aH do not have higher expected profit than the
quantities indexed by n* (from above), those indexed by
these higher n also cannot. Finally, for quantity pairs with
naB 4 0, the discussion in the text shows that only buy
orders can result in those order flows, so a price of VL as
304 A. Brav, R.D. Mathews / Journal of Financial Economics 99 (2011) 289–307

specified prevents any trading gains. This suffices to prove aH 4 aZ =ð2 þ aZ Þ (note that aZ ð1aH Þ ¼ 2aZ =ð2 þ aZ Þ when
the existence of the positive trading profit equilibrium aH ¼ aZ =ð2 þ aZ Þ, and is decreasing in aH ). Thus, as above, H
described in the result. is better off trading any quantity with zero expected trading
Now, consider the no trading profit cases aH o aZ =2 profits that keeps the sign of its stake constant. The pricing
and aH 4 aZ =ð2 þ aZ Þ, where H’s equilibrium strategy s scheme outlined above is therefore optimal for the market
places weight only on quantities that do not change the maker (again, there is no uncertainty about how H will vote
sign of H’s position (and where all such s are equally valid in equilibrium, and all trades t that do not change the sign of
equilibrium strategies). Take the case aH oaZ =2 first. We H’s position and can therefore have positive weight in s lead
specify beliefs (both on and off the equilibrium path) as to a price of VL, which is correct). &
follows: for n 4 0, mðnÞ ¼ 0 so p(n)= VL, and for n r0,
mðnÞ ¼ 1 so p(n) =VS. For n= 0 to occur with any probability, Proof of Proposition 1. Taking the derivative of VL  VS
H either must buy aZ shares or not trade. Thus, to reach with respect to aX using (5) yields
any higher order flow, it must always buy, and a price of     
@g @d
VL prevents any trading profits. To reach any lower order Dv fG ðgÞ fG ðdÞ : ð33Þ
@aX @aX
flow, it must either sell some quantity, in which case the
price of VS eliminates trading profits, or buy some Note that
quantity. A buy at a price of VS has positive expected
@g ð1Þ 12 1
profit if the buy quantity exceeds aH , so that H’s final ¼ ¼ ð34Þ
@aX ð1max½0, aH aX Þ2 2ð1max½0, aH aX Þ2
stake is positive and expected firm value is VL. The largest
possible expected trading profit occurs at the largest and
possible buy that sometimes leads to an order flow  
weakly less than zero, i.e., a buy of aZ . The expected @d ð1Þð1max½0, aH aX Þð1Þ 12 max½0, aH aX
¼
trading profit to that trade is 12 aZ ðVL VS ÞFhalf of the time @aX ð1max½0, aH aX Þ 2

atomistic holders will not trade and the trade will be 1

¼ ð35Þ
priced at VL. However, since a buy with positive trading 2ð1max½0, aH aX Þ2
profits changes the sign of H’s final position, H takes an
when aH o 12 aX , which is always true over the range of
expected loss on its existing stake of aH ðVS VL Þ, which
feasible positive trading profits. Using these in (33), (33)
loss is larger than the above trading gain since becomes
aH o aZ =2. Thus, H is better off trading any quantity !
with zero expected trading profits that keeps the sign of 1
Dv ðfG ð gÞ þ fG ð dÞÞ 4 0: ð36Þ
its stake constant. The pricing scheme outlined above is 2ð1max½0, aH aX Þ2
therefore optimal for the market maker (there is no The result for aX is then immediate conditional on
uncertainty about how H will vote in equilibrium, and all aX o1aH , which is shown to hold in equilibrium below.
trades t that do not change the sign of H’s position and can Next, note that when aH o 0,
therefore have positive weight in s lead to a price of VS, ! !!
1 1
2 aX
which is correct). VL VS ¼ Dv FG FG 2 ð37Þ
1aX 1aX
Finally, take the case aH 4 aZ =ð2 þ aZ Þ. We specify mðnÞ ¼ 0
so p(n)=VL for all n ZaZ ð1aH Þ, and mðnÞ ¼ 1 so p(n)= VS does not vary with aH . Given this, it is obvious that (31) is
for all lower order flows. For n ¼ aZ ð1aH Þ to occur with increasing in aH for all aH o0, so aH Z 0 follows.
any probability, H must either not trade or sell aZ ð1aH Þ Now, consider aH Z 0. Given that the second term in the
shares. Thus, to reach any lower order flow, it must always objective function (31) equals zero in equilibrium (since
sell, and a price of VS prevents any trading profits. To reach aX ¼ aX ), the entire objective function equals zero for
any higher order flow, it must either buy some quantity, in aH 4 aZ =ð2 þ aZ Þ. However, it is positive at aH ¼ 0 (it equals
which case the price of VL eliminates trading profits, or sell ðVL VS ÞaZ =8), which proves aH o aZ =ð2 þ aZ Þ. Taking the
some quantity. A sell at a price of VL has positive expected first derivative of (31) with respect to aH assuming aH 2
profit if the sale quantity exceeds aH , so that H ends up ½0, aZ =ð2 þ aZ Þ and aX ¼ aX yields (noting that
short and the expected firm value is VS. The most profitable VL VS ¼ DvðFG ðgÞFG ðdÞÞ and that @g=@aH ¼ @g=@aX and
possible such sale occurs at the largest sale quantity that can @d=@aH ¼ @d=@aX , which were calculated above for the
lead to an order flow weakly above aZ ð1aH Þ, i.e., a sale of assumed range of aH Þ:
aZ ð1aH Þ. The expected trading profit to such a sale is !  
a2Z ð1aH Þ2 4a2H @g @d
1 Dv fG ðgÞ fG ðdÞ þ DvðFG ðgÞFG ðdÞÞ
2 aZ ð1aH ÞðVL VS ÞFhalf of the time atomistic holders will 8aZ ð1aH Þ @aH @aH
also sell and the price will be VS. However, since a sell with !
positive trading profits changes the sign of H’s position, H ða2Z 2ð1aH Þ8aH Þ8aZ ð1aH Þð8aZ Þða2Z ð1aH Þ2 4a2H Þ
takes an expected loss on its existing stake of aH ðVL VS Þ, 64a2Z ð1aH Þ2
which loss is larger than the above trading gain since ð38Þ
 A. Brav, R.D. Mathews / Journal of Financial Economics 99 (2011) 289–307 305

! !
DvðfG ðdÞ þfG ðgÞÞ a2Z ð1aH Þ2 4a2H aZ 1aX ð1aX Þ
¼ þ DvðFG ðgÞ ¼ Dv 40: ð45Þ
2ð1aH aX Þ2 8aZ ð1aH Þ 8 ð1aX Þ2
!
aH ðaH 2Þ aZ To prove there is a unique maximum, it suffices to show
FG ðdÞÞ  : ð39Þ
2aZ ðaH 1Þ2 8 that (31) is concave in the relevant range,
Finally, we show that this is positive when aH ¼ aX ¼ 0, aH 2 ½0, aZ =ð2 þ aZ Þ. Taking the derivative of (43) with
implying that aH will be positive at least for sufficiently respect to aH yields
! ! !
small aX . First, note that g ¼ d ¼ 12 when aH ¼ aX ¼ 0, so a2Z ð1aH Þ2 4a2H 2 Dv aH ðaH 2Þ aZ
Dv þ 
FG ðgÞFG ðdÞ ¼ 0 and (39) reduces to 8aZ ð1aH Þ ð1aH aX Þ 3
ð1aH aX Þ 2
2aZ ðaH 1Þ2 8
  ! !  
aZ 1 aH ðaH 2Þ aZ ð1aH aX Þ þ ðaH þ aX Þ aH þ aX
Dvf G 4 0: & ð40Þ þ Dv  þ Dv
1aH aX
8 2 2aZ ðaH 1Þ2 8 ð1aH aX Þ2
!
ð2aH 2Þ2aZ ðaH 1Þ2 4aZ ðaH 1ÞaH ðaH 2Þ
2ð 4
ð46Þ
Proof of Proposition 2. If FG ðÞ ¼ FB ðÞ, then, as noted in 4a a Z H 1Þ

the text, FG ðÞ must be symmetric around 12. This implies

4aH ðð1aX Þð6aH 3ð1aX ÞÞa2H ð3aX ÞÞ þ aX ða2Z ð1aH Þ3 4ðaX 1Þ2 Þ
both that the probability of approval without H is 12 and ¼ Dv o 0:
4ðaH 1Þ3 ðaH þ aX 1Þ3 aZ
that FG ðdÞ ¼ 1FG ðgÞ. Thus, if aH ¼ 0, from (24) the optimal
ð47Þ
mixing probability is q ¼ 12, and the probability of a
correct decision becomes 12 FG ðdÞ þ 12 ð1FG ðdÞÞ ¼ 12, so H The sign is derived as follows. The denominator is clearly
does not affect efficiency regardless of aX . Thus, it suffices positive given aH þ aX o 12. The term 6aH 3ð1aX Þ is
to show that (32) is increasing in aH for aH 2 maximized when aH and aX are large, and with aX ¼ 12 it
½0, aZ =ð2 þ aZ Þ and any feasible aX . Taking the derivative is negative for all aH 2 ½0, 15, which is the maximum
of (32) with respect to aH yields
allowable range. The other terms in the numerator are
    clearly negative.
@q @g @d
ð1FG ðgÞÞ þ q fG ðgÞ þ ð1q Þ fG ðdÞ
@aH @aH @aH Next, note that VL  VS is independent of a (see the
  expression derived above), so aH will also be independent
@q
þð1FG ðdÞÞ  : ð41Þ of a. This proves the last comparative static result.
@aH
For the first two comparative static results, it suffices to
Now, note that @q =@aH ¼ ðaZ ð1aH ÞðaZ ÞaH Þ=a2Z
sign the appropriate cross-partial derivative. Taking the
ð1aH Þ2 ¼ 1=aZ ð1aH Þ2 . Also noting that fG ðgÞ ¼ fG ðdÞ
derivative of (43) with respect to aZ yields the cross-
when FG ðÞ is symmetric around 12 and using the expres-
sions derived in the proof of Proposition 1 above for partial
!
@d=@aH and @g=@aH , (41) simplifies easily to Dv 2aZ ð1aH Þ2 8aZ ð1aH Þ8ð1aH Þða2Z ð1aH Þ2 4a2H Þ
FG ðgÞFG ðdÞ fG ðgÞð12q Þ ð1aH aX Þ2 64a2Z ð1aH Þ2
2
þ 2
40, ð42Þ !
aZ ðaH 1Þ 2ð1aH aX Þ  
aH þ aX 2ðaH 1Þ2 aH ðaH 2Þ 1
þ Dv  ð48Þ
where the inequality follows from q o 12. & 1aH aX 4a2Z ðaH 1Þ4 8

" !
Proof of Proposition 3. First, note that given fG ðyÞ ¼ 1 Dv 4a2H
Ry ¼ 1aH þ
1 þaðy 12Þ, we have FG ðyÞ ¼ 0 ð1 þ aðw 12ÞÞ dw ¼ ðð2a 8 ð1aH aX Þ2 a2Z ð1aH Þ
  !#
ð1yÞÞ=2Þy. Thus, FG ðgÞFG ðdÞ ¼ FG ðgÞFG ð1gÞ ¼ ðð2a aH þ aX 4aH ð2aH Þ
ð1gÞÞ=2Þgðð2að1ð1gÞÞÞ=2Þð1gÞ ¼ 2g1. Using the þ Dv 1 þ 2 4 0: ð49Þ
1aH aX aZ ð1aH Þ2
definition of g and noting from the text that aH þ aX o 12
in equilibrium, this implies VL VS ¼ Dvð2g1Þ The inequality is derived as follows. First, note that
¼ DvðaH þ aX Þ=ð1aH aX Þ. Also note that since d ¼ 1g, Dv=ð1aH aX Þ2 Z4DvððaH þ aX Þ=ð1aH aX ÞÞ given aH þ
we have fG ðgÞ þ fG ðdÞ ¼ 1 þ aðg 12Þ þ1 þ aðð1gÞ 12Þ ¼ 2. aX o 12. The result then follows from the fact that
With these, (39) becomes
ð1 þ 4aH ð2aH Þ=a2Z ð1aH Þ2 Þ Z1 clearly always holds,
!
while ð1aH þ 4a2H =a2Z ð1aH ÞÞ Z 45 must hold given that
Dv a2Z ð1aH Þ2 4a2H
8aZ ð1aH Þ aH o 15 in equilibrium according to Proposition 1 given our
ð1aH aX Þ2
! assumption that aZ o 12 (Assumption 1).
 
aH þ aX aH ðaH 2Þ aZ Next, taking the derivative of (43) with respect to aX
þ Dv  : ð43Þ
1aH aX 2aZ ðaH 1Þ2 8 yields the cross-partial
! !
a2 ð1aH Þ2 4a2H 2
Dv Z
Now, to prove aH 40, we evaluate this setting aH ¼ 0. 8aZ ð1aH Þ ð1aH aX Þ3
This gives ! !
aH ðaH 2Þ aZ ð1aH aX Þ þ ðaH þ aX Þ
  þ Dv 
Dv aZ aX  aZ 2aZ ðaH 1Þ2 8 ð1aH aX Þ2
þ D v  ð44Þ
ð1aX Þ2 8 1aX 8 ð50Þ
306 A. Brav, R.D. Mathews / Journal of Financial Economics 99 (2011) 289–307

" !
Dv 2 a2Z ð1aH Þ2 4a2H Taking the derivative with respect to a yields
¼    
ð1aH aX Þ 2 ð1aH aX Þ 8aZ ð1aH Þ gð1gÞ dð1dÞ
!# q þ ð1q Þ : ð57Þ
aH ðaH 2Þ aZ 2 2
þ  o 0: ð51Þ
2aZ ðaH 1Þ2 8 Note that gð1gÞ=2 ¼ dð1dÞ=2 since d ¼ 1g, so, noting
that we always have aH þ aX o 12 in equilibrium, (57) equals
The inequality is derived as follows. Consider the terms in ! !
1 1
the square brackets. The last term is clearly negative, 2 2 aH aX
while the first is positive. Now, note that in the first-order gð1gÞ 1aH aX 1aH aX 12ðaH þ aX Þ
¼ ¼ 4 0:
condition [(43) =0], ðaH ðaH 2Þ=2aZ ðaH 1Þ2 aZ =8Þ multi- 2 2 8ð1aH aX Þ2
plies DvððaH þ aX Þ=ð1aH aX ÞÞ, while ðða2Z ð1aH Þ2 4a2H Þ ð58Þ
=8aZ ð1aH ÞÞ multiplies Dv=ð1aH aX Þ2 . As above, we
Finally, taking the derivative of this with respect to aH or
have Dv=ð1aH aX Þ2 Z 4DvððaH þ aX Þ=ð1aH aX ÞÞ, so,
aX yields
from the first order condition we must have ðaH ðaH 
2Þ=2aZ ðaH 1Þ2 aZ =8Þ Z 4ðða2Z ð1aH Þ2 4a2H Þ=8aZ ð1aH ÞÞ 2ð8ð1aH aX Þ2 Þ16ð1aH aX Þð1Þð12aH 2aX Þ
in equilibrium. This proves the result since 64ð1aH aX Þ4
2=ð1aH aX Þ r 4. aH þ aX
¼ o0: & ð59Þ
For the third comparative static result, note that when 4ð1aH aX Þ3
setting (39) equal to zero and solving for aH , Dv will drop
out of the solution. &
Proof of Proposition 6. With aX Z 12, votes are free up to
Proof of Proposition 4. The derivative of (32) with the point where H takes complete control of the vote. It is
respect to aH is given by (41) above. Evaluating easy to see that the value wedge H can create, VL  VS, is
this with linear density functions (and noting from above maximized when H can single-handedly determine the
that @q =@aH ¼ 1=aZ ð1aH Þ2 , FG ðgÞFG ðdÞ ¼ ðaH þ aX Þ= outcome. Furthermore, its maximized value equals Dv.
ð1aH aX Þ, @g=@aH ¼ 1=2ð1aH aX Þ2 , and @d=@aH ¼ 1=2 Thus, H will acquire at least 12 of the votes, and the value
ð1aH aX Þ2 Þ yields wedge will no longer depend on aH . H’s maximization
!  problem with respect to aH is therefore given by (note
1 aH þ aX 1 that there is still no incentive to go short, as the analysis
þ
aZ ð1aH Þ2 1aH aX 2ð1aH aX Þ2 in the proof of Proposition 1 above is valid here):
      
1 1
ð1q Þ 1 þ a d q 1 þ a g ð52Þ ðDvÞða2Z ð1aH Þ2 4a2H Þ
2 2 max : ð60Þ
aH 8aZ ð1aH Þ
aH þ aX aðaH þ aX Þ Taking the derivative of this with respect to aH , using
¼ 
aZ ð1aH Þ2 ð1aH aX Þ 4ð1aH aX Þ3 results derived above, yields
4aH !
þ 40: ð53Þ aH ðaH 2Þ aZ
4aZ ð1aH Þð1aH aX Þ2 Dv  o 0: ð61Þ
2aZ ðaH 1Þ2 8
The inequality is derived as follows. The last term is
clearly positive. The second term is negative and its Finally, with respect to the efficiency results, note that the
absolute value increases in a. Setting a= 2, the sum of probability of approval is 12 with or without H’s presence
the two terms is positive if aZ ð1aH Þ2 ð1aH when FG ðÞ ¼ FB ðÞ. On the other hand, when FG ðÞaFB ðÞ,
aX Þ o2ð1aH aX Þ3 ¼)2 4 aZ ð1aH Þ2 =ð1aH aX Þ2 , Proposition 4 implies that the probability of a correct
which always holds (the denominator is larger than 14 and decision is decreasing in aX when aH ¼ 0. &
the numerator is at most 12Þ.
Taking the derivative of (32) with respect to aX yields References
   
@d @g
ð1q Þ fG ðdÞ þ q fG ðgÞ ð54Þ Admati, A., Pfleiderer, P., 2009. The wall street walk and shareholder
@aX @aX
activism: exit as a form of voice. Review of Financial Studies 22,
! ! 2645–2685.
fG ðdÞ fG ðgÞ Allen, F., Gale, D., 1992. Stock price manipulation. Review of Financial
¼ ð1q Þ q : ð55Þ Studies 5, 503–529.
2ð1aH aX Þ2 2ð1aH aX Þ2 Attari, M., Banerjee, S., Noe, T., 2006. Crushed by a rational stampede:
strategic share dumping and shareholder insurrections. Journal of
Since fG ðÞ is linear and increasing, we have fG ðdÞ r fG ðgÞ. Financial Economics 79, 181–222.
Bagnoli, M., Lipman, B., 1996. Stock price manipulation through takeover
Also, as aH falls from aZ =ð2 þ aZ Þ to zero, q* rises from zero bids. RAND Journal of Economics 27, 124–147.
to 12. The result for aX follows. & Barclay, M., Holderness, C., 1989. Private benefits from control of public
corporations. Journal of Financial Economics 25, 371–395.
Proof of Proposition 5. Given the linear density, (32) can Blair, D., Golbe, D., Gerard, J., 1989. Unbundling the voting rights and
profit claims of common shares. Journal of Political Economy 97,
be written as 420–443.
    Brav, A., Jiang, W., Partnoy, F., Thomas, R., 2008. Hedge fund activism,
2að1gÞ 2að1dÞ
q 1 g þ ð1q Þ 1 d : ð56Þ corporate governance, and firm performance. Journal of Finance 63,
2 2 1729–1775.
 A. Brav, R.D. Mathews / Journal of Financial Economics 99 (2011) 289–307
Burkart, M., Gromb, D., Panunzi, F., 1997. Large shareholders, monitoring,
and the value of the firm. Quarterly Journal of Economics 112,
693–728.
Burkart, M., Lee, S., 2008. The one share-one vote debate: a theoretical
perspective. Review of Finance 12, 1–49.
Chakraborty, A., Yilmaz, B., 2004. Informed manipulation. Journal of
Economic Theory 114, 142–152.
Christoffersen, S., Geczy, C., Musto, D., Reed, A., 2007. Vote trading and
information aggregation. Journal of Finance 62, 2897–2929.
Clifford, C., 2008. Value creation or destruction: hedge funds as
shareholder activists. Journal of Corporate Finance 14, 323–336.
DeAngelo, H., DeAngelo, L., 1985. Managerial ownership of voting rights:
a study of public corporations with dual classes of stock. Journal of
Financial Economics 14, 33–69.
Edmans, A., 2009. Blockholder trading, market efficiency, and managerial
myopia. Journal of Finance 64, 2481–2513.
Edmans, A., Manso, G., forthcoming. Governance through trading and
intervention: a theory of multiple blockholders. Review of Financial
Studies.
Gilson, R., 1987. Evaluating dual class common stock: the relevance of
substitutes. Virginia Law Review 73, 807–844.
Goldstein, I., Guembel, A., 2008. Manipulation and the allocational role of
prices. Review of Economic Studies 75, 133–164.
Greenwood, R., Schor, M., 2009. Investor activism and takeovers. Journal
of Financial Economics 92, 362–375.
Grossman, S., Hart, O., 1988. One share-one vote and the market for
corporate control. Journal of Financial Economics 20, 175–202.
Harris, M., Raviv, A., 1988. Corporate governance: voting rights and
majority rules. Journal of Financial Economics 20, 203–255.
Hu, H., Black, B., 2006. The new vote buying: empty voting and hidden
(morphable) ownership. Southern California Law Review 79,
811–908.
Hu, H., Black, B., 2007. Hedge funds, insiders, and the decoupling of
economic and voting ownership: empty voting and hidden (morph-
able) ownership. Journal of Corporate Finance 13, 343–367.
Kahan, M., Rock, E., 2007. Hedge funds in corporate governance and
corporate control. University of Pennsylvania Law Review 155,
1021–1093.
307
Kahn, C., Winton, A., 1998. Ownership structure, speculation and
shareholder intervention. Journal of Finance 53, 99–129.
Kalay, A., Pant, S., 2009. One share-one vote is unenforceable and sub-
optimal. Unpublished working paper, University of Utah.
Khanna, N., Marietta-Westberg, J., 2005. Rational manipulation by large
shareholders: the hijacking of an election. Unpublished working
paper, Michigan State University.
Khanna, N., Sonti, R., 2004. Value creating stock manipulation: feedback
effect of stock prices on firm value. Journal of Financial Markets 7,
237–270.
Klein, A., Zur, E., 2009. Entrepreneurial shareholder activism: hedge
funds and other private investors. Journal of Finance 64, 187–229.
Kolasinksi, A., Reed, A., Ringgenberg, M., 2008. A multiple lender
approach to understanding supply and demand in the equity lending
market. Unpublished working paper, University of North Carolina at
Chapel Hill.
Kyle, A., Vila, J., 1991. Noise trading and takeovers. RAND Journal of
Economics 22, 54–71.
Manne, H., 1964. Some theoretical aspects of share voting: an essay in
honor of Adolf A. Berle. Columbia Law Review 64, 1427–1445.
Maug, E., 1998. Large shareholders as monitors: Is there a trade-off
between liquidity and control? Journal of Finance 53, 65–98.
Maug, E., 1999. How effective is proxy voting? information aggregation
and conflict resolution in corporate voting contests. Unpublished
working paper, University of Mannheim.
Musto, D., Yilmaz, B., 2003. Trading and voting. Journal of Political
Economy 111, 990–1003.
Neeman, Z., Orosel, G., 2006. On the efficiency of vote buying when
voters have common interests. International Review of Law and
Economics 26, 536–556.
Shleifer, A., Vishny, R., 1986. Large shareholders and corporate control.
Journal of Political Economy 94, 461–488.
Vila, J., 1989. Simple games of market manipulation. Economics Letters
29, 21–26.
Zachariadis, K., Olaru, I., 2010. Trading and voting in distressed firms.
Unpublished working paper, Northwestern University.
Zwiebel, J., 1995. Block investment and partial benefits of control.
Review of Economic Studies 62, 161–185.