Activist Hedge Funds: Who`s on the Other Side of Their Trades?

Activist Hedge Funds:
Who’s on the Other Side of Their Trades?
Nickolay Gantchev and Chotibhak Jotikasthira§
September 2012
ABSTRACT
Hedge fund activists achieve substantial improvements in the performance and governance of
target firms. These activists accumulate most of their holdings in the sixty days before a
campaign when other institutions heavily sell target shares. Does institutional exit facilitate the
emergence of value-enhancing activists? Three pieces of evidence indicate that the answer is yes.
First, at the daily frequency, we identify a strong positive relationship between institutional
selling and the activist’s purchases. Second, we find that institutional selling is driven by only a
few investors who dispose of a disproportionately large fraction of their other holdings. This
suggests that their sales of the target and the concurrent hedge fund purchases are not motivated
by common firm-specific news. Third, we instrument an institution’s daily trading in the target
by its trading in non-target stocks and establish a causal link between institutional sales and
hedge fund purchases. Institutional exit lowers prices and increases turnover, creating favorable
market conditions for an activist to acquire a block of target shares. Our results imply that even
non-informational institutional trading can play an important corporate governance role by
increasing the probability of activist interventions.
Keywords: Institutional investors, Hedge funds, Trading, Shareholder activism
JEL classification: G11, G12, G23, G34

We are grateful to Alon Brav, Paolo Fulghieri, Wei Jiang, Paige Ouimet, Anil Shivdasani, Geoffrey Tate, and
seminar participants at the University of North Carolina at Chapel Hill for useful comments. We also thank Oleg
Gredil for research assistance.
§
Gantchev and Jotikasthira are at the University of North Carolina at Chapel Hill. Corresponding author:
[email protected].
1. Introduction
Hedge funds have assumed the role of active monitors in the majority of shareholder campaigns
since 2000. Hedge funds suffer from fewer conflicts of interest, face minimal regulatory
restrictions, and have a better-aligned incentive structure than other institutional monitors such as
mutual funds, pension funds, and labor unions (see Kahan and Rock, 2006; Gillan and Starks,
2007). As a result, hedge fund activists have achieved significant improvements in the
performance and governance of target firms, resulting in high abnormal returns (see Brav, Jiang,
Partnoy, and Thomas, 2008; Becht, Franks, Mayer, and Rossi, 2008; Klein and Zur, 2009;
Clifford, 2008).1
It is well documented that hedge funds typically target firms with large institutional holdings.
The rationale for this is twofold: first, institutional investors are able to better evaluate the
success of an activist intervention, facilitating a faster convergence of the target’s share price to
its fundamental value; second, institutional voting directly impacts a campaign’s success.
Although these general facts are well established, the literature has been silent on how activists
accumulate their stakes, how other institutional investors trade before the public announcement
of a campaign, and whether the trading of these two sophisticated investor types is economically
related. Our paper aims to fill this gap.
Three of our main results deserve attention. First, at the daily frequency, we establish a strong
positive relationship between institutional selling of target shares and the activist’s accumulation
of a block position. Second, we show that institutional selling is driven only by a few investors
whose disproportionate selling of non-target stocks suggests that their trades are liquidity
motivated. Third, we instrument daily institutional trading volume by investor-specific trading in
non-target stocks to establish a causal link between institutional sales and hedge fund purchases
of target shares. The combined evidence presented in this paper indicates that even noninformational institutional exit can be a powerful corporate governance mechanism.
Our paper complements recent work in the corporate governance literature, which argues that
large institutional shareholders add value even without direct intervention in firm operations.
Edmans (2009) develops a theoretical model in which a privately informed blockholder trades on
negative news, encouraging management to improve long-term investment. Edmans and Manso
(2011) show that a multiple-blockholder structure increases the effect of governance through
trading while Kahn and Winton (1998) illustrate that information precision raises a blockholder’s
motive to speculate. Empirically, McCahery, Saunter, and Starks (2008) present survey evidence
that exit is the most prevalent institutional governance mechanism while Parrino, Sias, and
1
Karpoff (2001), Gillan and Starks (2007), and Yermack (2010) conclude that institutional activism by pension
funds, mutual funds, and unions has had limited impact on firm governance and performance. 1
Starks (2003) and Chen, Harford, and Li (2007) demonstrate its role in the context of CEO
turnover and acquisitions.
The above papers consider blockholders as informed traders who impound their private
information into stock prices and increase market efficiency. A common assumption is that block
size is positively associated with the shareholder’s incentives to gather and trade on private
information. We use the actual trading pattern of institutional investors (not their block size) to
demonstrate the importance of institutional exit. Most importantly, we show that even noninformational institutional trading plays an important corporate governance role by facilitating
the emergence of an activist shareholder with incentives to monitor.
The governance mechanism described in this paper emphasizes the importance of liquid stock
markets for the activist’s accumulation of a concentrated position. Maug (1998) focuses on the
role of liquidity in the formation of a large ownership block and Edmans (2009) shows that
liquidity is also vital for making a blockholder’s threat of exit credible. Our findings provide
empirical support for Maug’s theoretical argument that liquidity aids effective corporate
governance because it makes it easier for an activist to purchase a large stake in a target firm
with a limited price impact.2 Thus, liquid markets increase the likelihood that a large shareholder
will step in and actively monitor.
Our analysis starts by documenting that the average activist hedge fund purchases the majority of
its initial stake (4.2% of the target’s outstanding shares) in the open market in the two months
before the public announcement of activist intentions. During the same period, non-hedge fund
institutions sell a large fraction of their target holdings (1.5% of outstanding shares). 3 On the day
the activist crosses the 5% regulatory threshold, the average hedge fund purchases 1.02% and
institutions sell a net of 0.34% of the target’s outstanding shares. The combined trading of hedge
fund and non-hedge fund institutions accounts for 61.59% of the market volume on that day,
suggesting that they trade (indirectly) with each other.
[Insert Figure 1]
Generally, we observe that the trading patterns of hedge funds and other institutions are highly
synchronized in time, leading up to the campaign announcement. Mutual and pension funds start
selling about eight months before the activist filing but their selling dramatically accelerates
during the two-month period before the campaign. This evidence implies that institutional
trading may facilitate the activist’s accumulation of a concentrated position in the target.
2
Maug (1998) argues that liquidity helps a large shareholder pass on the costs of intervention to uninformed
liquidity traders. Gantchev (2012) provides empirical evidence that the costs of activist monitoring are substantial.
3
The institutional data are from Ancerno (formerly known as the Abel/Noser Corp.) and include a large subset of
mutual and pension funds. For a typical stock, these institutions account for up to 20% of its total CRSP volume. 2
[Insert Figure 2]
Our examination of the effect of institutional trading on hedge fund purchases of target shares
relies on three identification steps. First, matching daily institutional volume to net hedge fund
trades, we show that hedge funds are significantly more likely to trade on days of net
institutional selling than on days of net institutional buying or no trading. Moreover, the activist
purchases a larger quantity of shares when institutional selling volume is higher. The daily
frequency of our data makes it unlikely that this relationship is spuriously generated by the
aggregation of trades over time.
Second, we investigate whether institutional selling is motivated by fundamental firm-specific
news or need for liquidity. We find that only a few institutions (out of many that trade in a target
stock) are responsible for almost all non-hedge fund institutional selling on the event day and
two thirds of the selling in the sixty days before a campaign. Examining these institutions’
trading in non-target stocks, we find that they sell a disproportionately large fraction of their
other holdings suggesting that they are more ‘distressed’ than other institutional investors. We
interpret this as evidence that liquidity (rather than information) drives institutional selling.
Third, we instrument each institution’s daily trading in the target by its trading in non-target
stocks to establish a causal link between institutional sales and hedge fund purchases. Our
instrument is motivated by a robust positive (negative) relationship, at the institution level,
between the probability of selling (buying) the target stock and the fraction of non-target stocks
that the institution sells. We use this relationship to calculate the expected probabilities that each
individual institution will sell and buy the target stock on each trading day. We then multiply
these probabilities by the average trade size and sum the product across all institutions to obtain
our instruments, the expected total sell and buy volumes. Since the expected volumes are
institution-specific, rather than target-stock-specific, we achieve identification of the causal flow
from institutional selling to hedge fund buying.
Finally, we investigate the market conditions associated with institutional selling and their
relationship to the hedge fund’s accumulation of target shares. Our results show that non-hedge
fund institutional trading is associated with two effects, which may facilitate the hedge fund’s
acquisition of a block position. First, institutional trading correlates with high trading volume
allowing hedge funds to camouflage their trades. Second, institutional selling is also associated
with a significantly negative price impact helping hedge funds purchase shares at lower prices.
We confirm these relationships in a dynamic setting using Panel Vector Autoregression (PVAR),
allowing for persistence in liquidity, institutional trading, and hedge fund trading.
Our paper contributes to several strands of the finance literature. First, we complement recent
work in the corporate governance literature, specifically as it relates to shareholder activism.
3
Our results establish a positive causal relationship between institutional exit and the activist’s
ability to accumulate a block of target shares in a relatively short period and with a limited price
impact. We emphasize the role of liquidity in allowing an incoming blockholder to share the
costs of an intervention with other shareholders and raise the likelihood of activist monitoring.
Our evidence suggests that non-informational institutional trading plays an important corporate
governance role by facilitating activist monitoring.
Second, our paper contributes to a growing literature examining the trading behavior of hedge
funds and their role as liquidity providers.4 Chen et al. (2008) compare two hedge fund trading
strategies – liquidity provision and front-running – and provide indirect evidence that hedge
funds profit from front-running distressed mutual funds. Their paper shows a significant positive
correlation between the returns of equity hedge funds and contemporaneous mutual fund
outflows and argues against a liquidity provision interpretation of these results. Shive and Yun
(2012) find that hedge funds profitably trade on quarter-ahead predicted aggregate mutual fund
flows, especially in small and illiquid stocks. They combine quarterly lagged mutual fund flows
and returns with hypothetical stock allocation schemes to predict mutual fund flows.
Collin-Dufresne and Fos (2012) argue that standard liquidity measures fail to capture the
presence of informed traders. They examine the trades of a large sample of Schedule 13D owners
(many of them hedge funds) and provide evidence of selection bias in liquidity measures due to
the endogenous timing of informed trading. Their evidence is consistent with our results that
hedge fund trading is associated with better liquidity of target stocks. Campbell, Ramadorai and
Schwartz (2009) infer daily institutional trading from TAQ data and show that institutions
demand liquidity, especially when they sell.
Finally, our findings have important public policy implications regarding the disclosure
requirements for blockholder ownership. The U.S. Securities and Exchange Commission (SEC)
is currently considering whether reducing the ten-day period for reporting beneficial ownership
in Schedule 13D will protect investors from the arguably abusive trading of ‘aggressive’
activists. We find evidence consistent with the conclusion that the benefits of increased
transparency are outweighed by its costs in terms of lower incentives for a blockholder to
accumulate a large stake in a firm and monitor its operations.
The remainder of the paper is organized as follows. Section 2 describes the data used in this
study. Section 3 summarizes hedge fund and other institutional trading in target and non-target
stocks and Section 4 investigates the accumulation of concentrated positions by activist hedge
funds. Section 5 considers individual institutions’ daily trading in non-target stocks as an
4
An additional strand of the hedge fund literature considers the relation between the ownership of hedge funds and
other institutional investors (see Griffin and Xu, 2009; Jiao, 2012; and Ben-David, Franzoni and Moussawi, 2012).
4
instrument to establish a causal link between institutional selling and hedge fund purchases.
Section 6 outlines the economic mechanism and introduces panel vector autoregression allowing
for persistence in liquidity, institutional trading, and hedge fund trading. Section 7 discusses
public policy implications and Section 8 concludes.
2. Data
Our analysis combines three unique datasets covering hedge fund activist campaigns, daily
hedge fund trades in activist targets, and high frequency trades by non-hedge fund institutions.
The list of activist campaigns comes from Gantchev (2012) who uses data from SEC Schedule
13D, preliminary and definitive proxy statements, and SharkRepellent.net to construct a
comprehensive data set of hedge fund activist campaigns in 2000-2007. The activist sample
consists of 1,191 distinct campaigns involving 981 unique targets and 130 hedge fund families.
Any investor who acquires more than 5% of the voting stock of a public company with the
intention of influencing its operations or management must file Schedule 13D with the SEC. As
part of the filing, activists are required to report all transactions in the target’s shares in the 60
days before the filing date. We manually collect the date of each reported transaction, the amount
of shares purchased or sold, the price per share and the type of each transaction (open market,
private or other) for the original sample of activist campaigns. We have the hedge fund
transaction history for 813 of the activist events; the remaining campaigns do not provide
transaction details because of previous Schedule 13G filings, private placement or IPO
distributions, missing share or price information, etc.5
We combine the hedge fund transaction data with high-frequency institutional trading data from
Ancerno (formerly known as the Abel/Noser Corporation).6 Ancerno provides transaction cost
analysis to mutual funds, pension plan sponsors, and brokers representing (on average) 13.47%
of the total CRSP volume and 14.66% of the total CRSP principal traded during 2000-2007. The
dataset covers the trading activity of 94,776 managers associated with 914 unique client
institutions during the sample period.7 The institutions with the largest principal traded are
Fidelity, Wellington Management, Vanguard and AllianceBernstein. The data include the
identity of the trading institution, the execution date and time, the stock ticker and number of
shares traded, the price, commission and taxes per share, and the direction of each trade. As
5
The 13G filing is considered a more passive version of the 13D, and has fewer reporting requirements. Activist
practices are not permitted by 13G filers unless they refile as 13D owners.
6
See Chemmanur, He, and Hu (2009) and Puckett and Yan (2011) for a broad description of the data.
7
We conservatively define an institution as the unique combination of client code and client manager code from
Ancerno. However, multiple client manager codes are typically associated with the same client code.
5
discussed in Puckett and Yan (2011), the Ancerno data suffers from no significant survivorship
or selection biases.
We document that about 70 unique institutions sell and 56 institutions buy the average target
stock in the 240 days before the start of a campaign. However, the top 5 sellers (buyers) account
for about 68% (62%) of the total institutional sell (buy) volume in the sixty days before the
public announcement of a campaign. These top traders are mostly household names, such as
AllianceBernstein, AXA, Barclays Global Investors, Dimensional Fund Advisors, Fidelity,
Loomis, Sayles & Co., Putnam, State Street, Vanguard and Wellington Management.
We obtain stock market data from the Center for Research in Security Prices (CRSP) and
accounting data from Compustat. Our final sample consists of 643 activist campaigns with
hedge fund and institutional trading data as well as CRSP price and share information.
3. Institutional trading in target and non-target stocks
The typical analysis of institutional ownership in the literature uses data from the Thomson
Reuters 13F Institutional database, which contains the quarterly ownership reports of institutions
reporting to the SEC. Tracing the identity of 13F institutional owners around the start of activist
campaigns in 2000-2007, we find that the mean (median) ownership of hedge funds increases by
3.94% (4.44%) of shares outstanding while the ownership of other institutions decreases by
5.43% (5.31%) (see Figure 3). This finding suggests a significant churning of institutional
investors before the start of a campaign, with hedge funds replacing other institutions as major
blockholders of target firms.
[Insert Figure 3]
Given the speed with which activists accumulate concentrated positions in their targets, a natural
question is who is on the other side of their trades? More importantly, is there a relationship
between the disposal of target shares by non-hedge fund institutions in the months before the
start of a campaign and the accumulation of the hedge funds’ activist stakes? Does the trading
by non-hedge fund institutions create favorable market conditions, which facilitate the hedge
funds’ acquisition of target shares? Answering these questions using quarterly 13F data is
obviously problematic.
6
3.1. Hedge fund trading in target stocks
The daily transaction data we use in this study allows us to investigate the manner in which
hedge fund activists accumulate concentrated positions in target firms. The analysis in this
section focuses on the 60 days before the public announcement of a campaign in a Schedule 13D
filed with the SEC.
Table 1 summarizes the trading in target stocks by 643 activist hedge funds in 2000-2007. The
trading volume of our sample of activists represents an average of 15.78% of the total CRSP
volume in the shares of a target. This number goes up to 17.63% in the 10-day period between
the hedge fund’s crossing of the 5% reporting threshold (event date) and the filing date. On the
event date, the activist’s trades comprise 41.24% of the target’s total market volume.
[Insert Table 1]
Almost all hedge fund purchases of target shares (mean 97.51%, median 100%) are executed in
the open market in a small number of large trades. In the two months before the start of a
campaign, the average activist purchases 4.25% of the target’s outstanding shares representing
61.89% of the total ownership of the hedge fund on the filing date. About one-third of activists
acquire more than 5% of the target’s outstanding shares in the 60 days before filing a Schedule
13D (untabulated).
Hedge funds trade very actively on the event date. On that single day, the average hedge fund
activist acquires more than 1% of the target’s outstanding shares representing 13.68% of the
activist’s total ownership at the start of the campaign. Almost all of these transactions are
executed in the open market and are very large in size (the mean (median) number of trades is 14
(2)). Hedge funds continue to purchase shares after crossing the 5% ownership threshold and
accumulate another 1.28% of outstanding shares until the public announcement of their activist
intentions.
We also document a significant price run-up in the 60 days before the start of a campaign. The
average purchase price in that period is 94.12% of the target’s price on the filing date. On the
event date, activists acquire shares at an average discount of 2.42% to the price on the filing date.
Surprisingly, the large volume of hedge fund purchases on the event date (more than 1% of
outstanding shares) does not seem to have a significant price impact. The difference in the
average prices between the event day and the 10-day period until the filing date is only 1%. We
explore the effects of institutional trades on the targets’ returns and liquidity in Section 6.
Figure 4 clearly shows that the average cumulative abnormal returns (CARs) of targeted stocks
turn significantly negative in the 240 days leading up to the start of a campaign. This period is
7
characterized by heavy non-hedge fund institutional selling. Consistent with prior evidence in
the literature, we also observe a significant price run-up in the 60 days before a campaign,
especially in the 10 days between the event and filing dates. This coincides with the
accumulation of the hedge fund’s position in the target.
[Insert Figure 4]
3.2. Institutional trading in target stocks
As shown in Figure 1, the net volume of institutional trades turns negative about 240 days before
the public announcement of a campaign. The average activist acquires the majority of his
ownership in the target in the 60 days before the filing date (see Table 1) suggesting that the
selling by non-hedge fund institutions precedes the acquisition of the hedge fund’s stake.
Table 2 describes the institutional trading in activist targets in 2000-2007. The trading volume of
our sample of mutual and pension funds represents 13.46% of the average target’s market
volume in the period from t-240 to t-60 days and 14.36% in the 60 days leading up to the
campaign. Adding the trading volume of hedge fund activists from Table 1, we find that trading
by both institutional groups accounts for 30.14% of the target’s market volume in the 60 days
before the campaign and 61.59% of the market volume on the event day. The large fraction of
trading volume represented by our sample of institutions implies that these two market players
trade indirectly with each.
[Insert Table 2]
Non-hedge fund institutions sell an average of 2.52% of a target’s outstanding shares in the 240
days before the activist’s SEC filing. The majority of this selling (1.50% of shares outstanding)
occurs in the 60 days before the filing date. On the event day, these institutions sell a net of
0.34% of the target’s outstanding shares. Our sample of mutual and pension funds seem to be
providing a large proportion of the shares purchased by the activist on that day. The median
statistics suggest that the distribution of institutional trades is positively skewed with a few
institutions supplying the majority of shares.
The mean number of selling institutions exceeds the mean number of buying institutions in all
event periods. On the event day, the median number of selling non-hedge fund institutions is 2
and the median number of trades per institution is 1. The largest single seller represents 70% of
total institutional sell volume on that day (untabulated). It seems that only one or two institutions
are responsible for the majority of trading in the target on the event day. We interpret this as
evidence that institution-specific (rather than target-specific) events are driving institutional
8
trading.
We further investigate the determinants of institutional trading in activist targets by tracing the
activity of the top (event-day or sixty-day) sellers and buyers in the period between 240 days
before and 30 days after the start of a campaign. Table 3 presents a breakdown of the results.
[Insert Table 3]
In Panels A and B, we report the combined trading activity of the top two (event-day) sellers and
buyers of target stocks. On the event day, the top two sellers dispose of 0.44% and the top two
buyers purchase 0.13% of the average target’s outstanding shares. Compared to the cumulative
institutional selling of 0.46% and buying of 0.12% of shares outstanding (Table 2), we clearly
see that the top two institutional buyers and sellers account for virtually all of the institutional
trading activity on the event day.8
In the 60-day window to the filing date, the top two (out of 40 average) selling institutions
dispose of 1.02% of the target’s outstanding shares. Compared to the cumulative institutional
selling of 4.21% of outstanding shares (Table 2), the top two sellers account for close to onequarter of all selling. In the same period, the top two (out of 30 average) institutional buyers
purchase 0.53% of shares outstanding, or about 20% of cumulative institutional purchases.
The results so far suggests that almost all of the institutional selling on the event day and one
quarter of the selling in the sixty days before a campaign is driven by only two institutions (out
of the average of 40 institutions who trade during that window). This makes it unlikely that
stock-specific negative news is driving institutional trading. Rather, in the next section, we find
evidence that selling institutions are more distressed than the average trader, as they also sell a
higher fraction of non-target stocks. This is also confirmed by the fact that the top two sellers
continue to dispose of target shares in the 30 days after the public announcement of a campaign,
selling an additional 0.38% of the target’s outstanding shares.
The last two panels of Table 3 report the trading activity of the top 5 institutional sellers and
buyers of target stocks (determined during the sixty days before the filing date). The top 5 sellers
account for 68% (=2.86/4.21) of the cumulative institutional selling and the top 5 buyers
represent 62% of the cumulative institutional buying in the 60 days before a campaign. These
results confirm that only a handful of institutions are responsible for the majority of trading in
target stocks.
8
Note that the numbers of campaigns in calculating these statistics are different, as not all campaigns have
institutional sellers and buyers on the event day.
9
3.3. Institutional trading in non-target stocks
To further investigate mutual and pension funds’ motivation for trading in activist targets, we
study their trading activity in non-target stocks.9 Table 4 presents the results for the top 2 and
top 5 buyers and sellers.
[Insert Table 4]
Comparing the top 2 event-day sellers (in Panel A) to the top 2 event-day buyers (in Panel B),
we observe a significant difference in both the fraction of sell principal (measured as the total
sell principal divided by the combined buy and sell principal) and the proportion of stocks sold.
In the early period from t-240 to t-60 days, 50% of the principal and 48% of the stocks traded by
the top two sellers represent selling. On the other hand, only 41% of the principal and 40% of the
stocks traded by the top two buyers represent selling. The same pattern persists on the event day
(51% vs. 31%) and in the thirty days after the filing date (51% vs. 40%).
Panels C and D report the trading activity in non-target stocks of the top 5 sellers and buyers in
the sixty days leading up to the filing date. The results are consistent with our previous findings
– the selling activity of the top 5 sellers of target stocks represents a significantly larger portion
of their total principal and stocks traded. The widespread selling of the top sellers of target
stocks suggests that these funds are relatively more ‘distressed’ and trade (partially) for liquidity
reasons. These institutions continue to sell in the thirty days following the public announcement
of activism. Consequently, institutional selling in the days before a campaign is likely not driven
by adverse target-specific news.
4. Accumulation of concentrated positions by activist hedge funds
4.1. Hedge fund purchases as a function of institutional trading
Figures 1 and 2 reveal a startling synchronicity in the daily trading of hedge fund and non-hedge
fund institutions in activist targets. The combined trading of these two market players accounts
for 61.59% of the target’s market volume on the event day indicating that they trade indirectly
with each other. Moreover, only one or two institutions seem to be responsible for almost all of
the non-hedge fund institutional selling on the event day and one quarter of the selling in the
sixty days before a campaign. Since these sellers also sell the majority of their non-target stocks,
we argue that these institutions are relatively more ‘distressed’ and trade for liquidity reasons.
9
We rely on the finding of Coval and Stafford (2007) that funds experiencing large inflows (outflows) tend to buy
(sell) the majority of their stocks. However, unlike Coval and Stafford (2007), we do not have flow data at the daily
frequency; therefore, we rely on the trading behavior across a large set of stocks to infer institutions’ circumstances.
10
In this section, we explore the relationship between a hedge fund’s accumulation of a position in
a target firm and institutional net trading in the sixty days before the start of a campaign. Our
results show that hedge funds are significantly more likely to trade on days of net institutional
selling than on days of net institutional buying or no trading. Moreover, the activist purchases a
larger (smaller) quantity of shares when institutional selling (buying) volume is higher. In the
next section, we use instrumental analysis to establish a causal link between institutional selling
and the hedge fund’s acquisition of a concentrated stake in the target.
In Table 5, we present a ‘naïve’ analysis of the relationship between a hedge fund’s
accumulation of target shares and institutional trading. In Panel A, we report logistic regressions
with a dependent variable equal to one if the activist hedge fund trades on a trading day. In Panel
B, we present OLS estimation where the dependent variable is the daily net hedge fund volume
as a percent of shares outstanding.
[Insert Table 5]
We use two indicator variables to denote days with non-hedge fund institutional net selling and
net buying, and a continuous covariate measuring the ratio of institutional selling and buying
volume to outstanding shares.10 As controls, we include the target’s contemporaneous abnormal
return, turnover, Amihud ratio, as well as VIX.11 We calculate abnormal returns by the marketmodel adjustment approach, in which the CRSP value-weighted index is used as the market
portfolio. Abnormal turnover (Amihud ratio) is calculated by subtracting the mean turnover
(Amihud ratio) from the daily turnover (Amihud ratio). The estimation period is from t-600 to t240. For campaign-level controls, we use cumulative abnormal returns (CAR), cumulative
abnormal turnover (CAT), and cumulative abnormal Amihud (CAA) in the early period from t240 to t-60 in some specifications and campaign fixed effects in others. We also use robust
standard errors, clustered by campaign, in all regressions.
Panel A of Table 5 reports logistic regressions estimating the probability of observing a daily
hedge fund trade. Each observation is a campaign day. Column (1) uses indicators for nonhedge fund institutional net selling and net buying as the main regressors. Both institutional net
selling and net buying are positively correlated with contemporaneous hedge fund trading. Both
relationships are highly statistically significant (at 1%). In terms of economic magnitude, the
odds that hedge funds trade are 64% higher on days of net institutional selling and 37% higher
on days of net institutional buying, compared to days with no institutional trading. Both
10
We use the selling and buying volumes directly, as opposed to the net volume, to allow for potentially asymmetric
impacts of selling and buying, and to capture the overall amount of trading. 11
Following Acharya and Pedersen (2005), we winsorize abnormal Amihud at 0.3. Nagel (2012) uses a reversal
strategy to proxy for the returns from liquidity provision and shows that the time variation in this strategy can be
predicted with the VIX index.
11
institutional net buying and net selling are associated with high abnormal turnover allowing
hedge funds to camouflage their trades. Institutional net selling also correlates with negative
abnormal returns helping hedge funds accumulate a position at a lower cost (see Section 6).
In Column (2), the main regressors are institutional selling and buying volumes as a percentage
of shares outstanding. Our results confirm that hedge funds are significantly more likely to trade
on days with high institutional selling volume. However, unlike the results in Column (1),
institutional buying volume has no effect. This suggests that hedge funds are more likely to
trade on days of net institutional buying only when net institutional selling is also high on those
days. In economic terms, a one-standard deviation increase in institutional selling volume
increases the odds of hedge fund trading by about 70%.
Column (3) includes controls for the target’s contemporaneous abnormal return, turnover, and
Amihud ratio. The effect of institutional selling drops only slightly. However, we need to
exercise caution in interpreting this result as institutional trading volumes, returns, and liquidity
are simultaneously determined, and hence highly correlated. Column (4) further controls for five
lags of hedge fund trading. The results are generally the same as in Columns (2) and (3). An
important difference is that the effect of institutional buying volume is negative and statistically
significant. In economic terms, a one-standard deviation increase in institutional selling (buying)
volume increases (decreases) the odds of hedge fund trading by 17% (4%). It is interesting to
note that lagged hedge fund trading is significantly positively correlated with current hedge fund
trading. In a subsequent section, we present a vector autoregression (VAR) analysis, which
confirms that net hedge fund volume is persistent for up to three lags (see also Figure 6).
Columns (5) and (6) repeat the analyses in Columns (2) and (4), with campaign fixed-effects
replacing CAR, CAT, and CAA during t-240 to t-60 as campaign-level controls. The results
remain qualitatively the same.
Panel B of Table 5 reports OLS estimates for the effect of institutional trading on net hedge fund
trading volume. The sets of regressors are the same as in the comparable columns in Panel A.
Column (1) shows that hedge funds purchase 0.05% (of shares outstanding) more on days of net
institutional selling. No significant effect is observed on days of net institutional buying. Across
the other specifications with continuous trading regressors, a one-standard deviation increase in
institutional selling volume increases hedge fund daily purchases by 0.07-0.13% of shares
outstanding, statistically significant at 1%. The effect of buying volume is either insignificant or
significantly negative (about -0.03% for a one-standard deviation increase), depending on the
specification.
12
4.2. Persistence of hedge fund and institutional trading
Our search for an instrument for non-hedge fund institutional trading starts with regressions of
measures of hedge fund and institutional trading on their lags. We include five lags to reflect the
number of trading days in a week. We also include campaign fixed effects and cluster standard
errors by campaign. Table 6 reports the results.
[Insert Table 6]
As expected, both hedge fund trading volume and institutional sell and buy volumes are
positively serially correlated. Column (1) reports logit estimates for a specification with a daily
hedge fund trading dummy as the dependent variable. Column (2) reports OLS estimates with
hedge fund volume as the dependent variable. Both specifications show significant persistence
of hedge fund trading for up to four lags.
Columns (3) and (4) of Table 6 report OLS coefficient estimates for two regressions with
institutional selling and buying volume, respectively, as dependent variables. We find
significant persistence in both selling and buying volumes for up to four lags. Current
institutional selling (buying) volume is also negatively correlated with lagged buying (selling)
volumes. Although the cross lead-lag correlations are small, they highlight the need to use both
sets of lagged institutional volumes as instruments for current volumes.
The most important result of this analysis is that institutional trading is generally not
significantly correlated with lagged hedge fund trading, and neither is the reverse. As a
consequence, lagged institutional trading volumes are likely to fulfill the exclusion restrictions in
a regression of hedge fund trading on institutional trading (once lagged hedge fund trading is
controlled for). Coupled with the fact that institutional selling and buying volumes are
significantly serially correlated, their lags can be used to identify the causal effects of
institutional trading on hedge fund purchases.
5. Identification
5.1. Institutional trading of non-target stocks as an instrument
Table 4 revealed that the top sellers of target stocks are also likely to sell more of their other
holdings suggesting that fund distress may be a determinant of institutional trading in activist
targets. As a result, we use the institution-specific trading in other (non-target) stocks to
construct an instrument for non-hedge fund institutional trading in target stocks.
13
Figure 5 presents some preliminary evidence supporting the validity of this instrument. We plot
the percent of days on which institutions sell or buy target stocks as a function of their fraction of
non-target selling or buying principal (top panel), fraction of non-target stocks sold or bought
(middle panel), and percent of institution trading days during the sample period (bottom panel).
[Insert Figure 5]
Consistent with the results for top sellers and buyers (Table 4), institutions whose selling
represents a larger fraction of their total principal or holdings traded are also more likely to sell
target stocks. Conversely, funds whose buying constitutes a larger fraction of their total principal
or stocks traded are also more likely to buy target stocks. The extreme deciles in the top two
panels often show non-monotonicity of these effects, as the highest and lowest fractions of sell
principal and stocks sold (1 and 0) sometimes involve just one other stock.12
The last (right-most) bars in the top two panels show that when institutions do not trade other
(non-target) stocks, they are also unlikely to trade target stocks. In addition, the bottom panel
shows that institutions that trade more in general are also more likely to trade target stocks, both
buy and sell.
Overall, the evidence presented in Figure 5 implies that an institution’s combined trading in nontarget stocks may serve as an instrument for its trading in target stocks.
5.2. Instrumental variable analysis
We begin constructing instruments for institutional selling and buying volumes by estimating the
propensity that each institution will sell, buy, or not trade the target stock as a function of its
trading in other non-target stocks. This construction uses several variables suggested by our
prior findings: (i) a dummy that equals 1 if the institution trades any stock and 0 otherwise, (ii)
the interaction between the dummy in (i) and the fraction of sell principal or stocks sold, (iii) a
dummy that equals 1 if the institution only trades one other stock and 0 otherwise, and (iv) the
fraction of trading days during the sample period that the institution trades any stock. The
variables (i)-(iii) vary over time while (iv) is a constant characteristic of each institution. In all
specifications, we control for lagged trading in both target and non-target stocks, market returns,
and VIX.
Table 7 reports the results. The specifications in Panel A employ multinomial logistic functions
and are estimated by maximum likelihood. Buy and sell are considered jointly, with no trading
12
In this case, the trading is likely specific to one particular non-target stock and thus unrelated to target stocks. 14
as the reference outcome. We use CAR, CAT, and CAA during the earlier period from t-240 to
t-60 as campaign-level controls. The specifications in Panel B are linear and are estimated by
OLS. Propensities to buy and sell are separately estimated and are not bounded between 0 and 1.
We include campaign fixed-effects in the OLS regressions. In both Panel A and Panel B, we
cluster standard errors by campaign.
[Insert Table 7]
As a measure of sell fraction, we use the fraction of sell principal in Columns (1) and (2) of
Panel A and the fraction of stocks sold in Columns (3) and (4).13 Both models fit the data well
(with pseudo R2 of 25% and 27%, respectively), and the results are essentially the same.
Consistent with the univariate findings in Figure 5, an institution is more (less) likely to sell
(buy) target stocks if it sells a larger fraction of non-target stocks. These effects are highly
statistically significant. Moreover, an institution is more likely to sell and buy target stocks if it
also trades other (non-target) stocks.
The coefficients of the lagged covariates show that an institution is likely to trade the target stock
in the same direction as it did yesterday while its trading in non-target stocks appears to be
partially reversed. As for the control variables, we find that an institution is more likely to buy
when the market return is high, and less likely to buy and sell when VIX is high. Finally, the
results in Panel B confirm those in Panel A, indicating that the economic relationships we
identify are robust to changes in functional form.
We use the estimates in Panel A of Table 7 to obtain the predicted propensities that each
institution will sell and buy each target stock on each trading day. We then aggregate these
propensities across all institutions to obtain the expected total volumes. This is done in the
following steps. First, for each institution and each target stock, we calculate the average daily
selling (buying) volume, using only the days that it sells (buys) each target stock. Then, we
multiply the selling (buying) probability by the average selling (buying) volume. Finally, we sum
these products across all institutions trading in each target stock, separately for sells and buys, to
obtain the expected selling and buying volumes of all institutions.
We use the expected institutional selling and buying volumes (calculated above) together with
lagged institutional volumes as instruments to (over) identify the causal links between
institutional trading and hedge fund purchases. Table 8 reports the two-stage least squares
(2SLS) estimates. In all columns, the dependent variable is net hedge fund volume as a
percentage of shares outstanding. All models include campaign fixed-effects and VIX as
13
Note that the effect of sell fraction needs to be adjusted down if the fraction is obtained using just one non-target
stock. 15
controls. As a measure of sell fraction, we use the fraction of sell principal in Columns (1) and
(2) and the fraction of stocks sold in Columns (3) and (4).
[Insert Table 8]
All models pass the identification test as well as Hansen’s test of overidentifying restrictions.
The results in Columns (1) and (2) are essentially the same; so, we will only discuss those in
Column (2). Consistent with our earlier results, institutional selling volume has statistically
significant and positive effects on hedge fund purchases, while institutional buying volume has
no effect. In economic terms, a one-standard deviation increase in (exogenous) daily
institutional selling volume increases hedge fund purchases by approximately 0.18% of shares
outstanding (the standard deviation of hedge fund volume is approximately 0.37% of shares
outstanding).
In Columns (3) and (4), we add five lags of hedge fund net volume and (predicted) abnormal
return, abnormal turnover, and abnormal Amihud ratio as additional control variables.14 The
effect of both institutional selling and buying volumes appear smaller but they are similar in sign
and magnitude to those from the comparable specifications in Table 5.
The instrumental variable analysis confirms that institutional selling (buying) volume
significantly increases (decreases) hedge fund purchases. In term of economic magnitude, a onestandard deviation increase in institutional selling (buying) volume increases (decreases) the
odds of hedge fund trading by 0.06% (0.04%). Given that the standard deviation of hedge fund
volume is approximately 0.37% of shares outstanding, these causal effects are economically
significant.
6. Market Mechanism
6.1. Institutional trading, abnormal returns and liquidity: univariate analysis
In this section, we investigate whether the market conditions associated with institutional selling
are favorable for the hedge fund’s accumulation of a concentrated position in the target firm. We
differentiate our results from those in the previous literature by presenting novel evidence linking
the trading of hedge funds and other institutions to the abnormal returns and liquidity of activist
targets.
14
Abnormal return, turnover and Amihud are simultaneously determined with institutional and hedge fund trading.
We use their lags as instruments.
16
We estimate abnormal returns using the market model with the CRSP value-weighted return
index as the market portfolio. We use logged returns and calculate cumulative abnormal returns
(CARs) as the sum of abnormal returns in each event period. Abnormal turnover (Amihud ratio)
is calculated by subtracting the mean turnover (Amihud ratio) from the event period turnover
(Amihud ratio). The estimation period is from t-600 to t-240. With the exception of CARs, the
reported summary statistics are the cross-sectional means of mean daily campaign variables.
Table 9 presents our univariate analysis. The average target of an activist campaign experiences
negative cumulative abnormal returns (CARs) of -1.79% in the t-240 to t-60 days before the
activist’s public filing. These CARs turn significantly positive in the two months before the start
of the campaign. For example, the average target’s CAR is 4.42% in the 60 days before the filing
date and 3.21% in the 10-day period between the event and the filing dates. The two months
preceding the start of a campaign is when hedge funds accumulate the majority of their
concentrated positions.
[Insert Table 9]
We distinguish between days with hedge fund and/or non-hedge fund institutional selling and
days with no such trading. In Panel A, we find that the average target’s daily abnormal returns
are significantly higher on days with hedge fund trading than on days with no hedge fund trading
(e.g., 0.35% versus 0.06% in the sixty day period before filing). On the other hand, we document
significantly lower average daily abnormal returns on the days of non-hedge fund institutional
net selling (compared to days with no institutional selling). Panel B shows that abnormal
turnover is higher on days of hedge fund trading and net non-hedge fund institutional selling than
on other days.
Our univariate results reveal that hedge fund trading is associated with positive abnormal returns
whereas institutional net selling is associated with negative (or lower) abnormal returns. In
addition, both hedge fund and non-hedge fund trading lead to high abnormal turnover.
Consequently, the price impact of hedge fund and institutional trading is unclear depending on
whether the return effect (numerator) or the turnover effect (denominator) dominates.
Using abnormal Amihud ratio as a measure of price impact, Panel C shows that liquidity seems
to be consistently higher on days of hedge fund trading than on days with no such trading. The
association between liquidity and non-hedge fund institutional selling is less clear. Before the
event date, liquidity is higher on days with institutional net selling than on other days. However,
liquidity is lower on the event date for campaigns with institutional net selling implying that the
one or two institutional sellers on that day may demand liquidity.
17
Our results are broadly consistent with the findings of Collin-Dufresne and Fos (2012) who show
that standard liquidity measures (such as Kyle’s lambda and Amihud’s illiquidity measure)
improve on the days that hedge funds trade. However, unlike their focus on the hedge funds’
discretionary timing of trading strategies, we present evidence that hedge funds may actually
reduce the supply-demand imbalance caused by heavy institutional selling. Consequently, their
improved liquidity measures may be capturing the liquidity provision role of activist hedge
funds.15
6.2. Institutional trading, abnormal returns and liquidity: multivariate analysis
In this section, we ask whether the heavy selling by mutual and pension funds before the start of
a campaign creates market conditions that facilitate the hedge fund’s acquisition of a
concentrated stake. Our univariate findings show that net institutional selling correlates with
negative (or lower) abnormal returns and higher abnormal turnover. These conditions allow
hedge fund to both camouflage their trades and buy target shares at lower prices.
The results in Table 10 show the correlations between institutional trading and the abnormal
returns, turnover, and liquidity of activist targets in the period from t-240 to t-60 (before activist
hedge funds buy heavily and report their trades). In a standard OLS setting, we regress abnormal
returns, turnover, or liquidity on measures of institutional trading. We use two indicator
variables to denote days with non-hedge fund institutional net selling or net buying in columns
(1), (3), and (5), and two continuous variables measuring institutional selling and buying
volumes (as a percentage of shares outstanding) in columns (2), (4), and (6). In all columns, we
also control for VIX to capture market volatility and time-varying returns to liquidity provision,
include five lags of the dependent variable to control for its serial correlation, add campaign
fixed-effects, and cluster standard errors by campaign.
[Insert Table 10]
Column (1) of Table 10 shows that abnormal returns are negative when mutual funds and
pension funds sell and positive when they buy. The economic magnitude is about 20 basis points
in both cases. Using the continuous measures of institutional volume, we find in Column (2) that
it is only the institutional selling volume that significantly negatively impacts prices. A onestandard deviation increase in institutional selling volume is associated with approximately a 30
basis points drop in abnormal return. This is broadly consistent with our earlier evidence that, in
most cases, only institutional selling volume is associated with hedge fund purchases.
15
This ignores the fact that liquidity, institutional trading, and hedge fund trading are all endogenously determined. 18
Columns (3) and (4) of Table 10 estimate abnormal turnover as a function of non-hedge fund
institutional trading. As seen in Column (3), both net selling and net buying are associated with
higher turnover, even though the coefficient on net selling has double the magnitude of the
coefficient on net buying. The positive coefficients suggest that institutional trading does not
fully crowd out trading by other investors. In Column (4), we replace the two dummies with
institutional selling and buying volume (scaled by outstanding shares) and find a highly
statistically significant positive relationship. The coefficient on institutional selling volume is
close to 2, suggesting that most sell trades by Ancerno institutions are with non-Ancerno
investors.
The last two columns of Table 10 focus on the relationship between mutual and pension fund
trading and abnormal Amihud ratio. In Column 5, we find that abnormal Amihud is lower (i.e.,
liquidity is higher) when institutions buy or sell. This result may be partially driven by the high
turnover associated with institutional trading (as turnover is the denominator of the Amihud
measure). Replacing the net buy and sell dummies by institutional buying and selling volumes in
Column (6), we find similar results. Larger institutional selling and buying volumes are
associated with lower Amihud ratio, although the effect of selling volume is significantly larger
than that of buying volume.
The results in Table 10 provide evidence that non-hedge fund institutional trading is associated
with two effects, which may facilitate the hedge fund’s accumulation of a concentrated position.
First, institutional trading is correlated with high turnover, which allows hedge funds to hide
their trades. Second, institutional selling is also associated with a negative price impact helping
hedge funds purchase shares at lower prices.
The remaining regressions in this section focus on the sixty days before the start of a campaign,
during which we have both hedge fund and non-hedge fund institutional trades. Our goal is to
compare and relate the effects of hedge fund and non-hedge fund trading on abnormal returns,
turnover, and liquidity. The specifications are similar to those in Table 10, but include hedge
fund trading as an additional regressor. Table 11 reports the results.
[Insert Table 11]
In the sixty days before the announcement of a campaign, net institutional selling is associated
with negative abnormal returns and net institutional buying is correlated with positive abnormal
returns, consistent with the results in the earlier period. In addition, both net buying and net
selling correlate with high abnormal turnover and low Amihud ratio.
The effects of hedge fund trading on the abnormal returns, turnover, and liquidity of target stocks
are broadly similar to those of non-hedge fund institutional buying. On the days that hedge
19
funds trade target shares, abnormal returns increase by about 10 basis points, abnormal turnover
increases by about 0.4% and abnormal Amihud ratio drops by about 1.5% (Columns (1), (3), and
(5)). In Columns (2), (4) and (6), we measure hedge fund activity using net hedge fund volume
as a percentage of shares outstanding. The results are similar to those reported earlier. Column
(2) shows that hedge fund purchases are associated with positive abnormal returns. In terms of
economic magnitude, a one-standard deviation increase in net hedge fund volume is associated
with a 0.24 percent higher abnormal return. The return sensitivities to hedge fund purchases and
to institutional selling are similar but with opposite signs.
Column (4) shows that larger hedge fund purchases are associated with increased turnover, with
the multiplier slightly larger than one and about the same as the multiplier of institutional selling
volume. Column (6) provides evidence that hedge fund buying is associated with an
improvement in liquidity (i.e., lower abnormal Amihud). Comparing the coefficient on hedge
fund volume to those on institutional selling and buying volumes, we see that hedge fund trading
is associated with relatively better liquidity. This may be either a result of hedge funds directly
providing liquidity to other institutions or a result of hedge funds being better at strategically
timing their trades, or both.
All in all, the multivariate analysis confirms our univariate findings that net institutional selling
is associated with negative abnormal returns and higher abnormal turnover. These conditions
allow hedge funds to both camouflage their trades and buy target shares at lower prices. The
reverse is also true – hedge fund purchases increase prices and improve liquidity, allowing
plausibly distressed institutions to sell target shares at higher prices.
6.3. Vector Autoregression Analysis
We address concerns about the simultaneity among hedge fund and non-hedge fund institutional
trading, returns, and liquidity by using panel vector autoregression (PVAR). PVAR can be
helpful in teasing out the causal effects among these endogenous variables and in examining the
persistence of these effects using impulse response analysis.16
We use a recursive identification scheme (Sims (1980)) and arrange the endogenous variables in
the following order – institutional selling volume, institutional buying volume, hedge fund net
volume, abnormal turnover, abnormal returns, and abnormal Amihud ratio. Our results remain
qualitatively unchanged under alternative orderings of the variables (as long as hedge fund net
16
Note that identification of contemporaneous relationships still relies on exclusion restrictions, which in our case
are difficult to defend due to the lack of guidance from structural models. 20
volume follows institutional selling and buying). We use five lags to reflect the number of
trading days in a week.
In Figure 6, we present the responses of net hedge fund volume to a standard-deviation shock in
other endogenous variables. In the top left panel, we observe that a standard-deviation shock to
institutional selling (i.e., about 0.44% of shares outstanding) results in a 0.09% increase in hedge
fund volume (approx. 0.09/0.37 = 24% of its own standard deviation). This effect persists for
about two trading days, yielding a total response of about 0.10% of shares outstanding. In the
top right panel, we also see that institutional buying volume does not have a significant effect on
hedge fund volume.
[Insert Figure 6]
The left middle panel of Figure 6 shows that the innovation in hedge fund volume itself is
persistent for up to three lags. The remaining three panels describe the effects of lagged market
variables on hedge fund net volume (the contemporaneous effects are fixed at zero due to the
ordering). As expected, a positive shock to abnormal turnover has a positive but small and
insignificant effect on hedge fund trading (middle right panel) while a positive shock to liquidity
(i.e., lower Amihud) has a positive effect, significant up to 2 trading days (lower right panel).
Abnormal returns do not seem to have discernible impact on hedge fund trading.
In Figure 7, we focus on the effects of institutional and hedge fund trading on abnormal returns
and turnover. We observe that a one standard deviation shock to net institutional selling
(buying) has a positive contemporaneous effect of 0.69% (0.40%) of shares outstanding on
abnormal turnover which persists for 2 (1) days. A one-standard deviation shock to institutional
selling (buying) volume has a negative contemporaneous effect of -0.21% (essentially zero) on
abnormal returns which persists only for 1 day.
[Insert Figure 7]
In the bottom two panels, a one standard deviation shock to hedge fund volume has a positive
contemporaneous effect of 0.51% of shares outstanding on abnormal turnover and 0.20% on
returns. Both effects are similar in persistence to the effects of institutional selling volume.
In unreported results, we also find that a one standard deviation shock to net institutional selling
(buying) has a contemporaneous negative effect of -0.23% (-0.06%) on abnormal Amihud ratio
whereas a one standard deviation shock to net hedge fund volume has a contemporaneous
negative effect of -0.54% on abnormal Amihud ratio. These negative effects persist for several
days, much longer than the effects of trading on returns and turnover. These effects can be
mechanical as trading adds directly to the denominator of the Amihud ratio. However, the
21
significantly larger negative effect of hedge fund trading (despite its lower standard deviation
than that of institutional selling) suggests that activist hedge funds may provide liquidity to other
institutions.
7. Discussion: Policy implications
We show that one or two institutions are responsible for almost all of the non-hedge fund
institutional selling on the event day and one quarter of the selling in the sixty days before the
start of an activist campaign. Examining these funds’ trading in other non-target stocks, we find
that a significantly larger fraction of their trading constitutes selling suggesting that they are
relatively more ‘distressed’ and trade for liquidity reasons.
This evidence has public policy implications regarding the recent debate about tightening the
ten-day window before a 5% owner must disclose a beneficial position. In 2011, the law firm of
Wachtell, Lipton, Rosen & Katz submitted a petition to the SEC arguing that reducing the
reporting window will protect investors from aggressive activists who seek “to exploit this
period of permissible silence to acquire shares at a discount to the market price that may result
from its belated disclosures.”17 On the other hand, Bebchuk and Jackson (2011) argue that
increased transparency by tightening the disclosure requirements should be balanced against
reduced incentives for large shareholders to accumulate concentrated positions and monitor.
The results in this paper demonstrate that institutional rather than retail investors are typically on
the other side of hedge fund purchases. As sophisticated market players, institutional investors
are less likely to get harmed by trading with privately informed hedge funds. In addition, most
event-day institutional selling is by one or two institutions, suggesting that firm-specific news
concerning activist targets is less likely to explain institutional trading. An examination of the
trading pattern of the selling institutions also reveals that they are relatively more distressed as
evidenced by their wide spread selling of other holdings. In addition, these institutions continue
to dispose of target shares even after the public announcement of a campaign.
Even though the selling institutions potentially sell at prices that are not reflective of the
impending activist campaign, their trading before and after the event day reveals that a
heightened need for liquidity rather than private information is the main determinant of their
decision to dispose of target shares. Consequently, reducing the reporting window for disclosing
beneficial ownership is unlikely to significantly affect their decision to sell. In addition, the
17
Full text of the petition is available at http://www.sec.gov/rules/petitions/2011/petn4-624.pdf.
22
evidence suggests that hedge fund purchases increase prices and improve liquidity, allowing
these distressed institutions to sell target shares at higher prices.
8. Conclusion
In this paper, we ask whether the acquisition of a concentrated position by an activist hedge fund
is economically related to the trading behavior of other institutional investors. At the daily
frequency, we find a strong positive relationship between the net selling of pension and mutual
funds and the hedge fund’s purchase of target shares. On the day that the hedge fund crosses the
5% ownership threshold, the average activist purchases more than 1% while non-hedge fund
institutions sell a net of 0.34% of the target’s outstanding shares. The combined trading of hedge
fund and non-hedge fund institutions accounts for almost two-thirds of the market volume on the
event day suggesting that they trade indirectly with each other.
One or two institutions are responsible for almost all of the non-hedge fund institutional selling
on the event day and one quarter of the selling in the sixty days before a campaign. Examining
their trading in other (non-activist) stocks, we find that a significantly larger fraction of these
funds’ trading constitutes selling suggesting that they are relatively more ‘distressed’ and trade
for liquidity reasons. Using the trading of institutions in non-target stocks as an instrument, we
establish a causal link between institutional selling and hedge fund purchases of target stocks.
Investigating the economic mechanism, we find that institutional selling lowers prices and
increases turnover, creating a favorable market environment for the activist to acquire a large
quantity of target shares.
23
References
Acharya, V.V., Pedersen, L.H., 2005. Asset pricing with liquidity risk. Journal of Financial
Economics 77, 375-410.
Bebchuk, L.A., Jackson Jr., R.J., 2011. The law and economics of blockholder disclosure.
Harvard Business Law Review, forthcoming.
Becht, M., Franks, J., Mayer, C., Rossi, S., 2008. Returns to shareholder activism: evidence from
a clinical study of the Hermes UK Focus Fund. Review of Financial Studies 22, 3093-3129.
Ben-David, I., Franzoni, F., Moussawi, R., 2012. Hedge fund stock trading in the financial crisis
of 2007-2009. Review of Financial Studies 25, 1-54.
Brav, A., Jiang, W., Partnoy, F., Thomas, R., 2008. Hedge fund activism, corporate governance,
and firm performance. Journal of Finance 63, 1729-1773.
Campbell, J.Y., Ramadorai, T., Schwartz, A., 2009. Caught on tape: institutional trading, stock
returns, and earnings announcements. Journal of Financial Economics 92, 66-91.
Chemmanur, T.J., He, S., Hu, G., 2009, The role of institutional investors in seasoned equity
offerings, Journal of Financial Economics 94, 384-411.
Chen, J., Hanson, S., Hong, H., Stein, J.C., 2008. Do hedge funds profit from mutual-fund
distress? Unpublished working paper. NBER.
Chen, X., Hartford, J., Li, K., 2007. Monitoring: Which institutions matter? Journal of Financial
Economics 86, 279-305.
Clifford, C., 2008. Value creation or destruction? Hedge funds as shareholder activists. Journal
of Corporate Finance 14, 323-336.
Collin-Dufresne, P., Fos, V., 2012. Do prices reveal the presence of informed trading?
Unpublished working paper. Columbia University and University of Illinois at UrbanaChampaign.
Coval, J., Stafford, E., 2007. Asset fire sales (and purchases) in equity markets, Journal of
Financial Economics 86, 479-512.
Edmans, A, 2009. Blockholder trading, market efficiency, and managerial myopia. Journal of
Finance 64, 2481-2513.
Edmans, A., Manso, G., 2011. Governance through trading and intervention: a theory of multiple
blockholders. Review of Financial Studies 24, 2395-2428.
Gantchev, N., 2012. The costs of shareholder activism: evidence from a sequential decision
model. Journal of Financial Economics, forthcoming.
24
Gillan, S., Starks, L., 2007. The evolution of shareholder activism in the United States. Journal
of Applied Corporate Finance 19, 55-73.
Griffin, J., Xu., 2009. How smart are smart guys? A unique view from hedge fund stock
holdings. Review of Financial Studies 22, 2531-2570.
Jiao, Y., 2012. Hedge funds and equity prices. Review of Finance, forthcoming.
Kahan, M., Rock, E., 2007. Hedge funds in corporate governance and corporate control.
University of Pennsylvania Law Review 155, 1021-1093.
Kahn, C., Winton, A., 1998. Ownership structure, speculation, and shareholder intervention.
Journal of Finance 53, 99-129.
Karpoff, J., 2001. The impact of shareholder activism on target companies: a survey of empirical
findings. Unpublished working paper. University of Washington.
Klein, A., Zur, E., 2009. Entrepreneurial shareholder activism: hedge funds and other private
investors. Journal of Finance 63, 187-229.
Maug, E., 1998. Large shareholders as monitors: is there a trade-o§ between liquidity and
control? Journal of Finance 53, 65-98.
McCahery, J., Sautner, Z., Starks, L., 2010. Behind the scenes: the corporate governance
preferences of institutional investors. Working paper, Tilburg University.
Nagel, S., 2012. Evaporating liquidity. Review of Financial Studies, forthcoming.
Parrino, R., Sias, R., Starks, L., 2003. Voting with their feet: Institutional ownership changes
around forced CEO turnover, Journal of Financial Economics 68, 3-46.
Puckett, A., Yan, X.S., 2011. The interim trading skills of institutional investors, Journal of
Finance 66, 601-633.
Shive, S., Yun, H., 2012. Are mutual funds sitting ducks? Journal of Financial Economics,
forthcoming.
Yermack, D., 2010. Shareholder voting and corporate governance. Annual Review of Financial
Economics 2, 103-125.
25
Figure 1: Mean Net Trading Volume of Activist Hedge Funds and Other Institutions
The figure plots the average target’s mean net trading volume (as a percentage of shares outstanding) of
activist hedge funds and other institutions during the 60 days before the public announcement of activism.
The mean is calculated across 643 campaigns with available trading data in 2000-2007. Event date (day
0) refers to the date on which the hedge fund’s ownership crosses the 5% threshold. Hedge fund trading
data are collected from SEC filings and non-hedge fund institutional trades come from Ancerno.
1.20%
Mean net volume of non-HF institutions as % of
outstanding
1.00%
Mean net volume of activist HF as % of
outstanding
0.80%
0.60%
0.40%
0.20%
0.00%
-50
-40
-30
-20
-10
0
10
-0.20%
-0.40%
Days from Event Date
Figure 2: Mean Cumulative Ownership of Activist Hedge Funds and Other Institutions
The figure plots the average target’s mean cumulative ownership (as a percentage of shares outstanding)
of activist hedge funds and other institutions (starting from 0% on day t-360) in the one-year period
before the public announcement of activism. Event date (day 0) refers to the date on which the hedge
fund’s ownership crosses the 5% regulatory threshold. The mean is calculated across 643 campaigns with
available trading data in 2000-2007. Hedge fund trading data are collected from SEC filings and nonhedge fund institutional trades come from Ancerno.
2%
7%
1%
4%
0%
-360 -330 -300 -270 -240 -210 -180 -150 -120 -90
-1%
-2%
1%
-60
-30
0
-2%
Mean cumulative shares of non-HF
institutions as % of outstanding (LEFT)
-5%
Mean cumulative shares of activist HF as %
of outstanding (RIGHT)
-3%
-8%
Days from Event Date
26
Figure 3: Mean/ Median Institutional Ownership around the Public Announcement of Activism
The figure plots changes in the mean and median ownership of hedge funds and other institutions in target
stocks over the four quarters surrounding the start of a hedge fund activist campaign. The sample period is
2000-2007. The reference quarter (Quarter 0) contains the date of the public announcement (SEC filing).
The institutional ownership data come from Thomson Reuters 13F Institutional Database.
0.14
0.58
0.12
0.56
0.10
0.54
0.08
0.52
0.06
0.50
0.04
Mean Hedge Fund Share Ownership (LEFT)
Median Hedge Fund Share Ownership (LEFT)
Mean Other Institutional Ownership (RIGHT)
Median Other Institutional Ownership (RIGHT)
0.02
0.00
-­‐2 -­‐1 0 1 Quarters around Campaign Start
0.48
0.46
2 Figure 4: Mean Cumulative Abnormal Returns and Hedge Fund Trade Price to Price on File Date
The figure plots the average target’s mean cumulative abnormal returns (CARs) and mean ratio of the
hedge fund’s (trade size-weighted) trade price to the price on the filing date. CARs are calculated by the
market-model adjustment approach, in which the CRSP value-weighted index is used as the market
portfolio and the loading of each target stock return on the market return is estimated using the period
from t-600 to t-240 days before SEC filing. The mean is calculated across 643 campaigns with available
trading data in 2000-2007. Hedge fund trading data are collected from SEC filings and non-hedge fund
institutional trades come from Ancerno.
3%
2%
1%
0%
-240
-1%
1.01
Mean Cumulative Abnormal
Returns (CARs) (LEFT)
Mean activist HF trade price/
price on file date (RIGHT)
1.00
0.99
0.98
-210
-180
-150
-120
-90
-30
0
0.97
-2%
0.96
-3%
0.95
-4%
0.94
-5%
0.93
-6%
0.92
Days from Event Date
-60
27
Figure 5: Institutional Trading in Target and Non-Target Stocks
These figures plot the percentage of institution-target-days on which the target stocks are sold or bought
conditional on the institutions’ trading patterns in other non-target stocks. The observations are sorted by
the fraction of sell principal in non-target stocks (top panel), the fraction of non-target stocks sold (middle
panel), and the percentage of days in the sample on which each institution trades non-target stocks. The
sample includes 643 campaigns in 2000-2007. Non-hedge fund institutional trades come from Ancerno.
% Days Target Stocks Are Bought
% Days Target Stocks Are Sold
8.0%
6.0%
6.0%
4.0%
4.0%
2.0%
2.0%
0.0%
0.0%
1
3
5
7
9
No
Highest Decile by Fraction of Sell Principal Trading
in Other
Stocks
1
Highest
% Days Target Stocks Are Bought
% Days Target Stocks Are Sold
10.0%
10.0%
8.0%
8.0%
6.0%
6.0%
4.0%
4.0%
2.0%
2.0%
0.0%
0.0%
1
Highest
3
5
7
1
Highest
9
No
Decile by Fraction of Stocks Sold Trading
in Other
Stocks
4.0%
4.0%
3.0%
3.0%
2.0%
2.0%
1.0%
1.0%
0.0%
1
Highest
3
5
7
9
No
Decile by Fraction of Stock Sold Trading
in Other
Stocks
% Days Target Stocks Are Bought
% Days Target Stocks Are Sold
3
5
7
9
No
Decile by Fraction of Sell Principal Trading
in Other
Stocks
0.0%
3
5
7
9
Decile by % of Trading Days during Sample
1
Highest
28
3
5
7
9
Decile by % of Trading Days during Sample
Figure 6: Vector Autoregression (VAR) Analysis – Hedge Fund Net Volume
These figures plot impulse-response functions (IRFs) of activist hedge fund volume to institutional sell
and buy volumes, abnormal return (AR), abnormal turnover (AT), and abnormal Amihud ratio (AA). The
IRFs are calculated using the parameter estimates for a Panel VAR model with 5 lags. A recursive
identification scheme is used, in which institutional sell volume is ordered first, followed by institutional
buy volume, hedge fund volume, AT, AR, and AA. 90%-confidence intervals are calculated by Monte
Carlo simulation. All volumes are measured as a percentage of shares outstanding. The sample includes
643 campaigns in 2000-2007. Hedge fund trading data are collected from SEC filings and non-hedge fund
institutional trades come from Ancerno.
0.100%
0.080%
Inst sell vol. → HF net vol.
90% confidence interval
Inst buy vol. → HF net vol.
90% confidence interval
0.010%
0.005%
0.060%
0.000%
0.040%
0.020%
-0.005%
0.000%
-0.010%
-0.020%
0
1
2
3
4
5
6
7
8
9
0
10
1
2
3
5
6
7
8
9 10
Lag
Lag
HF net vol. → HF net vol.
0.400%
4
AT → HF net vol.
0.030%
90% confidence interval
0.300%
90% confidence interval
0.020%
0.200%
0.010%
0.100%
0.000%
0.000%
-0.010%
-0.100%
0
1
2
3
4
5
6
7
8
9
0
10
1
2
3
5
6
7
8
9 10
Lag
Lag
AR → HF net vol.
0.030%
4
AA → HF net vol.
0.010%
90% confidence interval
0.020%
90% confidence interval
0.000%
0.010%
-0.010%
0.000%
-0.020%
-0.010%
0
1
2
3
4 5
Lag
6
7
8
0
9 10
29
1
2
3
4 5
Lag
6
7
8
9 10
Figure 7: Vector Autoregression (VAR) Analysis – Returns and Liquidity
These figures plot impulse-response functions (IRFs) of abnormal returns (AR) and abnormal turnover
(AT) to institutional sell and buy volumes and activist hedge fund net volume. The IRFs are calculated
using the parameter estimates for a Panel VAR model with 5 lags. A recursive identification scheme is
used, in which institutional sell volume is ordered first, followed by institutional buy volume, hedge fund
volume, AT, AR, and AA. 90%-confidence intervals are calculated by Monte Carlo simulation. All
volumes are measured as a percentage of shares outstanding. The sample includes 643 campaigns in
2000-2007. Hedge fund trading data are collected from SEC filings and non-hedge fund institutional
trades come from Ancerno.
Inst sell vol. → AT
1.000%
Inst sell vol. → AR
0.100%
90% confidence interval
0.800%
90% confidence interval
0.000%
0.600%
-0.100%
0.400%
-0.200%
0.200%
-0.300%
0.000%
0
1
2
3
4
5
6
7
8
9
0
10
1
2
3
Inst buy vol. → AT
0.500%
4
5
6
7
8
9
10
Lag
Lag
Inst. net vol. (buy) → AR
0.050%
90% confidence interval
90% confidence interval
0.400%
0.300%
0.000%
0.200%
0.100%
-0.050%
0.000%
-0.100%
0
1
2
3
4
5
6
7
8
9
-0.100%
10
0
1
2
3
Lag
6
7
8
9
10
HF net vol. → AR
0.300%
90% confidence interval
0.400%
0.200%
0.200%
0.100%
0.000%
0.000%
90% confidence interval
-0.100%
-0.200%
0
5
Lag
HF net vol. → AT
0.600%
4
1
2
3
4
5
Lag
6
7
8
0
9 10
1
2
3
4
5
Lag
30
6
7
8
9 10
Table 1: Hedge Fund Trading in the 60 Days before the Public Announcement of Activism
This table presents summary statistics on activist hedge funds’ trading in the target stocks. The sample
includes 643 campaigns, over the period from 2000 to 2007, for which both hedge fund trading data and
Ancerno institutional trading data are available. The upper (lower) panel reports cross-sectional means
(medians). The statistics are reported both for the entire 60-day period in which the hedge funds report their
trades and for each of the three identifiable sub-periods. For each campaign, day t-60 refers to day -60 from
the file date, and event date refers to the date on which the hedge fund’s ownership crosses the 5% threshold.
% of
Shares
Purchased
in Open
Market
N
t-60 to Event Date
Event Date
Event to File Dates
589
581
452
12.53%
41.24%
17.63%
2.65%
1.02%
1.28%
41.08%
13.68%
16.93%
94.12%
97.58%
98.61%
185
14
72
185
14
71
98.79%
97.28%
98.70%
t-60 to File Date
643
15.78%
4.25%
61.89%
98.17%
232
232
97.51%
Shares Purchased as %
of
Total
Shares
Shares on
Outstanding File Date
MEDIAN
Number of
Average
Trades
Price/
Price on
File
Open
Date
Total Market
% of
Shares
Purchased
in Open
Market
N
Trade as
% of
Market
Volume
t-60 to Event Date
Event Date
Event to File Dates
589
581
452
12.53%
41.24%
17.63%
2.48%
0.40%
0.73%
36.79%
6.36%
11.49%
96.83%
97.72%
98.20%
24
2
7
23
2
7
100.00%
100.00%
100.00%
t-60 to File Date
643
15.78%
3.81%
61.83%
97.31%
30
30
100.00%
Period
MEAN
Number of
Average
Trades
Price/
Price on
File
Open
Date
Total Market
Trade as
% of
Market
Volume
Period
Shares Purchased as %
of
Total
Shares
Shares on
Outstanding File Date
31
Table 2: Institutional Trading in Activist Targets
This table presents summary statistics on non-hedge fund institutional trading in the targets of activist
campaigns. The sample includes 643 campaigns, over the period from 2000 to 2007, for which both hedge fund
trading data and Ancerno institutional trading data are available. The upper (lower) panel reports crosssectional means (medians). The statistics are reported both for the 60-day period in which the hedge funds
report their trades and for the prior 180 days. For each campaign, day t-60 (t-240) refers to day -60 (-240) from
the file date, and event date refers to the date on which the hedge fund’s ownership crosses the 5% threshold.
Institution is a unique combination of client code and client manager code in the Ancerno data.
MEAN
Number of
Trades per
Institution
Buy Sell
N
Trade
as % of
Market
Volume
t-240 to t-60
682
13.46%
6.41%
-7.43%
-1.02%
56
70
13
11
t-60 to Event Date
Event Date
Event to File Dates
625
447
518
15.14%
20.35%
14.53%
1.93%
0.12%
0.92%
-2.93%
-0.46%
-1.28%
-1.00%
-0.34%
-0.36%
25
3
14
33
5
15
9
3
5
10
3
5
t-60 to File Date
643
14.36%
2.71%
-4.21%
-1.50%
30
40
10
10
Period
Volume/ Shares
Outstanding
Buy
Sell
Net
Number of
Institutions
Net Buy Net Sell
MEDIAN
N
t-240 to t-60
682
13.46%
3.98%
-4.63%
-0.15%
24
27
10
8
t-60 to Event Date
Event Date
Event to File Dates
625
447
518
15.14%
20.35%
14.53%
0.90%
0.01%
0.11%
-1.56%
-0.05%
-0.22%
-0.19%
-0.02%
-0.03%
12
2
5
13
2
5
6
1
3
5
1
3
t-60 to File Date
643
14.36%
1.11%
-2.10%
-0.38%
14
15
7
6
Period
Number of
Trades per
Institution
Buy Sell
Trade
as % of
Market
Volume
Volume/ Shares
Outstanding
Buy
Sell
Net
32
Number of
Institutions
Net Buy Net Sell
Table 3: Trading in Target Stocks by Top Institutional Sellers and Buyers
This table presents statistics on the top non-hedge fund institutions’ combined trading, as a percentage of
shares outstanding, in the targets of activist campaigns. The sample includes 643 campaigns in 2000-2007. In
Panel A (Panel B), top institutions are the two largest sellers (buyers) in each target stock on the event date.
In Panel C (Panel D), top institutions are the five largest sellers (buyers) during the 60-day period in which
the hedge funds report their trades. For each campaign, days t-60, t-240, and t+30 refer to days -60, -240, and
+30 from the file date, and event date refers to the date on which the hedge fund’s ownership crosses the 5%
threshold. Institution is a unique combination of client code and client manager code in the Ancerno data.
Event Window
75%
90%
t-240 to t-60
Panel A: Top 2 sellers (combined) on the event date
278
0.00%
1.81%
-0.74%
-0.04%
0.03%
0.28%
1.01%
t-60 to Event Date
Event Date
Event to File Dates
309
365
224
-0.44%
-0.44%
-0.35%
1.31%
1.72%
0.67%
-1.55%
-0.96%
-1.12%
-0.43%
-0.29%
-0.35%
-0.05%
-0.07%
-0.08%
0.00%
-0.02%
-0.01%
0.11%
0.00%
0.00%
t-60 to File Date
File Date to t+30
365
207
-1.02%
-0.38%
2.72%
0.88%
-2.79%
-1.21%
-0.95%
-0.45%
-0.25%
-0.05%
-0.03%
0.00%
0.00%
0.02%
t-240 to t-60
Panel B: Top 2 buyers (combined) on the event date
242
0.26%
1.19%
-0.08%
0.00%
0.05%
0.20%
0.68%
t-60 to Event Date
Event Date
Event to File Dates
298
343
241
0.32%
0.13%
0.17%
1.21%
0.33%
0.62%
-0.01%
0.00%
0.00%
0.01%
0.00%
0.00%
0.05%
0.02%
0.01%
0.21%
0.10%
0.09%
0.72%
0.32%
0.46%
t-60 to File Date
File Date to t+30
343
266
0.53%
0.09%
1.55%
0.57%
0.00%
-0.13%
0.02%
0.00%
0.08%
0.01%
0.43%
0.04%
1.11%
0.22%
Panel C: Top 5 sellers (combined) during t-60 to the file date
536
0.26%
2.30%
-1.62%
-0.35%
0.06%
0.81%
2.32%
t-240 to t-60
Mean
St. Dev.
10%
25%
Median
t-60 to Event Date
Event Date
Event to File Dates
578
237
371
-2.11%
-0.69%
-0.85%
3.08%
2.48%
1.88%
-5.13%
-1.72%
-2.25%
-2.62%
-0.44%
-0.72%
-1.11%
-0.13%
-0.24%
-0.38%
-0.04%
-0.05%
-0.06%
-0.01%
-0.01%
t-60 to File Date
File Date to t+30
595
367
-2.86%
-0.50%
4.03%
1.14%
-6.72%
-1.46%
-3.70%
-0.57%
-1.59%
-0.09%
-0.51%
0.00%
-0.13%
0.05%
Panel D: Top 5 buyers (combined) during t-60 to the file date
489
0.72%
1.63%
-0.32%
0.01%
0.23%
0.96%
2.64%
t-240 to t-60
N
t-60 to Event Date
Event Date
Event to File Dates
570
235
409
1.36%
0.16%
0.44%
2.21%
0.41%
1.20%
0.06%
0.00%
0.00%
0.23%
0.00%
0.01%
0.65%
0.03%
0.10%
1.64%
0.15%
0.36%
3.56%
0.38%
1.05%
t-60 to File Date
File Date to t+30
587
464
1.69%
0.01%
2.56%
0.82%
0.09%
-0.57%
0.28%
-0.09%
0.84%
0.01%
2.09%
0.16%
4.40%
0.62%
33
Table 4: Trading in Non-Target Stocks by Top Institutional Sellers and Buyers
This table presents statistics on the top non-hedge fund institutions’ trading in stocks other than the target stocks
of activist campaigns. In Panel A (Panel B), top institutions are the two largest sellers (buyers) in each target
stock on the event date. In Panel C (Panel D), top institutions are the five largest sellers (buyers) during the 60day period in which the hedge funds report their trades. For each campaign, days t-60, t-240, and t+30 refer to
days -60, -240, and +30, respectively, from the file date, and event date refers to the date on which the hedge
fund’s ownership crosses the 5% threshold. Institution is a unique combination of client code and client
manager code in the Ancerno data.
Event Window
N
Days
Traded in
Period
Sell
Principal/
Total
Principal
# Stocks
Sold/ #
Stocks
Traded
Buy Trade Sell Trade
Size
Size
($ Million) ($ Million)
Panel A: Top 2 sellers (combined) on the event date
84
49.54%
48.47%
0.418
t-240 to t-60
343
t-60 to Event Date
Event Date
Event to File Dates
364
356
346
28
1
7
49.59%
56.47%
51.11%
49.25%
58.12%
50.44%
t-60 to File Date
File Date to t+30
365
350
35
17
50.41%
50.91%
49.98%
50.94%
# Days
Traded/
Total #
Days in
Sample
0.492
53.39%
0.395
0.390
0.404
0.448
0.366
0.429
52.92%
54.26%
54.41%
0.393
0.420
0.436
0.476
53.27%
53.75%
t-240 to t-60
Panel B: Top 2 buyers (combined) on the event date
325
87
40.83%
39.68%
0.342
0.719
53.38%
t-60 to Event Date
Event Date
Event to File Dates
340
336
332
28
1
8
39.34%
31.46%
38.21%
37.62%
28.10%
35.61%
0.360
0.311
0.354
0.850
0.727
0.965
53.36%
53.85%
53.99%
t-60 to File Date
File Date to t+30
343
337
36
17
38.77%
40.11%
36.85%
38.31%
0.340
0.392
0.858
1.231
53.25%
53.59%
t-240 to t-60
Panel C: Top 5 sellers (combined) during t-60 to the file date
586
83
49.44%
48.36%
0.459
0.688
52.32%
t-60 to Event Date
Event Date
Event to File Dates
592
572
565
28
1
11
50.32%
50.58%
50.42%
49.81%
50.04%
49.98%
0.466
0.422
0.449
0.543
0.455
0.517
52.20%
56.26%
53.62%
t-60 to File Date
File Date to t+30
595
580
38
17
50.55%
49.62%
50.10%
49.22%
0.460
0.493
0.533
0.561
52.11%
52.95%
t-240 to t-60
Panel D: Top 5 buyers (combined) during t-60 to the file date
578
83
43.53%
41.86%
0.396
0.685
52.53%
t-60 to Event Date
Event Date
Event to File Dates
585
558
557
28
1
9
43.11%
44.22%
43.20%
41.18%
41.89%
41.42%
0.446
0.415
0.451
0.896
0.625
1.102
52.17%
56.21%
53.83%
t-60 to File Date
File Date to t+30
587
579
36
17
43.07%
44.19%
41.14%
42.93%
0.444
0.445
1.205
1.275
52.03%
52.61%
34
Table 5: Relationship between Hedge Fund Trading and Institutional Trading
This table reports logit estimates (Panel A) and OLS estimates (Panel B) of the effect of institutional trading in
target stocks on activist hedge fund trading. The sample includes 643 campaigns, over the period from 2000 to
2007, for which both hedge fund trading data and Ancerno institutional trading data are available. In Panel A,
the dependent variable is a dummy that equals one if the activist hedge fund trades on a target-day, and zero
otherwise. In Panel B, the dependent variable is the total number of shares purchased by the hedge fund on
each target-day, as a percentage of shares outstanding. Standard errors, clustered by campaign, are in
parentheses. *, **, and *** refer to statistical significance at 10%, 5%, and 1% levels, respectively.
Panel A: Logit estimates for daily hedge fund trade dummy
(1)
Dummy (inst net sell)
Dummy (inst net buy)
(2)
(3)
(4)
(5)
(6)
130.697***
(15.924)
2.698
(25.558)
89.634***
(17.129)
-34.260
(26.341)
1.004*
(0.529)
15.389***
(2.658)
-2.831***
(0.403)
153.490***
(16.402)
1.411
(25.555)
-0.009
(0.007)
-0.007
(0.007)
85.076***
(13.748)
-44.560**
(22.451)
-0.283
(0.603)
15.168***
(2.218)
-2.738***
(0.300)
2.004***
(0.047)
0.741***
(0.045)
0.442***
(0.049)
0.247***
(0.049)
0.297***
(0.048)
-0.002
(0.003)
81.391***
(15.629)
-67.229***
(25.890)
-0.187
(0.688)
23.067***
(3.142)
-4.350***
(0.370)
1.710***
(0.047)
0.554***
(0.044)
0.282***
(0.047)
0.068
(0.048)
0.032
(0.048)
0.039***
(0.012)
0.494***
(0.086)
0.316***
(0.092)
Inst sell volume/SHROUT
Inst buy volume/SHROUT
Abnormal return
Abnormal turnover
Abnormal Amihud ratio
Dummy (HF trade) l1
Dummy (HF trade) l2
Dummy (HF trade) l3
Dummy (HF trade) l4
Dummy (HF trade) l5
VIX
Campaign-level controls
N
Pseudo R-squared
-0.006
(0.007)
CAR (t-240 to t-60), CAT (t-240 to t-60),
CAA (t-240 to t-60)
27,684
0.009
27,684
0.007
27,215
0.022
35
27,215
0.315
0.053***
(0.016)
Campaign dummies
30,602
0.010
27,179
0.214
Table 5, cont’d: Relationship between Hedge Fund Trading and Institutional Trading
Panel B: OLS estimates for daily net hedge fund volume as a percentage of shares outstanding
(1)
Dummy (inst net sell)
Dummy (inst net buy)
(2)
(3)
(4)
(5)
(6)
0.339***
(0.034)
0.003
(0.028)
0.171***
(0.041)
-0.144***
(0.035)
0.004*
(0.002)
0.060***
(0.013)
-0.004***
(0.001)
0.348***
(0.036)
0.034
(0.027)
0.000
(0.000)
0.000
(0.000)
0.172***
(0.042)
-0.147***
(0.037)
0.004*
(0.002)
0.059***
(0.014)
-0.003***
(0.001)
0.108***
(0.023)
0.048***
(0.012)
0.010*
(0.006)
0.031***
(0.010)
0.013
(0.010)
0.000
(0.000)
0.170***
(0.051)
-0.156***
(0.043)
0.003
(0.003)
0.068***
(0.016)
-0.005***
(0.001)
0.087***
(0.023)
0.030**
(0.012)
-0.005
(0.006)
0.017*
(0.010)
-0.002
(0.010)
0.000
(0.000)
0.0005***
(0.0001)
0.0001
(0.0001)
Inst sell volume/SHROUT
Inst buy volume/SHROUT
Abnormal return
Abnormal turnover
Abnormal Amihud ratio
HF volume/SHROUT l1
HF volume/SHROUT l2
HF volume/SHROUT l3
HF volume/SHROUT l4
HF volume/SHROUT l5
VIX
Campaign-level controls
N
Pseudo R-squared
0.000
(0.000)
CAR (t-240 to t-60), CAT (t-240 to t-60),
CAA (t-240 to t-60)
27,684
0.004
27,684
0.017
27,215
0.105
36
27,215
0.129
0.000
(0.000)
Campaign dummies
30,602
0.016
27,179
0.128
Table 6: Persistence of Hedge Fund and Institutional Trading
This table reports coefficient estimates of regressions of activist hedge fund trading and institutional
trading on their lags. The sample includes 643 campaigns, over the period from 2000 to 2007, for which
both hedge fund trading data and Ancerno institutional trading data are available. All regressions include
VIX, campaign fixed effects, and up to 5 lags of trading variables. Standard errors are clustered by
campaign but are omitted for brevity. *, **, and *** refer to statistical significance at 10%, 5%, and 1%
levels, respectively.
(1)
Dummy
(HF
trade)
Dummy (HF trade) l1
Dummy (HF trade) l2
Dummy (HF trade) l3
Dummy (HF trade) l4
Dummy (HF trade) l5
(4)
HF
volume/
SHROUT
Inst sell
volume/
SHROUT
Inst buy
volume/
SHROUT
0.121***
0.042***
0.002
0.026**
0.005
0.007**
0.001
0.003
0.003
0.003
0.001
0.000
0.002
0.002
-0.001
1.759***
0.568***
0.293***
0.103**
0.044
HF volume/SHROUT l1
HF volume/SHROUT l2
HF volume/SHROUT l3
HF volume/SHROUT l4
HF volume/SHROUT l5
Dependent Variable
(2)
(3)
Inst sell volume/SHROUT l1
Inst sell volume/SHROUT l2
Inst sell volume/SHROUT l3
Inst sell volume/SHROUT l4
Inst sell volume/SHROUT l5
3.926
4.306
-1.734
7.593
7.150
0.007
0.005
0.002
0.006
0.012
0.051***
0.016***
0.006**
0.002
0.009***
-0.007**
-0.001
-0.000
-0.002
0.002
Inst sell volume/SHROUT l1
Inst sell volume/SHROUT l2
Inst sell volume/SHROUT l3
Inst sell volume/SHROUT l4
Inst sell volume/SHROUT l5
-7.695
-0.144
4.108
10.150*
-1.131
-0.011
-0.014**
-0.004
0.000
-0.001
-0.020
-0.010**
0.001
-0.004
-0.009***
0.048**
0.016**
0.009
0.016**
0.004
Model
N
Pseudo R-squared/ R-squared
LOGIT
30,580
0.192
OLS
27,502
0.022
OLS
27,502
0.031
OLS
27,502
0.025
37
Table 7: Relationship between Individual Institution’s Trading in Target and Non-Target Stocks
This table reports multinomial logit estimates (Panel A) and OLS estimates (Panel B) of the probability
that each institution will buy, sell, or not trade the target stocks conditional on its trading in other nontarget stocks. The sample includes 6,035 institutions that trade in 643 activist campaigns over the period
from 2000 to 2007. For each campaign, institutions are included only if they trade at least twice during
the 60-day period in which the hedge funds report their trades. Each observation is stock-institution-day.
In Panel A, the odds of buy and sell are estimated relative to no trading (reference outcome). Campaign
characteristics are controlled for using the target’s prior six-month cumulative abnormal return (CAR),
cumulative abnormal turnover (CAT), and cumulative abnormal Amihud ratio (CAA). In Panel B, the
probabilities of buy and sell are estimated separately. Campaign characteristics are controlled for using
campaign fixed-effects. Standard errors, clustered by campaign, are in parentheses. *, **, and *** refer
to statistical significance at 10%, 5%, and 1% levels, respectively.
Panel A: Multinomial logit estimates for individual institution’s buy/sell in target stocks
Sell Fraction = Fraction of
Sell Principal
(1)
(2)
Buy
Sell
Dummy (trade other stocks)
Dummy (trade other stocks)
* Sell fraction
Dummy (trade only one other stock)
Dummy (sell target) l1
Dummy (buy target) l1
Dummy (trade other stocks) l1
Dummy (trade other stocks) l1
* Sell fraction l1
Fraction of trading days during sample
CRSP value-weighted return
VIX
Campaign-level controls
N
Pseudo R-squared
4.194***
(0.030)
-1.790**
(0.019)
-1.764**
(0.027)
0.224***
(0.026)
2.637***
(0.012)
-1.065**
(0.018)
0.187***
(0.019)
-0.310**
(0.037)
3.381***
(0.600)
-0.018**
(0.001)
2.331***
(0.028)
1.119***
(0.017)
-1.352**
(0.024)
2.956***
(0.011)
0.253***
(0.026)
-0.512**
(0.020)
-0.376**
(0.018)
-0.401**
(0.035)
0.024
(0.586)
-0.016**
(0.001)
Sell Fraction = Fraction
of Stocks Sold
(3)
(4)
Buy
Sell
4.466***
(0.030)
-2.810**
(0.022)
-1.939**
(0.027)
0.164***
(0.027)
2.654***
(0.013)
-1.202**
(0.019)
0.653***
(0.020)
-0.285**
(0.037)
2.691***
(0.605)
-0.016**
(0.001)
1.856***
(0.029)
1.909***
(0.020)
-1.488**
(0.025)
3.017***
(0.012)
0.165***
(0.026)
-0.259**
(0.020)
-0.801**
(0.020)
-0.340**
(0.036)
0.326
(0.591)
-0.016**
(0.001)
CAR (t-240 to t-60), CAT (t - 240 to t-60),
CAA (t-240 to t-60) in all models
945,861
0.249
38
945,861
0.268
Table 7, cont’d: Relationship between Individual Institution’s Trading in Target and Non-Target
Stocks
Panel B: OLS estimates for individual institution’s buy/sell in target stocks
Sell Fraction = Fraction of
Sell Principal
(1)
(2)
Buy
Sell
Dummy (trade other stocks)
Dummy (trade other stocks)
* Sell fraction
Dummy (trade only one other stock)
Dummy (sell target) l1
Dummy (buy target) l1
Dummy (trade other stocks) l1
Dummy (trade other stocks) l1
* Sell fraction l1
Fraction of trading days during sample
CRSP value-weighted return
VIX
0.143***
(0.009)
-0.101**
(0.006)
-0.063**
(0.004)
-0.011**
(0.002)
0.314***
(0.009)
-0.051**
(0.004)
0.011***
(0.002)
-0.003
(0.003)
0.171***
(0.053)
0.000
(0.000)
Campaign-level controls
N
R-squared
0.046***
(0.004)
0.070***
(0.005)
-0.051**
(0.004)
0.368***
(0.012)
-0.006**
(0.002)
-0.022**
(0.002)
-0.023**
(0.003)
-0.013**
(0.003)
0.044
(0.066)
0.000
(0.000)
Sell Fraction = Fraction
of Stocks Sold
(3)
(4)
Buy
Sell
0.164***
(0.010)
-0.148**
(0.009)
-0.062**
(0.004)
-0.011**
(0.002)
0.312***
(0.010)
-0.060**
(0.005)
0.030***
(0.003)
-0.004
(0.003)
0.147***
(0.051)
0.000
(0.000)
0.027***
(0.003)
0.113***
(0.007)
-0.052**
(0.004)
0.369***
(0.012)
-0.006**
(0.002)
-0.012**
(0.002)
-0.045**
(0.003)
-0.013**
(0.003)
0.053
(0.065)
0.000
(0.000)
Campaign dummies in all models
945,861
0.162
39
945,861
0.197
945,861
0.173
945,861
0.204
Table 8: Instrumenting Daily Trading in Activist Targets by Individual Institution’s Daily Trading
in Other Non-Target Stocks
This table reports 2SLS estimates of the effect of institutional trading in target stocks on activist hedge
fund trading. The sample includes 643 campaigns, over the period from 2000 to 2007, for which both
hedge fund trading data and Ancerno institutional trading data are available. The dependent variable is
the total number of shares purchased by the hedge fund on each target-day, as a percentage of shares
outstanding. In columns (1) and (2), institutional sell and buy volumes are instrumented by their lags,
plus expected institutional trading by all institutions. The expected trading volume for each institution is
calculated based on the models in Panel A of Table 7, in which trading in non-target stocks is used to
identify exogenous liquidity-motivated trades. In columns (3) and (4), contemporaneous abnormal return
(AR), abnormal turnover (AT), and abnormal Amihud ratio (AA) are also considered endogenous and are
instrumented by their own lags. Standard errors, clustered by campaign, are in parentheses. *, **, and
*** refer to statistical significance at 10%, 5%, and 1% levels, respectively.
Pred (Inst sell volume/SHROUT)
Pred (Inst buy volume/SHROUT)
(1)
(2)
(3)
(4)
0.460***
(0.062)
0.027
(0.072)
0.464***
(0.062)
0.044
(0.074)
0.000
(0.000)
0.000
(0.000)
0.137**
(0.069)
-0.212**
(0.085)
-0.001
(0.012)
0.015**
(0.007)
-0.006**
(0.003)
0.118***
(0.025)
0.041***
(0.015)
0.000
(0.007)
0.023**
(0.010)
0.003
(0.010)
0.000***
(0.000)
0.135**
(0.068)
-0.177**
(0.086)
-0.001
(0.012)
0.014**
(0.007)
-0.006**
(0.003)
0.118***
(0.025)
0.041***
(0.015)
0.000
(0.007)
0.023**
(0.010)
0.003
(0.011)
0.000***
(0.000)
Pred (Abnormal return)
Pred (Abnormal turnover)
Pred (Abnormal Amihud ratio)
HF volume/SHROUT l1
HF volume/SHROUT l2
HF volume/SHROUT l3
HF volume/SHROUT l4
HF volume/SHROUT l5
VIX
Sell fraction in other stocks used to
construct instrument
Other instruments
Observations
R-squared
Fraction
Fraction
of sell
of stocks
principal
sold
5 lags of Inst sell
volume/SHROUT and
Inst buy
volume/SHROUT
30,654
0.015
40
30,654
0.014
Fraction
Fraction
of sell
of stocks
principal
sold
5 lags of Inst sell
volume/SHROUT, Inst
buy volume/SHROUT,
Abnormal return,
Abnormal turnover, and
Abnormal Amihud ratio
23,252
23,252
0.069
0.069
Table 9: Abnormal Returns and Liquidity of Target Stocks
This table presents summary statistics on the abnormal returns, abnormal turnover, and abnormal Amihud
ratio of stocks targeted in activist campaigns. The sample includes 643 campaigns, over the period from
2000 to 2007, for which both hedge fund trading data and Ancerno institutional trading data are available.
Abnormal returns are calculated by the market-model adjustment approach, in which the CRSP valueweighted index is used as the market portfolio. Abnormal turnover (Amihud ratio) is calculated by
subtracting the mean turnover (Amihud ratio) from the period turnover (Amihud ratio). The estimation
period is from t-600 to t-240 days before the public announcement of activism (t=0 is the file date).
Means are calculated first across all days in the period for each stock and then across all stocks. The
means are reported both for the 60-day period in which the hedge funds report their trades and for the
prior 180 days. For each campaign, day t-60 (t-240) refers to day -60 (-240) from the file date, and event
date refers to the date on which the hedge fund’s ownership crosses the 5% threshold.
Period
t-240 to t-60
740
-1.79%
t-60 to Event Date
Event Date
Event to File Dates
640
630
613
0.81%
0.56%
3.21%
0.15%
0.56%
0.56%
0.01%
t-60 to File Date
643
4.42%
0.35%
Period
0.06%
0.61%
-0.04%
0.30%
0.42%
0.11%
0.77%
0.58%
0.06%
0.07%
0.20%
Panel B: Abnormal Turnover
Hedge Funds
N
All Days Trade No Trade
740
0.001
t-60 to Event Date
Event Date
Event to File Dates
640
630
613
0.002
0.021
0.005
0.009
0.021
0.007
0.001
t-60 to File Date
643
0.003
0.011
740
-0.001
t-60 to Event Date
Event Date
Event to File Dates
640
630
613
-0.004
-0.030
-0.013
-0.020
-0.030
-0.021
t-60 to File Date
643
-0.006
-0.023
41
Institutions
Net Sell Others
0.003
0.000
0.005
0.006
0.030
0.008
0.001
0.014
0.003
0.001
0.007
0.002
Panel C: Abnormal Amihud Ratio
Hedge Funds
N
All Days Trade No Trade
t-240 to t-60
Institutions
Net Sell Others
-0.14%
t-240 to t-60
Period
Panel A: Abnormal Returns
CARs
Hedge Funds
N
All Days Trade No Trade
Institutions
Net Sell Others
-0.006
0.000
-0.009
-0.012
-0.024
-0.014
-0.002
-0.035
-0.012
0.000
-0.014
-0.004
0.002
Table 10: Institutional Trading, Abnormal Returns, and Liquidity (t-240 to t-60)
This table reports coefficient estimates of regressions of target stocks’ abnormal returns, abnormal turnover,
and abnormal Amihud ratio on variables that reflect non-hedge fund institutional trading activity during the
period from t-240 to t-60 days before the public announcement of activist campaigns (t=0 is the file date).
The sample includes 643 campaigns, over the period from 2000 to 2007, for which both hedge fund trading
data and Ancerno institutional trading data are available. Abnormal returns are calculated by the marketmodel adjustment approach, in which the CRSP value-weighted index is used as the market portfolio.
Abnormal turnover (Amihud ratio) is calculated by subtracting the mean turnover (Amihud ratio) from the
period turnover (Amihud ratio). The estimation period is from t-600 to t-240 days before the file date. All
models include campaign fixed effects. Standard errors, clustered by campaign, are in parentheses. *, **,
and *** refer to statistical significance at 10%, 5%, and 1% levels, respectively.
Dummy (inst net sell)
Dummy (inst net buy)
Inst sell volume/SHROUT
Inst buy volume/SHROUT
Dependent variable l1
Dependent variable l2
Dependent variable l3
Dependent variable l4
Dependent variable l5
VIX
(1)
(2)
Abnormal
Return
Abnormal
Return
-0.002***
(0.001)
0.002***
(0.001)
Abnormal
Turnover
Abnormal
Turnover
0.002***
(0.000)
0.001***
(0.000)
-0.699**
(0.321)
0.002
(0.117)
-0.051*** -0.045***
(0.012)
(0.012)
-0.023*** -0.024***
(0.008)
(0.008)
-0.021** -0.023***
(0.008)
(0.008)
-0.016**
-0.017**
(0.008)
(0.008)
-0.017*** -0.015***
(0.006)
(0.005)
-0.000*** -0.000***
(0.000)
(0.000)
Campaign-level controls
N
R-squared
Dependent Variable
(3)
(4)
0.358***
(0.087)
0.084***
(0.013)
0.033***
(0.010)
0.055***
(0.011)
0.022**
(0.010)
0.000
(0.000)
(5)
(6)
Abnormal
Amihud
Abnormal
Amihud
-0.014***
(0.001)
-0.014***
(0.001)
2.179***
(0.274)
1.203***
(0.315)
0.353***
(0.077)
0.072***
(0.017)
0.030***
(0.011)
0.049***
(0.011)
0.020**
(0.008)
-0.000
(0.000)
0.131***
(0.008)
0.077***
(0.007)
0.054***
(0.007)
0.050***
(0.007)
0.045***
(0.007)
0.001***
(0.000)
-0.756***
(0.121)
-0.494***
(0.081)
0.134***
(0.008)
0.078***
(0.007)
0.054***
(0.007)
0.052***
(0.007)
0.045***
(0.007)
0.001***
(0.000)
Campaign dummies in all models
75,667
0.007
76,667
0.006
42
76,667
0.292
76,667
0.340
70,688
0.052
70,688
0.049
Table 11: Institutional Trading, Hedge Fund Trading, Abnormal Returns, and Liquidity (t-60 to t)
This table reports coefficient estimates of regressions of target stocks’ abnormal returns, abnormal turnover,
and abnormal Amihud ratio on variables that reflect institutional trading activity during the 60-day period
leading up to the public announcement of activist campaigns (t=0 is the file date). The sample includes 643
campaigns, over the period from 2000 to 2007, for which both hedge fund trading data and Ancerno
institutional trading data are available. Abnormal returns are calculated by the market-model adjustment
approach, in which the CRSP value-weighted index is used as the market portfolio. Abnormal turnover
(Amihud ratio) is calculated by subtracting the mean turnover (Amihud ratio) from the period turnover
(Amihud ratio). The estimation period is from t-600 to t-240 days before the file date. All models include
campaign fixed effects. Standard errors, clustered by campaign, are in parentheses. *, **, and *** refer to
statistical significance at 10%, 5%, and 1% levels, respectively.
Dummy (inst net sell)
Dummy (inst net buy)
Dummy (HF trade)
(1)
(2)
Abnormal
Return
Abnormal
Return
-0.002***
(0.001)
0.001**
(0.001)
0.001***
(0.001)
Inst sell volume/SHROUT
Inst buy volume/SHROUT
HF volume/SHROUT
Dependent variable l1
Dependent variable l2
Dependent variable l3
Dependent variable l4
Dependent variable l5
VIX
-0.045***
(0.014)
-0.026***
(0.009)
-0.029***
(0.009)
-0.020**
(0.009)
-0.020***
(0.008)
0.000
(0.000)
Campaign-level controls
N
R-squared
Dependent Variable
(3)
(4)
Abnormal
Turnover
Abnormal
Turnover
0.003***
(0.000)
0.002***
(0.000)
0.004***
(0.000)
-0.648**
(0.271)
0.031
(0.340)
0.676***
(0.208)
-0.030**
(0.014)
-0.024**
(0.010)
-0.038***
(0.013)
-0.031***
(0.011)
-0.013
(0.013)
0.000
(0.000)
0.276***
(0.033)
0.012
(0.015)
0.014*
(0.008)
0.035*
(0.018)
0.010
(0.008)
0.000
(0.000)
(5)
(6)
Abnormal
Amihud
Abnormal
Amihud
-0.013***
(0.002)
-0.012***
(0.002)
-0.015***
(0.001)
1.365***
(0.363)
1.094
(0.674)
1.383***
(0.167)
0.255***
(0.032)
0.004
(0.015)
0.001
(0.007)
0.026
(0.016)
0.004
(0.008)
0.000
(0.000)
0.122***
(0.013)
0.081***
(0.012)
0.039***
(0.013)
0.042***
(0.012)
0.063***
(0.012)
0.001***
(0.000)
-0.204**
(0.108)
-0.304**
(0.140)
-1.694***
(0.183)
0.127***
(0.013)
0.084***
(0.012)
0.042***
(0.013)
0.042***
(0.012)
0.062***
(0.012)
0.001***
(0.000)
Campaign dummies in all models
27,640
0.006
27,640
0.011
43
27,640
0.174
27,640
0.278
26,121
0.061
26,121
0.058