1

Institutional Trading around Repurchase Announcements:

An Uphill Battle

Pankaj Jain

Suchismita Mishra

Vinh Huy Nguyen

Abstract

Share repurchases create an asymmetric information environment for institutional

investors. The firm and its insiders know the announcement’s timing and enjoy regulatory

exemptions from securities law violations for the timing and pricing of share repurchases

implementation or lack thereof. Institutions do not have this information ex-ante, but do they have

the foresight to trade profitably? We test and cannot reject lack of institutional foresight around

repurchases using quarterly 13F institutional holding changes, daily intermarket sweep orders,

Abel Noser institutional trades, and biweekly short interest data. These results hold with control

sample analyses as well as after addressing potential endogeneity issues.

Key Words: Intermarket sweep order, Abel Noser, 13F, short seller, repurchase announcement,

institutional investor

JEL Classification: D82, G14, G20, G35

Pankaj K. Jain, University of Memphis, Fogelman College of Business and Economics, 3675

Central Avenue, Memphis, TN 38152, pankaj.jain@memphis.edu, 901-678-3810

Suchismita Mishra, Florida International University, College of Business Administration, 11200

S.W. 8th Street RB 204BA, Miami, Florida 33199, mishras@fiu.edu, 305-348-4282

Vinh Huy Nguyen, California State University, Fresno, Craig School of Business, 5245 N Backer

Ave. M/S PB7, Fresno, CA 93740, VinhN@csufresno.edu, 559-278-8214

For helpful comments and suggestions, we thank the seminar participants at the American

University, University of Miami, University of Utah, University of Memphis, Xavier Institute of

Management, Bhubaneswar, Indian Institute of Management, Lucknow, the 2015 Florida Finance

Conference, the 2016 Eastern Finance Association Conference, the 2016 Global Finance

Conference, the 2016 Financial Management Association Conference, and the 2018 Western

Decision Sciences Institute Annual Meeting.

2

1. Introduction

In the strategic trading model developed by Holden and Subrahmanyam (1992), there are

multiple informed traders competing aggressively, and in the process, they quickly reveal their

commonly shared information. Their private information is rapidly incorporated into stock prices

and the market is assumed to be strong-form efficient. This means that in a multi-period scenario

traders with private information do not have an information advantage to be able to sustain

profitable trading. We empirically test this model using share repurchase announcements as the

event with multiple periods: the pre-announcement, announcement, and post-announcement

period. The informed agents are the institutional investors, registered insiders, and the firm.

Outside the context of repurchases, several papers document the superior trading profitability

separately for institutions (e.g., Dechow et al., 2001; Ke and Petroni, 2004; and Christophe et al.,

2004; Chemmanur et al., 2009; Chemmanur et al., 2010; Chemmanur et al., 2015; Chemmanur

and He, 2016; Chakravarty et al., 2012; McInish et al., 2014;) and insiders (e.g., Lee et al., 1992

and Alldredge and Cicero, 2015). However, those situations do not create competition among

important informed traders as share repurchases announcements do. Unlike other corporate events

for which the firm itself does not engage in buying its own stocks, all three participants actively

participate around share repurchase announcements. Institutional investors are important because

they hold approximately 75 percent of U.S. stocks (Alexander et al., 2014). Share repurchase

announcements provide a good opportunity for insiders to sell their accumulated shares as the

stock prices increase due to the significant positive announcement effect (Ofer and Thakor, 1987;

Ikenberry et al., 1995; Peyer and Vermaelen, 2009; Bargeron et al., 2011). For the firm, this

corporate event is the only time it can buy back its shares, and it has the timing advantage.

Essentially, this is a battle of champions between the institutional investors, insiders and the firm.

Hoberg et al. (2018) note that funds generate future alpha when they face less competition from

3

rival institutions. However, we posit that at the time of share repurchases the institutions face

competition from the corporations themselves and their insiders even if the competition from the

rival institutions is low. Hence, we ask the question if the institutions have the foresight to trade

profitably around repurchase announcements when they are pitted against these other informed

and active players. Alternatively, will it be a three-way tie because in an efficient market all

information advantage is competed away rapidly?

The answers to these questions are further complicated by the repurchasing strategy of the

firm. While other corporate events, such as earnings announcements, tend to resolve uncertainty

(Lee et al., 1993; Jiang et al., 2012), share repurchase announcements seem to complicate matters

further for other market participants because the announcing firms are not required to follow

through with actual repurchases. In our sample, 72% of the announcements are followed through

with the actual implementation of share repurchases within two years, but the other 28% are not.

Some of these announcement-only firms may not need to follow through with actual repurchases

because the announcement may itself result in valuation correction. According to Bhattacharya

and Jacobsen (2016), the announcement can generate enough interest to the firm that retail and

institutional trading will adjust the stock price back to the equilibrium for heavily undervalued

firms. In the case that there is a large mispricing, the firm may choose to repurchase shares, but it

does not necessarily need to buy back the shares. For slightly undervalued firms, the announcement

is not enough, and the firm has to follow through with actual buybacks implementing a costly

signal. With overvalued firms, they should not announce since it may attract unwanted attention

from investors. Yet, the overvalued firms may still announce and repurchase shares for other

reasons highlighted in the literature. Asquith and Mullins (1986), Ikenberry and Vermaelen (1996),

Dittmar (2000), Lee and Rui (2007), Bhana (2007), De Ridder (2009), Lee et al. (2010), and

4

Punwasi and Brijlal (2016) indicate that some ulterior motives behind repurchase may include

reducing taxes, distributing excess cash, adjusting capital structure, creating an attractive selling

opportunity for insiders, and eliminating threats of unwanted takeover attempts. Thus, institutions

may not know for sure that the announcement is due to undervaluation or for other purposes. In

contrast, for the insiders and the firm, this uncertainty does not exist. In summary, the share

repurchase environment is quite complicated for institutional traders because the true intention of

the repurchase announcement is not clear. By analyzing this heightened uncertainty around

repurchase announcements, we are able to meaningfully extend the literature on trading activity

around repurchases that has focused mainly on actual repurchase implementations (e.g., De Lisle

et al., 2014).

Using a diverse set of data sources, we find that institutional investors’ trades are not

profitable because they are unable to overcome the information advantage of the firm and its

insiders with little notice about when the announcement will occur1. The timing of the repurchase

announcement appears to create an environment that is difficult for institutional investors to trade

profitably. Institutional investors are selling around the pre-announcement (t=-1) and

announcement (t=0) periods when they should be buying more shares to take advantage of the

post-announcement (t=+1) price run-up. An alternative interpretation of our findings is that

institutional investors sell because they have captured their target profit from a long-term

investment perspective. This explanation assumes that institutional investors have purchased

shares long before the announcement, and they view the event as a good time to capture the

required profit for their investment. Rather than making this assumption in our paper, we analyze

1 Chakrabarty, Moulton, and Trzcinka (2017) evaluate institutional trading performance outside the context of corporate announcements by

examining short-term, round-trip trades of money managers and pension funds. They question the prior literature on short-term trading skills of institutional investors and do not find them to be profitable in the short run due to behavioral biases, such as trading to look active and the

recency bias.

5

profitability as the institutional investor’s ability to earn a significant and positive return given the

evidence from prior research of a post-announcement price run-up (Stephens and Weisbach, 1998;

Babenko, 2009).

Our paper differs from prior research on repurchases in several important ways. First, we

focus on the complexity and dual nature of the repurchase signals due to the presence of the firms

and insiders as highly informed participants in the market. In De Cesari et al. (2012), the authors

explain that the repurchasing firms can buy their shares back at a bargain price if the firms have

little institutional interest. Without institutional involvement, the firms can take further advantage

of the information asymmetry by buying back shares from less informed retail traders. Also, Louis

and White (2007) find that insiders use fixed-priced tender offers to signal undervaluation and

Dutch-auction tender offers to reduce the actual repurchasing price. Therefore, we believe the

inclusion of the firm and its insiders will enrich our understanding of institutional trading ability.

Second, we focus on the sophisticated institutional investors’ decisions around the

announcement time, not around the actual implementation of share repurchases. Based on the

findings of De Lisle et al. (2014), institutional investors are active around actual repurchases; they

are net sellers when the firms are implementing repurchases. The authors attribute this trading

strategy mostly to the information asymmetry between institutions and individual investors. Our

contribution is different from theirs as they analyze institutional trading around the repurchase

implementation period, which resolves the uncertainty created by the initial repurchase

announcement. Because institutional investors hold the majority of U.S. equities, they would serve

as liquidity providers and take the opposite side of the firm’s actual repurchases for the right price

at the time of repurchase implementation, especially when individual investors are unwilling to

tender their shares. Also, on the liquidity demand side when institutions need to move large chunks

6

of investments, repurchase events often create counterparty trading interest and liquidity that they

need to execute large trades. In a non-repurchase related working paper by Chakravarty and Ray

(2018), the authors explain that liquidity trading is a significant motive for non-profit trading,

which could explain why institutions would sell their shares around actual repurchases.

The third contribution of our paper is that we directly examine institutions’ own

profitability around the announcements, including the announcement period (t=0). In a working

paper closely related to ours, Chemmanur et al. (2018) examine information production by Abel

Noser institutions, including hedge funds around open market share repurchase programs to profit

during the post-announcement quarters, Q+1 to Q+4. Their paper highlights the importance of

examining the role played by institutions in determining the market reaction to corporate

repurchase announcements. We contribute to their line of inquiry in several complementary ways.

Even if institutions produce information, it may be challenging to profit from it at the

announcement (t=0) because of the difficulties in timing the market in relation to price changes

and trading actions of the announcing firms and its insiders. Additionally, institutions face capital

constraints and diversification constraints that may prevent them from buying the stock of an

undervalued repurchase announcer; they also face short-selling constraints which may prevent

them from selling the stock of an overvalued repurchase announcer (Saar, 2001; Chiyachantana et

al., 2017). Such constraints are less likely to apply to repurchasing firm or even its insiders.

Fourth, given that the Abel-Noser institutions cover only about 15% of the overall

institutions (Hu et al. 2018), we broaden the sample of sophisticated institutions to include all 13F

institutions, intermarket sweep order traders in TAQ, and short sellers. Thus, we cover institutional

trading strategies related to both long and short positions as well as long and short horizons. We

also verify that our findings are robust to the inclusion of non-announcing firms in the control

7

sample analyses as well as after addressing potential endogeneity issues using instrumental

variables.

Besides adding to the repurchase and institutional investor literature, we believe our

research has important regulatory implications. Given that institutional investors are informed,

their inability to properly time their trades tells us that the firm and insiders may have an unbeatable

strategy. In fact, share repurchases have gained attention from regulators and lawmakers, and they

are considering curbing these advantages. Robert Jackson Jr., a member of the Securities and

Exchange Commission, believes that the exemptions firms have from securities law violations for

the timing and pricing of share repurchases should be reconsidered (Cox, 2018). Our research

supports the idea that in this competitive environment the firm’s advantages prevent investors from

gaining any significant profits. The competition appears to be an uphill battle for institutional

investors.

2. Hypothesis Development & Research Design

In contrast to the monopolistic informed trader model developed by Kyle (1985), the multi-

period model of Holden and Subrahmanyam (1992) relaxes this restriction and allows for at least

two informed agents. In this model, prices rapidly incorporate any information as soon as

competitive informed agents take action. The market is strong-form efficient in which the

advantage of having private information quickly dissipates. We believe this model holds true when

there is no uncertainty surrounding the information signal. This is not the case with share

repurchase announcements because the first source of uncertainty comes from the unknown timing

of the announcement, which has been linked to lower market efficiency (Bagnoli and Watts, 1998).

While there may be a way to identify when the announcement might occur by examining the

average volatility spreads using daily options trading data according to Hao (2016), share

8

repurchase announcements are still considered non-routine when compared to other corporate

events, such as earnings and dividend announcements. Secondly, the announcement signal

regarding the value of the firm may not be entirely true. For example, the information signaling

hypothesis in the share repurchase literature posits that if the firm believes its shares are

undervalued, it can signal such information to the market using repurchase announcements (Miller

and Rock, 1985; Vermaelen, 1981). However, the announcement may not be related to

undervaluation signaling. Overvalued firms may engage in share repurchases for other reasons. It

is possible that the announcing firms are employing a “signal-jamming” strategy considering that

a large proportion of announcement events are not followed through with actual share repurchases.

According to Fudenberg and Tirole (1986), firms can interfere with other participants’

information-gathering and decision-making process by providing signals that may not be entirely

true about the firm. It is also possible that the firm’s repurchase decision is related to funding

corporate acquisitions, managing the dilutive effects of employee stock options, boosting the

reported EPS, reducing excess cash available to management, and inhibiting overinvesting

(Jensen, 1986; Bens et al., 2003; Grullon and Michaely, 2004; Skinner, 2008). These disturbances

or noisy information may lead others to make poor decisions with unfavorable outcomes (Mirman

et al., 1993). In the context of share repurchases, the institutional investors are the information

gatherers and their decision-making process is affected by the signals sent by the firm and its

insiders. These dual signals can help the institutions make profitable trades if this information

accurately portrays the status of the firm. While there are papers related to undervaluation signaling

or signal jamming, our paper is unique because we are the first to study the interaction between

the institutions, firm, and insiders around repurchase announcements when all three parties are

actively competing against each other. We fill this gap in the literature by explaining if the

9

institutions have the skills to decipher the complex and compounded nature of the information

signaled by the firm and its insiders.

[Insert Figure 1]

For the purpose of developing our formal hypotheses, we consider a four-period timeline

in Figure 1. The time t = 0 represents the repurchase announcement. At the pre-announcement

period, time t = -1, if the firm is undervalued, then it would make more sense to announce share

repurchases. However, depending on the motives of the insiders, the firm may announce its

repurchase event when the firm is overvalued or fairly valued at time t = -1. Perhaps, the decision

to announce share repurchases even when the firm is overvalued may be a result of perceived

undervaluation by the managers (Chen et al., 2018). The other signal about the firm’s

undervaluation or overvaluation is contained in the insider’s decision to buy or sell, respectively,

at time t = -1. We expect prices to appreciate after the announcement at time t = 1 irrespective of

the undervaluation or overvaluation of the firm (Chan et al., 2010). For the overvalued firms, prices

would depreciate at time t = 2 because of the lack of follow through. Although the firm will not

follow through in an overvalued scenario, it may still announce repurchases for other ulterior

motives mentioned previously. However, these reasons are unrelated to undervaluation, and

therefore, the stock price would still adjust back to the pre-announcement equilibrium. For the

truly undervalued firms, prices would continue to appreciate at time t = 2 in response to the actual

implementation of repurchases.

Against the null hypothesis of no effects, if the institutional traders have foresight, then our

alternative hypothesis is that they purchase shares aggressively at time t=-1 or t=0 before the post-

announcement price run-up. Similarly, with good foresight, we expect the institutions to sell shares

of overvalued firms at time t = 1 and sell shares of initially undervalued firms at time t = 2 when

10

the stocks have reached full price appreciation due to the actual implementation of repurchases.

While there is ample evidence in the literature that institutions are skilled traders, we question their

ability to replicate the same success around share repurchase announcements when the insiders

and firm, who control the information release, actively participate in the market by executing their

own trades.

Hypothesis 1 (null form): Institutional investors do not trade profitably around share

repurchases announcements. The direction, pricing, and timing of their trades do not result in

significant positive returns.

For the second hypothesis, we turn our attention to the registered insiders. Registered

insiders are defined as individuals who directly or indirectly own more than 10% of the firm’s

equity or who are officers or directors of the company according to Section 16 of the Securities

Exchange Act of 1934. While lawmakers have established key regulations to increase transparency

and reduce market manipulation, these traders are still active and profitable around key corporate

events. For instance, lawmakers established Rule 10b-5, which requires insiders 1) to refrain from

trading the firm’s shares when they have “material” nonpublic information or 2) to disclose the

information. However, to be charged with breaking Rule 10b-5, the insiders have to intentionally

deceive others. Fraud due to negligent behavior will not invoke Rule 10b-5. Furthermore, the

information has to be “material,” giving the insiders an unfair advantage to unduly influence the

market. Otherwise, the insiders are free to trade. The regulation also allows insiders to establish

multiple 10b-5 plans, which facilitate the sales of a predesignated number of shares at regular

intervals. While the intentions are good, the insiders can still work around these restrictions by

canceling the planned sales if they perceive good information is in the near future. This does not

break insider trading laws as no transactions were executed. Hence, no liabilities were created. In

11

fact, according to Lee et al. (1992), insiders appear to increase their frequency of buying and

decrease their frequency of selling before the repurchase announcement. In addition to Rule 10b-

5, the insiders need to follow the SEC Section 16(b) short-swing profit rule, which states that the

insiders must return any profit gained from the buying and selling of the company stock within a

six-month period. This regulation is designed to discourage insider trading with non-public

information. Similar to the other regulations, there are ways to get around the short-swing rule.

The insiders can avoid violating Section 16(b) by waiting until the six-month period ends, allowing

them to keep all the profit. For example, insiders can buy shares months before the announcement

and sell these shares during the post-announcement price run-up to maximizing the selling price

or to exploit any mispricing (Louis et al., 2010). Due to their ability to trade profitably, we further

expand the main question to test the relevance of the second signal from the portfolio activities of

insiders around the time of the repurchase announcement.

We determine the insiders’ net trade direction based on their transactions during the

previous six months when they are found to be active (Chan et al., 2012). We believe that insider

trades are a valuable source of information to investors. The more often insiders trade, the more

information is revealed to the public, giving investors more opportunities to reallocate their

resources and potentially make some profits (Manne, 1966; Bernhardt et al., 1995). According to

Bonaime and Ryngaert (2013), actual repurchases accompanied by net insider buying result in

significantly higher and longer-lasting abnormal returns than when insiders are net sellers. We

believe that insider trading during the pre-announcement period could also provide information to

investors as it does during the actual share repurchase, and it would be wise for the institutions to

study the insider’s trading pattern so they can respond profitably.

Hypothesis 2 (null form): Institutional investors do not trade profitably in the complex signaling

12

environment of share repurchases taking into account insider trades.

3. Data

According to the amendments to Rule 10b-18 (Release Nos. 33-8335; 34-48766; IC-26252;

File No. S7-50-02) made effective in January 2004, the SEC requires that the announcing firm

discloses its repurchase activities every quarter. The firm must disclose the total number of shares

repurchased during the previous quarter, the average price paid for those shares, the number of

shares that were purchased as a part of a previously announced plan, and the maximum number of

shares that could be repurchased. If the firm has not made any share repurchases after the

announcement, it is not required to disclose any of the information mentioned previously or its

intention to follow through with the repurchase plan. A disclosure is only necessary if the firm

decides to terminate the repurchase program.

The data for share repurchases are from the Securities Data Company (SDC). Our full

repurchases sample has 4,051 repurchase announcements from 2,259 unique firms reported from

September 2007 to December 2013. The firms announce the repurchase of approximately 243

billion shares and actually repurchase 38 billion shares at an average repurchase price of $33.16.

In total, these firms spent $1.2 trillion to repurchase their shares (SDC). We exclude all tender

offers, which are guaranteed to be implemented, to properly assess institutional trading ability in

an environment with added uncertainty.

We evaluate institutional profitability using three different datasets: 1) Thomson Reuters

13F provides quarterly institutional holding which will also capture institutional trading activities

not included in the ISO and short interest data, such as the changes in institutional holding resulting

from trades executed in the upstairs market, 2) TAQ provides intra-day ISO trading activity

including the exact timestamp, price, quantity, and trade condition, 3) Abel Noser provides daily

13

institutional trades with exact execution price and quantity, and 4) Compustat provides biweekly

short interest position data. We calculate profit by comparing the actual trade prices to the CRSP

closing prices in a later period.

Our first dataset, Thomson Reuters 13F, provides a big-picture summary of institutional

trading. The 13F data provide required filings of institutional investment managers with over $100

million in assets. We use the quarterly updates to understand long-term institutional trading and to

determine if institutions profit in the quarters around the repurchase announcement. Using the

Bushee (2001) method of classifying the 13F institutions, we are able to study in-depth if

institutions with different investment time horizons and styles are able to profit in the presence of

the firm and its insiders. The first type is the transient institutional investors, who have a high

portfolio turnover and highly diversified portfolio holdings. These are the institutions most likely

to pay close attention to corporate announcements and respond to them by altering their holding

and position frequently. According to Bushee (2001) and Puckett and Yan (2011), this trading

strategy generates significant abnormal returns. The second type is the dedicated institutional

investors, who maintain a very low portfolio turnover and larger average portfolio investments.

Quasi-indexer institutional investors also have low portfolio turnover but highly diversified

portfolio holdings. Both dedicated and quasi-indexer investors have longer investment horizons.

In addition to 13F, we use ISOs, which are limit orders that automatically execute in

designated markets while simultaneously submitting orders in the markets with better prices. ISOs

represent 31% of the volume and 38% of trades in our sample. Fully integrated in September 2007,

ISOs are mainly used by informed institutional traders to sweep liquidity from multiple markets,

although possibly at prices inferior to NBBO (Chakravarty et al., 2012). The authors also provide

evidence that ISO trades, presumably used by buy-side institutions, are associated with larger

14

information share than NISO trades, which are mainly used by retail traders. While ISOs can be

used to take liquidity, they can also provide liquidity. Liquidity-providing ISO traders are

incentivized to do so in order to earn rebates to cushion their profitability. Regardless of liquidity

taking or provision, traders use ISOs more for the execution speed and order fulfillment. Faster

execution gives institutional investors more opportunities to trades profitably as price-sensitive

information is released, for example, in repurchase announcements. Because we suspect that ISO

traders will be able to properly time the execution of their orders, we focus on these relatively

more informed trades to determine if they can make a profit using this more aggressive trading

mechanism. An added benefit to using the ISO data is the ability to examine the immediate

response to the announcement at the daily level.

We also use data from Abel Noser to examine institutional trading at the daily level. Abel

Noser collects its data directly from portfolio managers of plan sponsors, who are required to report

their trades to a cost analysis service by the Department of Labor’s Employment Retirement

Income Security Act (ERISA) to ensure that investment portfolio managers demonstrate the best

execution for their clients (Hu et al. 2018). According to Puckett and Yan (2011) and Hu et al.

(2018), the data cover 8% to 15% of CRSP volume. The Abel Noser data have similar advantages

as the ISO data in that we can gather the exact executed trade price and volume to assess

institutional trades around the announcement day. Several leading papers have used the Abel Noser

data to analyze institutional trading behavior around key corporate decisions, such as seasoned

equity offering (Chemmanur et al., 2009), spin-offs (Chemmanur and He, 2016), IPOs

(Chemmanur et al., 2010), and stock splits (Chemmanur et al., 2015). Bhattacharya, Wei and Xia

use same data source in their working paper to determine that institutions can using credit ratings

by investor-paid agencies, specifically Egan-Jones Ratings Company to outperform other investors

15

that do not follow the same advice.

Our last data source is Compustat short interest file. We analyze the trading activities of

short sellers, which account for approximately 26% of the daily total trading volume (Alexander

et al., 2014), to determine if these sophisticated traders can benefit from repurchase

announcements. Considering that there is a price appreciation after the announcement, we expect

to see a significant decrease in short selling up to the price run-up associated with repurchase

announcement.

In our efforts to provide a diverse set of institutional data, we do acknowledge that there

are overlaps in the data. For instance, transient 13F institutions can use ISOs to execute their trades

around the announcement, short sellers can also be transient 13F traders, some shorting strategies

could include using ISOs, and Abel Noser trades are also reported in the 13F filings. Without the

information to identify these traders, we are unable to clearly separate the three institutional types

from each other but such a separation is not required to test our hypotheses. Instead, we believe

that our inclusion of these three types can shed light by providing a multiplicity of evidence on

how the ISO sweeping mechanism, Abel Noser portfolio managers’ trades, short seller’s timing,

and overall institution’s strategy can lead to profitability.

Lastly, for control variables, we use analyst forecast data from I/B/E/S, accounting data

from Compustat, and insider trading data from Thomson Reuter TFN U.S. Securities and

Exchange Commission Form 4.

4. Findings

Before we begin our investigation of institutional foresight, we first examine the premise

of the post-announcement price run-up documented in prior research. Using the same estimation

and testing periods as in Stephens and Weisbach (1998) and Babenko (2009), we too find in our

16

sample a significant price increase after the repurchase announcement as shown in Table 1. We

also find that the abnormal return exists around repurchase announcements for additional testing

periods. Modifying the testing period in Babenko (2009) to [0, +90] and [-5, +5], we find that the

cumulative abnormal returns are positive and significant in these extended windows as well at

2.42% and 1.76%, respectively.

[Insert Table 1]

4.1 Trade imbalance

Having confirmed the presence of the post-announcement price run-up, we proceed in our

research using the four different institutional datasets. The descriptive statistics are presented in

Table 2, which shows the variables of interest for our multivariate regression models.

[Insert Table 2]

Before we look at the multivariate regression analysis, we first examine the trading activities of

institutional investors at the quarterly level using the 13F data. We analyze institutional trading for

five quarters before and after the announcement quarter. In Table 3 Panel A, we show that the

announcing firms during the non-announcing time, quarters [-5, -1], are associated with significant

positive or neutral institutional trade imbalance. During the repurchase announcement quarter [0]

and the following five quarters [+1,+5] they become net sellers, which means that repurchase

announcements do affect institutional trading and the effects last for many quarters. We believe

that the positive or neutral trade imbalances in quarters [-5, -1] show that the significant

institutional selling or negative imbalance in subsequent quarters is in response to the

announcement and not the other way around. In other words, the firm did not announce share

repurchases because of pre-announcement selling pressure from institutional investors, ruling out

reverse causality as the explanation of our findings. More importantly, we fail to see an increase

17

in institutional buying during the repurchase announcement quarter, Q0, which is associated on

average with positive abnormal returns in near future.

To consider the effects of insiders for the dual signal hypothesis, we separate the full

sample based on whether the corporate insiders are net buyers or net sellers in the period

surrounding the repurchase announcement. Our method of classifying insider trade direction is

similar to that of Bonaime and Ryngaert (2013). The announcement event is considered net selling

if insider sales exceed insider purchases by at least 0.01% of the firm’s market capitalization. The

announcement event is considered net buying if purchases exceed sales by the same requirement.

Otherwise, the event is associated with neutral insider trading. These classifications are based on

the transactions of insiders during the previous six months relative to the announcement because

insiders are found to be most active during this period based on the findings of Chan et al. (2012).

When we divide the full 13F sample based on the insider trade direction, we observe similar

significant and negative trade imbalances for quarters [0, +4] following the repurchase

announcement.

[Insert Table 3]

Our examination of the 13F data extends further by breaking down the trading activities of

the institutional investors according to the classification method of Bushee (2001). We observe

that transient, dedicated, and quasi-indexer institutions are all net sellers exhibiting significant

negative trade imbalances from quarters [0, +1] as shown in Figure 2.

[Insert Figure 2]

To observe how institutional investors trade when there is no announcement effect, we

create a matching control sample of non-announcing firms based on closest matches to the industry

SIC, market capitalization, year and quarter to conduct placebo tests. We also impose a strict

18

matching requirement that the minimized absolute difference in the market capitalizations of

treatment and control firm be no greater than twenty-five percent of the sample firm’s value. In

the case that there is no matching firm, we drop the treatment firm. If any market-wide omitted

factors or economic reasons other than repurchases cause institutional selling, then the selling

imbalance would also show up in the control sample placebo tests using the treatment firm’s

repurchase announcement date as the pseudo-announcement date for matched control firms.

In sharp contrast to our sample of announcing firm, our control sample of non-announcing

firms shows that there is a positive buying trade imbalance for the quarters around the pseudo-

announcement with varying degrees of significance. The results are shown in Figure 2. This

control sample shows that the firms with no connections to repurchase announcements are

associated with net institutional buying. This is further evidence that there are no omitted market-

wide factors affecting institutional imbalance around repurchase announcements.

In Figure 3, when we examine the activities of the firm and its insiders, we find that they

are not trading in the same direction as institutions. The firm appears to be repurchasing shares

starting at the announcement quarter, and the insiders exhibit neutral or negative trade imbalances.

We attribute the insignificance to a passive buying strategy in which the insiders reduce their sales

significantly more than their purchases to have the appearance of indifference (Agrawal and

Nasser, 2012). While we can assume that their trading activities may be independent of each other,

the appearance of the firm’s repurchases to support insider selling should be acknowledged. For

instance, in Moore (2017), the author finds that there is a positive relationship between share

repurchases and CEO equity sales. Hence, the idea that share repurchases can be used strategically

by insiders is a plausible one.

[Insert Figure 3]

19

Next, we take a much closer look at how institutions trade during the days immediately

around the announcement. The ISO intra-day data allow us to examine how investors promptly

react to the event. Our analysis of daily ISO trading covers three distinct periods similar to Jain

and Wang (2013) and follows the general approach of Irvine et al. (2007): the pre-announcement

period is the five days [-5, -1] window leading up to the announcement date, day 0 is the

announcement day, and the post-announcement period is the five days [+1, +5] after the event

date. Although we are studying the perceptive of institutional investors before and on the

announcement date, we include the post-announcement period to evaluate the institutional

investor’s immediate reaction to the repurchase information.

Based on the well-documented, positive market reaction to the announcement, we expect

institutional investors to be net buyers in these three periods. In theory, purchasing these shares

before the price run-up reaches its peak is a profitable strategy. Rather than observing a net buying

trend, we actually find evidence of significant negative institutional trade imbalance for all 11 days

shown in Table 3 Panel B. We calculate daily ISO trade imbalance as the net shares traded

normalized by the number of shares outstanding in millions. During the pre-announcement period,

we expect institutional investors to take a neutral trading position if they are unable to predict the

timing of the announcement. If they do have the foresight, we expect them to be net buyers.

Therefore, to see that these investors are presenting themselves as sellers is rather surprising. The

statistically significant daily trade imbalance of -75.32 shares per million of shares outstanding for

days [-5, -1] relative to the announcement day shown in Table 3 Panel B suggests that the ISO

traders do not perceive value in these firms before the repurchase event. On the announcement day

and during the post-announcement period, institutional investors would still be profitable if they

are net buyers. Yet, we observe statistically significant daily trade imbalances of -87.06 and -75.86

20

for the announcement day and post-announcement period. The net selling trend suggests that

institutional investors do not foresee the possible gains that follow repurchase announcements and

are selling to the better-informed traders. Additionally, we find that institutional investors do not

take cues from insiders; institutional investors are still net sellers with statistically significant daily

trade imbalances ranging from -50.72 to -141.40 when insiders are net buyers as shown in Table

3 Panel B.

In our third measure using the Abel Noser data, we examine how institutions trade daily in

the thirty days before and after the announcement. The daily Abel Noser trade imbalance is

calculated as the actual number of buy shares minus the number of sell shares, normalized by the

number of shares outstanding in millions. In Table 3 Panel C, we show that institutional traders do

not exhibit a positive trade imbalance before, during and after the announcement. The average

daily trade imbalance during the thirty days before the announcement is -281.50. Similarly, after

the announcement, the average daily trade imbalance is -653.50. These institutional traders

actually exhibit the highest net selling trend on the day of the announcement with a trade imbalance

of -9,485.10. Furthermore, when we divide the full sample based on insider trading trends, we

find that institutional traders do not exhibit any significant buying. Once again, we have

confirmation that institutional traders are not buying more shares around the announcement to

resell at a higher price.

Lastly, we turn our attention to the short position traders. We expect the short sellers to

refrain from trading around the announcement period as stock prices have not fully appreciated.

Any abnormal shorting during this period could result in significant losses. Abnormal short interest

is calculated as the average number of shares sold short in the test period divided by the average

number of shares sold short in the benchmark period minus one (Christophe et al., 2004). The

21

benchmark period is six biweekly periods before the announcement period, ending approximately

one quarter before the announcement. We find the highest abnormal short selling in the immediate

five biweekly periods after the repurchase announcement as shown in Table 3 Panel D. The

statistically significant abnormal short interests in periods [+1, +5] relative to the benchmark are

18.01%, 19.00%, 18.40%, 17.23, and 16.89%. Again, we find evidence of an increase in selling

around the announcement period before stock prices have reached their peaks. Interestingly, when

the insiders are net buyers we see a significant reduction in short selling during the five periods

before the announcement. The abnormal short selling in during period [-5, -1] are -21.31%, -

21.59%, -20.75%, -20.85%, and -19.47%. From these numbers, which are all statistically and

economically significant, we see that short-sellers are taking cues from the insiders’ net buying

signal, and, consequently, reduce their shorting activities before the announcement.

4.2 Profitability

Next, we evaluate the profitability of institutional investors. We calculate institutional

profitability based on the percentage change in quarterly closing prices. For negative trade

imbalances, we calculate profitability as the difference between the closing prices of the

announcement quarter and the following quarter (to cover the position). Analogously, the opposite

difference is used for positive trade imbalances. The decision to sell by the institutions is dominant

during the announcement quarter and closing such positions by quarter +4 results in an average

loss of 0.75% as shown in Table 4.

[Insert Table 4]

Next, we evaluate if the ISO trades executed during the days around the announcement are

profitable. Daily institutional profit is determined using the exact ISO trade prices in each of the

[0, +5] days and CRSP daily closing prices on day +5 relative to the announcement. For sell trades,

profitability is calculated as the actual trade price minus the imputed closing price divided by the

22

trade price. For buy trades, profitability is calculated as the imputed closing price minus the actual

trade price divided by the trade price.

In Table 4, we find that the institutional ISO trades opened during day [0, +5] result in

significant losses if the positions are closed +5 days after the announcement. In the full sample of

all repurchase announcements, the average daily loss of the trades initiated during days [+0, +5] is

0.04%, statistically significant at one percent. When we examine institutional profitability

considering insiders’ buy versus sell signal, we find that the institutions are not profitable in either

category even with dual signals. For instance, the trades initiated in the subsample with net insider

buying has a daily loss of 0.07% also significant at the one percent level.

When we examine the daily profitability of institutional traders using the Abel Noser data,

we find that they are not profitable. From days [0, +30], the average daily institutional profitability

is 0.00%. We find that in the sample where the insiders are net buyers the institutions incur a daily

loss of 0.02%. This is most likely due to the non-positive trade imbalance during the [0, +30] day

period. Selling even a few shares at the announcement combined with insider buying appears to

be an unprofitable strategy for the institutions, due to large adverse price moves.

Finally, we determine if the short sellers’ poor timing relative to the repurchase

announcements adversely affects their profitability. We measure profit as the difference between

the proceeds from the sale and the cost to close the position using CRSP closing prices at the end

of the biweekly period [0, +8]. Table 4 shows that significant abnormal short selling in the post-

announcement biweekly periods leads to a significant loss because the cost to close the short

position is higher than the revenue from the opening trades. We find that by the time these traders

cover their short positions in the eighth biweekly period they have cumulated loss of 2.15%

significant at one percent. We also examine other time frames, such as [0, +4], [0,+5], [0,+6],

23

[0,+7], and find that short sellers are not profitable in any of these periods around repurchase

announcements. Similar to the ISO results, we conclude that short sellers are not able to trade

profitably around repurchase announcements regardless of the insider trade directions. Our

findings are different from the literature citing short seller’s foresight (Chakrabarty and Shkilko,

2013; Christophe et al., 2004) because we specifically study the short seller’s trading performance

in an environment with heightened uncertainty and multiple highly informed and actively trading

participants.

4.3 Robustness tests

To ensure that both institutional trading and profitability are not influenced by other

corporate events, we create two sub-samples with the ISO data. In one sample, the share repurchase

announcements do not coincide with other corporate news, such as dividend announcements,

earnings announcements, merger and acquisition announcements, stock dividends and stock splits.

In the second sample, all the share repurchase announcements coincide with at least one of the

corporate events listed. The results are shown in Table 5. Like before, we calculate the daily ISO

trade imbalance as the actual number of buy shares minus the number of sell shares, normalized

by the number of shares outstanding in millions. We calculate ISO profitability using the exact

trade prices and CRSP daily closing prices on day +5. First, we find that the ISO traders exhibit

statistically significant negative trade imbalances in both subsamples. Second, the trades result in

significant losses in both samples. Thus, the effects of the repurchase announcement on

institutional trading strategy and its profitability are independent of other corporate events.

[Insert Table 5]

Then, we analyze the profitability of institutional trades opened during the days around the

announcement and closed 5 days later in the full sample adjusting for quarter-end effects. Since

24

some institutions engage in quarter-end window dressing (He et al., 2004), we eliminate all

announcements and institutional trades during the last three days of each quarter using the ISO

data so our results are independent of these decisions, and we find that these sell trades still result

in significant losses. The average loss is statistically significant at 0.05% as shown in Table 6. In

summary, our findings using the ISO data reveal that daily institutional trades are not profitable.

[Insert Table 6]

Additionally, we entertain the possibility that a repurchase announcement may be a strategy

to counter institutional selling pressure. In this sense, institutional trade imbalance and the

announcement event may be endogenously determined. To examine this potential endogeneity

issue, we perform two tests. In our first test, we examine the differences between the institutional

trade imbalances in the shares of announcing firms around the repurchase announcement date

versus that in matching non-announcing firms around the matching pseudo-event date. We find

that in quarter [-5, -1] before the announcement, the differences are statistically insignificant.

Although we do not show the results in a table format, the similarity between the institutional trade

imbalances for the announcing and non-announcing firms before the event can also be seen in

Figure 2. Thus, institutional trading does not appear to be the catalyst for the repurchase

announcement.

In our second test, we formally evaluate if the institutional trade imbalance during the

announcement quarter and announcement signal can be endogenously determined by estimating

the following linear model:

25

𝑇𝑅𝐴𝐷𝐸𝐼𝑀𝐵𝐴𝐿𝐴𝑁𝐶𝐸 = 𝛼 + 𝛽1𝐴𝑁𝑁𝑂𝑈𝑁𝐶𝐸𝑀𝐸𝑁𝑇 + 𝛽2𝐼𝑁𝑆𝐼𝐷𝐸𝑅𝐵𝑈𝑌 +

𝛽3𝐼𝑁𝑆𝑇. 𝐻𝑂𝐿𝐷𝐼𝑁𝐺(𝑄−1) + 𝛽4𝐸𝑃𝑆𝑆𝑈𝑅𝑃𝑅𝐼𝑆𝐸 + 𝛽5𝐹𝐼𝑅𝑀𝑆𝐼𝑍𝐸 + 𝑢 (1)

The dependent variable, TRADEIMBALANCE, is the quarterly institutional trade

imbalance of the announcement quarter, Q0. ANNOUNCEMENT, the endogenous regressor, is a

dummy variable set to 1 if the firm announces share repurchases. INSIDERBUY is a dummy

variable set to 1 if insider purchases exceed insider sales during the six months before the

announcement day by at least 0.01% of the firm’s market capitalization. INST.HOLDING Q-1 is

calculated based on the reported holdings on Form 13F normalized by the number of shares

outstanding. EPS SURPRISE measures the difference between the actual reported EPS and the

mean analyst EPS forecast divided by the share price. FIRMSIZE is the natural log of the firm’s

market capitalization. u is the error term.

To determine if endogeneity exists meaning that the ANNOUNCEMENT is correlated with

the error term, u, we use the Hausman Test. The test requires two stages. In the first stage, we

regress the endogenous regressor, ANNOUNCEMENT, on the instrumental variables (IVs), which

are ΔCASH, DILUTION, and TAXGAP, to find the residuals, v. ΔCASH is the change in

EBITDA/Total Assets over the previous quarter. DILUTION is calculated as one minus the ratio

of diluted EPS/ basic EPS. TAXGAP is calculated as the difference between the average tax rate

on dividends and the average tax rate on capital gains. For these instruments, we follow the method

outlined in De Lisle et al. (2014).

ANNOUNCEMENT= 𝛼 + 𝛽1𝛥𝐶𝐴𝑆𝐻 + 𝛽2𝐷𝐼𝐿𝑈𝑇𝐼𝑂𝑁 + 𝛽3𝑇𝐴𝑋𝐺𝐴𝑃 + 𝑣 (2)

In the second stage, we include the residuals, v, from the first stage as an independent

26

variable in the following model:

𝑇𝑅𝐴𝐷𝐸𝐼𝑀𝐵𝐴𝐿𝐴𝑁𝐶𝐸 = 𝛼 + 𝛽1𝐴𝑁𝑁𝑂𝑈𝑁𝐶𝐸𝑀𝐸𝑁𝑇 + 𝛽2𝐼𝑁𝑆𝐼𝐷𝐸𝑅𝐵𝑈𝑌 +

𝛽3𝐼𝑁𝑆𝑇. 𝐻𝑂𝐿𝐷𝐼𝑁𝐺(𝑄−1) + 𝛽4𝐸𝑃𝑆𝑆𝑈𝑅𝑃𝑅𝐼𝑆𝐸 + 𝛽5𝐹𝐼𝑅𝑀𝑆𝐼𝑍𝐸 + 𝛽6𝑣 + 𝜀 (3)

where the error term, 𝑢 = 𝛽6

𝑣 + 𝜀. This means that the error term, u, depends on v and some other

residual, 𝜀. Therefore, we test the null hypothesis that the residual coefficient, 𝛽6, is statistically

equal to zero. If 𝛽6 = 0, then we can conclude that institutional trading during the announcement

quarter and the announcement signal from the firm are not endogenously determined (Hausman,

1978; Nakamura and Nakamura, 1981; Abrevaya et al., 2010). However, if 𝛽6 ≠ 0, then

endogeneity exists and OLS regressions will provide biased estimates. In Table 7, we show that

the residual, v, with a coefficient of -0.0006 is not statistically different from zero. From this result,

we conclude that institutional trading and the firm’s repurchase announcements are not

endogenously determined.

[Insert Table 7]

4.4 Factors affecting institutional profit

To understand the various factors that may affect how institutions perform around the

announcement, we model institutional profit as a function of the firm’s repurchase announcement

and the insiders’ trade direction controlling for institutional holdings, EPS surprise, firm size, beta,

SMB, HML, industry fixed effects, and year fixed effects.

𝐼𝑛𝑠𝑡𝑖𝑡𝑢𝑡𝑖𝑜𝑛𝑎𝑙 𝑃𝑟𝑜𝑓𝑖𝑡 = 𝛼 + 𝛽1𝐴𝑛𝑛𝑜𝑢𝑛𝑐𝑒𝑚𝑒𝑛𝑡 + 𝛽2𝐼𝑛𝑠𝑖𝑑𝑒𝑟 𝐵𝑢𝑦 𝑆𝑖𝑔𝑛𝑎𝑙 +𝛽3𝐴𝑛𝑛𝑜𝑢𝑛𝑐𝑒𝑚𝑒𝑛𝑡 ∗ 𝐼𝑛𝑠𝑖𝑑𝑒𝑟 𝐵𝑢𝑦 𝑆𝑖𝑔𝑛𝑎𝑙 + 𝛽4𝐼𝑛𝑠𝑡𝑖𝑡𝑢𝑡𝑖𝑜𝑛𝑎𝑙 𝐻𝑜𝑙𝑑𝑖𝑛𝑔𝑠𝑄−1 +

𝛽5𝐸𝑃𝑆 𝑆𝑢𝑟𝑝𝑟𝑖𝑠𝑒 + 𝛽6𝐹𝑖𝑟𝑚 𝑆𝑖𝑧𝑒 + 𝛽7𝐵𝑒𝑡𝑎 + 𝛽8𝑆𝑀𝐵 + 𝛽9𝐻𝑀𝐿 +𝐼𝑛𝑑𝑢𝑠𝑡𝑟𝑦 𝐹𝑖𝑥𝑒𝑑 𝐸𝑓𝑓𝑒𝑐𝑡𝑠 + 𝑌𝑒𝑎𝑟 𝐹𝑖𝑥𝑒𝑑 𝐸𝑓𝑓𝑒𝑐𝑡𝑠

(4)

27

Institutional profit for all three institutional types is calculated as presented previously in

Table 4. We calculate profits for trades in stocks of repurchase announcers (treatment stocks) and

matching non-announcer control stocks. ANNOUNCEMENT is a dummy variable set to 1 if the

firm announces share repurchases. INSIDER BUY is a dummy variable set to 1 if insider purchases

exceed insider sales during the six months before the announcement day by at least 0.01% of the

firm’s market capitalization. INSTITUTIONAL HOLDING Q-1 is calculated based on the

reported holdings on Form 13F and normalized across firms by dividing by the number of shares

outstanding. We include institutional holdings to account for the liquidity provision role that

institutions may play around repurchases (De Lisle et al., 2014). EPS SURPRISE measures the

difference between the actual reported EPS and the mean analyst EPS forecast for the

announcement quarter divided by the share price. We also include other common control variables,

such as the firm size, market risk premium, SMB, and HML. To address multicollinearity issues

when fitting our regression models, we examine VIF scores and find that all of them are below 10.

In Table 8, we report how the announcement and the insider buy direction affect

institutional profit from Q0 to Q+4 using the quarterly 13F data. We use a sample that includes

both announcing and non-announcing control firms, which are matched by industry and size. The

evidence shows that the announcement and the insider buy direction alone do not significantly

affect long-term profit. However, when the insiders are net buyers during the six months before

the announcement and then the firm actually announces buybacks, institutional profit significantly

decreases by 12.31%. These two signals from the firm and insiders suggest that institutions should

buy more shares, but from the evidence previously presented in Table 3 institutions exhibit

significant negative trade imbalances starting in Q0, ultimately resulting in trading losses for

institutions now reported in Table 8.

28

[Insert Table 8]

To examine how the announcement and the insider buy direction affect institutional profit

in the short term, we use the ISO data. Table 9 shows that the insider buy direction appears to have

a significant and negative relationship with short-term profit, reducing it by 0.14% daily or 40.03%

annualized with daily compounding. Instead, the overall institutional holdings in Q-1 appears to

be more significant in determining profit. Higher institutional holdings in the pre-announcement

period appear to be a profitable strategy, which aligns with our belief that buying before the price

run-up and selling during the price appreciation can be beneficial to institutional investors.

[Insert Table 9]

In Table 10, we show the results for the regression of profitability using the Abel Noser

data on the announcement and insider signal. We find that on average a repurchase announcement

has a negative impact on institutional profit. The impact on daily profit range from -0.0129% to -

0.0121%, significant at the five and ten percent levels, respectively.

[Insert Table 10]

Finally, we analyze the profitability of short sellers from biweekly 0 to +8 period shown in

Table 11, and we find that the announcement and insider buy direction do not help short-sellers

generate any statistically significant profit, either. In summary, all of the profitability regressions

indicate institutional investors’ lack of foresight around share repurchase announcement and the

insider buy direction fails to produce abnormal long- or short-term profit. Instead, some of the

institutional selling trends are counterintuitive, especially when there are buy signals from other

informed players.

[Insert Table 11]

29

5. Institutional Foresight Potential Explanations

While we acknowledge that prior research explains that institutional investors are informed

relative to average market participants (Daniel et al., 1997; Baker et al., 2010; Hoberg et al., 2018),

we find that in the battle of champions around a share repurchase announcement, there are several

factors working against these sophisticated traders. First, a repurchase announcement may or may

not be driven by undervaluation; other motives may include tax benefits, distributing excess cash,

adjusting capital structure, creating opportunities for insiders to sell, and eliminating threats of

unwanted takeover attempts according to Asquith and Mullins (1986), Ikenberry and Vermaelen

(1996), Dittmar (2000), Lee and Rui (2007), Bhana (2007), De Ridder (2009), Lee et al. (2010),

and Punwasi and Brijlal (2016). Second, share repurchase announcements do not occur at a

predetermined time interval. The timing of these announcements are unpredictable, and it leaves

the institutional investors little time to respond to the information. Third, the decision to follow

through with actual share repurchases and the timing of these buyback programs are also

unpredictable. Once announced, the firm has no legal obligations to follow through with actual

implementation nor does it have to buy back shares within a certain time period. This flexibility is

one of the main reasons for using repurchases rather than dividends as the main payout method.

Fourth, around share repurchase announcements, institutional investors are competing against the

firm and its insiders, who arguably are more informed than the institutional investors, at least,

when it comes to matters related to the valuation of the company. In this scenario of a share

repurchase announcement, both of these informed players are also actively trading their company’s

stocks. When we examine the dual signals given by these two players, it appears that they provide

mixed messages to outsiders. Our analysis shows that approximately 50% of the follow-through

sample is associated with net insider sell, 31% is associated with net insider buying, and 19% with

neutral insider trading. This contradiction between the firm and the insiders implies that the

30

insiders may be using the firm’s money to support insider selling. In fact, the Wall Street Journal

has taken notice of this strategy and has published articles explaining that repurchases can be used

to support sales, install an artificial price floor, and most interestingly, to help executives and board

members earn abnormal returns (Browning, 2015; Driebusch and Eisen, 2016; Strumpf, 2014;

Waggoner, 2015). The insiders can benefit from the initial post-announcement price run-up and

then from the actual implementation of the repurchase programs that could take years to complete.

The long-term strategy of using repurchase events from the announcement to the follow through

to manage price impact is completely plausible. Klein et al. (2017) find that when there are multiple

insiders trading simultaneously managing price impact and market liquidity is the main concern.

Because of these aforementioned reasons, it may be difficult for institutions to trade profitably

around the announcement.

6. Conclusion

Share buyback or repurchase announcements create heightened information asymmetry

about the motives and the final outcome of such announcements. We analyze the institutional

foresight about the dual signals from the actions of repurchase announcing firms and its insiders,

for various types of institutions. We find that institutions—13F institutional investors, ISO traders,

Abel Noser institutions and short sellers—sell their shares around the announcement. This is a

rather unprofitable strategy because of the poor timing and execution price of institutional trades

around repurchase announcements. Our paper provides evidence that although institutional

investors are sophisticated traders they may not be profitable when better-informed agents such as

the announcing firms and their insiders are actively pursuing their own interests.

7. References

Abrevaya, J., Hausman, J., Khan, S., 2010. Testing for casual effects in a generalized regression

31

model with endogenous regressors. Econometrica 78, 2043-2061.

Alldredge, D., Cicero, D., 2015. Attentive insider trading. Journal of Financial Economics 115,

84-101.

Agrawal, A., Nasser, T., 2012. Insider trading in takeover targets. Journal of Corporate Finance

18, 598-625.

Alexander, G., Peterson, M., Beardsley, X., 2014. The puzzling behavior of short sellers around

earnings announcements. Journal of Financial Intermediation 23, 255-278.

Asquith, P., Mullins, D.W., 1986. Signalling with dividends, stock repurchases, and equity issues.

Financial Management,15, 27-44.

Babenko, I., 2009. Share repurchases and pay-performance sensitivity of employee compensation

contracts. Journal of Finance 64, 117-150.

Bagnoli, M., Watts, S.G., 1998. Information acquisition, information release and trading dynamics.

Journal of Financial Markets 1, 221-252.

Baker, M., Litov, L., Wachter, J., Wurgler, J., 2010. Can mutual fund managers pick stocks?

Evidence from their trades prior to earnings announcements. Journal of Financial and

Quantitative Analysis 45, 1111-1131.

Bargeron, L., Kulchania, M., Thomas, S., 2011. Accelerated share repurchases. Journal of

Financial Economics 101, 69-89.

Bens, D., Nagar, V., Skinner, D., Wong, M.H., 2003. Employee stock options, EPS dilution, and

stock repurchases. Journal of Accounting and Economics 36, 51-90.

Bernhardt, D., Hollifield, B., Hughson, E., 1995. Investment and insider trading. Review of

Financial Studies 8, 501-543.

Bhana, N., 2007. The market reaction to open market share repurchases announcements: The South

32

African experience. Investment Analysts Journal 65, 25-36.

Bhattacharya, U., Jacobsen, S., 2016. The share repurchase announcement puzzle: Theory and

evidence. Review of Finance 20, 725-758.

Bhattacharya, U., Wei, K., Xia, H., 2018. Follow the money: Investor trading around investor-

paid credit rating changes. Available at SSRN:

https://papers.ssrn.com/sol3/papers.cfm?abstract_id=2474391.

Bonaime, A., Ryngaert, M., 2013. Insider trading and share repurchases: Do insiders and firms

trade in the same direction? Journal of Corporate Finance 22, 35-53.

Browning, E.S. “Is the surge in stock buybacks good or evil?” The Wall Street Journal, Dow Jones

& Company, 22 Nov. 2015, www.wsj.com/articles/is-the-surge-in-stock-buybacks-good-or-

evil-1448188684.

Bushee, B., 2001. Do institutional investors prefer near-term earnings over long-run value?

Contemporary Accounting Research 18, 207-246.

Chakrabarty, B., Moulton, P., Trzcinka, C., 2017. The performance of short-term institutional

trades. Journal of Financial and Quantitative Analysis 52, 1403-1428.

Chakrabarty, B., Shkilko, A., 2013. Information transfers and learning in financial markets:

Evidence from short selling around insider sales. Journal of Banking and Finance 37, 1560-

1572.

Chakravarty, S., Jain, P., Upson, J., Wood, R., 2012. Clean sweep: informed trading through

intermarket sweep orders. Journal of Financial and Quantitative Analysis 47, 415-435.

Chakravarty, S., Ray, R., 2018. On institutional trading, behavioral bias, and demand for liquidity.

Available at SSRN: https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3013436.

Chan, K., Ikenberry, D., Lee, I., Wang, Y., 2010. Share repurchases as a potential tool to mislead

33

investors. Journal of Corporate Finance 16, 137-158.

Chan, K., Ikenberry, D., Lee, I., Wang, Y., 2012. Informed traders: linking legal insider trading

and share repurchases. Financial Analysts Journal 68, 60–73.

Chemmanur, T., He, S., Hu, G., 2009. The role of institutional investors in seasoned equity

offerings. Journal of Financial Economics 94, 384-411.

Chemmanur, T., He, S., 2016. Institutional trading, information production, and corporate spin-

offs. Journal of Corporate Finance 38, 54-76.

Chemmanur, T., Hu, G., Huang, J., 2010. The role of institutional investors in initial public

offerings. Review of Financial Studies 23, 4496-4540.

Chemmanur, T., Hu, G., Huang, J., 2015. Institutional investors and the information production

theory of stock splits. Journal of Financial and Quantitative Analysis 50, 413-445.

Chemmanur, T., and Li, Y., Xie, J., and Zhu, A., 2018, Noisy Signaling through Open Market

Share Repurchase Programs and Information Production by Institutions. Available at

SSRN: https://ssrn.com/abstract=2543397.

Chen, H-C., Harper, J., Iyer, S., 2018. Economic shock and share repurchases. North American

Journal of Economics and Finance 44, 254-264.

Chiyachantana, C., Jain, P., Jiang, C., Sharma, V., 2017. Permanent price impact asymmetry of

trades with institutional constraints. Journal of Financial Markets 36, 1-16.

Christophe, S., Ferri, M., Angel, J., 2004. Short-selling prior to earnings announcements. Journal

of Finance 59, 1845-1875.

Cox, J. “Corporate Executives Are Using Stock Buybacks to Pad Their Own Compensation,

According to SEC Official.” CNBC, CNBC, 11 June 2018, www.cnbc.com/2018/06/11/sec-

official-clamp-down-on-officials-who-cash-in-on-stock-buybacks.html.

34

Daniel, K., Grinblatt, M., Titman, S., Wermers, R., 1997. Measuring mutual fund performance

with characteristic-based benchmarks. Journal of Finance 52, 1035–1058.

De Cesari, A., Espenlaub, S., Khurshed, A., Simkovic, M., 2012. The effect of ownership and

stock liquidity on the timing of repurchase transactions. Journal of Corporate Finance 18,

1023-1050.

Dechow, P., Hutton, A., Meulbroek, L., Sloan, R., 2001. Short-sellers, fundamental analysis, and

stock returns. Journal of Financial Economics 61, 77-106.

De Lisle, R.J., Morscheck, J.D., Nofsinger, J.R., 2014. Share repurchases and institutional supply.

Journal of Corporate Finance 27, 216-230.

De Ridder, A., 2009. Share repurchases and firm behavior. International Journal of Theoretical

and Applied Finance 12, 605-631.

Dittmar, A., 2000. Why do firms repurchase stock? Journal of Business 73, 331-355.

Driebusch, C., Eisen, B. “Buybacks pump up stock rally.” The Wall Street Journal, Dow Jones &

Company, 12 July 2016, www.wsj.com/articles/buybacks-pump-up-stock-rally-1468363826.

Fudenberg, D., Tirole, J., 1986. A “signal jamming” theory of predation. The RAND Journal of

Economics 17, 366-376.

Grullon, G., Michaely, R., 2004. The information content of share repurchase programs. Journal

of Finance 59, 651-680.Hao, Q., 2016. Is there information leakage prior to share repurchase

announcements? Evidence from daily options trading. Journal of Financial Markets 27, 79-

101.

Hausman, J., 1978. Specification tests in econometrics. Econometrica 46, 1251-1271.

He, J., Ng, L., Wang, Q., 2004. Quarterly trading patterns of financial institutions. Journal of

Business 77, 493-509.

35

Hoberg, G., Kumar, N., & Prabhala, N., 2018. Mutual fund competition, managerial skill, and

alpha persistence. The Review of Financial Studies, 31, 1896-1929.

Holden, C., Subrahmanyam, A., 1992. Long-lived private information and imperfect competition.

Journal of Finance 47, 247-270.

Hu, G., Jo, K., Wang, Y., Xie, J., 2018. Institutional trading and Abel Noser data. Working paper

series.

Ikenberry, D., Lakonishok, J., Vermaelen, T., 1995. Market underreaction to open market share

repurchases. Journal of Financial Economics 39, 181-208.

Ikenberry, D., Vermaelen, T., 1996. The option to repurchase stock. Financial Management 25, 9-

24.

Irvine, P., Lipson, M., Puckett, A., 2007. Tipping. Review of Financial Studies 20, 741-768.

Jain, P., Wang, Q., 2013. Credit-rating changes and institutional trading. Journal of Trading 8, 38-

47.

Jensen, M.C., 1986. Agency costs of free cash flow, corporate finance, and takeovers. American

Economic Review 76, 323–329.

Jiang, C., Likitapiwat, T., McInish, T., 2012. Information content of earnings announcements:

Evidence from after-hours trading. Journal of Financial and Quantitative Analysis 47, 1303-

1330.

Ke, B., Petroni, K., 2004. How informed are actively trading institutional investors? Evidence from

their trading behavior before a break in a string of consecutive earnings increases. Journal of

Accounting Research 42, 895-927.

Klein, O., Maug, E., Schneider, C., 2017. Trading strategies of corporate insiders. Journal of

Financial Markets 34, 48-68.

36

Kyle, A., 1985. Continuous auctions and insider trading. Econometrica 53, 1315-1336.

Lee, D., Mikkelson, W., Partch, M., 1992. Managers' trading around stock repurchases. Journal of

Finance 47, 1947-1961.

Lee, C., Ejara, D., Gleason, K., 2010. An empirical analysis of European stock repurchases.

Journal of Multinational Financial Management 20, 114-125.

Lee, C., Mucklow, B., Ready, M., 1993. Spread, depths, and the impact of earnings information:

An intraday analysis. Review of Financial Studies 6, 345-374.

Lee, C., Ready, M., 1991. Inferring trade direction from intraday data. Journal of Finance 46, 733-

746.

Lee, B., Rui, O., 2007. Time series behavior of share repurchases and dividends. Journal of

Financial and Quantitative Analysis 42, 119-142.

Louis, H., and White, H., 2007. Do managers intentionally use repurchase tender offers to signal

private information? Evidence from firm financial reporting behavior. Journal of Financial

Economics 85, 205-233.

Louis, H., Sun, A., White, H., 2010. Insider trading after repurchase tender offer announcements:

Timing versus informed trading. Financial Management, 39, 301-322.

Manne, H., 1966. Insider trading and the stock market. The Free Press 7, 93-158.

McInish, T., Upson, J., Wood, R., 2014. The flash crash: Trading aggressiveness, liquidity supply,

and the impact of intermarket sweep orders. Financial Review 49, 481-509.

Miller, M.H., Rock, K., 1985. Dividend policy under asymmetric information. Journal of Finance

40, 1031–1051.

Mirman, L., Samuelson, L., Urbano, A., 1993. Duopoly signal jamming. Economic Theory 3, 129-

149.

37

Moore, D., 2017. Managerial self-interest and strategic repurchases: Evidence from equity vesting

schedules. (Working Paper Series).

Nakamura, A., Nakamura, M., 1981. On the relationship among several specification error tests

presented by Durbin, Wu, and Hausman. Econometrica 49, 1583-1588.

Ofer, A.R., Thakor, A.V., 1987. A theory of stock price responses to alternative corporate cash

disbursement methods: stock repurchases and dividends. Journal of Finance 42, 365–394.

Peyer, U., Vermaelen, T., 2009. The nature and persistence of buyback anomalies. Review of

Financial Studies 22, 1693-1745.

Puckett, A., Yan, X., 2011. The interim trading skills of institutional investors. Journal of Finance

66, 601-633.

Punwasi, K., Brijlal, P., 2016. The market reactions to share repurchase announcements on the

JSE: An event study. Investment Management and Financial Innovations 13, 191-205.

Skinner, D., 2008. The evolving relation between earnings, dividends, and stock repurchases.

Journal of Financial Economics 87, 582-609.

Stephens, C., Weisbach, M., 1998. Actual share reacquisitions in open-market repurchase

programs. Journal of Finance 53, 313-333.

Strumpf, D. “Companies' stock buybacks help buoy the market.” The Wall Street Journal, Dow

Jones & Company, 15 Sept. 2014, www.wsj.com/articles/companies-stock-buybacks-help-

buoy-the-market-1410823441.

Vermaelen, T., 1981. Common stock repurchases and market signaling. Journal of Financial

Economics 9, 139-183.

Waggoner, J. “Beware the stock-buyback craze.” The Wall Street Journal, Dow Jones & Company,

19 June 2015, www.wsj.com/articles/beware-the-stock-buyback-craze-1434727038.

38

Figure 1: Timeline of Possible Actions Taken by Institutional Investors, the Firm, and its Insiders around Share Repurchase

Announcement and Actual Repurchases

Overvalued

Fairly Valued

Undervalued

Announcement

No Announcement

Institutions Should Sell

Actual Repurchases

No Repurchases

Insider Sell

Announcement

No Announcement

Announcement

No Announcement

Insider Buy

Institutions Should Buy

InstitutionsShould Sell

Announcement Period (t= 0)

Pre-Announcement Period (t=-1)

Post-Announcement Period (t=1)

Repurchasing Period (t=2)TIMELINE

InstitutionsShould Sell

No Repurchases Price Stabilizes

Actual Repurchases

Price Holds due to Firm's Commitment

Institutions Should Buy

Actual Repurchases

No Repurchases

Price Increases as Investors Discover Value

39

Table 1: Evidence of Post-Announcement Price Run-Up for Repurchase Announcements

For all 4,051 stock repurchase announcements from 2,259 unique firms from September 2007 to December 2013, we calculate the

cumulative total return (CRET) and cumulative abnormal return (CAR) using the Fama-French factors plus momentum risk model for

the estimation and testing periods corresponding to the periods used in the studies cited below. Date 0 is the repurchase announcement

day. Repurchase announcements are from SDC. Returns are from CRSP. The estimation and testing periods in the first two rows are

from Stephens and Weisbach (1998) and Babenko (2009). The testing periods in the last two rows are our modifications of Babenko

(2009) to include longer testing periods. ***, **, * stand for statistical significance at the 1%, 5%, and 10% level, respectively.

Method Corresponds to: Estimation

Period

Testing

Period CRET CAR

Obs = (4,051

Announcements) Mean p-value Mean p-value

Stephens and Weisbach (1998) [-165, -65] [-1, +1] 1.89%*** <.0001 1.75%*** <.0001

Babenko (2009) [-250 -50] [-1, +1] 1.89%*** <.0001 1.75%*** <.0001

Babenko (2009) w/new testing period [-250 -50] [0, +90] 6.06%*** <.0001 2.42%*** <.0001

Babenko (2009) w/new testing period [-250 -50] [-5, +5] 1.83%*** <.0001 1.76%*** <.0001

40

Table 2: Descriptive Statistics for Final Regression Datasets

This table reports the descriptive statistics for variables used in the next set of regression analysis. Profitability is measure as in Table

4. ANNOUNCEMENT is a dummy variable set to 1 if the firm announces share repurchases. INSIDER BUY is a dummy variable set

to 1 if insider purchases exceed insider sales during the six months before the announcement day by at least 0.01% of the firm’s market

capitalization. EPS SURPRISE measures the difference between the actual reported EPS and mean analyst EPS forecast divided by the

share price. INST. HOLDING Q-1 is calculated based on the reported holdings on Form 13F normalized by the number of shares

outstanding. EPS SURPRISE measures the difference between the actual reported EPS and mean analyst EPS forecast divided by the

share price. We also include other common control variables, such as the natural log of the firm’s market capitalization or the natural

log of the firm’s total assets as the FIRM SIZE during the announcement quarter, market risk premium (BETA), SMB and HML. The

final sample sizes based on data availability for all regression variables are 875 repurchase announcements for quarterly 13F institutional

holding changes regression, 679 announcements for intra-day ISO analysis regression, 1,154 announcements for Abel Noser institutional

trading imbalance regression, and 715 announcements for bi-weekly short seller regressions.

Panel A: Quarterly 13F

Mean Maximum Minimum Standard Deviation

PROFIT Q0 to Q+4 1.9730 105.0000 -93.4860 28.8940

ANNOUNCEMENT 0.5550 1.0000 0.0000 0.4970

INSIDER BUY 0.2180 1.0000 0.0000 0.4130

INST. HOLDING Q-1 0.2850 0.7580 0.0000 0.2930

EPS SURPRISE -0.0032 0.0663 -0.4754 0.0255

FIRM SIZE 19.0610 25.4058 12.6115 1.9957

BETA 1.6922 16.5500 -22.1900 10.2498

SMB 0.8097 12.0100 -6.0300 3.4163

HML -0.1512 23.8500 -13.6200 7.1290

41

Panel B: Daily Intermarket Sweep Order

Panel C: Daily Abel Noser

Mean Maximum Minimum Standard Deviation

PROFIT D0 to D+5 -0.0610 2.2410 -3.6960 0.5140

ANNOUNCEMENT 0.5270 1.0000 0.0000 0.5000

INSIDER BUY 0.3170 1.0000 0.0000 0.4660

INST. HOLDING Q-1 0.0390 0.3810 0.0000 0.0980

EPS SURPRISE -0.0004 1.9186 -1.8256 0.1277

FIRM SIZE 6.6190 13.2280 0.0000 2.6330

BETA -0.0622 11.3500 -8.9500 1.8483

SMB 0.0170 4.2900 -3.7900 0.7103

HML -0.0623 3.9500 -2.8000 0.7095

Mean Maximum Minimum Standard Deviation

PROFIT D0 to D+30 0.0000 0.2900 -0.3100 0.1000

ANNOUNCEMENT 0.5000 1.0000 0.0000 0.5000

INSIDER BUY 0.1900 1.0000 0.0000 0.4000

INST. HOLDING Q-1 0.5600 2.4000 0.0000 0.5000

EPS SURPRISE 0.0000 4.9700 -1.0200 0.1500

FIRM SIZE 6.4500 11.5900 1.3200 1.9900

BETA 0.0107 4.9700 -8.9500 1.5419

SMB 0.0064 3.6100 -3.6700 0.6961

HML 0.0476 3.0800 -3.5200 0.7216

42

Panel D: Biweekly Short Seller

Mean Maximum Minimum Standard Deviation

PROFIT BW0 to BW+8 -1.9200 26.1270 -46.8950 10.2720

ANNOUNCEMENT 0.5080 1.0000 0.0000 0.5000

INSIDER BUY 0.3100 1.0000 0.0000 0.4630

INST. HOLDING Q-1 0.2990 1.2470 0.0000 0.3880

EPS SURPRISE 0.0065 1.9186 -0.2371 0.0785

FIRM SIZE 7.2030 13.4670 2.4860 1.8460

BETA -0.0368 11.3500 -7.3600 1.7184

SMB 0.0398 4.2900 -3.4600 0.6835

HML -0.0221 3.5000 -2.6200 0.6516

43

Table 3: Institutional Trade Imbalance for 13F, ISO, Abel Noser and Short Selling

This table reports the trade imbalance for all four institutional types around stock repurchase announcements from September 2007 to

December 2013. The quarterly 13F institutional trade imbalance is calculated as the net shares purchased, the number of shares bought

minus the number of shares sold, in the quarter normalized by the number of shares outstanding in millions for 836 unique announcing

firms found in the 13-F dataset. The daily ISO trade imbalance is calculated as the actual number of buy shares (inferred by Lee and

Ready (1991) mechanism) minus the number of sell shares, normalized by the number of shares outstanding in millions, for all 1870

unique firms in the final sample. The daily Abel Noser trade imbalance is calculated as the actual number of buy shares minus the

number of sell shares, normalized by the number of shares outstanding in millions, for 543 unique firms found in Abel Noser dataset.

Abnormal short interest is calculated as the average number of shares sold short in the test period divided by the average number of

shares sold short in the benchmark period minus one (Christophe et al., 2004), for 1,753 unique firms found in short selling dataset. The

benchmark period is six biweekly periods before the announcement period, approximately one quarter before the announcement. Net

insider trading direction is separated into three categories: net buying, net selling and neutral broadly following Bonaime and Ryngaert

(2013). The announcement event is considered net selling if insider sales exceed insider purchases by at least 0.01% of the firm’s market

capitalization. The announcement event is considered net buying if purchases exceed sales by the same requirement. Otherwise, the

event is defined as neutral insider trading. These classifications are based on the transactions of insiders during the previous six months

relative to the announcement because insiders are found to be most active during this period based on the findings of Chan, Ikenberry,

Lee, and Wang (2012). ***, **, * stand for statistical significance at the 1%, 5%, and 10% level, respectively.

Panel A: Quarterly 13F Trade Imbalance Implied for Changes in Institutional Holdings

13F

Quarterly

Trade

Imbalance

Full

Sample

(Obs=1,265

Announcements )

Insider Buy

Subsample

(Obs=353

Announcements)

Insider Sell

Subsample

(Obs=403

Announcements)

Neutral Insider

Subsample

(Obs=509

Announcements)

Mean p-value Mean p-value Mean p-value Mean p-value

-5 1,100** 0.0351 -700 0.4461 4,700*** <.0001 -400 0.6478

-4 200 0.8276 -2,200* 0.0795 1,300 0.2312 900 0.4878

-3 500 0.4181 800 0.5494 200 0.8778 600 0.5014

-2 2,400*** <.0001 1,900** 0.0351 4,100*** <.0001 1,500** 0.0364

-1 -300 0.5440 -1,700 0.1072 1,600 0.1539 -900 0.1772

0 -4,600*** <.0001 -5,600*** <.0001 -3,300*** 0.0017 -5,000*** <.0001

+1 -4,700*** <.0001 -3,100*** 0.0008 -4,100*** <.0001 -6,300*** <.0001

44

Panel B: Intra-day ISO Trade Imbalance from TAQ

ISO Daily

Trade

Imbalance

Full

Sample

(Obs=3,334

Announcements)

Insider Buy

Subsample

(Obs=1,131

Announcements)

Insider Sell

Subsample

(Obs=1,576

Announcements)

Neutral Insider

Subsample

(Obs=627

Announcements)

Mean p-value Mean p-value Mean p-value Mean p-value

-5 to -1 -75.32*** <.0001 -68.46*** 0.0010 -87.85*** <.0001 -56.16*** 0.0097

0 -87.06** 0.0183 -141.40* 0.0828 -76.50 0.1139 -15.64 0.7344

+1 to +5 -75.86*** <.0001 -50.72*** 0.0007 -98.08*** <.0001 -65.36** 0.0461

Panel C: Daily Abel Noser Institutional Trade Imbalance

Abel Noser

Daily Trade

Imbalance

Full

Sample

(Obs=702

Announcements)

Insider Buy

Subsample

(Obs=192

Announcements)

Insider Sell

Subsample

(Obs=235

Announcements)

Neutral Insider

Subsample

(Obs=275

Announcements)

Mean p-value Mean p-value Mean p-value Mean p-value

-30 to -1 -281.50 0.6100 -1,980.40 0.1347 83.30 0.9350 595.30 0.3196

0 -9,485.10 0.1609 -28,524.90 0.2428 -1,377.60 0.5332 -3,107.50 0.1641

+1 to +30 -653.50 0.1672 -198.60 0.7114 -500.20 0.3988 -1102.00 0.2859

+2 -2,700*** <.0001 -3,000*** 0.0018 -1,700* 0.0565 -3,200*** <.0001

+3 -3,500*** <.0001 -3,700*** <.0001 -3,300*** 0.0002 -3,600*** <.0001

+4 -3,300*** <.0001 -2,900*** 0.0009 -2,400*** 0.0032 -4,200*** <.0001

+5 -1,500*** 0.0004 -1,600** 0.0416 0 0.9936 -2,700*** 0.0001

45

Panel D: Biweekly Abnormal Short Selling

Biweekly

Abnormal

Short Selling

Full

Sample

(Obs=3,196

Announcements)

Insider Buy

Subsample

(Obs=1,024

Announcements)

Insider Sell

Subsample

(Obs=1,465

Announcements)

Neutral Insider

Subsample

(Obs=707

Announcements)

Mean p-value Mean p-value Mean p-value Mean p-value

-5 7.23%* 0.0860 -21.31%** 0.0389 11.80%*** 0.0051 39.11%*** <.0001

-4 6.87%* 0.0950 -21.59%** 0.0306 11.31%*** 0.0060 38.90%*** <.0001

-3 7.23%* 0.0875 -20.75%** 0.0452 10.75%*** 0.0085 40.45%*** <.0001

-2 7.44%* 0.0801 -20.85%** 0.0459 11.52%*** 0.0052 39.96%*** <.0001

-1 8.13%* 0.0648 -19.47%* 0.0799 13.00%*** 0.0019 38.00%*** <.0001

0 15.55%*** 0.0019 -15.59% 0.1761 18.29%*** <.0001 54.99%*** <.0001

+1 18.01%*** 0.0002 -14.23% 0.2007 21.23%*** <.0001 58.05%*** <.0001

+2 19.00%*** 0.0001 -13.84% 0.2255 23.38%*** <.0001 57.47%*** <.0001

+3 18.40%*** 0.0003 -13.46% 0.2684 21.73%*** <.0001 57.65%*** <.0001

+4 17.23%*** 0.0005 -13.53% 0.2602 20.50%*** <.0001 55.02%*** <.0001

+5 16.89%*** 0.0006 -13.60% 0.2582 21.64%*** <.0001 51.19%*** <.0001

46

Figure 2: Quarterly Institutional Trade Imbalance Separated by Investment Behavior

This graph shows the cumulative trade imbalance of all institutional investors as well as separately for the transient, quasi-indexer, and

dedicated institutional investor around the announcement quarter (t=0). The trade imbalance is calculated as the net shares traded, the

number of shares bought minus the number of shares sold, in the quarter normalized by the number of shares outstanding. We separate

institutional trading by three types according to Bushee (2001). Transient institutional investors have high portfolio turnover and highly

diversified portfolio holdings. Transient investors are more focus on short-term gains. Dedicated institutional investors have very low

portfolio turnover and larger average portfolio investments. Quasi-indexer institutional investors also have low portfolio turnover but

highly diversified portfolio holdings. Both dedicated and quasi-indexer investors have longer investment horizons. The Institution line

below is the total institutional trade imbalance that can be separated into transient, dedicated, and quasi-indexer trade imbalances using

the method discussed in Bushee (2001), which means that cumulative Institution = TRA + DED +QIX. Our control firm quarterly

institutional trade imbalance is calculated using non-repurchase-announcing firms matched by industry and size.

-5,000

-4,000

-3,000

-2,000

-1,000

0

1,000

2,000

3,000

4,000

5,000

-5 -4 -3 -2 -1 0 1 2 3 4 5

Trad

e im

bal

ance

Quarter

Institution TRA DED QIX Control

Negative Trade Imbalance

Positive Trade Imbalance

47

Figure 3: Quarterly Trade Imbalance for the Firm, Insiders, and Institutions

This graph shows the trade imbalance of the firm, insiders and institutions around the announcement quarter (t=0). The trade imbalance

for the firm is calculated as the actual share repurchased normalized by the number of shares outstanding. The firm’s repurchases before

the announcement quarter (t=0) are from previous uncompleted repurchase programs. The trade imbalance for both the insiders and the

institutions is calculated as the net shares traded, the number of shares bought minus the number of shares sold, in the quarter normalized

by the number of shares outstanding.

-6,000

-4,000

-2,000

0

2,000

4,000

6,000

8,000

10,000

12,000

0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5

Trad

e im

bal

ance

Quarter

Institution Insider Firm

Positive Trade Imbalance

Negative Trade Imbalance

48

Table 4: Profitability for 13F, ISO, Abel Noser and Short Selling

This table reports the profitability for all four institutional types around stock repurchase announcements from September 2007 to

December 2013. For negative changes in Quarterly 13F holdings trade imbalances, we calculate profitability as the difference between

the closing prices of repurchase quarter and following quarter (to cover the position). Analogously, the opposite difference is used for

positive changes in holdings. For ISOs, daily institutional profit is determined using the exact ISO trade prices in each of the [0, +5] day

and CRSP daily closing prices on day +5 relative to the announcement: ISO Profitability = 𝑇𝑟𝑎𝑑𝑒 𝑝𝑟𝑖𝑐𝑒𝑡−𝐶𝑙𝑜𝑠𝑖𝑛𝑔 𝑝𝑟𝑖𝑐𝑒𝑡+5

𝑇𝑟𝑎𝑑𝑒 𝑝𝑟𝑖𝑐𝑒𝑡 for sells,

and ISO Profitability = 𝐶𝑙𝑜𝑠𝑖𝑛𝑔 𝑝𝑟𝑖𝑐𝑒𝑡+5−𝑇𝑟𝑎𝑑𝑒 𝑝𝑟𝑖𝑐𝑒𝑡

𝑇𝑟𝑎𝑑𝑒 𝑝𝑟𝑖𝑐𝑒𝑡 for buys. For Abel Noser, daily institutional profit is determined using the exact trade

prices in each of the [0, +30] day and CRSP daily closing prices on day +30 relative to the announcement. Similar to the ISO calculation,

the Abel Noser profitability is calculated according to the trade direction. Short-selling profit is calculated as the difference between the

proceeds from the sale and the cost to close the position using CRSP closing prices. Net insider trading direction is separated into three

categories: net buying, net selling and neutral broadly following Bonaime and Ryngaert (2013). The announcement event is considered

net selling if insider sales exceed insider purchases by at least 0.01% of the firm’s market capitalization. The announcement event is

considered net buying if purchases exceed sales by the same requirement. Otherwise, the event is defined as neutral insider trading.

These classifications are based on the transactions of insiders during the previous six months relative to the announcement because

insiders are found to be most active during this period based on the findings of Chan, Ikenberry, Lee, and Wang (2012). ***, **, * stand

for statistical significance at the 1%, 5%, and 10% level, respectively.

Institutional

Profitability

Full

Sample

Insider Buy

Subsample

Insider Sell

Subsample

Neutral Insider

Subsample

Mean Obs Mean Obs Mean Obs Mean Obs

13F: Q0 to Q+4 -0.75% 2,246 -0.79% 715 -1.05% 1,005 -0.12% 526

(0.1196) (0.3593) (0.1652) (0.8914)

ISO: D0 to D+5 -0.04%*** 3,174 -0.07%*** 935 -0.01% 1,580 -0.04%*** 659

(<.0001) (<.0001) (0.2158) (0.0005)

Abel Noser: 0.00% 702 -0.02% 192 0.00% 235 0.01% 275

D0 to D+30 (0.8522) (0.2034) (0.9268) (0.1984)

Short Seller: -2.15%*** 3,169 -2.81%*** 1,017 -2.27%*** 1,455 -0.94%** 697

BW0 to BW+8 (<.0001) (0.0005) (<.0001) (0.0114)

49

Table 5: ISO Trade Imbalance and Profitability around Share Repurchase Announcements and Contemporaneous Events

The table shows the trade imbalance and profitability of ISOs around share repurchase announcements on days with and without other

types of corporate events, such as dividend announcements, earnings announcements, M&A announcements, stock dividends and stock

splits. The daily ISO trade imbalance is calculated as the actual number of buy shares minus the number of sell shares, normalized by

the number of shares outstanding in millions. ISO profit is determined using the actual ISO buy or sell price for each given trade during

days [0, +5] as the opening price. We assume that the position is closed at the CRSP daily closing price on day t+5. ***, **, * stand for

the statistical significance of difference from zero at the 1%, 5%, and 10% level, respectively.

Panel A: Daily ISO trade imbalance around repurchase announcements:

ISO

Trade Imbalance

No Other Contemporaneous

Corporate Events

(Obs=1,259)

With Contemporaneous

Corporate Events

(Obs=853 )

Mean p-value Mean p-value

-5 to -1 -118.10*** <.0001 -77.99*** <.0001

0 -64.99 0.1664 -110.50 0.1008

+1 to +5 -82.86*** <.0001 -136.90*** 0.0006

Panel B: Profitability of ISO trades around repurchase announcements:

ISO

Profitability

No Other Contemporaneous

Corporate Events

(Obs=1,259)

With Contemporaneous

Corporate Events (Obs=853)

Mean p-value Mean p-value

0 to +5 -0.05%*** 0.0014 -0.03%* 0.0568

50

Table 6: ISO Trade Imbalance and Profitability around Share Repurchase Announcements Adjusting for Quarter Ends

The table shows the trade imbalance and profitability of ISOs around share repurchase announcements adjusting for quarter ends.

Announcements and trades during the last three days of each quarter are eliminated from the sample. The daily ISO trade imbalance is

calculated as the actual number of buy shares minus the number of sell shares, normalized by the number of shares outstanding in

millions. ISO profit is determined using the actual ISO buy or sell price for each given trade during days [0, +5] as the opening price.

We assume that the position is closed at the CRSP daily closing price on day t+5. ***, **, * stand for the statistical significance of

difference from zero at the 1%, 5%, and 10% level, respectively.

Panel A: Daily ISO trade imbalance around repurchase announcements:

ISO Trade Imbalance Adjusting Quarter Ends (Obs=2,078)

Mean p-value

-5 to -1 -100.20*** <.0001

0 -79.74** 0.0423

+1 to +5 -106.20*** <.0001

Panel B: Profitability of ISO trades around repurchase announcements:

ISO Profitability Adjusting Quarter Ends (Obs=2,078)

Mean p-value

0 to +5 -0.05%*** <.0001

51

Table 7: Test of Endogeneity

This table reports the results of the Hausman Test for endogeneity. Using this test, we can

determine if the announcement signal is indeed endogenous to the institutional trade imbalance

during the announcement quarter. First, we regress the endogenous variable on the instrumental

variables (IVs). The endogenous variable is ANNOUNCEMENT, which is a dummy variable set

to 1 if the firm announces share repurchases. We use three IVs, ΔCASH, DILUTION, and TAXGAP,

in our test. We calculate ΔCASH as the change in EBITDA/Total Assets over the previous quarter.

DILUTION is calculated as one minus the ratio of diluted EPS/ basic EPS. TAXGAP is calculated

as the difference between the average tax rate on dividends and the average tax rate on capital

gains. We follow the method outlined in De Lisle et al. (2014) for these IVs.

ANNOUNCEMENT= 𝛼 + 𝛽1𝛥𝐶𝐴𝑆𝐻 + 𝛽2𝐷𝐼𝐿𝑈𝑇𝐼𝑂𝑁 + 𝛽3𝑇𝐴𝑋𝐺𝐴𝑃 + 𝑣

Dependent Variable:

𝐴𝑁𝑁𝑂𝑈𝑁𝐶𝐸𝑀𝐸𝑁𝑇

Coefficient

Estimate t-statistic p-value

𝛥𝐶𝐴𝑆𝐻 -1.4051* -1.77 0.0779

𝐷𝐼𝐿𝑈𝑇𝐼𝑂𝑁 -0.8587 -1.23 0.2175

𝑇𝐴𝑋𝐺𝐴𝑃 -0.2801*** -7.37 <.0001

INTERCEPT 4.6132*** 8.42 <.0001

Adj. R-Square 0.0727

Observations 714

52

In the second stage, we save the residuals, v, and include the residuals as an extra variable in the

second estimation.

𝑇𝑅𝐴𝐷𝐸𝐼𝑀𝐵𝐴𝐿𝐴𝑁𝐶𝐸 = 𝛼 + 𝛽1𝐴𝑁𝑁𝑂𝑈𝑁𝐶𝐸𝑀𝐸𝑁𝑇 + 𝛽2𝐼𝑁𝑆𝐼𝐷𝐸𝑅𝐵𝑈𝑌 +𝛽3𝐼𝑁𝑆𝑇. 𝐻𝑂𝐿𝐷𝐼𝑁𝐺(𝑄−1) + 𝛽4𝐸𝑃𝑆𝑆𝑈𝑅𝑃𝑅𝐼𝑆𝐸 + 𝛽5𝐹𝐼𝑅𝑀𝑆𝐼𝑍𝐸 + 𝛽6𝑣 + 𝜀

In the second estimation, the dependent variable is the quarterly institutional trade imbalance in

the announcement quarter, TRADEIMBALANCE, which is calculated as the net shares purchased,

the number of shares bought minus the number of shares sold, in the quarter normalized by the

number of shares outstanding in millions. ANNOUNCEMENT is a dummy variable set to 1 if the

firm announces share repurchases. INSIDERBUY is a dummy variable set to 1 if insider purchases

exceed insider sales during the six months before the announcement day by at least 0.01% of the

firm’s market capitalization. INST. HOLDING Q-1 is calculated based on the reported holdings on

Form 13F normalized by the number of shares outstanding. EPSSURPRISE measures the

difference between the actual reported EPS and mean analyst EPS forecast divided by the share

price. FIRMSIZE is the natural log of the firm’s market capitalization. Finally, we examine the t-

statistics of 𝛽6 to determine if endogeneity exists. If 𝛽6 is equal to zero, the there is no endogeneity.

If 𝛽6 is not equal to zero, then endogeneity exists. Our data is Winsorized at the 1% and 99%

levels. ***, **, * stand for statistical significance of difference from zero at the 1%, 5%, and 10%

level, respectively.

Dependent Variable: TRADEIMBALANCE

Coefficient

Estimate t-statistic p-value

ANNOUNCEMENT -0.0081 -0.50 0.6201

INSIDERBUY 0.0032 0.69 0.4891

INST. HOLDING Q-1 -0.0116 -0.89 0.3749

EPSSURPRISE 0.0340 0.49 0.6229

FIRMSIZE 0.0013 0.99 0.3224

v -0.0006 -0.04 0.9687

INTERCEPT -0.0121 -0.43 0.6709

Adj. R-Square 0.0247

Observations 714

53

Table 8: Regression Analysis of 13F Profitability on the Announcement and Insider Signal

This table reports the regression results where the dependent variable is the quarterly institutional trade profit from Q0 to Q+4. Our

sample consists of 875 events from 307 announcing firms and 307 matching firms between 2007 and 2013. ANNOUNCEMENT is a

dummy variable set to 1 if the firm announces share repurchases. INSIDER BUY is a dummy variable set to 1 if insider purchases

exceed insider sales during the six months before the announcement day by at least 0.01% of the firm’s market capitalization. INST.

HOLDING Q-1 is calculated based on the reported holdings on Form 13F normalized by the number of shares outstanding. EPS

SURPRISE measures the difference between the actual reported EPS and mean analyst EPS forecast divided by the share price. We also

include other common control variables, such as the natural log of the firm’s market capitalization as the firm size during the

announcement quarter, market risk premium, SMB, and HML. To address multicollinearity issues when fitting our regression models,

we eliminated any variable with a VIF score higher than 10. Our data is Winsorized at the 1% and 99% levels. ***, **, * stand for the

statistical significance of difference from zero at the 1%, 5%, and 10% level, respectively.

Dependent Variable:

Quarterly 13F Profit

Announcement Signal

(Obs=875)

Insider Buy

Signal (Obs=875)

Announcement &

Insider Buy Interaction

(Obs=875)

Coefficient

Estimate p-value

Coefficient

Estimate p-value

Coefficient

Estimate p-value

ANNOUNCEMENT -10.2299 0.1621 -7.8186 0.2937

INSIDER BUY -0.2883 0.9290 4.6278 0.2738

ANNOUNCEMENT*

INSIDER BUY

-12.3125* 0.0504

INST. HOLDING Q-1 6.7484 0.5364 -5.0710 0.4677 5.7101 0.6007

EPS SURPRISE 115.5757 0.2214 125.6476 0.1837 112.3733 0.2342

FIRM SIZE -1.1022 0.4042 -0.1490 0.8959 -1.4240 0.2896

BETA 0.1484 0.5823 0.1905 0.4802 0.1097 0.6861

SMB -0.4073 0.6180 -0.4229 0.6062 -0.3177 0.6979

HML 0.2034 0.5739 0.1650 0.6483 0.1852 0.6082

INDUSTRY FIXED EFFECTS YES YES YES

YEAR FIXED EFFECTS YES YES YES

R-SQUARE 0.3949 0.3926 0.3995

54

Table 9: Regression Analysis of ISO Profitability on the Announcement and Insider Signal

This table reports the regression results where the dependent variable is the average daily ISO profit from day 0 to day +5. Our sample

consists of 679 events from 255 announcing and 255 matching firms between 2007 and 2013. ANNOUNCEMENT is a dummy variable

set to 1 if the firm announces share repurchases. INSIDER BUY is a dummy variable set to 1 if insider purchases exceed insider sales

during the six months before the announcement day by at least 0.01% of the firm’s market capitalization. INST. HOLDING Q-1 is

calculated based on the reported holdings on Form 13F normalized by the number of shares outstanding. EPS SURPRISE measures the

difference between the actual reported EPS and mean analyst EPS forecast divided by the share price. We also include other common

control variables, such as the natural log of the firm’s total assets as the firm size during the announcement quarter, market risk premium,

SMB, and HML. To address multicollinearity issues when fitting our regression models, we eliminated any variable with a VIF score

higher than 10. Our data is Winsorized at the 1% and 99% levels. ***, **, * stand for the statistical significance of difference from zero

at the 1%, 5%, and 10% level, respectively.

Dependent Variable:

Daily ISO Profit

Announcement Signal

(Obs=679)

Insider Buy

Signal (Obs=679)

Announcement & Insider

Buy Interaction (Obs=679)

Coefficient

Estimate p-value

Coefficient

Estimate p-value

Coefficient

Estimate p-value

ANNOUNCEMENT -0.1154 0.1092 -0.1030 0.2143

INSIDER BUY -0.1404* 0.0848 -0.1208 0.2414

ANNOUNCEMENT*

INSIDER BUY

-0.0450 0.7394

INST. HOLDING Q-1 0.8887** 0.0278 0.6943* 0.0601 0.9802** 0.0163

EPS SURPRISE -0.0679 0.8581 -0.1649 0.6666 -0.1654 0.6659

FIRM SIZE 0.0125 0.3696 0.0059 0.6768 0.0070 0.6206

BETA 0.0272 0.1692 0.0280 0.1562 0.0288 0.1451

SMB 0.0039 0.9280 0.0060 0.8906 0.0047 0.9136

HML 0.0604 0.2433 0.0510 0.3259 0.0529 0.3083

INDUSTRY FIXED

EFFECTS YES YES YES

YEAR FIXED EFFECTS YES YES YES

R-SQUARE 0.6174 0.6181 0.6223

55

Table 10: Regression Analysis of Abel Noser Profitability on the Announcement and Insider Signal

This table reports the regression results where the dependent variable is the average daily institutional profit from day 0 to day +30. Our

sample consists of 1,154 events from 577 announcing and 577 matching firms between 2007 and 2013. ANNOUNCEMENT is a dummy

variable set to 1 if the firm announces share repurchases. INSIDER BUY is a dummy variable set to 1 if insider purchases exceed insider

sales during the six months before the announcement day by at least 0.01% of the firm’s market capitalization. INST. HOLDING Q-1

is calculated based on the reported holdings on Form 13F normalized by the number of shares outstanding. EPS SURPRISE measures

the difference between the actual reported EPS and mean analyst EPS forecast divided by the share price. We also include other common

control variables, such as the natural log of the firm’s market value as the firm size during the announcement quarter, market risk

premium, SMB, and HML. To address multicollinearity issues when fitting our regression models, we eliminated any variable with a

VIF score higher than 10. Our data is Winsorized at the 5% and 95% levels. ***, **, * stand for the statistical significance of difference

from zero at the 1%, 5%, and 10% level, respectively.

Dependent Variable:

Daily Abel Noser Profit

Announcement Signal

(Obs=1,154)

Insider Buy

Signal (Obs=1,154)

Announcement & Insider

Buy Interaction

(Obs=1,154)

Coefficient

Estimate p-value

Coefficient

Estimate p-value

Coefficient

Estimate p-value

ANNOUNCEMENT -0.0129** 0.0222 -0.0121* 0.0678

INSIDER BUY -0.0058 0.4942 0.0014 0.9272

ANNOUNCEMENT*

INSIDER BUY

-0.0036 0.8493

INST. HOLDING Q-1 0.0222*** 0.0090 0.0189** 0.0245 0.0221*** 0.0095

EPS SURPRISE 0.0107 0.6153 0.0103 0.6278 0.0104 0.6239

FIRM SIZE -0.0003 0.9290 -0.0001 0.9598 -0.0003 0.9185

BETA -0.0017 0.6147 -0.0018 0.5914 -0.0017 0.6114

SMB 0.0080 0.2205 0.0083 0.2046 0.0080 0.2215

HML 0.0128* 0.0673 0.0126* 0.0725 0.0128* 0.0678

INDUSTRY FIXED EFFECTS YES YES YES

YEAR FIXED EFFECTS YES YES YES

R-SQUARE 0.3963 0.3925 0.3963

56

Table 11: Regression Analysis of Short Sales Profitability on the Announcement and Insider Signal

This table reports the regression results where the dependent variable is the biweekly short selling profit from biweekly 0 to biweekly

+8. Our sample consists of 715 events from 274 announcing and 274 matching firms between 2007 and 2013. ANNOUNCEMENT is a

dummy variable set to 1 if the firm announces share repurchases. INSIDER BUY is a dummy variable set to 1 if insider purchases

exceed insider sales during the six months before the announcement day by at least 0.01% of the firm’s market capitalization. INST.

HOLDING Q-1 is calculated based on the reported holdings on Form 13F normalized by the number of shares outstanding. EPS

SURPRISE measures the difference between the actual reported EPS and mean analyst EPS forecast divided by the share price. We also

include other common control variables, such as the natural log of the firm’s total assets as the firm size during the announcement

quarter, market risk premium, SMB, and HML. To address multicollinearity issues when fitting our regression models, we eliminated

any variable with a VIF score higher than 10. Our data is Winsorized at the 1% and 99% levels. ***, **, * stand for the statistical

significance of difference from zero at the 1%, 5%, and 10% level, respectively.

Dependent Variable:

Biweekly Short Seller Profit

Announcement Signal

(Obs=715)

Insider Buy

Signal (Obs=715)

Announcement & Insider Buy

Interaction (Obs=715)

Coefficient

Estimate p-value

Coefficient

Estimate p-value

Coefficient

Estimate p-value

ANNOUNCEMENT -1.5684 0.3526 -1.2119 0.5180

INSIDER BUY -0.5184 0.7092 0.0129 0.9942

ANNOUNCEMENT*

INSIDER BUY

-0.9153 0.6878

INST. HOLDING Q-1 2.9686 0.2129 1.2225 0.4214 2.7965 0.2465

EPS SURPRISE -5.4594 0.4141 -6.0541 0.3637 -5.3208 0.4283

FIRM SIZE 0.0302 0.9449 0.0449 0.9200 -0.0093 0.9835

BETA -0.3446 0.3501 -0.3329 0.3681 -0.3398 0.3592

SMB 1.7243** 0.0340 1.6479** 0.0421 1.7266** 0.0344

HML -0.5654 0.5579 -0.6589 0.4978 -0.5955 0.5415

INDUSTRY FIXED EFFECTS YES YES YES

YEAR FIXED EFFECTS YES YES YES

R-SQUARE 0.5828 0.5818 0.5831