Inattention to Merger Announcements and Strategic Disclosure Timing Fabian Dreher, 11085231 July 7, 2016 University of Amsterdam, Amsterdam Business School MSc Business Economics, Finance Track Master Thesis Supervised by Dr. Florian Peters
Inattention to Merger Announcements and Strategic
Disclosure Timing
Fabian Dreher, 11085231
July 7, 2016
University of Amsterdam, Amsterdam Business School
MSc Business Economics, Finance Track
Master Thesis
Supervised by Dr. Florian Peters
Abstract
Do firms announce mergers strategically in response to investor inattention? This thesis
examines the market reaction of stocks to merger announcements made on days of increased
inattention. By focusing on mergers with a negative immediate market response of the ac-
quiring firm, I find that those firms perform significantly worse if the merger is announced
on a high inattention day than if the merger is announced on a normal day. Mergers an-
nounced on a day of a high number of earnings announcements show 2.5% lower cumulative
abnormal returns in the 30-day period after the announcement and mergers announced on a
day of a major sports event show 2.9% lower cumulative abnormal returns over this period.
These findings are in line with previous literature on strategic disclosure timing of managers
and support the claim that market inattentiveness is being exploited by managers around
merger announcements. Prior results to strategic disclosure timing on inattention days like
Fridays and days before national holidays can not be confirmed for merger announcements.
Acknowlegdements
I would like to express my gratitude to my supervisor Dr. Florian Peters for the useful
comments, remarks and engagement throughout the learning process of this master thesis.
Without his assistance and constant feedback this thesis would have not been accomplished
in the way presented.
Statement of Originality
This document is written by student Fabian Dreher who declares to take full responsi-
bility for the contents of this document.
I declare that the text and the work presented in this document is original and that no
sources other than those mentioned in the text and its references have been used in creating
it.
The Faculty of Economics and Business is responsible solely for the supervision of com-
pletion of the work, not for the contents.
Contents
1 Introduction 1
2 Literature Review 6
3 Methodology and Hypotheses 11
3.1 Hypotheses . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
3.1.1 Hypothesis 1: Investors show less attention to merger announcements
when there is a high number of earnings announcements on the same
day and on days of major sports events . . . . . . . . . . . . . . . . 12
3.1.2 Hypothesis 2: Managers exploit investor inattention by announcing
less profitable mergers on days of a high number of earnings announce-
ments and on days of major sports events . . . . . . . . . . . . . . . 12
3.2 Methodology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
3.3 Empirical Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
3.4 Control Variables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
4 Data and Descriptive Statistics 17
4.1 Announcement Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
4.1.1 Control Variables from Dataset . . . . . . . . . . . . . . . . . . . . . 17
4.1.2 Control Variable Outside of Dataset . . . . . . . . . . . . . . . . . . 18
4.2 Stock Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
4.2.1 Volume . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
4.2.2 Returns . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
4.2.3 Data Preparation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
4.2.4 Independent Variables . . . . . . . . . . . . . . . . . . . . . . . . . . 20
4.2.5 Descriptive Statistics . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
5 Empirical Results 24
5.1 Limited Attention to Merger Announcements . . . . . . . . . . . . . . . . . 25
5.1.1 Trading Volume . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
5.1.2 Immediate Stock Return Response . . . . . . . . . . . . . . . . . . . 29
5.2 Strategic Disclosure Timing of Mergers . . . . . . . . . . . . . . . . . . . . . 31
6 Robustness Checks 37
6.1 Post-announcement Drift and Convergence . . . . . . . . . . . . . . . . . . 37
6.2 Extension of Observation Period . . . . . . . . . . . . . . . . . . . . . . . . 38
6.3 Terciles and Quartiles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
7 Conclusion and Discussion 44
References 53
Appendix 57
1 Introduction
This research investigates strategic management behavior around the announcement of
mergers and acquisitions1. With respect to the large volume of empirical work on inatten-
tiveness in the stock market which has been done over the last several years, it is especially
interesting to see the implications of these findings on management decision making and
disclosure timing in particular. Prior literature in this field documents strategic disclosure
timing on Fridays and before national holidays (Niessner, 2015) and when shareholders are
distracted (Kempf, Manconi and Spalt, 2015). My research focuses on the announcement
of mergers and acquisitions on days of an increased number of earnings announcements or
days of major sports events. I propose that on these days markets are less attentive to
merger announcements and managers strategically announce worse mergers at that time. I
find clear evidence for this hypothesis, which is visible in a significantly worse performance
of mergers that are announced on these inattention days, compared to mergers announced
on normal days.
The underlying assumption of this research is that capital markets are not efficient. As
a result, managers have goals besides maximizing shareholder value. In the context of merg-
ers and acquisitions, corporate actions can also have a detrimental effect on the shareholder
value as managers favor private utility maximization over the shareholders’ preferences.
An example is a case of diversifying mergers which are usually taken out to stabilize cash
flows but are not in the best interests of shareholders who seek out diversification on an
individual level. Since in most cases, merger announcements are being kept secret until
their official publication, managers face a certain time window for the disclosure. I pro-
pose that managers prefer to choose days of high inattention for the announcement of less
profitable mergers (seen from an investor’s perspective) and days of low inattention for the
announcement of more profitable mergers. In specific, I propose two sources of inattention.
First, clustering of earnings announcements on a single day leads to distraction from merger
announcements and second, days of major sports events have the same distraction effect.
1From here on, the term “merger” will be used in order to reference to both “mergers” and“acquisitions”
1
My methodology relies on the idea that inattention is visible in an under-reaction of the
stock market, as shown by DellaVigna and Pollet (2008). A merger announced on a day
of high inattention should show a lower abnormal trading volume and a lower abnormal
stock return response. This is summarized in my first hypothesis, which states that merg-
ers announced on days of earnings announcement concentration or days of major sports
events show lower abnormal trading volume and lower immediate stock return response on
the day of the announcement compared to other (non-inattentive) days. My second hy-
pothesis is that managers preferably announce less profitable mergers on these inattention
days. I hereby look at the long-run performance of those mergers and apply a difference-in-
difference methodology to compare announcements on inattention days and normal days.
In an attempt to differentiate between “good” and “bad” mergers, the sample is split into
mergers showing a positive immediate stock return response and mergers showing a nega-
tive immediate stock return response. The difference in performance is to be observed from
lower cumulative abnormal returns in the bottom group.
I first document that even though abnormal trading volume and immediate stock return
response are lower on inattention days, these differences do not prove to be significant.
These results are in contrast to the literature which reports significant differences for merger
announcements on Fridays and other days (Louis and Sun, 2010). However, the sole focus
on immediate stock return response can not be used as a proof or disproof of inattention as
inattention is ultimately observable in a post-announcement drift (DellaVigna and Pollet,
2009). The findings for long-run returns in fact show a clearer result of investor inattention
as well as strategic disclosure timing. The difference in the bottom group between merger
announcements on days of earnings announcement concentration and normal days is 2.5%
30-days after the announcement and 2.9% between merger announcements on days of major
sports events and other days. These results also persist for an extension of the observation
period to 60 days following the announcement and for a refinement of the classification into
terciles and quartiles. Controlling for firm and deal specific characteristics, the difference
in performance of mergers announced on days of major sports events jumps to a significant
2
9.3% lower cumulative abnormal return. For days of earnings announcement concentration,
significance does not persist if controlled for firm and deal specific characteristics.
The organization of the thesis is as follows. In chapter 2, an outlay of the most relevant
literature is provided with the associated contribution to the existing literature. I position
the thesis in the field of behavioral finance and explain the main theories existing in this
field. Prior studies document different patterns of inattention. The work of Barber and
Odean (2008) shows that less sophisticated investors tend to focus on salient stocks. Cohen
and Frazzini (2008) point out that investors also miss subtle links. The findings are closely
related to the finding of less attention payment towards the distant future (DellaVigna
and Pollet, 2007) and inattention to earnings announcements on Fridays (DellaVigna and
Pollet, 2009). Louis and Sun (2010) were able to document a similar effect for merger
announcements. All these findings of investor’s inattentiveness rais the question of whether
this inattention has an impact on corporate actions. More recent literature by Niessner
(2015) provides evidence that managers strategically disclose negative corporate events on
Fridays. Also Kempf, Manconi and Spalt (2015) show that managers strategically announce
value-destroying mergers when shareholders are distracted. The aim of this research is to
contribute to the existing literature by observing strategic behavior in disclosure timing of
merger announcements more closely. I introduce two distraction variables which lead to an
under-reaction of stock markets. First, the findings of Hirshleifer et. al (2006) indicate that
multiple earnings announcements at the same time can lead to a distraction of investors
and prices to “react sluggishly to relevant news”. Second, as observed by Ehrmann and
Jansen (2012) major sports events can be a significant distraction to shareholders.
Chapter 3 describes the methodology used in this research. As mentioned earlier, the
main hypothesis I propose is a strategic merger announcement timing by managers. In
specific, less profitable mergers are announced on high-inattention days. Since this research
does not focus on Friday inattention2, for which inattention is evidenced quite clearly, there
2An extension of this analysis to Friday inattention and inattention before national holidays can be found
in the Appendix
3
is the need to introduce another hypothesis on which the main hypothesis will be based
on. This basic hypothesis is that investors show less attention to merger announcements
on days of a high number of earnings announcements and days of major sports events.
These two hypotheses are assessed by means of an event study. Cumulative abnormal
returns are calculated from the market model in the immediate time window around the
announcement (i.e. 3-days around the announcement) and in the long-run (30-days after
the announcement). I use a difference-in-difference methodology to see how the performance
of mergers announced on inattention days differs from mergers announced on other days.
In order to get around the problem of a cancellation of negative and positive returns, the
sample is split into “good” mergers, which show a positive immediate return response and
“bad” mergers, which show a negative immediate stock return response. The hypothesis of
strategic disclosure timing focuses on the bottom group as significantly worse performance
is expected here.
In chapter 4 I characterize my sample of merger announcements from January 1985 to
December 2014. Two variables of inattention are constructed. First, earnings announce-
ment concentration days are defined as all days that are in the top 10th percentile of total
quarterly earnings announcements made on a day over the sample. Second, major sports
event days are defined as the finals of NBA, MLB and NFL. I also construct necessary
control variables for robustness checks. Summary statistics can be found under this section.
Chapter 5 presents the results of the analysis. The first hypothesis can not be supported
by the difference in abnormal trading volumes. Even though a visual inspection of mean
abnormal trading volumes on days of inattention and normal days shows that trading volume
is in fact lower on the proposed days of inattention, the following regression results do not
support these findings, as no statistical significance can be found. Moreover, the positive
correlation of trading volume and earnings announcements makes an analysis of the pure
effect of merger announcements on trading activity difficult. There is also no clear evidence
for the first hypothesis to be found in the analysis of abnormal returns. The regression
results of abnormal return in a 3-day window around the merger announcement show a
4
negative effect of the proposed inattention days on returns but are not statistically robust.
The results for the second hypothesis are clearer. Both the visual inspection as well as
the OLS regression give evidence for strategic disclosure timing. Plotting the cumulative
abnormal returns over a 60-day window shows a worse performance of mergers announced
on inattention days which even persists in the long-run without catching up to the control
group. The results can be verified by OLS regression, which shows a significant worse
stock performance in the bottom group if mergers are announced on inattention days. The
difference is -2.5% for earnings announcement concentration days and -2.9% for days of
major sports events. With the introduction of control variables, no significant difference is
to be observed for announcements made on earnings announcement concentration days but
for mergers announced on days of major sports events. The difference for the latter one is
9.4% between the variable of interest and the control group in the bottom group.
In chapter 6 I conduct some robustness checks for the provided results. It can be shown
that there is no clear catching up effect of mergers that are announced on high inattention
days. This supports the claim of strategic disclosure timing, as a difference in performance
solely due to inattention and a slower incorporation of news into prices would have meant a
post-announcement drift resulting in a convergence of the two groups. As this convergence
does not occur, I propose that mergers announced on inattention days are in fact worse and
have strategically been disclosed by managers on these days. I also analyze the performance
difference in the long run (60-day period instead of 30 days) and find that the regression
results confirm my prior results. Finally, as the classification of mergers into positive and
negative immediate stock return responses is merely rough, the classification is refined into
tercile and quartile grouping. Also for this refinement, the prior results can be achieved
and statistical robustness be confirmed. For all that, the introduction of control variables
has a detrimental effect on the significance of the results.
Chapter 7 covers the conclusion drawn from this analysis. I infer that managers indeed
exploit investor inattention in their disclosure timing decision. Nevertheless, I rate the
economic relevance as low since the observed inattention days do not happen frequently
5
enough to be chosen for merger announcements on a regular basis.
In the Appendix I show the results found for Friday merger announcements and merger
announcements made before national holidays. The strategic announcement timing hy-
pothesis can not be confirmed for these two variables, which is in contrast to the previous
literature.
2 Literature Review
The intention of this research is to contribute to the existing literature in several ways. On
a broader level it goes into the field of Behavioral Finance. As discovered by Kahneman
and Tversky (1979), individuals do not always behave fully rational which is a controversy
to traditional economic beliefs and assumptions. Simon (1957) argues that people have
limited cognitive capacity and therefore can overlook or disregard important information.
This inattention can lead to under-reaction in the stock market as shown by Hong and Stein
(1999) and is the main assumption underlying my thesis.
Investor inattention has been discovered by different studies, challenging the the Efficient
Market Hypothesis. Barber and Odean (2008) find that less sophisticated investors, like
retail investors, tend to buy salient stocks and while disregarding less attention-catching
stocks. This is due to limited attention, whereas institutional investors for example can focus
on more stocks at the same time. In summary, retail investors are net buyers of attention-
catching stocks as opposed to institutional investors who are net-sellers of attention catching
stocks. This ultimately leads to an upward price pressure in salient stocks (Da, Engelberg
and Gao, 2011).
Huberman and Regev (2011) outline inattention to news by examining stock price re-
actions of a pharmaceutical firm which had very early positive results on a cure for cancer.
The stock reaction was less pronounced when the article was featured in a science magazine
than when the same article was published in the New York Times and The Wall Street
Journal.
Cohen and Frazzini (2008) investigate investor inattention to subtle economic links. By
6
observing customer-supplier links they find that investor inattention leads to predictable
supplier returns and a resulting arbitrage opportunity. The time needed for information to
fully incorporate into prices is about one year. Other researchers had similar findings (e.g.
Hong, Torous and Valkanov, 2007; Driesprong, Jacobsen and Maat, 2008; Cohen and Lou,
2012). Inattention exists also with respect to the distant future as shown by DellaVigna
and Pollet (2007) by looking at demographic shifts and the predictability of industry-level
stock returns.
These patterns continue to exist even in times of global information accessibility. A
model by Nieuwerburgh and Veldkam (2009) tries to explain the home country bias puzzle,
investor’s natural tendency to invest in the domestic market instead of global markets.
They find that investors do not choose to learn what foreigners know, due to the fact that
they profit more from knowing information which others do not know. Nieuwerburgh and
Veldkam’s conclusion is that learning amplifies information asymmetry.
The main inspiration for this thesis is the discovery of inattention of investors to Friday
earnings announcements by DellaVigna and Pollet (2009). DellaVigna and Pollet examine
the stock market reaction to earnings surprises, comparing Friday to Non-Friday announce-
ments to see whether the weekend acts as a distraction factor. The earnings surprise is
calculated as the difference between the actual earnings per share and the consensus analyst
forecast (median of forecasts), normalized by the stock price. The corresponding abnormal
return is calculated from the market model and used to observe the short-run stock market
response as well as the post-announcement drift. In order to disentangle negative and pos-
itive surprises, DellaVigna and Pollet sort the earnings surprises into quintiles. What they
find is that inattention on Fridays can lead to a 15% lower immediate response and 70%
higher delayed response to earnings surprises as well as lower trading volume. Moreover, a
riskless portfolio investing into the Friday drift and shorting the non-Friday drift earns sub-
stantial abnormal returns. Their findings open a very important question which will be the
main topic of this thesis. Do firms time their announcements strategically? This thesis tries
to fill the research gap of this “largely unexplored field”, according to Baker and Wurgler
7
(2012). Existing literature already suggests strategic disclosure timing, for example Patell
and Wolfson (1982) state that “good news is more likely to be released when the security
markets are open while bad news appears more frequently after the close of trading”.
Niessner and Madsen (2015) indicate the strategic behavior of managers who increase
advertising ahead of positive earnings announcements. The authors identify the importance
of advertising in financial markets by documenting an increase in Google searches for com-
pany tickers after the company publishes print advertisements. These advertising efforts
are strategically being increased in the weeks around earnings announcement if the earnings
surprise is positive.
This is in line with the findings of Solomon (2012) on selective publicity. By examining
media coverage of good and bad corporate news, he finds that investor relations firms
“spin their clients news” and generate a higher media coverage of positive press releases
than negative press releases, resulting in increased announcement returns. This effect is
weakened around earnings announcements. Solomon draws the conclusion that “IR firms
causally affect both media coverage and returns”.
The goal of this thesis is to investigate a specific and also one of the most important
companies news - the announcement of mergers and acquisitions. Louis and Sun (2010) take
a similar approach to DellaVigna and Pollet (2009) and show that investor inattention on
Fridays exists around merger announcements as well. Their analysis focuses on stock swap
announcements only, arguing that cash acquisitions have an “economically small” market
reaction. I believe that even if this is the case in their dataset, there is no economic argu-
ment supporting this. If a company announces a stock swap merger, the market reaction
is naturally more negative than a comparable cash acquisition because the selling company
8
takes part in the shareholder value added (SVA)34. My thesis therefore looks at cash acqui-
sitions as well. The results of Louis and Sun document a lower abnormal trading volume of
acquirers to Friday announcements and less pronounced abnormal stock returns. They also
find evidence for the so called Merger-Monday effect, a phenomenon on Wall Street that
mergers are preferably announced on Mondays for better publicity. These two findings do
not contradict each other, but indicate that investor attention varies over different week-
days. CEOs have the incentive to draw advantage of that, for example through stock-based
compensation.
Current research by Niessner (2015) provides evidence that managers “strategically ma-
nipulate their company’s information environment to extract private benefits”. She makes
use of the SEC requirement for managers to disclose certain corporate events within five
business days. Since managers still have the freedom to choose the day of the week, she
hypothesizes that bad news is preferably disclosed on high inattention days like Fridays and
before holidays. In fact, she finds that negative events are disproportionally disclosed on
these days which results in a significant under-reaction of the market lasting up to three
weeks. In a second step, Niessner observes how managers profit from their strategic disclo-
3Consider the following example: Suppose there is a buyer company and a seller company. Market
capitalization of the buyer is $5 billion, made up of 50 million shares priced at $100 per share. The sellers
market capitalization stands at $2.8 billion - 40 million shares each worth $70. The assumed synergy value
is $1.7 billion. If the buyer offers to buy all the shares of the seller at $100 per share, the value placed on the
selling company is $4 billion, representing a premium of $1.2 billion over the companys preannouncement
market value of $2.8 billion. The shareholder value added (SVA) is simply the expected synergy of $1.7
billion minus the $1.2 billion premium, or $500 million, in a cash acquisition. But if the buyer finances the
acquisition by issuing new shares, the SVA for its existing stockholders will drop. If the buyer offers one of
its shares for each of the sellers shares and the offer places the same value on the seller as did the cash offer,
upon the deals completion, the acquiring shareholders’ ownership in the buyer company will be reduced.
They will own only 55.5% of a new total of 90 million shares outstanding after the acquisition. So their
share of the acquisition’s expected SVA is only 55.5% of $500 million, or $277.5 million. The rest goes to
sellers shareholders, who are now shareholders in an enlarged company.4Example taken from https://hbr.org/1999/11/stock-or-cash-the-trade-offs-for-buyers-and-sellers-in-
mergers-and-acquisitions
9
sure. She confirms that manager insider trading activities take advantage of the disclosure
timing. I take her findings as a foundation for my analysis as I expect to find similar
patterns for merger announcements.
Finally, and most important for this thesis, Kempf, Manconi and Spalt (2015) observe
the implications of distracted shareholders on corporate actions, hereby constructing a
measure of distraction. Distraction measures are constructed by using exogenous shocks to
unrelated parts of the shareholders’ portfolios. In line with their findings of the distraction
effect being concentrated around diversifying mergers, I expect the announcement of less
well performing mergers being concentrated on days of high inattention. Still, this “Strategic
Disclosure Hypothesis” is not undisputed. Doyle and Magilke (2008) also analyze disclosure
timing by using firm-level tests that focus only on firms that switch their disclosure timing.
The underlying argument is that firms which consistently report earnings at the same time
can not be considered to act strategically. As a result, they find “no evidence that managers
opportunistically report worse news after the market closes or on Fridays”. My thesis is
supposed to contribute to the current work on behavioral corporate finance by filling the
gap of strategic manager behavior in the context of M&A announcements and help to solve
this ongoing argument.
I propose four types of inattention days which I am going to observe in my analysis. First
of all, I assume that a source of distraction is a concentration of earnings announcements
as proposed by Hirshleifer et al. (2006) who find that multiple earnings announcements at
the same time can lead to distraction of investors and prices “react sluggishly to relevant
news”. I propose that on days of many earnings announcements, investors pay less attention
to merger announcements than on other days, due to the fact that cognitive capacities are
limited. Second, I assume days of major sports events to be distracting and leading to higher
inattention. Since the days of these sports events are known upfront, firms can strategically
make merger announcements on these days to exploit investor inattention. Evidence for
investor inattention during major sports events is given by an analysis of the European
Central Bank on investor inattention during FIFA world cup matches by Ehrmann and
10
Jansen (2012). The authors find that when the national team is playing, the number of
trades drops by 45% and volumes are 55% lower. Moreover, they find that co-movement
between national and global stock market returns decreases by over 20% during matches
of the national team. A paper on investor inattention to earnings announcements during
major sports events in the US by Santos (2009) documents more negative abnormal two-day
returns of announcements falling on sports event days. The observed sport events are the
Super Bowl, World Series of Baseball and NBA finals which I am also going to use in my
analysis.
Thirdly, and in line with DellaVigna and Pollet (2009), inattention is supposed to be
higher on Fridays compared to other weekdays. This is due to the fact that investors get
distracted by the upcoming weekend and pay less attention than on other days. Finally,
days before national holidays should have the same or an even more negative effect on
investor attention than Fridays. Evidence is given by Pantzalis and Ucar (2014) who find
that earnings announcements released during the Holy Week (Easter) are associated with a
significant post-earnings announcement drift. Positive earnings surprises show a drift of up
to 4% and negative earnings surprises even up to 10%, suggesting that negative events are
even less thoroughly incorporated into prices on high inattention days than positive events.
The latter two variables are covered in the appendix.
3 Methodology and Hypotheses
The thesis examines strategic behavior of managers around their timing of merger announce-
ments. My hypothesis consists of two parts, which will be further explained in the following.
The research focuses on two sources of inattention/ distraction. First, a high number of
earnings announcements and and second, days of major sports events.
11
3.1 Hypotheses
3.1.1 Hypothesis 1: Investors show less attention to merger announcements
when there is a high number of earnings announcements on the same
day and on days of major sports events
Hypothesis 1 aims to find inattention to merger announcements on certain days of assumed
inattention. First, when a lot of firms announce their earnings, the attention is shifted
towards those and reduced attention is available for merger announcements. This hypoth-
esis is based on Simons’ (1957) argument that the human minds capacity is limited. The
same accounts for days of major sports events, for which I assume investors being more
distracted than on other days and showing less attention towards merger announcements.
This hypothesis is closely related to the work of DellaVigna and Pollet (2009), but instead
of earnings announcements I analyze merger announcements. It is also related to the work
of Louis and Sun (2010) but looks at different inattention factors (for Friday announcements
see Appendix).
To evidence inattention, I am looking at trading volumes and initial stock returns. I
directly compare abnormal trading volume and cumulative abnormal returns around the
announcement of mergers on the proposed days of inattention and other days.
3.1.2 Hypothesis 2: Managers exploit investor inattention by announcing less
profitable mergers on days of a high number of earnings announcements
and on days of major sports events
If inattention exists around days of a high number of earnings announcements and on days
of major sports events, I hypothesize that managers take advantage by announcing less
profitable mergers on these days. There are two main reasons for that, first, major sports
events and earnings announcements are set in advance and managers can thereof plan their
merger announcements accordingly. Secondly, the managers are aware of this inattention
and act strategically to “hide” less profitable mergers on days of high inattention. Even
though there might be a post announcement drift and eventually a market correction, they
12
can profit in the short run. To evidence this hypothesis, I look at the performance of
mergers which are announced on days of inattention and compare it to mergers announced
on normal days.
Since there is the likelihood of a cancellation of positive and negative returns in the
aggregate, there is the need to to differentiate between “good” and “bad” mergers. Because
of that I split my sample into mergers showing a positive immediate stock return response
and mergers that showing a negative immediate stock return response. The difference in
performance should be seen from lower cumulative abnormal returns in the bottom group.
3.2 Methodology
The basic idea to assess the proposed hypotheses is to observe the effect a merger announce-
ment has on the stock return by means of an event study.
Before looking at stock returns, I start with the analysis of trading volumes and how
they differ between days of announcement. The inattention hypothesis suggests that trading
volume is lower on days of inattention than on other days. To assess this difference, I
examine abnormal trading volumes, i.e. the difference between the normal trading volume
and the increased volume due to the announcement of the merger. First a plotting of
abnormal trading volume will show whether it is lower on days of high inattention or not.
The findings are quantified in a regression of the following type:
∆v(0,1)t,k = α+ β1X + controls
where ∆v(0,1)t,k is the abnormal trading volume for firm k on the day of the merger
announcement and the subsequent day. α is the constant in the regression and β1 is the
coefficient for the inattention variable X which either indicates whether an announcement
was made on a day of high inattention (1) or not (0) or whether an announcement was
made on a day of a major sports event (1) or not (0).
To further assess hypothesis 1, I look at the immediate stock response by means of the
abnormal stock return around the day of the merger announcement. The event window to
13
be observed is -1/+1 day (3-day window) of the announcement. The abnormal return is
calculated from the market-adjusted model by taking the difference of the market return
and the stock return.
Ri = Rmt + εit
where Ri is the stock return, Rmt is the market return in period t and εit is the abnormal
return.
To measure the abnormal return over the observed time period, the cumulative abnormal
return (CAR) is calculated over the windows -1 to +1 leading to
CARi (−1,+1) =1∑
t=−1
εit
To see a difference in announcements made on days of inattention and normal days, this
thesis will borrow from the Difference-in-Difference methodology where an announcement
on a day of inattention is seen as the “treatment”. A dummy variable indicates whether an
announcement was made on a day of high inattention (1) or not (0). Announcements made
on a normal day act as a control group, as they are not affected by the treatment of interest,
but similarly affected by the announcement. Under the Efficient Market Hypothesis, the
announcements made on an inattention day should behave like the announcements made
on a normal day.
The OLS specification is
R(−1,+1)t,k = α+ β1X + controls+ εt,k
where R(−1,+1)t,k is the abnormal return of company k at time t one day before the
merger announcement, the day of the announcement and the following day. β1 measures
the difference in the variables observed. X constitutes a dummy variable for the inattention
variables being observed. In the case of earnings announcements, it is a dummy for high
earnings announcement concentration days and in the other case a dummy for days of major
sports events as before.
14
Moreover, if managers do not act strategically in their timing of merger announcements,
mergers announced on inattention days should perform similarly to mergers announced on
normal days without significant difference.
Finally, the long-term response is defined as the cumulative abnormal return over the
period -1 to +30 days around the announcement:
CARi (−1,+30) =1∑
t=−1
εit
The effect will be inspected in a graphical form by plotting the mean outcome per period
over the event time for the treatment and the control group.
Finally, a regression shows whether there are differences in the announcement days and
whether the mergers perform differently.
3.3 Empirical Model
To quantify the findings, I examine the stock performance of acquiring companies whose
merger announcement led to a negative immediate stock return response. The OLS speci-
fication is
R(−1,30)t,k = α+ β1X + β2bottom.group+ β3 (X ∗ bottom.group) + controls+ εt,k
where R(−1,30)t,k denotes the abnormal stock returns for company k at time t between 1
day before the announcement and 30 days after the announcement. α is the constant in the
estimated regression. X is the independent variable of interest which indicates the source of
inattention. In the case of earnings announcements, it constitutes a dummy variable taking
the value 1 for days of high number of earnings announcements (90th percentile) and 0
if not (below 90th percentile). The coefficient β1 measures the abnormal return difference
between those two announcement days. In case of a major sports event, X takes the value 1
if an announcement was made on such a day and 0 otherwise. The coefficient measures the
abnormal return generated due to an announcement on a day of a major sports event. The
second variable bottom.group indicates whether the merger announcement had a negative
15
immediate stock market response (1) or not (0). Most important, X ∗ bottom.group is
an interaction variable of the bottom group indicator and the inattention variable. The
coefficient β3 shows the difference in abnormal returns due to the fact that an announcement
was made on an inattention day.
3.4 Control Variables
In contrast to other corporate events like earnings announcements, merger announcements
are much more complicated to assess by the market. Different control variables are needed
to control for other factors which influence the market reaction to merger announcements.
First of all, the profitability of the target as well as the acquirer have to be controlled
for. This is self-explanatory as more profitable targets should have a more positive effect on
the stock price. Since the profitability is usually priced into the bid price it is also necessary
to control for the transaction value.
The size of the acquirer plays an important role in the magnitude that a deal of a
certain size can have on the stock price. I hereby control for the market size of the acquirer
four weeks prior to the announcements in order to have a less influenced price (due to
information leaks for example). A firm that holds high amounts of cash is also more likely
to do value-destroying acquisitions (Harford, 1999), which is why a control variable for cash
is also necessary. It has been shown that mergers which are done over different industries
are less profitable compared to mergers inside industries, so-called diversifying mergers.
These ones will be controlled for as well.
The likelihood of the outcome is a very important factor on the market reaction but
can not be directly observed at the time of the announcement. I control for this effect by
considering the actual outcome of the merger announcement, assuming that the market can
assess the likelihood of the announcement on the day of the announcement correctly.
Finally, it is necessary to control for certain factors which effect inattention. Firstly,
the number of bidders can give an idea of how important a deal is and how much attention
will be given to this deal. Secondly, the public awareness of a company shows how much
16
possible inattention can exist. A control variable for this is the analyst coverage as I expect
a company with higher analyst coverage to be less affected by inattention than companies
with lower analyst coverage.
4 Data and Descriptive Statistics
4.1 Announcement Data
The source for merger announcement data is Thomson One. I restrict the whole sample
of merger announcements to non-financial firms, since in most cases, financial firms do
acquisitions as part of their business model. Consequently, strategic behavior should not be
observed for financial firms. The sample is also restricted to publicly traded firms located
in the US. The minimum deal size taken into consideration is 5 million USD. Smaller deals
are less likely to have an observable impact on the market value. The deal size restriction
also leads to ignorance of all mergers for which no transaction value has been reported.
Finally, the data range is set to a 30-year period from 01/01/1985 to 12/31/2014.
The resulting sample consists of 41,662 announcements made in this time period.
4.1.1 Control Variables from Dataset
In addition to the announcement data the following control variables are introduced. The
first control variable is the status of the merger which is reported in Thomson One as
“Status”. The string variable is converted into a dummy variable, which takes the value 1
for completed mergers and 0 otherwise (including “Withdrawn”, “Intented”, “Unknown”,
etc.). The next control variable is the type of merger, separating into diversifying and non-
diversifying mergers. For this variable the Thomson Financial Macro Code (“TF Macro
Code”) of the target is being matched with the acquirer’s code and a dummy results taking
the value 1 for diversifying mergers (i.e. mergers over different macro industries) and 0 for
non-diversifying mergers (i.e. mergers within the same macro industry). Another control
variable is the number of bidders for a single deal. To control for the size of the transaction,
17
I calculate the ratio of the acquirer’s market value to the transaction value. The acquirer’s
market value is defined as the stock price times shares outstanding, four weeks prior to the
announcement. In addition, I calculate the acquirer’s market to book ratio by dividing the
market value (as described above) by the book value (total assets as reported in the balance
sheet). As a control variable for the profitability of the acquirer and the target, I take the
earnings per share as reported in Thomson One. Finally, I create a control variable for the
acquirer’s cash holdings by dividing the cash position by the total value of assets. This
variable and the market to book ratio are winsorized at the 5% and 90% level. All other
variables are winsorized at lower levels of 1%-5% to 95%-99%.
4.1.2 Control Variable Outside of Dataset
To control for the analyst coverage of an acquiring firm, I create another variable, namely
analyst coverage. Analyst coverage is defined as the number of analysts reporting a year
end earnings estimate for a company in a certain year. This data is taken from IBES and
matched with the announcement data using a link table5.
4.2 Stock Data
Corresponding data for stock prices is retrieved from CRSP. In a first step, the CUSIPs
from Thomson One are matched with the CRSP permanent number (PERMNO). I then
take stock data for the period 01/01/1984 to 12/31/2015 in order to also have lagged data
and forward-looking data.
4.2.1 Volume
In order to calculate daily volumes, I calculate an average daily price by taking the average
of the highest ask price and the lowest bid price during a day. The volume is calculated as
the product of the average price and the total number of shares traded that day.
5The link table was provided by supervisor Dr. Florian Peters.
18
The abnormal volume is defined as the difference of the volume of the day of announce-
ment and the normal volume of a stock. Normal volume is defined as the average arithmetic
volume over the time period -20 to -11. I calculate the abnormal volume for the time frame
0 to +1 of the announcement.
Consequently, the measure for abnormal volume is
∆v(0,1)t,k =
1∑u=0
log(V ut,k) −
−11∑u=−20
log(V ut,k
)/10
Where (Vt,k)u is the value of shares traded on the u-th trading day after the merger
announcement of company k at time t. ∆v(0,1)t,k is the immediate abnormal volume on the
day of the merger announcement and the following day.
4.2.2 Returns
I calculate the abnormal return according to the market model:
εit = Ri −Rmt
Where Ri is the daily stock return and Rmt is the value-weighted CRSP market return
on the same day. The short run response is calculated as the cumulative abnormal return
over days -1 to +1 around the announcement:
CARi (−1,+1) =1∑
t=−1
εit
I calculate the long-run response as the cumulative abnormal return over days -1 to +30
CARi (−1,+30) =30∑
t=−1
εit
For the return calculation I create a workweek calendar which only takes into account
trading days, consequently the lagged value of a Monday return is the prior Friday return.
19
4.2.3 Data Preparation
For the merge of the announcement data and the stock data the CRSP permanent numbers
(PERMNO) are used as identifiers. 3,341 observations can not be matched and are neglected
as the effect on the outcome is assumed to be insignificant. In line with DellaVigna and
Pollet (2009), supposedly illiquid stocks like penny stocks are dropped as well. The threshold
is set to a $5 minimum stock price. Finally, a small number of announcements made on
Saturday and Sunday are dropped as well. The resulting sample consists of 41,622 merger
announcements with corresponding stock data.
4.2.4 Independent Variables
Earnings Announcements. As stated before, a concentration of earnings announcements
is assumed to be a factor of distraction. I retrieve earnings announcement data via I/B/E/S
on quarterly earnings announcements. This data is contracted by day which results in the
total number of earnings announcements made on a single day. Since at some points of
this analysis real numbers are not handy enough to draw conclusions, a dummy variable is
created to indicate announcement concentration days. Therefore, the top 10% in numbers of
announcements are tagged as announcement concentration day (1) and the remaining 90%
non-announcement concentration day (0). The calculated threshold is 339 announcements.
20
Figure 1: Number of Earnings Announcements made per Day
Notes: The graph shows the total number of earnings announcements per day over the period from January 1985 to December 2014.
Quarterly earnings announcement data is downloaded from I/B/E/S and contracted by day. The turquoise line indicates the threshold
for the 90th percentile (highest 10% above this line).
Major Sports Events. Sport plays an important role in the US and most of the US
citizens follow the national leagues. I assume that major sports events are also factors of
distractions if they fall on a trading day. The most important events for the biggest sports
are the Super Bowl in American Football, the World Series of Baseball and the NBA finals in
Basketball. These events are considered in the analysis. Since the Super Bowl traditionally
falls on a Sunday night, I assume that the following Monday will be hallmarked by higher
inattention.
21
Table 1: Distribution of Sports Event Variable by Event Type
Notes: The table shows the number of major sports events in the period from 01/01/1985 to 12/31/2014 falling on the same day of at
least one merger announcement. Data is retrieved from sports-reference.com. NBA Finals are the championship series of the National
Basketball Association, World Series of Basketball the championship series of the Major Baseball League. Super Bowl is the final game
of the National Football League. For NBA and MLB games, only the days falling on a weekday are taken into account, dropping
all weekend games. Since the Super Bowl falls on a Sunday, the following Monday is considered as an inattention day. The column
“Number” shows the absolute number of events by type, the column “Percentage” shows the percentage falling onto each type of event.
Number Percentage
NBA Finals 116 52.489
World Series of Baseball 76 34.389
Super Bowl 29 13.122
Total 221 100
4.2.5 Descriptive Statistics
In Table 2, I present summary statistics for the two variables of interest, earnings announce-
ments and major sport events, and the control variables used in this analysis. 7.2% of the
merger announcements in the sample were made on a day of a high number of earnings
announcements. Considering the fact that the dummy variable for earnings announcement
concentration days covers 10% of all days in the sample, this number is lower than expected.
3.2% of the merger announcements were made on days of major sports events which is high
considering that only 2.8% of days in the observation period hosted a major sports event.
Finally, 42.3% of the merger announcements are categorized into the bottom group which
means that most merger announcements showed a positive immediate market response.
Approximate location for Table 2
Table 3 shows the differences in firm characteristics by announcement day. I merge the
variables Announcement Concentration and Major Sports Event into a single inattention
variable. As a result, one can observe differences in firm characteristics of firms announcing
mergers on inattention days and firms announcing on non-inattention days. Surprisingly,
the target as well as the acquirer have higher earnings per share for mergers made on the
22
Table 2: Summary Statistics
Notes: The table shows summary statistics of the independent and control variables. Data on merger announcements is taken from
Thomson One and covers the period from 01/01/1985 to 12/31/2014. Announcements on Saturdays and Sundays are not included the
sample. “Announcement Concentration” is a dummy variable indicating whether the number of quarterly earnings announcements on a
single day was in the 90th percentile of all daily quarterly earnings announcements (data for earnings announcements is retrieved from
I/B/E/S). “Major Sports Event Dummy” is a dummy variable indicating whether the announcement was made on a day of a major
sports event (NBA finals, MLB finals, Super Bowl). “Bottom Group Dummy” is a dummy variable indicating whether the immediate
stock return response (+1/+1 day of the announcement) was negative (1) or not (0). “EPS Target” are the earnings per share of the
target (sell-side) firm. “EPS Acquirer” are the earnings per share of the acquirer (buy-side) firm, respectively. “Log (Pre-announc. MV)
Acquirer” is the natural logarithm of the acquiror’s market value, defined as share price times number of shares outstanding in million
US$, 4 weeks prior to the announcement. “Deal Value to Acquirer Size” is the ratio of the deal (transaction) value to the acquirer size
(“Pre-announc. MV Acquirer” as defined above, not logarithmized). “Log (MV/BV) Acquirer” is the natural logarithm of the ratio of
the market value of assets (“Pre-announc. MV Acquirer”) to the book value of assets of the acquirer. “Acquirer Cash Holdings” is the
ratio of cash assets to net total asset value of the acquirer. “Number of Bidders” is the number of competing bidders for the target
company. “Dummy for completed Mergers” is a dummy variable that takes the value 1 for completed mergers and 0 for any other status.
“Dummy for diversif. Mergers” is a dummy variable which takes the value 1 for mergers over different macro industries (according to
the Thomson Financial Macro Hierarchy) or 0 otherwise. “Analyst Coverage” is the number of analysts who publish quarterly earnings
estimates of the acquiring company in the year of announcement. Data for this variable is taken from I/B/E/S.
observed inattention days. Also the market value of acquirers who make announcements
on inattention days is significantly higher by almost 100%. The deal size is more or less
equal. The market-to-book ratio of acquirers announcing on inattention days is higher than
for the counterpart, possibly due to the high observed market value. Cash holdings and
number of bidders are almost equal for both groups. Mergers that are announced on high
inattention days are completed less frequently and are only in 19% of the cases diversifying,
compared to a 22% of the rest of the sample. Interestingly, analyst coverage is higher for
the companies who announce their mergers on inattention days. This could be interpreted
as an indicator of strategic behavior. In particular, firms who are already at the center of
23
attention might try to make use of inattention days as they are aware of the attention they
usually receive.
Table 3: Differences in Company Characteristics
Notes: The table shows the differences in company characteristics (control variables) on whether the company announced a merger on
a high inattention day or not. Data on merger announcements is taken from Thomson One and covers the period from 01/01/1985 to
12/31/2014. Announcements on Saturdays and Sundays are not included the sample. Column (1) shows the arithmetic mean for the
variable on the left. The variables “Earnings Announcement Concentration Dummy” and “Major Sport Event Dummy” are summed
up into a single “Inattention” variable, a dummy variable that indicates whether a merger announcement was made on a day of a high
number of earnings announcements or on a day of a major sports event (column 1) or not (column 2). Column 3 shows the difference
in means with the corresponding significance level presented as p-values in column 4. “EPS Target” are the earnings per share of the
target (sell-side) firm. “EPS Acquirer” are the earnings per share of the acquirer (buy-side) firm, respectively. “Log (Pre-announc. MV)
Acquirer” is the natural logarithm of the acquiror’s market value, defined as share price times number of shares outstanding in million
US$, 4 weeks prior to the announcement. “Deal Value to Acquirer Size” is the ratio of the deal (transaction) value to the acquirer size
(“Pre-announc. MV Acquirer” as defined above, not logarithmized). “Log (MV/BV) Acquirer” is the natural logarithm of the ratio
of the market value of assets (“Pre-announc. MV Acquirer”) to the book value of assets of the acquirer. “Acquirer Cash Holdings”
is the ratio of cash assets to net total asset value of the acquirer. “Number of Bidders” is the number of competing bidders for the
target company. “Dummy for completed Mergers” is a dummy variable that takes the value 1 for completed mergers and 0 for any
other status. “Dummy for diversif. Mergers” is a dummy variable which takes the value 1 for mergers over different macro industries
(according to the Thomson Financial Macro Hierarchy) or 0 otherwise. “Analyst Coverage” is the number of analysts who publish
quarterly earnings estimates of the acquiring company in the year of announcement. Data for this variable is taken from I/B/E/S.
5 Empirical Results
In this section I present the empirical results. As explained under Methodology, there are
two main hypotheses. First, there is limited attention to merger announcements on days
of high number of earnings announcements and days of major sports events. The empirical
results for this hypothesis can be found under 5.1. I first look at abnormal trading volumes
24
on the day of the merger announcement and in a second step at the immediate stock return
response.
Second, the hypothesis that mergers are announced strategically on days of higher inat-
tention is covered under 5.2 by looking at the performance of different mergers.
5.1 Limited Attention to Merger Announcements
This section provides evidence that investors are distracted by a high number of earnings
announcements and major sports events. First, I look at the abnormal trading volume on
days of inattention and on normal days. I then examine whether the immediate stock return
response differs between inattention days and normal days.
5.1.1 Trading Volume
Trading volume is the underlying mechanism that causes prices to adjust (DellaVigna and
Pollet, 2009). If there is inattention in the market, this should also be visible in a lower
than normal trading volume. DellaVigna and Pollet find lower abnormal trading volumes
for earnings announcements on Fridays and interpret it as evidence for investor distraction.
These results are confirmed by Louis and Sun (2011) for merger announcements on Fridays.
Following these findings, I try to show limited attention for the proposed inattention days
by means of abnormal trading volume.
Graphical evidence. Figure 2 plots the average abnormal volume of acquiring compa-
nies to merger announcements made on days of a high number of earnings announcements
and days of a low/normal number of earnings announcements over the day of announce-
ment and the subsequent day. The vertical axis shows the logarithmic abnormal trading
volume, defined as the difference between the normal volume (arithmetic volume over the
time period -20 to -11) and the day of observation. The abnormal volume on the day of the
announcement is shown by the dark blue bar and the abnormal volume on the following
day by the turquoise bar. The sum of the abnormal volume on these two days is shown
by the grey bar. On the left side of the horizontal axis one can see the average abnormal
25
trading volumes for days of a high number of earnings announcements. On the right side of
the horizontal axis one can see the average abnormal trading volumes on days of a lower/
normal number of earnings announcements. It can be seen that abnormal volume is about
25% lower on days of a high number of earnings announcements, which is in line with prior
findings to inattention on Fridays.
Figure 2: Abnormal Volume at Announcement by Earnings Announcement Concentration
Days
Notes: The figure shows abnormal trading volume per day, differentiated by days of a high number of earnings announcements and days
of a low/normal number of earnings announcements. Merger announcements in Thomson One from January 1985 until December 2014
are matched to stock prices and volumes in CRSP. Time 0 in event time is the day of the merger announcement. The abnormal trading
volume is the average logarithmic volume on the day of the merger announcement and the following day, minus the average log volume
for the period -20 to -11. “High # of Earn. Announc.” indicates whether the number of quarterly earnings announcements on a single
day is in the 90th percentile of all daily quarterly earnings announcements (data for earnings announcements is taken from I/B/E/S).
Figure 3 shows the same graph for days of major sports events. Abnormal trading
volume is also positive, showing that merger announcements in general lead to an increased
trading volume. Comparing the two different types of announcement days, namely days
of major sports events and normal days, a lower abnormal trading volume can be seen for
26
acquiring firms who announce their mergers on days of major sports events. The difference
is about 10%, which is lower than the difference previously observed.
Figure 3: Abnormal Volume at Announcement by Major Sports Event Days
Notes: The figure shows abnormal trading volume per day, differentiated by days of major sports events and normal days. Merger
announcements in Thomson One from January 1985 until December 2014 are matched to stock prices and volumes in CRSP. Time 0
in event time is the day of the merger announcement. The abnormal trading volume is the average logarithmic volume on the day of
the merger announcement and the following day, minus the average log volume for the period -20 to -11. “Day of Major Sports Event”
indicates whether the announcement was made on a day of a major sport event (NBA finals, MLB finals, Super Bowl). Data for this
variable is taken from sports-reference.com.
Regression. To quantify the findings above I regress the abnormal trading volume
on days of merger announcement onto the type of announcement day, i.e. the inattention
source. The OLS specification is
∆v(0,1)t,k = α+ β1X + controls
where ∆v(0,1)t,k is the abnormal trading volume for firm k on the day of the merger
announcement and the subsequent day. α is the constant in the regression and β1 is the
coefficient for the inattention variable X, which is either a dummy variable for days of
27
earnings announcement clustering or a dummy variable for days of major sports events.
The coefficient β1 shows the percentage difference in abnormal trading volume between
merger announcements made on inattention days and announcements made on normal days.
Table 4 shows the regression results. Column 1 shows a highly significant coefficient for
days of an increased number of earnings announcements. The coefficient of 0.139 means
that on average, abnormal trading volume in acquiring firms is 13.9% higher if mergers are
announced on days of earnings announcement clustering. This is in contrast to prior findings
on inattention like DellaVigna and Pollet (2009), who find a negative effect of inattention
days on trading volume. A possible explanation is that trading activity generally increases
with the number of earnings announcements (Frazzini and Owen, 2007) and consequently
the inattention effect is getting diluted. An example would be a company who announces
their earnings and a merger at the same time, making a disentangling of trading volume
due to the merger announcement and trading volume due to the earnings announcement
impossible. Another explanation is correlation existent in the stock market, which is also
reflected by trading volume. If many companies announce their earnings, trading volume
increases also around companies who don’t announce earnings on that day, for example
because of ETF trading. The introduction of control variables for merger and company
specifications in column 2 changes the coefficient into a negative 10.7% which is more in
line with the inattention hypothesis. But with a standard error of -1.17 it is not statistically
significant.
In the case of major sports events (columns 3 and 4), the coefficients which show the
difference in abnormal trading volume depending on the day of announcement are also
positive but neither significant with nor without control variables. One can conclude that
the abnormal trading volume in merger announcing companies is not dependent on whether
the announcement was made on a day of a major sports event.
Columns 5 and 6 show regressions with both independent variables with and without
controls. The picture drawn is similar to the previous results described.
28
Approximate location for Table 46
Summary. A visual inspection of mean abnormal trading volumes on days of inattention
and normal days shows that trading volume is in fact lower on the proposed days of inat-
tention, a finding that is in line with previous literature. Regardless, the regression results
do not support these findings, as no statistical significance can be found. Moreover, the
positive correlation of trading volume and earnings announcements makes a pure analysis
of the effect on merger announcement trading activity difficult.
5.1.2 Immediate Stock Return Response
Next to abnormal trading volume, also the responsiveness of stock returns is affected by
inattention. The logical reasoning is similar to abnormal trading volume. If investors are
distracted, the immediate stock return response should be lower than on days without
distraction.
Regression. To observe this effect, I regress abnormal return, defined as the stock
return subtracted by the market return, on the inattention variables earnings announcement
clustering and days of major sports events. This is similar to the regression I did for
abnormal trading volume, except that the dependent variable is now the abnormal return
instead of the abnormal trading volume.
The OLS specification for the immediate stock response is
R(−1,+1)t,k = α+ β1X + controls+ εt,k
whereR(−1,+1)t,k is the abnormal stock return for company k at time t between 1 day before
the merger announcement and 1 day after the merger announcement. The sample includes
all merger announcements. α is the constant in the regression and β1 is the coefficient for the
inattention variable X which is either a dummy variable for days of earnings announcement
clustering or a dummy variable for days of major sports events. Under Hypothesis 1 of
6Table 4 can be found on page 47
29
inattentive markets, β1 should be significantly lower than 0 in the short-run. This means
that the impact, a merger announcement has on the acquiring company’s stock is less
pronounced on inattention days than on normal days.
The OLS specification allows for a set of control variables, which are described in section
4.1.
The regression results are presented in table 5. Column 1 and 2 show the results for
the variable earnings announcement concentration without and including control variables.
Column 3 and 4 show the regression results for the variable major sports events without and
including control variables respectively. Finally, columns 5 and 6 show regression results
for all variables.
As seen in column 1, the abnormal return of acquiring companies in a 3-day window
around the merger announcement is 0.3% lower on days of earnings announcement con-
centration days than on normal days. This figure is significant in a statistical sense. Even
though the economic impact is low, mostly due to the fact that positive and negative returns
cancel each other out, there is a negative effect of the inattention variable on returns on
average. Introducing control variables, the effect is lower but not significant anymore. The
same accounts for major sports events, which lead to a 0.4% lower immediate abnormal
return than on normal days. Still, the explanatory power of the regressions is low consid-
ering the adjusted R-squared figures. Even with the introduction of control variables, the
percentage variation, which can be explained by the regression is only 1.8%.
Approximate location for Table 57
Summary. The regression results of abnormal return in a 3-day window around the merger
announcement show a negative effect of the proposed inattention days on returns. These
can be seen as an indicator for inattentiveness, but is not statistically robust. A refinement
of the results can be found under Chapter 6, robustness checks.
7Table 5 can be found on page 48
30
5.2 Strategic Disclosure Timing of Mergers
The previous section gave evidence that earnings announcement clustering and major sports
events, in fact, act as a distraction factor to investors around merger announcements. This
leads to the second hypothesis I propose, namely that managers take advantage of this
inattention by announcing less profitable mergers on these days. If the hypothesis is true,
the long-term stock return of acquiring firms should be more negative if the announcement
was made on a day of inattention, than on other days. What I expect is that managers
preferably announce bad mergers on high inattention days, to exploit the fact that investors
are distracted and having a less pronounced negative effect on the company’s stock price.
On the other hand, if managers do not expect an major negative market reaction, they are
indifferent of the day of the announcement.
For a better differentiation I split the sample into two groups of merger announcements.
First, the ones that showed a positive immediate response (i.e. the mergers which had an
abnormal return smaller than 0 in the period -1 to +1 day of the announcement). I name
this group the top group. Second, the merger announcements which lead to a negative
immediate stock return response are grouped in the bottom group. I expect the hypothesis
of strategic disclosure timing to be verified in the bottom group.
First of all, I present graphical evidence by plotting the long-term cumulative abnormal
returns of mergers announced on inattention days and mergers announced on normal days.
Next, I present an OLS regression which shows the quantified difference in announcement
days.
Graphical evidence. The figures show the development of cumulative abnormal re-
turns (CARs) of the acquiring companies over the time period of 1 day before the announce-
ment to 60 days after the announcement. The calculation for the CARs can be found under
section 4.2.2. Each graph shows two lines for which each either presents the development
of stock returns of mergers made on inattention days or the development of stock returns
in the control group.
Figure 4 and 5 show the results for the variable earnings announcement concentration.
31
Figure 4 shows the stock reaction over time for announcements made on days of a high
number of earnings announcements compared to other days in the top group. Figure 5
shows the same development for the bottom group. In the top group, a clear performance
difference is visible. Announcements made on days of earnings announcement concentration
days show a better performance from day 1 onwards. The difference in returns is about
1 to 1.5% in cumulative abnormal returns. Interestingly, there is no catch-up in the long
run, but mergers in the top group announced on inattention days continue to outperform
mergers in the control group even after a period of 60 days. A possible explanation for this
is that strategic behavior might be two-sided. A concentration of earnings announcements
on a single day might have a distraction effect on some firms, but also leads to a shift
in attention to other firms. A company that is in the spotlight due to positive earnings
announcements might have the incentive to also announce a profitable merger at the same
time for better publicity.
Figure 5 shows the same development of CARs in the 60-day period following the merger
announcement for companies in the bottom group. The picture drawn is similar to the
findings in the top group, whereas the difference in returns varies more over time. Starting
from a difference of about 1.5% in immediate stock return response, it looks like there is
a catching up after around 20 days. But looking at the longer time periods of 30 and 60
days, the difference even increases more. After 60 days the CARs for mergers announced on
earnings concentration days is still about 2% lower than in the control group with no sign of
a catching up. With respect to the explanation above, I see evidence for strategic behavior
of managers in the bottom group as these companies try to hide their less profitable mergers
in a distraction environment when the focus is shifted to other (earnings announcements
reporting) companies.
Figure 6 and 7 show the same graph for days of major sports events. In figure 6 the
plots show the development of CARs of acquiring firms in the top group which announced
a merger either on a day of a major sports event (dark green line) or on a normal day (light
green line). In contrast to the findings for days of earnings announcement concentration,
32
Figure 4: Response to Merger Announcements on Earnings Announcement Concentration
Days From -1 to 60 (Top Group)
Notes: The figure shows performance of mergers in the top group (positive immediate stock return response). Merger announcements
in Thomson One from January 1985 until December 2014 are matched to stock prices in CRSP. Saturdays and Sundays are not included
in the sample. Time 0 in event time is the day of the merger announcement. Cumulative abnormal returns (CAR) are calculated from
the market model. The vertical axis shows the mean CAR and the horizontal axis the days after the merger announcement. The blue
line shows the performance for mergers announced on days of a high number of earnings announcements and the turquoise line shows
the performance of mergers announced on other days.
the difference in CARs in the top group are much smaller. In the 20-day period after the
announcement, the CARs between companies announcing on days of major sports events
do not differ from the control group. Only after 30 and 60 days, announcements made on
days of major sports events show a better performance, but without a clear trend as for
earnings announcement concentration days. The regressions below will show whether the
difference is significant in a statistical sense.
Figure 7 shows the development over time in the bottom group. Here the difference
is quite visible, starting from a similar performance of both mergers announced on days
of major sports events and mergers announced on other days, the CARs drift apart after
around 10 days. It can be seen that the performance of mergers which have been announced
on a day of major sports event is worse than the performance of the control group. This
finding underlines the hypothesis of strategic disclosure timing.
33
Figure 5: Response to Merger Announcements on Earnings Announcement Concentration
Days From -1 to 60 (Bottom Group)
Notes: The figure shows performance of mergers in the bottom group (negative immediate stock return response). Merger announcements
in Thomson One from January 1985 until December 2014 are matched to stock prices in CRSP. Saturdays and Sundays are not included
in the sample. Time 0 in event time is the day of the merger announcement. Cumulative abnormal returns (CAR) are calculated from
the market model. The vertical axis shows the mean CAR and the horizontal axis the days after the merger announcement. The blue
line shows the performance for mergers announced on days of a high number of earnings announcements and the turquoise line shows
the performance of mergers announced on other days.
Regressions. The second step in examining strategic behavior looks at the long-term
stock response by means of an OLS regression.
The OLS specification for the long-term stock response is
R(−1,+30)t,k = α+ β1X + β2bottom.group+ β3 (X ∗ bottom.group) + controls+ εt,k
where R(−1,+30)t,k is the abnormal stock return for company k at time t between 1 day
before the merger announcement and 30 days after the merger announcement. The sample
includes all merger announcements. Coefficient β1 measures the return to mergers made
on an inattention day as described above. The coefficient β2 measures the return difference
due to classification into the bottom group and is not a variable of interest. β3 indicates
the difference in returns between mergers in the bottom group that are announced on a
day of inattention and mergers that are not announced on a day of inattention. Under
Hypothesis 2 of strategic behavior of managers, β3 should be negative for the long-term
response (R(−1,+30)t,k ) as mergers announced on days of inattention perform worse in the
34
Figure 6: Response to Merger Announcements on Days of Major Sports Events From -1 to
60 (Top Group)
Notes: The figure shows performance of mergers in the top group (positive immediate stock return response). Merger announcements
in Thomson One from January 1985 until December 2014 are matched to stock prices in CRSP. Saturdays and Sundays are not included
in the sample. Time 0 in event time is the day of the merger announcement. Cumulative abnormal returns (CAR) are calculated from
the market model. The vertical axis shows the mean CAR and the horizontal axis the days after the merger announcement. The blue
line shows the performance for mergers announced on days of a major sports event and the turquoise line shows the performance of
mergers announced on other days.
long run.
As stated above, and for a better analysis of whether managers announce mergers strate-
gically, I observe more closely the mergers that showed a negative immediate market re-
sponse. To analyze the difference to non-inattention days, I introduce an interaction variable
of the bottom group indicator and the inattention variable.
Table 6 shows the regression results for the long-term (30-day) stock return response. As
before, columns 1 and 2 show the regression results for the variable earnings announcement
concentration without and including control variables. Columns 3 and 4 show the same
regressions for the independent variable major sports event days. Finally, columns 5 and
6 show regression results of all independent variables. In line 4 and 5 we can see the
coefficients for our main variable of interest, the interaction variable of the bottom group
dummy and the inattention variable.
35
Figure 7: Response to Merger Announcements on Days of Major Sports Events From -1 to
60 (Bottom Group)
Notes: The figure shows performance of mergers in the bottom group (positive immediate stock return response). Merger announcements
in Thomson One from January 1985 until December 2014 are matched to stock prices in CRSP. Saturdays and Sundays are not included
in the sample. Time 0 in event time is the day of the merger announcement. Cumulative abnormal returns (CAR) are calculated from
the market model. The vertical axis shows the mean CAR and the horizontal axis the days after the merger announcement. The blue
line shows the performance for mergers announced on days of a major sports event and the turquoise line shows the performance of
mergers announced on other days.
Approximate location for Table 68
What can be observed for earnings announcement concentration days is that the bottom
group shows a significant 2.5% worse performance if mergers were announced on these
days. With the introduction of control variables, the significance vanishes. Nevertheless,
the credibility of the first regression is already quite high with an R-squared of 7.5% and
only increases slightly with the introduction of control variables to 9.4%.
A similar effect can be observed for mergers announced on a day of a major sports
event. Performance in the bottom group is 2.9% worse than in the control group and highly
significant. With the introduction of control variables, the significance persists and the
difference becomes even higher, namely 9.4%.
8Table 6 can be found on page 49
36
A regression with both variables and the interaction variables comes to a similar result
as shown in columns 5 and 6.
Summary. Both the visual inspection as well as the regression give evidence for strate-
gic disclosure timing. The graphs show a worse performance of mergers announced on
inattention days which even persists in the long-run without catching up to the control
group. The results can be verified by OLS regression, which shows a significant worse
stock performance in the bottom group if mergers are announced on inattention days. The
difference is -2.5% for earnings announcement concentration days and -2.9% for days of
major sport events. With the introduction of control variables, the significance is not given
anymore for announcements made on earnings announcement concentration days, but still
persists for mergers announced on days of major sports events. The difference for the latter
one is 9.4% between the variable of interest and the control group in the bottom group.
6 Robustness Checks
6.1 Post-announcement Drift and Convergence
Chapter 5.2 evidences strategic behavior of managers around their timing of mergers. In
order to prove a worse performance of mergers announced on inattention days, I divide the
full sample into a group of “good” mergers (i.e. mergers that showed a positive immediate
stock return response) and “bad” mergers (i.e. mergers that showed a negative immediate
stock return response. To better understand the inattention argument, I show how the
CARs of the whole sample (without classification) develop over time.
Figure 8 and 9 show the stock return reaction over time for announcements made on
high inattention days (2c for earnings announcement concentration and 3c for major sports
events) compared to other days over the whole sample.
Figure 8 shows a clear delayed response for announcements made on earnings announce-
ment concentration days which persists until 20 days after the merger announcement. In-
terestingly, the CARs drift apart again after around 60 days, which could be either due to
37
random variation or a general worse performance of mergers announced on these inattention
days (and the market needs time to discover this fact).
Figure 8: Response to Merger Announcements on Earnings Announcement Concentration
Days From -1 to 60 (Whole Sample)
Notes: The figure shows performance of mergers over the whole sample. Merger announcements in Thomson One from January 1985
until December 2014 are matched to stock prices in CRSP. Saturdays and Sundays are not included in the sample. Time 0 in event time
is the day of the merger announcement. Cumulative abnormal returns (CAR) are calculated from the market model. The vertical axis
shows the mean CAR and the horizontal axis the days after the merger announcement. The blue line shows the performance for mergers
announced on days of a high number of earnings announcements and the turquoise line shows the performance of mergers announced
on other days.
For merger announcements made on days of major sports events the difference in returns
is even more obvious as seen in figure 3c. A catching up effect is starting to develop after
around 30 days. Still, even after 60 days, the CARs for mergers announced on days of
major sports events have not reached the level of the control group but show a difference
of around 4%.
6.2 Extension of Observation Period
The core findings of table 6 in chapter 5.2 show the stock return response regression for the
event time -1 to +30 days around the merger announcement. Since merger announcements
are not as easy to process by the market and merger negotiations and the closing process
can take up much more time, it is necessary to see whether the findings persist over an even
38
Figure 9: Response to Merger Announcements on Earnings Announcement Concentration
Days From -1 to 60 (Whole Sample)
Notes: The figure shows performance of mergers over the whole sample. Merger announcements in Thomson One from January 1985
until December 2014 are matched to stock prices in CRSP. Saturdays and Sundays are not included in the sample. Time 0 in event
time is the day of the merger announcement. Cumulative abnormal returns (CAR) are calculated from the market model. The vertical
axis shows the mean CAR and the horizontal axis the days after the merger announcement. The blue line shows the performance for
mergers announced on days of a major sports event and the turquoise line shows the performance of mergers announced on other days.
longer time period.
Regression. The following regression will look at the long-term stock response to
merger announcement in the period of 60 days after the announcement of the merger.
The OLS specification for the long-term stock response is
R(−1,+60)t,k = α+ β1X + β2bottom.group+ β3 (X ∗ bottom.group) + controls+ εt,k
where R(−1,+60)t,k is the abnormal stock return for company k at time t between 1 day
before the merger announcement and 60 days after the merger announcement, instead of
the previous 30-day period. The sample includes all merger announcements. As before,
coefficient β1 measures the return to mergers made on an inattention day as described above.
The coefficient β2 measures the return difference due to classification into the bottom group
and is not a variable of interest. β3 indicates the difference in returns between mergers in
39
the bottom group that are announced on a day of inattention and mergers that are not
announced on a day of inattention. If the findings of strategic behavior of managers also
apply for the extended time period, β3 should be negative for the 60-day return response
(R(−1,+60)t,k ).
All specifications from before remain the same, mergers that showed a negative imme-
diate market response are observed more closely and an interaction variable of the bottom
group indicator and the inattention variable shows the difference which is of interest.
Regression results are presented in table 7. As before, columns 1 and 2 show the regres-
sion results for the variable earnings announcement concentration without and including
control variables. Columns 3 and 4 show the same regressions for the independent variable
major sports event days. Finally, columns 5 and 6 show regression results of all independent
variables. In line 4 and 5 we can see the coefficients for our main variable of interest, the
interaction variable of the bottom group dummy and the inattention variable.
Approximate location for Table 79
What can be observed for earnings announcement concentration days is that the bottom
group shows a significant 3.2% worse performance if mergers were announced on these days
which is even worse than the 2.5% difference in the 30-day period. With the introduction
of control variables, the significance vanishes, which is the case in the previous regressions
as well.
For mergers announced on a day of a major sports event, the performance in the bottom
group is 2.6% worse than in the control group and significant at the 5% level. This is slightly
better than the difference in the 30-day period of 2.9%. With the introduction of control
variables, the significance does not persist.
A regression with both variables and the interaction variables comes to a similar result
as shown in columns 5 and 6.
9Table 7 can be found on page 50
40
Summary. The OLS regression for the 60-day period shows similar results to the
previous findings. Merger announcements in the bottom group perform 2.6% to 3.2% worse
if announced on high inattention days even after 60 days. The regression results are only
significant if not controlled for company and deal specific characteristics.
6.3 Terciles and Quartiles
The results under chapter 5.2 are based on a separation of the merger announcement sample
into two groups of positive immediate stock returns and negative immediate stock returns.
To check the validity of these results I refine the separation into even finer groups. First,
I split the sample into terciles based on their immediate stock return response. The first
tercile includes the highest 1/3rd of immediate stock return responses, the second tercile
the middle 1/3rd and the last tercile the bottom 1/3rd. The focus here is on the bottom
tercile, in line with the hypothesis of strategic disclosure timing of worse mergers.
The OLS specification is
R(−1,+30)t,k = α+ β1X + β2bottom.tercile+ β3 (X ∗ bottom.tercile) + controls+ εt,k
where R(−1,+30)t,k is the abnormal stock return for company k at time t between 1 day
before the merger announcement and 30 days after the merger announcement. As before,
coefficient β1 measures the return to mergers made on an inattention day as described
above. The coefficient β2 measures the return difference due to classification into the bottom
tercile (i.e. the lowest 1/3rd of immediate stock return responses) and is not a variable of
interest. β3 indicates the difference in returns between mergers in the bottom tercile that
are announced on a day of inattention and mergers that are not announced on a day of
inattention. I expect a negative coefficient to prove the findings under chapter 5.1.2. The
sample does not contain values in the middle tercile but only the top tercile and bottom
tercile.
Regression results are presented in table 8. As before, columns 1 and 2 show the regres-
sion results for the variable earnings announcement concentration without and including
41
control variables. Columns 3 and 4 show the same regressions for the independent variable
major sports event days. Finally, columns 5 and 6 show regression results of all independent
variables. In line 4 and 5 we can see the coefficients for our main variable of interest, the
interaction variable of the bottom tercile dummy and the inattention variable.
Approximate location for Table 810
What can be observed for earnings announcement concentration days is that the bottom
group shows a significant 2.9% worse performance if mergers were announced on these days
which is slightly higher than the 2.5% difference observed earlier. With the introduction of
control variables, the significance vanishes, which had been the case earlier as well.
For mergers announced on a day of a major sports event the performance in the bottom
group is 2.7% worse than in the control group and significant at the 5% level. This is slightly
better than the difference in the 30-day period of 2.9%. With the introduction of control
variables, the significance does persist at the 5% level.
A regression with both variables and the interaction variables comes to a similar result
as shown in columns 5 and 6.
Secondly, I split the sample into quartiles based on their immediate stock return re-
sponse. The first quartile includes the highest 1/4th of immediate stock return responses,
the second quartile the second highest 1/4th and so on. The focus here is on the bottom
quartile, in line with the hypothesis of strategic disclosure timing of worse mergers.
The OLS specification is
R(−1,+30)t,k = α+ β1X + β2bottom.quartile+ β3 (X ∗ bottom.quartile) + controls+ εt,k
where R(−1,+30)t,k is the abnormal stock return for company k at time t between 1 day
before the merger announcement and 30 days after the merger announcement. As before,
coefficient β1 measures the return to mergers made on an inattention day as described
10Table 8 can be found on page 51
42
above. The coefficient β2 measures the return difference due to classification into the bottom
quartile (i.e. the lowest 1/4th of immediate stock return responses) and is not a variable
of interest. β3 indicates the difference in returns between mergers in the bottom quartile
that are announced on a day of inattention and mergers that are not announced on a day
of inattention. I expect a negative coefficient β3 to prove the findings under chapter 5.1.2.
The sample does not contain values in the middle quartiles (second and third highest) but
only the top quartile and bottom quartile.
Regression results are presented in table 9. As before, columns 1 and 2 show the regres-
sion results for the variable earnings announcement concentration without and including
control variables. Columns 3 and 4 show the same regressions for the independent variable
major sports event days. Finally, columns 5 and 6 show regression results of all independent
variables. In line 4 and 5 we can see the coefficients for our main variable of interest, the
interaction variable of the bottom quartile dummy and the inattention variable.
Approximate location for Table 911
What can be observed for earnings announcement concentration days is that the bottom
group shows a significant 2.7% worse performance if mergers were announced on these days
which is slightly higher than the 2.5% difference observed earlier. With the introduction of
control variables, the significance vanishes.
For mergers announced on a day of a major sports event the performance in the bottom
group is 2.8% worse than in the control group and significant at the 10% level. This is
slightly lower than the difference in the 30-day period of 2.9%. With the introduction of
control variables, the significance jumps to the 1% level for a difference of 12.3%.
A regression with both variables and the interaction variables comes to a similar result
as shown in columns 5 and 6.
In summary, the refinement of the sample in terciles and quartiles shows the same results
as stated under chapter 5.2.
11Table 9 can be found on page 52
43
7 Conclusion and Discussion
At the outset, I positioned this thesis among related literature regarding strategic disclosure
timing of managers in response to investor inattentiveness and proposed two main hypothe-
ses: (1) investors are less attentive to merger announcements if they fall on days of high
number of earnings announcements or days of major sports events and (2) managers exploit
this inattention by announcing less profitable mergers on these days.
How well could these hypotheses be confirmed by empirical analysis? With respect to
the first hypothesis I find evidence for inattention on the proposed inattention days by
means of a lower abnormal trading volume observable through a plotting of mean abnormal
trading volumes around the day of the merger announcement. These findings can not be
confirmed by OLS regression due to a lack of significance. Moreover, the correlation of a
number of earnings announcements and trading volume makes a differentiation of trading
volume due to the announcement of a merger difficult. A similar picture is drawn by the
analysis of immediate stock return responses. Even though an expected negative effect of
inattention days on immediate stock returns can be found, the results do not hold if control
variables are introduced and are of low economic relevance.
Regarding the second hypothesis my findings are more robust. I show that a separation
of merger announcements into a top group (i.e. mergers that showed an immediate positive
stock return response) and bottom group (i.e. mergers that showed an immediate negative
stock return response can prove the hypothesis of managers announcing less profitable
mergers on days of high inattention. Mergers in the bottom group showed 2.5% lower 30-
day CARs if the announcement was made on a day of earnings announcement concentration
and 2.9% lower CARs if the announcement was made on a day of a major sports event. The
difference in long-term performance is also visible by plotting the performance of mergers
announced on inattention days and mergers announced on normal days.
As expected, the results are in line with previous literature on strategic disclosure timing
but have to be interpreted with caution. Niessner (2015) for example shows that managers
strategically announce bad news on high inattention days. My results show that this is
44
also the case for bad mergers, which are announced preferably on high inattention days. In
spite of that, the inattention variables as proposed by Niessner (2015), namely inattention
on Fridays and days before national holidays, do not lead to the same results for merger
announcements. I find no significant difference in performance of mergers announced on
Fridays or before national holidays and other days (shown in the Appendix). Other than
that, the findings on inattention around earnings announcement concentration and days of
major sports events can support the results of Manconi, Kempf and Spalt (2015) who find
that firms are more likely to announce bad mergers (i.e. diversifying mergers) when their
shareholders are distracted. I can even extend their claim that these mergers are “bad” by
analyzing cumulative abnormal returns in response to the merger announcement.
As mentioned before, the research results have to be considered with caution. Especially
the claim of inattention on days of high number of earnings announcements and days of
major sport events can not be proven with certainty. The predominant methodology for
showing inattention is to observe pre-announcement drift as used for example by DellaVigna
and Pollet (2009) and Louis and Sun (2010). My analysis does not document a clear
announcement drift and catching up effect in response to inattention. The ongoing worse
performance of mergers announced on the observed inattention days is in my opinion owed to
strategic disclosure timing as described earlier. This finding proves to be statistically robust
also in the long run (60-day period) and after a refinement of the grouping into terciles and
quartiles. I propose that there are two possible interpretations for those findings. First,
the unclear evidence of inattention might be simply owed to the fact that inattention only
gets discovered over time. In other words, the immediate trading volume and stock return
responses do not show a significant difference over announcement days because the true
nature of the merger has not yet been fully processed by the market. Only after some time
and as more information gets discovered, the difference in profitability becomes obvious. On
the other hand, it is also possible that there is no real inattention on days of high number of
earnings announcements and days of major sports events, but managers intuitively assume
there is. Consequently, they prefer to announce their less profitable mergers on these days
45
and that is why a performance difference becomes visible.
In summary, I see the biggest drawback in the low economic relevance. Looking at
the variable construction and resulting descriptive statistics it can be seen that only 3%
of all announcements happened on earnings announcement concentration days and 7% of
announcements on days of major sports events. Even though there is clear evidence for
strategic disclosure timing on these days, there is a certain limitation to which extend this
inattention is exploitable. Usually, mergers are not planned that far ahead, nor is it possible
to keep a major announcement secret from the market for too long a time period. As a
consequence, announcements of worse mergers might be preferably announced on these
inattention days if the decision falls into a period around one of these days, but in general
managers do not usually have that flexibility.
The main implications of the findings in this thesis are that investors should raise aware-
ness to merger announcements on days of earnings announcement concentration and days of
major sports events. The findings can also be used for an arbitrage strategy by means of a
long-short portfolio. Still, since mergers are much more complex and are heavily dependent
on other various factors besides the day of the announcement, this arbitrage strategy might
not prove as successful as other strategies that exploit market imperfections.
Promising avenues for future research include the analysis of management incentives for
strategic disclosure timing. There is also the question of whether the findings are indicative
of a certain management style.
46
Table 4: Abnormal Trading Volume Regression
Notes: *significant at 10%; **significant at 5%; ***significant at 1%. Merger announcements in Thomson One from January 1985 until
December 2014 are matched to stock prices in CRSP. Saturdays and Sundays are not included in the sample. Time 0 in event time
is the day of the merger announcement. The abnormal trading volume is the average logarithmic volume on the day of the merger
announcement and the following day, minus the average log volume for the period -20 to -11. “Announcement Concentration” is a dummy
variable indicating whether the number of quarterly earnings announcements on a single day is in the 90th percentile of all daily quarterly
earnings announcements (data for earnings announcements is taken from I/B/E/S). “Major Sports Event Dummy” is a dummy variable
indicating whether the announcement falls on a day of a major sports event (NBA finals, MLB finals, Super Bowl). “EPS Target” are the
earnings per share of the target (sell-side) firm. “EPS Acquirer” are the earnings per share of the acquirer (buy-side) firm, respectively.
“Log (Pre-announc. MV) Acquirer” is the natural logarithm of the acquirer’s market value, defined as share price times number of shares
outstanding in million US$, 4 weeks prior to the announcement. “Deal Value to Acquirer Size” is the ratio of the deal (transaction)
value to the acquirer size (“Pre-announc. MV Acquirer” as defined above, not logarithmized). “Log (MV/BV) Acquirer” is the natural
logarithm of the ratio of the market value of assets (“Pre-announc. MV Acquirer”) to the book value of assets of the acquirer. “Acquirer
Cash Holdings” is the ratio of cash assets to net total asset value of the acquirer. “Number of Bidders” is the number of competing
bidders for the target company. “Dummy for completed Mergers” is a dummy variable that takes the value 1 for completed mergers and
0 for any other status. “Dummy for diversif. Mergers” is a dummy variable which takes the value 1 for mergers over different macro
industries (according to the Thomson Financial Macro Hierarchy) or 0 otherwise. “Analyst Coverage” is the number of analysts who
publish quarterly earnings estimates of the acquiring company in the year of announcement. Data for this variable is taken from I/B/E/S.
47
Table 5: Immediate Stock Return Response Regression
Notes: *significant at 10%; **significant at 5%; ***significant at 1%. Merger announcements in Thomson One from January 1985 until
December 2014 are matched to stock prices in CRSP. Saturdays and Sundays are not included in the sample. Time 0 in event time
is the day of the merger announcement. Immediate stock return response is the cumulative abnormal return in the period -1 to +1
around the announcement, calculated from the market model. “Announcement Concentration” is a dummy variable indicating whether
the number of quarterly earnings announcements on a single day is in the 90th percentile of all daily quarterly earnings announcements
(data for earnings announcements is taken from I/B/E/S). “Major Sports Event Dummy” is a dummy variable indicating whether the
announcement falls on a day of a major sports event (NBA finals, MLB finals, Super Bowl). “EPS Target” are the earnings per share of
the target (sell-side) firm. “EPS Acquirer” are the earnings per share of the acquirer (buy-side) firm, respectively. “Log (Pre-announc.
MV) Acquirer” is the natural logarithm of the acquirer’s market value, defined as share price times number of shares outstanding in
million US$, 4 weeks prior to the announcement. “Deal Value to Acquirer Size” is the ratio of the deal (transaction) value to the
acquirer size (“Pre-announc. MV Acquirer” as defined above, not logarithmized). “Log (MV/BV) Acquirer” is the natural logarithm
of the ratio of the market value of assets (“Pre-announc. MV Acquirer”) to the book value of assets of the acquirer. ’Acquirer Cash
Holdings’ is the ratio of cash assets to net total asset value of the acquirer. “Number of Bidders” is the number of competing bidders for
the target company. “Dummy for completed Mergers” is a dummy variable that takes the value 1 for completed mergers and 0 for any
other status. “Dummy for diversif. Mergers” is a dummy variable which takes the value 1 for mergers over different macro industries
(according to the Thomson Financial Macro Hierarchy) or 0 otherwise. “Analyst Coverage” is the number of analysts who publish
quarterly earnings estimates of the acquiring company in the year of announcement. Data for this variable is taken from I/B/E/S.
48
Table 6: Long-term Stock Return Response Regression
Notes: *significant at 10%; **significant at 5%; ***significant at 1%. Merger announcements in Thomson One from January 1985 until
December 2014 are matched to stock prices in CRSP. Saturdays and Sundays are not included the sample. Time 0 in event time is the
day of the merger announcement. Long-term stock return response is the cumulative abnormal return in the period -1 to +30 around
the announcement, calculated from the market model. “Announcement Concentration” is a dummy variable indicating whether the
number of quarterly earnings announcements on a single day is in the 90th percentile of all daily quarterly earnings announcements
(data for earnings announcements is taken from I/B/E/S). “Major Sports Event Dummy” is a dummy variable indicating whether the
announcement falls on a day of a major sports event (NBA finals, MLB finals, Super Bowl). “EPS Target” are the earnings per share of
the target (sell-side) firm. “EPS Acquirer” are the earnings per share of the acquirer (buy-side) firm, respectively. “Log (Pre-announc.
MV) Acquirer” is the natural logarithm of the acquirer’s market value, defined as share price times number of shares outstanding in
million US$, 4 weeks prior to the announcement. “Deal Value to Acquirer Size” is the ratio of the deal (transaction) value to the
acquirer size (“Pre-announc. MV Acquirer” as defined above, not logarithmized). “Log (MV/BV) Acquirer” is the natural logarithm
of the ratio of the market value of assets (“Pre-announc. MV Acquirer”) to the book value of assets of the acquirer. “Acquirer Cash
Holdings” is the ratio of cash assets to net total asset value of the acquirer. “Number of Bidders” is the number of competing bidders for
the target company. “Dummy for completed Mergers” is a dummy variable that takes the value 1 for completed mergers and 0 for any
other status. “Dummy for diversif. Mergers” is a dummy variable which takes the value 1 for mergers over different macro industries
(according to the Thomson Financial Macro Hierarchy) or 0 otherwise. “Analyst Coverage” is the number of analysts who publish
quarterly earnings estimates of the acquiring company in the year of announcement. Data for this variable is taken from I/B/E/S.
49
Table 7: Long-term (60-day) Stock Return Response Regression
Notes: *significant at 10%; **significant at 5%; ***significant at 1%. Merger announcements in Thomson One from January 1985 until
December 2014 are matched to stock prices in CRSP. Saturdays and Sundays are dropped from the sample. Time 0 in event time is the
day of the merger announcement. Long-term stock return response is the cumulative abnormal return in the period -1 to +60 around
the announcement, calculated from the market model. “Announcement Concentration” is a dummy variable indicating whether the
number of quarterly earnings announcements on a single day is in the 90th percentile of all daily quarterly earnings announcements
(data for earnings announcements is taken from I/B/E/S). “Major Sports Event Dummy” is a dummy variable indicating whether the
announcement falls on a day of a major sport event (NBA finals, MLB finals, Super Bowl). “Bottom Group” is a dummy variable
indicating whether the immediate stock return response (3-day window) was below zero. “(Bottom Group)*(Announc. Conc.)” and
“(Bottom Group)*(Sports Event)” are interaction variables. “EPS Target” are the earnings per share of the target (sell-side) firm.
“EPS Acquirer” are the earnings per share of the acquirer (buy-side) firm, respectively. “Log (Pre-announc. MV) Acquirer” is the
natural logarithm of the acquirer’s market value, defined as share price times number of shares outstanding in million US$, 4 weeks
prior to the announcement. “Deal Value to Acquirer Size” is the ratio of the deal (transaction) value to the acquirer size (“Pre-
announc. MV Acquirer” as defined above, not logarithmized). “Log (MV/BV) Acquirer” is the natural logarithm of the ratio of
the market value of assets (“Pre-announc. MV Acquirer”) to the book value of assets of the acquirer. “Acquirer Cash Holdings” is
the ratio of cash assets to net total asset value of the acquirer. “Number of Bidders” is the number of competing bidders for the
target company. “Dummy for completed Mergers” is a dummy variable that takes the value 1 for completed mergers and 0 for any
other status. “Dummy for diversif. Mergers” is a dummy variable which takes the value 1 for mergers over different macro industries
(according to the Thomson Financial Macro Hierarchy) or 0 otherwise. “Analyst Coverage” is the number of analysts who publish
quarterly earnings estimates of the acquiring company in the year of announcement. Data for this variable is taken from I/B/E/S.
50
Table 8: Long-term Stock Return Response Regression Terciles
Notes: *significant at 10%; **significant at 5%; ***significant at 1%. Merger announcements in Thomson One from January 1985
until December 2014 are matched to stock prices in CRSP. Saturdays and Sundays are not included in the sample. Time 0 in event
time is the day of the merger announcement. Long-term stock return response is the cumulative abnormal return in the period
-1 to +30 around the announcement, calculated from the market model. ’Announcement Concentration’ is a dummy variable in-
dicating whether the number of quarterly earnings announcements on a single day is in the 90th percentile of all daily quarterly
earnings announcements (data for earnings announcements is taken from I/B/E/S). “Major Sports Event Dummy” is a dummy
variable indicating whether the announcement falls on a day of a major sport event (NBA finals, MLB finals, Super Bowl). “Bot-
tom Tercile” is a dummy variable indicating whether the immediate stock return response (3-day window) was in the lower 1/3rd
bracket. The second tercile (middle tercile) is not included in the regression. “(Bottom Tercile)*(Announc. Conc.)” and “(Bot-
tom Tercile)*(Sports Event)” are interaction variables. Control variables are standard controls, as used in all previous regressions.
51
Table 9: Long-term Stock Return Response Regression Quartiles
Notes: *significant at 10%; **significant at 5%; ***significant at 1%. Merger announcements in Thomson One from January 1985
until December 2014 are matched to stock prices in CRSP. Saturdays and Sundays are not included in the sample. Time 0 in event
time is the day of the merger announcement. Long-term stock return response is the cumulative abnormal return in the period -1
to +30 around the announcement, calculated from the market model. “Announcement Concentration” is a dummy variable indicat-
ing whether the number of quarterly earnings announcements on a single day is in the 90th percentile of all daily quarterly earnings
announcements (data for earnings announcements is taken from I/B/E/S). “Major Sports Event Dummy” is a dummy variable indi-
cating whether the announcement falls on a day of a major sports event (NBA finals, MLB finals, Super Bowl). “Bottom Quartile”
is a dummy variable indicating whether the immediate stock return response (3-day window’ was in the lower 1/4th bracket. The
second and third quartile (middle quartiles) are not included in the regression. “(Bottom Quartile)*(Announc. Conc.)” and “(Bot-
tom Quartile)*(Sports Event)” are interaction variables. Control variables are standard controls, as used in all previous regressions.
52
References
Acharya, V., DeMarzo, P. and Kremer, I. (2011). Endogenous Information Flows and the
Clustering of Announcements. American Economic Review, 101(7), pp.2955-2979.
Amihud, Y. and Lev, B. (1981). Risk Reduction as a Managerial Motive for Conglomerate
Mergers. The Bell Journal of Economics, 12(2), p.605.
Baker, M. and Wurgler, J. (n.d.). Behavioral Corporate Finance: An Updated Survey.
SSRN Electronic Journal.
Barberis, N. and Thaler, R. (n.d.). A Survey of Behavioral Finance. SSRN Electronic
Journal.
Cohen, L. and Lou, D. (2012). Complicated firms. Journal of Financial Economics, 104(2),
pp.383-400.
Da, Z., Engelberg, J. and Gao, P. (n.d.). In Search of Attention. SSRN Electronic Journal.
DellaVigna, S. and Pollet, J. (2009). Investor Inattention and Friday Earnings Announce-
ments. The Journal of Finance, 64(2), pp.709-749.
DellaVigna, S. and Pollet, J. (2007). Demographics and Industry Returns. American
Economic Review, 97(5), pp.1667-1702.
Doyle, J. and Magilke, M. (2009). The Timing of Earnings Announcements: An Examina-
tion of the Strategic Disclosure Hypothesis. The Accounting Review, 84(1), pp.157-182.
Driesprong, G., Jacobsen, B. and Maat, B. (2008). Striking oil: Another puzzle?. Journal
of Financial Economics, 89(2), pp.307-327.
Ehrmann, M. and Jansen, D. (n.d.). The Pitch Rather than the Pit: Investor Inattention
During FIFA World Cup Matches. SSRN Electronic Journal.
Fama, E. (1965). Random Walks in Stock Market Prices. Financial Analysts Journal,
21(5), pp.55-59.
53
Federalreserve.gov. (2016). FRB: K.8 - Holidays Observed by the Federal Reserve, 2014-
2018. [online] Available at: https://www.federalreserve.gov/aboutthefed/k8.htm [Accessed
6 Apr. 2016].
Frazzini, A. and Lamont, O. (2007), The Earnings Announcement Premium and Trading
Volume. National Bureau of Economic Research.
Haas.berkeley.edu. (2016). Initiative for Behavorial Economics & Finance. [online] Avail-
able at: http://haas.berkeley.edu/behavioral/articles/the friday effect.html [Accessed 15
Apr. 2016].
Harford, J. (1999). Corporate Cash Reserves and Acquisitions. The Journal of Finance,
54(6), pp.1969-1997.
Harvard Business Review. (1999). Stock or Cash?:The Trade-Offs for Buyers and Sellers
in Mergers and Acquisitions. [online] Available at: https://hbr.org/1999/11/stock-or-cash-
the-trade-offs-for-buyers-and-sellers-in-mergers-and-acquisitions [Accessed 10 May 2016].
Hirshleifer, D., Lim, S. and Teoh, S. (2009). Driven to Distraction: Extraneous Events and
Underreaction to Earnings News. The Journal of Finance, 64(5), pp.2289-2325.
Hong, H. and Stein, J. (1999). A Unified Theory of Underreaction, Momentum Trading,
and Overreaction in Asset Markets. The Journal of Finance, 54(6), pp.2143-2184.
Hong, H., Torous, W. and Valkanov, R. (2007). Do industries lead stock markets?. The
Journal of Financial Economics, 83(2), pp.367-396.
Huberman, G. and Regev, T. (2001). Contagious Speculation and a Cure for Cancer: A
Nonevent that Made Stock Prices Soar. The Journal of Finance, 56(1), pp.387-396.
Kahneman, D. and Tversky, A. (1979). Prospect Theory: An Analysis of Decision under
Risk. Econometrica, 47(2), p.263.
Kempf, E., Manconi, A. and Spalt, O. (n.d.). Distracted Shareholders and Corporate
Actions. SSRN Electronic Journal.
54
Loasby, B., McGuire, C. and Radner, R. (1972). Decision and Organisation: A Volume in
Honor of Jacob Marschak. The Economic Journal, 82(328), p.1414.
Loh, R. (2010). Investor Inattention and the Underreaction to Stock Recommendations.
Financial Management, 39(3), pp.1223-1252.
Louis, H. and Sun, A. (2010). Investor Inattention and the Market Reaction to Merger
Announcements. Management Science, 56(10), pp.1781-1793.
Lu, X. and White, H. (2014). Robustness checks and robustness tests in applied economics.
Journal of Econometrics, 178, pp.194-206.
Malkiel, B. (2003). The Efficient Market Hypothesis and Its Critics. Journal of Economic
Perspectives, 17(1), pp.59-82.
Malmendier, U. and Tate, G. (2005). CEO Overconfidence and Corporate Investment. The
Journal of Finance, 60(6), pp.2661-2700.
Mehra, R. and E. C. Prescott (1985), The equity premium: A puzzle. Journal of Monetary
Economics 15, 145161.
Niessner, M. (n.d.). Strategic Disclosure Timing and Insider Trading. SSRN Electronic
Journal.
Patell, J. and Wolfson, M. (1984). The intraday speed of adjustment of stock prices to
earnings and dividend announcements. Journal of Financial Economics, 13(2), pp.223-252.
Pantzalis, C. and Ucar, E. (2014). Religious holidays, investor distraction, and earnings
announcement effects. Journal of Banking & Finance, 47, pp.102-117.
Roll, R. (1986). The Hubris Hypothesis of Corporate Takeovers. The Journal of Business,
59(2), p.197.
Santos, F. (2009), Investors Attention to Earnings Announcements During Major Sport
Events in the U.S.. http://www.gsb.stanford.edu/sites/default/files/documents/InvestorsAttentiontoEarningsAnnouncementsDuringMajorSportEventsintheUS.pdf
55
Shiller, R. (2003). From Efficient Markets Theory to Behavioral Finance. Journal of Eco-
nomic Perspectives, 17(1), pp.83-104.
Simon, H. (1955). A Behavioral Model of Rational Choice. The Quarterly Journal of
Economics, 69(1), p.99.
Shleifer, A. and Vishny, R. (2003). Stock market driven acquisitions. Journal of Financial
Economics, 70(3), pp.295-311.
Solomon, D. (2012). Selective Publicity and Stock Prices. The The Journal of Finance,
67(2), pp.599-638.
Stanton, R. (2012). Writing Finance Papers using LaTeX. [online] Faculty.haas.berkeley.edu.
Available at: http://faculty.haas.berkeley.edu/stanton/texintro/ [Accessed 25 Jun. 2016].
Stock, J. and Watson, M. (2015). Introduction to econometrics. Boston: Pearson/Addison
Wesley.
Van Nieuwerburgh, S. and Veldkamp, L. (2009). Information Immobility and the Home
Bias Puzzle. The Journal of Finance, 64(3), pp.1187-1215.
56
Appendix
Previous literature documents inattention around Fridays and before national holidays. The
following part will examine this proposition around merger announcements. In general, I
use the same methodology and hypotheses as earlier. The previous inattention variables
are now changed for Fridays and days before national holidays.
Variable Construction. The dataset is the same as constructed in Chapter 4. The
new variable “Friday” is a dummy variable taking the value 1 for merger announcements
on Fridays and 0 for merger announcements on other days12. The second variable “Days
before National Holidays” is a dummy variable taking the value 1 for merger announcements
on days before national holidays and 0 for merger announcements on other days. Since in
the U.S. public holidays with a corresponding day off vary over different states, I follow
the holidays observed by the Federal Reserve System. This is due to the reliance of banks
and financial institutions on the U.S. Federal Reserve and usually follow their holidays.
The holidays considered in this analysis are New Year’s Day, Birthday of Martin Luther
KingJr., Washington’s Birthday, Memorial Day, Independence Day, Labor Day, Columbus
Day, Veterans Day, Thanksgiving and Christmas. The general rule for holidays falling on
Saturday is that the day off is moved to the preceding Friday and for holidays falling on a
Sunday, the day off is moved to the next Monday. Consequently, the day before a national
holiday falling on a Saturday is Thursday and the day before a national holiday that falls
on a Saturday is Friday.
Volume Regression. Table 10 shows the regression results for the OLS specification
∆v(0,1)t,k = α+ β1X + controls
where ∆v(0,1)t,k is the abnormal trading volume for firm k on the day of the merger
announcement and the subsequent day. α is the constant in the regression and β1 is the
coefficient for the inattention variable, which is either a dummy variable for Fridays or a
dummy variable for days before national holidays.
12Note that announcements made on Saturdays and Sundays are not considered in this analysis
57
The coefficient β1 shows the percentage difference in abnormal trading volume between
merger announcements made on inattention days and announcements made on normal days.
Column 1 in table 10 shows a highly significant negative coefficient for Friday merger
announcements of -0.177. This means that on average, abnormal trading volume in acquir-
ing firms is 17.7% lower if mergers are announced on Fridays. Controlling for firm and deal
specific characteristics, the difference is even higher with 30% lower abnormal trading vol-
ume. This is in line with the findings of DellaVigna and Pollet (2009), Louis and Sun (2010)
and Niessner (2015). The findings for days before national holidays are similar. Acquiring
firms announcing mergers on these days observe a 50.5% lower abnormal trading volume,
or 47.6% lower if controlled for firm and deal specific characteristics. Both regressions show
a high credibility with an adjusted R-squared of around 22% including control variables.
Approximate location for Table 1013
Immediate Stock Return Response Regression. Table 11 shows the regression
results for the OLS specification
R(d,D)t,k = α+ β1X + controls+ εt,k
where R(d,D)t,k is the abnormal stock return for company k at time t between 1 day before
the merger announcement and 1 day after the merger announcement. The sample includes
all merger announcements. α is the constant in the regression and β1 is the coefficient for the
inattention variable X which is either a dummy variable for Fridays or a dummy variable for
days before national holidays. As stated earlier, under Hypothesis 1 of inattentive markets,
β1 should be significantly negative in the short-run.
As seen in column 1 in table 11, this is not the case. The regression coefficient for
the Friday dummy is not significantly different from 0. The same applies for days before
national holidays.
13Table 10 can be found on page 61
58
Approximate location for Table 1114
Long-term Stock Response Regression. I try to find strategic management behav-
ior by looking at the long-term stock response by means of an OLS regression. The focus
lies on the bottom group of merger announcements. For a deeper explanation please refer
to Chapter 3 Methodology.
The OLS specification for the long-term stock response is
R(−1,+30)t,k = α+ β1X + β2bottom.group+ β3 (X ∗ bottom.group) + controls+ εt,k
where R(−1,+30)t,k is the abnormal stock return for company k at time t between 1 day
before the merger announcement and 30 days after the merger announcement. The sample
includes all merger announcements. Coefficient β1 measures the return to mergers made
on an inattention day as described above. The coefficient β2 measures the return difference
due to classification into the bottom group and is not a variable of interest. β3 indicates the
difference in returns between mergers in the bottom group that are announced on a day of
inattention and mergers that are not announced on a day of inattention. Under Hypothesis 2
of strategic behavior of managers, should be negative for the long-term response (R(−1,+30)t,k )
as mergers announced on days of inattention perform worse in the long run.
Regression results are presented in table 12. For Friday merger announcements, the
performance difference is not significantly different from merger announcements on other
days. Surprisingly, mergers in the bottom group announced on days before national holidays
show a significantly better performance of 9.4% higher cumulative abnormal returns over
the 30-day period following the announcement.
Approximate location for Table 1215
14Table 11 can be found on page 6215Table 12 can be found on page 63
59
Summary. The regression on abnormal volume showed clear evidence for market inat-
tention on Fridays and days before national holidays with significant lower abnormal trading
volumes for announcements on these days. In contrast, abnormal returns turned out to be
not significantly different between the proposed inattention days and normal days in the
3-day window around the announcement. Moreover, the long-term performance in the
bottom group also showed no significant differences, except for announcements on days
before national holidays, which actually showed up to perform significantly better than an-
nouncements made on normal days. My conclusion is that there is no evidence for strategic
disclosure timing of mergers on Fridays or days before national holidays.
60
Table 10: Abnormal Trading Volume Regression Additional Variables
Notes: *significant at 10%; **significant at 5%; ***significant at 1%. Merger announcements in Thomson One from January 1985 until
December 2014 are matched to stock prices in CRSP. Saturdays and Sundays are not included in the sample. Time 0 in event time
is the day of the merger announcement. The abnormal trading volume is the average logarithmic volume on the day of the merger
announcement and the following day, minus the average log volume for the period -20 to -11. “Friday” is a dummy variable indicating
whether the announcement is made on a Friday. “Before Holiday” is a dummy variable indicating whether the announcement is made
on a day before a national holiday (following the Federal Reserve calendar). “EPS Target” are the earnings per share of the target
(sell-side) firm. “EPS Acquirer” are the earnings per share of the acquirer (buy-side) firm, respectively. “Log (Pre-announc. MV)
Acquirer” is the natural logarithm of the acquirer’s market value, defined as share price times number of shares outstanding in million
US$, 4 weeks prior to the announcement. “Deal Value to Acquirer Size” is the ratio of the deal (transaction) value to the acquirer size
(“Pre-announc. MV Acquirer” as defined above, not logarithmized). “Log (MV/BV) Acquirer” is the natural logarithm of the ratio
of the market value of assets (“Pre-announc. MV Acquirer”) to the book value of assets of the acquirer. “Acquirer Cash Holdings”
is the ratio of cash assets to net total asset value of the acquirer. “Number of Bidders” is the number of competing bidders for the
target company. “Dummy for completed Mergers” is a dummy variable that takes the value 1 for completed mergers and 0 for any
other status. “Dummy for diversif. Mergers” is a dummy variable which takes the value 1 for mergers over different macro industries
(according to the Thomson Financial Macro Hierarchy) or 0 otherwise. “Analyst Coverage” is the number of analysts who publish
quarterly earnings estimates of the acquiring company in the year of announcement. Data for this variable is taken from I/B/E/S.
61
Table 11: Immediate Stock Return Response Regression Additional Variables
Notes: *significant at 10%; **significant at 5%; ***significant at 1%. Merger announcements in Thomson One from January 1985
until December 2014 are matched to stock prices in CRSP. Saturdays and Sundays are not included in the sample. Time 0 in event
time is the day of the merger announcement. Immediate stock return response is the cumulative abnormal return in the period -1
to +1 around the announcement, calculated from the market model. “Friday” is a dummy variable indicating whether the announce-
ment is made on a Friday. “Before Holiday” is a dummy variable indicating whether the announcement is made on a day before
a national holiday (following the Federal Reserve calendar). “EPS Target” are the earnings per share of the target (sell-side) firm.
“EPS Acquirer” are the earnings per share of the acquirer (buy-side) firm, respectively. “Log (Pre-announc. MV) Acquirer” is the
natural logarithm of the acquirer’s market value, defined as share price times number of shares outstanding in million US$, 4 weeks
prior to the announcement. “Deal Value to Acquirer Size” is the ratio of the deal (transaction) value to the acquirer size (“Pre-
announc. MV Acquirer” as defined above, not logarithmized). “Log (MV/BV) Acquirer” is the natural logarithm of the ratio of
the market value of assets (“Pre-announc. MV Acquirer”) to the book value of assets of the acquirer. “Acquirer Cash Holdings” is
the ratio of cash assets to net total asset value of the acquirer. “Number of Bidders” is the number of competing bidders for the
target company. “Dummy for completed Mergers” is a dummy variable that takes the value 1 for completed mergers and 0 for any
other status. “Dummy for diversif. Mergers” is a dummy variable which takes the value 1 for mergers over different macro industries
(according to the Thomson Financial Macro Hierarchy) or 0 otherwise. “Analyst Coverage” is the number of analysts who publish
quarterly earnings estimates of the acquiring company in the year of announcement. Data for this variable is taken from I/B/E/S.
62
Table 12: Long-term Stock Return Response Regression Additional Variables
Notes: *significant at 10%; **significant at 5%; ***significant at 1%. Merger announcements in Thomson One from January 1985 until
December 2014 are matched to stock prices in CRSP. Saturdays and Sundays are not included in the sample. Time 0 in event time is the
day of the merger announcement. Long-term stock return response is the cumulative abnormal return in the period -1 to +30 around
the announcement, calculated from the market model. “Friday” is a dummy variable indicating whether the announcement is made on a
Friday. “Before Holiday” is a dummy variable indicating whether the announcement is made on a day before a national holiday (follow-
ing the Federal Reserve calendar). “Bottom Group” is a dummy variable indicating whether the immediate stock return response (3-day
window) was below zero. “(Bottom Group)*(Friday)” and “(Bottom Group)*(Before Holiday)” are interaction variables. “EPS Target”
are the earnings per share of the target (sell-side) firm. “EPS Acquirer” are the earnings per share of the acquirer (buy-side) firm, respec-
tively. “Log (Pre-announc. MV) Acquirer” is the natural logarithm of the acquirer’s market value, defined as share price times number of
shares outstanding in million US$, 4 weeks prior to the announcement. “Deal Value to Acquirer Size” is the ratio of the deal (transaction)
value to the acquirer size (“Pre-announc. MV Acquirer” as defined above, not logarithmized). “Log (MV/BV) Acquirer” is the natural
logarithm of the ratio of the market value of assets (“Pre-announc. MV Acquirer”) to the book value of assets of the acquirer. “Acquirer
Cash Holdings” is the ratio of cash assets to net total asset value of the acquirer. “Number of Bidders” is the number of competing bid-
ders for the target company. “Dummy for completed Mergers” is a dummy variable that takes the value 1 for completed mergers and 0
for any other status. “Dummy for diversif. Mergers” is a dummy variable which takes the value 1 for mergers over different macro indus-
tries (according to the Thomson Financial Macro Hierarchy) or 0 otherwise. “Analyst Coverage” is the number of analysts who publish
quarterly earnings estimates of the acquiring company in the year of announcement. Data for this variable is taken from I/B/E/S.
63