Search Volume Index and Abnormal Returns Around Acquisition
Announcements
Andrea G. Armitano Track: Finance and Organization Student number: 11103108 Date Final Version: 31/01/2018 Bachelor Thesis
Track: Finance and Organization
Abstract
Search engines gather huge amount of data, describing users interest, that can be used for financial analysis. This paper investigate whether, before an acquisition announcement, the amount of attention that the target company receives from investors influences its abnormal returns. Investor attention is measured using Search Volume Index (SVI) from Google Trends. The results in this paper show that pre-announcement levels of SVI are correlated with the abnormal returns of the sample target companies around the announcement date. On the other hand, SVI change is found to be not significantly correlated to abnormal returns.
Statement of Originality
This document is written by Andrea G. Armitano who declares to take full responsibility
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 completion of the work, not for the contents.
Table of contents 1. Introduction 2. Literature review 3. Hypothesis
4. Data collection and methodology 5. Results
6. Conclusion 7. References
1. Introduction
The search for predictable patterns in the stock market inspired many papers in finance. According to the Efficient Market Hypothesis (EMH) stock prices immediately reflect new information (Fama, 1970). But this new information couldn’t be incorporated in stock values if no agent would be interested in it. Models like EMH are based on the assumption that investor attention is effortless and constant. In reality, attention is a limited resource (Kahneman, 1973). Before investing in a stock, in order to make an informed decision, investors gather financial news about it. The extent of this search for information mirrors the investors’ attention for the asset. It has been proven that the amount of information demand for a stock can influence it’s returns (Da et all. 2011, Rounsltone et all. 2012).
While the supply of financial information can be measured directly gathering the output of the main information sources (Mitchell and Mulherin, 1994), past research is not unanimous on which measure can better represent investors‘ attention. A recent development of the literature consisted in the introduction of online queries volume as measurement proxy for information demand.
Search engines have become the main way to access information. Many investors use this interfaces to gather financial news. As this people conduct different queries, data about their searches is gathered too. All this information can be used for analytical purposes. Some of the data gathered by this platforms are accessible to privates investors through the use of specific programs. One of this programs, Google Trends, allows to collect the amount of search volume related to specific topics. The output of this tool is the Search Volume Index (SVI).
Many researchers saw the introduction of this program, in 2006, as a new way to extract data about retail investors’ interests. Different studies showed that SVI could outperform interest proxies like volume and news (Mondria et all. 2010, Da et all. 2011). Following these findings some empirical papers used online queries volume as a measure to represent investor’ interest in event studies (Drake 2012, Vlstakis 2012). This literature found the search volumes to be correlated with both abnormal returns and stock volatility.
A lot of research has been conducted on the link between search volume changes and stock returns in periods of normal activity for a company (Vlastakis 2012, Preis 2013, Challet 2014). On the other hand, literature related to the same relation in periods characterized by abnormal return and abnormal search volume is more scarce. The effects of SVI around earnings announcements has been studied by Drake, Roulstone and Thornock (2012), while Da, Engelberg and Gao (2011) worked on the connection between SVI and first day returns after an IPO. Still, there is no previous literature focusing on the relation between SVI and abnormal returns around acquisitions.
The focus of this paper is to contribute to the existing literature by investigating the relation between search volumes and abnormal stock returns around acquisition announcements. More specifically this research will try to assess whether the abnormal returns of a target company around the announcement day is correlated with the pre-announcement , firm specific, search volume.
To do this, the daily returns, and SVI, of 60 acquisition target companies have been collected. The sample of companies was created by picking the 60 largest acquisition offers in the U.S. between 2012 and 2016. Acquisition announcements have been linked to abnormal levels of search activity and abnormal returns for the target firms. For this reason, the period prior and adjacent to the announcement of an acquisition will be analyzed.
This paper will be structured as an empirical study. First, some background knowledge of attention, acquisitions and SVI studies will be given in the theoretical background. Following that, the methodology used to collect the data and answer the research question will be explained. Next, the results will be shown and discussed. Based on the results found, a conclusion will be drawn and, finally, there will be some remarks for future research.
2. Theoretical framework
This paper relates to two groups of research: event studies on the behavior of abnormal returns around mergers and acquisitions announcements and papers on the relation between information attention and stock prices.
Investors Interest
Classic models of assets pricing are based on the Efficient Market Hypothesis (EMH). This hypothesis, developed by Fama in 1970, imply that, when available, new information is instantly incorporated in asset prices. This further entails, that the agents in the market have a constant interest in the new information released.
While traditional models in asset pricing consider investor attention to be a constant, Kahneman (1973) more recently demonstrated that it is a limited and costly cognitive resource. Based on this assumption, Grossman and Stiglitz (1980) proposed a new competitive equilibrium model based on noisy rational expectations equilibrium. According to this model, equilibrium is achieved when prices reflect only a portion of the information held by informed investors. This conclusion is based on the fact that, as the percentage of informed investors increases, prices become more informative, thus, the incentives to acquire information decrease.
Metron (1987) further elaborated on the market equilibrium with incomplete information and the relation between investor attention and stock markets. He developed the investor recognition
hypothesis, according to which, retail investors prefer to buy shares of companies of which they have
more information.
The investor recognition hypothesis has been empirically verified by Barber and Odean (2008), who also added to the previous literature, introducing the attention theory. Based on the limited attention conjecture of Kahneman (1973), this theory argues that retail investors buy more likely stocks of companies that recently received a lot of attention. The reasoning behind this argument is that investors can choose between thousands of stock when they have to buy, but can only sell what they already own. According to the studies from Metron (1987), investors prefer to buy stock that they know. As a result, an increase in retail attention will lead to net buying.
Literature on Acquisitions
The literature about the effects of takeovers on company’s value is plentiful. Share return studies examine the impact of acquisition announcement on share prices for target and acquirer firms (Guest et. all. 2010). The main methodology used for this studies is that of the event study, first introduced by Fama in 1969. Previous research on acquisition consistently found that, when looking and at the combined wealth of target and bidder shareholder, takeovers create value (Martynova &
Although takeovers were shown to have positive combined effect on shareholders, the abnormal returns for the targets were significantly larger than the ones of the acquirers. On average, target shareholders saw an increase in stock value around the takeover announcement date (Berk and DeMarzo 2014). Acquirer firms’ abnormal returns, on the other hand, were smaller, with many studies failing to find CAARs significantly different from 0 (Ascquith 1983, Schwert 1996, Andrade 2001). This can be explained by the fact that the acquirer company’s offer is normally higher than the estimated value of the target company. This premium is payed because the bidding company thinks adding acquiring the target will create synergies that will add to the current value of the standalone company (Berk 2014). Stock price then increases as investors anticipate that their the target stock will be swapped or payed for a value higher than the current value.
As shown by Schwert (1996) the reaction of the stock returns to acquisition announcement is not limited to the period nearly adjacent to the announcement. According to his studies, abnormal returns are produced also in the pre-announcement period. More specifically, significant CAARs were found in the period of 42 days preceding the public announcement. Schwert argued that this pre-bid abnormal returns are caused by leakages of information about the future deal or, possibly, insider trading.
Another common finding is that the kind of payment proposed in exchange for the ownership of the target significantly influence the amount of cumulative abnormal returns (Martynova & Rennboog, 2008). According to Travlos, bidding firms that announced an acquisition offer composed by stock exchange experienced significant losses (1987). Cash offers acquisitions, contrarily, didn’t produce abnormal returns.
Google Trends and Search Volume Index
Google is the leading search engine in the USA. On average, in the period from 2012 to 2016, 66.9% of search queries were conducted on Google sites. Google Trends is a ,publicly available, free access tool that allows users to extract search volume data related to the input keywords . This service, launched in 2006, can be used to collect data from 2004 to present time. It is possible to filter search volumes by location and argument. The output result is the weekly Search Volume Index (SVI). This index represents the relative, weekly, search frequency of the input term over the chosen period of time. The value can span from 0, which indicates the lowest weekly interest in the period, to 100, that coincide with the highest interest.
SVI and market activity
As, retail investors attention can’t be directly measured, many studies tried to find the best proxy for this variable. When investigating in investors decisions trading volume is commonly used proxy for investor attention (Hou 2006, Odean 2009). The reasoning behind using this variable as measure for attention is that investors trade more in stock to which they are interested. Odean also found extreme returns to be correlated to market activity (2008). Finally, advertising expenditure has been found to be correlated to investors’ attention (Lou 2014).
One major contribution to the study of the relation between information demand and the stock market came from Da et all. (2011). They were the first to research the relation between SVI and stock market behavior. Using a sample composed of the returns of all the companies included in the Russell 3000 between 2004 and 2008, they demonstrated that SVI is a robust indicator for information demand. Furthermore, Da et al. findings suggested that search volumes can predict stock prices. More precisely, an increase in SVI was linked to a 2 weeks increase period in the stock prices (2011). This is in line with Odean’s attention theory. They also proved that a high levels of abnormal SVI before an IPO announcement anticipated high first day returns.
Vlastakis and Merkellos (2012) studied the relation between information demand and stock volatility for the 30 largest companies traded in the NYSE between 2004 and 2009. They found that changes in information demands affect significantly stock volatility and trading volume. Furthermore, their finding suggest that information demand is bigger during period of high returns. This results were significant even after controlling for information supply changes.
Drake and Roulstone and Thornock used Google Trends to investigate on the levels of investors interested around earnings announcement (2012). They showed that there is a positive correlation between information demand and information supply. The more news are published about a certain company, the more investors are interested in that specific company. They also found that higher SVI before the earnings announcement lead to a lower changes in price at the announcement. This is in line with the competitive equilibrium model by Grossman and Stiglitz (1980).
As it can be seen in the previous discussion, the literature regarding acquisition and that linked to investors’ attention don’t intertwine. Previous attention studies shown the relation between SVI and abnormal returns for the company during regular activities, IPO’s and earning announcement. Nevertheless, no research has been conducted on acquisitions announcements. Similarly, M&A studies focused on factors such payment methods and information leakage to justify abnormal returns, but the effect of investor attention has not been included yet. This paper aims to clarify the relation between SVI and abnormal returns around acquisitions announcements, filling the gap in literature and linking the two branches of research.
3. Hypothesis
To study the relation between abnormal returns around the acquisition announcement and the levels of Search Volume Index this paper will use 3 hypothesis.
H1: SVI levels during the announcement period are higher than average.
This Hypothesis follows form Vlatakis’ argument that information demand is bigger during periods of high returns. If we combine this assumption with the finding that target companies achieve significantly positive abnormal returns around bid announcements, it is to expect that SVI around announcements will be higher than average.
H2: Average pre-announcement SVI scores are negatively correlated to the CARs in the announcement period.
The period around earnings announcements, as that surrounding acquisition announcements, is characterized by abnormal returns and search volumes. It could then be possible that Drake et. all’s findings, that higher levels of pre-announcement SVI are negatively related to announcement period abnormal returns, are also applicable to acquisitions.
H3: The change between average estimation period SVI and average pre-announcement period SVI is correlated to the announcement period CAR.
The last hypothesis follows from the findings of Da, Drake and Vlastakis, that an increase in SVI is positively correlated to subsequent abnormal returns.
4. Methodology
To answer the research question, this paper will first focus on the stand alone behavior of information demand (SVI), stock abnormal returns and stock volatility. Once understood how this variables are influenced by the acquisition bid, the relations between them will studies. Finally a regression will be run to ascertain witch variables influence abnormal returns.
Data collection
The sample used in this paper is constructed using the largest 60 acquisitions announcement of US companies disclosed to the public between 2012 and 2016. Only acquisition bids in which the target company was based in the US are included in the sample. The list of bid is acquired from Zephyr M&A database. Acquisition announcement are used instead of a list of completed acquisitions to avoid classification bias. For each target company, daily stock price and volume of a 300 trading days period are collected from the Center of Research and Security Prices (CRSP). Furthermore for each deal, the proposed bid payment method, and the acquisition premium is collected on Zephyr.
Google trend weekly Search Volume Index (SVI) is used to measure investors’ attention. For each company, weekly SVI was collected for the same 300 days period used to gather stock data. To extract the SVI related to the right company, the full company name is used for the search. When the company name is also a common name (e.g. APPLE), the company ticker is used instead. To avoid collecting searches non related to financial aspects of the company, when available, the query is run filtering for financial linked SVI.
Event Study
The abnormal returns used to answer the research question are found trough an event study, conducted according to the methodology developed by Fama (1969). Following this methodology, abnormal returns are found subtracting expected normal returns from realized returns:
𝐴𝑅𝑖𝑡 = 𝑅𝑖𝑡− 𝑁𝑅𝑖𝑡 To calculate abnormal returns the market model is used:
𝑅𝑖𝑡 = 𝛼𝑖+ 𝛽𝑖𝑅𝑚𝑡+ 𝜖𝑖𝑡
The event day (t) is the first trading day following, or coincident, with the public announcement. The estimation period used to calculate the regression coefficients spams from t-300 to t-260. Abnormal returns of each target are calculated over the interval t-40 to t+10. The event window is divided in announcement (t-40;t-10) and announcement (t-10;t+10) periods. The pre-announcement period is used to check whether some information about the acquisition was spilled before the announcement date.
Different Standard and Poor’s’ indexes are used as benchmark return. Using the S&P 500 for all the companies resulted in low levels of 𝑅2 for the estimation period. For this reason each company stock data is matched with either the related industry S&P index or the general S&P 500 index, depending on which one provided the highest 𝑅2.
The OLS estimates of the regression coefficient for the estimation period are found for all the targets. Using those coefficients, the normal returns during the event window are estimated.
𝑁𝑅𝑖𝑡 = 𝛼̂𝑖+ 𝛽̂𝑖𝑅𝑚𝑡
Normal returns are then subtracted from the realized returns to find the daily Abnormal Returns (ARs) for the event window. To get more insights on the effects of acquisition announcements on stock returns, daily ARs are cross-sectionally averaged using the formula:
𝐴𝐴𝑅𝑡 = 1
𝑁∑ 𝐴𝑅𝑖𝑡 𝑁
𝑖=1
According to the previous literature, acquisition announcements should create positive abnormal returns for the target company. For this reason, G-test are conducted on each daily AAR, to check whether, around the announcement date, some AAR was significantly bigger than 0. As the standard deviation of the population is not known, the cross-sectional standard deviation is used as estimate: 𝑠𝑡= √ 1 𝑁 − 1∑(𝐴𝑅𝑖𝑡 − 𝐴𝐴𝑇𝑡)2 𝑁 𝑖=1
The resulting cross-sectional standard deviations are used to calculate the G-statistics of all the AARs: 𝐺 = √𝑁𝐴𝐴𝑅𝑡
𝑠𝑡
~𝑡𝑁−1
A 5% significance level was used to evaluate the significance of the different results in this paper. As a large announcement period window is chosen testing the separate AARs instead of the Cumulative Abnormal Returns allows to get more precise results that give the chance to observe on which days, within the window, AARs are significant.
Correlations
To test whether there is some relation between abnormal returns and the level of SVI around the acquisition announcement, Cumulative Abnormal Returns (CARs) are separately calculated, for each target, during the announcement period. In order to test hypothesis 1, the SVI level in coincident with the announcement day is tested against the average of the estimation period (t-55w, t-8w) to check for significant changes in SVI.
h0: 𝑋𝑡0− 𝑋̅𝑡−55𝑤,𝑡+55𝑤= 0 h1: 𝑋𝑡0− 𝑋̅𝑡−55𝑤,𝑡+55𝑤> 0
Hypothesis 2 is tested using the Average Pre-Announcement weekly SVI (APASVI). This variable is the average of the weekly search volume from tw-7w to week tw-1. The correlation between the sample companies APASVIs and their CARs for (t-10,t+10) is found and tested:
h0: 𝑟𝐶𝐴𝑅,𝐴𝑃𝑆𝑉𝐼=0 h1: 𝑟𝐶𝐴𝑅,𝐴𝑃𝑆𝑉𝐼=<0
The following statistic was used: 𝑡 = 𝑟√𝑛−2 1−𝑟2
The last Hypothesis of this paper is tested using the variable 𝛥𝑆𝑉𝐼,which is the change between APASVI and the average SVI calculated in the estimation period (tw-55w, tw-2w). The correlation between the 60 companies 𝛥𝑆𝑉𝐼s and the corresponding announcement CARs is then tested using the same statistic used for H1:
h0: 𝑟𝐶𝐴𝑅,𝛥𝑆𝑉𝐼 = 0 h1: 𝑟𝐶𝐴𝑅,𝛥𝑆𝑉𝐼≠ 0
Following that, to gain further insights on the relations between the sampled targets announcement CARs and the variables created, the following multiple regression is run:
𝐶𝐴𝑅(𝑡−10,𝑡+10)= 𝑎0+ 𝑎1𝐶𝐴𝑆𝐻 + 𝑎2𝑃𝑅𝐸𝑀𝐼 + 𝑎4%𝛥𝑉𝑂𝐿 + 𝑎5𝛥𝑆𝑉𝐼+𝑎3𝐴𝑃𝐴𝑆𝑉𝐼
CASH is a dummy variable that assumes value 1 if the proposed method of payment for the acquisition bid was cash and 0 otherwise. Although cash was previously demonstrated to not produce abnormal returns, our preliminary results showed that inserting a dummy variable for cash gave more significant results than a dummy for share proposed bid payment.
PREM is the acquisition premium, calculated as the difference between the value of the bid and the market capitalization of the target firm.
%𝛥𝑉𝑂𝐿 is the difference between the average of the daily percentage change in volume in the pre announcement period (t-40,t-10) and that in the estimation period (t-300, t-41). This variable is inserted in the regression as volume is a commonly used proxy for information demand.
5. Results Stock returns
Table 1 presents daily average and cumulative average abnormal returns of the 60 companies sampled for the event window (t-40,t+10). Table 2 gives a visual interpretation of the abnormal returns in the same period.
During the pre-announcement period (t-40,t-10), at a significance level of 5%, no AAR was significantly different from 0. The average AAR for the period was 0.0324 %. The CAAR of this period (0.972%) was not significantly different than 0 at 5% (t = 0.150679). From the above results, it is assumed that there is no information spillage, as, if that would be the case, investors would that could result in significant abnormal returns in the pre-announcement period, is happening.
In the announcement period (t-10,t+10) the target firms experienced significant and positive AAR of, respectively, 1.885% and 5.662 % at event day -1 and 0. The AAR on day -1 was significantly bigger then 0 at 5%(t = 2.02) and that on day 0 at 1% (t = 4.05). The CAAR in this period was 12.051%, which is significant bigger than 0 at 5% (t = 1.871). This results show that announcement of acquisitions have a positive effect on the stock prices of the sampled target firms.
Table 1 Daily Average Abnormal Returns (AAR) and Cumulative Average Abnormal Returns
(CAAR) for the 60 sampled target firms during the event window. Pre-announcement period
Event Day Daily AAR (%) t-statistic Daily CAAR (%)
-40 0.14% 0.3847 0.14% -39 0.19% 0.8156 0.33% -38 -0.10% -0.3827 0.23% -37 0.28% 0.8713 0.51% -36 -1.40% -0.9807 -0.89% -35 -0.11% -0.6355 -1.00% -34 0.01% 0.0263 -1.00% -33 0.10% 0.4405 -0.90% -32 0.75% 1.4534 -0.15% -31 0.13% 0.5147 -0.02% -30 -0.27% -1.4412 -0.29% -29 0.08% 0.3325 -0.21% -28 -0.14% -0.6510 -0.35% -27 -0.35% -1.5046 -0.70% -26 -0.21% -1.0694 -0.91% -25 -0.27% -1.4231 -1.19% -24 0.42% 0.9696 -0.77% -23 0.23% 1.0356 -0.54% -22 -0.03% -0.1334 -0.57% -21 0.15% 0.5056 -0.43% -20 0.18% 0.5636 -0.25% -19 0.18% 1.0934 -0.06% -18 0.18% 0.9462 0.12% -17 0.10% 0.4654 0.22% -16 -0.30% -1.3354 -0.08% -15 0.46% 1.6952 0.38% -14 0.10% 0.3489 0.48% -13 0.51% 1.5312 0.99% -12 -0.21% -0.6392 0.79% -11 0.19% 0.9275 0.97% Announcement period
Event Day Daily AAR (%) t-statistic Daily CAAR (%)
-10 0.59% 1.2314 0.59% -9 0.24% 0.6839 0.83% -8 0.19% 0.5984 1.02% -7 0.24% 0.9568 1.26% -6 0.16% 0.6242 1.42% -5 0.56% 1.6428 1.97% -4 0.07% 0.2334 2.04% -3 -0.15% -0.7470 1.89% -2 0.89% 1.5217 2.78% -1 1.89% 2.2021 4.66% 0 5.66% 4.0553 10.33% 1 1.46% 1.1736 11.78% 2 0.25% 1.4916 12.04% 3 0.10% 0.5919 12.14% 4 0.07% 0.3141 12.21% 5 -0.22% -1.1996 11.99% 6 0.47% 1.5479 12.46% 7 0.11% 0.7151 12.58% 8 -0.26% -1.2941 12.32% 9 -0.29% -1.4745 12.02% 10 0.03% 0.1378 12.05%
-3% -1% 1% 3% 5% -40 -35 -30 -25 -20 -15 -10 -5 0 5 10 % R e tu rn Event Day Table 2
Cross-section daily abnormal returns
0 10 20 30 40 50 60 70 80 90 100 -55 -50 -45 -40 -35 -30 -25 -20 -15 -10 -5 0 R e lativ e in te re st Week Table 3 Weekly SVI
Search Volume Index
Table 3 shows the weekly cross-sectionals SVI for the 60 companies sampled during the event window. No weekly SVI in the pre-announcement window was significantly different form the complete sample average. Contrarily, all weekly SVI in the announcement period were significantly different. Weekly volumes at event week -1 and 1 were significant at 5% level (t = 1.702, t = 1658) and at 1% at week 0 (t = 3.431).
According to this results it is assumed that investor attention changes only during the announcement period. This is in line with the finding from table 1 and confirms the inference that there is no leakage of information about the acquisition announcement in the pre-announcement period.
Table 4 Average weekly Search Volume Index
(SVI) for the 60 sampled target firms during the event window. Pre-announcement Window
Event Day Weekly SVI t-test
-9 34.5 1.350 -8 37.02 1.448 -7 33.98 1.329 -6 34.28 1.341 -5 35.96 1.407 -4 36.1 1.412 -3 36.08 1.411 -2 39.48 1.544 Announcement period -1 43.52 1.702 0 87.7 3.431 1 42.38 1.658
SVI and Cumulative Abnormal returns
For each company, the difference between SVI during the announcement week and the average SVI in the estimation period (t-55w,t-8w) is calculated. t-test scores highlight that, at a significance level of 5%, all the changes in SVI were significantly bigger then 0. Thus, the null hypothesis of H1 is rejected. There is enough statistical evidence to prove that there is a significant increase of SVI during the announcement week.
Table 5 shows the correlation between the main variables used in this paper.
Out of the variables used in this research APASVI has the highest correlation with CAR. More specifically there is a negative correlation of 0.636 between the two variables. The 𝑅2 of the regression for these variables is 0.405. This is in line with hypothesis 2. The t-value on this correlation (t = -5.532) is extreme enough to reject h0 of H2 at 1% significance. There is enough statistical evidence to prove that there is a significant negative correlation between the average pre-announcement SVI and the cumulative abnormal returns. This could be explained using the work of Grossman et all.. A higher pre-announcement level of attention indicate that investors are closely following a stock. A closer attention to the stock lead to more complete asset prices. Perhaps, the stocks that receive more attention are better priced, leading to a lower price change around the acquisition announcement.
The correlation between the sample CARs and 𝛥𝑆𝑉𝐼 is -0.096 and the 𝑅2 of the regression is 0.009. The t-statistic (t = -0.648758553) is not extreme enough to reject the null hypothesis of H3. There isn’t enough statistical evidence to prove that a correlation exist between Cumulative Abnormal Returns and the change in average SVI from pre-announcement and announcement period.
Table 5 Correlation Matrix
CAR APASVI ΔSVI %ΔVOL PREM CASH
CAR 1.00000 -0.63629 -0.09626 -0.10510 -0.07658 0.18196 APASVI -0.63629 1.00000 0.32801 0.36961 0.01938 0.05610 ΔSVI -0.09626 0.32801 1.00000 0.56508 0.01985 -0.05937 %ΔVOL -0.10510 0.36961 0.56508 1.00000 0.65092 0.08459 PREM -0.07658 0.01938 0.01985 0.06509 1.00000 0.08595 CASH 0.18196 0.05610 -0.05937 0.08459 0.08594 1.00000
In table 6, the coefficients and t-statistics of the regressions can be found. From there, it can be seen that regression 1, which uses only CASH, PREM and %ΔVOL as dependent variables doesn’t explain much of the announcement CARs. None of the independent variables of regression 1 are significant at a 5% level. Adding ΔSVI to regression 1 lead to an increase of 𝑅2: from 0.085 to 0.1348. Still, ΔSVI is not significant at 5% in regression 2. Finally, in regression 3, APASVI is added. The addition of this variable increased significantly the 𝑅2 score: from 0.1348 to 0.4079. In this regression, both CASH and APSVI are significant at 5%. This confirms the previous findings that APASVI has a significant influence on announcement CARs, which, on the other hand, is not found for ΔSVI. Another finding is that %ΔVOL, which has been previously used to estimate does not seem to influence the level of abnormal returns.
Table 6 Estimated Coefficients and t-statistics form regressing the 60 CARs of the period (t-10, t+10) for the target companies on CASH, PREM, APASVI, %ΔVOL and ΔSVI
Period 2012-2016 .
INTERCEPT CASH PREM %ΔVOL ΔSVI APASVI R^2 F Significance F
Regression 1 0.1247 0.0376 -0.0010 -0.0948 0.0850 1.3308 0.2768 6.6097 1.4402 -0.6883 -1.3952 Regression 2 0.1319 0.0338 -0.0009 -0.0570 -0.0031 0.1348 1.6357 0.1832 6.8946 1.3095 -0.6382 -0.8005 -1.5552 Regression 3 0.2122 0.0392 -0.0008 -0.0038 -0.0012 -0.0025 0.4079 5.6499 0.0005 8.6809 1.8137 -0.6922 -0.0629 -0.6696 -4.3492
6. Conclusions
This paper studies the relation between investor attention and abnormal returns for target companies around an acquisition announcement. Google Trends’ Search Volume Index is used as proxy for the level of attention. The research sample is composed of the abnormal returns of US target companies linked to the 60 largest acquisition deals in the period 2012-2016.
A significant negative correlation is found between pre-announcement SVI and abnormal returns at announcement. This finding is in line with previous research, and suggest that a higher level of attention before the acquisition announcement lead to more complete stock prices that reflected better the true value of the company.
The relation between the change in SVI before the announcement and abnormal returns is also tested. Contrarily to the findings from previous studies, no significant correlation is found between the two variables. This could be because of the different estimation periods used in the paper and the strong negative correlation between SVI and abnormal returns.
One limitation of this study is the limited amount of financials retrievable form Zephyr database. Future studies could gather more financial information about the target and bidding companies to test the explanatory power of SVI on more complete models for abnormal stock returns created by acquisition announcements.
Investigating on the effects of Information supply on abnormal returns could also be a valuable addition to this research. Adding a information variable to the model could help to clarify whether investors’ attention can explain more of the abnormal returns when controlled for information supply.
7. References
Andrade, G., Mitchell, M., & Stafford, E. (2001). New evidence and perspectives on mergers.
Journal of economic perspectives, 15(2), 103-120
Asquith, P., Bruner, R.F. and Mullins Jr, D.W., 1983. The gains to bidding firms from merger.
Journal of Financial Economics, 11(1-4), pp.121-139.
Barber, B. M., Odean, T., & Zhu, N. (2008). Do retail trades move markets?. The Review of
Financial Studies, 22(1), 151-186.
Berk, J. B., & DeMarzo, P. M. (2014). Corporate finance. Pearson Education. Challet, Damien, and A. B. Hadj Ayed (2014). "Do Google Trend data contain more predictability than price returns?." arXiv preprint arXiv:1403.1715.
Da, Z., Engelberg, J., & Gao, P. (2011). In search of attention. The Journal of Finance, 66(5), 1461-1499.
Drake, Michael S., Darren T. Roulstone, and Jacob R. Thornock. "Investor information demand: Evidence from Google searches around earnings announcements." Journal of Accounting
Research 50, no. 4 (2012): 1001-1040.
Fama, E. F. (1970). Efficient capital markets: A review of theory and empirical work. The
journal of Finance, 25(2), 383-417.
Grossman, S. J., & Stiglitz, J. E. (1980). On the impossibility of informationally efficient markets. The American economic review, 70(3), 393-408.
Guest, P. M., Bild, M., & Runsten, M. (2010). The effect of takeovers on the fundamental value of acquirers. Accounting and Business Research, 40(4), 333-352.
Hou, K., Xiong, W., & Peng, L. (2009). A tale of two anomalies: The implications of investor attention for price and earnings momentum.
Kahneman, D. (1973). Attention and effort (Vol. 1063). Englewood Cliffs, NJ: Prentice-Hall. Lou, D. (2014). Attracting investor attention through advertising. The Review of Financial
Studies, 27(6), 1797-1829
Martynova, M., & Renneboog, L. (2008). A century of corporate takeovers: What have we learned and where do we stand?. Journal of Banking & Finance, 32(10), 2148-2177.
Merton, R. C. (1987). A simple model of capital market equilibrium with incomplete information. The journal of finance, 42(3), 483-510.
Mitchell, M. L., & Mulherin, J. H. (1994). The impact of public information on the stock market. The Journal of Finance, 49(3), 923-950.
Mondria, J., Wu, T., & Zhang, Y. (2010). The determinants of international investment and attention allocation: Using internet search query data. Journal of International Economics, 82(1), 85-95.
Odean, T. (1999). Do investors trade too much?. American economic review, 89(5), 1279-1298.
Preis, T., Moat, H. S., & Stanley, H. E. (2013). Quantifying trading behavior in financial markets using Google Trends. Scientific reports, 3, srep01684.
Schwert, G. W. (1996). Markup pricing in mergers and acquisitions. Journal of Financial
economics, 41(2), 153-192.
Travlos, N. G. (1987). Corporate takeover bids, methods of payment, and bidding firms' stock returns. The Journal of Finance, 42(4), 943-963.
Vlastakis, N., & Markellos, R. N. (2012). Information demand and stock market volatility.
