Can Rumors Predict Performance?

(1)

Can Rumors Predict Performance?

Testing the rumor-effect for long-term post-M&A performance prediction

Master Thesis MSc BA Corporate Finance

EWM066A20

Weeda, C.B. s1404296

s1404296@student.rug.nl

Under supervision of:

Tra Pham

Word Count: 8704

Rijks Universiteit Groningen

Faculty of Economics and Business

WSN Building, Landleven 5

Groningen, 9747AD

(2)

Abstract

(3)

1. Introduction

The amount of merger and acquisition (M&A) activity has surged within the European Union since the early 1990s. Not only technological and financial innovations fueled this process, but deregulation on mergers and acquisitions by the European Union and the later introduction of the Euro made it easier for many firms to pursue economies of scale or scope through M&A activity. By increasing their efficiency through synergies, many firms tried to uphold their competitive advantage (Ayadi & Pujals, 2005).

After the economic downturn at the beginning of the 20th century made it difficult for many companies to pursue M&A activities, the amount of M&A activity has been increasing in both number and value during the last couple of years (Goergen & Renneboog, 2004). Because of the regulated economic freedom of movement among member states, cross-border M&A activity has become more common since 1993 and is becoming a larger part of all M&A activity within the European Union.

(5)

studied extensively, using various methodologies. However, the efficiency of the capital markets has not been tested for the ability to correctly predict the future performance of the newly consolidated firm on basis of rumors. Although this might seem only a trivial matter, its implications can be extensive, both for the efficient market hypothesis (which states that information will be reflected in the stock prices as soon as it is freely available), as well as for the reliability of shareholders’ predictions on firm performance. Therefore, the main research question is:

Do firms that have a positive (negative) rumor effect outperform (underperform) the market?

In order to answer this research question, several more specific research questions need to be answered, which can be found below as hypotheses in the literature review.

(6)

2. Literature Review

The total value of cross-border mergers and acquisitions has reached more than 2% of the total USD value of M&A activity in 2000, compared with only 0.5% in the 1980s (Goergen & Renneboog, 2003). These figures clearly show an increase in the value and number of cross-border M&A activity. The enormous scale of M&A activities has a large potential for firms to be used in either a profitable, or a value-destructing way. Whether M&A activities have proven to be profitable in the short- and long-run, and what drives this value creation or –destruction, will be discussed in this section.

2.1 Performance of M&A Activities

2.1.1 Theoretical Arguments for M&A Activity

(7)

increases the size of the combined-firm synergies. As more competition exists to take over the target firm, the premium paid by the bidding firm for corporate control increases (Seth, Song & Pettit, 2000). According to the hubris hypothesis (Roll, 1986), firm managers fail to correctly value the target firm, thereby on average overvaluing the synergies that can be attained. As a result, in its most extreme form the hubris hypothesis predicts that a premium paid to shareholders of the target firm is merely a transfer from the bidding to the target firm’s shareholders, resulting in a loss for the shareholders of the bidding firm. This is also known as the winner’s curse, as the winning firm (i.e. the firm that pays the highest price) is most likely to overpay, thereby resulting in a zero net effect of the M&A activity once the firm is merged or acquired (Roll, 1986). In case of cross-border M&A activity, the information asymmetry between the bidding and acquiring firm becomes even larger, possibly resulting in even larger premiums paid (Seth, Song & Pettit, 2000).

However, not all theories argue in favor of the existence of profitable synergies, or that the driving force behind the undertaking of M&A activity is the interest of the shareholders. According to the managerialism hypothesis, managers of firms will undertake negative net present value M&A activity in order to increase their own utility. Such ‘empire-building’ is often the result of a large free cash flow within the firm and low internal control (Jensen, 1986), leading to an agency problem between the managers of the firm and its shareholders, as management seeks larger growth in assets rather than profits (Marris, 1964), in order to increase their own prestige and pay (Seth, Song & Pettit, 2000). Another reason for firms to engage in negative-NPV mergers or acquisitions is to diversify their own risk of joblessness (Amihud & Lev, 1981). In this way, managers create value for themselves at the cost of the shareholders, in order for them to keep their position (Shleifer & Vishny, 2003). For such managers, takeovers can also act as an ex ante disciplinary motive as they become anxious to be taken over themselves, or an ex post rationale as they are replaced by non-value destroying managers after a merger or acquisition (Alchian & Demsetz, 1972).

(8)

performance expectations, the long-run performance of a firm is expected to have a larger chance to outperform the market.

These theories clearly show a large impact of M&A activity on the firm. However, whether this impact will be positive or negative can not be established. We therefore propose:

Hypothesis 1: The portfolio of firms with M&A activity has significant under- or outperformance of the market.

2.1.2 Empirical Research on Long-Run Performance

(9)

Kennedy & Limmack, 1996; Loughran & Vijh, 1997; Rau & Vermaelen, 1998; Mitchell & Stafford, 1998), leading Agrawal and Jaffe (1999) to conclude that “the work starting with FHT shows strong evidence of abnormal under-performance following mergers,” despite methodological issues raised on long-run return studies (Barber & Lyon, 1997; Lyon, Barber & Tsai, 1999). For a complete review, see the article by Agrawal and Jaffe (1999), which reviews 22 articles from the 1980s and 90s. The literature reviewed in Agrawal and Jaffe (1999) can be complemented by a number of articles, although the general conclusion seems to be in support of the long-run post-M&A underperformance. Kennedy and Limmack (1996) find significant negative abnormal returns when looking at a 1- or 2-year post-M&A period. Similarly, Conn, Cosh, Guest and Hughes (2004) find negative abnormal returns of almost 20% in a 3-year post-M&A period. Some of the methodological arguments used to attribute long-run post-M&A negative returns to chance, such as by arguing the use of calendar rather than event returns (Fama, 1998), can be refuted or prove to show even larger negative post-M&A long-run performance (Gregory, 1997).

2.2 Short-Term Performance of M&A Activity

(10)

2.2.1 The Announcement Effect and Market Efficiency

One of the most important questions in financial markets’ performance, is whether investors can “beat the market” by actively managing their investments and using various analyses to earn excess returns. The belief that there are no possibilities to earn such return for a longer period of time is summarized in the efficient market hypothesis (EMH) (Fama, 1973). Although the extreme form of EMH contends that “security prices fully reflect all available information” (Fama, 1991), this is under the assumption that there are no costs to changing the prices so that it reflects newly available information, which would imply zero transaction costs. As this assumption is clearly non-realistic, a milder and more sensible version of the EMH argues that prices reflect all available information up to the point that the potential profits are larger than the marginal costs of acting on the information (Jensen, 1978). The specific form of the efficiency of the market – weak-form, semi-strong-form or strong-form – depends on the speed and availability of the information, ranging from the simple prediction of no “patterns” in stock prices in weak-form efficiency, through quick and unbiased adjustments to new publicly available information without available excess returns in semi-strong-form efficiency, to stock prices that reflect all public and private information without any opportunity to earn excess returns (and no legal barriers to publicizing private information) in the strong-form efficient market.

(11)

the announcement date (Halpern, 1983). The either positive or negative rumor-effect may reflect the expected outcome of investors, as they either perceive the M&A activity to increase the value of the combined firm above the expected value without the M&A activity, or perceive the M&A activity to be the result of value-destroying behavior (e.g. manager empire-building) or poor strategic and financial decision-making on a large scale.

Rumors on the financial markets can originate from various sources. Typically, there has been a distinction between two various types of rumors. First, the rumors that have been given special attention in widely available media, such as the “Heard on the Street” column in the Wall Street Journal, which has been used for empirical research (Pound & Zeckhauser, 1990). Secondly, the rumor may come from private information that has found its way to the public market, also called “information leakage” (Aktas, de Bodt & Declerck, 2005). This type of rumor usually entails a legal problem, as it contains information that has leaked from official sources and therefore could imply insider trading or the passing on of information that is illegal to make public. The economical value of rumors and rumor-mongering has, after many years of being accepted at face value, been proven in a theoretical setting (Van Bommel, 2003). This paper looks at the abnormal returns after a rumor of M&A activity has surfaced, using the definition of a rumor according to Zephyr: “A deal status indicating that there is an unconfirmed report,

or an announced deal but the identity of one of the parties is not known e.g. Company A is to buy a German engineering firm for GBP 5 million.” This definition fits perfectly with our research objectives, as it does not focus primarily on information that has been leaked by official sources. In order to circumvent the role of information leakage, the event window for the rumor-effect has a minimum of ten days before the official announcement of the M&A activity. More specific information on this topic can be found in section 3.1.2 of this research.

2.2.2. Empirical Findings

(12)

unanimous, as well as topics that are still being researched and debated. One finding has become widely accepted and replicated, which is the large size and positive sign of the premiums shareholders of target firms receive compared to the pre-announcement price. This value mostly lies between 20% to 40%, and has been found for the 1960s to 80s (Jarrell & Poulsen, 1989) as well as the 1990s (Mulherin & Boone, 2000). The announcement effect for the shareholders of the bidding firm has been less unanimous, as results differ from small negative abnormal returns (Mitchell & Stafford, 2000) to small positive returns (Eckbo and Thorburn, 2000). Bruner (2001) summarizes these results in his expansive review, giving a total of 13 researches that find significantly negative results, and 17 reports of significantly positive results. Of all the literature, only the paper by Pound and Zeckhauser (1990) focuses on the rumor effect of an acquisition. Unfortunately, it also focuses on the returns for shareholders of the target firm, as opposed to the focus on the returns for shareholders of the bidding firm in this research. Although the significance and sign of the announcement effect has not been established, various value drivers have been identified and researched, in order to better understand their influence on the size and sign of any announcement effect. The hostility of the takeover (in the form of a tender offer or hostile acquisition) has been shown to influence the abnormal returns for shareholders, as hostile acquisitions and tender offers increase the return for both target firm- and bidder firm shareholders (Loughran & Vijh, 1997, among others). The strategic implications of the M&A activity also influence returns, as diversification by acquiring or merging with unrelated business is related to lower returns for bidding firms (Maquieria, Megginson & Nail, 1998) or a loss in value for the combined firm (Berger & Ofek, 1995).

(13)

value-oriented (have low book-to-market ratios), and are therefore judged to make better acquisitions which lead to better result, a theory that has been called the performance extrapolation hypothesis (Rau & Vermaelen, 1998).

The importance of rumors on the financial market has been confirmed as well, showing that during the pre-announcement period there are abnormal returns before any official publication on the M&A activity has been made, having a size of approximately one third of the target takeover premium (Jarrel & Poulsen, 1989). This result shows that the financial market not only reacts to an announcement of an M&A activity, resulting in an announcement effect, but that the financial markets even respond to rumors of M&A activity, resulting in a so-called rumor-effect. This rumor-effect follows the same reasoning as the announcement effect, as it is a redistribution of shareholder wealth and expected firm performance, based on uncertain information. Following empirical research that has found strong evidence for an abnormal return on the day of the announcement (Agrawal & Jaffe, 1999), and following similar reasoning for the rumor-effect, we propose:

Hypothesis 2: The rumor of a European merger or acquisition will create a significant reaction in the value of the stocks of the bidding firm.

2.3 From Rumor-Effect to Long-Run Performance

(14)

and may even lead to failure of the merger or acquisition (Kole & Lehn, 2000). The uncertainty of the successfulness of the integration process can be measured as an additional risk, called integration risk, which has been found to be significant and positive (Evans, Kim & Patro, 2007). The result of this can be an immediate drop in the value of the stock; an effect that has frequently been documented by researchers as a negative announcement or rumor effect (as discussed in section 2.2).

The increase in return investors demand for firms, as a result of a rumor, is due to the uncertainty that the rumor brings with it, as the future of the acquiring firm becomes less certain. As, in time, the future becomes more clear for the acquiring firm, the systematic and thereby total risk of the firm decreases after the integration period, which is usually 6 months after the M&A activity (Barath & Wu, 2005). After this period, the decrease in risk and therefore required rate of return by the investors leads to an increase in the stock price as cash flows are discounted by a lower demanded rate of return. This would mean that, after a 6-month period of higher demanded returns, stock prices increase due to decreasing demanded returns by investors. This would therefore result in price increases of the stock, which leads to outperformance of firms that do not have such a decrease in perceived risk, as the higher demanded rate of return was priced into the stock price at the time of the rumor- or announcement effect. Any decrease in the integration risk would therefore correspond with an increase in stock price, leading to abnormal performance in the long-term post-acquisition period. We therefore propose:

Hypothesis 3: The portfolio of firms with M&A activity and a negative rumor-effect outperforms the portfolio of firms without M&A activity.

Some earlier research, however, has found a positive rumor- or announcement effect for the acquiring firm. A positive rumor- or announcement effect can be attributed to an increase in the expected cash flows of the acquiring firm due to the M&A activity, and therefore lead to an increase in the stock price of the acquiring firm. Although the

(15)

the increase in demanded return. Therefore, in the short run the stock price of the acquiring firm increases.

In the longer run, namely after the 6-month integration period, the integration risk of the acquiring firm falls as the uncertainty of the cash flows is diminished, leading to lower systematic and therefore total risk (Bharath & Wu, 2005). As the demanded rate of return for the acquiring firm falls, its future cash flows are discounted by a lower rate of return, leading to an increase in stock price. As similar, non-acquiring, firms do not face such a fall in demanded return, as these firms never faced any integration risk nor the

corresponding increase in demanded return, the acquiring firms outperform the non-acquiring firms in the long run. We therefore propose:

Hypothesis 4: The portfolio of firms with M&A activity and a positive rumor-effect outperforms the portfolio of firms without M&A activity.

(16)

3. Methodology

3.1 Determining the Rumor Effect

3.1.1 Data

(17)

The firms for which data were available and all requirements are met, can be found below in Table 1 and 2.

Table 1. Events in Sample by Industry (Subcategories with more than 8 categories are shown

Table 2. Events in Sample per Country

US SIC Industry Classification Country

Agriculture, Forestry, and Fishing 0 Austria 2

Mining 8 Belgium 2

Construction 5 Germany 13

Manufacturing 61 Denmark 3

Food and Kindred Products 14 _Spain ₁₅

Printing, and Publishing 9 _Finland ₄

Transportation 46 France 34

Communications 18 _{Great Britain} ₄₈

Electric, Gas and Sanitary Services 14 _Greece ₅

Wholesale Trade 7 Ireland 4

Retail Trade 3 Italy 23

Finance, Insurance, and Real Estate 27 Luxembourg 2

Depository Institutions 12 _Netherlands ₄

Services 12 Portugal 4

Business Services 10 _Sweden ₆

Public Administration 0

(18)

3.1.2 Statistical Analysis

The rumor-effect will be researched using an event study, for which the estimation window typically is 252 days (the number of trading days in a year). During this period, which should have a minimum of 120 days (Mackinley, 1997), the normal return of the firm can be determined before the event window. However, in order to rule out any “information leakage” from the official announcement of the M&A activity, an event window is chosen from twenty days before the rumor, until ten days after the rumor in order to allow for any slow responses of the market, while keeping a minimum of ten days between the event window and the official announcement Thus, this study uses an event window of 31 days. During this event window, the abnormal returns of firms of which a rumor exists of M&A activity are measured. The time frame can be seen in Figure 1.

Figure 1. Time frame of event study.

By using a market and risk adjusted model, chosen among the models presented by Brown and Warner (1985) and McKinley (1997), the normal returns of the market will be calculated. The return of a firm’s stock can be defined as:

1 1) ( − − − = t t t it I I I R ,

where Rit is equal to the return of the stock of company i on time, and It and It-1 are the

(19)

The market and risk adjusted model can be written as: it mt i i it R R =

α

+

β

( )+

ε

,

where the variables

α

iand

β

i are estimates of the stock variables using a linear regression,

while the variable Rmt is the return of the market at time t. The market used in this

research is the FTSE Euro 100 index. The error term

ε

it has an expected mean of zero and

a variance of

σ

2

it. Considering the rumor date is t=0, the timeframe for estimating the

normal return is t=-252 until t=-21. Using above-mentioned model for all companies included in the sample, the model variables can be estimated, and then be used in order to forecast the normal returns during the event window. The abnormal returns can be defined as: ) ( it it it R E R AR = − , with E(Rit)=E(

α

i)+E(

β

i)Rmt ,

where ARit are the abnormal returns of firm i on date t, Rit are the observed returns of firm

i on date t, and E(Rit) are the normal returns of firm i on date t. We can than calculate the

average abnormal return, using:

∑

= = N i it t AR N AR 1 1 ,

This figure shows the abnormal return averaged over day t. If the average abnormal return ARt differs significantly from zero, there has been a reaction of the market on that day due to the event (i.e. the rumor). Once the average abnormal return has been calculated, we can calculate the cumulative abnormal return, which is defined as:

(20)

While the ARt can show significant abnormal returns on a specific day, the CAR(t1,t2)

can show a significant prolonged reaction of the market during a specific interval, in this research during t1 = -20 and t2 = +10.

Using the CAR(t1,t2), we can check whether the rumor of M&A activity of a firm has led

to significant abnormal returns during the event window. Following hypothesis 2, the specification to test for a significant rumor effect, following Goergen and Renneboog (2003) is: 0 ) 2 , 1 ( ) 2 , 1 ( > t t t t CAR CAR σ , where: T AR CAR t t ) ( ) ( (1,2) σ σ = and

∑

− − = − = 21 252 2 ) ( 231 1 ) ( t t AR AR AR

σ

(21)

3.2 Long-run performance

3.2.1 Data

Now that the sample has been divided into three groups (a positive, non-significant and negative rumor effect), the long-run performance of these groups compared to the market can be researched. The data for researching this long-run performance has been gathered using Datastream, where the monthly returns of the firms that have performed M&A activity, the size of the firm (measured as market valuation) and the book-to-market ratio have been collected. In order to compute the returns of the market, along with the size and book-to-market premium, additional data has been downloaded on 250 European firms that make up the Euronext 100 and Euronext 150, in order to reflect the market as complete as possible while limiting the amount of time necessary to do so. The data collected on these firms comprises the monthly adjusted returns, the value of each firm (measured as market valuation) and the book-to-market ratio. Also, the monthly risk-free rate has been collected using Datastream, by downloading the 3-month EURIBOR bond rate.

The data files of the monthly returns are used to create a portfolio of firms that had a rumor on an upcoming M&A activity in the three years before the event. In this way, a portfolio of monthly returns is created that is updated every month in order to include firms that have had a rumor on upcoming M&A activity, and exclude firms of which the rumor on an M&A activity has been more than 36 months ago. Three portfolios have been made: one with all firms that have had a rumor on M&A activity in the past three years, and two for firms that had a significant rumor-effect due to their M&A activity, split in two samples according to the sign of their respective rumor-effect.

(22)

The results of the event-study can be linked to the long-term performance of the firms with M&A activity by using a calendar-time model (Kothari and Warner, 1997; Barber and Lyon, 1997). In order to find these results, a calendar time Fama-French three-factor model will be used. In this model, above-mentioned portfolio of three-year post-M&A is benchmarked against the performance of the market, in order to calculate the long-term abnormal returns of the portfolio. The period of three years is defined following recent literature, such as André, Kooli and L’Her (2004). By regressing monthly excess returns of a portfolio of firms that have performed M&A activity in the preceding three years on a Fama-French three-factor model, significant values of the intercept of the Fama-French three-factor model can be defined as abnormal returns. Each month, the portfolio is refreshed in order to include new firms that have had an M&A activity in the previous month, and exclude firms whose M&A activity has been more than three years ago. For instance, Gas Natural SDG SA was included in the portfolio starting October 2003, as a rumor of an M&A activity surfaced on September 15th 2003. In that month, Schneider Electric SA was no longer included in the rumor portfolio, as a rumor of its M&A activity surfaced on September 3rd 2000, 37 months earlier.

The calendar time Fama-French three-factor model can be expressed in the following way: it Ft mt t t Ft pt R HML SMB R R R − =α +β₀ +β₁ +β₂( − )+ε ,

where Rpt is the return of the acquisitive portfolio (or, equivalently, the positive or negative rumor-effect portfolio), RFt is the risk-free rate (defined as the 3-month EURIBOR bond rate),

α

are the abnormal returns of the portfolio, while β₀,

β

₁ and

β

₂

are the coefficient estimates of the HML, SMB and market premium factors.

(23)

market-to-book firms (HML), and the difference between a portfolio of small firms and one of large firms (SMB). These so-called mimicking portfolios are used to calculate the market premiums for these factors, and are set up in the following manner. Using the EuroNext 100 and the EuroNext 150 index firms, in order to fully capture the different segments and diversity of the European markets while limiting the amount of time necessary for data processing (considering the enormous amount of firms available on all European stock markets), a division is made in the size of the firms. Using the median size of all 250 firms, the firm is divided into two groups, a small (S) and big (B) one. For the book-to-market ratio, the total number of firms is divided into three groups: one for the upper 30% (H), one for the middle 40% (M), and one for the bottom 30% (L). The book-to-market ratios used are those available from December in the previous year, for information availability purposes. In June of each year, the portfolios are reweighted in order to reflect any changes in book-to-market values, and to make sure these new figures are available. For each month, the SMB factor is calculated using the difference in returns between the average of the three small-size portfolios (which are S/H, S/M and S/L) and the average of the three large-size portfolios (which are B/H, B/M and B/L). Similarly, for the HML factor the difference in returns between the average of the two high market portfolios (which are H/S and H/B) and the average of the two low book-to-market portfolios (which are L/S and L/B) are calculated. Finally, the excess book-to-market return is calculated as the value-weighted return of all 250 stocks minus the risk free rate for each month.

The portfolio returns are calculated in two ways: equally and value-weighted. In case of an equally-weighted portfolio, the monthly returns of the acquisitive firms and non-acquisitive firms are averaged in each portfolio. In the value-weighted analysis, the returns of the acquisitive firms are averaged after multiplying each return with the firm size as measured at the end of the previous year, and dividing the portfolio return by the total size of all the combined firms. The same practice is used for the non-acquisitive firms (as well as for the negative- and positive rumor-effect portfolios).

(24)

(25)

4. Results

4.1 Event Study

In this section, the results of the event study and the long-term performance study will be presented, as well as the statistical properties of the found results.

After calculating the α and β for each firm in the sample, the expected returns, the abnormal returns, the average abnormal returns, the cumulative abnormal returns and the variance for the average abnormal returns and the cumulative abnormal returns have been calculated. Table 2 shows the average α and β for the sample.

Table 2. Average αααα and ββββ of firms in sample

alpha beta

0.000288 0.534785

In the Appendix additional statistical properties can be found of the sample that has been used.

(26)

Figure 2: Distribution of the significant cumulative abnormal returns 0 2 4 6 8 10 12 14 -20% -18% -16% -14%-12% -10% -8 % -6% -4% -2% 0% 2% 4% 6% 8% 10% 12% 14% 16% 18% 0.2 0

Of the 121 firms that had a significant rumor-effect, 67 firms had a positive abnormal return during the event window, while 54 had a significant negative abnormal return during the event window. Of the sample, 48 firms had a non-significant rumor-effect, indicating that any abnormal returns during the event window of the rumor are not significantly different from zero and are therefore in line with the fluctuations of the market itself. These results clearly support hypothesis 2, as over 70% of the firms in this sample have a significant rumor-effect, opening the way to research of the performance of these firms after this rumor-effect

4.2 Long-Run Performance

(27)

firms with a negative rumor-effect, and one containing all firms of which a rumor of M&A activity did not lead to any significant abnormal returns in the financial markets. The return of these portfolios has been calculated using both equally-weighting (i.e. simple averages of the returns of the firms in the portfolio) and value-weighting (i.e. the returns of each firm have been multiplied with the size of the firm, after which the total of the returns is divided by the total size of all the firms in the portfolio). Using market data, the HML and SMB factors have been calculated, as well as the market excess returns. Using the portfolio returns, a linear regression can be done on all four acquisitive portfolios.

As Loughran and Ritter (2000) argue, an Ordinary Least Squares (OLS) regression will lead to inefficient results, as the variance of the error term is dependent on the (highly varying) number of firms per month, leading to a possible problem of heteroskedasticity. Therefore, they argue in favor of the use of Weighted Least Squares (WLS) using the square root of the number of observations per month as weighting factor, which we will use for our analysis. However, Francoeur (2005) argues in favor of using the reciprocal of the square root of the number of observations per month ( 1/ N ) as a weighting factor, which we will use as well. Although this decreases the explanatory power of the WLS regression, the significance of the independent variables greatly increases. For a comparison of the explanatory power of the models using the two different weighting factors, see Table 3.

(28)

Non-Sign. 0.50 0.53 0.39 0.76

As we find, the use of Ordinary Least Squares, which has no disadvantage compared to WLS according to Mitchell and Stafford (2000), leads to statistically less significant results with lower explanatory power (as evident in the R2 statistic) when compared to the Weighted Least Squares regression when using the square root of the number of observations per month included in the portfolio as weighting factor. A comparison between the results can be found in Table 4.

Table 4. R2 of the portfolios using two different regression analyses.

Regression Weighted Least Squares Ordinary Least Squares

Value-Weighted Equally Weighted Value-Weighted Equally Weighted

Acquisitive 0.70 0.88 0.58 0.81

Positive 0.79 0.82 0.75 0.77

Negative 0.74 0.83 0.62 0.76

Non-Sign. 0.39 0.76 0.42 0.62

Based on these results, we will only use the Weighted Least Squares Regression using two different weighting factors for both the value-weighted and the equally weighted portfolios.

4.2.1 Equally Weighted Portfolio with WLS regression

(29)

be found in other research, but this is no cause for any lower statistical significance of the results of this model. When looking at the other three portfolios in the reciprocal weighting factor model, we see that the portfolio of firms with a positive rumor-effect has a significant positive abnormal return, while the portfolio with firms with a negative rumor-effect has a significant negative abnormal return. This finding is not in line with our hypothesis 3, as we expected firms with a negative rumor-effect to have a significant positive abnormal return. However, this finding does seem to be in line with the proposed hypothesis 4. The beta of the different portfolios varies as well, the lowest result being 0.8 for the acquisitive portfolio, implying that this portfolio has a lower market risk than the market, while the portfolio of firms without a significant rumor-effect has a beta of 1.13, which means that it carries more risk than the market.

Table 5. Regression of the Equally Weighted Portfolio of acquiring firms on the market premiums using Weighted Least Squares Regression with 1/ N as Weighting Factor

Acquisitive Positive Negative Non-Significant

α (intercept) -0.011 (-2.60)** 0.017 (2.93)*** -0.011 (-1.89)* 0.006 (0.50) f m R R − 0.802 (7.09)*** 1.117 (9.54)*** 0.984 (8.02)*** 1.128 (6.02)*** SMB -0.302 (-1.67)* -0.625 (-2.22)** 0.053 (0.22) -0.528 (-1.04) HML -0.054 (-0.49) -0.275 (-1.10) -0.253 (-1.84)* 0.579 (1.20) R2 0.513 0.664 0.509 0.527 Adjusted R2 0.494 0.649 0.489 0.506

*=significant at the 10% level, **=significant at the 5% level, ***=significant at the 1% level

(30)

under- or outperformance when compared to the market, but also that there is no significant difference in returns between a portfolio consisting of firms with a positive rumor-effect and a portfolio consisting of firms with a negative rumor-effect.

Table 6. Regression of the Equally Weighted Portfolio of acquiring firms on the

market premiums using Weighted Least Squares Regression with N as Weighting

Factor

α (intercept) 0.003 (1.19) 0.002 (0.53) 0.002 (0.58) 0.005 (1.61) f m R R − 0.897 (22.53)*** 0.883 (16.39)*** 1.031 (19.23)*** 0.734 (13.63)*** SMB 0.303 (2.76)*** 0.489 (3.58)*** 0.112 (0.73) 0.271 (1.73)* HML 0.001 (0.01) -0.097 (-0.84) -0.147 (1.22) 0.237 (1.93)* R2 0.879 0.819 0.834 0.764 Adjusted R2 0.874 0.811 0.827 0.753

For all four portfolios in both models a significant beta has been found, although there is some deviation between the two models. The largest beta can be found in the portfolio of firms without a rumor-effect: according to the reciprocal weighting factor model, the beta of this portfolio is 1.12, while the model of Table 6 shows a beta of 0.73, a difference of 0.4 times the market return.

(31)

implying a premium due to additional risk because of more risky (i.e. smaller) firms within the portfolio. Again, the influence of the weighting factor is clearly demonstrated. In conclusion for the equally weighted WLS regression results, we can say that the differences due to the weighting factor make it difficult to draw any conclusions, but that all portfolios have a significant beta, and the reciprocal weighting factor model supports both hypothesis 1 (underperformance of the acquisitive portfolio compared to the market), as well as hypothesis 4 (outperformance of the portfolio consisting of firms with a positive rumor-effect compared to the market) of this research, while hypothesis 3 shows a negative abnormal return instead of the expected positive abnormal return.

4.2.2 Value-Weighted Portfolio with WLS regression

The results of the value-weighted WLS regression can be found in Table 7 and Table 8, where the weights of the firms in the various portfolios have been used to create a more pronounced “hot-market”-effect of the returns of the firms within the portfolio, so that months with a lot of firms in the acquisitive portfolios will not be averaged with months with only a few firms in the acquisitive portfolio.

The regression in Table 7, using the reciprocal of the square root of the number of observations per month as weighting factor, shows a significant abnormal return for the acquisitive portfolio, as the coefficient of the intercept is -0.039 and is significant at the 1% level. This result indicates that a portfolio of firms with recent M&A activity underperforms when compared to the market. In this model, the beta’s of each portfolio are all highly significant at the 1% level, indicating a strong relation between the market excess returns and the return of the portfolio. The beta of each portfolio varies from 0.54 to 1.06, indicating that all portfolios are at most as (beta) risky as the market, but the acquisitive portfolio consisting of all firms with recent M&A activities is only half as risky as the market.

(32)

at the 1% level. These results, as well as the abnormal return for the acquisitive portfolio, are all in line with hypothesis 1 and hypothesis 4 of this research, while there is again a reversal of the expected long-run abnormal returns for firms with a negative rumor-effect. The coefficient of the SMB factor, which is again significant at the 10% level, is lower than in the equally weighted portfolio in the acquisitive portfolio WLS regression. This is expected as larger firms have a more pronounced influence on the outcome in this value-weighted model.

In Table 7, the coefficient estimates and t-values of the WLS regression using value-weighted returns can be found, which uses the square root of the number of observations per month as weighting factor. In this model, the acquisitive portfolio no longer has any significant abnormal return, indicating that there is no significant out- or underperformance compared to the market when using this type of regression analysis. However, apart from the significant beta’s of each of the portfolios, a significant positive abnormal return for the portfolio consisting of firms with a positive rumor-effect can be found. This is in contradiction with hypothesis 3, while the portfolio consisting of firms with a negative rumor-effect does not have any significant negative abnormal returns.

Table 7. Regression of the Value-Weighted Portfolio of acquiring firms on the market premiums using Weighted Least Squares Regression with 1/ N as Weighting Factor

α (intercept) -0.039 (-5.36)*** 0.019 (3.83)*** -0.033 (-4.17)*** 0.004 (0.32) f m R R − 0.539 (2.69)*** 0.967 (9.52)*** 0.842 (4.82)*** 1.058 (4.85)*** SMB -0.556 (1.73)* -1.155 (-4.73)*** 0.056 (0.17) -1.129 (-1.92)* HML 0.339 (1.75)* -0.285 (-1.32) -0.050 (-0.26) 0.952 (1.69)* R2 0.385 0.718 0.377 0.504 Adjusted R2 0.361 0.706 0.353 0.481

(33)

Table 8. Regression of the Value-Weighted Portfolio of acquiring firms on the market premiums using Weighted Least Squares Regression with N as Weighting Factor

α (intercept) 0.001 (0.23) 0.007 (2.27)** -0.002 (-0.34) -0.001 (-0.01) f m R R − 0.915 (12.85)*** 0.809 (15.68)*** 1.134 (14.52)*** 0.738 (5.70)*** SMB -0.035 (-0.18) 0.058 (0.44) 0.056 (0.25) 0.086 (0.23) HML 0.219 (1.39) -0.102 (-0.93) 0.086 (0.49) 0.684 (2.32)** R2 0.700 0.794 0.744 0.392 Adjusted R2 0.688 0.786 0.734 0.364

When looking at the results of all four regressions, several conclusions can be drawn. First of all, the choice between 1/ N or N as weighting factor largely influences the outcome, for reasons not entirely clear. When looking at the reciprocal weighting factor model, which has highly significant results in both equally weighted- and value-weighted WLS, we find that there is a significantly negative return for the acquisitive portfolio. Therefore, we can accept hypothesis 1 which states that the portfolio of acquisitive firms will perform significantly different from the market, as firms that have recently performed M&A activity will underperform compared to the market, and is in line with other research (André, Kooli & L’Her, 2004).

(34)

(35)

5. Discussion

This research aims to provide an insight into the relevance of the rumor-effect for long-term performance, by comparing rumor-effects due to rumors of M&A activity with the 36-month post-acquisition abnormal performance, in order to answer the following research question:

Do firms that have a positive (negative) rumor effect outperform (underperform) the market?

(36)

3, as the long-run abnormal performance of firms with a significant negative rumor-effect was found to be negative in two of the four regression analyses, while being non-significant in the other two regression analyses. This result suggests that for firms with a negative rumor-effect, there is an increase in demanded return by investors due to an increase in risk around the time of the rumor of M&A activity, which leads to a drop in the value of its stocks. Since the long-term abnormal returns are negative as well, the fall in demanded return over time is not sufficient to increase the performance of the new combined firm. This implies that for firms which bear an increase in firm risk which is not offset by high expected future cash flows around the time of a rumor of M&A activity, the decrease in demanded return over time does not offset the lower-than-expected cash flows. Therefore, the financial markets that prices in the risk of a firm when a rumor of M&A activity surfaces correctly predicts that cash flows will be low, leading to a price drop at the time of the rumor (the so-called rumor-effect), but despite these low expectations such firms still underperform. The prediction of the future performance of the combined firm is thus positively biased, leading to too high expectations of the cash flows earned in the post-acquisition period.

For those firms with a positive rumor-effect, we find a highly significant positive abnormal return in the long-run post-acquisition period, as predicted by hypothesis 4. Such firms have an expected cash flows which offsets the increase in demanded return, and the decrease of integration risk in the post-acquisition period leads to a significant positive abnormal return, which is in line with other research (Bharath and Wu (2005).

(37)

reverse is true for value firms. For glamour firms, because of good past performance, the M&A rumor will lead to greater abnormal returns than for value firms (Agrawal & Jaffe, 1999).

Figure 3. Division by Sign of Significant Rumor-Effects per Book-to-Market Ratio

0 5 10 15 20 25 30 35 40 45 50 H M L

Book -to-M ark e t Ratio

N u m b e r o f F ir m s Negative Positive

(38)

by the performance extrapolation hypothesis. The difference in results with Rau and Vermaelen (1998) may be attributed to a difference in sample, as they have studied M&A performance with bids announced or completed between 1980 and 1991. As this study uses data of M&A activity that has been rumored or announced mainly after 2001, the sharp fall of stock prices in 2000 has been a harsh lesson for many investors, which may have caused them to be wary of firms that are overvalued. Therefore, firms with a high book-to-market ratio might have been regarded with less positive expectations due to previous performance, moreover, even with fear of making the same mistakes as before by over-valuating the firm and its M&A activity. For value firms, the opposite might be true as investors value the decision of a value firm to engage in M&A activity highly, due to the potential it has in unlocking the firm value, for instance for firms who are conservative with their investment and can use their assets in a more efficient way due to the M&A activity.

(39)

6. Conclusion

This research focuses on the long run performance of firms with M&A activity, as it tries to establish a link between the earliest moment of future performance of M&A activity in the form of a rumor, and the long-run performance up to 3 years after the M&A activity. Following the efficient market hypothesis, which does not state that the market is always right, but states that some outcomes are more likely than others, financial markets might be able to predict future long-term performance of firms with M&A activity. This research is the first to directly combine short-term rumor-effects that have been found using an event study with the long-term performance of the firms in order to establish the analytical and predictive powers of the financial markets.

The results of this research are surprisingly clear. Not only did more than 70% (121 out of 169 firms) of the firms in the final sample have either a positive or negative rumor-effect, indicating the (semi-) strong form efficiency of the market, but in three of a total of four regressions did the portfolio consisting of firms with a positive rumor-effect have significant positive abnormal returns compared to the portfolio consisting of firms with a negative rumor-effect. In two of the four regressions, significantly negative abnormal returns were also found for the portfolio of negative rumor-effect firms, further establishing the difference in abnormal returns between the two portfolios. The zero abnormal returns found for the portfolio consisting of firms without a rumor-effect underlines these findings as well, since the abnormal returns can only be found in the portfolios of the firms with a significant rumor-effect.

(40)

between ordinary least squares (OLS) and weighted least squares (WLS) regression analysis, in order for the most suitable methodology to be chosen independent of the results.

Although attention has been given to details in order to obtain as significant and reliable results as possible, this research has not been done before and may therefore suffer from errors in methodology and the data that was used. Future research could be in line with this research, but by using easier-to-obtain samples, for instance in the U.S., which may yield a larger sample and more complete market data, the problem that this research has faced by using data from 15 different financial markets can be overcome.

(41)

References

Agrawal, A., Jaffe, J.F., Mandelker, G.N. (1992). “The Post-Merger Performance of Acquiring Firms: A Re-examination of an Anomaly”, Journal of Finance 47(4): 1605-1621.

Agrawal, A., and Jaffe, F.E. (1999). “The Post Merger Performance Puzzle”, in Advances

in Mergers and Acquisitions, Cooper, C. and Gregory, A. (eds), JAI Press: New York.

Aktas, N., De Bodt, E., and Declerck, F. (2002). “Is there Information Leakage around Business Combinations on the French Market?”, EFMA 2002 London Meeting.

Alchian. A. A., and Demsetz, H. (1972). “Production, Information Cost and Economic Organization”, The American Economic Review 62 (5): 777-795.

Amihud, Y., and Lev, B. (1981). “Risk Reduction as a Managerial Motive for Conglomerate Mergers”, Bell Journal of Economics 12(2): 605-617.

Anderson, C., and Mandelker, G.N. (1993). “Long Run Return Anomalies and the Book-to-Market Effect: Evidence on Mergers and IPOs”, Unpublished Paper Joseph M. Katz

Graduate School of Business, University of Pittsburgh.

Andre, P.A., Kooli, M. and L’Her, J-F. (2004). “The Long-Run Performance of Mergers and Acquisitions: Evidence from the Canadian Stock Market”, Financial Management 33(4): 27-43.

(42)

Barber, B.M and Lyon, J.D. (1997). “Detecting Long-Run Abnormal Stock Returns: The Empirical Power and Specification of Test Statistics”, Journal of Financial Economics 43(3): 341-372.

Berger, P.G., and Ofek, E. (1995). “Motives for Takeovers: An Empirical Investigation”,

Journal of Financial and Quantitative Analysis 28(3): 347-362.

Bharath, S. T. and Wu, G. (2005). “Long-Run Volatility and Risk around Mergers and Acquisitions”, AFA 2006 Boston Meetings Paper

Bommel, J. van. (2003). “Rumors”, The Journal of Finance 63(4): 1499-1519

Bradley, M., Desai, A., and Kim, E.H. (1988). “Synergistic Gains from Corporate Acquisitions and their Division between the Stockholders of Target and Acquiring Firms”, Journal of Financial Economics 21(1): 3-40.

Conn, R.L., Cosh, A., Guest, P.M., and Hughes, A. (2004), “The Impact on UK Acquirers of Domestic, Cross-border, Public and Private Acquisitions”, Journal of

Business Finance and Accounting 32(5-6): 815-870.

Eckbo, E., and Thorburn, K. (2000). “Gains to Bidder Firms Revisited: Domestic and Foreign Acquisitions in Canada”, Journal of Financial and Quantitative Analysis 35(1): 1-25.

Eiteman, D.K., Stonehill, A.I., and Moffett, M.H. (2006). Multinational Business

Finance, Pearson Addison-Wesley: Boston, MA, 11th edition.

(43)

Fama, E.F. (1973). “Efficient Capital Markets: A Review of Theory and Empirical Work”, Journal of Finance 25(2): 282-417.

Fama, E.F. (1991). “Efficient Capital Markets: II”, The Journal of Finance 46(5): 1575-1617.

Fama, E.F. (1998). “Market Efficiency, Long-Term Returns and Behavioral Finance”,

Journal of Financial Economics 49(3), 283-306.

Francoeur, C. (2005). “The Long-Run Performance of Cross-Border Mergers and Acquisitions: The Canadian Evidence”, HEC Montreál Working Paper.

Franks, J., and Harris, R.S. (1989). “Shareholder Wealth Effects of Corporate Takeovers: The UK Experience 1955-1985”, Journal of Financial Economics 23(2): 225-249.

Franks, J., Harris, R.S., and Titman,S. (1991). “The Post Merger Share-Price Performance of Acquiring Firms”, Journal of Financial Economics 29(1): 81-96.

Gregory, A. (1997). “An Examination of the Long Run Performance of UK Acquiring Firms”, Journal of Business Finance and Accounting 24(7&8): 971-1002.

Halpern, P. (1983), “Corporate Acquisitions: A Theory of Special Cases, A Review of Event Studies Applied to Acquisitions”, Journal of Finance 38(2): 297-317.

Jarrell, G., and Poulsen, A. (1989). “The Returns to Acquiring Firms in Tender Offers: Evidence from Three Decades”, Financial Management 18(3): 12-19.

Jensen, M.C. (1978). “Some Anomalous Evidence Regarding Market Efficiency”,

(44)

Jensen, M.C., and Ruback, R. (1983). “The Market for Corporate Control: The Scientific Evidence”, Journal of Financial Economics 11(1-4): 5-50.

Jensen, M.C. (1986). “Agency Costs of Free Cash Flow, Corporate Finance, and Takeovers”, The American Economic Review 76(2): 323-329.

Kennedy, V.A., and Limmack, R.J. (1996). “Takeover Activity, CEO Turnover, and the Market for Corporate Control”, Journal of Business Finance and Accounting 23(2): 267- 285.

Kole, S., and Lehn, K. (2000), “Workforce Integration and the Dissipation of Value in Mergers: The Case of USAir’s Acquisition of Piedmont Aviation”, in Kaplan, S.N. (ed.),

Mergers and Productivity, University of Chicago Press: Chicago.

Koller, T., Goedhart, M., and Wessels, D. (2005). Valuation: Measuring and Managing

the Value of Companies, Wiley: Hoboken, NJ, 4th edition.

Kothari, S.P., and Warner, J.B. (1997). “Measuring Long-Horizon Security Price Performance”, Journal of Financial Economics 43(3): 301-339.

Loughran, T., and Vijh, A.M. (1997). “Do Long-Term Shareholders Benefit from Corporate Acquisitions”, Journal of Finance 52(5): 1765-1790.

Loughran, T., and Ritter, J.R. (2000). “Uniformity Least Powerful Test of Market Efficiency”, Journal of Financial Economics 55(3): 361-389.

Lyon, J.D., Barber, B.M., and Tsai, C. (1999). “Improved Methods for Tests of Long-Run Abnormal Stock Returns”, Journal of Finance 54(1): 165-201.

MacKinlay, A.C. (1997). “Event Studies in Economics and Finance”, Journal of

(45)

Maquieria, C., Megginson, W., and Nail, L. (1998). “Wealth Creation versus Wealth Redistributions in Pure Stock-for-stock Mergers”, Journal of Financial Economics 48(1): 3-33.

Marris, R.L. (1964). The Economic Theory of Managerial Capitalism, Macmillan: London, UK.

Mitchell, M.L., and Stafford, E. (1998). “Managerial Decisions and Long-Term Stock price Performance”, Working Paper, Harvard Business School.

Mulherin, J., and Boone, A. (2000). “Comparing Acquisitions and Divestitures”, Journal

of Corporate Finance 6(2): 117-139.

Penrose, E.T. (1959). The Theory of the Growth of the Firm, Oxford University Press: Oxford, UK.

Pound, J., and Zeckhauser, R.J. (1990). “Clearly Heard on the Street: The Effect of Takeover Rumors on Stock Prices”, The Journal of Business 63 (3): 291-308.

Rau, P.R., and Vermaelen, T. (1998). “Glamour, Value and the Post-Acquisition Performance of Acquiring Firms”, Journal of Financial Economics 49(2): 223-253.

Roll, R. (1986). “The Hubris Hypothesis of Corporate Takeovers”, Journal of Business 59(2): 197-216.

(46)

Seth, A., Song, K.P, and Pettit, R. (2000). “Synergy, Managerialism or Hubris? An Empirical Examination of Motives for Foreign Acquisitions of U.S. Firms”, Journal of

International Business Studies 31(3): 387-405

Shleifer, A. and Vishny, R. (2003). “Stock Market Driven Acquisitions”, Journal of

Financial Economics 70(3): 295-311.

Singh, H. and Montgomery, C. (1987). “Corporate Acquisition Strategies and Economic Performance”, Strategic Management Journal 8(4): 377-386.

Tirole, J. (2006). The Theory of Corporate Finance, Princeton University Press: Princeton, NJ.

Weston, J.F., Mitchell, M.L., and Mulherin, J.H. (2003). Takeovers, Restructuring, and

Corporate Governance, Pearson/Prentice Hall: Upper Saddle River, NJ, 4th edition.

(47)

Appendix

Figure A1. Number of New Rumors and Announcements per Year in Sample

0 10 20 30 40 50 60 2000 2001 2002 2003 2004 2005 Announcement Rumor

Can Rumors Predict Performance?