SECURITY-PRICE PERFORMANCE OF RIVALS AFTER
MERGERS WITHIN THE EUROPEAN UNION
ING. RUBEN BOS
University of Groningen. Department of Business and Economics Final. 7th of June 2012
S2188449 [Words: 3996]
ABSTRACT
This paper presents empirical evidence about the negative effects of mergers on rivals within the European Union, by the use of an event study. A sample of 106 European firms is used to investigated to which extent the effects are negative for the rivals within its economic union. Furthermore the negative effect of mergers within service and manufacturing sectors are investigated and even so the negative effects of vertical and horizontal mergers. This paper shows that mergers within the European Union have no significant negative effects on rivals. A result that contradicts with the evidence for a positive effect on rivals after cross-border mergers.
I. Introduction
destroy market value when both firms experience the negative effects of the merger. The reason why negative effects of mergers are studied so extensively is because the extent of losses are quite substantial or even fatal for the firms.
Mergers within an economic union, like Europe, have not been researched until this paper. The effect of mergers within Europe on the market value of company’s rivals will lead to my research question. What are the effects on rivals when it comes to mergers within Europe?
II. Literature overview
All types of mergers have their own positive and negative effects on rivals. The outcome of a merger on the rival firm’ abnormal returns have a strong relation with the type of merger of the acquiring firm. Several types of mergers can be identified, like horizontal, vertical mergers or cross-border mergers (Clougherty and Duso (2009) and Duso, Guler and Yurtoglu (2010)). Since a negative abnormal return destroys market value, it would be a situation that a manager of the merging firm would avoid for his shareholders. But the companies outside of the merger cannot control the decision of other players in the market. They can only choose be part of a merger or to stay out of it (Halpern (1983)).
Oxley, Sampson and Sliverman (2007) states that cross-border, horizontal and horizontal alliances have a negative effects on domestic rival firms’ abnormal returns. A non-horizontal alliance can have a negative effect on the domestic rival, because the aim of the alliance is to create synergy for the merging firm which increases her turnover at cost of her rivals (see Oxley, Sampson and Silverman (2007) or Halpern (1983)). A horizontal alliance can have a negative effect on the rival firms’ market value, because the merging firm increases it market share at cost of the rivals (Oxley, Sampson and Silverman (2007)). Usually a horizontal merger within the nation’s borders will take away a rival firm and creates more market power for the rivals (Oxley, Sampson and Silverman (2007)). In the case of a cross-border merger the domestic rivals will not benefit from an increase in their market power, because the amount of companies stays the same (Clougherty and Duso (2009)).
laborers cannot easily work in a different country (Lodorfos and Boateng (2006)). Within a economic union, like Europe, laborers are free to work everywhere in the union (Marrewijk (2007)). This enlarged working area might reduce the working cultural differences, which increases the success of mergers simultaneously (Lodorfos and Boateng (2006)). On balance, if the chance of success increase the magnitude of the negative effects on rivals might increase as well. Which makes it interesting to investigate mergers within the European Union.
A. Main hypothesis
For European firms the effects found by Oxley, Sampson and Silverman (2007) might have a different outcome. The main hypothesis in this paper is enlarged to a specific area, Europe.
H1a: There is no significant negative effect for rivals if the merging company creates a strategic alliance within the European Union.
H1b: There is a significant negative effect for rivals if the merging company creates a strategic alliance within the European Union.
B. Sub-hypothesis on sectors
Clougherty and Duso (2009) state that a distinction of a sector can be based on the competition level of the industry. They also state that the service sector is more profitable than in the manufacturing sector. Given that the manufacturing sector in Europe experiences a greater competition from outside of Europe than the services sector (Andrade and Staffords (2004)). The interesting point is if this fact also holds when mergers within Europe are investigated. Since a merger in the service industry seems to be more profitable, it might lead to more negative effects on rivals compared to the manufacturing sector.
H2a: An alliance is not significantly more negative for the rivals in the service sector than the manufacturing sector within the European Union.
C. Sub-hypothesis on merger type
Oxley, Sampson and Sliverman (2007) investigated the effects of non-horizontal and horizontal mergers on rivals. Oxley, Sampson and Sliverman (2007) also investigated the effect of cross-border mergers on rivals. Both had a significant effect on rivals. But the combination of both isn’t investigated yet. In this paper the effects of non-horizontal and horizontal cross-border mergers on rivals will be investigated by the following hypothesis.
H3a: A vertical cross-border alliance is not significantly more negative for the rivals than a horizontal cross-border alliance the European Union.
H3b: A vertical cross-border alliance is significantly more negative for the rivals than a horizontal cross-border alliance the European Union.
III. Data collection and methodology
A. Data
of empirical evidence on the market model varies with the market index used, the time period analyzed, the frequency with which data are observed and the number of companies sampled. The non-trading dates were excluded from the raw data of Datastream® to enable the researcher to infer an estimation window over actual trading days.
B. Methodology
Within the research area of event studies several models can be chosen to infer the abnormal returns to explain the change in market value of the company. Namely, the Mean Return Model and the Market Return Model. Both models are explained and analyzed on their inference power by Brown and Warner (1980). Both models infer the abnormal return from the actual return on the event date minus the mean or expected return from the market. In order to calculate the mean, market Alfa or the market Beta, an estimation window is needed. The estimation window (L1) used in this paper starts at t = -250 until t = -11. Where the event windows (L2 and L3) are analyzed from a Cumulative Abnormal Return (CAR) example. In this paper t = 0 (the event date) is the announcement date of the merger.
1. Mean Return Model
Let μi be the mean return for asset i. Then the constant mean return model is
(1)
Where Rit is the period-t return on security i and ζit is the disturbance, and ζit has expectation of zero and the variance is σ2ζi.
(2)
As i is estimated over the estimation window of -250 to -11 trading days from t=0. Since merger announcements are rumor sensitive it is necessary to cumulate the AR around the event date t =0, on the lengths defined as L2 and L3. To capture the influence before and after
-250 -11 -5 -2 0 2 5
Estimation window = L1 Event window = L2
Alternative event window = L3
the actual announcement on t=0. A total abnormal return is the sum of the abnormal returns within the event window:
(3)
2. Market Return Model
For doing inferences with the Market Adjusted Model the general conditions for the ordinary least squares (OLS) must be satisfied. The market model parameters must be estimated from the estimation window of each stock and correlated stock index on which each rival is listed. Since the sample used in this paper contains companies that are listed on various indices around Europe. The main-index for each country is chosen to be the corresponding market from where the parameters are estimated:
(4) where, (5) and, (6) and for, (7)
In the equations above, and are the returns of the stock i and the market m on time t. In order to calculate to the abnormal return on time t in the event window, the expected value of the market must be subtracted from the return of a stock on time t:
(8)
To draw inferences on the event window the abnormal returns over the event window must be aggregated to get data on each individual stock.
3. Normality check
Before the statistical assumptions of normal or non-normal distribution can be stated the normality of the sample must be tested on the distribution of abnormal returns. Since there is a multiple days event window, an average abnormal return must be used for each firm. The normality is tested by using the Kilimogorov-Smirnov test.
4. Statistical properties of Abnormal Returns
If the abnormal returns are not normal the Corrado-test of Charles Corrado (1989) is used to check for abnormal performance of the rivals stock caused by the announcement of a cross border merger. Since the event has a multiple day time span the notation of the Corrado test is used to rank each actual or abnormal return and is extended by Arnold Cowan (1992) to apply Corrado’s notation for a multiple day period.
Corrado describes a rank test to test for significance on the event period. The ranks will be given to all returns, actual or abnormal, of the estimation window and the event period together. Since the estimation period used in this paper is from 248 days before the event period (L1) and 5 days for the event (L2). The maxim rank will be 253 and the mean rank Ki will be . (10)
Where is the average rank of the actual return of stock n on time t, is the average rank of stock n during the event window, d is the length of the event window in days and represent the Z-value of the test with actual returns.
(11)
The outcome represent the Z-value of the test with abnormal returns.
5. Statistical properties of the location of populations
Both sub-hypothesis are about the location of the two independent populations. One populations lies on the right-hand side or on the left-hand side of the other. The locations of both populations is already specified in both sub-hypothesis. Prove for those locations will be given by executing the multiple day Cowan-test on both populations and both identities. So on mergers in the service and manufacturing sector, as well as horizontal and vertical mergers. The data used for the Cowan test is based on market-adjusted abnormal returns because that data shows the least variance, which implies that the statistical properties are more reliable.
When both populations are considered to be non-normal distributed the statistical properties on the location can be withdrawn from the Mann-Whitney U-test. This test uses ranks to identify the means of both populations.
IV. Data analysis
A. Main hypothesis
For the analysis on the main hypothesis, the descriptive statistics were drawn from the sample based on different adjusting models and cumulative periods. In table I the main statistics can be found. From the results it can be concluded that the sample of 106 items adjusted for the mean will show relatively higher returns compared to the market adjusted model. But for both models the mean is positive but doesn’t deviate from zero significantly as shown in the third row of table I. The standard deviations are relatively large, especially when the result of the CAR’s are observed. The minimum or maximum abnormal returns of some rivals are quite extreme, some events lead to an aggregated increase in market value of 64.03% while another has an aggregated decrease of 14.48% when the returns are corrected by the mean.
Table I. Descriptive statistics of abnormal returns based on the Mean Adjusted Model (1) and the Market
Adjusted Model (2)
Model number (1) (2)
Adjusting model Mean Adjusted Market Adjusted
Event window 5-day AAR 5-day CAR 11-day AAR 11-day CAR 5-day AAR 5-day CAR 11-day AAR 11-day CAR N 106 106 106 106 106 106 106 106 Mean .0014 .0062 .0010 .0105 .0006 .0008 .0002 .0005
One sample t-test .820 .987 1.220 1.220 .618 .284 .735 .176
Asymp. Sig. (1-tailed) .207 .163 .113 .113 .268 .389 .232 .430 Std. dev .0179 .0650 .0081 .0888 .0095 .0291 .0032 .0297 Min -.0612 -.1067 -.0132 -.1448 -.0208 -.1055 -.0150 -.1366 Max .0883 .4416 .0582 .6403 .0758 .1895 .0146 .1384 Non-normality Kolmogorov-Smirnov Z 1.634 1.724 1.411 1.411 2.554 2.389 1.897 1.648 Asymp. Sig. (2-tailed) .010* .005* .037* .037* .000* .000* .001* .009*
Notes: the dependent variables are AAR and CAR (both 5-day and 11-day) for the rivals observations. * represents 5% significance.
For the rivals of the merging firms within the European market the negative z-score of the Cowan-test shows that there are negative returns. But the results in table II show that the ranked returns (R and AR) do not deviate from zero. The magnitude of the negative effect on the 5-day period around the announcement date of the merger is stronger than on the 11-day period, but still not significant. These results were are in contrast with the expectations raised by previous research of Clougherty and Duso (2009). Clougherty and Duso (2009) state that the cross-border merger effect on rivals should have a positive effect, although insignificant.
Table II. Non-parametric test statistics on the Actual Returns (1) and on the Abnormal Returns based on the
Market Adjusted Model (2)
Model number (1) (2)
Returns Actual (R)
Abnormal (AR) Market Adjusted
Event window 5-day 11-day 5-day 11-day
No. of ranks 253 256 253 256
Cowan Z-score -1.100 -.478 -1.344 -.413
Asymp. Sig. (1-tailed) .136 .316 .092 .341
Notes: the dependent variables are R (5-day and 11-day) and AR (5-day and 11 day) for the rivals observations. None are significant.
B. Sub-hypothesis
In this subsection, we check if there is any additional information gained out of grouped rivals on their sectors identity and the merger type they were subjected to.
The second null hypothesis states that there are no differences on abnormal returns between the service and the manufacturing sector. As you can see in the second column of table III, in both sectors the rivals experience a negative influence on their market value. But none of those two are significantly negative. When the populations are compared to each other in order prove if the service sector is indeed more vulnerable to negative influences as Clougherty and Duso (2009) states. The results show that indeed the service sector have a lower mean rank, which implies that the mean abnormal returns is lower than the one of the manufacturing sector, but it is not significantly different.
Table III. Non-parametric test statistics on rivals grouping identity (sectors or merger types)
Hypothesis 2a and 2b 3a and 3b
Grouping identity Sectors Merger type
Populations Service
Manufactu ring
Both Horizontal Vertical Both
Magnitude No. of ranks 253 253 253 253 Cowan Z-score -.720 -1.498 -0.509 -1.833 Sig. (1-tailed) .236 .067 .305 .034* Location Mean rank 51.71 56.45 54.70 51.17 Z-score -.769 -.560
Sig. one tailed (Hxa
> Hxb)
.442 .575
Notes: The symbol * represent significance at 5%.
V. Conclusion
The negative effects of mergers within the European Union on the rivals market value have not been investigated until this paper. This paper studies three issues on this subject.
First, using a sample of large mergers with expertly chosen rival identities and by the event-study methodology, the empirical results show that the mergers within the European borders have abnormal returns that not deviate from zero for the rivals of the merging company.
Second, based on the sample used in this paper it can be assured that the effects on rivals within Europe for the service sector are not more negative than within the manufacturing sector.
VI. Discussion and further research
Some acknowledgements must be made regarding to limitations of this paper. At first, the sample used in this paper is relatively small, 106 rivals. The selection of rivals is based on the information from Datamonitor Group for which the selection procedure is not published. An alternative method to identify rivals could be based on their covariance with the merging firm. Furthermore, there are some limitations on the reliability of the addressed grouping names of service and manufacturing sectors. The used databases have different sector codes. Same comment on reliability holds for the identification of merger types because within the study reports of the European Commission, it is not always quite clear if it was a merger within same sector or with another. Overall limitation on event studies in general is that results on stock prices can be misleading according to Duso et. al (2010). They prove that empirical research based on accounting data is more reliable.
VII. Literature list Periodicals:
Andrade, George, and Erik Stafford, 2004, Investigating the economic role of mergers,
Journal of Finance 10, 1-36.
Brown, Stephan J., and Jerold B, Warner, 1980, Measuring Security Price Management,
Journal of Financial Economics 8, 205-258.
Clougherty, Joseph A., and Thomas Duso, 2009, The Impact of Horizontal Mergers on Rivals: Gains to Being Left Outside a Merger, Journal of Management Studies 46, 1365-1395.
Corrado, Charles J., 1989, A Nonparamatic Test for Abnormal Security-price Performance in Event Studies, Journal of Financial Economics 13, 385-395.
Cowan, Arnold R., 1992, Nonparamatic Event Study Tests, Review of Quantative Finance
and Accounting 2, 343-358.
Drori, Isreal, Amy Wrzesniewski and Shmuel Ellis, 2011, Cultural Clashes in a “merger of equals”: The case of High-Tech Start-ups, Human Resource Management 50, 625-649. Duso, Tomaso, Klaus Gugler, and Burcin Yurtoglu, 2010, Is the event study methodology useful for merger analysis? A comparison of stock market and accounting data, International
Review of Law and Economics 30, 186-192.
Halpern, Paul, 1983, Mergers and acquisitions, Journal of Finance 2, 297-317.
Lodorfos, George, and Agyenim Boateng, 2006, The role of culture in the merger and acquisition process, Management Decision 10, 1405-1421.
Oxley, Joanne E., Rachelle C, Sampson, and Brian S, Silverman, 2009, Arms Race or Détente? How Interfirm Alliance Announcements Change the Stock Market Valuation of Rivals, Management Science 55, 1321-1337.
MacKinlay, Craig A., 1997, Event Studies in Economics and Finance, Journal of Economic
Literature 35, 13-39.
Unal, Haluk, and Edward Kane, 1988, Change in Market Assesments of Deposit-Institution Riskiness, Journal of Financial Services Research 1, 207-229.
Research/Data Sources:
Datamonitor, various years.
Books:
