Competition effects of mergers: An event study of the German electricity market

ARTICLE IN PRESS

Energy Policy 38 (2010) 5264–5271

Contents lists available at ScienceDirect

Energy Policy

0301-42

doi:10.1

n Tel.:

E-m1 Th

with a

domina

VIAG w

journal homepage: www.elsevier.com/locate/enpol

Competition effects of mergers: An event study of the Germanelectricity market

Andreas Keller n

Department of Economics, University of Oldenburg, 26111 Oldenburg, Germany

a r t i c l e i n f o

Article history:

Received 26 February 2009

Accepted 11 May 2010

Keywords:

Electricity market

Event study

Horizontal merger

15/$ - see front matter & 2010 Elsevier Ltd. A

016/j.enpol.2010.05.034

+49 441 798 4445; fax: +49 441 798 4116.

ail address: [email protected]

e German wholesale electricity market is d

combined market share of more than 50%

nce in 1999/2000: E.ON was created out of

hile RWE strengthened its position by mergin

a b s t r a c t

This paper investigates the competition effects of the entry of Vattenfall into the German electricity market.

While the competition authorities supported the entry by approving Vattenfall’s acquisition of three

regional utilities, other market participants raised concerns over the emergence of an upcoming oligopoly in

the German market for power generation. We contrast the efficiency hypothesis postulating pro-

competitive effects of mergers with the market power hypothesis postulating anti-competitive effects. For

the analysis of the two opposing hypotheses, we use an event study approach to the stock prices of

Vattenfall’s competitors in the German market. While we find no empirical evidence for increased market

power in the German electricity market due to Vattenfall’s mergers, there is some indication for efficiency

increases. We therefore cannot oppose the view of the competition authorities predicting an overall positive

effect for consumers as a result of Vattenfall’s entry into the German electricity market.

& 2010 Elsevier Ltd. All rights reserved.

1. Introduction

During the period 1999–2002, the Swedish company Vattenfallentered the German electricity market by acquiring three regionalutility companies (HEW, VEAG and Bewag). The German competi-tion authority (Bundeskartellamt) and the European Commissionboth approved without any conditions the acquisitions and thesubsequent merger of the three formerly separate German utilities.The Bundeskartellamt expressed the view that Vattenfall as afourth player in the German electricity market (in addition to E.ON,RWE and EnBW) would be able to compete in prices against theexisting big players.1 However, the question whether the formationof Vattenfall in the German market indeed resulted in lower pricesfor consumers, or whether it brought about higher prices byreducing the number of players and thereby increasing the marketpower of the new oligopoly, is a controversial issue today.

Against the background of rising electricity prices, severalempirical studies recently analysed whether power generation inGermany is a competitive market (e.g. Schwarz et al., 2007; Weigtand von Hirschhausen, 2008; Ockenfels, 2007). Although there issome indication that the large utilities possess market power, it iscontroversial whether prices charged for power generation areindeed set above the competitive levels. Given this dissent and thedifficulties of estimating price-cost margins due to limited data

ll rights reserved.

ominated by E.ON and RWE

. These two players gained

a merger between VEBA and

g with VEW.

availability, it is a challenging task to assess whether the entry ofVattenfall into the German market had a pro-competitive or an anti-competitive effect. We use an alternative approach in tackling thisquestion by applying an event study, which analyses the reaction ofthe stock market to the merger announcements by Vattenfall. Weconsider the three cases in which Vattenfall acquired regional utilitycompanies, and identify eight event dates on which important newsabout the takeovers was released. Following Eckbo (1983) andStillman (1983), we expect that an analysis of the stock returns ofthe merging parties’ competitors will reveal the competition effectsof the merger. Our results support the view of the competitionauthorities, since we find no indication that the formation ofVattenfall had anti-competitive consequences.

This paper is organised as follows: Section 2 presents thebackground of the liberalisation process in the German electricitymarket. Section 3 describes the competition effects of horizontalmergers. Section 4 derives our methodology and the data used tocreate the event study. Section 5 presents the results and relatesthem to the debate on efficiency gains versus market power.Section 6 concludes.

2. Liberalisation of the electricity market in Europe

2.1. Liberalisation and market power

The electricity sector can be subdivided into four verticallylinked stages: generation, transmission, distribution and retailing.While stages two and three represent classical network industriesand will continue to be regulated throughout Europe, competition

www.elsevier.com/locate/enpol

dx.doi.org/10.1016/j.enpol.2010.05.034

mailto:[email protected]

ARTICLE IN PRESS

3 For an overview of the methods to detect market power in electricity

markets see Twomey et al. (2004). A comprehensive application to six European

wholesale electricity markets including Germany is given in London Economics

(2007).4 These thresholds are: 33% market share of the biggest supplier, 50% market

share of the three biggest suppliers together and 66% market share of the five

A. Keller / Energy Policy 38 (2010) 5264–5271 5265

has been introduced at both the generation and the retail stage.The European Commission started the EU-wide policy of liberal-isation in 1996 with Directive 96/92/EC and speeded up theprocess with Directive 2003/54/EC. The economic rationales forliberalisation are expectations for lower prices and higherefficiency in the electricity industry, which would lead to aneconomic surplus through direct and indirect consumer priceeffects (see e.g. Martin et al. 2005). Although the potentialbenefits of liberalisation are well acknowledged, an oligopolisticmarket structure with only few suppliers can represent a majorobstacle. This is especially the case for the stage of powergeneration, which we will examine in more detail.

For an initial analysis of the effects of an oligopolistic marketstructure on prices, we employ the standard Cournot–Nashequilibrium. Although its application to electricity markets iscontroversial and alternative concepts like supply functionequilibrium (SFE) exist, the Cournot framework provides areasonable approach and is still widely used in electricity marketmodelling due to its advantages in computational manageability(Ventosa et al., 2005).2 Proceeding from the Cournot oligopoly, theLerner index for firm i, measuring how strong price exceedsmarginal cost is given by

p�C0ip¼

si

Z, ð1Þ

where p is the price of power in the wholesale market, Ci0 is the

marginal cost of the firm i, si is the share of firm i in total outputand Z is the price elasticity of demand. It is obvious from (1) thatthe margin over marginal cost is higher when the market share ofa company is high and when the price elasticity of demand is low.An aggregate Lerner index over all firms weighted by marketshares can be related to the Herfindahl–Hirschmann Index (HHI)of market concentration (e.g. Vives, 1999)

Xn

i ¼ 1

si

p�C 0ip

� �¼

HHI

Z with HHI¼Xn

i ¼ 1

s2i : ð2Þ

Thus, an increase in concentration in the market, measured bythe HHI, pushes the price further away from the marginal cost.The price elasticity of demand is typically low in the electricitymarket, increasing the potential to raise price over marginal cost.This tendency is aggravated by the absence of real-time billing toend-users, which causes a severe lack of demand elasticity in theshort-run (Stoft, 2002). Another issue, which can potentiallystrengthen market power in generation, is its close relation withtransmission and distribution networks that connect generationand final demand. A transmission service operator (TSO) managestransmission and allocates capacity to generators. If a generatorowns the network or especially the TSO, the integrated companyhas the possibility to discriminate against other generators.Furthermore, it has an incentive to reallocate profits fromgeneration to transmission in order to keep margins at thegeneration level low and deter new entrants.

In general, two issues can restrict the potential for exercisingmarket power: competition from abroad via imports and theexistence of contestable markets. In power supply, competitionfrom abroad or outside specific regions is typically insufficientdue to regional networks and limited capacity of interconnec-tions. At the European level, the relevant geographic markets forelectricity are still of limited size, often coinciding with nationalborders. This is due to few and congested interconnections at theborders of the national states. The theory of contestable marketsclaims that even if firms possess large market shares and thus

2 For a discussion on Cournot versus supply functions in electricity markets

see Willems et al. (2009).

market power, they may be deterred from setting price abovemarginal cost because this would attract new entrants into themarket. In power generation, however, new entrants are typicallydiscouraged by large sunk costs and a complicated approvalprocedure so that incumbents are able to set prices higher thanmarginal costs up to a certain extent.

We conclude from this first examination that in the case ofmarket power, the prediction of falling prices due to liberalisationmight turn out premature. Newbery (2007) claims that electricityrestructuring in Europe – as opposed to the US – has tended tooverlook issues of market power. This may also apply to thespecific case of Germany since several mergers after liberalisationled to an increase in the concentration of the market.

2.2. Market power in the German electricity market

Several indicators and techniques can be used to analysemarket power in the German electricity market.3 The marketshares reveal that the largest producers exceed the shares definedby German antitrust law as thresholds for the supposition ofmarket power.4 As shown in Fig. 1, the five largest producers inGermany, among them Vattenfall, have a cumulative sharein generation capacity of about 68% and in actual powergeneration of about 87% (Zimmer et al., 2007).

Similarly, the Herfindahl–Hirschmann Index (HHI) of 1840points indicates high concentration (Zimmer et al., 2007).5 Whenaccounting for criticism that the conventional measures of marketpower underestimate the potential market power in the elec-tricity market due to inelastic demand (e.g. Stoft, 2002), thesenumbers cause even more concern. The Residual Supply Index(RSI) measures the potential market power of a single company. Itdetermines if and for how long the capacity of one producer isessential to satisfy the final demand, thus if the company is apivotal supplier. Schwarz et al. (2007) calculate the RSI values forthe two biggest German companies RWE and E.ON and find a highpotential for the exertion of market power.

While the previous indicators assess only potential marketpower, a different approach uses a simulation model of marginalcosts to calculate the price-cost margins as an indicator for theactual exertion of market power. Schwarz and Lang (2006) havecalculated such a model for the German wholesale electricitymarket in the period 2000–2005 and come to the conclusion thatwhile in 2000 marginal cost pricing could be assumed, prices roseabove marginal costs in 2003 and thereafter. Weigt and vonHirschhausen (2008) find in their analysis that wholesale marketprices in 2006 were on average 12% higher than predicted by thecompetitive benchmark model. Though appealing, studies simu-lating price-cost margins of power generators are criticised fordifficulties in identifying the true cost structures and in copingwith the dynamics on electricity markets (Ockenfels, 2007).

From the evidence presented for the German market, it can besupposed that the major utilities, especially E.ON and RWE,possess considerable market power. There is also some indicationthat they at least partially exert this market power to set pricesabove the marginal costs. However, while evidence for market

biggest suppliers together (Gesetz gegen Wettbewerbsbeschrankungen y19, 3).5 While the German authorities mention no explicit thresholds for the HHI, the

US Department of Justice considers a market to be highly concentrated for a HHI of

more than 1,800 points.

ARTICLE IN PRESS

22.1% 26.5%

18.9%26.5%

12.6%

16.9%8.0%

10.3%

6.1%

6.5%32.3%

13.3%

0%

10%

20%

30%

40%

50%

60%

70%

80%

90%

100%

noitcudorPyticapaC

Other suppliersSTEAGEnBWVattenfallRWEE.ON

Fig. 1. Market shares in the German wholesale electricity market 2006.

Source: Zimmer et al., 2007.

A. Keller / Energy Policy 38 (2010) 5264–52715266

power can especially be found for the two largest players, E-ONand RWE (as for example indicated in the RSI analysis), it isunresolved whether the formation of Vattenfall, which wasactively supported by the competition authorities, enhancedcompetition through the presence of a large and efficientcompetitor or, in contrast, harmed competition through theincreased market power of the oligopoly. Before approaching thisquestion, we first report on the history of Vattenfall’s entry intothe German market.

2.3. The entry of Vattenfall into the German electricity market

Shortly after the German electricity market was liberalised in1998, the market experienced major changes. New electricityretailers and foreign companies entered the market and retailprices declined strongly, especially for industrial customers (seee.g. Brunekreeft and Keller, 2000). However, the German incum-bents reacted soon and a merger wave changed the industrystructure significantly. Especially remarkable are two largemerger cases that were announced in autumn 1999: theacquisition of VEW by RWE and the merger between VEBA andVIAG to create E.ON. After these transactions, the two majorutilities controlled more than half of the market, raising seriousconcerns about the exertion of market power. Nevertheless, theEuropean Commission (in the case of VEBA/VIAG) and the GermanBundeskartellamt (in the case of RWE/VEW) both approved thedeals in 2000 conditioned to remedies, which included the sale ofthe companies’ stakes in the three regional utilities HEW, VEAGand Bewag. Although other foreign companies were also inter-ested in acquiring one of the regional utilities, Vattenfall wassuccessful in creating its subsidiary Vattenfall Europe during1999–2002 by taking over all three utilities. The acquisitions tookplace according to the events described in the followingparagraphs; for a more in-depth report on Vattenfall’s expansioninto Germany see Hogselius (2009).

The entry of Vattenfall into the German market for electricitygeneration started with the takeover of the HamburgerElectricitats-Werke (HEW). After an announcement by the cityof Hamburg to sell 25.1% of its utility, Vattenfall declared itsinterest in September 1999 and acquired the stake in November.In October 2000, Vattenfall announced that it would acquireanother 37.2% in HEW from E.ON and Sydkraft and 7.8% from

other investors. Vattenfall finalised the deal with the acquisitionof the remaining 25.1% from the city of Hamburg in June 2002.

The second acquisition by Vattenfall was the East German utilityVEAG and its lignite supplier LAUBAG. In April 2000, when it becameclear that RWE and E.ON would have to sell their stakes in VEAG dueto obligations imposed by the competition authorities, HEW withsupport by Vattenfall confirmed its interest in the company. Otherpotential investors appeared as well but in December 2000 it wasannounced that HEW/Vattenfall would acquire VEAG.

The third company that was later integrated into VattenfallEurope was the Berlin utility Bewag. In August 2000, E.ONannounced that it would sell its 49% stake in Bewag to Vattenfalland not, as expected from many market participants, to Mirant (atthat time called the Southern Energy) that already owned a stake inBewag. However, Mirant blocked this deal with a veto one day laterand afterwards quarrelled with Vattenfall over the control of Bewag.During 2001, both companies controlled Bewag collectively, eachowning a 45% stake, but in December 2001, Mirant finallyannounced to sell its stake in Bewag to Vattenfall. The last dealcleared the way for fully integrating the three formerly separateregional utilities into a single large player in the North and East ofGermany in 2002, named Vattenfall Europe. Both the Bundeskartel-lamt and the European Commission raised no objections against thecreation of the new large utility in the German market.

3. Competition effects of horizontal mergers

In this section, we review the possible effects of horizontalmergers on competition and how event studies are used to test forthese effects empirically. We treat the mergers of Vattenfall withregional German utilities as horizontal mergers because allcompanies involved (HEW, VEAG and Bewag) as well as the largecompetitors E.ON and RWE were active in all stages of electricitysupply. None of the mergers were an explicit vertical integrationand furthermore, our focus lies on the stage of generation, whichtriggers the largest competition concerns.

3.1. Efficiency gains versus increase in market power

In our study, we analyse the competition effects of a horizontalmerger by examining the effects of the merger on other competitors

ARTICLE IN PRESS


in the market. The idea behind it is that a merger that harmscompetitors tends to be pro-competitive and a merger that benefitscompetitors tends to be anti-competitive. Using this insight, bothEckbo (1983) and Stillman (1983) started to analyse changes instock prices of competitors at the announcement date of a merger,an event study, to test whether there is a pro- or anti-competitiveeffect. The theoretical underpinnings for the competitive effects ofhorizontal mergers on rivals’ present discounted value, as measuredby stock prices, are changes in efficiency and market power thatinfluence product prices.

3.1.1. Efficiency hypothesis

Increased efficiency due to a merger that leads to the reductionof average costs can put downward pressure on product prices.According to the Cournot oligopoly model of horizontal mergersby Farrell and Shapiro (1990), this is possible for mergers witheconomies of scale and for mergers with learning, i.e. whenknowledge or technology is transferred within the new mergedentity.6 The efficiency hypothesis predicts that the stock prices ofrival firms decrease since the market expects that the mergedcompany will produce more efficient in the future (Eckbo, 1983).

3.1.2. Market power hypothesis

Under the market power hypothesis, events that make themerger more likely will cause rivals’ share prices to rise, reflectingincreases in expected future cash flows. In the Cournot model, thenew merged entity with an increased market share will typicallyrestrict its combined output. As shown in Eq. (2), a higherconcentration of market shares at a single firm leads, atunchanged costs, to an increase in the equilibrium price.7 Thismeans that all competitors of the merging firms will also benefit.As Duso et al. (2007, p. 462) mention, the ‘‘change in the value ofcompetitors’ equity can then be taken as a measure of the(discounted) additional profits that are expected to accrue tothem as a consequence of the merger.’’ It is then obvious thatevents that make a proposed merger less likely will be associatedwith a share price decline for rivals (Eckbo, 1983).

3.1.3. Consequences

To evaluate the consequences of a merger for consumer pricesand thus consumer surplus, it is necessary to check which of thetwo hypotheses dominates. Schumann (1993, p. 680f.) points outthat ‘‘whether a merger that creates efficiencies results in higheror lower prices depends on how large those efficiencies are and towhat extent increased concentration may affect how the firms inthe industry interact to determine market prices.’’ Combining thisinsight with the consequences of a merger for the stock prices ofnon-merging competitors, positive stock returns to competitorswould harm consumers while negative stock returns wouldbenefit them (cf. Duso et al., 2007).

6 For an overview of empirical studies estimating economies of scale in power

generation see Ramos-Real (2005). Unfortunately, there are almost no recent

studies on this issue and furthermore, the potential effects of learning due to a

utility merger are rather unexplored. The pioneering study by Christensen and

Greene (1976) estimates a minimum efficient scale of about 20 TWh of electricity

generation for U.S. utilities in 1970. Looking at the firms in our study, HEW and

Bewag were both operating below this threshold.7 The Herfindahl–Hirschmann Index of market concentration used in Eq. (2)

always rises in consequence of a merger in the relevant market since the squared

market share of the merged entity is higher than the sum of squares of the market

shares of the individual entities.

3.2. Criticism of event studies in merger cases

Right after the time event studies were first employed toanalyse competition effects of horizontal mergers, the methodol-ogy has also been criticised (e.g. by McAfee and Williams, 1988;Fridolfsson and Stennek, 2006). This criticism is often related toproblems arising from studies that cover a large number ofmergers in different industries and over a long time horizon,where it is not possible to consider merger specific information. Incontrast to these large-sample studies, we focus in our analysis onthree mergers in the German electricity market and are able toinclude case-specific information.

The criticism by McAfee and Williams (1988) focuses on theproblem of multi-product companies. A merger in an area whereonly one of the various products of a competitor is involved mightnot have a significant influence on its share price. This problem isnot relevant in our analysis since electricity supply is a corebusiness for the utilities we consider and furthermore, electricityis a homogenous good without product differentiation. Therefore,a horizontal merger will have a direct influence on thecompetitors in the market and their share prices.

Another interpretation of stock price movements of competi-tors was given by Eckbo (1983) and Eckbo and Wier (1985): thenews of a merger might not only imply efficiency gains for themerging parties but also spread the information to competitorsthat they are likewise able to benefit from efficiency gains infuture mergers. This so-called in-play effect could imply positiveabnormal returns to the rivals of merging firms, which are notconsistent with the market power hypothesis. In our case,however, the large competitors E.ON and RWE were already inthe process of consolidation themselves, so the merger announce-ments by Vattenfall provided no new information regardingfuture efficiency gains through mergers. Moreover, it was clearthat the competition authorities would take up a very rigid stanceon any further horizontal merger proposal by E.ON or RWE.

As pointed out by Fridolfsson and Stennek (2006, p. 26), ‘‘eventstudies are sensitive to the exact anticipations and uncertaintiespresent in the stock market prior to the merger’’. This is the casewhen existing rumours about a merger, which are alreadyincorporated in the stock prices, lead to a different interpretationof the actual merger announcement than if it would have come as acomplete surprise. As an example, suppose that a rumour of amerger already exists, but the allocation of roles regarding themerging parties is still uncertain. Then the announcement of thespecific merger can lead to an out-of-play effect for the competitorswho are not involved in the merger. The stock prices would declinebecause market expectations that the respective company isinvolved in the merger are now disappointed. It is necessary toconsider this out-of-play effect and therefore, detailed informationabout each event is needed. In our study, we include rumour datesas well as announcement dates of a merger. Furthermore, the out-of-play effect has no influence in the cases of VEAG and Bewag becauseit was clear that neither E.ON nor RWE would be allowed to controlthose companies due to the obligations of the competitionauthorities who actually required them to divest their stakes. Onlyin the case of HEW might have existed an out-of-play effect for thecompetitors when Vattenfall started to bid for HEW. We willconsider this possibility in the interpretation of these results.

4. Methodology

In this section, we present the data selection and methodologyof our event study. The application of an event study is based onthe validity of the semi-strong form of the efficient markethypothesis, i.e. that all new publicly available information is

ARTICLE IN PRESS

Table 1Event dates for the entry of Vattenfall into the German electricity market.

No Date Event

Case 1: Vattenfall—HEW1a 29-September-99 Vattenfall announces interest in 25.1% of HEW

1b 16-November-99 Vattenfall�s bid is accepted and it buys 25.1% in

HEW

1c 19-October-00 Vattenfall acquires 45% in HEW from E.ON,

Sydkraft and others

1d 11-June-02 Vattenfall buys the remaining 25.1% in HEW

from the city of Hamburg

Case 2: Vattenfall—VEAG2a 18-April-00 HEW, supported by Vattenfall, declares interest

in VEAG

2b 12-December-00 HEW/Vattenfall will take over VEAG from RWE

and E.ON

Case 3: Vattenfall—Bewag3a 09-August-00 HEW/Vattenfall announces the acquisition of

49% in Bewag from E.ON (one day later Mirant,

another shareholder of Bewag, blocks the deal)

3b 03-December-01 Mirant sells stake in Bewag to Vattenfall


reflected by corresponding changes in stock prices (see Fama,1970). Furthermore, it is assumed that stock prices represent anunbiased estimate of a firm’s future profits.8

Applying the theoretical considerations from Section 3.2, weuse changes in expected future profits of competitors – which arereflected by changes in stock prices at the announcement date of amerger – as indicator for the competition effects of the respectivemerger. Thereby, we do not attempt to quantify the changes inexpected future profits but are only interested in the sign: anegative sign indicates a pro-competitive effect of the proposedmerger while a positive sign indicates an anti-competitive effect.Regarding the entry of Vattenfall into the German market, weidentify three major cases with altogether eight event dates. Theywere obtained by screening the financial press (via the FACTIVAdatabase) for any news about the mergers between Vattenfall andHEW, VEAG or Bewag. A date was identified as an event date if therelease by a news agency contained the information that Vatten-fall either publicly announced its interest in, or was already in theprocess of, acquiring a stake in one of the other three utilities. Theapproach of looking in detail on each merger case and identifyingseveral event dates differentiates our study from many otherevent studies that only employ the announcement day of amerger and thus lose important information in many mergercases. Each event date we examine is presented in Table 1together with a short description.

In case an event took place on a weekend, we employ thefollowing trading day as the event day. We identify E.ON9 andRWE as relevant competitors since they are the two largestplayers in the German market. In addition, the share prices of

8 For the case of stock returns evolving according to a fair game, Samuelson

(1973) showed that stock prices equal the expected present value of future dividends.

A fair game is given when the expected return of the stock – under consideration of

all publicly available information – less a constant is equal to zero. The theorem of

Samuelson is important since, while it is difficult to show that the price of a security is

directly related to the expectations of future dividends, it is easier to analyze

empirically whether the expected return of any security is indeed zero (or more

precise, equal to a constant average return). Although traders and investors typically

try hard to beat the market, it is difficult to find any who consistently outperform the

market. For an extensive review on capital market efficiency and the development of

its theoretical underpinnings and empirical findings see LeRoy (1989).9 E.ON was officially created in June 2000 out of VEBA and VIAG but merger

talks started already in July 1999 and were officially confirmed in the beginning of

September 1999. Since E.ON shares were not yet traded at our first event dates, we

then use the share price of VEBA, which brought the majority of capital into the

new company. VEBA shares were renamed as E.ON in June 2000.

E.ON and RWE are well suited for an event study since theyexhibit frequent trading activities and thus react instantaneouslyto new information.

The measurement of the event-induced changes in stock pricesis based on the market model of the event study methodology aspresented in MacKinlay (1997). This model is frequently used inevent studies (for recent applications see Aktas et al., 2007;Clougherty and Duso, 2009) since it provides the basis for areasonable prediction of the normal return of a specific security,given the return of the market portfolio. The only difference hereis that we augment the market model by controlling for the oilprice, since its changes are likely to affect the electricity stocksdifferently than the market index. This assertion is comfirmed byOberndorfer (2009) who finds that the stock returns of Europeanelectricity corporations are negatively affected by a rising oilprice. Therefore, to estimate normal performance, the daily stockreturns of the examined company (adjusted for dividendpayments) are regressed on a constant, the market return andthe return of the oil price during an estimation window that startsand ends before an event:

rt ¼ aþbrDAX,tþgrOIL,tþet : ð3Þ

For calculating the market return we employ the DAX, which isthe German primary index measuring the performance of the 30largest Prime Standard companies. For calculating the oil return,we use the spot price of Brent, the North Sea crude oil benchmark.The residuals et are assumed to be normally distributed with theproperties

EðetÞ ¼ 0, ð4Þ

VARðetÞ ¼1

L�2

XT1

t ¼ T0

ðetÞ2, ð5Þ

where L is the length of the estimation window and T0 and T1 arethe first day and the last day of the estimation window,respectively. Our estimation window includes 90 trading days:it starts 100 days and ends 11 days before each event. This gives areasonably large estimation window while ensuring that thepotential information leakage before the event itself has noimpact on the regression. In order to further improve the qualityof the regressions within the estimation window and obtain morereliable predictions for normal performance, an in-depth checkwas performed: first, for each regression in an estimationwindow, we statistically identified those dates at which theabsolute value of the error term exceeded two-times the standarddeviation of the respective regression. Then, for the datesidentified we checked the financial press for important extra-ordinary news about the respective company (RWE or E.ON). Ifboth the statistical test and the inspection of the financial pressidentify an extraordinary event that unambiguously caused theabnormal return, the respective date was dropped from theestimation of normal performance (with this procedure, up to fivedates were dropped in a regression). This procedure is valid sinceit is the aim of the regressions to estimate the normalperformance of the stock without consideration of any extra-ordinary events. The coefficients a, b and g, which are estimatedby Eq. (3), can then be used to calculate the abnormal return ar foreach event day t:

art ¼ rt�a�brDAX,t�grOIL,t : ð6Þ

This means that the abnormal return is calculated as thedifference between actual return rt and the return predicted bythe market model. As illustrated in Fig. 2, we employ an eventwindow that ranges from two trading days before to four tradingdays after the event day and calculate abnormal returns for eachof the seven days. This extension is useful to incorporate insider

ARTICLE IN PRESS

-100 -11 -2 0 4

Estimation Window Event Window

Event

t

Fig. 2. Time path around an event day.

Table 2HEW-Case, 7-day cumulative abnormal returns in percentage points.

Event E.ON RWE Portfolio Evaluation

1a �7.29 (�1.61) �7.52n (�1.80) �7.21n (�2.06) Pro-competitive1b 1.44 (0.28) 4.04 (0.94) 2.90 (0.78) –1c �3.46 (�0.84) �0.46 (�0.14) �2.05 (�0.75) –1d 4.55n (1.96) 2.34 (1.04) 3.36n (1.71) (Anti-competitive)

Note: t-statistics in parentheses.

n Denotes significance at the 10% level.

Table 3VEAG-Case, 7-day cumulative abnormal returns in percentage points.


2a �0.89 (�0.11) 1.08 (0.16) 0.09 (0.01) –2b �3.67 (�0.92) �10.31n (�2.81) �6.99n (�2.41) (Pro-competitive)




knowledge that may influence the stock return prior to theannouncement of the event as well as to incorporate delays in thereaction of the market due to uncertainties. At the same time, it isnecessary to assure that there is no overlap of the event windowwith other events, which could distort the reaction of the shareprice. We checked the financial press and found two overlaps atevents 1d and 2b, which we will consider in the interpretation ofthe results. As an additional robustness check, we alternativelyreduce the event windows to four days, ranging from one daybefore to two days after the event.

In order to test whether an event has a statistically significantinfluence on stock returns, we follow the strategy proposed byMacKinlay (1997). Under the null hypothesis of no influence ofthe event on stock prices, the abnormal returns on the days duringan event window are jointly normally distributed with zeroconditional mean and conditional variance VAR(art). As discussedin MacKinlay (1997), under H0, the conditional variance can beapproximated by the disturbance variance from the estimation ofthe market model VAR(et), as expressed in (5). This approximationis valid as long as the estimation window is large, which is thecase in our study. Next, the abnormal returns of each day duringthe event window are aggregated to draw an overall inference forthe respective event. The variance has to be adjusted by multi-plication with the number of event days. From the cumulativeabnormal return and the variance, we calculate the test statisticsto check whether the abnormal returns of a company’s stockduring the event window are significantly different from zero.

For a second test, we construct a portfolio equally weighting thedaily returns of E.ON and RWE and repeat the above described testingprocedure for each event. Hence, we get a combined assessment ofthe stock price reactions of both companies. The resulting teststatistics verifies if the cumulative abnormal return of the portfolioduring the event window is significantly different from zero.

5. Results and analysis

For our analysis, we successively investigate the three mergercases with their respective event dates as listed in Table 1.Thereby, we evaluate a merger announcement as pro-competitivewhen the abnormal returns of competitors are significantlynegative and as anti-competitive when the abnormal returns aresignificantly positive. The latter would imply that the competitionauthorities were probably wrong when they expected no forma-tion of a dominant oligopoly due to these mergers (see mergerdecision by Bundeskartellamt, 2001).

5.1. Case 1: Vattenfall—HEW

On event date 1a, Vattenfall announced its interest to buy25.1% in HEW, which were offered by the city of Hamburg(Table 2). The cumulative abnormal return of the portfolio of E.ONand RWE during the event window was significantly negative andamounted to more than 7%. When using the shorter eventwindow of 4 days, this result is even more significant (seeAppendix, Table A1). The negative returns might be explainedwith a pro-competitive effect, caused by the possible creation of amore efficient competitor in the German electricity market. Still it

is important to point out that also an out-of-play effect could haveinfluenced the market reaction since the possibility that E.ON orRWE would be able to acquire the stake in HEW decreasedsignificantly. However, this explanation is rather unlikely becauseat the time of this event, it already became clear that neither E.ONnor RWE would have good prospects in acquiring the majority inHEW due to substantial concerns of the competition authorities.

Actually, if the out-of-play effect was important, it should haveplayed a role at the second event when Vattenfall’s bid wasaccepted because then, E.ON and RWE were definitely out of play.However, there is no significant result at this event and it isneither possible to identify an out-of-play effect nor a competitioneffect empirically in this case.

At event 1c, when Vattenfall obtained the majority in HEW withfurther acquisitions, there is a negative but statistically insignif-icant tendency of the share prices of E.ON and RWE. One possibleexplanation could be that the market had already anticipated theseacquisitions and that the change in expected future profits due tothis event was already included in the share price.

Event date 1d, when Vattenfall finalised the acquisition of HEWand took over the whole company, delivers a robust positiveabnormal return for E.ON, which would indicate an anti-competitiveeffect of this event. However, there is an overlap of the event withE.ON’s takeover of the British company Powergen, which wasenabled on June 12 by a decision of the U.S. Securities and ExchangeCommission (SEC) approving the takeover of a U.S. affiliate ofPowergen. We suppose that the positive abnormal return for E.ON israther caused by this decision, which was important for E.ON. Thisconjecture is confirmed by the stock price of RWE, which shows nosignificant abnormal return. Furthermore, the new information atevent 1d was rather limited since it was expected before thatVattenfall would finalize the takeover of HEW some day.

In general, the expectations of the stock market in the HEW-case are in line with the assessment by the competitionauthorities that expected no anti-competitive effects from theentry of Vattenfall into the German electricity market.

5.2. Case 2: Vattenfall—VEAG

When HEW/Vattenfall announced interest in acquiring VEAGon event date 2a, it is not possible to identify any significantabnormal behaviour of the stocks of E.ON and RWE (Table 3). Apossible explanation could be that many companies besidesVattenfall declared interest in VEAG. It should be noted that the

ARTICLE IN PRESS

Table 4Bewag-Case, 7-day cumulative abnormal returns in percentage points.


3a �4.46 (�0.91) 3.80 (0.94) �0.38 (�0.12) –

3b �7.00n (�1.92) �6.15n (�1.69) �6.49n (�2.00) Pro-competitive




regressions of normal return that were performed within theestimation window before event 2a are characterized by a poor fitand insignificant beta coefficients (the only regressions withinsignificant beta coefficients). This problem is probably causedby the stock market impacts of the boom and bust of the dot-combubble during that time. However, even with a better fit of theestimation of normal performance, abnormal returns in this casewould be too low to give statistically significant results.

On event date 2b, when HEW/Vattenfall took over VEAG byacquiring the shares from E.ON and RWE, which had to divesttheir stake as a precondition for their own merger interests, weidentify a significant negative abnormal return to RWE. In thiscase, however, the negative return might have been caused by theoverlapping announcement of RWE one day prior to the eventdate to sell its chemical division in order to focus on corebusiness. Nonetheless, according to the financial press, theaccomplishment of this deal as well as the price of sale was asexpected. The stock price of E.ON shows a negative butstatistically insignificant abnormal return. From these findings,we infer a slightly negative tendency of the stock returns ofVattenfall’s competitors in the German market.

5.3. Case 3: Vattenfall—Bewag

Event date 3a is closely connected to the VEAG-Case (Table 4).Mirant (at that time called the Southern Energy), a shareholder ofBewag, was also interested in taking over VEAG and was one ofHEW/Vattenfall’s competitors in the acquisition. At event date 3a,HEW/Vattenfall announced to buy 49% of the shares of Bewagfrom E.ON. The estimation result shows no significant abnormalstock market reaction of the relevant competitors. However, it islikely that the formal objection by Mirant against this acquisitionright after its announcement is an explanation why no significantabnormal returns to rivals occurred.

A significant reaction followed at event date 3b, when Mirantsold its shares in Bewag to Vattenfall after the two companiesquarrelled for more than one year over the control of Bewag. Thedecision by Mirant to sell its shares to Vattenfall is probably themost important event of the eight dates identified in this studybecause it cleared the way for the creation of the new playerVattenfall Europe. Only now, Vattenfall was able to completelyintegrate all three formerly separate utilities into one singlecompany. For this event, we find significantly negative abnormalreturns to the rivals of around 7% during the 7-day event window.This negative reaction is consistent with the assessment of theBundeskartellamt that the final integration of Vattenfall into asingle company in the German electricity market had pro-competitive effects and harmed the competitors.

6. Conclusion

In this paper, we analyse the competition effects resulting fromthe acquisition of three German regional utilities by the Swedishutility company Vattenfall. We first discuss two contrasting viewswith regard to horizontal mergers: the efficiency hypothesis, which

postulates the pro-competitive effects due to economies of scaleand learning, and the market power hypothesis, which postulatesanti-competitive effects due to increased concentration. Based onthe overall influence of these two effects on competitors of themerging firms and the validity of the semi-strong form of theefficient market hypothesis, it is possible to infer the consequencesof a merger on consumer welfare by means of an event study. Wespecifically assess whether the German competition authority(Bundeskartellamt) was right when it approved without anyobjections the consolidation of the three formerly separate regionalutilities into one single company headed by Vattenfall.

While we could not find any evidence that the increase in marketpower benefited the large competitors E.ON and RWE, there is someindication that potential increases in the efficiency of the new playerVattenfall Europe harmed the competitors. The expectations of thestock market were largely consistent with the assessment by theBundeskartellamt and none of the events could support the fears ofan anti-competitive effect due to the entry of Vattenfall. It istherefore not possible to oppose the view of the competitionauthorities who predicted an overall positive effect for consumers.

While the event study methodology for evaluating competitioneffects of mergers was rightly criticised after its emergence in the1980s, our case study approach to specific mergers in the Germanelectricity sector can largely resolve the criticism: First, in the case ofmulti-product companies and product heterogeneity, it would notalways be obvious that a merger actually affects the competitors.However, for the utilities we consider, electricity supply is a corebusiness and furthermore a market for a homogenous good. Next, it ispossible that an event contains no new information or that existinginformation biases the interpretation of an event by the market.Therefore, in our analysis, we take rumours and potential in-play andout-of-play effects into account for each event. Finally, if an eventwindow overlapped with other events that influence the examinedcompanies, abnormal returns might be wrongly attributed to amerger announcement. We check for overlapping events in thefinancial press and consider them in our interpretation.

There is one potential caveat to this study that has not yet beenaddressed. It is possible that the announcements, reports andinterviews of the competition authorities influenced stock marketexpectations about the competition effects of the mergers. Sincethe Bundeskartellamt explicitly welcomed the formation ofVattenfall Europe, it might have influenced the market with thisview and created a bias in expectations about future profits. Thisinfluence would be especially relevant if the competitionauthority had a high credibility in the public.

Our analysis also has a policy implication regarding mergers inthe electricity sector. Although market power is an importantissue, our analysis suggests that it should not be considered as theonly one. Besides the increased market power resulting frommergers, there is also the possibility that the merging companiesbecome more efficient. This argument might be especially relevantif small regional utilities merge to a larger entity as in the caseof Vattenfall Europe. Therefore, competition authorities shouldnot a priori rule out the efficiency defence in the assessmentof mergers. Concerning this matter, it would furthermore bevery helpful if future research on a firm-level basis would beable to quantify the efficiency gains of mergers in the electricitymarket.

Acknowledgements

I would like to thank Tim Steffens for valuable research assistance.Helpful comments and suggestions by Christoph Bohringer, Edwinvan der Werf and Henning Klodt are gratefully acknowledged.

ARTICLE IN PRESS

Table A1HEW-Case, 4-day cumulative abnormal returns in percentage points.


1a �9.08n (�2.65) �3.86 (�1.22) �6.41n (�2.42) Pro-competitive1b �4.66 (�1.19) 2.11 (0.65) �1.15 (�0.41) –1c �2.25 (�0.72) �3.87 (�1.49) �3.10 (�1.51) –1d 4.50n (2.57) 0.85 (0.50) 2.62n (1.77) (Anti-competitive)



Table A2VEAG-Case, 4-day cumulative abnormal returns in percentage points.


2a 3.09 (0.51) �2.24 (�0.44) 0.43 (0.09) –

2b �2.55 (�0.84) �7.66n (�2.77) �5.11n (�2.33) (Pro-competitive)



Table A3Bewag-Case, 4-day cumulative abnormal returns in percentage points.


3a �4.72 (�1.27) 2.96 (0.97) �0.90 (�0.38) –

3b �4.09 (�1.49) �3.09 (�1.12) �3.59 (�1.46) –



Appendix

See Tables A1–A3.

References

Aktas, N., de Bodt, E., Roll, R., 2007. Is European M&A Regulation Protectionist? TheEconomic Journal 117 (522) 1096–1121.

Brunekreeft, G., Keller, K., 2000. The electricity supply industry in Germany:market power or power of the market? Utilities Policy 9 (1) 15–29.

Bundeskartellamt, 2001. Beschluss B8-40100-U-15/01. 8. Beschlussabeilung.Available at: /http://www.bundeskartellamt.de/wDeutsch/download/pdf/Fusion/Fusion01/B8_15_01.pdfS.

Christensen, L.R., Greene, W.H., 1976. Economies of scale in U.S. electric powergeneration. The Journal of Political Economy 84 (4), 655–676.

Clougherty, J.A., Duso, T., 2009. The impact of horizontal mergers on rivals: gainsto being left outside a merger. Journal of Management Studies 46 (8),1365–1395.

Duso, T., Neven, D.J., Roller, L.-H., 2007. The political economy of European mergercontrol: evidence using stock market data. The Journal of Law and Economics50 (3), 455–489.

Eckbo, E., 1983. Horizontal mergers, collusion, and stockholder wealth. Journal ofFinancial Economics 11, 241–273.

Eckbo, E., Wier, P., 1985. Antimerger policy under the Hart–Scott–Rodino Act: areexamination of the market power hypothesis. The Journal of Law andEconomics 28 (1), 119–149.

Fama, E.F., 1970. Efficient capital markets: a review of theory and empirical work.The Journal of Finance 25 (2), 383–417.

Farrell, J., Shapiro, C., 1990. Horizontal mergers: an equilibrium analysis. TheAmerican Economic Review 80 (1), 107–126.

Fridolfsson, S.-O., Stennek, J., 2006. Industry concentration and welfare–on the useof stock market evidence from horizontal mergers. Discussion Paper No. 5977.Centre for Economic Policy Research, London.

Hogselius, P., 2009. The internationalization of the European electricity industry:the case of Vattenfall. Utilities Policy 17 (3–4), 258–266.

LeRoy, F.S., 1989. Efficient capital markets and martingales. Journal of EconomicLiterature 27 (4), 1583–1621.

London Economics, 2007. Structure and performance of six european wholesaleelectricity markets in 2003, 2004 and 2005. Study for the EuropeanCommission, prepared by London Economics in association with GlobalEnergy Decisions.

MacKinlay, A.C., 1997. Event studies in economics and finance. Journal ofEconomic Literature 35 (1), 13–39.

Martin, R., Roma, M., Vansteenkiste, I., 2005. Regulatory Reforms in Selected EuNetwork Industries. Occasional Paper Series No. 28/April. European CentralBank, Frankfurt.

McAfee, R.P., Williams, M.A., 1988. Can event studies detect anticompetitivemergers? Economics Letters 28 (2), 199–203.

Newbery, D., 2007. What are the issues in mergers and acquisitions arising fromelectricity market restructuring? EUI Working Paper RSCAS 2007/01. FlorenceSchool of Regulation.

Oberndorfer, U., 2009. EU emission allowances and the stock market: evidencefrom the electricity industry. Ecological Economics 68 (4), 1116–1129.

Ockenfels, A., 2007. Stromborse und Marktmacht. Energiewirtschaftliche Tages-fragen 57 (5), 44–58.

Ramos-Real, F.J., 2005. Cost functions and the electric utility industry. Acontribution to the debate on deregulation. Energy Policy 33 (1), 69–87.

Samuelson, P.A., 1973. Proof that properly discounted present values of assetsvibrate randomly. The Bell Journal of Economics and Management Science 4(2), 369–374.

Schumann, L., 1993. Patterns of abnormal returns and the competitive effects ofhorizontal mergers. Review of Industrial Organization 8 (6), 679–696.

Schwarz, H.-G., Lang, C., 2006. The rise in german wholesale electricity prices:fundamental factors, exercise of market power, or both? IWE Working Paper02/2006. Institut fur Wirtschaftswissenschaft, Universitat Erlangen-Nurnberg.

Schwarz, H.-G., Lang, C., Meier, S., 2007. Market power in the german wholesaleelectricity market: what are the political options? Paper prepared for theenerday 2007 in Dresden. Institut fur Wirtschaftswissenschaft, UniversitatErlangen-Nurnberg.

Stillman, R., 1983. Examining antitrust policy towards horizontal mergers. Journalof Financial Economics 11, 225–240.

Stoft, S., 2002. Power System Economics—Designing Markets for Electricity. IEEEPress, Wiley-Interscience, Piscataway, New Jersey.

Twomey, P., Green, R., Neuhoff, K., Newbery, D., 2004. A review of the monitoringof market power. Cambridge Working Papers in Economics CWPE 0504.University of Cambridge.

Ventosa, M., Baıllo, A., Ramos, A., Rivier, M., 2005. Electricity market modelingtrends. Energy Policy 33 (7), 897–913.

Vives, X., 1999. Oligopoly Pricing: Old Ideas and New Tools. The MIT Press,Cambridge, Massachusetts.

Weigt, H., von Hirschhausen, C., 2008. Price formation and market power in theGerman wholesale electricity market in 2006. Energy Policy 36 (11), 4227–4234.

Willems, B., Rumiantseva, I., Weigt, H., 2009. Cournot versus supply functions:what does the data tell us? Energy Economics 31 (1) 38–47.

Zimmer, M., Lang, C., Schwarz, H.-G., 2007. Marktstruktur und Konzentration inder deutschen Stromerzeugung 2006. Zeitschrift fur Energie, Markt, Wettbe-werb 5 (2), 64–69.

http://www.bundeskartellamt.de/wDeutsch/download/pdf/Fusion/Fusion01/B8_15_01.pdf

http://www.bundeskartellamt.de/wDeutsch/download/pdf/Fusion/Fusion01/B8_15_01.pdf

Documents

Competition effects of mergers: An event study of the German electricity market