Frequent News and Pure Signals: Evidence of a Publicly ...Georg Stadtmann⁄ November 2003 Abstract...
Transcript of Frequent News and Pure Signals: Evidence of a Publicly ...Georg Stadtmann⁄ November 2003 Abstract...
Frequent News and Pure Signals:
Evidence of a Publicly Traded Football Club
Georg Stadtmann∗
November 2003
Abstract
We use stock market data for Borussia Dortmund GmbH & Co.KGaA – one of the leading German football clubs – for an applicationof the news model. Due to the specific characteristics of the newsgenerating process, the case of a publicly traded sport club is a veryappropriate candidate for testing this model. We elaborate whethernew information regarding the sporting success helps to explain sub-sequent changes in the stock price of Borussia Dortmund. In a firststep, we develop an industry model that links the sporting success tothe economic success. In a second step, we quantify the effect of matchoutcomes in the different competitions on the subsequent change inthe stock price.
Keywords: News Model, Football Industry, Betting Odds, Stock Market
JEL classification: G14
∗ Institute for Development Strategies, Indiana University, Bloomington,USA and
Department of Economics, WHU Koblenz, Burgplatz 2, 56179 Vallendar,Germany, Phone: +49-261-6509-273, Fax: +49-261-6509-279E-mail address: [email protected]
Frequent News and Pure Signals:
Evidence of a Publicly Traded Football Club
November 2003
Abstract
We use stock market data for Borussia Dortmund GmbH & Co.KGaA – one of the leading German football clubs – for an applicationof the news model. Due to the specific characteristics of the newsgenerating process, the case of a publicly traded sport club is a veryappropriate candidate for testing this model. We elaborate whethernew information regarding the sporting success helps to explain sub-sequent changes in the stock price of Borussia Dortmund. In a firststep, we develop an industry model that links the sporting success tothe economic success. In a second step, we quantify the effect of matchoutcomes in the different competitions on the subsequent change inthe stock price.
Keywords: News Model, Football Industry, Betting Odds, Stock Market
JEL classification: G14
1
1 Introduction
According to the news model of asset price determination, changes in asset
prices are the outcome of the emergence of new, non-expected information
that was not considered in asset prices so far. This information is frequently
labeled as a signal with respect to the fundamental value of an asset. When
testing the news model, it would be ideal to have signals available that
are very frequent, easy to quantify, occur solely when financial markets are
closed, become publicly available to all agents at the same point in time,
and have ex-ante observable expectations.
Regularly, signals such as earnings announcements are used to test the news
model. The disadvantage of using earnings announcements is that such
information occurs only infrequently, since such an analysis is based on
quarterly reports. Furthermore, some agents have access to this information
in advance which can lead to substantial problems due to insider trading.
Additionally, information regarding quarterly reports is already expected to
some extent, and is therefore at least partially reflected in market prices.
All the characteristics mentioned above lead to a scenario, where earnings
announcements cannot be regarded as pure signals so that the news content
is not easy to quantify.1
A stock market segment where signals come close to fulfilling the above men-
tioned criteria is the sports industry with publicly traded sports clubs, like
football teams. Football teams participate in different competitions like the
national championships, national cup competitions, as well as international
competitions such as, on an European level, the Champions League, UEFA,
and UI-Cup. Therefore, financial agents receive information on the strength
of a team regularly and frequently. Furthermore, the outcome of a game is
easy to quantify and the influence of a game’s outcome on league competition
is very transparent due to the overall league ranking. Another important
feature is that matches normally take place on weekends or at night, so that
the outcome of the games materialize when financial markets are closed.
Additionally, the outcome of the games becomes public knowledge at the
1See Brown/Hartzell (2001).
2
very same time and it can thus be excluded that financial agents act on inside
information. A very important aspect is also the fact that betting odds are
available for all matches. These betting odds can be used to extract market
expectations and ex-ante winning/losing probabilities. Therefore, one can
control for the ex-ante expected match outcome. Given these industry
characteristics, it becomes clear that publicly traded sport clubs can be re-
garded as a very appropriate candidate for an application of the news model.2
Until now, empirical evidence of the link between sporting success and
subsequent stock performance of publicly listed football companies is only
limited. Lehmann/Weigand (1998) as well as Renneboog/Vanbrabant
(2000) analyze the performance of the British football clubs. Dahlke/Rott
(2001) who focus on Borussia Dortmund do not use econometric methods
to examine the stock performance but concentrate on corporate governance
issues related to Borussia Dortmund. A study which comes closest to this
study is an empirical work of Feddersen/Maennig (2003). They elaborate
whether the stock market for shares of Borussia Dortmund is efficient in its
weak and semi-strong forms as defined by Fama (1970). The only Borussia
Dortmund specific sport related variable considered in the empirical analysis
is the actual ranking of Borussia Dortmund in the Bundesliga competition.
Hence, this study neglects the fact that Dortmund also competes in the
DFB Cup competition as well as competitions on an European level. As we
show, the European competitions are a very important factor to generate
revenues and should therefore, not be neglected in an empirical analysis of
publicly traded football clubs.
The remainder of the paper is organized as follows. In Section 2, we introduce
the theoretical concept of the news model. We show how new information
2Nevertheless, one may argue that the football industry may also have some disadvan-tages. Football fans are often regarded to act irrationally and one cannot rule out thatfootball fans also engage in stock market trading. However, some institutional investorshold major stakes in Borussia Dortmund GmbH & Co. KGaA. For example, in 2000/2001the Deutsche Bank was holding a stake of 10 % and an international investment fund washolding a 5 % in the company (Borussia Dortmund 2001, p. 9). Largest single shareholderis still the Ballspielverein Borussia 09 e.V. Dortmund which holds a stake of more than25 % (Borussia Dortmund 2002, p. 30). Furthermore, the Deutsche Bank was the mainunderwriter during the IPO process.
3
about future fundamentals influences the current stock price and highlight
the role of the expectation formation process. In Section 3, we focus on the
football industry. We develop a theoretical framework that highlights the
link between the sporting success and the economic success. Equipped with
this theoretical framework, we derive the main hypotheses which are tested
in the empirical part of this paper. In Section 4, we present the results of
our empirical analysis and perform several statistical tests to confirm the
robustness of our main results. In particular, we focus on the role of the
expectation formation process by using betting odds information to control
for the ex-ante expected fundamentals. The last section concludes.
2 The News-Model
The corner stone of the news model is an asset evaluation equation.3 It
is assumed that there exist some fundamental variables (zt) – expressed in
natural logarithms – that explain the log of the asset price (st). Let t index
time and Et(st+1 − st) the expected change in the asset price for the next
period conditional on all information available at time t. We then have
st = α1z1,t + α2z2,t + α3z3,t + ...αnzn,t + γEt(st+1 − st). (1)
It is assumed that all n−coefficients (αi) which express the impact of the
fundamental variables on the asset price are constant over time. Besides the
impact of the fundamental variables, it is also assumed that the expected
change in the asset price also affects the level of the current asset price.
The expected relative change in the asset price influences the current asset
price through the parameter γ which is assumed to be positive and constant
over time. Collecting all fundamental variables in a column vector Zt of
dimension (n× 1) and all coefficients in a row vector A of dimension (1×n),
yields st = AZt + γ[Et(st+1)− st]. Solving this equation for st, we obtain
st =1
1 + γAZt +
γ
1 + γEt(st+1). (2)
3The news model was initially introduced by Frenkel (1981) and Mussa (1982) as amodel to explain exchange rate movements. See also Copeland (1994) for a good overviewand textbook version of the model.
4
The current asset price can, therefore, be expressed as a function of the
fundamental variables and the expected asset price of the next period. The
following transformations are performed to eliminate the expected asset price
parameter from equation (2) so that the current asset price can be written as
a function of current and expected fundamentals. To this end, the expression
containing the expected asset price has to be eliminated from equation (2).
Leading the time index t of this equation by one period, and taking in period
t the expectations of the asset price for period t + 1, one gets
Et(st+1) =1
1 + γAEt(Zt+1) +
γ
1 + γEt(st+2). (3)
Using equation (3) to eliminate the expectations Et(st+1) from equation (2),
we have
st =1
1 + γAZt +
γ
1 + γ
1
1 + γAEt(Zt+1) +
(γ
1 + γ
)2
Et(st+2). (4)
Taking in period t – in analogy to equation (3) – expectations about the asset
price in period t+2 and eliminating the expectation expression Et(st +2) in
equation (4) leads to:
st =1
1 + γAZt +
γ
1 + γ
1
1 + γAEt(Zt+1) + (5)
(γ
1 + γ
)21
(1 + γ)AEt(Zt+2) +
(γ
1 + γ
)3
Et(st+3).
Performing this kind of substitution n times and letting n approach infinity,
the last expression will converge towards zero, since γ > 0.4 We obtain:
st =1
1 + γAZt +
1
(1 + γ)A
∞∑
j=1
(γ
1 + γ
)j
Et(Zt+j). (6)
Equation (6) shows that the asset price of period t is solely determined by
the realized fundamentals in period t and the expected fundamentals for all
4To be more precise, the condition limn→∞[γ/(1 + γ)]nEt(st+n) = 0 only holds ifthe expectation expression does not grow faster towards infinity than the discount fac-tor [γ/(1 + γ)] converges towards zero. This assumption is known as the transversalitycondition (see Mark 2001, p. 68).
5
future periods. Furthermore, it should be emphasized that expected changes
in fundamentals enter the asset price equation only at a discounted value,
with [γ/(1+γ)] serving as the discount factor. This implies, that the further
a change in the fundamentals lies in the future, the lower its impact on the
current asset price. Taking again in period t−1 expectations about the asset
price in period t as given in equation (6) and subtracting the expectations
Et−1(st) from the asset price st, yields:
st − Et−1(st) =1
(1 + γ)A[Zt − Et−1(Zt)] + (7)
1
(1 + γ)A
∞∑
j=1
(γ
1 + γ
)j
[Et(Zt+j)− Et−1(Zt+j)] .
Equation (7) shows that a divergence of the observed asset price in t from
its in t− 1 expected value can be due to two factors:
• On the one hand, it is possible that the fundamentals realized in period
t differ from the expected fundamentals.
• On the other hand, it is possible that new information arrives so that
expectations about future fundamentals are adjusted.
Both factors can lead to a scenario where agents in the financial markets
change their evaluation of the fair value of an asset due to the arrival of new
information.
3 The Professional Football Industry
3.1 The Structure of the German and European Foot-ball Competitions
The top football division in Germany is the Bundesliga, which was founded
in 1963. It has 18 teams that play each other for the German championship
in home and away matches, alternating between their home and their
opponent’s stadiums. Thus, each team must play 34 Bundesliga matches
per season. In addition to the Bundesliga matches, a competition is held
6
each year for the DFB Cup. Under the current rules, 64 teams take part in
the first round of the cup competition. In addition to all teams from the 1st
and 2nd Bundesliga, clubs from the two regional leagues and amateur teams
that may reach the main draw of the DFB Cup competition via qualification
matches, perticipate in the main round. The DFB Cup competition is
played using a knockout system, i.e., only the winning side in a given
match qualifies for the next round. The pairings are drawn by lot (Borussia
Dortmund 2000, pp. 21 – 22).
Since the beginning of the 1999/2000 season, 32 European top teams have
been participating in the Champions League, a championship organized
by the UEFA.5 In addition to the competition’s sporting dimension, the
significant financial value of participation is of great importance to football
clubs. Under the current rules, the first two teams in the Bundesliga
automatically qualify to play in the Champions League. The third ranked
team may also qualify for the Champions League via a qualifying round.6
The competition initially begins with two rounds of matches where the
teams in each group play each other. The best eight teams then qualify for
the quarterfinals. Thereafter, the competition is continued under a knockout
system with home and away matches. The final is held at a neutral stadium.
Since the 1999/2000 season and the abolition of the European Cup Winners’
Cup competition, the only competition on an European level other than the
Champions League is the UEFA Cup. The competition is conducted using
a knockout system with home and away matches. In the third round, eight
teams that were knocked out following the first round of group matches
in the Champions League are added to the UEFA Cup competition. The
fourth to sixth placed teams from the Bundesliga qualify for the UEFA Cup,
5The Union des Associations Europeennes de Football (”UEFA”), an association formedunder the law of Switzerland, was founded in 1954. It is the umbrella organization of theEuropean football associations, and today has 51 members. In addition to the EuropeanChampionship, which is held every four years, UEFA organizes club championships suchas the Champions League and the UEFA Cup.
6In the 2000/2001 and 2001/2002 season even the fourth placed team had the chanceto qualify for the Champions League (CL). Those teams that fail to qualify for the CL viaqualification matches are automatically qualified for the UEFA Cup competition (Kicker2000, p. 3 and 2001, p. 3).
7
as well the DFB Cup winner.7 Two other Bundesliga teams may qualify to
participate in the competition via a further qualification round, the so-called
UEFA Intertoto Cup (”UI-Cup”).
– Insert Table 1 here –
Table 1 gives an overview of the sporting success of Borussia Dortmund for
the time span under consideration (11/2000 – 11/2002). In the Bundesliga,
Borussia Dortmund was quite successful with a 3rd place at the end of the
2000/2001 season and winning the national championship at the end of the
following season. Taking all observations for this competition together, one
can analyze the influence of the outcome of 70 matches, 37 of them were
won while 11 games were lost.
The good performance in the national competition allowed Dortmund to
qualify two times for the Champions League. Although they dropped out
after the first round in the Champions League in 2001/2002, they took
the chance to compete in the UEFA Cup competition subsequently and
made it to the final. Therefore, we observe 23 signals of the international
competitiveness of Borussia Dortmund.
While being highly successful in the two competitions mentioned before,
Dortmund failed early in the DFB Cup competition against ’underdogs’ of the
second or third league. Hence, we only have a limited number of observations
for this competition.
3.2 The Link Between the Sporting Success, Revenues,and Profits
Figure 1 highlights the link between the success in the sports league and the
financial success of a football club (see Lehmann/Weigand, 1997). If a team
7Should the winner of the DFB Cup competition be qualified for the CL, the loser of theDFB Cup final is qualified for the UEFA Cup. The number of starting places per nationalfootball association depends upon the clubs’ rating in the UEFA Five Year Evaluation,depending on the results of the respective national club teams during the preceding fiveyears.
8
is successful and wins some games in a row or has a good position in the
overall ranking,
• the club has the chance to qualify for an European competition like the
Champions League, UEFA or UI-Cup,
• the club will attract more spectators and gate attendance will be higher,
• and the club will have a larger presence in the mass media like daily
broadcasts in the TV news or in sport magazines, leading to a stronger
brand name.
– Insert Figure 1 here –
This success will lead to higher revenues due to the following linkages:
1. In Germany the television rights of the Bundesliga and the DFB Cup
are centrally marketed by the Deutsche Fußball Bund (DFB) and the
Deutsche Fußball Liga (DFL).8 Before the breakdown of Kirch Media
the contract foresaw to pay – on average – approximately 383 million
Euro per season. Under the terms of this agreement, the first division
receives 80 % of the total amount. Half of this income is divided such
that each member of the first division receives a fixed amount of about
8.5 million Euro. The other half is paid according to performance-based
criteria, depending upon sporting success during the last three as well
as the current season (Borussia Dortmund 2000, p. 24).
2. With a qualification for a European competition, the club can generate
additional funds from selling the broadcasting rights of this compe-
tition. While the broadcasting rights of the Champions League are
centrally marketed by the UEFA, every participant in the UEFA Cup
competition markets the broadcasting rights of its home games indi-
vidually (see Elter 2002, p. 86).
3. A successful team is able to generate higher advertising and sponsoring
revenues, because most sponsoring agreements provide for graduated
8See Kruse/Quitzau (2002) and Parlasca/Szymanski (2002) for a critical analysis ofthe central marketing of television rights to football matches.
9
revenues based on the team’s performance. Especially, a participa-
tion in European competitions can generate additional funds (Borussia
Dortmund 2000, p. 29 and 2001, p. 15).
4. Furthermore, a higher gate attendance will lead to higher
ticket and merchandising revenues. Gartner/Pommerehne (1978),
Lehmann/Weigand (1997) as well as Czarnitzki/Stadtmann (2002) an-
alyze the link between sporting success and attendance figures for the
German football industry. Both studies find a significant positive link
between the two variables.9
The link between sporting success and higher gate attendance is only valid
if the stadium capacity is not a binding constraint. Of the 35 Bundesliga
home games of Borussia Dortmund under consideration, 11 matches were
sold out. The average usage of the stadium capacity (68,600 spectators)
in the underlying time horizon is 96 %. The lowest number of spectators
during the underlying time horizon was 60,500 (Kicker 2003). Hence, the
stadium capacity is a binding constraint for Borussia Dortmund. Therefore,
the possibility to generate additional revenues by attracting a larger crowd
seems to be restricted – at least in the short run.10 Nevertheless, in times
of lacking sporting success, revenues could go down due to lower gate at-
tendance. Furthermore, the argument that sporting success leads to more
revenues stemming from ticket sales holds at least for the DFB and the UEFA
Cup competition where games are organized in a knock out modus.
– Insert Table 2 here –
Table 2 focuses on the revenue structure of Borussia Dortmund.11 The
balance sheet figures highlight the importance of those revenues that can be
9For example, Czarnitzki/Stadtmann (2002) find that every position gained in theoverall ranking leads to a higher attendance of about 230 spectators, i.e., the team whichis ranked first can expect approximately 3,910 additional spectators than the worst team.These authors explicitly control for the stadium capacity by using a generalized Tobitestimator for individual and known cut of points.
10To meet excess demand in the future, Borussia Dortmund plans to increase stadiumcapacity until the beginning of the 2003/2004 season. The enlarged arena will have acapacity for 80,000 spectators in Bundesliga matches (Borussia Dortmund 2002, p. 42).
11Since the fiscal year of Borussia Dortmund ends as of June 30 each year, the balancesheet figures for the fiscal year 2002/2003 are not available yet.
10
generated through selling broadcasting rights and advertising/sponsoring.
Furthermore, this table also shows the immense decrease of revenues during
the 2000/2001 season, when Dortmund did not qualify for an European
competition. When controlling for those revenues stemming from transfer
operations, revenues decreased by 41.5 % during fiscal year 2000/2001
compared to the fiscal year 1999/2000.12 Despite the immense sporting
success, overall profits can be regarded as relatively low. So far, Borussia
Dortmund has not paid any dividends even after winning the German
championship.
With a lot of funds at hand, a football club can invest in ”bricks and
bones”. It can increase the capacity or the comfort of the stadium.13 It can
also hire better players to increase the overall strength of the team. These
investments can have feedback effects on game attendance and the success
of the team, indicated by the dotted arrows in Figure 1.14 Furthermore,
the club has the opportunity to invest in sports-related fields, such as TV
stations, travel agencies, hotels and restaurants, or sportswear companies.15
12The high amount of transfer revenues stem mainly from selling the transfer rights ofplayer Evanilson to AC Parma. Nevertheless, Evanilson is still in the team of Dortmunddue to a licensing agreement (see Borussia Dortmund 2001, p. 16).
13See Dietl/Pauli (2002) for a detailed analysis of the economics of the German footballclubs with respect to their stadiums. They argue that Germany’s professional footballteams need modern stadiums in order to increase ticket revenues but that a high investmentrisk can lead to an underinvestment in stadium projects. Furthermore, they focus on theconflict between clubs that invest in their own arenas, like Borussia Dortmund, and clubsthat receive either direct or indirect stadium subsidies.
14Szymanksi/Smith (1997, p. 136) find in an empirical analysis of the English footballindustry that the function between profit and position in the league has a negative slope.This negative slope indicates that increased spending on players is – on average – not selffinancing through higher performance and revenue.
15For example, Borussia Dortmund (2000, p. 43 – 44) argues that it tries to investin lines of business related to football to become less dependent on the sporting successof the professional team. To this end, Dortmund established their own line of sportsclothing which is marketed under the brand name goool.de. Furthermore, Borussia Dort-mund owns part of their home arena ’Westfalenstadion’, holds a 51 % stake in the travelagency B.E.S.T, a strategic alliance with Euro Loyd Reiseburos, and a 33.4 % stake inOrthomed GmbH, a company that is specialized in medical rehabilitation and physicaltherapy services. Dahlke/Rott (2001, p. 10) critically remark, that Borussia Dortmundmay not have core competencies in these related fields. Furthermore, they emphasize thatit should be the shareholder – and not the company itself – who is responsible for portfoliodiversification.
11
Last but not least, higher revenues should also increase the profitability of the
club which should also lead to higher (expected) dividend payments. Higher
expected dividend payments will – according to the standard theory of fi-
nance – lead to higher stock prices. Hence, if all relations hold as supposed
in our framework, the outcome of a game should influence stock prices. How-
ever, the simple model neglects some special industry characteristics that are
reviewed shortly in the next subsection.
3.3 Deriving the Hypotheses
The literature on sport economics provides major arguments that question
the causal relationship between sporting and financial success as outlined
in the industry model. Some factors are company or Bundesliga specific,
while others are related to the overall industry characteristics and the
league structure of team sports.16 In Germany, the 50 % + 1 rule assures
that the voting majority of all clubs remains in the hand of a non-profit
organization (eingetragener Verein, ”e.V.”). 17 Therefore, we cannot rule
out that the members of the non-profit organization try to maximize their
personal welfare which may be directly related to the sporting, but not to
the financial success of the club. Since the members are not entitled to
receive dividends or similar kinds of payments, they may prefer to reinvest
all revenues in players in order to maximize team strength. If this is the case,
sporting success would positively influence revenues, but not necessarily
profits and therefore stock prices.18
16See also the discussion in Dahlke/Rott (2001), p. 6 as well as Lehmann/Weigand(2002).
17In case of Borussia Dortmund, the company is structured as a partnership limited byshares which is labeled as Kommanditgesellschaft auf Aktien (KGaA). The legal form ofa KGaA is uncommon in an international setting. Therefore, Borussia Dortmund informsthe potential shareholders in its Offering Circular Prospectus (2000, pp. 14 – 15): ”Thefact that, in comparison to the shareholders of a stock corporation, the Shareholders ofBorussia Dortmund have only a very limited ability to influence management at the Gen-eral Shareholders’ Meeting or indirectly through the Supervisory Board, as well as the factthat foreign investors in particular are largely unfamiliar with the KGaA as a corporatelegal form, could affect investor interest and lead to a decrease in the value of the shares.”
18The low degree of profit orientation among the clubs of the 1st and 2nd GermanBundesliga is backed by a recent survey among the club managers. Of the 21 managerswhich answered the survey 14 reported that maximizing team strength under the side
12
Franck/Muller (2000, p. 153) argue that a share of Borussia Dortmund may
– due to industry characteristics – represent fewer property rights compared
to a share in a company of an other industry. The football industry like
any other sports league, is characterized by a two stage production process:
On the club level – the first stage – different input factors are bundled to
form a team. On the league level – the second stage – the DFB/DFL set
the overall rules, structure the different competitions and thereby create
a marketable product (Franck/Muller 1998, p. 123). All the rules and
restrictions set by the DFB/DFL – which seem to be necessary to guarantee
league competition – lead to a scenario in which the different rights of the
football club, and, therefore, the rights of the shareholders, too, are limited
compared to other industries.
Franck/Muller (1997) compare league competition with a rat race where
the chance of getting the cheese (higher revenues for the individual club)
increases with the speed of the rat (investments in better players), although
no additional cheese is produced (revenues for the whole league are constant).
They argue that the main problem is the rank interdependence within the
sports league. A team that gains a higher rank and receives higher revenues
creates automatically a negative external effect for the displaced club which
is not taken into consideration by the successor. As a consequence the
displaced club may also increase its investments in better players to restore
the former ranking.19
Furthermore, these theoretical arguments are backed by empirical evidence
with respect to the most mature market for publicly listed football clubs:
The majority of all British clubs that went public showed a dramati-
cal decline in stock prices and under performed the overall market (see
condition of the budget constraint is the objective of the club. Only three managersindicated that their objective is profit maximization (Swieter 2002, p. 63).
19In a setting of rational agents, relatively weak clubs should accept the relative strengthof the other clubs and anticipate the counter of the opponents. For the sport industry,Frank/Muller (1997) find several theoretical arguments why club managers do not actrationally, but overinvest and become part of a rat race.
13
Lehmann/Weigand 1998 as well as Renneboog/Vanbrabant 2000).20 These
industry characteristics lead to a scenario where the hypothesis that the
sporting success does automatically also lead to additional profits is not
backed by existing empirical findings. Nevertheless, a few British clubs won
the race, are profitable, and even outperformed the overall stock market.
Furthermore, Lehman/Weigand (1998) show that Borussia Dortmund is
managed with a high degree of professionalism and financial soundness and is,
therefore, considered as one of the two candidates of the First German Bun-
desliga that seems to be prepared to perform as a publicly traded company.21
Since Borussia Dortmund is the first German club that went public, it could
be that their move to the stock market may turn out to be successful –
despite the overall finding that an investment in football clubs is, on average,
unprofitable. Due to a first-mover advantage Borussia Dortmund may be
even able to materialize an unsuitable part of the goodwill which was build
up by the German Bundesliga since its foundation in 1963 (Franck/Muller
2000, p. 156).
The discussion of the different arguments highlights a very important ques-
tion: Does the link between higher revenues – generated through sporting
success – and higher profits holds for the industry under consideration? As
pointed out in this paragraph, theoretical and empirical findings do not ex-
clusively support the one or the other hypothesis. Therefore, it is very in-
teresting to analyze how the agents at the financial markets are judging this
issue. Hence, we test the following hypotheses as suggested by the industry
model (Figure 1):
20Currently, the shares of 36 European football clubs are listed at different stock ex-changes. England is the most matured market with 20 clubs listed, followed by Denmark(7), Italy (3), Portugal (2), Turkey (2), The Netherlands (2), and Germany (1). Only fourclubs (Manchester United, Tottenham Hotspur, Brondbyernes I.F. Fodbold, and AGFKontraktsfodbold A/S) were traded at prices which were above the public offering priceas of August 2002 (WGZ-Bank 2002, p. 123).
21Lehman/Weigand (1998) use their insights from the stock market performance ofthe British football clubs to evaluate the chances and readiness for all clubs of the FirstGerman Bundesliga to go public. Using a four rank scale (IPO possible without problems,possible, no chance right now, no chance even in the medium term) only two clubs wereranked in the highest group: Bayern Munich and Borussia Dortmund.
14
• H1: A won match should influence stock returns positively.
• H2: A lost match should influence stock returns negatively.
As stressed above when analyzing the revenue structure of Borussia Dort-
mund, the high amounts of money that can be earned in the European com-
petitions should be considered. Hence, the third hypothesis is:
• H3: A won/lost game in a European competition will influence stock
returns to a larger extent than a win/defeat in the national competition.
4 The Empirical Analysis
4.1 Controlling for Expectations
The main idea of the news model is that only the difference between the
realized fundamentals and the expected fundamentals has to be regarded
as the news component. Putting it differently, only the expectation error
should influence stock prices. This argument can easily be formalized by
using the news model introduced in Section 2. Referring to equation (7),
the first component of the right-hand side reflects that the asset price will
only be influenced if the realized fundamentals differ from the expected ones.
Therefore, if a home win was already anticipated to a large extent (there
always remains some kind of uncertainty), stock prices should not react that
much. One method to control for the expectations of relative strength of
the two teams involved in a match and, therefore, in the expected match
outcome is to use betting odds information. As many studies show,22 the
football betting market seems to be pretty efficient in the sense that it seems
to be impossible to beat the market by using information such as rankings
of the teams involved, home-away performance, and injuries of important
players. Hence, it is interesting to employ these implicit expectations in
the current analysis. The betting odds used in this analysis were kindly
provided by gamebookers.com, an online bookmaker. Using the conventional
22See for example Pope/Thomas (1989), Williams (1999), and Forrest/Simmons (2000)for a literature survey. The finding of an efficient betting market is not common for allsports. See Thaler (1992) for an extensive literature survey of the horse race bettingmarket, which shows some degree of market inefficiencies.
15
abbreviation for a home win (1), a draw (0), and an away win (2), Table 3
highlights the background of the ’news proxy’ used in the following analysis.
– Insert Table 3 here –
For example, for the match Dortmund versus Koeln quotes prior to the game
were 1.35 for a home win, 4.25 for a draw, and 6.90 for an away win. This
means a bettor who put 1 Euro on a draw received 4.25 Euro. Comparing
the different quotes for this match already highlights that Dortmund was
regarded as the favorite in this match. Summing up the inverse of the quotes
(1/1.35 + 1/4.25 + 1/6.90), yields the mark-up of the betting company.
The higher this mark-up, the higher the price for the bet. As becomes
evident in Table 3, the mark-up is around 12 % for this betting company.
By controlling for this mark-up, one can compute the probability implicit in
the betting odds for a home win which amounts to 66 % [1/(1.35 ∗ 1.12)],
for a draw which is equal to 21 % [1/(4.25 ∗ 1.12)], and the away win of 13
% [1/(6.10 ∗ 1.12)], respectively.23 These implicit probabilities show that
Dortmund was indeed regarded as the favorite.
Having these probabilities at hand, it is possible to control for the expecta-
tions of the financial agents. Because the winning team receives 3 points and
a draw will lead to 1 point, the expected number of points for Dortmund in
the match against Koln is equal to 2.19 (3 ∗ 0.66+1 ∗ 0.21). Since Dortmund
drew in this match, the outcome has to be interpreted as a negative informa-
tion and the expectation error amounts to a negative 1.19 (1 – 2.19). Only
this expectation error has to be regarded as the new information which has
to be priced in on Monday morning. Due to the fact that Dortmund under-
performed in this match, stock prices should decrease. In a second example,
the expected number of points in the away match where Dortmund plays in
Munich is equal to 1.24 Hence, the outcome of the match was in line with
the prior expectations and should therefore have no impact on stock prices.
23See Bosch (2000), pp. 91 – 110 for an analysis how to extract probabilities frombetting odd information.
24The team which plays at home always has a home field advantage. SeeSchwartz/Barsky (1977) and Vergin/Sosik (1999). See also Skarupke (2000) for an empir-ical analysis of home advantage in the German Bundesliga.
16
4.2 Description of the Data Set
In addition to the match outcome variables described in the previous para-
graph, we also consider ad-hoc announcements of Borussia Dortmund as a
signal which could influence stock prices. Ad-hoc news are taken from the
homepage of Borussia Dortmund and are classified according the following
news groups:
• A player renews his contract (LONG),
• a new player is hired (HIRE),
• a player is sold to another club (SOLD), and
• on 12/20/2001 the contract of the coach Matthias Sammer was ex-
tended until 2006 (COACH).
It is possible that some (private) information regarding the incidences men-
tioned above was already in the market before being publicly released. Thus,
the dummy variables take on the value of 1 on the day of the announcement
as well as +/– 2 days before and after the announcement day of the ad-hoc
news.
– Insert Figure 2 here –
Daily stock data of closing prices as well as the development of the SDAX
are taken from Datastream. Figure 2 highlights the development of the stock
price of Borussia Dortmund after going public at the end of October 2000.
With an exception of a short period in the summer of 2001, the stock under
performed the SDAX and lost about 60 % of its value. Although this seems
to be much for the two-year horizon under consideration, the loss is (more
or less) in line with the overall market performance, since the SDAX also
decreased by about 50 %.
4.3 Regression Results
In a first step, we estimate the following Model 1:25
25Since financial time series like stock prices or stock indices are often non-stationary,we tested for stationarity by using the Augmented Dickey Fuller Test. It turned out that
17
∆DORTt = β0 + β1∆SDAXt + β2BUND EX ERt + (8)
β3EU EX ERt + β7DFB EX ERt +
β8SOLD + β9HIRE + β10LONG + β11COACH + ε.
Where ∆DORT denotes the change in stock prices and ∆SDAX the change
in the relevant stock market index.26 The regression results are presented
in Table 4. The estimated beta parameter takes the value of β1 = 0.6274,
meaning that a 1 % change of the SDAX only leads to an under-proportional
change in the stock price.27 The expectation error variable for the Bundesliga,
the EU competition and the DFB Cup competition are highly significant
and positive, meaning that points that were not expected (due to a win
of a game) had a positive impact and points that were already expected
but not generated (due to a loss of a game) had a negative impact on the
stock value. These effects are statistically highly significant. Hence, taking
the estimated coefficients into consideration, we cannot reject hypotheses H1
and H2. As the estimated coefficients of the expectation error variable for the
Bundesliga and the European competitions are nearly the same, the results
do not support hypothesis H3 that European matches are more important
than Bundesliga matches. The coefficients of all dummy variables, created
to measure the impact of the ad-hoc news, are not statistically different
from zero. On the one hand, this could be interpreted in a way that this
information does not influence the fundamental value of the asset. On the
other hand, since it is possible that these events were anticipated by the
public, we cannot rule out that the signal was already reflected in market
prices when it became public knowledge.
the log of the time series of the share prices and the index series are not stationary. Hencewe tested whether the first difference of the logged time series are stationary, which turnedout to be the case. The test results are available from the authors upon request.
26We estimate Models 1 – 3 over the full sample of trading days. Hence, our data setcontains trading days where a match took place the day before and trading days where nomatch was played the day before. This methodology allows to incorporate ad-hoc newsvariables as well as match specific variables. We drop this methodology in Model 4 wherewe condition on the fact that a match was played the day before.
27As a sensitivity analysis, we also regress the change in the asset price an a constantand the change in the SDAX. The estimated coefficient took the value of β = 0.591 andtherefore, does not vary significantly from the coefficient determined in Model 1. The R2
of this specification takes the value of 0.025.
18
– Insert Table 4 here –
As a sensitivity analysis we estimate Model 2
∆DORTt = β0 + β1∆SDAXt + β∗2BUND POINTSt (9)
+β∗3EU POINTSt + β∗7DFB OUTt +
β8SOLD + β9HIRE + β10LONG + β11COACH + ε,
where the variable BUND POINTSt (EU POINTSt) measures the
number of points gained in a Bundesliga (European) match. The variable
DFB OUTt denotes a dummy variable that assumes the value one on
trading days, following a day when Dortmund dropped out of the DFB-Cup
competition. Hence, in Model 2, we regress the stock price return on the
points gained instead of the unexpected number of points gained as an
additional specification. The estimation results for Model 2 show that the
estimated β∗2 coefficient is positive and significantly different from zero.
Furthermore, the β∗7 coefficient is significantly negative, implying that a
knock out in the DFB-Cup competition leads to a decrease in stock prices
of about 4.1 %. All other sport specific coefficients are not significantly
different from zero. As compared to Model 1, the goodness-of-fit decreases,
so that we conclude that Model 2 is not superior to Model 1. Therefore, we
continue to use Model 1 as our basis specification.
One may argue that the variable that measures the outcome of the European
matches is a combination of games that are played in the Champions League
and games that are played in the UEFA Cup competition. Hence, it may
bring further insights to separate the outcomes of the two competitions.
While all UEFA Cup rounds – with an exception of the final – are played
under a knockout system with home and away matches, we consider all UEFA
Cup games up to the final in a variable labeled UEFA EX ER and isolate
the effect of the lost UEFA Cup final in a separate dummy variable.28 Hence,
the regression equation of Model 3 is as follows:
28Since the betting odds prior to the game were nearly pari, it would not lead to furtherinsights to condition on the expectations.
19
∆DORTt = β0 + β1∆SDAXt + β2BUND EX ERt + (10)
β4UEFA EX ERt + β5UEFA FINALt +
β6CL EX ERt + β7DFB EX ERt +
β8SOLD + β9HIRE + β10LONG + β11COACH + ε.
The estimated coefficients reveal, that the outcome of regular UEFA
Cup matches did not influence the stock market in the expected pattern.
Nevertheless, the lost final led to a decrease in stock prices by about 9 %.
Furthermore, the estimated coefficient for the Champions League variable is
positive and larger than the coefficient estimated for Bundesliga matches.
However, the gap is not statistically different from zero. All other coefficients
lie in the same value range as in Model 1.
Another interesting question is, whether the stock market adjusts to new
information during the trading day or whether there is a measurable impact
of match outcome for more than one trading day. To address this point,
we also include the lagged independent variables of the match outcome
variables as additional variables. The estimated coefficients are presented as
Model 4 in Table 4. The lagged variables for Bundesliga, UEFA Cup as well
as DFB Cup matches are not significantly different from zero. This means,
that with respect to these signals, information is priced in within the trading
day following the day of the match. However, the lagged dummy variable for
the lost UEFA Cup final is significantly negative, meaning that this event
influenced stock prices for more than one day. Furthermore, the lagged CL
variable is significant, but has a negative sign meaning that there exists some
kind of overshooting behavior on the first trading day. Since the number
of observations for this competition is very limited, one should not drive
this interpretation too far. The results of Model 4 show that information
is regularly priced in on the subsequent trading day. Nevertheless, some
extraordinary events may also have an impact on the subsequent trading
days.
The regression results presented so far are based on daily observations. This
means that the data set incorporates trading days following days for which
20
we were able to observe the outcome of a match as well as trading days on
which no match was played the day before. Only this methodology allows us
to consider the effect of the ad-hoc news variables in the regressions analysis.
Since the estimated coefficients for the ad-hoc news variables turn out to be
not significantly different from zero, it seems to be interesting to drop the ad-
hoc variables and run the regression again for those trading days following
playing days. Hence we dropped the ad-hoc variables from Model 1 and
estimated the same specification for the smaller sub sample (97 observations).
All coefficients presented in Model 5 turn out to be in the same range as
compared to Model 1.29 Since the R2 is 0.4021, the goodness-of-fit can be
regarded as exceptionally high for this kind of stock market study. Hence, we
can conclude that new information contained in the match outcome variables
is the main factor influencing the stock market on trading days, following a
match.
5 Conclusion
We have applied the news model to the football industry to analyze, whether
new information regarding the sporting success can explain subsequent
changes in the stock price of Borussia Dortmund. The football industry
proves a very appropriate candidate for applying this model due to specific
characteristics: Signals are very frequent and easy to quantify, occur solely
when the markets are closed, become publicly available to all agents at the
very same time, and have observable expectations due to the existence of
betting odds.
According to the news model, only the expectation error should influence
stock prices. Hence, we use betting odds information to control for the
ex-ante expected match outcome. We show that there exists a close link
between the sporting success and subsequent changes in the stock market.
Therefore, the main hypotheses H1 and H2 derived from a theoretical model
cannot be rejected. However, hypothesis H3, predicting the outcome of
European matches should have a higher impact on the stock market than
29One may also choose Model 2 as a candidate for this kind of sensitivity analysis. Theresults did not change significantly.
21
the outcome of Bundesliga matches, is not supported by our regression
results. Although the estimated coefficient for Champions League matches
is larger, the difference between this coefficient and the one estimated for
Bundesliga matches is not statistically different at conventional error levels.
Future research could extend this study in several interesting directions: One
direction could be to test the several links of the industry model – outlined
in Figure 1 – empirically. A very important link in the model is whether
higher revenues, stemming from sporting success, also lead to higher profits.
One crucial factor is, therefore, how the cost structure responds to sporting
success. Hence, future research could focus on how player contracts are
negotiated and how sporting success leads to cost increases via contractual
fixed bonuses for the players.
Another direction for future research could be to extend this study for the
only German publicly traded football club is broadened by using stock
market information of international clubs. In doing this, one would be
able to check whether the results of this study are supported by a broader
international data set.
In the future, there is also a potential to update the existing study by us-
ing new observations. This would open up the possibility to test whether
estimated coefficients are stable over time. Specifically, because Borussia
Dortmund has been the first German football club that went public, it could
be that traders needed some time to become familiar with the football busi-
ness. In this case it could be that market efficiency was not constant over
time, but improved as traders became more familiar with the specific factors
important in pricing the stocks of the first publicly traded German football
club.
22
References
Bosch, Karl (2000): Glucksspiele – Chancen und Risiken, Oldenbourg Ver-lag Munchen.
Borussia Dortmund (2000): Offering Circular, 28.10.2000.
Borussia Dortmund (2001): Annual Report, July 2000 – June 2001.
Borussia Dortmund (2002): Annual Report, July 2001 – June 2002.
Brown, Gregory W. and Hartzell, Jay C. (2001): Market reaction to publicinformation: The atypical case of the Boston Celtics, in: Journal ofFinancial Economics, Vol. 60, pp. 333 – 370.
Copeland, Laurence S. (1994): Exchange Rates and International Finance,Second Edition, Addison-Wesley.
Czarnitzki, Dirk and Stadtmann, Georg (2002): Uncertainty of Outcomeversus Reputation: Some Empirical Evidence for the German Premier-League Football, Empirical Economics, Vol. 27(1), pp. 101 – 112.
Dahlke, Mirko; Rott, Armin (2001): Fußballaktionare in der Abseitsfalle?Erste Erfahrungen mit dem BVB-Papier, Dortmunder Diskussions-beitrage zur Wirtschaftspolitik, Nr. 106, Januar 2001.
Dietl, Helmut M. and Pauli, Markus (2002): Die Finanzierung vonFußballstadien – Uberlegungen am Beispiel des deutschen Profifußballs,Zeitschrift fur Betriebswirtschaft, Special Edition 4/2002, pp. 239 –262.
Elter, Vera-Carina (2002): Vermarktung medialer Rechte im Fußball, in:FC Euro AG, Analyse der borsennotierten europaischen Fußballun-ternehmen – Entwicklung und Chancen des deutschen Fußballmarktes,WGZ-Bank, 3. Edition, pp. 75 – 91.
Fama, E.F. (1970): Efficient Capital Markets: A Review of Theory andEmpirical Work, in: The Journal of Finance, Vol. 25, pp. 282 – 417.
Feddersen, Arne; Maennig, Wolfgang (2003): Sportlicher Erfolg und Kap-italmarktbewertung – Das Beispiel der Borussia Dortmund GmbH& Co. KgaA, Working Paper; forthcoming in: Helmut Dietl(Hrsg.), Globalisierung des wirtschaftlichen Wettbewerbs im Sport,Sportokonomie 4, Tagungsband zur Jahrestagung des ArbeitskreisSportokonomie.
23
Forrest, David and Simmons, Robert (2000): Forecasting Sport: The Be-haviour and Performance of Football Tipsters, International Journal ofForecasting, Vol. 16, pp. 317 – 331.
Franck, Egon and Muller, Jens Christian (1997): Zur okonomischen Struk-tur des sogenannten Rattenrennens, Freiberger Arbeitspapiere 97/15.
Franck, Egon and Muller, Jens Christian (1998): Kapitalgesellschaftenim bezahlten Fußball, in: Zeitschrift fur Betriebswirtschaft,Erganzungsheft 2/1998, pp. 121 – 141.
Franck, Egon and Muller, Jens Christian (2000): Fußball-Aktien: Nur aus-nahmsweise ein Renner, in: Die Bank 3/2000, pp. 152 – 157.
Frenkel, Jacob A. (1981): Flexible Exchange Rates, Prices, and the Roleof ”News”: Lessons from the 1970s, Journal of Political Economy, Vol.89(4), pp. 665 – 705.
Gartner M. and Pommerehne W.W. (1978): Der Fußballzuschauer –ein homo oeconomicus? Eine theoretische und empirische Analyse,Jahrbuch fur Sozialwissenschaften Vol. 29, pp. 1089 – 1095.
Kicker (2000): Sonderheft Finale 1999/2000.
Kicker (2001): Sonderheft Finale 2000/2001.
Kicker (2002): Sonderheft Finale 2001/2002.
Kicker (2003): Kicker Homepage [http://www.kicker.de].
Kruse, Jorn and Quitzau, Jorn (2002): Zentralvermarktung derFernsehrechte an der Fußball-Bundesliga, Zeitschrift fur Betrieb-swirtschaft, Special Edition 4/2002, pp. 63 – 82.
Lehmann, Erik and Weigand, Jurgen (1997): Money Makes the BallGo Round – Fußball als okonomisches Phanomen, in: ifo Studien –Zeitschrift fur empirische Wirtschaftsforschung, Vol. 43(3), pp. 381 –409.
Lehmann, Erik and Weigand, Jurgen (1998): Wieviel Phantasie braucht dieFußballaktie?, in: Zeitschrift fur Betriebswirtschaft, Erganzungsheft2/1998, pp. 101 – 120.
24
Lehmann, Eirk and Jurgen Weigand (2002): Mitsprache und Kontrolle improfessionellen Fußball: Uberlegungen zu einer Corporate Governance,in: Zeitschrift fur Betriebswirtschaft, Erganzungsheft 4/2002, pp. 43 –61.
Mark, Nelson C. (2001): International Macroeconomics and Finance, Black-well Publishers.
Mussa, Michael (1982): A Model of Exchange Rate Dynamics, Journal ofPolitical Economy, Vol. 90(1), pp. 74 – 104.
Parlasca, Susanne and Szymanski, Stefan (2002): When the whole is lessthan the sum of the parts: The negative effects of central marketing offootball television rights on fans, media concentration and small clubs,Zeitschrift fur Betriebswirtschaft, Special Edition 4/2002, pp. 83 – 103.
Pope P.F. and Thomas D.A. (1989) Information, Prices and Efficiency in aFixed–Odds Betting Market, Economica, Vol. 56, pp. 323 – 341.
Renneboog, Luc and Vanbrabant, Peter (2000): Share Price Reactions toSporty Performances of Soccer Clubs Listed on the London Stock Ex-change and the AIM Center for Economic Research Working Paper No.2000-19.
Schwartz, Barry and Barsky, Stephen F. (1977): The Home Advantage,Social Forces, Vol. 55(3), pp. 641 – 661.
Skarupke, Robert (2000): Quantifizierung des Heimvorteils im deutschenProfifußball: Eine empirische Untersuchung fur die 1. Fußball-Bundesliga, Arbeitspapier Nr. 20 des Instituts fur Statistik undOkonometrie and der Johannes Gutenberg-Universitat Mainz Fachbere-ich Rechts- und Wirtschaftswissenschaften.
Swieter, Detlef: Eine okonomische Analyse der Fußball Bundesliga, Duncker& Humblot, Berlin, 2002.
Szymanski, Stefan; Smith, Ron (1997): The English Football Industry:profit, performance and industrial structure, in: International Reviewof Applied Economics, Vol. 11(1), pp. 135 – 153.
Thaler, Richard H. (1992): The Winner’s Curse – Paradoxes and Anomaliesof Economic Life, The Free Press Macmillan.
25
Vergin, Roger C. and Sosik, John J. (1999): No Place like Home: An Exam-ination of the Home Field Advantage in Gambling Strategies in NFLFootball, Journal of Economics and Business, Vol. 51, pp. 21 – 31.
WGZ-Bank (2002): FC Euro AG, Dusseldorf 2002.
Williams, Leigthon Vaughan (1999): Information Efficiency in Betting Mar-kets: A Survey, Bulletin of Economic Research, 51(1), pp. 1 – 30.
26
Table 1: Overview of the Different Competitions
German Cup CompetitionSeason German Bundesliga European Competitions
’DFB Pokal’
· Not qualified, since Dortmund2000 – 2001 · 3rd rank at the end of season was only on 11th rank at the end
· Out in 3rd round against
of 1999 – 2000 seasonSchalke 04 on 11/29/2000
· Winner of the qualification games tothe Champions League against Donezk(8/7/2001 and 8/22/2001)
· 1st rank at the end of season · Out after 1st round in the CL · Out in 1st round against2001 – 2002 · Winning of the National · Switch to 3rd round of the UEFA Cup Wolfsburg Amateure
Championship (5/4/2001) competition on 8/25/2001· Loosing of the UEFA Cupfinal against Feyenoord Rotterdamon (5/8/2002)
· 2nd rank in the first phase of CL,· currently 2nd rank, after half qualification to 2nd phase · Out in 2nd round against
2002 – 2003of the season · until now: 1 match won and 1 SC Freiburg on 5/11/2002
lost in the 2nd phase
70 Matches: 23 Matches: 4 Matches:a)Summary
37 Win, 22 Draw, 11 Lost 12 Win, 6 Draw, 5 Lost 1 Win, 3 Lost
Note: a) The first two matches of the DFB Cup in the 2000 – 2001 season took placebefore November 2000 – the starting point of our empirical study – and are therefore, notconsidered in this analysis.
Table 2: Revenues and Profits of Borussia Dortmund GmbH & Co. KGaA
1999/2000 2000/2001 2001/2002Match Operations 17,624.7 13,829.9 17,841.7Advertising 26,048.8 13,233.4 27,951.9Radio, TV 35,001.1 19,339.1 45,975.7Transfer Revenues 3,047.4 25,259.6 1,235.5Merchandising 9,701.7 5,212.9 8,821.0Others 12.3 109.3 532.3Total Revenues 91,436.0 76,984.2 102,358.1Net Income 1.0 -9.0 1.4
Source: Borussia Dortmund (2001 and 2002). Figures are in thousand Euro.
27
Table 3: Betting Odds and Implicit Probabilities for the Outcome of a Game
Date Teams Result Betting Odds Mark-up Implicit Probability
05/19/2001 Dortmund vs. 1. FC Koeln 3:3 1.35 4.25 6.90 1.12 66 % 21 % 13 %02/09/2002 Bayern Muenchen vs. Dortmund 1:1 1.85 3.25 3.72 1.12 48 % 28 % 24 %
28
Table 4: Regression Results
Model 1 Model2 Model 3 Model 4 Model 5
β0 Constant-0.0016 -0.0018 -0.0013 -0.0013 -0.0135***
(-1.33) (-1.39) (-1.10) (-1.08) (-4.31)
β1 S DAX(t)0.6274*** 0.5751*** 0.6111*** 0.6433*** 0.7503*
(4.11) (3.57) (4.07) (4.18) (1.90)
β′1 SDAX(t-1) – – –-0.1392 –
(-0.89)
β2 BUND EXP ERROR(t)0.0187***
–0.0187*** 0.0192*** 0.0208***
(7.08) (7.20) (7.36) (6.49)
β′2 BUND EXP ERROR(t-1) – – –0.0025 –
(0.98)
β∗2 BUND POINTS(t) –0.0032**
– ––
(2.20)
β3 EU EXP ERROR(t)0.0181***
– – –0.0207***
(3.91) (3.71)
β∗3 EU POINTS(t) –-0.0004
– ––
(-0.16)
β4 UEFA EXP ERROR(t) – –-0.0051 -0.0049 –
(-0.69) (-0.67)
β′4 UEFA EXP ERROR(t-1) – – –-0.0068 –
(-0.93)
β5 UEFA FINAL(t) – –-0.0906*** -0.0900*** –
(-3.72) (-3.73)
β′5 UEFA FINAL(t-1) – – –-0.0571*** –
(-2.37)
β6 CL EXP ERROR(t) – –0.0283*** 0.0285*** –
(4.65) (4.71)
β′6 CL EXP ERROR(t-1) – – –-0.0140*** –
(-2.32)
β7 DFB EXP ERROR(t)0.0622***
–0.0626*** 0.0620*** 0.0460*
(3.05) (3.13) (3.12) (1.86)
β′7 DFB EXP ERROR(t-1) – – –-0.0130 –
(-0.65)
β∗7 DFB OUT(t) –-0.0413***
– ––
(-2.73)
β8 AD HOC SOLD0.0024 0.0048 0.0010 0.0020 –
(0.46) (0.88) (0.21) (0.40)
β9 AD HOC HIRE-0.0022 -0.0025 -0.0010 -0.0007 –
(-0.53) (-0.56) (-0.24) (-0.18)
β10 AD HOC LONG0.0032 0.0042 0.0023 0.0012 –
(0.56) (0.70) (0.40) (0.21)
β11 AD HOC COACH-0.0071 -0.0110 -0.0074 -0.0072 –
(-0.64) (-0.93) (-0.68) (-0.66)
R2 0.1521 0.0532 0.1839 0.2078 0.4021Adjusted R2 0.1390 0.0386 0.1682 0.1830 0.3761Durbin Watson Stat. 2.32 2.30 2.35 2.39 2.06Number of Observations 529 529 529 528 97
Figure 1: Stylized Model for the Football Industry
Higher Gate Attendance
Higher Profit =>
Higher Dividend =>
Higher Stock Price
Investment in Related Fields – TV-Stations – Travel Agencies – Hotels/Restaurants – Sportswear
Investment in the Size or Quality of the Facilities – Stadium Capacity – VIP-Lounges
Investment in Better Players
Higher Advertising / Sponsoring Revenues
Higher Merchandising
Revenues
Higher Ticket Revenues
Higher Broadcasting
Revenues
Qualification Champions League
UEFA/UI-Cup
Larger Attention in Mass Media
Success in the German Bundesliga
Figure 2: Performance of Borussia Dortmund AG Compared to the Benchmark SDAX
20.0%
30.0%
40.0%
50.0%
60.0%
70.0%
80.0%
90.0%
100.0%
110.0%
Oct.00
Dec.00
Feb.01
Apr.01
Jun.01
Aug.01
Oct.01
Dec.01
Feb.02
Apr.02
Jun.02
Aug.02
Oct.02
SDAX
Borussia Dortmund AG