The effect of the FIFA World Cup on the DAX

(1)

! ! ! ! ! ! ! ! ! ! ! ! !

!

The!Effect!of!the!FIFA!World!Cup!

on!the!DAX

! ! Bachelor!Thesis!Economics! ! ! ! ! ! ! ! ! ! ! ! ! ! Student:!Evelien!Rouwen! Student!Number:!10001875! University!of!Amsterdam! Faculty!of!Economics!and!Business! Supervisor:!Jan!Lemmen! Bachelor!Thesis!BSc!Economics! Submission!Date:!February!2nd_!2016!

(2)

Statement'of'Originality'

I hereby declare to take full responsibility for the contents of this thesis.

I declare that the text and work presented in this thesis are original and that no other sources than mentioned in the text and corresponding references have been used.

The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the content.

Signed by:

(3)

Word'of'Acknowledgement'

I would like to thank anyone who has been involved in the process of writing my bachelor thesis. Particularly, I would like to thank my supervisor, Jan Lemmen, for his fast responses that made it easy to stay in contact with him, his provided feedback and all other provided input; you have been a great help.

(4)

Abstract'

The aim of this thesis is to examine whether the FIFA World Cup has an effect on the German stock exchange (DAX). Data is used over a period of twenty-four years. During this period a total of six World Cup tournaments were played. In order to analyze if there is a so-called World Cup effect, two experiments are performed. The first experiment involves an OLS regression, and the second experiment involves an event study. To see whether the effect is only valid for Germany, the Belgium stock exchange (BFX) is used as a control variable. Both experiments do not show a statistically significant effect of the FIFA World Cup on the DAX or BFX. However, it does show a difference return on the DAX compared to the BFX.

Keywords: FIFA, Event Study, Germany JEL Classification: G14

(5)

Table'of'Contents'

Statement'of'Originality'...'1! Word'of'Acknowledgement'...'2! Abstract'...'3! List'of'Tables'and'Figures'...'5! 1.'Introduction'...'6! 2.'Literature'Review'...'8! 2.1!Investor!sentiment!...!8! 2.2!Impact!of!sports!on!the!economy!...!10! 3.'Methodology'...'14! 3.1!Research!Design!...!14! 3.2!Dataset!...!19! 3.3!Hypothesis!...!19! 4.'Model'Specifications'...'21! 4.1!OLSUAssumptions!...!21!

4.2 Calculating Abnormal Return!...!21!

4.3!Regression!Models!...!23! 5.'Results'...'25! 5.1!Regression!Results!...!25! 5.2!Results!Event!Study!...!29! 6.'Conclusion'and'Discussion'...'30! Bibliography'...'31! Appendix'...'33! ! '

(6)

List'of'Tables'and'Figures'

Tables:

Table 1: Regression output calculating abnormal return

Table 2: Regression output abnormal return and dummy variables FIFA and GER Table 3: Regression output abnormal return and dummy variables WIN and LOSS Table 4: Outcome event study

Table 5: Outcome Welch t-test

Table B.1 Summary statistics of the DAX return and the S&P500 Table B.2 Regression output of the DAX return and the S&P500 Table B.3 Summary statistics of the BFX return and the S&P500 Table B.4 Regression output of the BFX return and the S&P500

Table C.1 Summary statistics of the DAX abnormal return and the FIFA and GER Table C.2 Regression output of the DAX abnormal return and the FIFA and GER. Table C.3 Summary statistics of the DAX abnormal return and the WIN and LOSS Table C.4 Regression output of the DAX abnormal return and the WIN and LOSS Table C.5 Summary statistics of the BFX abnormal return and the FIFA and GER Table C.6 Regression output of the BFX abnormal return and the FIFA and GER Table C.7 Summary statistics of the BFX abnormal return and the WIN and LOSS Table C.8 Regression output of the BFX abnormal return and the WIN and LOSS Table D.1 Table event study

Table E.1 Correlation table

Figures:

(7)

1.'Introduction'

Every four years the FIFA organizes the FIFA World Cup. Together with the

Olympic games the FIFA World Cup is considered as one of the most important sport events in the world. During a period of a month, the qualified teams decide who is the best soccer nation in the world. The FIFA World Cup is not just about soccer,

commercially it is an important event and success at the tournament may lead to confidence about the future. Due to the size of the tournament it has shown to be an interesting research topic. Besides studying the economic effect of the World Cup, multiple studies found a relation between stock returns and investor sentiment. Edmans, Garcia and Norli (2007) find a strong relation between stock returns and the results of soccer games. Besides Edmans, Garcia and Norli (2007), Hirshleifer studied the effect of investor psychology and asset pricing (2001). Hirshleifer (2001) found a connection between investor psychology and asset pricing.

Previous studies have shown that there is indeed a movement at the stock market during a FIFA World Cup. Ehrmann and Jansen (2012) examine the trading activity during a World Cup. They find that the number of trades drop during a country’s World Cup match. Besides that, they found that match events (a goal, yellow/red card, halftime etc.) influence trading activity as well. Ashton, Gerrard and Hudson (2010) studied the economic impact of national sporting success on de London stock exchange. They found a statistically significant relationship between the performance of the English team and the price of shares traded on the London stock exchange. Baade, Victor and Matheson (2007) studied the economic impact of the World Cup. They found that the positive economic influence of hosting the World Cup is bigger for developing countries, but only fade away after the event.

Considering the size of the FIFA World Cup and the previously found relations between sport sentiment and stock returns, it will be interesting to examine the effect of the World Cup on the German stock exchange. As Germany is the current holder of the prestigious cup and a big soccer nation as well, the event could have a significant effect on the stock returns. This leads to the following research question:

(8)

This thesis is structured as follows; in chapter two previous studies and literature is reviewed. In chapter three the research design and dataset are explained. Chapter four provides the specifications about the constructed model. Chapter five consists of the results of the empirical research, and in chapter six the conclusion can be found. Furthermore, a bibliography is added containing all references to previous studies used for this thesis. And finally, all the tables and outcomes of this research can be found in the Appendix.

(9)

2.'Literature'Review'

This part of the thesis gives some insights in previous studies. Chapter two is divided into two parts. The first part is about investor sentiment and the stock market. The second part is about the impact of sports on the economy.

2.1'Investor'sentiment'

A fundamental assumption in economics is that agents are rational in decision-making. However, more and more recent studies disagree with this assumption and find increasing evidence of irrational behavior by investors. According to Beremut and Ceylan (2012) behavioral economics and behavioral finance are trying to explain the influence of emotions and cognitive errors on the decision-making process of investors. According to psychologists ‘mood’ is one of the triggers of agent

irrationality, and several studies confirm this claim. In a recent article Edmans, Garcia and Norli investigated the reaction of the stock market by unforeseen changes the mood of investors (2007). They used soccer results as a variable to measure mood, and recorded a significant market decline after a lost game. The recorded loss-effect is stronger in small stocks, and also depent on the importance of the match. Besides their results concerning soccer matches, Edmans, Garcia and Norli (2007) found a loss effect after international cricket, rugby and basketball games as well. However, the effect is smaller but still statistically significant. Besides the study conducted by Edmans, Garcia and Norli in 2007, Hirshleifer analyzed investor psychology and asset pricing as well (2001). He concluded that returns are determined by both risk and misvaluation. Hence, there is indeed some irrational behavior among investors. According to a study from Berument and Ceylan (2012) investor behavior is

influenced by match results as well. Merely, investors become more risk averse after a loss and less risk averse after a win.

In a study by Ashton, Gerrard and Hudson (2003) the impact of soccer results on the London stock exchange is examined. Their results showed a link between the performance of the England soccer team and daily changes in the FTSE 100 index1

. In 2008 Klein, Zwergel and Henning criticized the work of Ashton et al. They found a number of inconsistencies in the conducted study. First of all, a holiday-return effect !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!

(10)

was indicated, meaning that the World Cup is played every four years in the same period. This could also be the reason for slightly different stock market reactions. Secondly they argued for a so-called copy-and-paste-effect, meaning that Ashton et al. (2003) identified games that are won by penalties as a draw. According to Klein et al. (2008) these games should be corrected as penalties lead to a decisive outcome (win or loss). And finally the globalization-effect, as the international matches are played across the globe in different time zones, the data should be corrected for this time difference. Furthermore Klein, Zwergel and Fock (2009) conducted a study where they took into account that a lot of stocks on the market are traded internationally, therefore investors are not influenced by the performance of their national soccer team. Klein et al. (2009) showed that there was no connection between a specific national soccer team’s win or loss and stock index prices, but at the same time they showed how a few modifications in the model suggested that there was indeed a significant result. Therefore they argue that studies that suggest the influence by World Cup matches should be analyzed carefully due these inconsistencies

One of the questions that rose from the Ashton, Gerrard and Hudson (2003) study is; as national team performance during a World Cup tournament has an effect on the local stock market, is arbitrage possible? Kaplanski and Levy (2010) counter this claim. They studied the US stock market to examine whether arbitrage is a possibility. They do find a so-called World Cup effect on the US stock market, but it is a negative average return compared to an all-days average return. However, according to Kaplanski and Levy (2010) arbitrage is not possible due to the foreign investors on stock markets. On the US stock market numerous of investors with different nationalities trade. Therefore, all of these investors are influenced differently by the outcome of a World Cup game. Some win, some lose and some simply do not care.

Furthermore, trader attention during World Cup matches is analyzed as well. Ehrmann and Jansen (2012) studied the effects or shifts in investor attention during the 2010 World Cup in South Africa. The South Africa World Cup was a perfect event to estimate the investor attention, because many soccer matches were played during stock market trading hours. They used trading data for fifteen stock exchanges, and came to three important findings. First of all, when the national team played, the number of trades made dropped by forty-five percent, while volumes were fifty-five

(11)

percent lower. To continue, match events, like a goal, red/yellow card or a penalty kick, influenced trading activity as well. For instance, a goal dropped trading activity by five percent. And lastly they found that the co-movement between national and global stock market returns decreased by more than twenty percent during World Cup matches.

Besides the effect of the FIFA World Cup on a stock market, numerous studies examined the effect of a national soccer competition on a stock market. Berument, Ceylan and Gozpinar (2006) did research on the effect of tree major Turkish teams (Fenerbache, Galatasary and Besiktas) on the stock market returns. They found that a victory for Besiktas against a foreign opponent increases the returns, but the same effect is not found for Fenerbache and Galatasary. Stadtmann (2006) examined the stock price for Borussia Dortmund GmbH & Co.2

Stadtmann estimated a model to analyze if sporting success explains subsequent changes in the stock price of Borrussia Dortmund. He found a close link between sporting success and the stock price. He also found a higher estimated coefficient on Champions League matches compared to Bundesliga or Europa League matches, although this difference is not statistically significant. Scholtens and Peenstra (2009) did a similar type of study, but they selected eight teams from different countries in order to examine whether match results influence stock returns. They have chosen clubs that are listed on the stock exchange, namely; Ajax from the Netherlands, Borussia Dortmund from Germany, Lazio Roma, AS Roma and Juventus from Italy, Manchester United from England and Porto and Sporting Lissabon from

Portugal. !They analyzed both national and European competition and found a positive response on the stock market in case of a victory and a negative response in case of a defeat. However, the response found is strong if a European match is played.

All studies, except the one by Klein, Zwergel and Heiden (2009), found an effect of sports on the stock market. Most studies pointed out that the effect in case of a decrease was more significant than the effect in case of a victory.

2.2'Impact'of'sports'on'the'economy'

Besides an impact on the stock market, the FIFA World Cup has an impact on the economy of the hosting country as well. As Germany hosted the World Cup in 2006, !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!

(12)

it will be interesting to consider to what extent this influence on to economy is relevant.

Not only the effect of the World Cup, mostly for developing countries, is examined in previous studies but the effect of other mega-events as well. Rose and Spiegel (2010) studied the effect of the Olympics on the economy of the host country. They argued that hosting the Olympics would cost a lot, and test the claim of the IOC3

. The IOC states that hosting the Olympics promotes the export of the hosting country, especially tourism (Rose and Spiegel, 2009). Their findings report the high costs for a hosting country, but at the same time Rose and Spiegel report an increase of twenty percent in export. Remarkably, countries with an unsuccessful bid to host the Olympics show similar impact on exports. Therefore they conclude that the change in export is not due to hosting the Olympics, but more due to the policy signal towards future liberalization a country emits when they are in the running to host a mega-event.

According to Baade and Matheson (2007) hosting the World Cup brings significant costs, but potentially large benefits as well. The FIFA requires the host country to provide at least eight, but preferably ten, soccer stadiums to be available. Those stadiums need to have a capacity of 40,000 till 60,000 seats (FIFA, 2002). These requirements lead to high costs for the host country. As an example, in 2002 for the Japan and South Korea bid, both countries provided ten stadiums. South Korea had build ten brand new stadiums for a total cost of approximately two billion dollars, and Japan built seven new stadiums and renovated three others at a cost of four billion dollars. Moreover, Japan improved their infrastructure as well, for more than five billion dollars. Besides these costs prior to the event, operating costs during the event are also expected to be high. Because of the high costs of organizing the World Cup, Baade and Matheson studied if the host country can compensate for these costs. They found a huge income stream for a hosting government, but at the same time a lot of costs for the host cities4

. They concluded that it is far more likely that hosting a World Cup has a negative effect on the overall economy.

To host the South Africa World Cup in 2014 the government spent over three billion dollar on transportation, telecommunication and costs of building new

stadiums according to the Department of Sport and Recreation of South Africa !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!

3_{International Olympic Committee}

(13)

(SRSA). The investments for hosting the event were high, but at the same time it provided some great lessons for the country, like how to budget, organize, manage and implement projects of this size (OECD Observer and Nene, 2013). The same study showed a positive impact on the South African economy; the World Cup contributed a total of 509 dollar to the GDP in 2010. It also created numerous jobs that supported low-income households (SRSA, 2012). The numbers seem promising but the main problem is that these numbers are just from the period before the World Cup, during the World Cup and slightly after the World Cup. When the tournament is over it is just a matter of time that the government in a developing nation falls into old habits, and the jobs gained during the World Cup fade away.

For the Brazil World Cup in 2014 the Brazilian government also recorded a lot of spending (FIFA, 2014). The FIFA reported a total amount of fifteen billion dollar spent on stadiums, transportation and security matters. Brazil is, just like South Africa, a developing nation. Before and during the World Cup a lot of new jobs opened up in Brazil and it seemed that the country was making progress due the event. Tourism increased and tourist spending was estimated at 365 million dollars (EMBRATUR, 2014). The government expects that tourism will increase thanks to the World Cup. However, a study by Fourie and Santana-Gallego (2011) shows that there are only gains from tourist in the three years after the event. According to a McKinsey report published after the World Cup in Brazil the effects on the economy are significant, but at the same time the effects fade after a while. That is mostly due to the corrupt and inefficient government.

Furthermore, the impact of national competitions on the economy is studied. Berument and Yucel (2005) examined the success of the soccer team Fenerbache and the performance of the Turkish industrial production. They found a small increase in monthly growth every time Fenerbache won a European Cup match. Besides

Berument and Yucel their study, McKinsey & Company (2015) published a comprehensive study of the economic importance of soccer in Germany. Just like Berument and Yucel they researched soccer on a national level, namely the Bundesliga and 2.Bundesliga5

. The McKinsey study measured a growing impact of German professional soccer on the economy. They based their results on three factors; added value, jobs and net income for the government. By added value McKinsey !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!

(14)

means the contribution of German professional soccer to the GDP; by jobs they mean the impact of professional soccer on employment; and by net income for the

government they mean the contribution of professional soccer to taxes and social insurances. Besides measuring a growth-rate over the past years, the study cannot tell with certainty that the growth-rate is going to maintain.

So far, based on previous studies, it is hard to determine to what extent hosting the World Cup has on a nation like Germany. Previous research mostly examines the effect of hosting the FIFA World Cup for developing countries. However, the

McKinsey report gives some insights on soccer and economic performance in Germany. ! !

!

'

! ! !

(15)

3.'Methodology'

This part of the thesis is about the research performed to answer the research question. The first part is about the research design, the second part is about the data collection, and the third part lists the hypothesis.

3.1'Research'Design' '

In order to give a well-informed answer to the research question two experiments are performed on the dataset. The first experiment examines the effect of the World Cup on the German stock exchange in the case of a won or lost match. The second experiment contains an event study in order to examine the effect of the World Cup on the German stock exchange. For both experiments the abnormal return needs to be calculated.

3.1.1$Calculating$Abnormal$Return$

Not only in the case of an event stock prices can encounter variations, in the absence of an event changes in variation are possible as well. The estimated price changes in the absence of an event are not overlooked, and are the so-called normal returns. The normal returns serve as an essential benchmark in the model. There are multiple methods and models developed over the year to estimate the normal returns, all with its advantages and disadvantages. Examples of those models are the capital asset pricing model (CAPM), the mean-adjusted model, the (multi or one-factor) market model, the Fama and French three-factor model and the matching model, which implements only the movement of the stock price of a comparable firm or portfolio as benchmark. The CAPM is criticized for its assumptions and for its restriction on the intercept term. The latter equals the risk-free rate, and this additional restriction has a bigger variance as a consequence. Because of that bigger variance, the standard deviation of the error term that is needed for the test statistics (see formula 9), leads to a less powerful test. The mean adjusted model does not allow for general market movements, while the (one-factor) market model does take into account market trends and firm-related risks.

(16)

The one-factor market model to explain daily returns is used in this study in

agreement with the vast majority of other researches. Daily return is calculated with the following formula:

! !

! ! ! !" =(!!!!!!!!)

!!!! ! ! ! ! ! (1)!

Where Rt is the return on day t, and Rt-1 is the return on day t-1.

!

The approach with short-term returns is favored, because with this method the

weighted sum of returns of the individual stocks in a portfolio equals the return of the same portfolio. Moreover, returns present the opportunity to aggregate the abnormal returns across the individual stock while continuously compounded returns do not allow for this option in the cross-section (Van der Sar, 2011 and Brooks, 2008).

The transformation of the actual returns provides the possibility to estimate the predicted parameters in the sample, with the use of ordinary least squares (OLS) regression. The formula used for the regression is as follows:

!

R_j,t =αj+βjRm,t+εj (2) Where Rj,t represents the actual return on the individual stock at time t, αj

indicates part of the stock return that represents the constant influence of firm-specific factors over time,βj denotes part the stock return that depends on

market movements, Rm,t expresses the actual return on the market index at time

t and εj symbolizes the error term, which captures everything that is not

explained by the independent variable, in this case the actual return on the market index.

As the market index the S&P500 is chosen. The S&P500 is a good market proxy because the American stock exchange dominates the global market exchanges and therefore can explain for change in the DAX and BFX indexes.

To derive the normal returns, the actual return on the market index and the estimated parameters derived from (2) are required. The following formula reflects the scientific notation towards obtaining the normal returns:

(17)

ˆ

R_j,t = ˆαj+ ˆβjRm,t (3)

Where Rˆj,tillustrates the normal return on the stock at time t (of the DAX index and the BFX index), ˆαjequals the predicted constant from (2) and βˆj equals the predicted coefficient from (2) and is related to the selected market index (S&P500).

The final step is to calculate the abnormal returns (ARj,t,) for every event by means of

the one-factor market model. The abnormal return is the difference between the actual return, and the estimated normal return calculated in (3). The formula of the abnormal return is as follows:

AR_j,t = Rj,t− ˆRj,t = Rj,t− ( ˆαj+ ˆβjRm,t) (4) Where ARj,t is the abnormal return on the stock at time t, Rj,t is the actual

return at time t and Řj,t is the estimated abnormal return at time t.

3.1.2$Experiment$1:$Regression$

To examine whether there is indeed a so-called World-Cup-Effect in the DAX 30 two regressions are performed on the dataset. In both regressions the abnormal return is used as dependent variable. Because the World Cup is every year in the same period (the beginning of summer), it is important to rule out any possible seasonal effects (Kim and Park, 1994). Therefore Belgium is used as control variable. Belgium will be a good proxy to show trader (ir)rationality because they were not part in most

previous World Cup tournaments.

The first regression examines whether there is a difference between a World Cup match played by Germany and a World Cup match played by any other team than Germany. The estimated regression formula looks as follows:

AR_! = ! γ_!+!γ_!FIFA +!γ_!GER +!!_! (5)

Where Rt is the daily return on day t, FIFA is a binary dummy variable added

(18)

involve the German team, GER is a binary dummy variable as well, added that takes the value of 1 if a FIFA World Cup match was played by the German team.

The second regression examines whether there is a different influence on the stock exchange for different match outcomes (a win, a lose or a draw). The estimated regression formula looks as follows:

AR!= ! β!+!β!WON +!β!LOSS +!!! (6)

Where Rt is the daily return on day t, WON is a binary dummy variable taking

the value of 1 if a match is won, LOSS is a binary dummy variable as well, taking the variable 1 if a match is lost, and therefore if a match ended in a draw, both variables WON and LOSS take up a value of 0.

!

3.1.3$Experiment$2:$Event$Study$

Besides the regression to see if there is a World-Cup-effect, an event study is

performed as well. An event study is another way to measure the relationship between World Cup results and stock returns.

For the event study the abnormal return is used as well. However, the event window does not comprise more than one trading day. Therefore, the cumulative abnormal returns are not calculated but the average abnormal return (AAR) is. The AAR is a better estimate to assess the general validity of the return reaction in relation to the event. It totals every abnormal return in the sample and is divided by the

amount of events in the cross-section. The formula looks as follows:

AAR_t = 1

N _i=1ARi,t

N

∑

(7)

Where N represents the amount of events in the sample.

In order to test the significance of the abnormal returns the TAAR is calculated. This is done to evaluate the level of statistical significance, whether the events causes systematic abnormal returns. The fundamental assumption of independent and

(19)

identical distributed abnormal returns, following a normal distribution with mean 0, lies at the heart of the trustworthiness of this test (Van der Sar, 2011). A statistical significant AAR indicates that there are positive or negative systematic abnormal returns in the event period. This is calculated by the following formula:

TAARt=

AARt

st / N

(8)

Where St portrays the estimated standard deviation and is derived by taking

the square root of (9).

s_t2 = 1

N −1 (ARi,t− AARt) 2

i=1 N

∑

(9)

Every model has its strengths and weaknesses. The first assumes cross-sectional independence, but at the same time does not neglect the variance in the estimation period, while the latter two tests rise from the assumption of requiring a normal distribution. One can loosen the assumption of normal distribution in the parametric tests, as well with a proper large amount of events in the cross-section, as according to the Central Limit Theorem the asymptotic distribution becomes standard normal (Van der Sar, 2011). To address the variation in abnormal returns in all of the

abovementioned regressions, the ordinary least squares (OLS) method is used. By using the OLS-method, several assumptions have to be fulfilled to get unbiased, efficient and consistent outcomes. Besides ensuring that there is no perfect

multicollinearity and that the average value of the errors is equal to zero by including a constant, the regressions are inspected for the presence of heteroscedasticity and serial correlation as well.

3.1.4$Welch$tBtest$

Besides the regressions and the event study to see whether the World Cup has a significant effect on the stock exchange, a Welch t-test is performed as well. The Welch t-test examines whether there is a significant difference between the DAX and the BFX. It compares the means of two samples. The Welch t-test assumes that both

(20)

samples are normally distributed. Furthermore, it accounts for unequal variances between the two samples. Hence, the test is suitable to compare the TAARs of Germany and Belgium. The t-statistic can be calculated with the following formula:

! = !̅!!!̅! !!! !!!!! ! !! (10)

!̅!, !!and !! refer to the average, the standard deviation and the sample size

of the German sample, whereas !̅_!, !_!and !_! represent the average, the variance and the sample size of the Belgium sample.

3.2'Dataset'

The data used to perform both experiments is retrieved from Yahoo finance. Data over a period of 24 years is used (09/04/2015 – 09/04/1991). During this period a total of six World Cup tournaments were held. In this period Germany not only hosted the World Cup (2006) but they have won the World Cup as well (2014). It will be interesting to see whether these years have a slightly higher (or lower) effect on the returns in the DAX 30 index. For the same period the returns of the S&P500 are retrieved from Yahoo finance, as the S&P500 is used as the market proxy in order to estimate the abnormal return.

3.3'Hypothesis''

As previous studies have shown an effect between investor sentiment and stock return a so-called World-Cup-Effect is expected. This can be explained due to irrational choices people make when they are influenced by other factors, like for instance sports.

H_! = ! β_!"!#= ! β_!"#= !0 H_! = ! β_!"!#= ! β_!"# ≠ 0

Besides the World-Cup-Effect, I expect a more significant effect on the DAX 30 index when the German team played a match on the day before, than if another team played.

(21)

H_! = ! β_!"!# = ! β_!"# H_! = ! β_!"!# < β_!"#

As a result of the event study, a World-Cup-Effect is expected as well. Therefore the hypothesis of the event study are as follows:

H_! = ! γ_!"# = ! γ_!"## = !0 H_! = ! γ_!"# = ! γ_!"## ≠ !0

Besides expecting significant outcomes of the variables, I expect the following hypothesis to hold:

• H1: Wins result in positive abnormal returns on the subsequent trading day. • H2: Draws and losses result in negative abnormal returns on the subsequent

trading day.

(22)

4.'Model'Specifications'

This part of the thesis is about whether the OLS-assumptions hold for both

experiments. The first part is about the OLS-assumptions in general, the second part is about whether the OLS-assumptions hold for experiment one, and the third part is about whether the OLS-assumptions hold for experiment two.

4.1'OLSJAssumptions'

In order to get statistically reliable results from the regressions the OLS-assumptions must hold for both experiments. The OLS-assumptions are as follows (Stock and Watson, 2012):

1. The error term is a random variable with E(εi | Xi) = 0.

2. (Yi, Xi) are independent and identically distributed.

3. Large outliers are unlikely.

4. There errors are homoscedastic; meaning, var(εi | Xi) = σ 2

5. The covariance between the errors terms is zero; cov(εi, εj) = 0, i ≠ j.

Besides the assumptions, it is also important that there is no perfect multicollinearity. If multicollinearity occurs, it is impossible to compute the OLS estimator. For both experiments unbiasedness and consistency must be proven for the OLS-assumptions to hold.

4.2 Calculating Abnormal Return

The model of experiment one is as follows:

R_!,! = ! β_!+!β_!R_!,!+!ε_!

The first step is to prove that the error term is a random variable with E(εi | Xi) = 0.

The method to prove this is by minimizing the least squared residuals:

(23)

ε_! = ! R_!,!− β_!−!β_!!_!,!

Minimize!SSR = ! (R!,!− R!,!)!

( R_!,!− β_!−!β_!R_!,!)!

2( R_!− β_!−!β_!R_!,!)(−1)

ε_! = 0

Hence εi = 0, and therefore the average OLS-residuals are equal to zero. Besides

proving that the average residuals are equal to zero, consistency and unbiasedness need to be proven as well.

E β_!",! !X = E S!" S_!! X!) E 1 n − 1 (X!− X)(R!− R)! 1 n − 1 (X!− X)! X) E (X!− X)(R!− R)! (X_!− X)! X) X_! − X E β_!+ β_!",!X + ε_!− β_!− β_!",!X − ε X) (X_!− X)! X!− X β!",!X + E ε! !X −!β!",!X −1_{n! E ε}! !X)] (X_!− X)! X_!− X β_!",![X_!− X] (X_!− X)! = ! β!"!#

(24)

E β_!",! = !E[E β_!",! !X)] → E[β_!",!] = ! β_!",!

Var β_!",! = ! σ!!

nσ_!!! → 0, when!n →⋈

Thus, both consistency and unbiasedness are proven for βRm,t.

4.3'Regression'Models'

The model of experiment two is as follows:

AR_!"# = ! γ_!+!γ_!FIFA +!γ_!GER +!!_!

Just as in experiment 1 the first step is to prove that the error term is a random variable with E(μi | Xi) = 0. The method to prove this is by minimizing the least

squared residuals: AR_!"# = ! γ_!+!γ_!FIFA +!γ_!GER +!!_! u_! = ! AR_!"#− γ_!−!γ_!FIFA −!γ_!GER Minimize!SSR = ! (R_!− R_!)! ( AR!"#− γ!−!γ!FIFA −!γ!GER)! 2( AR!"# − γ!−!γ!FIFA −!γ!GER)(−1) u_! = 0

(25)

Hence μi = 0, and therefore the average OLS-residuals are equal to zero. Besides

proving that the average residuals are equal to zero, consistency and unbiasedness need to be proven as well.

E γ_!"!# !X = E S!" S_!! X!) E 1 n − 1 (X!− X)(R!− R)! 1 n − 1 (X!− X)! X) E (X!− X)(R!− R)! (X_!− X)! X) X_!− X E γ_!+ γ_!"!#X + u_!− γ_!− γ_!"!#X − u X) (X_!− X)! X!− X γ!"!#X + E u! !X −!γ!"!#X −1_{n! E u}! !X)] (X_!− X)! X_!− X γ_!"!#[X_!− X] (X_!− X)! = ! γ!"!# E γ_!"!# = !E[E γ_!"!# !X)] → E[γ_!"!#] = ! γ_!"!# Var γ_!"!# = ! σ!! nσ_!!! → 0, when!n →⋈

Thus, both consistency and unbiasedness are proven for γFIFA, all is equal for γGER. As:

AR_!= ! β_!+!β_!WON +!β_!LOSS +!!_!

is the same multiple linear regression line as the regression line with the variables γFIFA and γGER, the same conditions hold. Therefore βwon and βLOSS are unbiased and

(26)

5.'Results'

This part of the thesis shows the results of the tests. Chapter five is divided into two parts. The first part shows the results of the regressions. The second part shows the result of the event study.

5.1'Regression'Results'

In the first part of the results the results of the different regressions are discussed. As the correlation table in appendix D1 shows, there is no case of multicollinearity. Therefore the regressions can be performed.

5.1.1$Regressions$to$calculate$abnormal$return$

In order to calculate the abnormal return two regressions were performed. One with the DAX as dependent variable and the S&P500 as independent variable, and one with BFX as dependent variable and the S&P500 as independent variable. The most important results can be found in the table below:

β0 βRm,t R 2 t-value DAX 0.000 (0.000) 0.027 (0.016) 0.000484 β0: 2.286 βRm,t: 1.677 BFX 0.000 (0.000) -0.019 (0.013) 0.000361 β0: -1.733 βRm,t: -1.446

Table 1: regression output calculating abnormal return

The estimated regression line of the DAX is as follows:

R_!"# = 0.000 + 0.027β_!&"#$$

This means that every time the S&P500 goes up by 1, the DAX goes up by 0.027. However, with a significance level of five percent the estimated coefficient βS&P500 is

not statistically significant. Therefore the null-hypothesis cannot be rejected6

. !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!

6_{!In!order!to!reject!the!nullUhypothesis!with!a!significance!level!of!five!percent,!the!tUvalue!found!needs!to!be!larger!than!}

(27)

Another way to evaluate the model is with the R-squared. The R2

is an indication of how close the data is fitted to the regression line. The estimated R2

=0.000484, this means that 0.0484% of the fluctuations in the DAX index are explained by the S&P500. This result clearly shows that the association between the DAX and the S&P500 is weak.

The estimated regression line of the BFX is as follows:

R!"# = 0.000 − 0.019β!&#$%%

This means that every time the S&P500 goes up by 1, the DAX drops by 0.019. However, with a significance level of five percent the estimated coefficient βS&P500 is

not statistically significant. Therefore the null-hypothesis cannot be rejected.

This regression can also be evaluated with the R2_{. The estimated R}2_=0.000361,

this means that 0.0361% of the fluctuations in the BFX index are explained by the S&P500. This result shows that the association between the BFX and the S&P500 is weak.

5.1.2$Regression$with$the$variables$FIFA$and$GER$

To examine whether the games played at a World Cup have any effect on the return of the DAX two regressions were performed as well. The first one with the abnormal return on the DAX as dependent variable and the dummy variables FIFA and GER as independent variables. The second one is a regression with the abnormal return on the BFX as dependent variable, and FIFA and GER as independent variables. The results are listed in the following table:

β0 βFIFA βGER R 2 t-value DAX 0.000 (0.000) -0.002 (0.002) 0.001 (0.003) 0.000196 β0: 2.408 βFIFA: 0.315 βGER: 0.660 BFX 0.000 (0.000) 0.000 (0.001) 0.000 (0.002) 0.000004 β0: -1.676 βFIFA: -0.079 βGER: -0.121

(28)

The estimated regression line of the abnormal return on the DAX is as follows:

AR_!"# = 0.000 − 0.002β_!"!#+ 0.001β_!"#

The estimated regression line shows that every time a World Cup match is played return on the DAX index drops by 0.002, and whenever the German team plays a match the DAX index increases by 0.001. Nonetheless, the corresponding t-values are low and rejected at a significance level of five percent. Therefore the results are not statistically significant, and the null-hypothesis cannot be rejected.

The R2

of the regression takes up a value 0f 0.000196. This means that 0.0196% of the fluctuations on the DAX index can be explained by a match on a World Cup (played by Germany or another team). This R2

shows a weak association between the DAX index and played World Cup matches.

The estimated regression line of the abnormal return on the BFX is as follows:

AR_!"# = 0.000 − 0.000β_!"!#+ 0.000β_!"#

The estimated regression line shows that every time a World Cup match is played return on the BFX index stays the same, and whenever the German team plays a match the BFX stays the same as well. These results may suggest that there is no link between World Cup games played and the BFX index. However, the t-statistics are low and the null-hypothesis is considered to be low so the results are not statistically significant.

The R2

of the regression is 0.000004. This means that 0.0004% of the fluctuations on the BFX index can be explained by a match on a World Cup. This R2

shows a weak association between the BFX index and played World Cup matches.

5.1.2$Regression$with$the$variables$WIN$and$LOSS$

In order to examine whether match outcomes of World Cup games played by

Germany influence the German stock exchange, two regressions were performed. The first one with the abnormal return on the DAX as dependent variable and the dummy variables WIN and LOSS as independent variables. The second one is a regression

(29)

with the abnormal return on the BFX as dependent variable, and WIN and LOSS as independent variables. The results are listed in the following table:

γ0 γWIN γLOSS R 2 t-value DAX 0.000 (0.000) 0.002 (0.004) -0.002 (0.007) 0.000081 γ0: 2.314 γWIN: 0.608 γLOSS:-0.314 BFX 0.000 (0.000) 0.000 (0.003) -0.004 (0.006) 0.000001 γ0: -1.677 γWIN: -0.150 γLOSS:-0.727

Table 3: regression output abnormal return and dummy variables WIN and LOSS

The estimated regression line of the abnormal return on the DAX is as follows:

AR_!"# = 0.000 + 0.002γ_!"#− 0.002γ_!"##

The estimated regression line shows that every time a World Cup match is won, the DAX index increases by 0.002. It also shows that when a World Cup match is lost, the DAX index drops by 0.002. However, the t-values corresponding with the

estimated coefficients are considered to be low. Therefore, with a significance level of five percent the null-hypothesis cannot be rejected.

The corresponding R2

with this regression is 0.000081; this means that

0.00081% of the fluctuations in the DAX index can be explained by the outcome of a World Cup match. Considering the R2

, the match outcome shows a weak association with the DAX index and the outcome of a World Cup game.

The estimated regression line of the abnormal return on the BFX is as follows:

AR_!"# = 0.000 + 0.000γ_!"#− 0.004γ_!"##

The estimated regression line shows that every time a World Cup match is won, nothing happens on the BFX index. At the same time it shows that when a World Cup match is lost, the BFX index drops by 0.004. However, the t-values corresponding

(30)

with the estimated coefficients are considered to be low. Therefore, with a significance level of five percent the null-hypothesis cannot be rejected.

The corresponding R2

with this regression is 0.000001; this means that 0.00001% of the fluctuations in the BFX index can be explained by the outcome of a World Cup match. Considering the R2

, the match outcome shows a weak association with the BFX index and the outcome of a World Cup game.

5.2'Results'Event'Study'

Besides the regressions an event study was performed as well. The outcome of the event study can be found in the table below:

AAR S TAAR

DAX 0.0433 15.1043 0.0703

BFX -0.2556 8.9104 -0.0703

Table 4: outcome event study

The value of the DAX is 0.0703, and the value of the BFX is -0.0703. Both t-values are low, and at a significance level of five percent the null-hypothesis cannot be rejected. Resulting from this event study it cannot be concluded that the World Cup has an effect on neither the DAX nor the BFX.

Besides the event study a Welch t-test is performed. The Welch t-test examines whether there is a significant difference between the DAX and the BFX. It compares the means of two samples. The outcome of the Welch test can be found in the table below:

DAX-BFX

Welch t-score 3.0631

Table 5: outcome Welch t-test

The t-score of the Welch test is 3.0631. This shows a statistically significant result at a significrance level of five percent. This means that there is a significance difference between the DAX and the BFX.

(31)

6.'Conclusion'and'Discussion

This thesis examines whether the FIFA World Cup has an effect on the German stock exchange (DAX). To estimate if there is any effect, two regressions were performed on the data. The first regression performed tested whether a World Cup match played by Germany, or a World Cup match played by any other team but Germany has an effect on the DAX. The second regression performed examined whether a win, draw or loss by Germany in a World Cup game has any effect on the DAX. Besides the regressions an event study, to estimate the so-called World Cup effect, was done as well. As a control variable the same tests were run on Belgian stock exchange (BFX). Finally, a Welch test is performed to check whether there is a significant difference between the DAX and the BFX.

Both regressions did not show a statistically significant relation between the variables. The first regression does not provide enough evidence to assume that World Cup games, either played by Germany or any other team, have any statistically

significant effect on the DAX or BFX. The second regression does not show a statically significant relationship either. It does not provide enough evidence to assume that a win, a loss or a draw in a World Cup game has an effect on the DAX or the BFX.

Secondly, the event study was performed. The event study resulted, just like the regressions, in low t-values and therefore no statistically significant evidence to conclude that the World Cup has an effect on the DAX or BFX. After the event study the Welch t-score was calculated. The Welch t-score did result in a statistically significant outcome. Therefore it can be concluded that there is a significance difference between the DAX and the BFX.

Further research may investigate why there is a difference between the DAX and the BFX. As the games played and the match outcome do not show a significant

statistically relationship, it can investigate whether that difference is explained by match events (a red or yellow card, a penalty-kick given, a goal scored etc.) or that the result is an outcome of the difference in trading frequency during the World Cup.!

(32)

Bibliography

Ashton, J. K., Gerrard, B. and Hudson, R. (2003). Economic Impact of National Sporting Success: Evidence from the London Stock Exchange. Applied

Economics Letters, 10 (12), 783-785.

Ashton, J. K., Gerrard, B. and Hudson, R. (2010). Do National Soccer Results Really Impact on the Stock Market? Applied Economics. 43 (26), 3709-3717.

Baade, R. A. and V. A. Matheson (2004). The Quest for the Cup: Assessing the Economic Impact of the World Cup. Regional Studies, 38 (4), 343‐354. Beremut, M. H. and Ceylan, N. B. (2012). Effects of Soccer on Stock Markets:

the Return-Volatility Relationship. The Social Science Journal, 49 (3), 368- 374.

Bernile, G. and Lyandres, E. (2011). Understanding investor sentiment: The case of soccer. Financial Management, 40 (2), 357-380.

Edmans, A., Garcia, D. and Norli, O. (2007). Sports Sentiment and Stock Returns.

The Journal of Finance, 62 (4), 1967–1998.

Ehrmann, M. and Jansen, D. (2012). The Pitch Rather Than The Pit: Investor Attention During FIFA World Cup Matches. Working Paper. Available at: https://www.ecb.europa.eu/pub/pdf/scpwps/ecbwp1424.pdf

EMBRATUR.!(2014).!A!Copa!das!Copas!foi!um!sucesso.!Available!at:!

http://www.embratur.gov.br/piembratur/opencms/salaImprensa/artigo s/arquivos/A_Copa_das_Copas_foi_um_sucesso.html.!!

FIFA. (2014). FIFA World Cup Brazil™ in numbers. Available at:

http://www.fifa.com/worldcup/news/y=2014/m=9/news=2014-fifa-world-cup-braziltm-in-numbers-2443025.html

FIFA. (2002). Preliminary matches for 2002 FIFA World Cup Korea/Japan. Available at: http://www.fifa.com.

Fourie, J. and Santana-Gallego, M. (2011). The impact of mega-events on tourist arrivals. Tourism Management, 32 (6), 1364-1370.

Fock, H., Klein, C. and Zwergel, B. (2008). Reconsidering the impact of national soccer results on the FTSE 100. Applied Economics, 41 (25), 3287-3294.

(33)

Government of South Africa. (2014). About Government. Available at: http://www.gov.za/about-government

Hirshleifer, D. (2001). Investor Psychology and Asset Pricing. Journal of Finance,

56 (4), 1533-1597.

Hudson Company. (2006). The impact of sport on the work place. Available at: http://www.sirc.org/publik/sport_and_the_workplace.pdf.

Kaplanski, G. and Levy, H. (2010). Exploitable Predictable Irrationality: The FIFA World Cup Effect on the U.S. Stock Market. Journal of Financial and

Quantitative Analysis, 45 (2), 535-553.

Kim, C. and Park, J. (1994). Holiday Effects and Stock Returns: Further Evidence.

Journal of Financial and Quantitative Evidence, 29 (1), 145-157.

Klein, C., Zwergel, B. and Heiden, S. (2009b). On the existence of sports sentiments: the relation between football match results and stock index returns in Europe.

Review of Managerial Science, 3 (3), 191-208.

McKinsey & Company (2015). The Bundesliga as a growth engine, the significance of professional football for the German economy.

OECD Oberserver, and Nene, N. (2013). After the final: reflections on South Africa’s soccer World Cup. Available at:

http://www.oecdobserver.org/news/fullstory.php/aid/4181/After_the_final Rose, A. and Spiegel, M. (2009). The Olympic Effect. National Bureau of Economic

Research Working Paper No. 14654.

Rose, A. and Spiegel, M. (2010). The Olympic trade effect. Journal of Finance and

Development, 121 (553), 652-677.

Scholten, B. and Peenstra, W. (2009). Scoring on the stock exchange? The effect of football matches on stock market returns: an event study. Applied Economics,

41 (25), 3231-3237.

Sports and Recreation South Africa (SRSA). (2012). 2010 FIFA World Cup Country Report. Available at:

http://www.ssra.gov.za/MediaLib/Home/DocumentLibrary/SRSACountryRep ort2013-withcover.pdf

Stadtmann, G. (2006). Frequent news and pure signals: evidence of a publicly trade football club. Scottish Journal of Political Economy, 53 (4), 485–504.

(34)

Appendix'

A.1 Graph of the abnormal return.

This graph shows the movement of the abnormal return between the DAX and the BFX.

B.1 Summary statistics of the DAX return and the S&P500.

This table shows the summary statistics of the variables DAX return and S&P500.

B.2 Regression output of the DAX return and the S&P500.

These tables show the regression output of the regression in which, DAX return is used as the dependent variable and the S&P500 as the independent variable.

B.3 Summary statistics of the BFX return and the S&P500.

This table shows the summary statistics of the variables BFX return and S&P500.

B.4 Regression output of the BFX return and the S&P500.

These tables show the regression output of the regression in which, BFX return is used as the dependent variable and the S&P500 as the independent variable.

C.1 Summary statistics of the DAX abnormal return and the dummy’s FIFA and GER.

This table shows the summary statistics of the variables DAX abnormal return, the dummy-variable FIFA and GER.

C.2 Regression output of the DAX abnormal return and the dummy’s FIFA and GER.

These tables show the regression output of the regression in which, DAX abnormal return is used as the dependent variable and the dummy-variables FIFA and GER as the independent variables.

C.3 Summary statistics of the DAX abnormal return and the dummy’s WIN and LOSS.

This table shows the summary statistics of the variables DAX abnormal return, the dummy-variable WIN and LOSS.

C.4 Regression output of the DAX abnormal return and the dummy’s WIN and LOSS.

These tables show the regression output of the regression in which, DAX abnormal return is used as the dependent variable and the dummy-variables WIN and LOSS as the independent variables.

(35)

C.5 Summary statistics of the BFX abnormal return and the dummy’s FIFA and GER.

This table shows the summary statistics of the variables BFX abnormal return, the dummy-variable FIFA and GER.

C.6 Regression output of the BFX abnormal return and the dummy’s FIFA and GER.

These tables show the regression output of the regression in which, BFX abnormal return is used as the dependent variable and the dummy-variables FIFA and GER as the independent variables.

C.7 Summary statistics of the BFX abnormal return and the dummy’s WIN and LOSS.

This table shows the summary statistics of the variables BFX abnormal return, the dummy-variable WIN and LOSS.

C.8 Regression output of the BFX abnormal return and the dummy’s WIN and LOSS.

These tables show the regression output of the regression in which, BFX abnormal return is used as the dependent variable and the dummy-variables WIN and LOSS as the independent variables.

D.1 Table event study.

This table shows the outcome of the event study. It shows the AAR, s2

, s, and TAAR.

E.1 Correlation table.

This table shows the correlation between the used variables; DAX return, BFX return, S&P500 return, DAX abnormal return, BFX abnormal return, FIFA, GER, WIN and LOSS.

(36)

A.1'Graph'of'the'abnormal'return' '

'

!

B.1 Summary statistics of the DAX return and the S&P500

N Minimum Maximum Mean

Std. Deviation

Varia nce Stati

stic Statistic Statistic Statistic Std. Error Statistic

Statis tic Return S& P500 6047 -.0903497781 550307 .1158003696 072271 .0003504152 87734 .0001471765 18956 .0114448079 92929 .000 Return DAX 6080 -.0939938430 000000 .1140195390 000000 .0004370471 99258 .0001827058 70956 .0142463723 98410 .000 Valid N (listwise) 6047

B.2 Regression output of the DAX return and the S&P500

The estimated regression equation is as follows: R_DAX =αj+βjRm,t+εj U0.15! U0.1! U0.05! 0! 0.05! 0.1! 0.15! 1! 278! 555! 832! _1109! _1386! _1663! _1940! _2217! _2494! _2771! _3048! _3325! _3602! _3879! _4156! _4433! _4710! _4987! _5264! _5541! _5818! Abnor mal'Retur n'

Abnormal'Return'Germany'and'

Belgium'

ARUDAX! ARUBFX!

(37)

B.2.1 Variables Entered Model Variables Entered Variables Removed Method 1 Return S&P500b . Enter

a. Dependent Variable: Return DAX b. All requested variables entered.

B.2.2 Model Summary Model R R Square Adjusted R Square Std. Error of the Estimate 1 .022a _.000 _.000 .014267328815 083 a. Predictors: (Constant). Return S&P500

B.2.3 ANOVA

Model Sum of Squares df Mean Square F Sig.

1 Regression .001 1 .001 2.812 .094b

Residual 1.231 6045 .000

Total 1.231 6046

a. Dependent Variable: Return DAX

b. Predictors: (Constant). Return S&P500

B.2.4 Coefficients Model Unstandardized Coefficients Standardized Coefficients t Sig. B Std. Error Beta 1 (Constant) .000 .000 2.286 .022 Return S&P500 .027 .016 .022 1.677 .094

(38)

B.3 Summary statistics of the BFX return and the S&P500

Std. Deviation

Varia nce Stati

Statis tic Return S& P500 6047 -.0903497781 550307 .1158003696 072271 .0003504152 87734 .0001471765 18956 .0114448079 92929 .000 Return BFX 6074 -.0978347467 221706 .0798263491 685047 -.0002591556 25279 .0001480614 28750 .0115392964 21996 .000 Valid N (listwise) 6047

B.4 Regression output of the BFX return and the S&P500.

The estimated regression equation is as follows: RBFX =αj+βjRm,t+εj B.4.1 Variables Entered Model Variables Entered Variables Removed Method 1 Return S&P500b . Enter

a. Dependent Variable: Return BFX b. All requested variables entered.

B.4.2 Model Summary Model R R Square Adjusted R Square Std. Error of the Estimate 1 .019a .000 .000 .011561172930 593 a. Predictors: (Constant). Return S&P500

(39)

B.4.3 ANOVA

1 Regression .000 1 .000 2.090 .148b

Residual .808 6045 .000

Total .808 6046

a. Dependent Variable: Return BFX

b. Predictors: (Constant). Return S&P500

B.4.4 Coefficients Model Unstandardized Coefficients Standardized Coefficients t Sig. B Std. Error Beta 1 (Constant) .000 .000 -1.733 .083 Return S&P500 -.019 .013 -.019 -1.446 .148

a. Dependent Variable: Return BFX

C.1 Summary statistics of the DAX abnormal return and the dummy’s FIFA and GER

Std. Deviation

Varia nce Stati

Statis tic AR-DAX 6080 -.09376018987 12556 .11346576463 82648 .0004276373 38467 .0001826635 53226 .0142430727 00656 .000 X-FIFA 6080 0 1 .01 .001 .111 .012 X-GER 6080 0 1 .00 .001 .061 .004 Valid N (listwi se) 6080

(40)

C.2 Regression output of the DAX abnormal return and the dummy’s FIFA and GER

The estimated regression equation is as follows:

AR_!"# = ! γ_! +!γ_!FIFA +!γ_!GER +!!_! C.2.1 Variables Entered Model Variables Entered Variables Removed Method

1 X-GER. X-FIFAb . Enter

a. Dependent Variable: AR-DAX b. All requested variables entered.

C.2.2 Model Summary Model R R Square Adjusted R Square Std. Error of the Estimate 1 .014a .000 .000 .014244021366 601 a. Predictors: (Constant). X-GER. X-FIFA

C.2.3 ANOVA

1 Regression .000 2 .000 .595 .552b

Residual 1.233 6077 .000

Total 1.233 6079

a. Dependent Variable: AR-DAX

b. Predictors: (Constant). X-GER. X-FIFA

C.2.4 Coefficients Model Unstandardized Coefficients Standardized Coefficients t Sig. B Std. Error Beta 1 (Constant) .000 .000 2.408 .016 X-FIFA -.002 .002 -.013 -1.006 .315 X-GER .001 .003 .006 .440 .660

(41)

C.3 Summary statistics of the DAX abnormal return and the dummy’s WIN and LOSS

Std. Deviation

Varia nce Stati

Statis tic AR-DAX 6080 -.09376018987 12556 .11346576463 82648 .0004276373 38467 .0001826635 53226 .0142430727 00656 .000 X-WIN 6080 0 1 .00 .001 .051 .003 X-LOSS 6080 0 1 .00 .000 .026 .001 Valid N (listwi se) 6080

C.4 Regression output of the DAX abnormal return and the dummy’s WIN and LOSS

AR_!"# = ! β_!+!β_!WON +!β_!LOSS +!!_! C.4.1 Variables Entered Model Variables Entered Variables Removed Method 1 LOSS. X-WINb . Enter

a. Dependent Variable: AR-DAX b. All requested variables entered.

(42)

C.4.2 Model Summary Model R R Square Adjusted R Square Std. Error of the Estimate 1 .009a .000 .000 .014244867094 092 a. Predictors: (Constant). X-LOSS. X-WIN

C.4.3 ANOVA

1 Regression .000 2 .000 .234 .791b

Residual 1.233 6077 .000

Total 1.233 6079

a. Dependent Variable: AR-DAX

b. Predictors: (Constant). X-LOSS. X-WIN

C.4.4 Coefficients Model Unstandardized Coefficients Standardized Coefficients t Sig. B Std. Error Beta 1 (Constant) .000 .000 2.314 .021 X-WIN .002 .004 .008 .608 .543 X-LOSS -.002 .007 -.004 -.314 .754

(43)

C.5 Summary statistics of the BFX abnormal return and the dummy’s FIFA and GER

Std. Deviation

Varia nce Stati

Statis tic AR-BFX 6080 -.09789456942 06757 .07964936600 78230 -.0002522781 25706 .0001478897 49992 .0115316078 31044 .000 X-FIFA 6080 0 1 .01 .001 .111 .012 X-GER 6080 0 1 .00 .001 .061 .004 Valid N (listwi se) 6080

C.6 Regression output of the BFX abnormal return and the dummy’s FIFA and GER

AR_!"# = ! γ_! +!γ_!FIFA +!γ_!GER +!!_! C.6.1 Variables Entered Model Variables Entered Variables Removed Method

1 X-GER. X-FIFAb . Enter

a. Dependent Variable: AR-BFX b. All requested variables entered.

C.6.2 Model Summary Model R R Square Adjusted R Square Std. Error of the Estimate 1 .002a .000 .000 .011533485120 296 a. Predictors: (Constant). X-GER. X-FIFA

(44)

C.6.3 ANOVA

1 Regression .000 2 .000 .011 .989b

Residual .808 6077 .000

Total .808 6079

a. Dependent Variable: AR-BFX

b. Predictors: (Constant). X-GER. X-FIFA

C.6.4 Coefficients Model Unstandardized Coefficients Standardized Coefficients t Sig. B Std. Error Beta 1 (Constant) .000 .000 -1.676 .094 X-FIFA .000 .001 -.001 -.079 .937 X-GER .000 .002 -.002 -.121 .904

C.7 Summary statistics of the BFX abnormal return and the dummy’s WIN and LOSS

Std. Deviation

Varia nce Stati

Statis tic AR-BFX 6080 -.09789456942 06757 .07964936600 78230 -.0002522781 25706 .0001478897 49992 .0115316078 31044 .000 X-WIN 6080 0 1 .00 .001 .051 .003 X-LOSS 6080 0 1 .00 .000 .026 .001 Valid N (listwi se) 6080

(45)

C.8 Regression output of the BFX abnormal return and the dummy’s WIN and LOSS

AR_!"# = ! β_!+!β_!WON +!β_!LOSS +!!_! C.8.1 Variables Entered Model Variables Entered Variables Removed Method 1 LOSS. X-WINb . Enter

a. Dependent Variable: AR-BFX b. All requested variables entered.

' ! C.8.2!Model!Summary! Model R R Square Adjusted R Square Std. Error of the Estimate 1 .010a .000 .000 .011532982626 372 a. Predictors: (Constant). X-LOSS. X-WIN

' '

C.8.3!ANOVA!

1 Regression .000 2 .000 .275 .759b

Residual .808 6077 .000

Total .808 6079

b. Predictors: (Constant). X-LOSS. X-WIN

(46)

C.8.4!Coefficients! Model Unstandardized Coefficients Standardized Coefficients t Sig. B Std. Error Beta 1 (Constant) .000 .000 -1.677 .094 X-WIN .000 .003 -.002 -.150 .881 X-LOSS -.004 .006 -.009 -.727 .467

'

! !

D.1 Table event study

AAR S2

S TAAR

DAX 0.0433 228.1401 15.1043 0.0703

(47)

! !

The!Effect!of!the!FIFA!World!Cup!on!the!DAX! 46!

!

Return DAX Return BFX X-FIFA X-GER X-WIN X-LOSS AR-DAX AR-BFX

Return S&P500

Return DAX Pearson Correlation 1 .007 -.013 .006 .008 -.004 1.000** _.008 _.022

Sig. (2-tailed) .576 .314 .665 .533 .751 .000 .554 .094