• No results found

S PORT SENTIMENT AND STOCK RETURNS IN E UROPE

N/A
N/A
Protected

Academic year: 2021

Share "S PORT SENTIMENT AND STOCK RETURNS IN E UROPE"

Copied!
41
0
0

Bezig met laden.... (Bekijk nu de volledige tekst)

Hele tekst

(1)

1

S

PORT SENTIMENT AND STOCK RETURNS IN

E

UROPE

THE CASE OF EXPLOITING THE SOCCER SENTIMENT EFFECT

Thesis MSc Finance

Rijksuniversiteit Groningen

Peter Weststeijn s2050153 MSc Finance Supervisor: Dr. R.O.S. Zaal Date: July 2015 ABSTRACT

(2)

2

I.INTRODUCTION

(3)

3

theory entails that soccer results influence investors mood which in turn influences investors sentiment, their feeling on how the market will behave in the future. This investor sentiment of course affects the market.

In this paper I will look at this soccer sentiment effect from an investors point of view. An anomaly only becomes really interesting when it also has implications for the market and becomes practically useful. I will therefore study if the effect could also be used by investors to benefit from, something that previous paper did not fully cover. The asymmetrical characteristic of this effect should make it ideal for investment purposes. If, for example, one would invest in both teams playing against each other, the winning team will incur no effect from the sentiment effect whilst the losing one will incur a significant negative return. Edmans et al. (2007) however also note that the existence of transaction costs probably makes investing in it not worthwhile. This might explain why few articles go into the subject of investing with this knowledge. Only Kaplanski & Levy (2010) and Pantzalis & Park (2014) briefly do so. Kaplanski & Levy (2010) try to overcome this transaction-cost problem by looking at a period as a whole instead of per single match. The investment advice which is ultimately given entails shifting from the stock market to government bonds during an event period (World Cup) in order to avoid the financial markets during these sporting events and miss out on the negative returns. They therefore not exploit the effect as they claim with the title of their paper. I try to see whether there is a more lucrative and active investing strategy than the passive one recommended before. I do this by looking at elimination matches in international soccer tournaments. Next to this, my study differs from previous papers by only looking at European markets. I create a portfolio of six prominent soccer countries and study the effect of elimination in their home markets. If the soccer sentiment theory indeed exists it must at least be visible in these European markets I study. This because changes in mood, which is the core of the theory, should be biggest in countries where soccer lives most. Based on merchandize sales and TV audience it can be concluded that the countries I choose are on the top of that list.

The main research question of this thesis can be phrased as follows:

How can the soccer sentiment effect be exploited during the World Cup or European Championship?

(4)

4

methodology I employ to be able to answer the main research question. My first sub question therefore is:

Does an elimination from an international soccer tournament result in significant returns in the local market?

Kaplanski & Levy end both their articles (2010 and 2014) with a very short hypothesis about the future use of the World Cup anomaly. These, however, seem to contradict each other by first saying that the anomaly will probably not disappear over time and this “irrational market behavior will probably be with us for a long period to come” (2010, p.552) and later on stating that the anomaly might actually disappear. This brings me to the following sub question:

Is the effect still present in the European Championship of 2012 and the World Cup of 2014?

Different papers on this subject clearly have different hypothesis on how the soccer sentiment effect will behave in the future. Because of the growing attention scholars have given to the sport sentiment and the World Cup effect, sophisticated investors might have already reduced the effect of the anomaly or at least changed the price pattern of the index with respect to previous World Cups. It might also be the case that the anomaly is intensified as Kaplanski & Levy (2010) state and the market hasn’t become more efficient. Next to this there are no papers written about the latest European Championship and World Cup. The strategy my thesis proposes can only be relevant when the effect is still present and therefore the last two tournaments need to be studied more closely. My data consist of daily market index returns in combination with soccer results from international tournaments played from 1984 until 2014. Because I only consider six European countries I can also look at the European Championships. In order to estimate the impact of an elimination by one of the six countries I employ a regression using the OLS method.

I find that a national soccer loss indeed leads to a significant negative next-day return in the local stock market. This means that the soccer sentiment effect is present in the dataset which is used in this thesis. It is however not fully clear whether or not the effect is still present in the last two international tournaments, although it does seem like it. The effect is not large enough to be exploited using short selling, put options or future contracts in the local stock markets of the six considered European countries.

(5)

5

chapter IV which is followed by the results of this paper in chapter V. This paper will be concluded by answering the research questions and summarizing my main findings together with a short discussion part.

II.LITERATURE REVIEW

In this section I will discuss the relevant literature on the sport sentiment effect. I will start by explaining what some of the psychology literature reports about the impact of sport results on mood. After this, the link between mood and stock returns is investigated. In this way the indirect link between sport results and stock returns is shown which is one of the more recent area’s within behavioral finance. I will end with discussing papers that directly investigate the link between sport results and stock returns, especially in the case of soccer which is the focus of my thesis.

2.1SPORTS AND MOOD

Different articles in psychology discuss the link between sport results and a person’s mood. For example, Wann et al. (1994) shows that fans experience an increase in positive emotions after their team wins and an increase in negative emotions following a loss. In the latter, fans who identify themselves with their team experience emotions such as sadness and anger. Furthermore Schwarz et al. (1987) show that the results of the German national team during the World Cup of 1982 changed the subjects’ mood which even affects their global life satisfaction.

(6)

6

An interesting characteristic that is present in the relationship between sports and mood is the asymmetrical effect that occurs after wins and losses. Edmans et al. (2007) describes it as following: “..while an increase in heart attacks, crimes, and suicides is shown to accompany sporting losses,

there is no evidence of improvements in mood of a similar magnitude after wins.” (p.1971). This

characteristic is especially present within a tournament format where a win only ensures a country of the next round and a loss can immediately eliminate a country from the tournament. This is the reasoning behind the assumption used in the soccer sentiment effect that a loss has a larger effect on the local stock market than a win.

2.2MOOD AND RETURN

(7)

7

changes. Kamstra, Kramer & Levi find a statistically and economically significant effect implying a one-day loss of 31 billion on the NYSE, NASDAQ and AMEX at the day the time change occurs. Based upon the literature discussed it can be concluded that in many studies a relationship is found between a mood variable and stock returns. Changes in investors’ mood can apparently impact the financial markets.

2.3SPORT AND RETURN

The last few years sports has been a much-discussed topic in this field of behavioral finance. Within these studies sports forms a proxy for mood instead of the weather or time changes. Ashton, Gerrard and Hudson (2003) were the first to document the effect of sports in this way by looking at the performance of the national English soccer team and the London stock exchange. In case of a win (loss) they found a statistically significant positive (negative) return on the stock exchange. Next to this finding, they show that more important matches, such as those on a tournament, have greater impact than less important games, such as friendlies. Edmans et al. (2007) went deeper into this subject by looking at the effect of international soccer, cricket, rugby and basketball games. They find a loss effect, significant negative returns after a loss, for all four sports but show that especially the soccer match outcomes lead to a strong market reaction. They find no significant returns after a win, making the effect asymmetrical. Because they control for pre-game expected outcomes they conclude that the effect is not driven by economic factors such as lost revenues or reduces productivity or because of rational investors acting on relevant information but due to the impact of the loss on investors’ mood. They study 1162 matches of 39 different countries from 1973 until 2004 and find next-day abnormal stock returns of -49 basis point, which is more than -7% in monthly terms. Similarly as with the study of Ashton et al. (2003), the authors find that important matches, such as in an elimination stage, have greater impact. The same holds for countries where soccer is of greater importance.

(8)

8

prices of the publicly traded Boston Celtics. Losing a basketball match leads an asymmetrical reaction and important matches have greater impact.

Several other sports were also investigated, documenting the same results as above. Chang, Chen, Chou & Lin (2012) found that the stock of firms with their headquarters geographically close to an American football (NFL) team react to match results due to investor sentiment. Especially a surprising or critical game loss leads to lower next-day returns. Pantzalis & Park (2014) looked at the four major professional sports leagues in the U.S. (NFL, MLB, NBA and NHL) and found high correlation between the sport performances and stock returns of firms with geographical proximity. Apart from this they conclude that the returns of those firms aren’t always fully information-based and often mispriced due to investor sentiment. This is especially the case for firms in areas where fan base support appears to be stronger.

Still most articles written are focused on soccer. Especially the international soccer events, such as the European Championship or World Cup, are ideal for investigating. This is the case because the enormous attention surrounding it makes it almost impossible to escape investors’ attention. Kaplanski & Levy (2010) continue on the work of Edmans et al. (2007) by looking at the presence of the soccer sentiment effect in the U.S. market from 1950 until 2007 and find a very large and highly

Figure 1

(9)

9

significant effect during the effect period of -2.58% compared to a normal return of +1.21% over the same period length. They show that for 12 out of 15 events (World Cups) the return is less than the riskless interest rate and at 14 out of the 15 events it is lower than the average return on equity (see Figure 1). With this in mind they advise investors to avoid the stock markets during a World Cup and invest that money in short-term treasury bills. When investing $1 in 1950 this strategy leads to $6.948 in 2007. To compare, a normal buy-and-hold strategy yields $4.386 during those 57 years. Four years later they published another article about the 2010 World Cup to see if inventors had already made use of this previously unknown anomaly (Kaplinki & Levy, 2014). They conclude that indeed the price pattern during the World Cup in 2010 was different than the previous ones, presumably due to sophisticated investors making use of the anomaly. Making an abnormal profit was, however, still possible.

Although the evidence of this sport sentiment anomaly is overwhelming, some authors also found contradicting results. Klein, Zwergel & Heiden (2009) and Viera (2012) for example find no significant results when testing for the effect of soccer results on stock returns and Gerlach (2011) does find the same returns as previously mentioned papers but doesn’t assign that to the match results. Difference in statistical models used (as Klein et al. (2009) mentions as a reason) and the fact that still many articles on this subject were published after these three appeared show that the sport sentiment effect is far from being declared untrue. It is still a hot topic for scholars, with far more publications concluding its existence then the other way around, that it remains interesting to investigate further. Based on this existing and above mentioned literature I present the following hypotheses.

H1: An elimination event creates a significant negative return in the local stock market of the eliminated European country.

This is the hypothesis drawn from the first sub question which tries to answer the question of whether or not the soccer sentiment effect is significant. The second hypothesis is based on the second sub question which looks into the existence of the effect during the last two international soccer tournaments.

H2: The soccer sentiment effect is present in the European Championship and World Cup of 2012 and 2014 respectively.

(10)

10

also mentions. The following hypothesis (H3) is based on the main research question of this thesis, whether or not the sentiment effect is the exploitable.

H3: A profitable investment strategy, which is superior in terms of return to the strategy proposed by

Kaplanski & Levy (2010), can be created by going short in several European indices during an international soccer tournament until the specific country is eliminated.

III.METHODOLOGY

The method and investment strategy I employ is based on a combination of the methods that Edmans et al. (2007) and Kaplanski & Levy (2010) use to treat this subject of investor mood and subsequently their sentiment on stock returns. Kaplanski & Levy (2010) investigated this effect by looking at the U.S. stock market using the New York Stock Exchange. The effect of losing however is larger in the local market of the losing country than the effect of the spill-over on the NYSE. I will therefore focus on the local stock markets, just as Edmans et al. (2007) does. Edmans et al., however, look at the effect on the stock market after every single game played. This gives them a large amount of data and many events to study. Although this is desirable when doing an event study, it is hardly useful for investors in practice because they can never alter their investment portfolio after every single game played in the world. This is why they also concluded that “even traders who face low transaction costs would find it challenging to take advantage of the

price drop” (p. 1992). Kaplanski & Levy (2010) recognized this problem and overcame this by

investing in a longer period than just the one single match. They considered investment periods, Event Period Effect Days (EPED) as they call it. This period is defined as all days of the World Cup, from the first day of the World Cup to the first day after the final game plus two additional trading days. With this approach of looking at periods as a whole they drastically reduced the number of investment steps, and thus transaction costs, compared to a strategy which entails adjustments after every match. With the latter you need to invest in both competing local markets and clear out your position after the match to profit from the asymmetrical negative return. This entails four investment steps per match. To illustrate, Edmans et al. (2007) looked at 1162 soccer matches. Following this strategy, consisting of 4648 actions, would therefore be quite time consuming and expensive with respect to transaction costs. Using periods, like Kaplanski & Levy (2010) do, only results in two investment steps per period (World Cup) per country.

(11)

11

investing until the specific country is eliminated it is certain that this impact is caught. Because of the asymmetric characteristics of the effect no money will be lost when the chosen country actually wins the World Cup. Following the methodology of Edmans et al. (2007) I will describe an event day as the day after the lost game to make sure the local market at least had a full trading day to process the elimination. In this way we ensure that we get a return which is based on a situation where the game outcome is known during the full day. From now on the game day will be described by t-1 and the day after the game day, the event day, by t.

To have a certain return I will assume a portfolio of several (N) countries. In this way I will at least have N-1 eliminated countries in the portfolio generating a negative return and at most 1 winning country whose market isn’t influenced by the game outcomes. Bearing transaction costs in mind I will only look at countries were football lives most, because here it has the largest chance to affect investors and thus stock returns (Edmans et al. shows that the ‘top 7’ global football countries induce significantly lower returns after a loss compared to the rest) . I also take European countries so there is more data available because I can include the European Championships. The countries that satisfy those criteria are Germany, England, The Netherlands, Spain, Italy and France. All of these countries are almost always in the top 10 of the FIFA world ranking or the ELO ratings ranking.

I will only look at the periods itself because I only care about the return that is realized between the first and last day of the period. What happens in the two years in between tournaments doesn’t matter for the strategy I am proposing. A period is defined as the period in which a country participates in the tournament and is still able to win (i.e. from the first tournament day until elimination). Next, I’ll gather the returns from the chosen countries during the time they participate in the World Cup or European Championship. The returns are taken because the value of the indices of the different countries are numerically different and not comparable. In order to have enough events for an event study, the different periods with returns of different years and countries are pooled together and treated as panel data. By using the elimination date from a country in a certain year as a dummy variable, the significance and magnitude of the market reaction can be determined. Using panel data has a few advantages. First of all it increases sample size because different cross-sections can be added together (Studenmund, 1997). Panel data can also be used to tackle a broader range of problems compared to normal time-series data and the combination of time-series and cross-sectional data can help to mitigate problems with multicollinearity and increase the power of the test (Brooks, 2002).

(12)

12

necessary. A special event could be the start of a war or perhaps the death of a monarch during the time of elimination. Other possible variables which could be thought of to affect the results but were already investigated by previous papers and proven to be insignificant, are omitted in my model. These are variables such as day-of-the-week or June-July effect which were already tested by Edmans et al. (2007) and Kaplanski & Levy (2010) and deemed insignificant. Next to this I will look at the influence of some outlier years by running the regressions with and without these outliers. It could be that 1 or 2 tournament periods with extreme return values are the cause of the (in)significant overall results.

ECONOMETRIC APPROACH

The null hypotheses of the regressions used are that the local stock markets are unaffected by the elimination of their national football team. This hypothesis relates to the theory that stock markets are efficient and investors rational. Any potential effects from the results of the matches are too small to affect the stock market or for investors to benefit from. The alternative hypothesis represents the theory earlier explained which stated that game outcomes affect investors’ mood which impacts the stock market. This results in an elimination leading to negative stock returns as is already hypothesized in H1.

To test this, the following regression is ran:

(13)

13

as an ordinary least squares (OLS). Stock index returns often also have time varying volatility which in panel data can be addresses with GLS using period weights. As a robustness check, the regression will be run with and without outliers to check whether a single extraordinary event day is distorting the results or not.

In order for an investor to profit from this effect in the future it is important that this soccer sentiment effect still exists. If the results from the regression described above are significant, they could still be caused by highly significant effects from previous tournament while the effect is already arbitraged away by investors during the last ones. Kaplanski & Levy (2014) for example already recognized changing prize patterns of the NYSE during the World Cup of 2010 compared to previous ones. If the market is efficient, as often assumed, investors should have recognized the opportunity for making profits and in this way dissolved the arbitrage. As far as I know, the World Cup of 2010 is the latest tournament studied by academics. The European Championship of 2012 and World Cup of 2014 can therefore be used to study the existence of the effect after the academic attention. First of all I can study the results of a regression where the events from these last two tournaments are defined as a unique variable. The regression now becomes:

𝑅

𝑖𝑡

= 𝑦

0

+ ∑

2

𝑦

1𝑖

𝑖=1

𝑅

𝑖𝑡−𝑖

+ 𝑦

2

𝑅

𝑚𝑡−𝑖

+ 𝑦

3

𝑅

𝑚𝑡

+ 𝑦

4

𝑀

𝑡

+ 𝑦

5

𝐸

𝑖𝑡1984−2010

+

𝑦

6

𝐸

𝑖𝑡2012−2014

+ 𝜀

𝑡

(2)

Where 𝐸𝑖𝑡1984−2010are the events from 1984 until 2010 and 𝐸𝑖𝑡2012−2014 the ones from 2012 and 2014.

Next to running the regression, the price patterns of the last two tournaments can be studied by looking at the change in returns at elimination and studying the graph of the index. When comparing those with previous tournaments, change in investor behavior may perhaps be visible.

(14)

14

IV.DATA

For this study data is needed on national soccer results in previous European Championships and World Cups and on national stock indices. The data on the indices of the six countries under consideration can be found on Datastream. The commonly used and large index of the specific country is taken, which are shown in table 1. I chose to use the price index instead of the return index because for most countries this index contains more data (older values). This gives us the opportunity to include more tournaments resulting in more events, which is desirable since the lack of it is often the pitfall in an event study. The assumption is made that the paid out dividends, which are concluded in the return index, don’t significantly influence the (magnitude of the) sentiment effect. The continuously compounded daily returns of the indices are then calculated using the following equation:

𝑅

𝑖𝑡

= ln(

𝑃𝑖𝑡

𝑃𝑖𝑡−1

)

(3)

(15)

15

Table 1 Data overview

Considered countries with corresponding indices, their continuously compounded mean daily returns (percentage) and amount of relevant and used eliminations. The mean returns and eliminations are gathered during tournament periods (European Championshipsand World Cup) from 1984 until 2014.

Country Index Time Series Begins Mean Return Elimination events

Netherlands AEX 12 – 06 – 1984 0,0548 % 9

England FTSE 100 12 – 06 – 1984 -0,0439 % 10

Germany DAX 30 12 – 06 – 1984 -0,0289 % 11

France CAC 40 10 – 06 – 1988 -0,0478 % 7

Italy Milan Comit 30 17 – 16 – 1994 -0,0084 % 8

Spain IBEX 35 10 – 06 – 1988 0,1365 % 6

All countries 0,0062 % 51

(16)

16

the number of countries times the number of tournaments (6*16 = 96). Ultimately, 51 elimination events remained to be used as data in this study.

Table 2

Elimination Events criteria

This table shows the criteria I used to get to the elimination events that were used as data. There are 96 possible eliminations during the tournaments from 1984 until 2014. 12 times a reviewed country didn’t manage to qualify and 11 times a country ended up winning the tournament. Omitting events because of data availability, the next day being a weekend day and two other cases (discussed in the text above), resulted in 51 usable events.

Elimination Events N

Total Possible Events 96

Did Not Qualify - 12

Winning countries - 11

Total Eliminations 73

No index data available - 6

Next day is weekend day - 14

Other cases - 2

Total usable eliminations 51

The specific elimination dates per country and tournament which were used can be found in Appendix B. Note that the tournaments from 1984 until 1990 documented under Germany were actually played by West-Germany.

The descriptive statistics of the data is shown in table 3. The data consist of 823 tournament days without an elimination event and 51 with an elimination events. Furthermore, the continuously compounded daily mean returns are given with their standard deviation, maximum and minimum values. The average daily mean return in the non-elimination game days is 0,0175% compared to a return of -0,1762% for a trading day after an elimination. An elimination event thus seems to cause a lower index return than other tournament days do.

Table 3 Descriptive statistics

The number of elimination events and tournament days without an elimination, from 1984 until 2014, which are used as data. Accompanied by their continuously compounded daily mean returns, standard deviation, maximum and minimum values.

No elimination games Elimination

N 823 51

Mean Return 0,0175 % -0,1762 %

SD 1,1897 % 1,2804 %

Maximum 5,7218 % 4,2401 %

(17)

17

V.RESULTS

This chapter will present and discuss the results found in the analysis. These will be used to answer the research questions. The first research question is whether or not the soccer sentiment effect creates significant returns, and is elaborated on in part A. This is followed by part B where the existence of the effect today is discussed. Finally, part C looks for ways to exploit the sentiment effect.

A. ARE THERE SIGNIFICANT RETURNS AFTER AN ELIMINATION?

In order to study the first research question, a regression is ran using Eviews. The correlation matrix of the variables used in this regression (formula 1) can be found in table 4. There appears to be a strong correlation (ρ = 0,6909) between the (first order) lagged return of the local indices, 𝑅𝑖𝑡−1, and the lagged return of the European Market index, 𝑅𝑚𝑡−𝑖. Although GLS usually provides unbiased estimates in the case of autocorrelation, I will still account for this multicollinearity manually to see how robust the results are. In order to deal with a strong correlation between two independent variables, one of them needs to be dropped out of the regression. Because 𝑅𝑖𝑡−1 has the lowest impact (lower coefficient) on the dependent variable (𝑅𝑖𝑡), it is removed from the regression. The results of this regression, denoted as regression 2, can be found in table 5.

Table 4 Correlation Matrix

This matrix gives the correlations between the variables in formula 1 (the first regression in the methodology part).

Ri Rit-1 Rit-2 Rmt-1 Rm M Ei Ri 1 0,0204 -0,0071 0,0914 0,6979 -0,0641 -0,0402 Rit-1 0,0204 1 0,0201 0,6909 0,0763 0,0222 0,0354 Rit-2 -0,0071 0,0201 1 0,0732 -0,0184 -0,0297 0,0795 Rmt-1 0,0914 0,6909 0,0732 1 0,0960 0,0264 0,0291 Rm 0,6979 0,0763 -0,0184 0,0960 1 0,0262 -0,0195 M -0,0641 0,0222 -0,0297 0,0264 0,0262 1 0,0920 Ei -0,0402 0,0354 0,0795 0,0291 -0,0195 0,0920 1

(18)

18

(which is approximated by Datastream’s European market index). The Monday effect also is significant at a 1% level and has a coefficient of -0,2398. This means that on Mondays the return is on average 0,02398% lower than on other days. The Monday having, on average, lower returns than other days is in agreement with the Monday effect theory. The most important result is the value and probability of the elimination factor, 𝐸𝑖. The coefficient is -0,1416 and the t-statistic -2,3613, making it statistically significant at a 5% confidence level. It means that the an elimination leads to a 14,16 basis point decrease in the local stock market. That an elimination lead to a negative next day index return is in accordance with the theory tested in this thesis. We can reject the null hypothesis that a loss in an elimination game has no effect on index returns. The R-squared is 0,8134 and the probability of the F-statistic highly significant.

Table 5

Main Regression Results

This table reports the results of the following regression: 𝑹𝒊𝒕 = 𝒚𝟎+ ∑ 𝒚𝟏𝒊

𝟐

𝒊=𝟏

𝑹𝒊𝒕−𝒊+ 𝒚𝟐𝑹𝒎𝒕−𝒊+ 𝒚𝟑𝑹𝒎𝒕+ 𝒚𝟒𝑴𝒕+ 𝒚𝟓𝑬𝒊𝒕+ 𝜺𝒕

where 𝑹𝒊𝒕 is the continuously compounded daily return; 𝒚𝟎is the intercept; 𝑹𝒊𝒕−𝟏 and 𝑹𝒊𝒕−𝟐 are the lagged index returns

(one and two days); 𝑹𝒎𝒕 is the return of Datastream’s European market index; 𝑹𝒎𝒕−𝒊 the previous day European market

return; 𝑴𝒕 is a dummy variable to account for the Monday effect and 𝑬𝒊 is a dummy variable for an elimination event. The

regression coefficients, t-values and corresponding probabilities are given. *, ** and *** indicate a 10%, 5% and 1% level of significance, respectively. The results are obtained using Ordinary Least Squares with panel data. The data consist out of continuously compounded daily returns of indices from six prominent European soccer countries (the Netherlands, England, Germany, France, Italy and Spain) together with their eliminations from European Championships and World Cups from 1984 until 2014. The observed period includes 823 non-elimination games and 51 elimination events. The null hypothesis is that the returns are unaffected by eliminations from their national soccer team. In Regression 1 all the variables are included, regression 2 accounts for multicollinearity.

(19)

19

When looking at regression 2 in table 5, the one without the first order lagged index return 𝑅𝑖𝑡−1, almost all the independent variables included are still statistically significant. The intercept is significant at a 1% level and causes the local index return (dependent variable) to increases with 0,0432% when all the other independent variables stay unchanged at zero. The remaining lagged index return variable, 𝑅𝑖𝑡−2, is not significant anymore. Both the European index variable and its lagged component stay highly significant at a 1% confidence level, as well as the Monday variable which still yields a negative return. 𝐸𝑖, the elimination dummy variable, becomes a bit less significant compared to before but still remains significant at a 5% level. The coefficient now is -0,1387. The R squared is still high and the probability of the F-statistic is still highly significant.

To test the robustness of the results, a regression without the outliers is ran. For every cross-section (country), the lowest and highest returns are taken out of the elimination event sample (𝐸𝑖) and put into a new created variable. This new created variable, which stands for the outliers, is denoted as 𝐽𝑡. The regression now looks as follows:

𝑅

𝑖𝑡

= 𝑦

0

+ ∑

2𝑖=1

𝑦

1𝑖

𝑅

𝑖𝑡−𝑖

+ 𝑦

2

𝑅

𝑚𝑡−𝑖

+ 𝑦

3

𝑅

𝑚𝑡

+ 𝑦

4

𝑀

𝑡

+ 𝑦

5

𝐸

𝑖𝑡

+

𝐽𝑡+

𝜀

𝑡

(4)

There are now 39 elimination events gathered in the 𝐸𝑖 variable and 12 (6 times 2) in the outlier variable 𝐽𝑡. In table 6, the correlation matrix of the variables in this regression can be found. Of course, only the last two rows and columns differ from the previous correlation matrix (the correlation between the other variables stay the same). There isn’t a very high correlation present between the 𝐸𝑖 or 𝐽𝑡 and the other variables. Therefore, the same approach as before is used, and an extra regression without 𝑅𝑖𝑡−1 is run.

Table 6

Correlation Matrix with outliers as separate variable

This matrix gives the correlations between the variables in formula 4.

Ri Rit-1 Rit-2 Rmt-1 Rm M Ei Jt Ri 1 0,0204 -0,0071 0,0914 0,6979 -0,0641 -0,0405 -0,0092 Rit-1 0,0204 1 0,0201 0,6909 0,0763 0,0222 0,0040 0,0642 Rit-2 -0,0071 0,0201 1 0,0732 -0,0184 -0,0297 0,0504 0,0708 Rmt-1 0,0914 0,6909 0,0732 1 0,0960 0,0264 0,0092 0,0424 Rm 0,6979 0,0763 -0,0184 0,0960 1 0,0262 -0,0386 0,0291 M -0,0641 0,0222 -0,0297 0,0264 0,0262 1 0,0862 0,0324 Ei -0,0405 0,0040 0,0504 0,0092 -0,0386 0,0862 1 -0,0259 J -0,0092 0,0642 0,0708 0,0424 0,0291 0,0324 -0,0259 1

(20)

20

ones from the same regression but without 𝑅𝑖𝑡−1. The results for the first 6 independent variables of regression 3 are quite the same as with the previous shown regressions (just as the R-squared), with the exception that the lagged index return, 𝑅𝑖𝑡−2, isn’t statistically significant anymore at a 1, 5 or 10 percent level. It becomes interesting when looking at the elimination variables 𝐸𝑖 and 𝐽𝑡. The standard elimination events variable (𝐸𝑖) reduced in significance but can still be considered significant at a 10% level. From this it can be concluded that the results presented in table 5 keep being significant without their extreme values and are therefore robust. The results achieved are not the product of a few extraordinary event days. The sentiment effect still has a magnitude of 11,9 basis points, which is comparable to the previous findings in table 5. The outlier variable 𝐽𝑡 is highly significant, which could be expected. Because only the most extreme events (which significantly differ from zero) are separated from the rest and put together, the results are bound to be more significant than the dataset it was extracted from. These outliers have a coefficient of -0,3121 which is relatively large and indicates a relatively strong sentiment effect. As an investor however, these extreme events cannot be anticipated or differentiated from the rest and are therefore on their own not useful for investment purposes.

The results from regression 4 differ little compared to those of regression 3. All the variables which were significant with regression 4 still are, with the same confidence levels. It can be concluded that even without outliers and collinearity, an elimination event causes significant negative returns in the local index. To test the robustness of the results further, only part of the variables in formula 4 are regressed. Table 7 gives the results of two of such regressions. The results denoted by regression 5 are without the Lagged return variables and just as before only one elimination event variable is used (𝐸𝑖). Regression 4 gives the results of a regression that only consists out of the intercept and the elimination variable. Neither of the two regression results gives new insights. Both regressions result in a highly significant elimination coefficient indicating the presence of the sentiment effect.

(21)

21

Table 7 Robustness Checks

This table reports the results of the following regression:

𝑅𝑖𝑡 = 𝑦0+ ∑ 𝑦1𝑖 2

𝑖=1

𝑅𝑖𝑡−𝑖+ 𝑦2𝑅𝑚𝑡−𝑖+ 𝑦3𝑅𝑚𝑡+ 𝑦4𝑀𝑡+ 𝑦5𝐸𝑖𝑡+ 𝐽𝑡+ 𝜀𝑡

where 𝑹𝒊𝒕 is the continuously compounded daily return; 𝒚𝟎is the intercept; 𝑹𝒊𝒕−𝟏 and 𝑹𝒊𝒕−𝟐 are the lagged index returns (one and two days); 𝑹𝒎𝒕 is the return of Datastream’s European

market index; 𝑹𝒎𝒕−𝒊 the previous day European market return; 𝑴𝒕 is a dummy variable to account for the Monday effect, ; 𝑬𝒊is a dummy variable for an elimination event and 𝑱𝒕 represents

outliers. Several variables may be omitted in running the regression in order to test the robustness of the results. The regression coefficients, t-values and corresponding probabilities are given. *, ** and *** indicate a 10%, 5% and 1% level of significance, respectively. The observed period includes 823 non-elimination games, 39 elimination events and 12 outliers. The null hypothesis is that the returns are unaffected by eliminations from their national soccer team. In Regression 3 all the variables are included, regression 4 accounts for multicollinearity. Variable 𝑱𝒕 is left out

in regression 5 and 6. In these regressions, the elimination event variable 𝑬𝒊 therefore contains all 51 elimination events.

Regression 3 Regression 4 Regression 5 Regression 6

(22)

22

B.IS THE EFFECT STILL PRESENT IN THE EUROPEAN CHAMPIONSHIP OF 2012 AND THE WORLD CUP OF

2014?

The second sub question tries to answer the question whether or not the sentiment effect is still present in the last European Championship and World Cup. I therefore ran the regression mentioned in the methodology part, formula 2, in which the elimination events of those last two tournaments are defined as a separate independent variable, 𝐸𝑖𝑡2010−2014. This variable contains the 8 elimination events from 2012 and 2014. Table 8 shows the correlation matrix in which no remarkable changes occurred in comparison with previous shown correlation matrices. Similar to before, a second regression in which 𝑅𝑖𝑡−1 is left out is also ran.

The results of both regressions can be found in table 9. The aim of running these regression is to see whether or not the elimination event variable is still significant for the tournaments of 2012 and 2014. As can be seen in table 9, it is not. Although the coefficient is negative, it isn’t significant at a 1%, 5% or 10% significance level in either of the two regressions. The elimination events which occurred during the tournament before that, denoted by 𝐸𝑖𝑡1984−2010, still are. This indicates that the soccer sentiment has changed during the last two tournaments and isn’t economically nor statistically significant anymore. It could be that smart investors, after reading the previous studies on this subject as well, took the sentiment effect into account and in this way arbitraged it away. The results being insignificant might however also be caused by the low number of events available, especially for the last two tournaments. MacKinlay (1997) states that the power of an event study test with 45 events is about 3,5 times higher than with 8 events (when assuming the mean return of the effect to be 0,5 percent, the standard deviation to be 2 percent and significance level 5%). The same effect (i.e. same return, same SD, etc) is therefore less likely to be deemed statistically significant when is has less events to be studied. The results obtained will therefore not necessarily

Table 8

Correlation Matrix with last two tournaments’ eliminations as a separate independent variable

(23)

23

completely rule out the possibility of the effect still being present today. The last two tournaments need a closer look before the existence of the effect can be fully discarded.

Table 9

Regression Results with last two tournaments’ eliminations as a separate independent variable

This table reports the results of the following regression: 𝑹𝒊𝒕 = 𝒚𝟎+ ∑ 𝒚𝟏𝒊

𝟐

𝒊=𝟏

𝑹𝒊𝒕−𝒊+ 𝒚𝟐𝑹𝒎𝒕−𝒊+ 𝒚𝟑𝑹𝒎𝒕+ 𝒚𝟒𝑴𝒕+ 𝒚𝟓𝑬𝒊𝒕𝟏𝟗𝟖𝟒−𝟐𝟎𝟏𝟎+ 𝒚𝟔𝑬𝒊𝒕𝟐𝟎𝟏𝟐−𝟐𝟎𝟏𝟒+ 𝜺𝒕

where 𝑹𝒊𝒕 is the continuously compounded daily return; 𝒚𝟎is the intercept; 𝑹𝒊𝒕−𝟏 and 𝑹𝒊𝒕−𝟐 are the lagged index returns

(one and two days); 𝑹𝒎𝒕 is Datastream’s European market return; 𝑹𝒎𝒕−𝒊 the previous day European market return; 𝑴𝒕 is a

dummy variable to account for the Monday effect; 𝑬𝒊𝒕𝟏𝟗𝟖𝟒−𝟐𝟎𝟏𝟎 is a dummy variable which represents an elimination event

from 1984 until 2014 and 𝑬𝒊𝒕𝟐𝟎𝟏𝟐−𝟐𝟎𝟏𝟒 represents an elimination event which happened during the tournaments in 2012 and

2014. The regression coefficients, t-values and corresponding probabilities are given. *, ** and *** indicate a 10%, 5% and 1% level of significance, respectively. The results are obtained using Ordinary Least Squares with panel data. The data consist out of continuously compounded daily returns of indices from six prominent European soccer countries (the Netherlands, England, Germany, France, Italy and Spain) together with their eliminations from European Championships and World Cups from 1984 until 2014. The observed period includes 823 non-elimination games, 43 elimination events between 1984 and 2012 and 8 elimination events between 2012 and 2014. The null hypothesis is that the returns are unaffected by eliminations from their national soccer team. In Regression 7 all the variables are included, regression 8 accounts for multicollinearity.

Regression 7 Regression 8 Independent variables Coefficient t-Statistic & (Prob.) Coefficient t-Statistic & (Prob.) y0 0,0480 3,4809 (0,0005)*** 0,0435 3,3338 (0,0009)*** Rit-1 -0,0930 -5,4515 (0,0000)*** Rit-2 -0,0189 -1,6446 (0,0982)* -0,0092 -0,7749 (0,4386) Rmt-1 0,1075 6,2602 (0,0000)*** 0,0299 2,9286 (0,0035)*** Rm 0,8183 57,3259 (0,0000)*** 0,8252 60,0064 (0,0000)*** M -0,2382 -7,2157 (0,0000)*** -0,2250 -6,5441 (0,0000)*** 𝐸𝑖𝑡1984−2010 -0,1498 -2,1808 (0,0295)** -0,1415 -2,0013 (0,0457)** 𝐸𝑖𝑡2010−2014 -0,1126 -0,9305 (0,3524) -0,1138 -0,9921 (0,3214) R-squared 0,8137 0,8340 F-statistic Prob(F-stat) 532,9037 0,0000 716,1669 0,0000

(24)

24

the sentiment effect. After eliminations of the Netherlands and England in 2012 and France in 2014, negative returns can also be seen. Although here it could be argued that the index was already in a negative trend beforehand. The figures of Germany and Italy in 2012 seem to contradict the theory of the sentiment effect by showing a clear positive change in direction after the eliminations. These result eventually don´t give undisputed evidence for the existence of the effect, nor do they totally deny it. In most of the presented cases, elimination are still followed by negative index trends. Studying the movement of the graphs, as mentioned above, is not very accurate and quite arbitrary. Because the sentiment effect is present in the tournaments from 1984 until 2010 (with a significance level of 5%, see table 9), a better way is to compare the latest eliminations with that data and see in which degree it differs. Table 10 shows descriptive statistics of both groups of eliminations. When comparing both elimination categories, it is clear that the eliminations from 1984 until 2010 are followed by negative returns whereas the returns after eliminations in 2012 and 2014 are positive. This seems to support the conclusion that indeed the sentiment effect has vanished during the last two tournaments. The positive mean return is however mainly due to a strong positive return after the elimination of Germany in the World Cup of 2012 (+4,24%). The return following this elimination is the most positive return of all the 51 eliminations found from 1984 until 2014. Without this positive outlier, the mean return of the remaining eliminations drops to -0,577%. This is far more negative than the mean of the eliminations from 1984-2010, which in turn seems to confirm the presence of soccer sentiment effect. This could also be seen from fact that the median is negative. The average daily return (continuously compounded) of the six countries over the whole period of 2012, 2013 and 2014 is 0,0239%. If we compare this with the mean daily return without the German outlier, it could be argued that the sentiment effect is still present because eliminations are followed by far lower returns than on average.

Table 10

Descriptive statistics of last two tournaments’ eliminations as a separate independent variable

(25)

25

Next to this, 63% of the eliminations from 2012 and 2014 yield a negative return whereas this is only 56% from 1984 until 2010. The percentage of returns which are lower than -1% is also higher. This also supports the view that the sentiment effect is at least as much present as before. Still, the low number of observation in the second group makes drawing conclusions difficult. Therefore it cannot be said with complete certainty that the effect is still there, although it does seem like it.

C.THE EXPLOITABILITY OF THE EFFECT

(26)

26

investment in the German index in 2008, for example, generated a return of $-5,76. Because the strategy entails going short during this time, it accumulated these $5,76 as profit. Together with the other investments that year, the strategy yields $22,31 which is 3,72% of the invested $600. Note that the $600 isn’t actually invested by the investor. He first of all sells (shorts) the indices, which yields him $600 to e.g. invest in risk-free government bonds. With this $600 he then buys back the indices later on. Whatever remains, is profit (together with possible revenues from the time he was investing in a riskless funds). The highlighted cells indicate returns of countries which won that tournament. The returns are, as expected, mostly positive. When looking at the grand total at the bottom of the table, a negative total return is shown. This means that shorting several local indices until the moment of elimination doesn’t yield positive returns but even leads to a loss of $2,82. The sentiment effect is probably not strongly enough present in the countries considered to overcome the general upward trend of stock indices. The strategy of short selling in this way is therefore not viable, even without considering any transaction costs.

Table 11

Dollar returns during tournaments from 1984 until 2014 per country, together with the total returns ($ and %) generated by the short selling strategy

This table reports the amount of money ($) which could have been made per tournament and per country using the short selling strategy. This strategy entails shorting $100 per country per tournament. The Total Strategy Return ($) sums up these amounts per tournament to represent the profit per tournament expressed in dollars. The Total Strategy Return (%) expresses this dollar amount as a percentage of the total shorted amount ($600). Finally, this table adds the amounts up to present the total profit (loss) of this strategy from 1984-2014.

Year

RNED ($) RENG ($) RGER ($) RFR ($) RIT ($) RSP ($)

(27)

27

There might still be other ways to exploit the sentiment effect. As mentioned earlier, a second possibility for investing is doing so with the use of derivatives. Two (at first sight) possible derivatives to use are put options and future contracts. A put option gives the owner the right to sell his security for a pre-specified price at any moment he pleases. A future contract is an agreement which entails buying a security at a pre-specified price at a specified moment. A put option is similar to short selling in the way that you hope that the security is going to decline in value. In this case you can sell the security for the pre-specified price which is higher than it costs to buy it at that moment. The advantage of a put option over short selling however is that you don’t have downside risk. You are not exposed to the losses when the underlying increases in value, as you are with short selling. Because of the limited risk and the choice aspect, a put option costs money to buy. In order to make a profit using put options, the money made with in the money puts must compensate for the prices of the call options (even the ones that turn out to be out of the money). All the negative returns occurring after an eliminations summed up are -$80,978 (table 11: -1,4716 + -1,8717 + -2,0233 + …). This is the total profit which is made from the 28 put options which are in the money after elimination. In total you need 60 put options to cover all tournaments. This means that a put option is not allowed to cost more than $1,35 apiece (80,978/60). Using the Black-Scholes-Merton formula for option pricing however, it can be calculated that a put option in this situation1 will cost about $1,45 apiece. Executing put options at elimination is therefore not a profitable strategy. It is in practice, however, less negative as it seems. There will be a few options which, at elimination, are out of the money but still have a few days until maturity. This options are still worth something because the index might drop below the strike price in those last few days. I however don’t expect this (small) gain to completely turn the profitability of the strategy around.

Although a futures contract can also be used for making profits with securities which are declining in value, it is not very useful with the soccer sentiment effect. A futures contract ends on the delivery date, which is determined at the moment the contract is negotiated. Because the exact moment of elimination is unknown and match outcomes cannot be predicted, it is impossible to match the delivery date with the moment of elimination. If the delivery date is before the elimination, you miss out on the negative return caused by the loss. If it is after the elimination, the index has probably already recovered from the drop. A futures contract can therefore not be used to exploit the soccer sentiment effect.

1

(28)

28

The trading strategy proposed in this theses did (unfortunately) not prove to be useful. Mainly because the sentiment effect isn’t large enough to outpace the growth of the indices. One way of overcoming this is by using Edmans et al. (2007) approach of investing per single match played. In this case the positive returns occurring in between matches is avoided. The authors however also concluded that in practice this strategy probably isn’t useful because of transaction costs and the amount of work. Still it is interesting to see how high transaction costs have to be in order for this strategy to be generating a loss. Because it’s unclear when the elimination will occur, it is necessary to take a short position at every match played by a country in a tournament. For simplicity, I will assume that every matched played until elimination is won. In total there are 291 matches played by the countries and in the tournaments I consider. In 51 of them, one of the countries considered is eliminated. This means that 240 matches are won, generating no negative returns caused by the sentiment effect. I assume that the returns which do occur is the average return found during tournament days which weren’t elimination events, i.e. 0,0175 % (as can be found in table 3, descriptive statistics). In the case of short selling, this leads to a total loss of 4,2% (240*-0,0175). This loss is added to the transactions costs needed to enable the short selling. Shorting 240 matches results in 480 investment steps. This loss and these transaction costs must be compensated by elimination matches. These elimination matches lead to negative returns and therefore contribute to a profit. As found in the beginning of this chapter, an elimination event leads to a negative return of 0,1416% (table 5). 51 eliminations therefore result in a return of -7,2% (51*-0,1416). It however does need 102 investment steps (51*2) to realize. The costs and benefits of shorting every match during an international tournament perfectly compensate each other when transaction costs are 0,00519%. This means that when the transaction costs are about 0,5 basis points or less, a profit is made. In this case about 1 basis point (2*0,5) is spend on transaction costs every match. Facing transaction costs as low as these, however, is practically impossible as an investor. Kaplanski & Levy (2010) e.g. calculate the profit of their strategy assuming transaction costs which are 20 to 100 times this value. Hirshleifer and Shumway (2003) calculate that transaction costs for trading in the U.S. market are in the order of one basis point per transaction. They however also conclude that trading other market indices, like the strategy above entails, results in costs which may be significantly higher. Either way it can be concluded that profiting from trading on every single match is practically impossible, which is in line with the conclusion of Edmans et al. (2007).

(29)

29

VI.CONCLUSION AND DISCUSSION

Previous papers, such as the one of Edmans et al. (2007) or Kaplanski & Levy (2010), used psychological insights to study investor behavior on the financial markets and made the connection between sport results and stock returns. In this thesis this connection is studied in six European countries using an elimination from an international soccer tournament as a proxy for mood. The underlying theory is derived from the soccer sentiment effect which entails that a soccer loss results in negative stock returns. The main goal of this thesis is to see whether it is possible to employ a profitable investment strategy based on this soccer sentiment effect. The main research question therefore is:

How can the soccer sentiment effect be exploited during the World Cup or European Championship?

The results in this study show that the soccer sentiment effect is present in the data and with the strategy that is used. An elimination from an European Championship or World Cup leads to an abnormal next-day return of -14,16 basis points, which is more than -4% in monthly terms. Although it is three and a half times lower than the effect Edmans et al. (2007) found after an elimination, it is still statistically significant. The results remain significant when leaving out outliers. These results are consistent with what was hypothesized in this paper. Whether or not the effect is still present in the last two tournaments, couldn’t be answered with full certainty. Comparing descriptive statistics of eliminations before and after 2012 do, however, seem to indicate that the effect is still there.

(30)

30

The main conclusion of this paper therefore is that no profitable investing strategy can be employed using the strategy proposed. Hence, the strategy of Kaplanski & Levy (2010) is proven to be better suited for exploitation of the effect. Although the soccer sentiment effect is shown to be statistically significant, it might just be insufficiently present in Europe and therefore not exploitable. Maybe the only way to make a profit of the soccer sentiment effect is to invest in a foreign index which partly incurs the negative effects of all losses and eliminations during a tournament, just as Kaplanski & Levy (2010) do. In this way there are more moments of stock-price decline which can compensate for the general upward trend of stock prices.

A few remarks can be made on this study and its results. Eliminations which took place during Fridays and some weekend-days were deleted from the data sample. This resulted in less elimination events and in additions may have resulted in extra significance for the Monday effect (in the case the sentiment effect reaches further than just the day after elimination). I also only looked at elimination events, and ignored other lost games. These lost games also cause a significant sentiment effect, according to Edmans et al (2007). These games are part of the ‘no games’ data-group and its negative returns could therefore have made the elimination events less significant (because these are compared with the ‘no games’ group). The last factor that could have slightly impacted the results is the use of Datastream’s European Market index to simulate the economic circumstances. This index consists out of local indices so probably also incurres some of the sentiment effect, resulting is a lower significance for the elimination events.

(31)

31

REFERENCES

Ackert, L. F., Deaves, R. (2010). Behavioral Finance: Psychology, Decision-Making and Markets. Mason, OH: South-Western Cengage Learning

Ashton, J., Gerrard, B., & Hudson, R. (2003). Economic impact of national sporting success: evidence from the London stock exchange. Applied Economics Letters, 10(12), 783-785.

Bernile, G., & Lyandres, E. (2011). Understanding Investor Sentiment: The Case of Soccer. Financial

Management, 40(2), 357-380.

Berthier, F., & Boulay, F. (2003). Lower myocardial infarction mortality in French men the day France won the 1998 World Cup of football. Heart (British Cardiac Society), 89(5), 555-556.

Brooks, C. (2002). Introductory econometrics in Finance. Cambridge: Cambridge University Press. Brown, G. W., & Hartzell, J. C. (2001). Market Reaction to Public Information: The Atypical Case of the

Boston Celtics. Journal Of Financial Economics, 60(2-3), 333-370.

Cao, M., & Wei, J. (2005). Stock market returns: A note on temperature anomaly. Journal Of Banking

& Finance, 29(6), 1559-1573.

Carroll, D., Ebrahim, S., Tilling, K., Macleod, J., & Smith, G. D. (2002). Admissions for myocardial infarction and World Cup football: Database survey. BMJ: British Medical Journal, 325(7378), 1439-1442.

Chang, S., Chen, S., Chou, R. K., & Lin, Y. (2012). Local Sports Sentiment and Returns of Locally Headquartered Stocks: A Firm-Level Analysis. Journal Of Empirical Finance, 19(3), 309-318. EDMANS, A., GARCÍA, D., & NORLI, Ø. (2007). Sports Sentiment and Stock Returns. Journal Of

Finance, 62(4), 1967-1998.

French, K. R. (1980). Stock returns and the weekend effect. Journal Of Financial Economics, 8(1), 55-69.

French, K. R., & Poterba, J. M. (1991). Investor diversification and international equity markets.

American Economic Review, 81(2), 222.

Gerlach, J. R. (2011). International sports and investor sentiment: do national team matches really affect stock market returns?. Applied Financial Economics, 21(12), 863-880.

Hirshleifer, D. (2001). Investor Psychology and Asset Pricing. Journal Of Finance, 56(4), 1533-1597. Hirshleifer, D., & Shumway, T. (2003). Good Day Sunshine: Stock Returns and the Weather. Journal Of

Finance, 58(3), 1009-1032.

Kamstra, M. J., Kramer, L. A., & Levi, M. D. (2000). Losing Sleep at the Market: The Daylight Saving Anomaly. American Economic Review, 90(4), 1005-1011.

Kamstra, M. J., Kramer, L. A., & Levi, M. D. (2003). Winter Blues: A SAD Stock Market Cycle. American

(32)

32

Kaplanski, G., & Levy, H. (2010). Exploitable Predictable Irrationality: The FIFA World Cup Effect on the U.S. Stock Market. Journal Of Financial & Quantitative Analysis, 45(2), 535-553.

Kaplanski, G., & Levy, H. (2014). Sentiment, Irrationality and Market Efficiency: The Case of the 2010 FIFA World Cup. Journal Of Behavioral And Experimental Economics, 4935-43.

Klein, C., Zwergel, B., & Heiden, S. (2009). On the existence of sports sentiment: The relation between football match results and stock index returns in Europe. Review Of Managerial

Science, 3(3), 191-208.

Lee, Y., & Wang, K. (2010). The Asymmetric Impulse of the Sunshine Effect on Stock Returns and Volatilities. Amfiteatru Economic, 12(28), 606-633.

Lucey, B. M., & Dowling, M. (2005). The Role of Feelings in Investor Decision-Making. Journal Of

Economic Surveys, 19(2), 211-237.

MacKinlay, A. C. (1997). Event studies in economics and finance. Journal of economic literature, 13-39.

Palomino, F., Renneboog, L., & Zhang, C. (2009). Information Salience, Investor Sentiment, and Stock Returns: The Case of British Soccer Betting. Journal Of Corporate Finance, 15(3), 368-387. Pantzalis, C., & Park, J. C. (2014). Exuberance out of left field: Do sports results cause investors to

take their eyes off the ball?. Journal Of Economic Behavior & Organization, 107760-780. Saunders, E. M., & Jr. (1993). Stock prices and the Wall Street weather. American Economic Review,

83(5), 1337.

Schwarz, N., Strack, F., Kommer, D., & Wagner, D. (1987). Soccer, rooms, and the quality of your life: Mood effects on judgments of satisfaction with life in general and with specific domains.

European Journal Of Social Psychology, 17(1), 69-79.

Schweitzer, K., & Zillmann, D. (1992). Perception of threatening events in the emotional aftermath of a televised college football game. Journal Of Broadcasting & Electronic Media, 36(1), 75. Shu, H., & Hung, M. (2009). Effect of wind on stock market returns: evidence from European markets.

Applied Financial Economics, 19(11), 893-904.

Social Issues Research Centre (2006). The impact of sports on the workplace. Retrieved June 01, 2015 from http://www.sirc.org/publik/sport_and_the_workplace.pdf

Studenmund, A. H. (1997). Using econometrics: A practical guide. Reading, Mass: Addison-Wesley. Trovato, F. (1998). The Stanley Cup of Hockey and Suicide in Quebec, 1951-1992. Social Forces, 77(1),

105-126.

Vieira, E. S. (2012). Investor Sentiment and Market Reaction: Evidence on 2010 FIFA World Cup.

(33)

33

Wann, D. L., & McGeorge, K. K. (1994). Relationships Between Spectator Identification and Spectators' Perceptions of Influence, Spectators' Emotions, and Competition Outcome.

Journal Of Sport & Exercise Psychology, 16(4), 347-364.

Wann, D. L., Melnick, M. J., Russell, G. W., & Pease, D. G. (2001). Sport fans: The psychology and

social impact of spectators. New York, NY, US: Routledge.

White, G. F. (1989). Media and Violence: The Case of Professional Football Championship Games.

Aggressive Behavior, 15(6), 423-433.

Witte, D. R., Bots, M. L., Hoes, A. W., & Grobbee, D. E. (2000). Cardiovascular mortality in Dutch men during 1996 European football championship: Longitudinal population study. BMJ: British

(34)

34

APPENDIX

In this appendix several tables are shown which are used in this thesis to present and come up with the results.

Appendix A Tournament periods

(35)

35

Appendix B Elimination dates

All the elimination dates of the considered countries for the tournaments of 1984 until 2014 are shown here. The dates given are the dates of the elimination matches. These are therefore t-1 because the day after the match is specified by t. When a country wins the tournament this is denoted by ‘WIN’; when it wasn’t able to qualify it is shown with the abbreviation for Did Not Qualify, ‘DNQ’. ‘NA’ means that the index data was Not Available for that day. ‘Weekend’ denotes days for which the following day falls in a weekend (the match day is on Friday or Saturday). These games are also omitted as events. The two omitted events described by ‘Deleted’ were events which are left out as data because of other reasons (mentioned in chapter IV, Data).

1

West-Germany

Elimination (t-1)

Tournament Netherlands England Germany France Italy Spain

2014 9-7-2014 Deleted WIN 4-7-2014 24-6-2014 18-6-2014 2012 17-6-2012 24-6-2012 28-6-2012 WEEKEND 1-7-2012 WIN 2010 11-7-2010 27-6-2010 7-7-2010 22-6-2010 24-6-2010 WIN 2008 WEEKEND DNQ 29-6-2008 17-6-2008 22-6-2008 WIN 2006 25-6-2006 WEEKEND 4-7-2006 9-7-2006 WIN 27-6-2006 2004 30-6-2004 24-6-2004 23-6-2004 WEEKEND Deleted 20-6-2004 2002 DNQ WEEKEND 30-6-2002 11-6-2002 18-6-2002 WEEKEND 2000 29-6-2000 20-6-2000 20-6-2000 WIN 2-7-2000 25-6-2000

1998 7-7-1998 30-6-1998 WEEKEND WIN WEEKEND 24-6-1998

1996 WEEKEND 26-6-1996 WIN 26-6-1996 19-6-1996 WEEKEND

(36)

36

Appendix C

Price index during the World Cups and European Championships with elimination moment indicated

The daily values of the price indices of the countries considered (Netherlands, England, Germany, France, Italy and Spain) per tournament (2012 and 2014). The moment of elimination is indicated with a circle and an arrow.

280 285 290 295 300 305 310 315

Price index Netherlands 2012

PI NL 2012 395 400 405 410 415 420 425 12 -6-201 4 14 -6-201 4 16 -6-201 4 18 -6-201 4 20 -6-201 4 22 -6-201 4 24 -6-201 4 26 -6-201 4 28 -6-201 4 30 -6-201 4 2- 7-2014 4- 7-2014 6- 7-2014 8- 7-2014 10 -7-201 4 12 -7-201 4 14 -7-201 4

Price index Netherlands 2014

(37)

37 5300 5350 5400 5450 5500 5550 5600 5650 5700

Price index England 2012

PI ENG 2012 5900 6000 6100 6200 6300 6400 6500 6600

Price index Germany 2012

PI GER 2012 4150 4200 4250 4300 4350 4400 4450 4500 4550 4600 12 -6-201 4 14 -6-201 4 16 -6-201 4 18 -6-201 4 20 -6-201 4 22 -6-201 4 24 -6-201 4 26 -6-201 4 28 -6-20 1 4 30 -6-201 4 2- 7-2014 4- 7-2014 6- 7-2014 8- 7-2014 10 -7-201 4 12 -7-201 4 14 -7-201 4

Price index France 2014

(38)

38 135 140 145 150 155 160 165

Price index Italy 2012

PI IT 2012 210 215 220 225 230 235 240 12 -6-201 4 14 -6-201 4 16 -6-201 4 18 -6-201 4 20 -6-201 4 22 -6-201 4 24 -6-20 1 4 26 -6-201 4 28 -6-201 4 30 -6-201 4 2- 7-2014 4- 7-2014 6- 7-2014 8- 7-2014 10 -7-201 4 12 -7-201 4 14 -7-201 4

Price index Italy 2014

PI IT 2014 10200 10400 10600 10800 11000 11200 11400 12 -6-20 1 4 14 -6-201 4 16 -6-201 4 18 -6-201 4 20 -6-201 4 22 -6-201 4 24 -6-201 4 26 -6-201 4 28 -6-201 4 30 -6-201 4 2- 7-2014 4- 7-2014 6- 7-2014 8- 7-2014 10 -7-201 4 12 -7-201 4 14 -7-201 4

Price index Spain 2014

(39)

39

Appendix D

$100 invested during the World Cup or European Championship until elimination

These graphs show the how an investment of $100 at the beginning of an international soccer tournament performs as the tournament progresses. The straight dashed line in the middle of the graphs represents the $100 line. Tournaments from 1984 until 2014 are considered. The countries used are The Netherlands, England, Germany, France, Italy and Spain.

(40)
(41)

Referenties

GERELATEERDE DOCUMENTEN

Based on the literature review conducted earlier in this thesis, seven hypotheses were drafted, suggesting effects between and among type of monitoring (behaviour or output),

Een aantal bewerkingen is geschikt voor verscheidene doeleinden: voor elke provincie apart, voor vergelijking tussen provincies en voor het gehele land, geagrregeerd over

Ongevallen (inclusief UlYIS) per miljoen voertuigkilometers op drie- en vierstrooks rijbanen van autosnelwegen, op jaarbasis in 1993 en 1994, wegsecties uitgesplitst

The sections are separated by a transpar- ent Nation (Du Pont) membrane. The counter electrode is placed against the membrane, while the distance between the

Is the DOW-effect present in returns that are adjusted to the market beta, market capitalization and book-to-market ratio of firms listed on the Dutch

Psychological literature predicts that the result of international football matches causes an up- or down beat in mood. This thesis examines the relationship between

As stated, operations usually take place far beyond our national borders. To enable deployment over great distances, transport capacity is required: by sea and by air. This is one

One can conclude that rumors before officially announcing an international acquisition could influence the abnormal returns after the announcement date for a target and or acquiring