Rivalry in the football industry and its impact on the stock
prices of listed football clubs
Master Thesis Finance
BY W.J.Tankink Supervisor: J.H. von Eije Groningen, January 11, 2018 University of Groningen
Faculty of Economics and Business MSc Finance
2
Abstract
Rivalry in the football industry is examined in this paper as it analyses rivalry effects on stock price performance of football clubs. It does not matter which sport is exercised and at what level, rivalry among clubs is one of the main sources of attractiveness of a league. The rivalry among football clubs leads to a positive “mood” in case the rival loses and leads to a negative “mood” in case the rival has a for them positive match result. It is hypothesised that these emotions due to the performance of the period rival affect the stock price performance of the focal football club. This study shows that there is evidence that the results of the period rival can have an impact on the investment decisions of club supporters.
3
1. Introduction
Football, or what they call it in the United States soccer, is a well-known sport exercised all around the world. It all started in 1887. From then on it was allowed by the Fédération Internationale de Football Association (FIFA) to recruit football players as an individual football club. This was the start of, what is familiar to us now as, professional football. In other words, money made its entrance here and the role of money has increased a lot since then (Dobson and Goddard, 1998). One of the largest causes for this significant role of money is the Bosman arrest, also known as the Bosman Ruling. Simply stated, after the expiration of a European football players contract, the player is free to move from his previous club to another club within the European Union. As a consequence of different regulations within national competitions, in combination with the large differences in budgets, Kesenne (2007) states that the gap between the “rich” and “poor” countries has clearly widened, budget-wise as well as performance-wise. In the world of football, four nations are considered to be the “Big four”, in terms of money. These countries are: England, Germany, Italy, and Spain.
Other causes for the increasing role of money are the enormous increases in media contracts and sponsoring contracts, which is one of the main sources of income in the football industry. The other sources are merchandising income and income from match receipts (Scholtens and Peenstra, 2009). One clear example for such enormous media contracts is the Skysports contract for the English Premier League. Skysports has the right to broadcast the English Premier League matches for which it paid 7 billion euros, for a period of 3 years (2016 till 2019). As a consequence of these media contracts and sponsoring contracts, the football industry attracted relatively recently also the interest of large investors (Demir and Rigoni, 2017), by which the football industry eventually resulted in a stand-alone sector (Ecer and Boyukaslan, 2014).
4
to understand what the drivers are behind the fluctuations in share prices for listed football clubs.
Several studies have examined how stock prices are affected by the results and performance of football clubs (Demir & Danis, 2011; Scholtens & Peenstra, 2009; Stadtmann, 2006). These studies purely examine what the effect of the match outcome is on the stock prices of that particular football club. Other studies find that stirred up emotions due to football have also an impact on the stock prices of football clubs (Demir and Rigoni, 2017; Edmans, Garcia, and Nørli, 2007; Palomino et al., 2009). Based on the findings in the literature, there are two ways a match outcome affects the stock price of a particular team. The match outcome itself and the mood of the fans resulting from the match outcome. Palomino et al. (2009) is one of the few studies that took both of these effects into consideration. They, however, only examined the effects in isolation of each other.
Only one recent study took the two match outcome effects out their isolation and examined their combined effect on the stock price of a particular football club (Demir and Rigoni, 2017). This combined effect is examined by studying the effect of rivalry in the football industry on the stock price of a listed football club. This rivalry is represented by a focal team and its rival. The listed football club for which the presence of rivalry in its stock price is measured, is called the focal team. The combined effect of the performance of the focal team as well as the effect of its rival’s performance represents this rivalry effect. The rival team is the football club which is most feared by the focal team.
5
club teams to differ (a lot) from year to year. Therefore is the rivalry score based on only two years of performance, causing the examined rivals to be “period rivals”. Secondly, the study that examined the rivalry effect on stock prices in the football industry (Demir and Rigoni, 2017) focuses on the rivalry between AS Roma and Lazio Roma. Here the focus is on multiple European rivalries.
All football clubs have their centuries old “rival”, based on multiple and varied reasons. Some rivalries are based on their final position in the league. For example, both battling for the championship trophy, year in year out. Others are based on political reasons, or from both coming from the same city, or because of religious reasons. Many reasons thus exist for supporters of “their” team to consider a particular other team to be their arch-rival. The question arises, however, if the right football club is considered to be the arch-rival in terms of stock price affection. As a consequence, I will try to answer the following research question:
Is period rivalry in the football industry present in the stock prices of listed football clubs?
The following section will discuss the literature that can be related to this study and it presents hypotheses. In the subsequent section, “Data collection”, the required data to answer this research question are described and it is explained how and where these data are collected. Then, “Methodology” section describes how I test the hypotheses. In the “Results” section, the results following from the hypothesis testing are provided and analysed. Finally, in the “Conclusion” section, the research is concluded, limitations and suggestions for future research are given.
2. Literature Review
6
of listed Turkish football clubs are affected by their match results. These studies however purely examine what the effect of the match outcome will be on the stock prices of that particular football club.
Edmans et al. (2007) find that after football losses the market significantly declines attribute it to sudden changes in fan investor mood due to football outcomes. In addition, Palomino et al. (2009) find that stock prices are influenced by the mood of investors which resulted from match results, especially a positive abnormal return related to a win. Based on these findings in the literature, there are two ways a match outcome affects the stock price of a particular team. The match outcome itself and the mood of the fan investors resulting from the match outcome. Palomino et al. (2009) is one of the few studies that took both these effects into consideration. These effects are, however, only examined in isolation of each other. Only one recent study took these two match outcome effects on the stock price of a particular football club (Demir and Rigoni, 2017).
According to Edmans et al. (2007) it is rationalized to study the link between a mood variable and stock returns, in case the variable satisfies the following three key characteristics:
A large proportion of the population’s mood should be affected by the variable
French and Poterba (1991) studied the relationship between the individuals affected and the investments in the domestic stock market. They found that these individuals are also the ones that invest in a marginal way in the domestic stock market. In other words, the international football matches are perceived to be important to a large fraction of the population in multiple countries, while these matches take place at regular intervals. This characteristic is even more strengthened by Boyle and Walter (2003). They state that in general, affiliation for a football club is assumed to be more important than national identity.
Mood must be driven in a substantial and unambiguous way by the variable
Carroll et al. (2002) studied the change to have a heart attack during the World Cup tournament in 1998 between England and Argentina. They found a 25% increase in admissions for heart attacks in the 3-day period after the World Cup final, in which England lost to Argentina in a penalty shoot-out.
The effect of the mood-variable on the stock returns should be correlated across the majority of events within an examined football club.
7
countries such as Greece, Italy, Portugal, and Spain, newspapers that are exclusively dedicated to sports, in particular football, are best-selling newspapers.
These two ways of how a match outcome affects the stock price of a football club in combination is defined as rivalry. Rivalry is an important driver behind the fluctuations in share prices for listed football clubs. One message of the study by Mason (1999) is that it does not matter at what level football is exercised, rivalry among clubs is one of the main sources of attractiveness of a league. Still, only a limited number of researchers have examined rivalry in football, or in sports in general. For example Leach et al. (2003) found that the football fans of the Dutch national team were very happy in the World Cup tournament of 1998, at which the German national team lost against Croatia. While Germany and Holland were both placed in another grouping of teams, and the Germans were eliminated earlier than the Dutch. In General, the Dutch football fans regard the Germans as their arch-rival, by which the results of Leach et al. (2003) indicate that a disappointment of a rival team is psychologically more beneficial in case the interest in sports become larger. Stadtmann (2006) found a similar argument. He studied the effect of unexpected positive results of Bayern Munich on the share price of Borussia Dortmund. As these clubs are rivals of each other in terms of the championship trophy in the German Bundeliga. Stadtmann (2006) found that a (unexpected) success of Bayern Munich results in a negative effect on Borussia Dortmund’s share price.
8
those rivalries. Schadenfreude is namely described as some kind of mood resulting from the performance of a particular team that is considered to be important for an investor.
Demir and Rigoni (2017) examined only one single rivalry. The rivalry between Lazio Roma and AS Roma, two clubs listed on the Italian stock exchange. However, many more rivalries exist between football clubs. A highly ranked rivalry is the rivalry between Fenerbahçe and Galatasaray (www.footballderbies.com). This rivalry is called the “Kitalar Arasi Derbi”, which is Turkish for “Intercontinental Derby”. This rivalry has its name due to Fenerbahçe coming from the Asian part of the capital Istanbul and Galatasaray coming from the European part of the capital Istanbul. Other highly ranked rivalries, according to www.footballderbies.com are; “El SuperClasico” between Boca Juniors and River Plate (both football clubs coming from the Argentina capital Buenos Aires); “The Old Firm” between Celtic and Glasgow Rangers (both football clubs coming from the Scottish capital Glasgow); and “El Clasico” between Barcelona and Real Madrid (both coming from Spain, and battling for the Championship trophy in the Primera Division).
The expectation here is that the performance of the rival in terms of sportive success is also considered to be relevant to football investors. Thus, the positive mood impact of the fan investors which resulted from a win should increase in case the period rival lost. Contrary to this, the negative mood impact of the fan investors which resulted from a loss should increase in case the period rival won.
Hypothesis 1: If both the focal team and the period rival team win their match the share price of the focal team is positively influenced.
Hypothesis 2: If the focal team wins their match and the period rival team loses their match the share price of the focal team is strong positively influenced.
Hypothesis 3: If the focal team loses their match and the period rival team win their match the share price of the focal team is strong negatively influenced.
Hypothesis 4: If both the focal team and the period rival team lose their match the share price of the focal team is negatively influenced.
9
next two hypotheses, hypothesis 5 and hypothesis 6, are testing this word “strong” more explicitely. More specifically, hypothesis 5 is testing the difference between hypothesis 1 and hypothesis 2, and hypothesis 6 is testing the difference between hypothesis 3 and hypothesis 4.
Hypothesis 5: In case the focal team wins and the period rival loses the share price of the focal team is more positively influenced than when the period rival wins as well.
Hypothesis 6: In case the focal team loses and the period rival wins the share price of the focal team is more negatively influenced than when the period rival loses as well.
3. Data Collection
As the subject of this research is the effect of period rivalry in the football industry on stock price performance, it is necessary to identify the football clubs that have a listing on a stock exchange. Such a list is provided by the Stoxx Football Index (Appendix A). According to this Stoxx Football Index list there exist 22 European football clubs that have a listing on a stock exchange. Besides this European listing, the football club Manchester United is listed as well, however, they are listed on the New York Stock Exchange. That is why this football club is not included in the Stoxx Football Index. This research will, therefore, exclude the football club Manchester United from this examination.
10
Table 1 – Listed football clubs and their listing characteristics Football Club Stock Code Country Stock Exchange
AS Roma BIT: ASR Italy Italy Stock Market (FTSE MIB)
Besiktas IST: BJKAS Turkey Turkey Stock Market (XU100) – BIST 100 Borussia Dortmund ETR: BVB Germany German Stock Market (SDAX)
FC Porto ELI: FCP Portugal Portugal Stock Market (PSI All Share Index) Fenerbahçe IST: FENER Turkey Turkey Stock Market (XU100) – BIST 100 Galatasaray IST: GSRAY Turkey Turkey Stock Market (XU100) – BIST 100 Juventus BIT: JUVE Italy Italy Stock Market (FTSE MIB)
Lazio Roma BIT: SSL Italy Italy Stock Market (FTSE MIB) SL Benfica ELI: SLBEN Portugal Portugal Stock Market (PSI20) Sporting CP ELI: SCP Portugal Portugal Stock Market (PSI20)
Trabzonspor IST: TSPOR Turkey Turkey Stock Market (XU100) – BIST 100
The clubs presented in table 1 are all listed on their domestic stock exchange. The Italian football clubs: AS Roma, Juventus, and Lazio Roma have a listing on the Italian Stock Exchange, the Turkish football clubs: Besiktas, Fenerbahce, Galatasaray, and Trabzonspor have a listing on the Turkish Stock Exchange, the Portuguese football clubs: FC Porto, SL Benfica, and Sporting CP have a listing on the Portuguese Stock Exchange, and Borussia Dortmund has a listing on the German Stock Exchange. Except for the Portuguese football clubs, the focus of this research, to test the hypotheses, is on the seasons 2011/2012 till 2016/2017, as all football clubs have a listing from 2010 onwards. The focus for the Portuguese football clubs is on the seasons 2013/2014 till 2016/2017 as the market data for the PSI All-Share Index is only available from October 15 2012.
11 Table 2 – Rivals based on rivalry score
Football Club Two-season period Period Rival
AS Roma (Ita) 2011/2012 Juventus 2012/2013 2013/2014 Juventus 2014/2015 2015/2016 Atalanta 2016/2017 Besiktas (Tur) 2011/2012 Fenerbahce 2012/2013 2013/2014 Galatasaray 2014/2015 2015/2016 Istanbul Basaksehir 2016/2017
Borussia Dortmund (Ger)
2011/2012 Schalke 04 2012/2013 2013/2014 Borussia Mönchengladbach 2014/2015 2015/2016 Bayern München 2016/2017 FC Porto (Por) 2011/2012 SL Benfica 2012/2013 2013/2014 SL Benfica 2014/2015 2015/2016 Sporting Lisbon 2016/2017 Fenerbahce (Tur) 2011/2012 Karabukspor 2012/2013 2013/2014 Akhisar Belediyespor 2014/2015 2015/2016 Antalyaspor 2016/2017 Galatasaray (Tur) 2011/2012 Gaziantepspor 2012/2013 2013/2014 Kasimpasa 2014/2015 2015/2016 Istanbul Basaksehir 2016/2017 Juventus (Ita) 2011/2012 AC Milan 2012/2013 2013/2014 Fiorentina 2014/2015 2015/2016 Palermo 2016/2017
Lazio Roma (Ita)
2011/2012
Genoa 2012/2013
12 2014/2015 2015/2016 Napoli 2016/2017 SL Benfica (Por) 2011/2012 FC Porto 2012/2013 2013/2014 Sporting Lisbon 2014/2015 2015/2016 FC Porto 2016/2017
Sporting Lisbon (Por)
2011/2012 Maritimo Funchal 2012/2013 2013/2014 FC Porto 2014/2015 2015/2016 SL Benfica 2016/2017 Trabzonspor (Tur) 2011/2012 Fenerbahce 2012/2013 2013/2014 Besiktas 2014/2015 2015/2016 Fenerbahce 2016/2017
www.soccerway.com provided all the match information (such as the goals scored, the minutes the goals were scored, and the results) that was needed to test the hypotheses. Other data that is needed here are market indices, share prices, alphas, and betas that are collected from Datastream at the RUG University Library. Due to the available information and the criteria to which this research is subject to, this database consists of 4200 matches that are analysed for the hypothesis testing which are spread over 11 different football clubs.
4. Methodology
4.1 Rivalry Score
Multiple factors could be considered as important in order to consider a particular football club as the “arch-rival” of your focal team. Such as, city pride (Guschwan, 2007) or political beliefs (Demir and Rigoni, 2017) etc. However, the question that arises here is whether the considered “arch-rival” is actually the rival in economic terms. Therefore, a rivalry score is given to the competitors of the examined teams.
13
total lists of rivalry scores per examined team are given in Appendix C. The rival teams that get a rivalry score are only the teams that played the examined team two years in a row.
The rivalry score consists of multiple scores. First of all, the match result could be a win, draw or loss for a particular team. In general this results in three points, one point, or no points respectively. Therefore the score a team is given for match result is thus three points, one point, or no points in case of a win, draw, or loss respectively. Another score that is given to a rival is the goal difference. Because in simple terms, the difference between the goals scored is representative for the ease of the match. However, a 2-0 match result is not the other 2-0 match result. To give an example: it could be that a team scores the 1-0 in the 3’ minute and the 2-0 in the 15’. Based on the 2-0 goal difference early in the match the winning team completes the match at his dead ease. The other case could be that the one team scores the 1-0 in the 68’ minute and the final 2-0 in the 90’+ 2’ minute. This second example indicates that the 2-0 win was much more difficult than the first 2-0 win example. Therefore the percentage of time that still has to be completed after scoring the goal is added as well to the rivalry score. This results in the following equation:
𝑟𝑖𝑣𝑎𝑙𝑟𝑦 𝑠𝑐𝑜𝑟𝑒 = 𝑚𝑎𝑡𝑐ℎ 𝑟𝑒𝑠𝑢𝑙𝑡 + 𝑔𝑜𝑎𝑙 𝑑𝑖𝑓𝑓𝑒𝑟𝑒𝑛𝑐𝑒 + 𝑝𝑙𝑎𝑦𝑖𝑛𝑔 𝑡𝑖𝑚𝑒 𝑠𝑐𝑜𝑟𝑖𝑛𝑔 (1)
Match result is a score of 0, 1, or 3 points. In case one of the focal teams loses, the “match
result” is 3 points, in case of a draw 1 point and in case of a win 0 points. Goal difference is
14
Rivalry Score: Juventus
“Match Result” = 0 “Goal Difference” = -1 “Playing time scoring” = -0,411 ((90-87)/90) - ((90-61)/90) - ((90-79)/90)
Rivaly Score = -1.411 Rivalry Score: Juventus
“Match Result” = 3 “Goal Difference” = 1 “Playing time scoring” = 0.844 ((90-14)/90)
Rivaly Score = 4.844
Rivalry Score: Juventus
“Match Result” = 3 “Goal Difference” = 1 “Playing time scoring” = 0.144 ((90-77)/90)
Rivaly Score = 4.144 Rivalry Score: Juventus
“Match Result” = 1 “Goal Difference” = 0 “Playing time scoring” = 0.156 ((90-64)/90) - ((90-78)/90)
Rivaly Score = 1.156
15
4.2 Hypothesis Testing
As described in the previous sections the period rivalry effect in the football industry is examined by using the effect of match results of the focal team in combination with the match results of the period rival team on the stock price of the focal team. As these match results are a specific result on a particular point in time, these match results could be considered as events. The literature in Finance, especially about stock prices, is familiar with event studies as a stock price is present for each day and innumerable factors could have an influence on the stock price for a particular company or business.
According to Becker and Suls (1983) the performance of a football club increases attendance rate of that football club, and thus the stock price should increase. Sponsoring contracts are also positively influenced by increasing sportive performance (Ngan et al., 2011). Lucifora and Simmons (2001) state that merchandise income is increased due to increasing positive sporting performance in the football industry. In summary, the main sources of income for a professional football club (Scholtens and Peenstra, 2009) are directly positively affected by the performance of that professional football club. However, stock prices are influenced by innumerable factors. Therefore the event study methodology is used as the effect of a specific event on the stock market is tested by which match outcome is the event which could be a win, draw or loss. Besides, the effects that are tested here are the sudden mood changes of investors due to the performance of the focal team as well as the period rival team and according to Edmans et al. (2007) this is clearly identified by an event approach.
In order to capture the effect of an event on the stock prices, the return that could not be explained by the market should be filtered out. This part is called the abnormal return in the Finance literature. This abnormal return is the difference between actual return of a particular stock and the part that can be explained by the market, which results in the following equation:
𝐴𝑅𝑖𝑡 = 𝑅𝑖𝑡− 𝑅̂𝑖𝑡 (2)
𝐴𝑅𝑖𝑡 is the abnormal return at time t for stock i, by which the normal estimated return for a particular football club i at time t (𝑅̂ ) is given by the formula: 𝑖𝑡
𝑅̂ = 𝑎𝑖𝑡 𝑖 + 𝛽𝑖𝑅𝑚𝑖𝑡+ 𝜀𝑖𝑡 (3)
This is the market model. 𝑎𝑖 is the alpha of stock i, 𝛽𝑖 is the stock prices’ sensitivity to the market return, also defined as the beta of stock i, and 𝑅𝑚𝑖𝑡 is the market return of stock i at time
16
(4)
Italia All Share Index. Therefore this market index is used for the Italian football clubs. Sporting CP, Porto FC, and SL Benfica are all included in the PSI All Share Index. Therefore this market index is used for the Portuguese football clubs. The Turkish football clubs, Fenerbahçe, Besiktas, Trabzonspor, and Galatasaray are all included in the BIST 100 index. Therefore this market index is used for the Turkish football clubs. Finally, Borussia Dortmund is included in the Small-cap Deutscher Aktieindex. Therefore this market index is used for the German football club.
The first trading day after the match was played is used to calculate the abnormal returns. Football clubs play their matches frequently, sometimes with only 2/3 days between the matches, therefore the event period of a one-day window is used in order to prevent the effect of an overlapping game. Due to this frequency in matches played an estimation period of pre-event data cannot be used here to estimate the parameters in the market model which are used in the abnormal return equation. Multiple other studies (Demir & Rigoni, 2017; Palomino et al., 2009; Scholtens & Peenstra, 2009) have tackled this problem by using a whole sample period available as the estimation period. Therefore, an estimation period of a two season sample period is used here.
Based on this two year sample period as estimation period, the alpha (𝑎̂𝑖) and beta (𝛽̂𝑖) are estimated, which are then used to calculate the abnormal return (𝐴𝑅𝑖𝑡) at day t for each team (i) separately. Subsequently, an ordinary least square (OLS) regression is run to capture the effect of game results on abnormal returns. At first, the literature is verified by analysing what the effect of the match results of the focal team is on its own stock price. This effect is analysed by making use of the following OLS regression:
𝐴𝑅𝑖𝑡 = 𝑎 + 𝛽1𝑊𝑖𝑛𝑖𝑡+ 𝛽2𝐿𝑜𝑠𝑠𝑖𝑡 + 𝜀𝑖𝑡
The 𝑊𝑖𝑛𝑖𝑡 and 𝐿𝑜𝑠𝑠𝑖𝑡 regression only measure the end result of the game by which the 𝑊𝑖𝑛𝑖𝑡
variable takes 1 in case the examined team wins and the 𝐿𝑜𝑠𝑠𝑖𝑡 variable takes 1 in case the examined team loses, in both cases the variable takes 0 otherwise. Furthermore a robustness check is run for this match result effect by using the goal difference score as variable instead of the dummy variables 𝑊𝑖𝑛𝑡 and 𝐿𝑜𝑠𝑠𝑡. This results in the following OLS regression:
𝐴𝑅𝑖𝑡 = 𝑎 + 𝛽1𝐺𝑜𝑎𝑙 𝐷𝑖𝑓𝑓𝑒𝑟𝑒𝑛𝑐𝑒𝑖𝑡+ 𝜀𝑖𝑡
The difference between those OLS regressions is the way they analyse the match result effect on the abnormal stock returns. Each abnormal return (𝐴𝑅𝑖𝑡) is of the first trading day
17
after the game day. Both regressions tackle the match result effect of the focal team on the stock price of that examined team, however, the second OLS regressions which includes the 𝐺𝑜𝑎𝑙 𝐷𝑖𝑓𝑓𝑒𝑟𝑒𝑛𝑐𝑒𝑖𝑡 variable includes the magnitude of a victory or defeat. The 𝐺𝑜𝑎𝑙 𝐷𝑖𝑓𝑓𝑒𝑟𝑒𝑛𝑐𝑒𝑖𝑡 variable represents a win in case the variable has a positive value and
represents a loss in case the variable has a negative value. A draw is represented by the number 0, as both teams scored the same amount of goals. Therefore the magnitude of the match result is considered here as well. In case with a higher difference, the 𝐺𝑜𝑎𝑙 𝑑𝑖𝑓𝑓𝑒𝑟𝑒𝑛𝑐𝑒𝑖𝑡 variable takes a higher value. This robustness test is examined because of investors probably reacting more strongly to a higher/lower goal margin.
The two equations (4) and (5) are used for confirming or rejecting the literature, as these equations analyse what effect the performance of the focal team has on its own stock price. Thus, this part is a confirmation (rejection) of the literature.
The next part makes a contribution to the literature, in particular to the finance research in the football industry with respect to rivalry as it is going to focus on period rivals in the European football industry. Similar to the research by Demir and Rigoni (2017) a dummy interaction variable method is used. Such that the combined performance of both the focal team as well as the period rival team is captured in the analysis. A similar equation is used as in the research by Demir and Rigoni (2017):
𝐴𝑅𝑖𝑡 = 𝑎 + 𝛽1𝑊𝑖𝑛𝑊𝑖𝑛𝑖𝑡+ 𝛽2𝑊𝑖𝑛𝐿𝑜𝑠𝑠𝑖𝑡+ 𝛽3𝐿𝑜𝑠𝑠𝑊𝑖𝑛𝑖𝑡+ 𝛽4𝐿𝑜𝑠𝑠𝐿𝑜𝑠𝑠𝑖𝑡+ 𝜀𝑖𝑡
The variables WinWin, WinLoss, LossWin, and LossLoss are the dummy interaction variables. The variables are relatively simple to explain. In case both considered teams, focal team as well as the period rival team, wins, the WinWin variable takes a value of 1. Otherwise the value for this variable would be 0. In case both considered teams, focal team as well as the period rival team, loses, the LossLoss variable takes a value of 1. Otherwise the value for this variable would be 0. The other two opposite variables are the WinLoss and LossWin variables. If the focal team wins its match and the period rival team loses its match in the similar playing round, the WinLoss variable takes a value of 1. In case of the opposite event, where the focal team loses and the period rival team wins its match in the similar playing round, LossWin variable takes a value of 1.
Furthermore a robustness check is run for this interaction effect by using the goal difference scores as variables instead of the 𝑊𝑖𝑛𝑡 and 𝐿𝑜𝑠𝑠𝑡 combination variables. This results in the following OLS regression:
18
(7)
𝐴𝑅𝑖𝑡 = 𝑎 + 𝛽1𝑃𝐺𝐷/𝑃𝐺𝐷𝑖𝑡+ 𝛽2𝑃𝐺𝐷/𝑁𝐺𝐷𝑖𝑡+ 𝛽3𝑁𝐺𝐷/𝑃𝐺𝐷𝑖𝑡+ 𝛽4𝑁𝐺𝐷/𝑁𝐺𝐷𝑖𝑡+ 𝜀𝑖𝑡
PGD stands for positive goal difference and NGD stands negative goal difference. This means that the variable 𝑃𝐺𝐷/𝑃𝐺𝐷𝑡 takes a value larger than zero in case both the focal team as well as the period rival team wins and the goal difference for the focal team is larger positive than the goal margin for the period rival team. Actually, the PGD of the period rival team is the opposite of the PGD of the focal team. To give an example, in case the focal team wins 2-0, the PGD score for this team is 2. In case the period rival team wins 1-0, the PGD score for this team is -1. Resulting in an overall 𝑃𝐺𝐷/𝑃𝐺𝐷𝑡 variable score of 1 for this example. If one of the two considered teams do not win its match, this 𝑃𝐺𝐷/𝑃𝐺𝐷𝑡 variable takes a value of 0. In case the focal team wins and the period rival team loses, the 𝑃𝐺𝐷/𝑁𝐺𝐷𝑡 variable takes a positive value
due to a win of the focal team, resulting in a PGD score which is positive (+) and the period rival loses, resulting in a NGD score which is positive (+) as well. In all other cases, this variable 𝑃𝐺𝐷/𝑁𝐺𝐷𝑡 is 0. In case the focal team loses and the period rival team wins, the 𝑁𝐺𝐷/𝑃𝐺𝐷𝑡
variable takes a negative value due to a loss of the focal team, resulting in a NGD score which is negative (-) and the period rival team wins, resulting in a PGD score which is negative (-) as well. Again, in all other cases, this variable 𝑁𝐺𝐷/𝑃𝐺𝐷𝑡 is 0. The variable 𝑁𝐺𝐷/𝑁𝐺𝐷𝑡 takes a value higher than zero in case both the focal team as well as the period rival team loses and the goal difference for the focal team is less negative than the goal margin of the period rival team. Finally, to test hypotheses 5 and 6 a t-test is run between the averages of the four variables; WinWin, WinLoss, LossWin, and LossLoss. First regression (6) is analysed for all eleven football clubs. This regression presents a coefficient for the four aforementioned variables for all eleven football clubs. Such that there are eleven WinWin beta’s, eleven WinLoss beta’s etc. Based on these eleven beta’s for each variable, an average beta is given, resulting in an average beta for the WinWin variable, an average beta for the WinLoss variable, an average beta for the LossWin variable, and an average beta for the LossLoss variable. To test hypothesis 5, the average beta of the WinWin variable and the average beta of the WinLoss variable are compared by making use of a t-test. On basis of this t-test result, hypothesis 5 is rejected or not. The same method is used for testing hypothesis 6, however, the t-test representing this hypothesis is comparing the average beta’s of the LossWin variable and the LossLoss variable.
5. Results
19
only is examined, to analyse this effect on the stock price of the focal team. Then the hypothesis testing is analysed.
To have a clear and easy to read chapter, an example of how the regression results should be interpreted is given in this section. The remaining regression results and explanations are presented in Appendix D. The regression results are presented in statistical tables. These statistical tables contain several data. The alpha and the beta for each variable are given in these tables together with the significance of this variable. This significance level is indicated with *’s, by which * represents a statistical level of 10%, ** represents a 5% significance level and *** represents a 1% significance level.
The variables that are given in these statistical tables are the 𝑊𝑖𝑛𝑡, 𝐿𝑜𝑠𝑠𝑡 and 𝐺𝑜𝑎𝑙 𝐷𝑖𝑓𝑓𝑒𝑟𝑒𝑛𝑐𝑒𝑡 to reject or not reject previous literature. The variables WinWin, WinLoss,
LossWin, and LossLoss are the dummy variables and the 𝑁𝐺𝐷/𝑃𝐺𝐷𝑡, 𝑁𝐺𝐷/𝑃𝐺𝐷𝑡, 𝑁𝐺𝐷/𝑃𝐺𝐷𝑡
and 𝑁𝐺𝐷/𝑃𝐺𝐷𝑡 variables are the other dummy interaction variables in order to test my hypothesis. Furthermore the standard errors are given in parentheses. The number of observations are given by “N”. Finally, the R² is given for each regression, which is an explanation of the particular regression.
5.1 AS Roma
20
Table 3 – Market Reaction (AS Roma) AR (1) AR (2) Constant -0.0158*** (0.000) -0.0049** (0.022) Win 0.0256*** (0.000) Loss 0.0055 (0.331) Goal-Difference 0.0060*** (0.000) N 228 228 R² 0.14 0.11
The way the market reacts to the performance of the period rival is also found evidence for. The WinWin variable as well as the WinLoss variable both are statistically significant at a 1% significance level. In case of the WinWin variable, if both the considered team as well as its period rival team wins the return of the considered team increases by 0.0167, thus hypothesis 1 is not rejected. In case the supported team wins and the period rival loses, the return of the supported team increases even more (0.0225). This information causes hypothesis 2 also not to be rejected. Both the LossWin and the LossLoss variable are insignificant at all significance levels, whereby hypothesis 3 as well as hypothesis 4 are rejected.
In case of the robustness check, only the PGD/PGD variable is statistically significant at a 1% significance level, while the other variables are statistically insignificant at all significance levels. This variable states that in case the focal team as well as the period rival team wins, and thus the goal difference increases by 1, the return of the focal team increases by 0.0031. Therefore, hypothesis 1 is considered to be correct. The other three variables,
21
Table 4 – Rival Influence (AS Roma) AR (1) AR (2) Constant -0.0077** (0.013) -0.0042* (0.085) WinWin 0.0167*** (0.000) WinLoss 0.0225*** (0.001) LossWin -0.0027 (0.649) LossLoss 0.0015 (0.921) PGD/PGD 0.0031*** (0.002) PGD/NGD 0.0024 (0.562) NGD/PGD -0.0014 (0.762) NGD/NGD 0.0012 (0.776) N 228 228 R² 0.10 0.05
The results for the other examined listed football clubs, which are presented in statistical tables, and the explanation of these results are presented in Appendix D. Table 5, hereunder, presents the hypotheses results for the eleven examined football clubs:
Table 5: Hypothesis Rejection/No Rejection
Rivalry Hypothesis 1 Hypothesis 2 Hypothesis 3 Hypothesis 4 AS Roma Not Rejected Not Rejected Rejected Rejected Besiktas Rejected Rejected Not Rejected Rejected Borussia Dortmund Not Rejected Rejected Not Rejected Not Rejected FC Porto Rejected Rejected Rejected Rejected Fenerbahce Rejected Not Rejected Rejected Rejected Galatasaray Rejected Rejected Rejected Rejected Juventus Rejected Rejected Not Rejected Rejected Lazio Roma Not Rejected Not Rejected Not Rejected Not Rejected SL Benfica Rejected Rejected Rejected Rejected
Sporting Lisbon N/A N/A N/A N/A
Trabzonspor Rejected Rejected Not Rejected Rejected
Total Not Rejected: 3 3 5 2
22
times for these 10 football clubs. Resulting in little evidence found for the presence of the period rival’s performance in the stock market.
5.2 T-test
In order to test hypothesis 5 a t-test is run between the average of the WinWin variable and the average of the WinLoss variable. Subsequently, a t-test is run between the average of the LossWin variable and the average of the LossLoss variable in order to test hypothesis 6. The following table presents the results for the t-tests:
Table 6 – T-testing
WinWin Probability WinLoss LossWin Probability LossLoss AS Roma 0.0167*** (0.000) 0.0000 0.0225*** (0.001) -0.0027 (0.649) 0.0000 0.0015 (0.921) Besiktas 0.0002 (0.974) 0.0000 0.0059 (0.534) -0.0196** (0.034) 0.0023 0.0031 (0.821) Borussia Dortmund 0.0096*** (0.002) 0.0000 0.0062 (0.126) -0.0092** (0.032) 0.0002 -0.0295*** (0.000) FC Porto 0.0185 (0.127) 0.0000 -0.0220 (0.350) -0.0137 (0.493) 0.0031 -0.0124 (0.778) Fenerbahce 0.0052 (0.340) 0.0038 0.0095* (0.051) -0.0098 (0.157) 0.0003 -0.0052 (0.687) Galatasaray 0.0049 (0.397) 0.0001 0.0038 (0.582) 0.0053 (0.554) 0.0123 -0.1390 (0.296) Juventus -0.0187** (0.055) 0.4607 -0.0104 (0.293) -0.0327* (0.098) 0.3340 -0.0096 (0.678) Lazio Roma 0.0117** (0.045) 0.0010 0.0136** (0.049) -0.0183*** (0.008) 0.0326 -0.0178** (0.032) SL Benfica -0.0004 (0.978) 0.0000 0.0234 (0.311) -0.0021 (0.868) 0.0000 -0.0304 (0.219) Sporting CP - - - - Trabzonspor 0.0047 (0.330) 0.0000 0.0058 (0.374) -0.0182*** (0.000) 0.0000 -0.0042 (0.612) Full Sample: 0.0026 (0.262) 0.0000 0.0066** (0.024) -0.0100*** (0.001) 0.0000 -0.0121** (0.020) Hypothesis 5 is testing whether the WinLoss variable is significantly different from the
WinWin variable, and whether the Winloss variable is more positive than the WinWin variable.
Hypothesis 6 is testing whether the LossWin variable is significantly different from the LossLoss variable, and whether the LossWin variable is more negative than the LossLoss variable.
23
for which the probability of the t-test regarding hypothesis 5 is 0.0038, whereby the WinLoss variable is significant at a 10% level, while the WinWin variable is insignificant. Therefore hypothesis 5 is not rejected for Fenerbahce as well. Regarding Lazio Roma, the t-test representing hypothesis 5 gives a probability of 0.0010. Thus the WinWin variable and the
WinLoss variable are significantly different from each other, whereby both variables are
significant at a 5% significance level and the WinLoss variable is more positive than the WinWin variable. Therefore hypothesis 5 is not rejected for Lazio Roma.
The probability resulting from the Besiktas t-test regarding hypothesis 6 is 0.0023, which means that the LossWin and LossLoss variable are significantly different from each other for all significance levels. The LossWin variable is significant at a 5% significance level and more negative than the LossLoss variable, while the LossLoss variable is insignificant at all significance levels. Therefore hypothesis 6 is not rejected for Besiktas. The same holds for Trabzonspor, for which the probability of the t-test regarding hypothesis 6 is 0.0000, whereby the LossWin variable is significant at a 1% significance level, while the LossLoss variable is insignificant. Therefore hypothesis 6 is not rejected for Trabzonspor as well. Regarding Lazio Roma, the t-test representing hypothesis 6 gives a probability of 0.0326. Thus the LossWin variable and the LossLoss variable are significantly different from each other, whereby the
LossWin variable is significant at a 1% significance level and the LossLoss variable is
significant at a 5% significance level. Furthermore, the LossWin variable is more negative than the LossLoss variable. Therefore hypothesis 6 is not rejected for Lazio Roma. Finally, the
LossWin variable for Juventus is significant at a 10% significance level, while the LossLoss
variable is insignificant at all significance levels whereby the LossWin variable is more negative than the LossLoss variable. However, the probability of Juventus’ t-test regarding hypothesis 6 is 0.3340, and thus hypothesis 6 is rejected for Juventus.
Comparing the Full Sample WinLoss variable and the Full Sample WinWin variable by running a t-test results in a probability of 0.0000, which means that the averages of the WinLoss and WinWin variable are significantly different from each other for all significance levels. Furthermore, the WinLoss variable is more positive than the WinWin variable. Therefore hypothesis 5 is not rejected. Comparing the Full Sample LossWin variable and the Full Sample
LossLoss variable by running a t-test results in a probability of 0.0000, which means that the
24
Table 7: Hypothesis Rejection/No Rejection Rivalry Hypothesis 5 Hypothesis 6 AS Roma Not Rejected Rejected Besiktas Rejected Not Rejected Borussia Dortmund Rejected Rejected FC Porto Rejected Rejected Fenerbahce Not Rejected Rejected Galatasaray Rejected Rejected Juventus Rejected Rejected Lazio Roma Not Rejected Not Rejected SL Benfica Rejected Rejected Sporting Lisbon N/A N/A
Trabzonspor Rejected Not Rejected Full Sample Not Rejected Rejected
Total Not Rejected: 4 3
Regarding hypothesis 1, there is found evidence for multiple football clubs that their stock price increases by a win of the focal team combined with a win of the period rival. However, in case of Juventus, this variable gives a remarkable result. For this football club a win of Juventus combined with a win of their period rival results in a decrease in the share price of Juventus. A possible conclusion for this finding could be that investors consider a win of Juventus’ period rival as more important than a win of Juventus.
25
6. Conclusion
This study examines whether rivalry is present in the stock market regarding the football industry. Specifically, whether the performance of the period rival is present in the stock market. The expectation is that the performance of the supported team and the performance of the period rival(s) could both have an impact on the mood of the fan investor, and therefore could have an impact on the stock price of the focal club.
Gross of the 11 examined football clubs confirm the literature that a win of the supportive team has a positive effect on their stock price and in case a loss of the supported team, their stock price decreases. These findings are even more strengthened by the goal-difference robustness check, that the more positive the goal margin is the more positive the supported team’s stock price is affected.
All four dummy variables, WinWin, WinLoss, LossWin, and LossLoss, is found evidence for. Furthermore, comparing the WinWin variable with the WinLoss variable and comparing the
LossWin variable with the LossLoss and testing whether they are significantly different from
each other is also found evidence for.
Based on these result, I can conclude that investors are driven by mood swings due to rivalry in football. Thus, Is period rivalry in the football industry present in the stock prices of listed football clubs? Yes! Presumably could these results and conclusions be strengthened if more data were available. This data availability is one of the limitations for this research. Only one database program was freely accessible for RUG students by which the data for the Portuguese stock exchange was, for example, only available from 15 October 2012 onwards. Demir and Rigoni (2017) argue that fan investors could weigh the performance of their rival more heavily if the ranking difference between the rival and the considered team decreases. This possibility is, however, not controlled for as the information was not available. Furthermore, the focus of this research is on football clubs coming from the top ten (appendix B) best performing European football competitions. Of course the research could be extended by including more European football competitions and/or football clubs coming from other continents. Which leads to suggestions for future research.
6.1 Future Research
26
27
References
Becker, M. A., Suls, J., 1983. Take me out to the ballgame: The effects of objective, social, and temporal performance information on attendance at major league baseball games. Journal of Sport Psychology, 5(3), 302-313.
Boyle, G., Walter, B., 2003. Reflected glory and failure: international sporting success and the stock market. Applied Financial Economics, 13(3), 225-235.
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.
Demir, E., Danis, H., 2011. The effect of performance of soccer clubs on their stock prices: evidence from Turkey. Emerging Markets Finance and Trade, 47(sup4), 58-70.
Demir, E., Rigoni, U., 2017. You lose, I feel better: rivalry between soccer teams and the impact of schadenfreude on stock market. Journal of Sports Economics, 18(1), 58-76.
Dobson, S.M., Goddard, J.A., 1998. Performance and revenue in professional league football: evidence from Granger causality tests. Applied Economics, 30(12),1641-1651.
Dobson, S., Goddard, J.A., 2001. The economics of football. Cambridge: Cambridge University Press.
Ecer, F., Boyukaslan, A., 2014. Measuring performances of football clubs using financial ratios: the gray relational analysis approach. American Journal of Economics, 4(1), 62-71.
Edmans, A., Garcia, D., Norli, Ø., 2007. Sports sentiment and stock returns. The Journal of Finance, 62(4), 1967-1998.
French, K.R., Poterba, J.M., 1991. Investor diversification and international equity markets. American economic review, 81, 222-226.
Guschwan, M., 2007. Riot in the curve: soccer fans in twenty‐first century Italy. Soccer & Society, 8(2-3), 250-266.
28
Leach, C.W., Spears, R., Branscombe, N.R., Doosje, B., 2003. Malicious pleasure: schadenfreude at the suffering of another group. Journal of personality and social psychology, 84(5), 932-943.
Lucifora, C., Simmons, R., 2001. Superstar effects in Italian football: an empirical analysis. Università cattolica del Sacro Cuore.
Mason, D.S., 1999. What is the sports product and who buys it? The marketing of professional sports leagues. European Journal of Marketing, 33(3/4), 402-419.
Mitchell, H., Stewart, M.F., 2007. A competitive index for international sport. Applied Economics, 39(5), 587-603.
Ngan, H. M., Prendergast, G. P., Tsang, A. S., 2011. Linking sports sponsorship with purchase intentions: team performance, stars, and the moderating role of team identification. European Journal of Marketing, 45(4), 551-566.
Palomino, F., Renneboog, L. and Zhang, C., 2009. Information salience, investor sentiment, and stock returns: the case of British soccer betting. Journal of Corporate Finance, 15(3), 368-387.
Panagiotis, D.E., 2011. Profitability of the Greek football clubs: implications for financial decisions making. Operations Management: A Modern Approach, 159-169.
Samagaio, A., Couto, E., Caiado, J., 2009. Sporting, financial and stock market performance in English football: an empirical analysis of structural relationships. Centre for Applied Mathematics and Economics (CEMAPRE) working papers.
Scholtens, B., 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. Stadtmann, G., 2006. Frequent news and pure signals: the case of a publicly traded football club. Scottish Journal of Political Economy, 53(4), 485-504.
Websites:
29
Appendix A
All European football clubs that have a listing on a European stock exchange.
Football Club Country Listing date
Aalborg Denmark 14-09-1998
Arhus Denmark 20-05-1988
AIK Sweden 31-07-2006
Ajax Netherlands 11-05-1998
AS Roma Italy 22-05-2000
Borussia Dortmund Germany 30-10-2000
Besiktas Turkey 19-02-2002
Brøndby Denmark 05-04-1988
Celtic United Kingdom 28-09-1995
FC Porto Portugal 01-06-1998 Fenerbahçe Turkey 17-09-2004 Galatasaray Turkey 19-02-2002 Juventus Italy 19-12-2001 Lazio Italy 06-05-1998 Lyon France 08-02-2007 Parkensport1 Denmark 13-11-1997
Ruch Chorzow Poland 31-12-2009
Silkeborg Denmark 07-10-1991 SL Benfica Portugal 21-05-2007 Sporting Portugal 02-06-1998 Teteks Macedonia 24-08-2009 Trabzonsport Turkey 15-04-2005 https://www.stoxx.com/index-details?symbol=FCTP
30
Appendix B
UEFA rankings for club competitions
Rank Country Points*
1 Spain 91.855 2 England 64.748 3 Germany 63.998 4 Italy 63.082 5 France 48.415 6 Russia 45.382 7 Portugal 41.582 8 Belgium 37.100 9 Ukraine 35.733 10 Turkey 31.200
* Based on the previous five seasons
31
Appendix C
AS Roma
2011/2012 2013/2014 2015/2016
2012/2013 2014/2015 2016/2017
32
Besiktas
2011/2012 2013/2014 2015/2016 2012/2013 2014/2015 2016/2017 Fenerbahce 13.0333 Galatasaray 22.0111 Istanbul Basaksehir 7.6556 Kayserispor 12.4778 Fenerbahce 11.8556 Fenerbahce 5.8000 Galatasaray 11.0667 Sivasspor 4.4222 Kasimpasa 5.7444 Bursaspor 5.7444 Eskisehirspor 3.0000 Belediyespor 0.1444 Trabzonspor 4.9889 Rizespor 2.4778 Trabzonspor -1.5000 Istanbul Basaksehir 4.1667 Karabukspor 1.6778 Galatasaray -1.6889 Gaziantepspor 4.0556 Genclerbirligi 1.5222 Genclerbirligi -2.3556 Sivasspor 1.1111 Konyaspor -0.6667 Konyaspor -5.5222 Eskisehirspor 0.7444 Belediyespor -5.3111 Rizespor -6.2889 Genclerbirligi -0.1667 Gaziantepspor -5.5778 Bursaspor -6.6222 Mersin Idmanyurdu -3.8667 Kayseri Erciyesspor -5.6111 Antalyaspor -9.6889 Karabukspor -5.5444 Trabzonspor -7.6111 Osmanlispor -13.0444 Antalyaspor -7.4333 Bursaspor -7.8778 Gaziantepspor -17.388933
Borussia Dortmund
2011/2012 2013/2014 2015/2016
2012/2013 2014/2015 2016/2017
Schalke 04 5.7000 Mönchengladbach 15.6000 Bayern München 15.7333 Hamburger SV 2.1222 Bayer Leverkusen 13.0778 Köln 7.1556 Hoffenheim 1.9111 Bayern München 9.9444 Hertha Berlin 4.0222 Stuttgart 1.0667 Wolfsburg 8.4111 Eintracht Frankfurt 2.8222 Bayern München -0.4778 Hamburger 7.3778 Schalke 04 0.7778 Hannover 96 -2.0111 Hertha Berlin 0.2556 Hoffenheim -0.9000 Wolfsburg -2.3000 Schalke 04 -0.5222 Bayer Leverkusen -5.0333 Mainz 05 -6.9444 Hoffenheim -0.9000 Darmstadt -5.1889 Bayer Leverkusen -7.3889 Mainz 05 -1.5333 Augsburg -5.6778 Mönchengladbach -8.4111 Augsburg -3.1111 Ingolstadt -6.4000 Nürnberg -8.6000 Hannover 96 -3.3111 Werder Bremen -6.7111 Augsburg -10.3667 Werder Bremen -4.8333 Hamburger -6.9444 Werder Bremen -13.7778 Eintracht Frankfurt -5.3444 Mainz 05 -7.4556 Freiburg -19.0556 Stuttgart -7.2444 Mönchengladbach -15.1111
34
FC Porto
2011/2012 2013/2014 2015/2016
2012/2013 2014/2015 2016/2017
SL Benfica -0.6667 SL Benfica 11.4444 Sporting Lissabon 13.9444
Olhanense -1.7444 Nacional 3.1667 SC Braga 6.7778
Gil Vicente -2.9667 Maritimo Funchal 1.9111 Moreirense 1.9444 Sporting Lissabon -3.0778 Estoril-Praia -1.0222 Pacos Ferreira 0.0444 SC Braga -7.3556 Sporting Lissabon -1.1111 SL Benfica -0.1889 Rio Ave -7.6333 Vitoria Guimaraes -1.6111 Tondela -1.5667 Pacos Ferreira -8.9778 Belenenses -3.9778 Maritimo Funchal -1.5889 Academica Coimbra -9.0000 Academica Coimbra -6.2556 Vitoria Setubal -1.9000 Maritimo Funchal -11.3222 SC Braga -7.6222 Vitoria Guimaraes -6.4556 Vitoria Setubal -13.5000 Gil Vicente -13.5556 Rio Ave -6.5222 Beira-Mar -14.4778 Pacos Ferreira -13.5667 Estoril-Praia -7.9556 Nacional -14.7667 Vitoria Setubal -14.8333 Arouca -8.3444 Vitoria Guimaraes -15.9778 Rio Ave -14.9333 Belenenses -10.2889
Arouca -16.0556 Boavista -16.1778
35
Fenerbahçe
2011/2012 2013/2014 2015/2016
2012/2013 2014/2015 2016/2017
36
Galatasaray
2011/2012 2013/2014 2015/2016
2012/2013 2014/2015 2016/2017
37
Juventus
2011/2012 2013/2014 2015/2016
2012/2013 2014/2015 2016/2017
AC Milan 3.0111 Fiorentina 2.3000 Palermo 7.0333
Genoa 0.7000 Udinese 0.1111 AC Milan 2.8444
Bologna -1.4667 Genoa -2.0778 Genoa 1.0667
Internazionale -1.7667 Hellas Verona -2.8222 Lazio Roma 0.6778 Cagliari -2.3778 AC Milan -3.7556 Empoli -0.5222 Catania -3.1444 Parma -5.6333 AS Roma -2.7778 Chievo -3.1778 Sassuolo -5.8444 Fiorentina -3.6222 Napoli -3.4667 AS Roma -6.6111 Sassuolo -4.3556 Parma -5.4667 Sampdoria -7.4889 Torino -6.0444 Lazio Roma -5.4778 Lazio Roma -8.3000 Bologna -7.0778
Siena -5.6778 Torino -8.5222 Chievo -8.3222
38
Lazio Roma
2011/2012 2013/2014 2015/2016
2012/2013 2014/2015 2016/2017
Genoa 18.6778 Internazionale 21.9889 Napoli 25.0111 Juventus 16.4778 Genoa 19.6000 Juventus 24.8778 Siena 16.0444 AC Milan 13.3333 AS Roma 15.8333 Catania 13.3556 Parma 10.8000 AC Milan 13.4556 Fiorentina 9.2667 Hellas Verona 9.8556 Chievo 11.6556 Palermo 7.5000 Juventus 6.6778 Torino 8.4333 Napoli 6.4667 Atalanta 6.0889 Internazionale 8.0444 Udinese 6.1333 Sampdoria 5.3556 Fiorentina 4.4889
Parma 4.2778 Udinese 1.5222 Bologna -0.8111
39
SL Benfica
2011/2012 2013/2014 2015/2016
2012/2013 2014/2015 2016/2017
FC Porto 10.6667 Sporting Lissabon 0.7000 FC Porto 10.1889 SC Braga -1.0889 FC Porto -0.4444 Sporting Lissabon 6.1889 Sporting Lissabon -1.3333 SC Braga -2.1667 Vitoria Setubal 1.0778 Academica -2.3222 Pacos Ferreira -5.2333 Boavista 0.5778 Vitoria Guimaraes -6.5444 Vitoria Guimaraes -6.0222 Arouca -5.6556 Olhanense -7.0222 Rio Ave -6.3778 Rio Ave -8.0222 Beira-Mar -7.7889 Estoril-Praia -7.7000 Estoril-Praia -8.1444 Nacional -10.5333 Arouca -7.9556 Pacos Ferreira -10.0556 Maritimo Funchal -10.7889 Belenenses -8.4222 Moreirense -10.9444 Pacos Ferreira -10.9333 Gil Vicente -9.2333 Maritimo Funchal -11.7556 Rio Ave -13.4778 Maritimo Funchal -9.3333 Vitoria Guimaraes -14.0333 Gil Vicente -15.1444 Nacional -11.4778 SC Braga -14.3111 Vitoria Setubal -18.3778 Vitoria Setubal -14.0000 Nacional -15.8111
Academica -19.5111 Tondela -18.4667
40
Sporting Lisbon
2011/2012 2013/2014 2015/2016
2012/2013 2014/2015 2016/2017
Maritimo Funchal 15.6667 FC Porto 12.1111 SL Benfica 4.8111 SL Benfica 13.3333 SL Benfica 8.3000 Rio Ave 3.7111 FC Porto 13.0778 Estoril-Praia 3.0778 FC Porto -1.9444 Pacos Ferreira 7.0444 Vitoria Guimaraes -0.0444 Belenenses -2.1444 Rio Ave 4.3667 Nacional -2.2222 SC Braga -2.7778 Academica 3.1222 Belenenses -3.2778 Tondela -3.2778 Vitoria Setubal 1.5000 Rio Ave -4.2444 Vitoria Guimaraes -4.8778 SC Braga -0.2889 Academica -6.1444 Maritimo Funchal -6.0333 Olhanense -2.0333 SC Braga -6.3444 Pacos Ferreira -7.4222 Gil Vicente -2.8778 Pacos Ferreira -7.1111 Nacional -9.2556 Nacional -4.0444 Maritimo Funchal -8.9778 Estoril-Praia -9.9556 Vitoria Guimaraes -6.4333 Arouca -9.4444 Boavista -10.0333 Beira-Mar -8.7222 Vitoria Setubal -9.7333 Moreirense -10.4889
Gil Vicente -14.6444 Arouca -13.7111
41
Trabzonspor
2011/2012 2013/2014 2015/2016
2012/2013 2014/2015 2016/2017
Fenerbahce 15.1000 Besiktas 18.6111 Fenerbahce 23.8667 Galatasaray 14.8556 Eskisehirspor 12.5667 Galatasaray 18.6889 Bursaspor 8.4556 Belediyespor 11.4444 Osmanlispor 14.2889 Karabukspor 8.0333 Gaziantepspor 11.2444 Besiktas 13.5000 Kayserispor 6.8556 Fenerbahce 9.0000 Istanbul Basaksehir 11.5889 Antalyaspor 6.6667 Kasimpasa 6.4444 Antalyaspor 10.3778
Mersin 5.6222 Galatasaray 5.9222 Konyaspor 10.0667
42
Appendix D
D.1 Besiktas
Testing the literature again finds evidence for the way the market reacts to the performance of the considered football club. The Win variable as well as the Goal-Difference is statistically significant at a 1% significance level. In case of the Win variable, the return increases by 0.0186 in case the considered team wins. In case of the Goal-Difference variable, the return increases by 0.0048 in case the considered team scores one goal more. The Loss variable is insignificant and thus in case of Besiktas, their stock price is not affected due to a loss of Besiktas.
Table D.1 – Market Reaction (Besiktas)
AR (1) AR (2) Constant -0.0110** (0.047) -0.0036 (0.234) Win 0.0186*** (0.005) Loss 0.0017 (0.843) Goal-Difference 0.0048*** (0.006) N 204 204 R² 0.05 0.04
Only the LossWin variable is statistically significant at a 5% significance level. The other variables are insignificant at all significance levels. This LossWin variable states that the market return of the supported team decreases by 0.0179 in case the considered team loses and the rival wins which is according to the literature. Therefore hypothesis 3 is not rejected here.
43
Table D.2 – Rival Influence (Besiktas)
44
D.2 Borussia Dortmund
The literature is proven to be present for the Win variable, the Loss variable, as well as the Goal-Difference variable. In case Borussia Dortmund wins their market return increases by 0.0094. In case Borussia Dortmund loses their market return decreases by even more, as their market return then decreases by 0.0122. In case Borussia Dortmund scores one goal more, their market return increases by 0.0034. These results are in line with the literature.
Table D.3 – Market Reaction (Borussia Dortmund)
AR (1) AR (2) Constant -0.0032 (0.251) -0.0038** (0.015) Win 0.0094*** (0.004) Loss -0.0122*** (0.003) Goal-Difference 0.0034*** (0.000) N 204 204 R² 0.18 0.10
Multiple evidence is found here for the relevance of rivalry in football. The WinWin variable is statistically significant at a 1% significance level and results in an increased market of 0.0096 if both, Borussia Dortmund and its rival, win. Therefore, hypothesis 1 is not rejected. However, the WinLoss variable is insignificant at all confidence levels. Thus hypothesis 2 is rejected. The LossWin variable is significant at a 5% significance level and represents a decrease of the market return of Borussia Dortmund by 0.0092 if Borussia Dortmund loses its match and their rival wins that playing round, thus hypothesis 3 is not rejected. The LossLoss variable is statistically significant as well, even at a 1% significance level, and represents a decrease of 0.0295 if both football clubs lose their match. Therefore hypothesis 4 is not rejected as well.
The PGD/PGD variable is statistically significant at a 1% significance level and results in an increased market of 0.0015 if both, Borussia Dortmund and its rival, wins and the goal difference increases by 1, and thus hypothesis 1 is correct. However, the PGD/NGD variable is insignificant at all confidence levels. Making hypothesis 2 incorrect. The NGD/PGD variable is significant at a 5% significance level and represents a decrease of the market return of Borussia Dortmund by 0.0089 if Borussia Dortmund loses its match and their rival wins that playing round and the goal difference increases by 1, and thus hypothesis 3 is correct. The
45
represents an increase of 0.0062 if both football clubs lose their match and the goal difference increases by 1.This is actually the opposite of what is expected, therefore hypothesis 4 is incorrect.
Table D.4 – Rival Influence (Borussia Dortmund)
46
D.3 FC Porto
Unfortunately, in case of FC Porto, the literature is contradicted, as it is expected that fan investors react positively (negatively) to a win (loss) of the supportive team, but all the variables, Win, Loss, and Goal-Difference, are insignificant at all significance levels.
Table D.5 – Market Reaction (FC Porto)
AR (1) AR (2) Constant 0.0013 (0.915) -0.0016 (0.816) Win 0.0024 (0.865) Loss -0.0212 (0.263) Goal-Difference 0.0011 (0.705) N 132 132 R² 0.02 0.00
Both the WinWin as well as the WinLoss variable are insignificant and thus hypothesis 1 and 2 are rejected. Both the LossWin as well as the LossLoss variable are insignificant and thus hypothesis 3 and 4 are rejected as well. Thereby, the PGD/PGD, PGD/NGD, NGD/PGD, and NGD/NGD are statistically insignificant as well at all confidence levels, making hypothesis 1,2,3, and 4 to be incorrect.
Table D.6 – Rival Influence (FC Porto)
47
D.4 Fenerbahce
Again, the literature is confirmed here. The Win variable is statistically significant at a 1% significance level. If Fenerbahce wins its match, their return increases by 0.0205. The
Goal-Difference variable is statistically significant at a 1% significance level as well. Meaning that
the market return of Fenerbahce increases by 0.0061 in case Fenerbahce scores one goal more. The Loss variable is insignificant and thus in case of Fenerbahce, their stock price is not affected due to a loss of Fenerbahce.
Table D.7 – Market Reaction (Fenerbahce)
AR (1) AR (2) Constant -0.0123*** (0.002) -0.0052** (0.016) Win 0.0205*** (0.000) Loss -0.0001 (0.989) Goal-Difference 0.0061*** (0.000) N 204 204 R² 0.13 0.11
48
Table D.8 – Rival Influence (Fenerbahce)
49
D.5 Galatasaray
Unfortunately, in case of Galatasaray, the literature is contradicted, as it is expected that fan investors react positively (negatively) to a win (loss) of the supportive team, but all the variables, Win, Loss, and Goal-Difference, are insignificant at all significance levels.
Table D.9 – Market Reaction (Galatasaray)
AR (1) AR (2) Constant -0.0026 (0.589) -0.0012 (0.637) Win 0.0051 (0.371) Loss -0.0020 (0.786) Goal-Difference 0.0015 (0.283) N 204 204 R² 0.01 0.01
Both the WinWin as well as the WinLoss variable are insignificant and thus hypothesis 1 and hypothesis 2 are rejected. Both the LossWin as well as the LossLoss variable are insignificant and thus hypothesis 3 and hypothesis 4 are rejected as well. Thereby, the
PGD/PGD, PGD/NGD, NGD/PGD, and NGD/NGD are statistically insignificant as well at all
confidence levels. Thus hypothesis 1,2,3, and 4 are likely to be incorrect. Table D.10 – Rival Influence (Galatasaray)
50
D.6 Juventus
Unfortunately is the literature not confirmed here as the Win variable, the Loss variable, and the Goal-Difference are all three statistically insignificant at all significance levels.
Table D.11 – Market Reaction (Juventus)
AR (1) AR (2) Constant 0.0135 (0.142) -0.0004 (0.942) Win -0.0158 (0.125) Loss -0.0224 (0.163) Goal-Difference 0.0003 (0.915) N 228 228 R² 0.01 0.00
Regarding the market reactions to rivalry are the WinWin variable and the LossWin variable statistically significant at a 10% significance level. The WinWin variable represents a decrease of 0.0187 in market return if both Juventus and its rival win their match. Therefore hypothesis 1 is rejected. The LossWin variable represents a decrease of 0.0327 in market return if Juventus loses their match and its rival win their match. The PGD/PGD, PGD/NGD,
NGD/PGD, and NGD/NGD are all insignificant variables which causes actually causes
hypothesis 1,2,3, and 4 to be incorrect.
Table D.12 – Rival Influence (Juventus)
51
D.7 Lazio Roma
Testing the literature again finds evidence for the way the market reacts to the performance of Lazio Roma. The Win variable as well as the Goal-Difference is statistically significant at a 1% significance level. In case of the Win variable, the return increases by 0.0234 in case Lazio Roma wins. In case of the Goal-Difference variable, the return increases by 0.0063 in case the considered team scores one goal more. The Loss variable is insignificant and thus in case of Lazio Roma, their stock price is not affected due to a loss of Lazio Roma.
Table D.13 – Market Reaction (Lazio Roma)
AR (1) AR (2) Constant -0.0089* (0.054) -0.0021 (0.341) Win 0.0234*** (0.000) Loss -0.0070 (0.237) Goal-Difference 0.0063*** (0.000) N 228 228 R² 0.17 0.12
Regarding the way the market reacts to the performance of the considered rival is found evidence for. The WinWin variable is statistically significant at a 5% significance. In case Lazio Roma as well as their rival wins their match the market return of Lazio Roma increases by 0.0117. Therefore hypothesis 1 is not rejected. The WinLoss variable is statistically significant at a 5% significance level. In case Lazio Roma wins and their rival loses the market return of Lazio Roma increases even more, by 0.0136 and thus hypothesis 2 is not rejected. The LossWin variable is statistically significant at a 1% significance level. In case Lazio Roma loses and their rival wins the market return of Lazio Roma decreases by 0.0183. Therefore hypothesis 3 is not rejected. In case both Lazio Roma and their rival loses their match, the market return of Lazio Roma decreases by 0.0178, as the LossLoss variable is statistically significant at a 5% confidence level causing hypothesis 4 not to be rejected.
For the robustness check, the variables that are statistically significant are PGD/PGD and NGD/NGD. PGD/PGD is statistically significant at a 10% significance level and the
NGD/NGD variable is statistically significant at a 1% significance level. In case Lazio Roma
52
This data causes hypothesis 1 to be correct, while hypothesis 3 and 4 are considered to be incorrect.
Table D.14 – Rival Influence (Lazio Roma)