• No results found

(2)‘TESTING THE UNILATERAL RELATIONSHIP BETWEEN INVESTOR SENTIMENT AND STOCK RETURNS: AN EVENT-STUDY ON PUBLICLY LISTED EUROPEAN FOOTBALL CLUBS E.R

N/A
N/A
Protected

Academic year: 2021

Share "(2)‘TESTING THE UNILATERAL RELATIONSHIP BETWEEN INVESTOR SENTIMENT AND STOCK RETURNS: AN EVENT-STUDY ON PUBLICLY LISTED EUROPEAN FOOTBALL CLUBS E.R"

Copied!
32
0
0

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

Hele tekst

(1)

‘TESTING THE UNILATERAL RELATIONSHIP BETWEEN INVESTOR SENTIMENT AND STOCK RETURNS: AN EVENT-STUDY ON PUBLICLY LISTED

EUROPEAN FOOTBALL CLUBS 2005-2010’

 

         

Student: E.R.J. Stals Student number: S1944592

T: 0031 6 51761899 E-mail: rj.stals@gmail.com Supervisor 1: prof. dr. S. Beugelsdijk

Supervisor 2: prof. dr. R.H. Koning Rijksuniversiteit Groningen Faculty of Economics & Business

Nettelbosje 2 9747 AE Groningen

Acknowledgement

I would like to thank prof. dr. S. Beugelsdijk and prof. dr. R.H. Koning for their great support and feedback during my master thesis. Especially the econometric input and critical view of both supervisors helped me forward. They always supported and reviewed my ideas by doing this particular research and I am very grateful for that.

(2)

‘TESTING THE UNILATERAL RELATIONSHIP BETWEEN INVESTOR SENTIMENT AND STOCK RETURNS: AN EVENT-STUDY ON PUBLICLY LISTED EUROPEAN FOOTBALL

CLUBS 2005-2010’

E.R. STALS

University of Groningen (RUG) ABSTRACT

This paper investigates the unilateral relationship between investor mood and stock returns of publicly listed football clubs. Proxies scrutinized for investor mood are ‘results on the field’ and ‘coach replacements’. Previous studies mainly concentrated on the results on the field and found different results. Therefore, this paper first conducts replication analyses to validate previous results with an extensive sample of 20 football clubs over five years. This study contributes to existing literature in threefold: first, a derby variable is added to the results on the field, second, a new formula for unexpected success is defined and, third, the announcement of coach replacements, i.e. appointment and dismissal, in relationship to stock returns is investigated. The replication results show, when controlled for all games, that wins result in greater abnormal returns (β = .174) than defeats (β = -.150) or draws (β

= -.050). This is also the case when controlled for competition, but not for venue. Away defeats trigger greater abnormal returns (β = -.105) than an away win (β = .094). Mixed results with respect to significance were found after draws. The results for the derby matches show, as expected, significance at 5% and 1% after respectively wins (β = .035) and defeats (β = -.066). No significant results were found after draws. The unexpected success of a game outcome shows highly significant results, excluding expected draws. Interestingly, an expected loss (β = -.119) is penalized more than an unexpected loss by the investors (β = -.075). The results were robust after controlling for a commonly used model to arrive at the unexpected success as well as for outliers. Coach replacement effects on stock returns did not significantly result in abnormal returns. This indicates that the market reacts irrational to results on the field, i.e. abnormal returns are realized, and rational to coach replacements, hence the stock does not outperform the market. These findings indicate that investors in football stock react emotional rather than rational, i.e. irrational, because these ‘fan-investors’ do not seem to trade on new publicly available information that may affect future cash flows, except for some results on the field.

INTRODUCTION

Professional football, soccer in American denotation, was first practiced in England in 1888. The amount of money circulating in football did not decline since then (Dobson & Goddard, 1998). The immense popularity of football all over the globe is exhaustively underpinned by facts of the Fédération Internationale de Football Association (FIFA). In 2007, the FIFA announced that 265 million people played football worldwide and 270 million when referees and officials are included. This indicates that football is one of the most attractive sports in the world. An annual research of Deloitte & Touche (2010) about the financial situation of football clubs reveals that the combined revenue of the top-20 football clubs exceeded four billion euro for the first time in 2009/2010 which indicates that money is exceedingly significant in current professional football. The studies of Deloitte and Touche (2010; 2011) also indicate that merchandizing, sponsor contracts, media contracts (e.g. broadcasting rights), and game merits are the most important regular incomes for football clubs. The need for money and the enormous competition culminated into public offerings of football clubs (Mitchell & Stewart, 2007). Tottenham Hotspur F.C.

was the first football club that issued stocks on the market by an initial public offering (IPO) in 1983 (Baur & McKeating, 2009). This football club was listed on the London Stock Exchange to generate direct money supply and permit preferred access to capital markets in the future. This IPO resulted in 41% equity that assured the ‘company’ ₤3.3 million. The IPO of Tottenham Hotspur was succeeded by clubs like Millwall F.C. and Manchester United in respectively 1989 and 1991, but also in Denmark where Bröndby (1987) and Silkeborg (1989) went to the market. Millwall and Manchester United

(3)

generated both 38% of equity which was turned into respectively ₤4.8 million and ₤6.7 million net proceeds (Andreff & Szymanski, 2006). These enormous supplies of money were a signal to other European football clubs. These football clubs generated millions of euros by their public offering and, between 1995 and 1998, another 15 IPOs were founded. A downward trend was perceived after this boom due to ownership changes and insolvency problems of the football clubs (Andreff & Szymanski, 2006).

The IPOs of the football clubs in this sample are regularly divided over time (see appendix 1). It is clear that an IPO generates an injection of money for football clubs, but the long-term effect is not clear.

Palomino et al (2005) showed that almost every British listed football club suffered negative operational revenues for the last three years after their IPO. Several football clubs already de-listed again and 24 clubs are listed on primary stock exchanges in 2011.

The results on the field of football clubs should possibly affect the mood of the fans, i.e.

investors, and, consequently, the stock returns. Several researches have already been conducted on the existence of investor mood in financial markets (Nofsinger, 2005; Gregory-Allen et al, 2010; Shu, 2010;

Al-Haijeh, 2011; Levi & Yagil, 2011). One of the most outstanding studies on the existence of a relationship between investor mood and sport stocks is conducted by Edmans et al (2007). They find that results on the field during international tournaments affect national stock index returns and elucidate this by fluctuations in investor mood. Other studies tested this relationship as well and reveal mixed results (Renneboog & Vanbrabant, 2000; Ashton et al, 2003; Brounen et al, 2004; Stadtmann, 2006; Edmans et al, 2007; Benkraiem et al, 2009; Berument et al, 2009). Brounen et al (2004) indicate that most investors in these football clubs are emotionally linked to the football clubs. Investors interpret the results on the field as new available information to the market and should take this into consideration when revaluating the firm. This feature makes it interesting to investigate the influence of another proxy that may affect investor mood, like coach replacements. Financial and economic research is already conducted on the market value changes of industrial and commercial companies due to management changes (Brown, 1982; Beattie & Zajac, 1987; Lubatkin et al, 1989; Worrell et al, 1993; Fizel & D’Itri, 1997; Niño &

Romero, 2008). Recent literature started to investigate the effect of coach replacements on the operational performance, i.e. the performance of the football team on the field, and found mixed results (Van Dalen, 1994; Dawson et al, 2000; Boyle & Walter, 2003; Bruinshoofd & Ter Weel, 2003; Hope, 2003; Koning, 2003; Balduck et al, 2010; Hughes et al, 2010; Heuer et al, 2011; De Schryver & Eisinga, 2011).

However, no research is conducted on the relationship between coach replacements and stock returns. All previous studies that investigated the relationship between investor mood and stock returns tested the efficient market hypothesis (Fama, 1970). Fama (1970) designed three directions of this hypothesis: the weak, semi-strong, and strong hypothesis. Previous research did use different forms of this hypothesis which elucidate the different conclusions found in these studies. Most research is based on results on the field and captures historical prices only; hence they test the soft version. When research is based on information from betting companies as well, then the semi-strong hypothesis is tested, like in this research. Under the null hypothesis, the market participants react rational and the market is efficient.

Consequently, no abnormal returns exist, because the economic benefits do not exceed the market returns, hence the market is capable enough to integrate all new publicly available information immediately. This also indicates that, when translating this to the regression results of previous studies, no significant results can be found. The alternative hypothesis states that, according to Fama (1970), the market is inefficient or that the market participant react irrational. Consequently, abnormal returns exist and significant results are found. The relationship between investor mood changes and abnormal stock returns is interesting for football clubs, because, in most cases, the performance is realized on non-trading days. Therefore, the

(4)

market should have anticipated to the news in the stock returns, because the results on the field form a weekly news release. The stock price can increase or decrease, depending on the result, which affect the income of the football clubs, e.g. after some successive bad performances, attendance will decrease; as well as merchandize incomes, and on the long-term, sponsor contracts may be lower. This market reaction is not always common, because football stocks are not just an investment due to investors’ emotional link (Brounen et al, 2004). This allegiance bias (Hirt et al, 1992) and loss-aversion (Kahneman & Tversky, 1979) result from investors who expect their ‘company’ to win and they reduce the probability ex-ante that a football club will lose.

This study researches the relationship between investor mood and stock returns by testing the efficient market hypothesis (Fama, 1970). Investor mood is distributed in two proxies: results on the field, testing the semi-strong hypothesis, and trainer replacements, which tests the weak hypothesis. This paper first conducts replication analyses to discover whether or not results correspond with previous studies.

This is tested by several subsamples for robustness and the usefulness of this study, i.e. to generalize the results. This replication analyses is not tested with hypotheses. Several previous studies did show that mood is influenced by exogenous factors, like the hours of sun per day (Hirshleifer & Shumway, 2003), temperature (Cao & Wei, 2005), and daylight (Kamstra et al, 2000). This makes it interesting to link investor mood to sports, like Edmans et al (2007) did. This research adds a new control group to results on the field by derby games. High emotional investments are common in these games which make them interesting to look after. This research contributes to existing literature by adding a new formula to arrive at the unexpected success of the results on the field and the corresponding influence on investor mood and stock returns. This new formula defines ex-ante whether or not a result was unexpected and not ex-post, like all previous research. Therefore, this formula can be generalized and adapted to other research areas.

In sports, high standards of disclosure are common which make it interesting to investigate the impact of the announcement of coach replacements. Previous research suggests that only a short-term, or none, shock-effect in operational performance is created when a new coach takes over (Van Dalen, 1994; Hope, 2003; Koning, 2003). This paper fills a research gap by evaluating the impact of coach dismissals and appointments on stock returns of the football clubs. So, this paper is based on four pillars: replication analyses, derby games, unexpected success, and coach replacements. This paper contributes to existing literature by adding the last three pillars.

An event-study is conducted and this paper makes use of multiple regression analyses to arrive at the results. The first proxy for investor mood, results on the field, is tested by replication analyses and two subsamples. The replication analyses include: (1) all games, (2) competition, and (3) venue. The results show that almost every subsample generates abnormal returns, just like previous research did. Only the national cup matches do not seem to impact investor mood, just like several drawn games. The derby games in subsample 1a, show that losses are penalized more than wins, but draws, again, are not resulting in abnormal returns. The most interesting subsample of the results on the field is subsample 1b, the unexpected success. This subsample is based on a new formula and the results show that every result, except for an unexpected draw, generates abnormal returns. These results indicate that the efficient market hypothesis does not hold for unexpected success, because abnormal returns are generated.

Investors are subject to several behavioral biases and react irrational. The second investor mood proxy, coach replacements, is tested by dividing the replacements into appointment and dismissal of the coaches.

Then, three subsamples are tested for robustness. The results indicate that no abnormal returns can be generated. This means that the market reacts rational and the efficient market hypothesis applies. So, when the efficient market hypothesis of Fama (1970) is not supported, in the case of several results on the

(5)

field, it does not elucidate that the market is inefficient, but that the investors react irrational due to some behavioral biases. Consequently, current research in this area questions the definition of the efficient market hypothesis; hence not the efficiency of the market itself but the definition of rationality of the market participants and the lack of behavioral input, so this hypothesis does not take behavioral biases into account and, in this study, the market is subject to such biases. This paper is inclined to support the adaptive market hypothesis (Lo, 2005). The adaptive market hypothesis indicates that investors may react irrational due to several behavioral biases. The findings indicate that investors in football stock react emotional and factual rather than rational, because these ‘fan-investors’ do not seem to trade on new publicly available information that may affect future cash flows, i.e. coach replacements, excluding results on the field.

The outline of this paper is as follows: First an extensive theoretical background is provided from economic and sports literature to discuss existing theories and concepts. The latter one encompasses almost every study that is conducted in this research area. Then the hypotheses are formulated. This study consists mainly of four parts: the replication analyses, the derby games, the unexpected success, and the coach replacements. The hypotheses are followed by methodology. Fourth, the results are presented and this paper concludes with a discussion.

THEORETICAL BACKGROUND Theoretical Concepts

The efficient market hypothesis (Fama, 1970) suggests that an efficient market fully reflects the available information. Fama (1970) designed three directions of this hypothesis: (1) the weak hypothesis represents information that only encompasses historical prices, (2) the semi-strong hypothesis includes the level of adjustment of the specific market to publicly available information, e.g. stock splits, and (3) the strong hypothesis which indicates that investors are permitted access to all information that is relevant for possible price fluctuations. Consequently, no abnormal returns can be gathered, because all information is available. This does also mean that no significant results can be found in the regression analyses of this study. Another interesting theory in this respect is an extension of the efficient market hypothesis.

Prospect theory of Kahneman & Tversky (1979) suggests that people make underweighted decisions about the outcomes that are simply possible and outcomes that are gathered with more certainty. This line of reasoning is summarized in the alternative market hypothesis of Lo (2004). Lo (2004) states that the efficient market hypothesis limits the persistence and the degree of behavioral biases, like contradicting probability beliefs, and, consequently, the efficient market should re-price the stock to a decent level of rationality again, just like Hirschleifer (2001) and Baker & Nofsinger (2002) argued for. This implies that no abnormal returns can be gathered, because the market price fully reflects the available information that is publicly available, i.e. the market is efficient. This powerful feature indicates that irrational beliefs are not pervasive enough to exceed the ability of arbitrage capital dedicated to take advantages from. Lo (2004) states that the efficient market hypothesis does not take behavioral factors into consideration, like investor mood. This research tests whether or not the semi-strong efficient market hypothesis is supported for results on the field and the weak hypothesis for coach replacements, or that the stock returns of publicly listed football clubs are subject to behavioral concepts.

Investor Mood and Economic Background

Investor mood is researched thoroughly in the past. Barret et al (1987) indicated that industry specific reactions on the stock market were detected after airplane crashes. Research is also conducted on

(6)

cataclysm, like hurricanes (Lamb, 1998) and earthquakes (Shelor et al, 1992). A more recent study of Gregory-Allen et al (2010) shows that the daylight saving time does not impact stock returns, unlike previous studies that argue that investor mood alters due to induced sleeping patterns (Kamstra et al, 2000). Investor mood is also tested in different areas, like air pollution. Levy & Lagil (2011) researched the relationship between air pollution and stock market returns of four countries by using the Air Quality Index. They found a negative relationship between the degree of air pollution and stock market returns and the relationship attenuates as the air pollution area gets further away from the stock exchange. These studies show that interesting non-economic events impact investor mood and stock returns. In football, Edmans et al (2007) were the first who conducted an in-depth study on the relationship between results on the field, triggering investor mood, and the stock market returns.

Results on the Field on ‘Club-Level’

One of the first studies which tested the fluctuations in stock returns of football clubs was conducted by Corman & Huisman (1998). They map short-term observations of results on the field and trainer replacements and indicate that the share price of Chelsea fell by 6% after the unexpected dismissal of coach Ruud Gullit. Furthermore, Corman & Huisman researched the income flow of AFC Ajax, AC Milan and Manchester United. They concluded that publicly listed football clubs need more stable long- term income flows. Additionally, they question the long-term performances and attractiveness of these funds for institutional investors. Koning (2004) also questioned the long-term profitability of football stock and concluded that results on the field are uncertain and the stock returns volatile. Koning (2004) also inferred a development towards an entertainment industry, where healthy business should be developed more radical.

Renneboog & Vanbrabant (2000) conducted a research on the existence of a relationship between stock returns and performances on the field of 17 British listed football clubs in 1995-1998. They indicated that performance on the field result in a frequent news flow and, consequently, affect the income of the football club. Regression analyses were used to test four subsamples: (1) victory, loss, draw, (2) cup, competition and European games, (3) promotion and relegation, and (4) clubs listed on the Alternative Investment Market (AIM) and the London Stock Exchange (LSE). Renneboog & Vanbrabant (2000) found 1% abnormal return as a result of a victory and abnormal losses of 2.5% and 1.7% for respectively defeats and draws. These results were robust for all competitions. Another interesting finding is the bigger increase after a victory on the LSE and a bigger loss recorded for clubs listed on the AIM.

Also larger abnormal returns were gathered after promotion and relegation games. Brounen et al (2004) considered the same relationship as Renneboog & Vanbrabant. Their sample consisted of 5,127 games of 38 publicly listed football clubs. Brounen et al (2004) argue that investors react rational to performance and, hence, the efficient market hypothesis holds. This different conclusion can be elucidated by the different econometric framework used; hence Brounen et al (2004) are investigating normal returns instead of abnormal returns. Consequently, the investors react rational in a way that negative results are penalized more by investors resulting from their emotional link to the football club.

Zuber et al (2005) argue for investor-fans in this industry. Their sample consisted of 10 publicly listed football clubs in the English Premier League. Zuber et al (2005) concluded that stock returns are very insensitive to game results by mean and volume when controlled for surprise variables, goal difference, and game-related variables. They document that this new type of investors do not benefit from cash flow information, but more from solitary ownership. The study of Stadtmann (2006) concentrates

(7)

only on the results of Borussia Dortmund from 2000-2005 so these results are not generalizable and do not have enough explanatory power to come at valuable conclusions in relation to this study.

An extensive study was conducted by Scholtens & Peenstra (2008) who tested the efficient market hypothesis by using information of publicly listed football clubs. This resulted in positive and significant abnormal returns after victories and negative abnormal returns after draws and defeats. An interesting finding is the negative significant abnormal return after a draw. Scholtens & Peenstra (2008) argue that this is due to the loss aversion theory. This theory (Kahneman & Tversky, 1979) indicates that people, i.e. investors, reply more emotionally after defeats than victories. This can also be underpinned by the allegiance bias (Hirt et al, 1992; Markman & Hirt, 2002), indicating that people are subject to overvaluations of the probabilities that their company will perform above standard. Palomino et al (2008) also argue for the existence of investor sentiment. In their study on football clubs listed on the LSE, they found high abnormal stock returns, indicating that the market reacts irrational due to overreactions caused by sentiment. They did not find the same impact after a loss. Palomino et al (2008) also researched the impact of the release of betting odds, but did not find any market reaction. This discrepancy in results between news releases concerning betting quotes and game results is due to overreactions and information salience of betting quotes. One of the most recent studies is prepared by Berument et al (2009). Their sample consisted of Besiktas, Fenerbahce, Galatasaray, and Trabzonspor. One interesting feature of this research is that the authors’ link the quantity of abnormal returns gathered after performance was realized to the degree of fan fanaticism. Berument et al (2009) concluded that Besiktas had the highest rate of fan fanaticism, because the stocks were more volatile after derby games.

Benkraiem et al (2009) did research on publicly listed football clubs from all over Europe by using the methodology of the Dow Jones Football Index. Their findings indicate that sporty performances affect abnormal returns and trading volumes around match dates. The impact of the sporty performance differs, like previous research indicated (Renneboog & Vanbrabant, 2000; Scholtens & Peenstra, 2008), that the abnormal returns are different for match venue and performance.

Results on the Field on ‘Nation-Level’

Research is also conducted on national level, like the influence of national team game outcomes on investor mood in the form of an increase or decrease of the market index return the trading day after the performance was realized. A study with a small sample was conducted by Ashton et al (2003). They researched the performances of the national football team of England and the reaction of the stock market the day after an important match on an international tournament as well as friendly games. They found that important games caused more movements on the stock market, here the FTSE 100. These movements were positive after a win and negative after a draw and defeat. Interestingly, Klein et al (2009a) detected several mistakes in the empirical set-up of the study of Ahston et al (2003). Klein et al (2009a) imply that robustness checks validated the results, like outliers, sub-samples, and seasonal effects. Boyle and Walter (2003) did research on a single country as well. Their sample consisted on the sporty performances of the national rugby team of New Zealand. They concluded that the irrationality of investor responses to sporty performances is transitory at best, hence no relationship between stock and results on the field were detected.

A more outstanding research in this area was conducted by Edmans et al (2007). These researchers also argued for significant negative abnormal returns after losses, i.e. a loss effect. These results were stronger after important matches and on smaller stock exchanges. No similar results were found after winning games. This indicates that losses are perceived more powerful by investors than wins,

(8)

which can be explained by the allegiance bias, just like Ashton et al (2003) concluded. Another interesting feature of Ashton et al study (2003) is the greatest mean return after draws. This can be due to the use of statistically different measures and techniques by using the mean return, instead of the market model like Edmans et al (2007). Edmans et al (2007) checked robustness by evaluating the same relationship in different areas, like cricket, rugby, ice hockey, and basketball. The loss effect appeared in the field of basketball, cricket, and rugby like the soccer results. Kaplanski & Levi (2010) reviewed Edmans et al paper and designed a new novel approach. The most important finding of their study is that the stock market reaction is not due to game results, unlike Edmans et al (2007) indicate, because it is always negative.

Klein et al (2009b) conducted an event-study on the national football performance of countries which played at the European and World Championships between 1990 and 2006. They tested this to verify whether or not football could have a psychological impact on investors resulting in changes in the price configuration on capital markets. Unlike Ashton et al (2003) and Edmans et al (2007), Klein et al (2009b) support the effective market hypothesis, because no significant relationship could be found. Klein et al (2009b) used betting odds to arrive at a surprise variable, because surprising results could trigger the mood of investors more, but no significant results were found. A reason for the different findings of the studies of Ashton et al (2003) and Edmans et al (2007) in comparison to Klein et al (2009b) can be found in the econometric framework of both types of studies. Where Edmans et al (2007) make use of regression equations with panel-controlled errors and using GARCH (1,1) methods, Klein et al (2009b) use a constant mean model and a refined Markov-switching MM. The latter study is the first by using this technique, almost every other study in this respect make use of regression analyses or ordinary least squares (Renneboog & Vanbrabant, 2000; Scholtens & Peenstra, 2008).

A recent study of Smith & Krige (2010) is also based on the same methodology used by the study of Edmans et al (2007). They evaluate the impact of sporting performance on the national stock exchange of South-Africa by using investor mood variables. The main conclusion is a moderate win effect, oppositely to what Edmans et al (2007) found, i.e. a loss effect. Some limitations of the study of Smith &

Krige are the lack of intra-day market data to discover short-term effects, no control for the weekend- effect, and they did not control for the expected and unexpected results. An interesting study of Gerlach (2011) researched the effect of world championships in countries which did not participate on these championships. Gerlach (2011) concluded that fluctuations in investor sentiment did not affect share prices significantly the day after international sport results were realized.

Coach Replacements and Economic Background

It is difficult for economists and researchers to assess whether or not the profitability of the firm improves when a manager gets replaced, because these replacements are not always observable from available firm- level data sources. This research intricacy does not exist in the football industry due to high disclosure standards. Ordinary multinational companies, like those in the S&P500, replace managers to increase the revenues of the company. In the case of football, it is slightly different, because coaches are responsible for the operational performance on the field. These performances influence financials, like merchandise, attendance, sponsor contracts etc. (Renneboog & Vanbrabant, 2000). Consequently, it is very important for a football club to have a good coach managing the team.

The idea of researching the effect of coach dismissals on stock returns of publicly listed football clubs is motivated by existing economic literature. There is a dichotomy recognizable in existing literature between leadership effects and leadership succession. This research concentrates on the latter.

(9)

Beattie & Zajac (1987) indicate that CEO changes are characteristically translated into negative stock returns. Lubatkin et al (1989) show that investors are predisposed to changes by which outsiders are attracted to the healthy firm. The dismissal of a CEO is reviewed and studied by Worrell et al (1993).

They found a positive market reaction following announcement of replacements of CEOs, where other types of CEO dismissal did not cause market reactions, like insider/outsider denotations. Warner et al (1988) researched the effect of management changes on stock returns for companies listed on the NYSE and the AMEX. They cannot conclude that abnormal stock returns were gathered after the announcement of top management changes. Denis & Denis (1994) found that forced resignations of top management were followed by decreasing operating performance and also large improvements. A recent practical example of Apple perfectly summarizes the relationship between the importance of a manager and stocks.

The CEO of Apple, Mr. S. Jobs, revealed on 18 January 2011 a medical leave and the share price of Apple dropped immediately by 6%. This exemplifies the important position of managers and, when translating it to football, coaches (BBC, 2011).

Coach Replacements and Football

Coaches are held responsible for the performances on the field. Consequently, when a football team loses some matches in a row, the management board and the investors lose their confidence in the abilities of the coach to change the situation. When a coach gets replaced, new information comes available to the market. This new information can change the future cash flows of the company, as well as investor mood.

An interesting research area is the field of insiders and outsiders and is taken into consideration in this study (Koning, 2003). Insiders might be in a favorite role, because they are familiar with the club culture and do not have to adapt a lot, unlike outsiders, who are new to the football clubs. Koning (2003) already mentioned that industrial firms use financial information to evaluate management changes and those financial measures for football clubs is less relevant. This changed over time and now it became more important (Deloitte & Touche’s Football League, 2011). The football coach is an important person in the organization of a football club, because he has an important role in determining the status of the football club. A coach’s main tasks are to determine which players are hired, bought and sold, provide guidance to wins and provide training sessions (Koning, 2003). Coaches get replaced by management for several reasons, like (1) induce a commonly known shock-effect, (2) managers are acquired by other clubs, (3) the managers get fired, or (4) the manager resigns. A recent example has taken place in the Netherlands where coach John van den Brom is bought as a trainer of Vitesse from ADO Den Haag for €500.000 (De Telegraaf, 2011). Another way is to resign from the club and hand-in your contract, just like Villas-Boas of FC Porto did lately. Chelsea wanted to approach the coach from Porto, but Porto did not want to let him go and Chelsea did not want to pay an extensive fee (Nusport, 2011). Villas-Boas was the new coach of Chelsea just a few days after he handed-in his contract.

The effect of changing a coach is researched before in sports literature. Brown (1982) researched 26 football teams in the National Football League between 1970 and 1978. He used random effect panel data models in which he explains the percentage of wins by lagged performance, off-field components of the organization, and succession and turnover processes. Brown (1982) concluded that a small succession effect exists.

A Dutch study of Van Dalen (1994) is based on the Dutch ‘Eredivisie’. Van Dalen investigated whether or not the replacement of the football coach led to an improvement of team performance measured in goal difference and factors influencing this were defined by the quality of the referee, home advantage, team superiority, and the results of previous games. Van Dalen (1994) concluded that coach

(10)

replacements had positive effects on the performance of the football teams in 60% of his cases (3 out of 5). A drawback of this study is the period under study which is one year. Koning (2003) evaluated the firing of a coach and the effect on team performance during the seasons 1993-1998, also in the Dutch Football League. The performance measures were based on operational performance and not on financial performance. Koning (2003) indicates that it is hard to measure the effect of team performance, because, both old and new, managers do not play against the same opponents and, consequently, goal difference alone is not sufficient to evaluate coach replacements. The main finding of this research is that the operational performance does not always improve when coaches are replaced. Bruinshoofd & Terweel (2003) researched whether or not forced resignations improve team performance in the short-term. They used the Dutch Football League for 12 years. The main finding consisted of a decline in performance followed by improvements. This improvement did not exceed the seasonal average performance of the football club. Surprisingly, Bruinshoofd & Terweel (2003) found that, after controlled for several variables, when a coach was not replaced and the old manager would have been coaching the team, the performance would be better. Consequently, they conclude that the dismissal of a coach is neither efficient nor effective.

Dios Tena & Forrest (2007) researched the causes and consequences of coach dismissals in the Spanish Football League from 2002 till 2005. One of the main findings is that coaches got replaced due to relegation danger and the corresponding desirable shock-effect was not found. De Schryver & Eisinga (2011) evaluate the time that coaches get replaced. The results indicate that coaches get replaced when bad performance is confirmed; hence some negative results in succession are realized. An interesting finding is that in the Dutch Football League, coaches get quicker replaced when they lose a match that they should not lose. Wagner (2010) researches the changes ex-ante replacements in the environment of the football club and whether or not these changes influence the replacement. His sample consisted of the German Football League. One of the most important findings of Wagner (2010) is that the shock-effect in performance depends on rewarding schemes.

Theoretical Framework

This study is the first that implements a new formula to arrive at the unexpected success of results on the field and link this to stock returns, develops a derby dummy, as well as linking coach replacements to stock returns of publicly listed football clubs. The replication analyses are not tested separately.

Therefore, five hypotheses are tested.

FIGURE 1 Theoretical Framework

Investor mood Investor Mood Proxies Stock Returns

Results on the Field Derby

Unexpected Success

Stock returns Appointed and Dismissed

Coach replacements Pre and Mid-Season

In- and Outsider

(11)

THEORY AND HYPOTHESES DEVELOPMENT Replication Analyses

The replication analyses encompass proxies for results on the field which were studied in previous research as well. In these studies, researchers used proxies like performance, competition, and venue. This part elaborates on the theoretical and practical relevance of these proxies, but do not test them with separate hypotheses. Therefore, these variables were not added in the theoretical framework.

The shareholder structure of publicly listed football clubs primarily consists of a few stable (controlling) shareholders, a number of institutional investors, and the greatest part consists of individual investors (Renneboog & Vanbrabant, 2000). Most of these individual investors have an emotional link with the football club, e.g. Brounen et al (2004) argued for an emotional stock investment by these individual investors. Consequently, these investors react more emotional and factual rather than rational.

Rationality was tested by Fama (1970) in his paper about the efficient market hypothesis (EMH). This EMH states that in an efficient market the prices always fully reflect the available information to the market. Fama (1970) defined three applicable information flows, as aforementioned. The semi-strong EMH tests whether or not the prices efficiently change to the publicly available information. When these prices react rational, the markets are efficient and no abnormal returns can be gathered. In the case of football, frequently publicly information becomes available to the market. This information consists of the results of the operational performances of the football clubs, hence the game results. A football team plays a game every week, sometimes even twice a week, which results in a constant information flow to the market (Scholtens & Peenstra, 2008). The investors are expected to react to these performances on the field and, consequently, trigger changes in firm value of the football clubs. When a football club loses a game, the investors receive negative information. Consequently, attendance decreases in number, as well as merchandising products, the catering incomes, and, when a team loses some games in succession, lower sponsor contracts. All these fluctuations in income impact the stock returns of especially firms like football clubs with their emotional investors, because the fluctuations influence their mood (Renneboog

& Vanbrabant, 2000). The reaction of these investors are expected to be positive (negative) after a win (lose and draw) which will result in positive (negative) abnormal returns.

This will be even more when one game is more important than one another. This can be related to the stock prices in a way that these trigger higher degrees of abnormal returns when a game becomes more important. The importance of a game depends on the outcome consequences, which is most appropriate in international games. The UEFA (2011) gives every football club participating in the Champions League €7.2 million when they qualified. AFC Ajax received in December 2010 a total of

€9.2 million, due to the UEFAs bonuses of €800,000 after a win and €400,000 after a draw. Juventus FC also boosted their revenue by qualifying and playing in this prestigious Champions League. Their revenue increased from €35.7 million to €203.2 million in 2009 (Deloitte’s Football Money League, 2010). These earnings will in some cases be fully settled in the share price, as the case for the Turkish listed football clubs (Berument et al, 2009). It is expected that this triggers more attendance, more income from merchandise, and it will positively influence the tendency of the clubs’ sponsors. The increase in merchandise can be explained by ‘impulse purchasing’, e.g. AFC Ajax won the championship in 2010- 2011 after 7 years without any championship and they finally reached 30 championships in their history, which results in three stars on their tricot. The official AFC Ajax fan-shop reported a positive chaos in their office after all inquiries for merchandise, season tickets and business seats (AFC Ajax NV, 2011).

The share price of AFC Ajax NV also increased by 8.7% the first trading day after the championship was won (De Financiële Telegraaf, 2011). All these events influence the investor mood which results in higher

(12)

abnormal returns when a match becomes more important, i.e. in this paper controlled for by competition dummies.

The venue where the game is played is already calculated in the odds given by the betting offices.

Overall, these offices are well-informed and the quotes seem to be a good reflection of the outcome prediction of the games (Palomino et al, 2008). This paper adds a venue variable as well to capture the particular reaction after performance was realized at home ground or away. Koning (2010) researched the home advantage for men and women in tennis and found different results, hence there exists a home advantage for men, but there is not a home advantage for women in tennis.

Derby

Brounen et al (2004) argue that investors have emotional ties to the football clubs in which they invest.

Therefore, it is expected that, hence the individual investors form a large proportion of the ownership of the football club, derby games trigger abnormal returns. It is not expected that this exceeds the abnormal returns when controlled for all games, because the derby games form only a fraction of the total sample.

The draws are not taken into consideration here, because they only form 66 games. The weak efficient market hypothesis is tested here, because this analysis is based on historical prices only. The hypothesis follows:

Hypothesis 1a: Investor mood is positively (negatively) influenced when a derby game was won (lost) resulting in positive (negative) abnormal returns.

Unexpected Success

Theoretical and empirical evidence shows that investors respond stronger to losses than to wins (Kahneman & Tversky, 1979; Boyle & Walter, 2003). A negative reaction and abnormal return is expected after a draw and loss due to the allegiance bias (Hirt et al, 1992). This allegiance bias explains that individuals, with psychologically investment in a particular outcome cause biased expectations which are the result of overweening conceit of investors. Consequently, investors of football clubs expect their team to win as a result of this biased pre-expectation of the game. Consequently this is translated into negative reactions when these pre-expectations cannot be satisfied (Hirt et al, 1992; Edmans et al, 2007).

Klein et al (2009b) researched the relationship between the surprising outcome of a football game and the stock market return. Klein et al (2009b) argue that surprising outcomes trigger more national media coverage resulting in more impact on investor mood. Edmans et al (2007) indicate that losses are almost always unexpected due to allegiance bias. Palomino et al (2009) argue that investors overreact to unexpected outcomes as well. The efficient market hypothesis of Fama (1970) states that share prices reflect the rational expectations, but did not take investor sentiment into account. Edmans et al (2007) argue that market efficiency forecasts that investors must price in the expected economic impact of football ex-ante the football game; hence they do not take unexpected results into account. This suggests that unexpected results trigger higher degrees of abnormal returns of the football club when the semi- strong hypothesis is tested. The hypothesis follows:

Hypothesis 1b: Investor mood is triggered more after unexpected results were realized resulting in higher abnormal returns than after expected results.

Coach Replacements

Managers get replaced when the performances do not satisfy the expectations of the board. The same happens within football clubs. When a coach cannot manage the team properly and the team loses several

(13)

matches in succession, the chance of replacement increases. Koning (2003) investigates sample selectivity for measuring the effect of firing a coach and whether or not firing a coach and appointing a new coach improves the operational performance of the football team. Van Dalen (1994) also investigated this issue.

A very recent paper of De Schryver & Eisinga (2011) investigates the timing to dismiss a coach. All these papers show that typically the operational performance change ex-ante and ex-post the replacement-event is measured. This paper reviews the effects of pre and mid-season coach replacements on the stock returns of the professional football clubs, as well as the effects of insiders and outsiders in relation to the stock returns. The coach replacements are divided in appointment and dismissal of coaches. The investors receive new information and these investors will react to this by interpreting it as positive or negative news. The event differs per replacement, sometimes a coach is immediately replaced, sometimes after two or three games. Therefore, this paper looks into the stock price movements on the day that the coach is dismissed and the day that the new coach is appointed.

Replacing coaches on the field does not always result in the commonly known ‘shock-effect’ in operational performance of the football clubs (Van Dalen, 1994; Koning, 2003). Warner et al (1988) researched the relationship between stock prices and management changes and did not discover significant stock price movements following management changes. This could be different for the football stock reactions, because the investment in stocks is made with different perceptions. The football stocks can be described as emotional investment stocks. This could lead to the importance of insider/outsider determination when new coaches are appointed to the club. Insiders do already have inside-knowledge and know what is expected from them and what the insider can expect from the club as well. An interesting study of Holmes (2010) indicates that insider head coaches get less quickly replaced than outsiders. This leads to high level of confidence within the football club and investors are expected to react to this positive news of attracting an insider. Outsiders are new to the football club and are not familiar with the club’s culture and stakeholders. This may lead to a lower level of confidence ex-ante in the outsiders’ ability to adapt to the new club. Therefore, I expect that the new insider coach appointment will gather greater abnormal returns than outsiders will. The hypotheses follow:

Hypothesis 2a: Investor mood is positively influenced when a coach is appointed or dismissed which results in abnormal returns.

Hypothesis 2b: Investor mood is more positively influenced when an insider is appointed or dismissed than when an outsider is appointed or dismissed resulting in higher abnormal returns.

Hypothesis 2c: Investor mood is more positively influenced when a coach is appointed or dismissed in the pre-season than in the mid-season resulting in higher abnormal returns.

METHODS Sampling Frame and Empirical Setting

This research is based on the relationship between stock returns of football clubs and different proxies of investor mood. These proxies are divided in results on the field and coach replacements. The sample consists of football clubs listed on the Dow Jones Football Index. Dow Jones uses only football clubs that are listed on primary stock exchanges and, therefore, excludes clubs listed on stock exchanges like the Alternative Investment Market in Britain. The clubs are selected from the publicly available monthly report from March 2011. Further selection was necessary to create satisfactory comparisons and only football clubs with the IPO before 2005 were selected. Ruch Chorzow from Poland was excluded due to

(14)

information inconsistency and non-availability. This leaves a sample of 20 football clubs of 8 countries:

AFC Ajax (The Netherlands), Borussia Dortmund (Germany), Besiktas JK, Fenerbahce, Galatasaray, Trabzonspor (all Turkey), Aalborg, Aarhus GF, Bröndby IF, FC Copenhagen, Silkeborg IF (all Denmark), AS Roma, Juventus FC, SS Lazio Roma (all Italy), Glasgow Celtic (Scotland), Millwall FC, Tottenham Hotspur FC, Watford (all England), FC Porto, and Sporting Club de Portugal (both Portugal).

The football club names are used in this research instead of their investment company names to avoid confusion. The market index returns of the countries’ market indices are used to define the abnormal returns. The following market returns are used in the corresponding countries: AEX (The Netherlands), DAX 30 (Germany), ISE 100 (Turkey), OMXC 20 (Denmark), Milan 30 (Italy), FTSE 100 (Great Britain), and the PSI 20 (Portugal). The overall sample consists of 4619 matches, which is the total of all national and international games of all football clubs. This sample size differs per hypothesis, because different subsamples are defined. These subsamples and replication analyses are created for consistency and robustness of the results.

Data Sources

The stock returns of the football clubs are gathered from Thomson’s DataStream. This database provided all stock returns for all football clubs between 2005 and 2010 as well as all corresponding national market returns and the market returns of the Dow Jones Football Market Index. Due to the use of football clubs from different countries across Europe, the stock and market returns are translated to the € at the exchange rate of that particular day which were available from this database as well. Game results of all football clubs were generated from the freely available public website www.footballdatabase.eu. This website did not provide sufficient results on national cup matches for Turkish and Danish football clubs.

This problem was solved by using the national football federations’ websites. Other match results were randomly double-checked by the official websites of the countries football federations, e.g. www.knvb.nl for AFC Ajax (The Netherlands). Betting odds are used to arrive at the expected and unexpected game results and were gathered from www.betexplorer.com. Another important variable concerns derby matches. This information was gathered from www.footballderbies.com. The coach replacements were gathered from multiple data sources, to avoid inconsistency. The most reliable and extended data source used is LexisNexis News portal. This database provides all Dutch and foreign news releases. The dates of coach dismissals and appointments were double-checked by reliable nationwide specialized football websites, like Voetbal International (www.vi.nl), Goal (www.goal.com), and Kicker (www.kicker.de).

This research is based on multiple data-sources that have been double checked by different data sources.

This is important with respect to postponed or withdrawal of football matches, otherwise the corresponding return of stock, exchange rate and market return are biased.

Dependent Variable and Analytical Technique

This research conducts an event-study methodology to evaluate the impact of football games on the stock market returns of 20 publicly listed football clubs. This methodology ensures to capture the impact of each single event i.e. the football game or the dismissal of a coach. The main benefit from this type of methodology, in contrast to a continuous variable, is that it evidently spots changes in investor mood.

Consequently, this provides a large signal-to-noise ratio in the returns (Edmans et al, 2007). MacKinlay (1997) and Brown & Warner (1985) review this methodology. The outcome of the event can be treated as a win, draw are defeat. The following formula (1) is used to arrive at the normal return Rit of club i at time t.

(15)

=

=

Δ +

it it it

it P

P R P 1

% (1)

When Rit is a non-trading day, the trading day before the event is measured (t); Rit+1 is the stock return of football club i at the day after the event, or, when this is a non-trading day, the first following trading day (t+1). The normal return should be settled, i.e. the return that is the expected return when the event would not have taken place. The difference between the normal and expected returns is called the abnormal return. The market model should be defined to come at expected returns. This paper follows Brown &

Warner (1985) and MacKinlay (1997) to define a market model (2):

it mt i i

it R

R =α +β +ε (2)

Where Rmt is the percentage change of the market index return of the country’s market as well as the returns of the DJSTOXX Football index at time t and alpha and beta are estimated parameters by using ordinary least square methods in SPSS. The beta defines the risk of the football clubs’ share in relation to the market. This paper is based on in-sample method to arrive at abnormal returns and these returns are calculated for the whole sample period, hence not for a particular number of trading days. Several researchers showed that is does not make any difference in the outcomes of similar research (Renneboog

& Vanbrabant, 2000; Zuber et al, 2005; Scholtens & Peenstra, 2008). The market model (2) classifies the following formula to arrive at abnormal returns (3):

mt i i

it Rit R

AR^ = α^ β^ (3)

Where ARit is the abnormal return of club i at time t, Rit is the normal return and Rmt is the expected return.

It is expected that the abnormal returns will be positive after a win, hence the investors receive positive news, and negative news after a draw or a lost game, due to the loss-aversion and allegiance biases. The event period for this study is one day, where t is zero at the day of the event and t is one at the trading day after the event. This very short interval in event period provides to avoid overlapping event periods.

Overlapping should be common in the case of football, because events are on a weekly basis. Further analysis is based on multiple regression analyses.

Independent Variables and Model Specification

This part describes the empirical set-up of this research concerning the replication analyses and subsamples. First, the replication analyses are determined with several models followed by respectively, the derby games, unexpected success, and coach replacements.

Replication Analyses

The performance of each football club is divided into a dummy variable to indicate what performance was achieved. In Football, three game results exist: win, draw or lose. Several researchers create three dummy variables to arrive at exact performance (Renneboog & VanBrabant, 2000; Scholtens & Peenstra, 2008;

Berument et al, 2009), others use one dummy variable divided into a minus one for a loss, zero for a draw and a one for a win. This paper uses three different dummy variables to arrive at further data construction and the creation of interaction variables. The win variables (DitW) equal one if a game was won and zero

Referenties

GERELATEERDE DOCUMENTEN

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

clinical  inflammatory  processes  of  RA  [29].  Also,  GH  ratings  have  been  shown  to  be  different  across  patients  with  similar  DAS28  scores, 

This paper reviews role of physical, digital and mixed prototyping in validation tasks.. Prototypes have been extensively employed as transmitters in concept design and

It is interesting to note that whilst the number of narrative fiction films screened at this year’s festival is less than half the number of non-fiction documentary

45 Nu het EHRM in deze zaak geen schending van artikel 6 lid 1 EVRM aanneemt, terwijl de nationale rechter zich niet over de evenredigheid van de sanctie had kunnen uitlaten, kan

Attack step parameters Attacker parameters Attacker skill (β) Attack step difficulty (δ) Attacker speed (τ ) Attack step labor intensity (θ) Outcome / Result Execution time..

X Evaluation of the data dissemination protocol for different wireless sensor network applications: We focus on the performance evaluation of the data dissemination pro- tocol,

In other words, when full scalar invariance holds across groups and over-time, the differences between mean political trust levels among high and low sophistication groups are