Sport sentiment and stock return : do national football results have impact on the stock market?

(1)

1

Master Thesis

Sport Sentiment and Stock Return: Do

National Football Results Have Impact on

the Stock Market?

Author: Shuhao Gu (11292865)

MSc. in Finance, Asset Management Track

Amsterdam Business School, University of Amsterdam

Faculty of Economics and Business

Supervisor: Liang Zou

(2)

2

Statement of Originality

This document is written by Student Shuhao Gu who declares to take full

responsibility for the contents of this document.

I declare that the text and the work presented in this document is original

and that no sources other than those mentioned in the text and its references

have been used in creating it.

The Faculty of Economics and Business is responsible solely for the

supervision of completion of the work, not for the contents.

(3)

3

Abstract

This thesis investigates whether investor sentiment can have impact on stock return. Here we choose national football results as the sentiment variable because football results have obvious influence to people’s mood.

According to the results, we document a negligible positive stock market effect after a win however a significant negative stock market effect after a loss in international football match. For example, a loss in the elimination games of World Cup leads to abnormal stock return of -25 basis point on the next trading day. Moreover, this loss effect is stronger in more important games. In details, the statistical significance experiences a decline through elimination games, group games and qualifying games and World Cup has larger loss effect than European Cup for all three game groups.

(4)

4

We all know that Investor sentiment has no role in classic finance theory. Classic finance theory thinks that all the investors in the stock market are rational while doing decisions. This theory also indicates that investors are in an efficient market where people are able to process all available information. That means asset prices are rational so that can reflect all information relative to future economic prospects. (Fama, 1991) However, many recent studies pointed that classis finance theory ignores the importance of investor sentiment and mood, which is the core of behavioral finance theory. The fact is that investors are not completely rational, they may be impacted by their mood so that errors may be systematically and continuously created, and thereby resulting pricing deviation from the pure rationality. With more studies providing evidence that investor sentiment brings serious consequence to stock market, investor sentiment is becoming an important factor on investment decision and behaviors.

Providing a significant challenge to standard finance theory, these studies drive people started to investigate what investor sentiment is, how mood impacts stock market and what particular factors are considered as component of sentiment. Then many factors such as Friday or non-Friday, weather, sunshine, family are researched particularly to find the relation between stock return and these factors. Some previous work examined the relationship between weather in a city and daily market return and proved that sunshine is significantly correlated with stock returns (see Hirshleifer, D. and Shumway, T., 2003). This result is explained that investors are more likely to be in a good mood in a sunny day so that they intend to be more optimistic and investor more in local stock market. In the contrast,

(6)

2

investors are more likely to be sad and pessimistic with a terrible weather so that they intend to sell more in stock market. Similarly, sports seem to significantly affect people’s mood just like weather and sunshine. People are generally in a happy mood after a win of their national sports team and otherwise in a bad mood after a loss. Therefore, can sport results also be considered as a factor to impact stock returns in many works?

Football economics is growing rapidly in applied economics because football is the most popular sport in the world. Billions of people concentrate on football matches and may be affected by the final football results. Our study selects football section to investigate whether National Football Results have Impact on the Stock Market. The psychological literature on football results, mood and sport sentiment is presented in the nest section. In this paper, we use investors’ mood as connection. First relation is the impact of football results on investor’s mood that is proved by some works.( see Baker, M. and Wurgler, J., 2006) People will be happy when their national team win a match especially some important matches, vice versa. The other one is the impact of investor sentiment on stock returns that is a significant topic in behavioral finance theory and has been presented in many works. This paper does empirical analysis with football results and stock returns to investigate the relationship between them.

An alternative view is that people will be excited and satisfied when their approved national team wins a match. Good news makes them more optimistic and this mood may lead them to be also optimistic when doing investment decision. As a consequence, they are more willing to invest and buy stocks. By contrast, people will become disappointed and sore when national football team gets a loss. As a consequence, these people will be more pessimistic in daily life and also in investment thus more selling orders will exist. In a word, we predict

(7)

3

that a win of national team can bring positive impact on stock returns, a loss can bring negative impact on stock returns.

To validate this explanation, we choose five countries in Europe (country lists and stock returns in these countries are shown in data part). Football is quite popular in Europe where people attach high importance to their national football team. Because of their high attention to football, national football results can have a significant impact on their moods. Therefore people in Europe are ideal research subjects in this paper. We select important international football results in Europe as sentiment variable to study the correlation between football and stock return. Important international football matches mentioned here are qualifying games, group games and eliminated games for Word Cup and European Championship. National friendly games are excluded. We also collect local stock return for the first tread day after games as research samples. To study whether football results have impact on local stock market, we firstly construct a model through using world stock market, lagged stork return, weekday as well as non-weekend holiday to explain local stock return. The residuals in this model and can be defined as abnormal normalized returns which may have relations to football results. After regressing wins and losses in international football matches on abnormal normalized returns, we can examine our null hypothesis according to the regression results. Detailed approach and results are can be found in the rest of this paper.

The content of the paper is organized as follows. Section 2 presents previous studies in relationship between investor sentiment and stock returns. Section 3 describes the data, and especially distinguish different types of football competitions. In section 4 we show our approach to do empirical analysis. Section 5 analyses economically and statistically significant of results. In section 6 we do the statistical robustness checks by using larger sample with

(8)

4

longer period and controlling time-clustering effect. Finally we provide economic explanation and summarize our findings and conclusions.

(9)

5

2. Literature Review

A diverse range of literature has provided the evidence of whether the relationship between football results and stock returns exist. However the original topic is whether investor sentiment is incorrected with stock returns which has also been discussed in a large number of literatures. For example, Baker, M. and Wurgler, J. (2006) provided evidence that a wave of investors’ mood has great impact on security whose asset prices are highly subjective and difficult to arbitrage. This means investor sentiment, which is ignored in classic finance theory, has significant effects on the cross-section of stock prices. However, to investigate the relationship between the investor sentiment and stock market return, we should firstly research the definition of investor sentiment.

2.1 The definition of investor sentiment

In the present behavioral finance theory, we have no standard for the definition of investor sentiment. Through worldwide studies we can divide investor sentiment into broad sense and narrow sense.

The broad sense defines investor sentiment with the respect to psychology that investor sentiment is a kind of investment mood derived from investors’ cognitive bias, investors’ mood and against Bayesian Rationality. In this sense, investor sentiment is difficult to measure due to the involvement of psychology and mood value. Lee and Shleifer (1991) defined individual investor sentiment as mood based judgement because of cognitive bias when investigating the position that fluctuations in discounts of closed- end funds are driven by

(10)

6

changes in individual investor sentiment. However, Shleifer and Vishny (1997) considered investor sentiment as a procedure when traders making decision.

The narrow sense defines investor sentiment with the respect to expectation on investment return as the expectation from investors on risky capital prices as well as demand of speculation. This understanding involves empirical analysis on the relationship between investor sentiment and capital markets. Stein (1996) defined investor sentiment as systematic error of expectation by investors. Baker and stein (1996) presented that investor sentiment is the incorrect judgement when investors pricing capital values, particular is the error between the value of capital priced by investors and the ture value. Brown and Cliff (2004) considered investor sentiment as the investors’ degree of optimist or pessimist on stocks. However Baker and Wurgler (2004) defined investor sentiment as Speculative tendency that motivates investors’ investment demand thus affecting stock prices. Mehra and Sah (2004) presented that investor sentiment reflects investors’ subjective preferences to future volatility of stock prices.

This paper uses behavioral finance theory and the relationship between investor sentiment and stock markets to do empirical analysis on the impact of national football results on stock returns. Therefore the investor sentiment is broad sense. We think that investor sentiment consists two factors: the first is investors’ individual judgement on information, that is different investors because of different attitudes and options have different psychology reflection to information; the second factor is the interaction among investors, that is influenced by other investors’ preference or other investors’ previous investment results, investors do their individual investment decisions. The combination of these two factors is investor sentiment.

(11)

7 2.2 The measurement of investor sentiment

There are four types of measurement of investor sentiment, direct index, indirect index, composite index and sentiment proxy index.

Direct index is to reflect investor sentiment through directly doing questionnaire for investors and thus directly reflecting investors’ expectation of the future stock market according to survey results. Common direct index includes Investors Intelligence (II), American Association of individual investors (AAII). In particular, Investors Intelligence is achieved through the survey for the expectation of future stock market from over 130 newspapers’ stock reviewers. This index is measured by the difference between the percentage of expecting rising and declining to reflect investor sentiment. American Association of individual investors (AAII) is measured by doing questionnaire for its members and collecting their expectation of the stock market in 6 months (expecting rising, declining and no change) to reflect investors’ mood. In addition, Consumer confidence index (CCI), Sentiment of Wall Street sell-side strategists and JF Investor Confidence Index can also reflect investor sentiment. Kenneth L. Fisher and Meir Statman (2003) used Sentiment of Wall Street Sell-side Strategists, II and AAII to represent the sentiment of large, medium and small investors respectively and investigated the relationship between these three investor groups and stock return. They found the evidence that these three kinds of investors have negative relationship with future stock market return. This suggested that institutional investors which are considered as large investors are also affected by mood just like medium and small investors. They also found that the relation of Wall Street Sell-side Strategists is smaller than that of Investors

(12)

8

Intelligence (II) and American Association of individual investors (AAII), suggesting that sentiment has lower impact on institutional investors than medium and small investors.

Indirect index is structured by collecting mood related trading data on the stock market to reflect investor sentiment. The major of trading data is activeness index on stock market including Closed-End Fund Discount (CEFD), turnover rate, Advanced-Decline Line (ADL), IPO quantity and first day yield, trading volume, change in margin debt, fund flow, volatility index and so on.

Composite investor sentiment index uses principal component analysis to combine direct index and indirect index together to reflect investor sentiment. Baker and Wurgler (2006) used CEFD, turnover rate, monthly IPO quantity and first trading day yield, fund cash holding ratio and bonus premium index to construct a composite index. The empirical analysis proved that this composite index can better reflect investors’ mood than simple index. Chinese researcher Zhigao Yi and Ning Mao (2009) modified the way of construction on basis of Baker and Wurgler by using six Chinese adjusted index (CEFD, trading volume, IPO quantity and first day yield, Chinese consumer confidence index and new account number). The empirical analysis finally tested the significance of CISI.

The researches mentioned above analyse the relationship between investor sentiment and stock return through constructing investor sentiment index. However, some extrinsic factors that have effect on investors’ mood can also be researched to analyze the relation with stock return. These factors which can be considered as proxy of investor sentiment are usually non-economic factors such as weather, temperature, season and so on. Kamstra M. J., Kramer L. A., Levi M. D(2000, 2013) related daylight saving time with stock returns, and found that when daylight time declines with season, stock returns drop significantly. Melanie Cao and Jason

(13)

9

Wei(2005) provided evidence for the negative relationship between temperature and stock returns. Christos Floros(2008) also proved this relationship in European stock market, and used Generalized Autoregressive Conditional Heteroscedasticity (GARCH) method to provide new evidence of this topic. Moreover, studies found that sunshine which is closely linked with mood has a significant positive relation with stock returns. (Hirshleifer, D. and Shumway, T., 2003).

Apart from extrinsic factors like weather mentioned above, sport matches, serious earthquakes and other big events can also affect stock returns. A large number of psychical researches provide the evidence that when people are in a good mood, they will have more optimistic expectation on the future, and have stronger risk tolerance. On the contract, when people feel dejected or down, they are more likely to have pessimistic expectation on the future and have weaker risk tolerance. Therefore, investor sentiment has the trend to affect investors’ expectation of stock market value, then affect their treading behaviors in the market, and finally affect the stock prices in the stock market. Guo Xin-Li (2010) selected Chenyu Price Index of China as samples to investigate whether the May 12 Wenchuan Earthquake had impact on China’s stock market. Results indicated that pessimistic sentiment caused by earthquake could affect stock returns in short duration, but the long-term impact on Chinese stock markets is limited.

Similarly, sport results can also affect investor sentiment and therefore have significant impact on stock returns. Wann (2001) pointed that fans feel delighted and proud when their team behave well and thus have some positive influence. In contrast, when their team experiences continuously losses, fans will feel depressed and sad, and then bring some negative influence. More importantly, the two reflections trend to rise or decrease fans’

(14)

self-10

esteem and their positive attitude to their lives, resulting in the impact on the evaluation and determination in daily life especially in investment process. Hirt (1992)found that students in American Indiana University have significantly better grades after their college’s basketball team winning than losing. Schwarz provided the evidence that Germany national football team’s two matches in the 1982 World Cup had significant impact on their fans’ assessment on their personal situation as well as their options on their national issues. A research from Schweitzer (1992) suggested that for a live American football match, winner’s supporters had significant lower assessment of the probability of 1990 Iraq War would break out and the potential numbers of casualties caused by the war than loser’s supporters. In addition, investor sentiment can also bring influence on economic behaviours. For example, the sales of lottery tickets in Ohio will have a significant increase after a win of the Ohio State University’s football team. (Arkes, 1988)

As a result, we can conclude that results of sports matches can have an impact on investors’ assessment on their personal capability as well as their optimistic or pessimistic attitude in their daily life. In other world, that is the impact on investor sentiment.

2.3 The relationship between sports results and stock market returns

Empirical evidence presented that the cross-section of future stock returns is conditional on sentiment (Baker, M. and Wurgler, J., 2006). In details, when sentiment is relatively high, optimistic investors tend to earn relatively low subsequent returns. Stocks here are more risky such as younger stocks, non-dividend-paying stocks, small stocks, unprofitable stocks and so on. These stocks have low attraction to arbitrageurs but definitely attract optimists and

(15)

11

speculators. Ashton et al. (2003) selected UK as researching field to investigate the impact of national football results on local stock returns. Through researching the 100 largest companies traded on the London stock exchange (FTSE 100 index), he found strong association between the performance of England football team and daily change in London stock market. Edmans et al. (2007) selected national football results as sentiment variable and collected 1162 football matches results from 39 different countries as relevant mood events. The finding indicated that stock returns can react significantly to national football results. Edmans et al. (2007) also did empirical analysis on cricket, rugby, ice hockey, and basketball to expand samples. He found that these sports have similar relationship with stock market as football: Loss of national football results causes significant drop in stock returns, however no significant reaction can be found when national football teams win a match. Based on the finding of Edmans et al. (2007), Chang et al. (2006) performed some broaden study. They chose professional American football as object of study. After testing the relationship between team results in American National Football League and returns of listed companies located in different teams’ cities in the Nasdaq stock market, they got the same conclusion that losses will bring negative impact. However different from Edmans, they also found that those companies with small size, highly fluctuated yields, low rate of profit and without dividend experience more significant loss effect. Hakan Berument et al. (2006) investigated the effect of three major Turkish football teams performance on Istanbul stock market. They presented that win has positive association with stock returns and this association becomes stronger when their fans are crazier. Kaplanski, G. and Levy, H. (2010) also provided evidence that FIFA World Cup affects U.S stock market, with a significant negative effect on loss and an insignificant effect on win. However, he also found another

(16)

12

important but unique conclusion that “unlike the local effect, the aggregate effect does not depend on the game results, as it is always negative.”

However, not all studies agree on the relationship between football results and stock returns. Klein, Zwergel, and Henning (2008) rebuilt the study of Ashton et al. (2003) and detected several mistakes in the empirical set-up. They found in the previous model even the mirror flaws can have a crucial impact on stock market. After the robustness checks and modifying the analysing model, the relationship between football results and stock returns is proved to be rejected. Then Ashton, Gerrard, and Hudson (2011) did study to respond to the critical work from Klein et al. (2008). Through extending the database from period 1984-2002 to 1984-2009, employing new range of tests and allowing outliners, the link between national football results and stock returns did indeed exist, however the significance of effect declined particularly the effect of wins. Gerlach (2011) examined stock returns in matching countries and found the existence of unusual returns in countries even though their national football team did not play. This result indicated that national football results did not cause impact on stock returns. Boyle, Glenn, and Brett Walter (2002) put study samples in New Zealand which has a single dominant sport. They presented that stock return in this country is independent on the performance of national sport results.

2.4 The relationship between investor sentiment and stock market returns

According to the present relative studies in this topic, the impact of investor sentiment on stock return is divided into two parts. One is systematical influence caused by investor

(17)

13

sentiment, which is also called overall effect. The other part is otherness influence on stocks with different features, which is called cross-section effect.

A large number of studies have proved the existence of overall effect and provided the evidence that investor sentiment can help forecast the stock market. Gallimore and Gray (2002) thought that investor sentiment act a significant role while doing investment decision. Black (1986) named investors who have no internal information but irrationally collect noise as useful information as “noise traders”, and investigated them in stock market. Empirical analysis provided that the existence of noise traders can improve the liquidity but decline the effectiveness of stock market. Lee et al. (1991) pointed that the sentiment of noise traders are negative relative to fund rate discount. Based on the theory of noise traders, De Long (1990) built an asset pricing model (named DSSW Model for short) to test whether the sentiment of noise traders affects stock market. The findings shown that in the limited arbitrage market, investor sentiment is a systematical factor on stock equilibrium price. Barberies et al. (1998) built a belief related investor sentiment model (BSV Model) to explain the shape of sentiment and the impact on stock prices. They considered that the optimistic and pessimistic attitude can drive stock price away from true value, however when investors realize their extremes in previous behaviors stock price will go to the reverse trend. Brown and Cliff (2004) did an overall analysis on the causal relationship between various investors’ mood and short-term stock return. They found that investor sentiment has little predictability on stock return but investor sentiment indeed is an important factor affecting stock return. Burghardt (2008) constructed a sentiment index of individual investors by using the data in Euwax. His findings proved the negative relationship between sentiment index and potential stock return.

(18)

14

Apart from the research of overall effect, the cross-section effect on stock return also attracted people’s attention. In fact, early empirical studies on investor sentiment were derived from the impact on small sized stocks. Lee et al. (1991) studied the question of closed-end fund discount (CEFD) on the basis of DSSW and stated that affected by individual mood, yield change of small firms has positive relations with closed-end fund discount. In other word, investor sentiment theory can explain closed-end fund discount. Neal and Wheatley (1998) investigated the relationship between stock return and three sentiment variables: closed-end fund discount, net redemptions of mutual funds and buying and selling interest rate. They provided the evidence that closed-end fund discount and net redemptions of mutual funds can explain small firms’ stock return and size premium. However, they found little evidence that can prove the relationship of buying and selling interest rate and stock return.

(19)

15

3. Methodology and Hypothesis

Our first null hypothesis is that national football results have no impact on stock returns. This null hypothesis indicates that national football results are uncorrelated with stock markets, investors are rational in the efficient market. The alternative hypothesis is that stock markets are affected by national football results: wins lead to an increase in stock returns and losses lead to a decline in stock returns. This is motivated by the previous literature indicating that wins are linked with good mood and losses are linked with bad mood. Our second null hypothesis is that win and loss in international football matches have similar impact on the daily return after a match. This null hypothesis indicates that it is the football match but not the result of match that can affect stock returns.

Edmans et al. (2007) estimate the following model:

𝑹𝒊𝒕= 𝜸𝟎𝒊+ 𝜸𝟏𝒊𝑹𝒊𝒕−𝟏+ 𝜸𝟐𝒊𝑹𝒎𝒕−𝟏+ 𝜸𝟑𝒊𝑹𝒎𝒕+ 𝜸𝟒𝒊𝑹𝒎𝒕+𝟏+ 𝜸𝟓𝒊𝑫𝒕+ 𝝐𝒊𝒕 (1)

Where 𝑹𝒊𝒕 is the stock return in local currency on local stock market index for country i on

day t, 𝑹𝒎𝒕 is the continuously compounded local stock return in dollar on Datastream’s

World Market Index for country i on day t. 𝑫𝒕 = {𝑫𝟏𝒕, 𝑫𝟐𝒕, 𝑫𝟑𝒕, 𝑫𝟒𝒕} are the dummy variables

for Monday to Thursday. For example, let 𝐷1𝑡 equal one if the first trading day after the match

is Monday and zero otherwise.

In regression (1), the lagged stock return 𝑹𝒊𝒕−𝟏 is included to control first-order serial

correlation. We know that international stock markets are integrated so that stock markets in different countries may correlate with each other. Therefore we include 𝑹𝒎𝒕 to control this

(20)

16

some may lead the world index. We can test the statistical and economical significance of the coefficient of one-day-before local market return (𝑹𝒊𝒕−𝟏), present world market return (𝑹𝒎𝒕),

one-day-before world market return (𝑹_{𝒎𝒕−𝟏}), one-day-later world market return (𝑹_{𝒎𝒕+𝟏}) and constant to discover the existence of other factors that are not included. The R square and adjusted R square can also help us test the fitness of this model.

^

𝝐𝒊𝒕 = 𝜷𝟎+ 𝜷𝑾𝑾𝒊𝒕+ 𝜷𝒍𝑳𝒊𝒕+ 𝒖𝒊𝒕 (2)

^

𝝐𝒊𝒕 is computed as the residuals in regression (1) and can be defined as abnormal normalized

returns. 𝑾𝒊𝒕 is dummy variable for wins in national games that equals to one if country i wins

a football match on a day with the first trading day t after the match and zero otherwise. And 𝑳𝒊𝒕 is dummy variable that equals to one if country i loses a football match on a day for the

first trading day t after the match and zero otherwise. In the model (1) we can get the estimate point of 𝑹𝒊𝒕−𝟏, 𝑹𝒎𝒕−𝟏, 𝑹𝒎𝒕, 𝑹𝒎𝒕+𝟏, 𝑫𝒕, 𝑸𝒕 as well as constant. We use the equation to

compute the ^

𝝐𝒊𝒕 for each date with international football match and then regress the

information of wins and losses in World Cup and European Cup from January 197 through

December 2016 on ^

𝝐𝒊𝒕. We can investigate the relationship between football match results

and stock market returns through testing whether 𝛽𝑊 and 𝛽𝐿 are both significantly different

from zero. Our third null hypothesis is that 𝛽𝑊= 0 at 0.05 level and the alternative

hypothesis is that 𝛽𝑊≠ 0. This null hypothesis indicates that wins in international football

matches have no significant effect on local stock market returns. Similarly, we have another null hypothesis that 𝛽𝐿 = 0 and the alternative hypothesis is 𝛽𝐿 ≠ 0. This null hypothesis

(21)

17

4. Data and descriptive statistics

4.1 Data

We collected national football results during the period between January 1987 and December 2016 from the website www.fifa.com and www.uefa.com. The results can be defined as the dummy of win and loss mentioned in section 3. The data includes matches from the World Cup and European Cup. Both World Cup and European Cup are held every 4 years. For World Cup, The current format involves a qualification phase, which currently takes place over the preceding three years, to determine which teams qualify for the tournament phase. European national teams that win the qualification, as well as national teams in other geographic regions, form the total 32 teams to participate the World Cup final. In the group stage, these teams are divided into eight groups of four teams each. Teams in each group play against each other and the top two teams from each group advance to the “elimination games”. In this stage, half of the remaining teams will be eliminated in each round and the last one who survive in all elimination games win the World Cup. The European has some differences in team numbers but the general format is similar. We select Netherlands, Germany, France, England and Italy as sample countries. For 𝑫𝒕 = {𝑫𝟏𝒕, 𝑫𝟐𝒕, 𝑫𝟑𝒕, 𝑫𝟒𝒕} , we can collect the

weekday data of match date, and then compute the weekday of first trading day after match.

The data of market indices requested in this paper are collected from Datastream. Stock returns in 5 countries are computed in the local currency. We calculate index returns using total return index. We can download daily total return in five countries’ national stock market from January 1987 through December 2016 from Datastream. The national stock market for

(22)

18

England, France, Germany, Italy and Netherlands are FTSE 100, FRANCE CAC 40, DAX 30, FTSE ITALIA MIB STORICO and AEX INDEX respectively. Then we can calculate the daily change of total return index and use the log of daily change as the data of 𝑹_𝒊𝒕. We can also download

the data of daily return in world market from Datastream which is 𝑹_𝒎𝒕 in equation (2). Following the definition, we can get the lagged stock return (𝑹_{𝒊𝒕−𝟏}), one-day-before world market return (𝑹_{𝒎𝒕−𝟏}), one-day-later world market return (𝑹_{𝒎𝒕+𝟏}). After the regression of model (1), we can get the estimate point of 𝑹_{𝒊𝒕−𝟏}, 𝑹_{𝒎𝒕−𝟏}, 𝑹_𝒎𝒕, 𝑹_{𝒎𝒕+𝟏}, 𝑫_𝒕, 𝑸_𝒕 as well as

constant. We use the equation to compute ^

𝝐𝒊𝒕for each date with international football match

and remove the date without an international football match.

4.2 Descriptive Statistics

TableⅠ provides information of total number of days with and without international football games included within our studying sample as well as the mean daily log stock market return on the first trading day after match. For the sample of 5 different European countries, Netherlands, France, Germany, England and Italy, 37421 days are not associated with international football matches. The mean of log daily return for non-game days is 0.683 basis point and the standard deviation for these days is 55.45778 basis point. There are total 548 wins and 138 losses in our sample from January 197 through December 2016. The mean of log daily return for first trading day after all wins is 8.425 basis point, significantly higher than that in non-game days. The average return following a loss in international football match is -16.37 basis point, negative and significantly higher than that after a win. The standard deviation of daily return following an international football match is slightly higher than that

(23)

19

in non-game days (71.55 and 62.93 respectively after a win or loss). Within all sample with matches, we collect the information of average daily return by different game types and different cups. Cups are divided in World Cup and European Cup and game types are divided in elimination games, group games and qualifying games. From the Table Ⅰ, we can find that in general, World Cup and elimination games have larger loss effect than other football matches. The mean of log daily return (-32.5 basis point) following losses of European Cup elimination games shows the highest negative effect of loss outcome. The average return following a loss of World Cup elimination games is -18.14 basis point, which is more significant than most other match types. The mean log daily return is declining from elimination games, group games, qualifying games for World Cup. However we can find that average return following wins of European Cup elimination games as well as European Cup group games (-13.6, -2.13 respectively) pronounces negative win effects. The similar situation occurs in the first trading day after European Cup qualifying games. The mean of log daily return following European Cup qualifying games is 7.65 basis point which reflects a positive loss effect.

We have total 6 independent subsets of international football matches. Therefore we can assume that all the samples within these 6 groups are independent with each other. Although wins as well as losses have opposite impact on average e return for first trading day, we can conclude that the difference between average return following wins and losses is always positive, with the maximum of 33.5 basis point for World Cup group games. This means that the average return following a win is always higher than that following a loss. In other words, loss in international football games has more negative effect on daily return than win. In a word, Table Ⅰ suggests that the results of international football games indeed have

(24)

20

relationship with stock market return, and thus the null hypothesis of win and loss have similar impact on the daily return after match can be rejected.

However, this descriptive statistics cannot illustrate the accurate relation between match results and stock market without interruption. An important feature of international football matches we study is that they take place majorly in a few months, weeks, and weekdays. For example, the group games and elimination games of World Cup and European Cup are mostly in in the month of June and July, and qualifying games mostly take place on Wednesday and weekends. That means there are more than one match happening in the same day, which can be considered as a factor of interruption. We have total 74 wins and 45 losses for elimination games, however in the full sample there are only 55 different days with winning and only 32 different days in which at least one country lost. Moreover, most international football games especially qualifying games take place between Friday to Sunday, however we collect information of Monday for all these games because next Monday is the first trading day after them. This may cause spurious relation because the impact of weekend on stock market may be introduced in our study as an interruption. In the next section, we will introduce detailed econometric approach which can remove the effect of these factors of interruption and show the statistic relation between match results and stock market.

Table Ⅰ

Number of wins and losses in international football matches for different match types and percent mean daily return for the first trading day after matches

The table reports the number of five European countries’ wins and losses in international football matches and the average return following these matches. The football matches are played from January 1987 through December 2016 in the World Cup and European Cup, including elimination games, group games and qualifying games. In this table the average returns are calculated from log daily return for respectively countries on the first trading day following international football matches. Elimination games are matches where losers will be dismissed from further play. Group games are played to qualify teams for elimination games. Qualifying games are matches in which teams compete to get the qualification for championship.

(25)

21

No Games Wins Losses

N Mean SD N Mean SD N Mean SD

No games 37421 0.00683 0.5545778

All games 548 0.08425 0.7154901 138 -0.163747 0.6293613

World Cup elimination games 48 0.0433149 0.5127532 24 -0.181383 0.6345172

World Cup group games 48 0.015379 0.503618 14 -0.319184 0.6239506

World Cup qualifying games 188 0.015546 0.8773262 21 -0.147943 0.54203 European Cup elimination games 26 -0.136482 0.7931326 21 -0.325229 0.6174922

European Cup group games 48 -0.021333 0.4377706 25 -0.085579 0.3640187

(26)

22

5. Results

We use daily return for five different countries’ stock market indices on the first trading day after international games to measure the effect of international football results on stock return. For example, we collect the data of stock return on next Monday if the football game takes place on Friday, Saturday or Sunday because Monday is the first following trading day. Although some national stock markets are still open after the announcement of results for those weekday games, we choose to use data on first trading day because we would like to ensure that each game outcome has a full day impact on stock market.

5.1 Econometric Statistics of five countries together

Table Ⅱ reports the regression result of first estimation using national stock market return during the period of January 1987 to December 2016 in five countries. In this table, rt1 which is Rt-1 in regression (1), is the variable of national stock market return one trading day before. Rmt_1 and rmt_2 are the daily return in world market index for one day before and one day later respectively. 𝑫𝒕 is the dummy variable for Monday to Thursday. We can find that in this

table, the coefficient of present world market return (0.0016) and one-day-later world market return (0.0005) are both statistically significant. The constant in this model is also statistically significant, suggesting the existence of other factors that are not included. We will than do the second regression to study the relation between errors and match results, thus studying whether wins and losses are significant factors in the first model.

(27)

23 Table Ⅱ

The regression analysis is based on the national stock market return on the day with international football matches for five different European countries (Netherlands, Germany, France, Italy and England) from January 1987 through December 2016. The full sample collects a total of 874 trading days with international football matches observations. The table reports the regression results from

𝑹𝒊𝒕= 𝜸𝟎𝒊+ 𝜸𝟏𝒊𝑹𝒊𝒕−𝟏+ 𝜸𝟐𝒊𝑹𝒎𝒕−𝟏+ 𝜸𝟑𝒊𝑹𝒎𝒕+ 𝜸𝟒𝒊𝑹𝒎𝒕+𝟏+ 𝜸𝟓𝒊𝑫𝒕+ 𝝐𝒊𝒕 VARIABLES rt rt1 0.0112 (0.0346) rmt 0.00160*** (0.000183) rmt_1 -0.000231 (0.000186) rmt_2 0.000515*** (0.000196) d1t 0.0966* (0.0544) d2t 0.0879 (0.105) d3t 0.0610 (0.0799) d4t 0.0356 (0.0562) Constant -0.397*** (0.0816) Observations 874 R-squared 0.097

Standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1

Table Ⅲ reports detailed regression results of the second regression model which states the main finding of this paper. We regress the information of wins and losses in World Cup and European Cup from January 1987 through December 2016 on E which is defined in first regression model to study the impact of match results on local stock market return. We can find that in general national stock markets experience a negative effect on the first trading

(28)

24

day after a loss in football match, and a positive return on the day after a win in international football match.

Focusing on the effect of wins on stock market return, the ordinary least square (OLS) estimate point of the win is 9.19 basis point within the full sample of 548 games. Except European Cup group games (-3.3 basis point), all other subsets of games have positive impact on stock market return. The coefficient of win dummy is a positive 14.7 basis point for Elimination games and positive 16.9 basis point for World Cup elimination games, much larger than the estimate for group games and qualifying games. We can hereby conclude that Elimination games have larger impact on stock market return than group games as well as qualifying games, and World Cup has larger impact than European Cup. However except the estimate of World Cup elimination games, other estimates cannot be statistically distinguished from 0. Therefore, we cannot reject theH0: 𝛽𝑊 = 0 at 0.05 level.

Except European Cup qualifying games (positive 2.15 basis point), losses in international football matches have consistently negative effect on national stock market return. Within the full sample of 138 losses, the coefficient of loss is -15.15 basis point, highly significant both in statistical and economic terms. Therefore we can easily reject the null hypothesis of 𝛽𝐿 =

0 at 0.05 level. Similar with the findings in wins, the point estimate of losses is increasing with the importance of international matches. First, elimination games have larger loss effect on stock returns than group games and qualifying games. The coefficient of elimination games is -24.78 basis point, larger than that of group games (-17.84 basis point) and qualifying games (2.11 basis point). We can also get that the statistical significance experiences a decline through elimination games, group games and qualifying games. It seems that the final match in the elimination stage has the greatest mood effect, because this match is the most

(29)

25

important and concentrates the greatest media coverage. If a team wins the final game, all people in this nation will share the greatest honour, otherwise they will experience the strongest disappointing. Moreover, a loss in World Cup brings more negative effect than a loss in European Cup in all subsets. Both the point estimate (-22.4 basis point) and the statistical significance of World Cup elimination games are greater than that of European Cup elimination games (-21.5 basis point). The difference becomes larger in group games and qualifying games. We can see that the coefficient of European Cup qualifying games is positive (2.15 basis point). In fact there may be some irrelevant losses in qualifying games because a team already qualified or dismissed due to earlier performance may have little win or loss effect.

Table Ⅲ

Impact of Wins and Losses on Abnormal Daily Stock Market Return after International Football Matches

The table reports regression results based on the international football match results for five different European countries (Netherlands, Germany, France, Italy and England) from January 197 through December 2016. The full sample collects a total of 37421 daily return observations. This table reports the ordinary least square (OLS) coefficient of the win dummy (𝛽𝑊) and loss dummy (𝛽𝐿) from

^

𝜖𝑖𝑡= 𝛽0+ 𝛽𝑊𝑊𝑖𝑡+ 𝛽𝑙𝐿𝑖𝑡+ 𝑢𝑖𝑡

Where 𝑾_𝒊𝒕 is dummy variable for wins in national games that equals to one if country i wins a football match on a day with the first trading day after the match t and zero otherwise. And 𝑳_𝒊𝒕 is dummy variable that equals to one if country i loses a football match on a day with the first trading day after the match t and zero otherwise. The ^

𝜖𝑖𝑡 is defined by the regression model

(30)

26 Wins Losses Number of games 𝛽𝑊 t-Values Number of games t-Values All games 548 0.0918961 1.75 138 -0.151512 -2.17 Elimination games 74 0.147027 1.87 45 -0.247803 -2.18

World Cup elimination games 48 0.1689 2.08 24 -0.224 -1.96

European Cup elimination games 26 0.02973 0.34 21 -0.214889 -1.08

Group games 96 0.0220398 0.28 39 -0.178395 -1.83

World Cup group games 48 0.0734033 0.6 14 -0.306977 -1.81

European Cup group games 48 -0.03306 -0.34 25 -0.133963 -1.18

Qualifying games 378 0.1411464 2.07 54 -0.021083 -0.19

World Cup qualifying games 188 0.1761511 1.73 21 -0.090105 -0.5 European Cup qualifying games 190 0.1059872 1.17 33 0.0214681 0.16

5.2 Econometric Statistics of five countries respectively

After testing the relationship between abnormal returns and results of international football matches over five countries together, we regress this model for each country alone to compare the results in different countries. Table Ⅳ reports the detailed regression results of ^

𝝐_𝒊𝒕 = 𝜷𝟎+ 𝜷𝑾𝑾𝒊𝒕+ 𝜷𝒍𝑳𝒊𝒕+ 𝒖𝒊𝒕 for England, France, Germany, Italy and Netherlands respectively. We can find that in England, Italy and Netherlands, the results are similar to what we get in Table Ⅱ: National stock markets earn a negative return after a loss in football match, and a positive return on the day after a win in international football match.

For the effect of wins on stock market return, the ordinary least square (OLS) coefficient on the win dummy for England, Italy and Netherlands is 11.9, 17.0 and 4.2 basis point respectively. The ordinary least square (OLS) coefficient on the loss dummy for England, Italy

(31)

27

and Netherlands is -16.7, -25.6 and -13.1 basis point respectively. In other words, the stock markets in these three countries all experience a positive impact after a win and a negative impact after a loss.

However, the results of Germany and France seem to be inconsistent with what we get in Table Ⅲ. For Germany, the coefficients of win dummy and loss dummy are both positive (27.3 basis point and 39.5 basis point respectively). In contrast, the ordinary least square (OLS) coefficients on the win dummy and loss dummy for France are both negative (-14.9 basis point and -26.3 basis point respectively). Although the results are not perfectly consistent with the previous findings, we can find that the difference between the point estimate of wins and losses for both Germany and France are positive. This means that in Germany and France a loss have more negative effect on national stock market returns than a win.

Table Ⅳ

Abnormal Daily Stock Market Return after International Football Matches for five different countries

This table reports the ordinary least square (OLS) coefficient of the win dummy (𝛽𝑊) and loss dummy (𝛽𝐿)

from

^

Where ^

𝜖𝑖𝑡 are the “abnormal normalized returns” defined in Section 3 and described in Table Ⅲ. 𝑊𝑖𝑡 and

𝐿𝑖𝑡 are both dummy variable described in section 3 and Table Ⅲ. The five European countries are England,

France, Germany, Italy and Netherlands. We collect the number of games, the value of 𝑊𝑖𝑡 and 𝐿𝑖𝑡 and t-

statistics for five countries respectively as well as all countries. The sample is the data of all game types including qualifying games, group games and elimination games for different countries.

(32)

28

Wins Losses

Number of games 𝛽𝑊

t-Values Number of games 𝛽𝐿

t-Values All countries 548 0.0918961 1.75 138 -0.151512 -2.17 England 91 0.1187005 1.08 29 -0.167239 -1.2 France 82 -0.148586 -1.41 27 -0.263424 -1.93 Germany 130 0.2726792 1.96 22 0.0394548 0.2 Italy 121 0.169888 1.7 23 -0.25593 -1.75 Netherlands 124 0.0421912 0.31 36 -0.131432 -0.8

(33)

29

6. Statistical Robustness Checks

6.1 Check time-clustering problem

This section attempts to avoid statistical robustness in the win and loss effect through removing the interruption of outliers and through controlling the effect of clustering of matches in different countries in certain dates. Although equation (1) in section 3 includes𝑹_{𝒊𝒕−𝟏},𝑹_𝒎𝒕, 𝑹_{𝒎𝒕−𝟏} and 𝑹_{𝒎𝒕+𝟏} to control the effect of market movements and lag of market reaction among different countries’ stock indices, we may be overvaluing the statistical significance in the estimating because our model may fail to dismiss the correlations among different countries on a given date. To solve the problems created by time-clustering, we form the portfolio returns for winning and losing. For each date for country i with

international football matches, we average ^

𝝐𝒊𝒕for all countries winning the football match,

and average ^

𝝐𝒊𝒕for all countries losing the match. Then we will get two time series of portfolio

returns 𝛽𝑊 and 𝛽𝐿 for winning and losing respectively. Under the null hypothesis of

both 𝛽_𝑊 𝑎𝑛𝑑 𝛽_𝐿 eqauls 0, the average of these series both should be zero

Table Ⅴ reports the number of wins and losses, the mean return within two time series of portfolio for three games stages respectively, and standard t-values for the mean returns. According to the results in Table Ⅴ, we can document a negligible positive stock market effect after a win as well as a significant negative stock market effect after a loss which is consistent with the findings in section 5. For the all game sample, group game sample and qualifying game sample, the estimate point of 𝛽𝑊 is positive (8.5, 1.6 and 6.8 basis point

(34)

30

respectively). However the t-value is so low that 𝛽𝑊 cannot statistically distinguish from zero.

The estimate point of win for elimination games is even negative (1.6 basis point with the t-statistic is -0.08). So we cannot reject the null hypothesis that 𝛽_𝑊= 0 at 0.05 level. However, the loss effect remains statistically significant under levels of 5% for all matches within both elimination and group games subsets. The total 125 loss observation, which reduced from full sample of 138 losses, remain highly significant. The point estimate of loss is −16.4 basis points and the t-value is −3.502. Therefore we can easily reject the null hypothesis of 𝛽_𝐿 = 0 at 0.05 level.

Table Ⅴ

Abnormal Daily Stock Returns after International Football Matches Using Portfolio Returns for Winning and Losing

Just as the definition in Section 3 and described in Table Ⅲ, ^

𝜖𝑖𝑡 are the abnormal normalized returns. For

each date for country i with international football matches, we average ^

𝜖𝑖𝑡for all countries with 𝑊𝑖𝑡= 1,

and average ^

𝜖𝑖𝑡for all countries with 𝐿𝑖𝑡= 1. Then we will get two time series of portfolio returns 𝛽𝑊 and

𝛽𝐿 for winning and losing countries respectively. In this table, “N” reports the number of dates under each

game stage. The t-values reports the estimate of the standard error of the average by using SD(𝛽𝑗)/√𝑁 − 1.

6.2 Check by using larger sample

Table Ⅵ repeated the regression of Table Ⅲ by using the information of wins and losses in World Cup and European Cup from January 1974 through December 2016. This means we

Wins Losses N 𝛽𝑊 t-Values N 𝛽𝐿 t-Values All games 317 0.085 0.03 125 -0.164 -3.502 Elimination games 55 -0.016 -0.085 40 -0.314 -2.343 Group games 76 0.068 0.057 36 -0.152 -2.631 Qualifying games 188 0.11 0.039 49 -0.052 -1.114

(35)

31

broaden our samples and period to test our null hypothesis again. We can find the similar results to that in section 5 that national stock markets earn a negative return on the first trading following a loss in football match, and a positive return on the day after a win in international football match.

For the effect of wins on stock market return, the ordinary least square (OLS) estimate point of the win dummy is 5.2 basis point within the full sample of 710 games. Except qualifying games (-3.8 basis point), all other subsets of games have positive impact on stock market return. The win estimate point is a positive 4.7 basis point for Elimination games and positive 9.2 basis point for World Cup elimination games. Although the estimate of win in elimination group is economically significant, we still cannot reject the null hypothesis of 𝛽𝑊 = 0 because

these estimates are not statistically distinguished from 0.

Losses in international football matches with all match groups have consistently negative effect on national stock market return. For the all 184 football losses, the point estimate is -20.5 basis point, and the t-value is -2.17. This means that the loss effect is highly significant both in economic and statistical terms. Therefore we can easily reject the null hypothesis of 𝛽_𝐿 = 0 at 0.05 level. Similar with the findings in section 5, the point estimate of losses is increasing with the importance of international matches. The coefficient of elimination games is -36.4 basis point, larger than that of group games (-17.8 basis point) and qualifying games (15.7 basis point). We can also get that the statistical significance experiences a decline through elimination games, group games and qualifying games. Moreover, World Cup has larger loss effect than European Cup at all match stages. Both the point estimate (-47.3 basis point) and the statistical significance (-2.66) of World Cup elimination games are greater than that of European Cup elimination games (-30.9 basis point and -1.98 respectively).

(36)

32

Table Ⅵ

Abnormal Daily Stock Market Return after International Football Matches

The regression analysis is based on the international football match results for five different European countries (Netherlands, Germany, France, Italy and England) from January 1974 through December 2016. The full sample collects a total of 37421 daily return observations. This table reports the ordinary least square (OLS) coefficient of the win dummy (𝛽𝑊) and loss dummy (𝛽𝐿) from

^

Where 𝑊𝑖𝑡 is dummy variable for wins in national games that equals to one if country i wins a football

match on a day that makes t the first trading day after the match and zero otherwise. And 𝐿𝑖𝑡 is dummy

variable that equals to one if country i loses a football match on a day that makes t the first trading day after the match and zero otherwise. The ^

𝜖𝑖𝑡 is defined by the regression model

𝑅𝑖𝑡= 𝛾0𝑖+ 𝛾1𝑖𝑅𝑖𝑡−1+ 𝛾2𝑖𝑅𝑚𝑡−1+ 𝛾3𝑖𝑅𝑚𝑡+ 𝛾4𝑖𝑅𝑚𝑡+1+ 𝛾5𝑖𝐷𝑡+ 𝛾6𝑖𝑄𝑡+ 𝜖𝑖𝑡 Wins Losses Number of games 𝛽𝑊 t-Values Number of games t-Values All games 710 0.052 0.75 184 -0.205 -2.17 Elimination games 111 0.047 1.43 60 -0.364 -3.24

World Cup elimination games 63 0.092 1.53 32 -0.473 -2.66

European Cup elimination games 48 0.039 0.12 28 -0.309 -1.98

Group games 184 0.042 0.43 52 -0.178 -1.43

World Cup group games 112 0.007 0.52 19 -0.480 -2.21

European Cup group games 72 0.084 0.05 33 -0.032 -1.28

Qualifying games 415 -0.038 -0.52 72 -0.157 -1.02

World Cup qualifying games 218 -0.075 -0.73 28 -0.192 -0.52 European Cup qualifying games 197 0.029 0.18 44 -0.140 -0.66

(37)

33

7. Concluding Remarks

Motivated by the conflict in whether sports results can have a significant effect on national stock market returns, this paper collect the national stock returns from January 197 through December 2016 and the results of international football matches (including elimination games, group games and qualifying games in World Cup and European Cup) over five European countries (England, France, Germany, Italy and Netherlands) to investigate the effect of international football results on stock markets. We use the regression results to compute abnormal returns and regress win and loss on it. According to the results, we can document a negligible positive stock market effect after a win in international football match. In details, Elimination games have larger impact on stock market return than group games as well as qualifying games, and World Cup has larger impact than European Cup. However, these estimates are not statistically distinguished from zero. Therefore, we cannot reject the null hypothesis of 𝛽𝑊 = 0 at 0.05 level. However we find a strong negative stock market

reaction to losses in international football matches. The size of this loss effect is economically and statistically significant at 0.05 level. In details, the statistical significance experiences a decline through elimination games, group games and qualifying games and World Cup has larger loss effect than European Cup at all match stages.

This findings are constant with the conclusion of Edmans et al (2007) that there will be a significant market decline after football losses. This loss effect is stronger in small stocks and in more important games, and is robust to methodological changes. The difference is that Edmans et al also investigate the relationship between sports results and stock returns for cricket, rugby, and basketball games. This paper only chooses football as objective but using

(38)

34

latest period (from January 1987 through December 2016) and includes all the qualifying games in World Cup and European Cup into the research sample (Edmans only collected data of closed qualifying games).

(39)

35

References

1. Arkes, Hal, Lisa Herren, and Alice Isen, 1988, The role of potential loss in the influence of effect on risk-taking behavior, Organizational Behavior and Human Decision Processes 42, 181–193

2. Ashton, J.K., Gerrard, B., and Hudson, R., 2003 Economic impact of national sporting success: evidence from the London stock exchange, Applied Economics Letters, 10:12, 783-785, DOI: 10.1080/1350485032000126712

3. Ashton, J.K., Gerrard, B., and Hudson, R., 2011. Do national football results really impact on the stock market_ Applied Economics, 43(26), pp.3709-3717.

4. Barberis, Nicholas Shleifer, Andrej ; Vishny, Robert W,1998, A model of investor sentiment, Journal of financial economics, Vol.49(3), pp.307-343 [Peer Reviewed Journal]

5. Baker, M. and Wurgler, J., 2006. Investor Sentiment and the Cross-section of Stock Returns. Journal of Finance, 61, pp.1645–1680.

6. Baker, Malcolm ; Wurgler, Jeffrey, 2006,Investor Sentiment and the Cross-Section of Stock Returns, Journal of Finance, August 2006, Vol.61(4), pp.1645-1680 [Peer Reviewed Journal].

7. Black F, 1986, Noise, [J]. Journal of Finance. 1986, 41, 529-543.

8. Boyle, Glenn, and Brett Walter, 2002, Reflected glory and failure_ International sporting success and the stock market, Applied Financial Economics 13, 225–235.

9. Bizman, Aharon, 2002, Engaging in distancing tactics among sport fans : effects on self-esteem and emotional responses, Journal of Social Psychology 142,3 (2002) 381-392 10. Brown, Gregory W ; Hartzell, Jay C, 2001, Market reaction to public information: The

atypical case of the Boston Celtics, Journal of Financial Economics, 2001, Vol.60(2), pp.333-370 [Peer Reviewed Journal].

11. Brown, Gregory W. ; Cliff, Michael T, 2004, Investor sentiment and the near-term stock market, Journal of Empirical Finance, 2004, Vol.11(1), pp.1-27 [Peer Reviewed Journal] 12. Brown, Gregory W., and Michael T. Cliff, 2004, Investor sentiment and the near-term stock

(40)

36

13. Brown, Stephen J., William N. Goetzmann, Takato Hiraki, Noriyoshi Shiraishi, and Masahiro Watanabe, 2003, Investor sentiment in Japanese and U.S. daily mutual fund flows, Working paper, Yale University.

14. Burghart M., Czink M. Riordan R. 2008, Retail investor sentiment and the stock market [J]. SSRN.

15. Christos Floros, 2008, Stock market returns and the temperature effect: new evidence from Europe, Applied Financial Economics Letters, 2008, 4, 461–467.

16. De Long, J. Bradford ; Shleifer, Andrei ; Summers, Lawrence H. ; Waldmann, Robert J, 1990, Noise Trader Risk in Financial Markets, Journal of Political Economy, 1 August 1990, Vol.98(4), pp.703-738 [Peer Reviewed Journal]ISSN: 00223808 ; E-ISSN: 1537534X.

17. Edmans, A., Garcia, D., and Norli, O., 2007. Sports sentiment and stock returns. Journal of Finance, 62(4), pp.1967-19998.

18. Gallimore P., Gray A. 2002, The Role of Invest or Sentiment in Property investment decision, Journal pf Property Research, 19(2):111-120.

19. Gerlach, J.R., 2011. International sports and investor sentiment_ do national team matches really affect stock market returns_ Applied Financial Economics, 21(12), pp.863-880.

20. GUO Xin-li, 2010, The Impact of the May 12 Wenchuan Earthquake on China's Stock Market---As An Example to the Chenyu Price Index of China, 2010 International Conference on Management Science & Engineering (17th )November 24-26, 2010.

21. Hakan Berument a, Nildag Basak Ceylan , Esin Gozpinar, 2006, Performance of football on the stock market: Evidence from Turkey, The Social Science Journal 43 (2006) 695–699. 22. Hirshleifer, D. and Shumway, T., 2003. Good day sunshine_ stock returns and the weather.

Journal of Finance, 58(3), 1009–1032.

23. Hirt, Edward R., Grant A. Erickson, Chris Kennedy, and Dolf Zillmann, 1992, Costs and benefits of allegiance: Changes in fans’ self-ascribed competencies after team victory versus defeat, Journal of Personality and Social Psychology 63, 724–738.

24. Huber, Peter J., 1996, Robust statistical procedures, second edition (Society for Industrial and Applied Mathematics, New York).

25. Kahneman, Daniel, and Amos Tversky, 1979, Prospect theory: An analysis of decision under risk, Econometrica 47, 263–292.

26. Kamstra, Mark J., Lisa A. Kramer, and Maurice D. Levi, 2000, Losing sleep at the market: The daylight saving anomaly, American Economic Review 12, 1000–1005.

27. Kamstra, Mark J., Lisa A. Kramer, and Maurice D. Levi, 2003, Winter blues: A SAD stock market cycle, American Economic Review 93, 324–343.

(41)

37

28. Kaplanski, G. and Levy, H., 2010. Exploitable Predictable Irrationality_ The FIFA World Cup Effect on the U.S. Stock Market. Journal of Financial and Quantitative Analysis, 45(2), pp.535-553.

29. Kenneth L. Fisher and Meir Statman, 2003, Consumer confidence and stock returns: what sentiment tells us.(The Market), Journal of Portfolio Management, Fall, 2003, Vol.30(1), p.115(13) [Peer Reviewed Journal]

30. Kenneth L. Fisher and Meir Statman, 2000, Investor Sentiment and Stock Returns, Financial Analysts Journal, March-April, 2000, Vol.56(2), p.16(8) [Peer Reviewed Journal].

31. Klein, C., Zwergel, B. and Henning, F. J. 2008 Reconsidering the impact of national football results on the FTSE 100, Applied Economics, 41, 3287–94.

32. Lee, Charles M. C., Andrei Shleifer, and Richard H. Thaler. 1991. “Investor Sentiment and the Closed-End Fund Puzzle.” The Journal of Finance 46 (1) (March): 75. doi:10.2307/2328690. [Peer Reviewed Journal]

33. Melanie Cao, Jason Wei, 2005, Stock market returns: A note on temperature anomaly, Journal of Banking & Finance 29 (2005) 1559–1573.

34. Neal, Robert ; Wheatley, Simon M, 1998, Do Measures of Investor Sentiment Predict Returns?, Journal of Financial and Quantitative Analysis, 1998, Vol.33(4), pp.523-547 [Peer Reviewed Journal].

35. Palomino, Frederic ; Renneboog, Luc ; Zhang, Chendi, Information salience, investor sentiment, and stock returns: The case of British football betting, Journal of Corporate Finance, 2009, Vol.15(3), pp.368-387

36. Schwarz, Norbert, Fritz Strack, Detlev Kommer, and Dirk Wagner, 1987, Football, rooms, and the quality of your life: Mood effects on judgements of satisfaction with life in general and with specific domains, European Journal of Social Psychology 17, 69–79.

37. Schweitzer, Karla, Dolf Zillmann, James Weaver, and Elizabeth Luttrell, 1992, Perception of threatening events in the emotional aftermath of a televised college football game, Journal of Broadcasting and Electronic Media 36, 75–82.

38. Swaminathan, Bhaskaran, 1996, Time-varying expected small firm returns and closed-end fund discounts, Review of Financial Studies 9, 845–887.

39. Wann, Daniel L. ; Dolan, Thomas J. ; Megeorge, Kimberly K. ; Allison, Julie A, 1994, Relationships between Spectator Identification and Spectators' Perceptions of Influence, Spectators' Emotions, and Competition Outcome, Journal of Sport and Exercise Psychology, 12/1994, Vol.16(4), pp.347-364 [Peer Reviewed Journal]

40. Wann, Daniel, and Nyla Branscombe, 1995, Influence of identification with a sports team on objective knowledge and subjective beliefs, International Journal of Sport Psychology 26, 551–567.

41. Wann, Daniel, Merrill Melnick, Gordon Russell, and Dale Pease, 2001, Sport Fans: The Psychology and Social Impact of Spectators (Routledge, London).

(42)

38

42. Wachter, Jessica, 2000, Habit formation and the cross-section of asset returns, Chapter 4 of Ph.D. dissertation, Harvard University.

43. Wurgler, Jeffrey, and Katia Zhuravskaya, 2002, Does arbitrage flatten demand curves for stocks?, Journal of Business 75, 583–608.

44. Zweig, Martin E., 1973, An investor expectations stock price predictive model using closed-end fund premiums, Journal of Finance 28, 67–87.

Sport sentiment and stock return : do national football results have impact on the stock market?

Master Thesis