The impact of a sponsorship announcement on the sponsor’s stock price

(1)

The impact of a sponsorship announcement on

the sponsor’s stock price

Examination of Olympic, World Cup Football and European Cup Football

1. Introduction

What do a bank, a brewing company, a brand of sportswear and an automobile company have in common? They are some of the frequently featured sponsors of sports organizations and events. They invest large amounts of money, products or services to obtain the right to be a sponsor. Sponsorship is a growing form of communication which can be defined as ‘the provision of resources (i.e. money, people, or equipment) to an event or activity in exchange for a direct association to the event or activity’ (Sandler and Shani, 1989; Olson, 2010). The annual worldwide spending on sponsorships has been increasing rapidl y since the 1990s. Nowadays, spending on sponsorship support by corporations exceeds $5.1 billion annually in the USA within the sports industry alone (Olson, 2010), constituting about two thirds of total sponsorship dollars in the USA (Verity, 2002; Smiths, 2004; Crompton, 2004). The worldwide spending on

sponsorships was approximately $34 billion in2006; which grew further to $37 billion in 2007 (IEG 2005, 2007). This growth in sponsorship spending can be explained by the increase in the number of sponsoring opportunities, as well as the perception of an increase in the effectiveness of sponsorship as a marketing tool, and the increased media coverage of sports events (Meenaghan, 1991; Meenaghan, 1999).

Sponsorship poses a very important marketing tool for companies (Meenaghan, 1999): since the mid 1990s it represents one of the most rapidly growing fields of marketing communication (Javalgi, Traylor, Gross, et al. 1994). Due to the ever pressing need for financing, the dependence of sport organizations and events on corporate sponsorship has grown as well. The trend has led to an increasing interest in sponsorship research. To enable a company to make a fair decision as to whether to sponsor an event, the benefits of sponsorship should be thoroughly evaluated. During the 1980s, only half of sponsoring corporations measured sponsorship outcomes in terms of financial value (Gardner and Shuman, 1987). However, since the turn of the millennium, the awareness of the benefits and pitfalls of sponsorship has been growing (Amis and Slack, 1999). The financial added value is of ultimate importance for the sponsoring corporations and their shareholders.

(3)

determinants of the degree of change in shareholders’ value. I specifically look at the causes of differences in abnormal returns, since the impact of the sponsorship announcements has been shown to differ across companies (Farrel and Frame, 1997); Clark, Cornwell and Pruitt, 2002). Thus, the two central questions in this paper are: “Does a sponsorship announcement have a positive influence on the sponsoring firm’s stock return?” and “Which factors are able to explain the degree of abnormal return?”.

To test whether the companies show significant abnormal returns on and directly after the date of sponsorship announcement, this paper makes use of the efficient capital markets theory developed by Fama (1970). Since the net present value of all future expected returns is reflected in the stock price of the company, an announcement with a positive effect on shareholder value should result in an immediate price increase. Using this theory, Brown and Warner (1985) developed the parametric event study methodology, which is a guide for researchers to test whether an announcement or event actually had an impact on the shareholder wealth. In their buying and selling decisions, investors make judgments concerning the impact of various market events upon the sales, net revenues, and riskiness of the affected companies (Clark, Pruitt and Van Ness, 2001). Since stock price changes offer a measure of sponsorship success, free of some of the biases inherent in more subjective metrics, the event study methodology fits this research.

(4)

This research is different from earlier papers in several respects. First, hardly any research has been conducted regarding the sponsoring of World Cup and European Cup Football; as mentioned earlier, most research to date has focused on the Olympics exclusively. Second, earlier research focused on a single sports event: this paper is the first to investigate the influence of sponsoring various different sports events on shareholder value. This makes an interesting extension to this field of study, since the value added through sponsoring can differ depending on the sponsored event (Olson, 2010). In addition - to my knowledge - the relationship between the stock price and the announcement of becoming a sponsor of the FIFA world Cup or the UEFA European Cup has never been researched before.

The rest of this paper is built up as follows: in section 2, the literature is reviewed ; in section 3 and 4 the methodology and data are outlined. Section 5 discusses the results. Section 6 contains the conclusions and limitations of this research and presents suggestions for further research.

2. Literature Review

Since the mid 1980s, the interest amongst the academic and financial community as to the effectiveness of sponsorships has gone up dramatically (Walliser, 2003). The publication of academic papers led to new insights regarding the financial impact of the decision to become a sponsor. They showed that sponsorship results in increasing brand name awareness, increasing sales and that it enhances the company image and/or shareholder value (Gardmer and Schuman, 1988; Farrell and Frame, 1997).

(5)

of those papers are mixed: Farrel and Frame (1997) find a negative impact on the shareholder wealth, while the other papers like Miyazaki and Morgan (2001) and Samitas, Kenourgios and Zounis (2008) find a positive effect. In general, these papers conclude that sponsorship has a positive effect on the stock price of the sponsoring company. That is, at the moment a company announces that it will sponsor an event, stadium or athlete, the stock price of the company generally increases. An overview of the conclusions of the empirical papers can be found in table 1. The rest of this section presents the hypotheses attended to in this study, while referring to the relevant literature in table 1.

Research methods employed in the earlier literature

As can be seen in Table 1, the events investigated in the main literature are all sports related, varying from the Olympic Games to stadium and tournament sponsorships. The methodology employed in those studies is generally the OLS market model event study methodology developed by Brown and Warner (1985). However, some papers make use of the bootstrapping method or the Scholes-Williams standardized cross-sectional market model to test for abnormal returns. The Scholes-Williams approach was developed to eliminate the problems associated with non-synchronous trading that sometimes occurred in event based studies with firms of widely varying market values (Clark, Cornwell and Pruitt, 2008). The bootstrapping method is preferred in some cases if the researcher wants to provide more precise tests for small samples of firms, and this method is robust to time series correlation in returns (Samitas et al. , 2008). There are only a few papers that investigate a large sample of sponsors. In sports it is hard to conduct research on a large sample, since most large tournaments opt to have only a limited number of sponsors. They prefer to have a few large sponsors over a prolonged time period (Gillis, 2005).

(6)

Table 1 Overview of earlier empirical studies This table presents empirical studies regarding the influence of sponsorship announcements on the sponsoring firm’s stock price. For each paper, the authors, the investigated event, the number of observations, the event and estimation windows and the most important

conclusions are displayed.

Paper by Event Data Event Window Estimation Window Conclusion

Agrawal and Ka makura (1995) Sports celebri ty sponsors 110 events +/- 10 da ys -244 to -6 the onl y signifi cant resul t was for t=[-1,0]

whi ch had a positi ve return of 0.54%

Cla rk, Cornwell and Prui tt (2002) Sta dium Sponsors 49 Sponsors -20 to +50 da ys -175 to -26 La rge and si gnifi cant abnormal returns immedia tel y after the announcement: 1.4% duri ng t= [0,1] Cla rk, Cornwell and Prui tt (2008) Tournament sponsorship 114 events +/- 20 da ys Not mentioned No signi fica nt positi ve or nega ti ve abnormal returns Renewals nega ti ve signi fi cant influence on AR of 3% Cornwell, Pruitt and Cla rk (2005) Ma jor Lea gue 53 Sponsors +/- 25 da ys -275 to -26 Abnormal return of 0,5% duri ng t = [-5,5]

Fa rrel and Frame (1997) Ol ympi cs 1996 26 Sponsors +/- 5 da ys -250 to -10 In event window insigni fica nt nega ti ve return of -0.1880

t=[0,2] ha ve a signifi cant nega ti ve return of -0.4323

Mishra , Bobinski and Bha bra (1997) Special events 76 events +/- 5 da ys -147 to -22 t=[0,1] signifi cant posi ti ve return of 0.56%

Sports , Ol ympi cs and Other AR was positi vel y rela ted to ROA

Mi ya zaki and Morgan (2001) Ol ympi cs 1996 27 Sponsors +/- 5 da ys -125 to -6 1,24 % abnormal return duri ng t = [0,1]

Russell, Ma ha r and Drewniak (2005) Athleti c endorsers 137 events +/- 5 da ys Not used ‘Good’ and ‘bad’ events affect the s tock pri ce Ba d events ha ve a s tronger influence than good events

T=[-1;3] bad events : -2,45%, good events : 1,4% Sa mi tas, Kenourgios and Zounis (2008) Ol ympi cs 2004 21 Sponsors +/- 19 da ys Not mentioned Effect is hi gher among s mall fi rms

(7)

The size of the estimation windows that are used in the studies described in table 1 depend on the size on the event windows. Generally, the event period itself is not included in the estimation period to prevent the event from influencing the normal performance model parameter estimates (MacKinlay, 1997). Only in the paper of Agrawal and Kamakura (1995) the event window captures part of the estimation window. The length of the event window depends on the objective of the paper. The papers with an event window length over 20 days also test the long-term impact of the sponsorship announcements.

Abnormal returns

The outcomes of the papers regarding the value of sponsorships are somewhat mixed. Some papers find a positive return on the day of the sponsorship announcement, the event day, while others only find such an effect within a small event window directly following the event date. Most papers find the strongest stock price effect in the few days directly following or surrounding the event date, while a few papers do not find any significant effects. When a larger event period (of more than 20 days) is considered, most papers show a small, but significantly negative abnormal return, both before and after the event date. Nevertheless, the papers generally conclude that a sponsorship announcement has a significantly positive effect on firm stock price. The earlier findings lead me to hypothesize the following:

H1: A sponsorship announcement has a positive influence on the sponsoring company’s

stock price on the date of the announcement

H2: A sponsorship announcement has a positive influence on the sponsoring company’s

stock price during the days directly following the announcement

Determinants of sponsorship’s impact on firm market value

(8)

Firm specific factors

Firm specific variables that are used in earlier studies of cross-sectional sponsorship effectiveness are firm size, market value, market share, firm cash flow, ownership structure and whether or not a firm can be classified as being ‘high-technology’. In two independent studies, the market share and the high-technology dummy are found to have a significant influence on the effectiveness of sponsorship announcements (Cornwell, Clark and Pruitt, 2002 and 2005). In addition, the nature of a company - whether it serves other companies or consumers - is shown to have a significant influence on the degree of abnormal return after a sponsorship announcement, with consumer related companies showing a higher return (Mahar, Paul and Stone, 2004). Farrel and Frame (1997) find that the percentage ownership of the firm by large outside block-holders has a significant positive influence on sponsorship effectiveness. According to Samitas, Kenourgios and Zounis (2008) the effect on the stock price of domestic sponsors, when sponsorship is announced, is more significant than the effect on the stock price of international sponsors. In their sample, all the domestic sponsors are smaller than the international sponsors. Thus, they conclude that sponsorship announcements have a larger impact on small firms. On the other hand, Mishra et al. (1997) find no significant relationship between event-window abnormal return and firm size. Lastly, there appears to be a positive relationship between firm profitability, as measured by return on assets (ROA), and the event-window abnormal return after a sponsorship announcement (Mishra et al. 1997). This finding implies that event sponsorship activities are viewed more positively by the market for those firms that have been (more) profitable in the past. Taking into account data availability restrictions, these earlier findings lead me to raise the following two hypotheses for testing in this paper:

H3: The degree of abnormal return after a sponsorship announcement is negatively

related to the market value of a company

H4: The abnormal return after a sponsorship announcement is higher for a

(9)

Contract related factors

Examples of contract related factors that may influence the impact of sponsorship activity on stock returns are sponsorship spending, sponsorship type and sponsoring contract length. First, the duration of a sponsorship contract is shown to be positively related to the sponsor’s abnormal stock return (Clark, Cornwell and Pruitt, 2002). Second, it appears that the contract length of the sponsorship has a negative influence on the abnormal return. The price of title sponsorships is increasing, which might suggest that renewals are perceived less positively than first signings (Clark, Cornwell and Pruitt, 2008). On the other hand, from a strategic communications perspective, continuing in a sponsorship relationship over time offers important benefits such as image continuity.

The sponsors of the Olympic Games, as well as those of the World cup and the European cup games are split up into different categories. A couple of papers look at the difference in the effect of sponsorship announcements on stock price by sponsorship type . Farrel and Frame (1997) cannot conclude that different sponsor categories have differing abnormal returns. Their finding contradicts with the conclusion of Samitas, Kenourgios and Zounis (2008), namely that different types of sponsors have a diverse impact on the sponsor’s stock return. Specifically, their results suggest that ‘partner-type’ sponsors have a higher abnormal return than other types of sponsors. A possible explanation is that, because partners together invest the largest amount of money in the events, and since they have the highest rights for advertising, they benefit more from a sponsorship contract when compared to other sponsor types (Samitas, Kenourgios and Zounis, 2008). Regarding the relationship between the type of sponsorship and the degree of abnormal return after a sponsorship announcement, I raise the following hypothesis:

H5: Partner-type sponsors have higher abnormal returns after a sponsorship

announcement than other types of sponsors Socio-economic factors

(10)

population size of the host country seems not to be related to abnormal returns. On the other hand, local sponsors have significantly higher abnormal returns than sponsors from foreign countries (Clark, Cornwell and Pruitt, 2002). Furthermore, as Walliser (2003) suggests, the effectiveness of a sponsorship depends on the wealth within the home country, i.e. in the market in which the sponsoring company is primarily active. One of the variables that can be used as a proxy for home country wealth is its Gross Domestic Product (GDP). Walliser (2003) claims that there exists a positive relationship between the wealth of the country a company operates in and the effectiveness of a sponsorship in terms of market share and sales. To test whether this relationship also holds for the abnormal return, the following hypothesis is used:

H6: There exists a positive relationship between the GDP of the sponsor’s home country

and the abnormal return following a sponsorship announcement

Last, regarding the location of the sponsoring firm, I hypothesize the following:

H7: Domestic sponsors have higher abnormal returns after a sponsorship announcement

than do foreign sponsors

3. Methodology

(11)

Abnormal returns can be obtained in several ways. In their seminal article , Brown and Warner (1985) outline three different abnormal return models. Specifically, the abnormal return can be computed using the market model, or one can rely on the mean adjusted return or the market adjusted return. The mean adjusted return is the easiest to obtain and Brown and Warner (1980, 1985) find that it often yields results similar to those of more sophisticated models. However, I choose to compute abnormal returns according the market model in this paper. In this model, the return of any given security is related to the return on the market portfolio. The market model represents a potential improvement over the two excess returns models. By removing the part of the return that is related to variation in the market’s return, the variance of the abnormal returns is reduced. This in turn can lead to increased ability to detect event effects. (MacKinlay, 1997)

The first step in the application of the market model is one of defining the estimation-, event- and post-event window. When using daily stock returns, as in this study, it is common to have 250 daily return observations for the period around its respective event (Brown and Warner, 1985). Following Brown and Warner (1985), the estimation period is day -244 through -6, when day 0 is the event day. The event period is day -5 through +5 and the post event is day 5 through 30. This is illustrated in Figure 1. Before implementing this procedure, all daily stock returns that were gathered are converted into Euros, since without this adjustment the market model could overestimate the changes in the security price (Park, 2004).

T

0

T

1

τ

= 0

T

2

T

3

(post-event window]

(

event window]

(

estimation window]

Figure 1 Estimation and event window The estimation window, which is used to estimate the parameters of the model , runs from T0 to T1 [-244,-6]. The

abnormal stock returns in the event window from T1 to T2 [-5,+5] are calculated

(12)

Firm daily stock returns are obtained by taking the log-returns per day, applying the following formula: iτ+1 iτ iτ

ln RI

R =

ln RI

(1)

Where RIiτ+1 is the beginning-of-the-day total return index of firm i on day τ+1 and RIiτ is the

total return index at the beginning of day τ.

The second step is the estimation of the market model parameters. Under general conditions, the ordinary least squares (OLS) regression method is a consistent estimation procedure to obtain these parameters. The following regression is run:

R

iτ

=

i

+ β

i

R

mτ

+ ε

iτ (2)

where

Riτ is the return on the shares for firm i at time (date) t

Rmτ is the return on the market portfolio m at time (date) t

i , βi are the parameters of the model

εiτ is the statistical margin of error for which the expected value E(εiτ) = 0 and the

variation VAR(εiτ) = σ²εi.

The OLS estimators of the market model parameters for an estimation window of observations are described by the following formulae:

(13)

ˆ

1 t 0 T 2 2 ε iτ i i mτ τ=T +1 1 0

1 σ =

(R - α -βR )

T - T - 2

(5) where ˆ 1 0 T i iτ τ=T +1 1 1 μ = R L (6) and ˆ 1 0 T m mτ τ=T +1 1 1 μ = R L (7)

Abnormal returns can be measured and analyzed using the parameters calculated according to equation (3)-(7) as inputs. A proxy for the market portfolio is needed in applying the formulae. MacKinlay (1997) uses the total return index values of the S&P 500 for this purpose, but his research sample includes only US firms. The sample used in the present paper covers sponsoring firms from all over the world. Therefore, I choose to use the FTSE world index as a proxy for the market portfolio. The FTSE world index is a broad based stock index, which is necessary in order to regard it a good market portfolio (MacKinley, 1997). This world index is converted into Euros, so that its values are consistent with the currency of the securities’ data. The use of the world index as the market proxy in combination with a multi-country event may cause a bias in the abnormal firm return values (Park, 2004). I therefore cross-check the use of this proxy by creating a different set of abnormal returns, where equations (4)-(7) are applied using the domestic market index of each sponsoring company as proxies for the (domestic) market portfolio. This approach is suggested by Park (2004). A t-test for the equality of means is used to check whether the results obtained when the FTSE world index is used as the market proxy differ from those obtained when country-specific market indices are used as market proxies.

When the outcomes of equations (3) and (4) - the parameter estimates - are inserted in the formula in equation (2), one obtains the expected return. The abnormal return (AR) at any time can then be calculated as the difference between the actual return and the expected return.

That is:

ˆ

iτ iτ i i mτ

(14)

The average abnormal return (AAR) across the sample at a particular time (day) τ is equal to:

N

τ iτ

i=1

AAR = 1 / N AR

(9)

As MacKinlay (1997) and Samitas, Kenourgios, and Zounis (2008) explain, the abnormal return observations must be aggregated in order to draw overall inferences for an event. This aggregation takes place in two dimensions, i.e. across securities and over time. The cumulative average abnormal return (CAAR) is the sum of all daily average abnormal returns over the period event window:

t

i x t τ

x=1

CAAR τ , τ =

AAR

(10)

To test whether significant abnormal returns exist over a particular event period, the following test-statistic is used: x t x t τ ,τ τ ,τ CAAR θ = N(0,1) var(CAAR ) (11)

(15)

Diagnostic tests

Since the abnormal returns are calculated using OLS model estimation, is it important to perform diagnostic tests on the data. The first check is to see whether the regression residuals are homoscedastic, i.e. whether they have a constant variance. If the errors do not have a constant variance, they are said to be heteroscedastic. When the errors are heteroscedastic, but this fact is ignored and the researcher proceeds with model estimation, the OLS estimators will still give unbiased coefficient estimates, but they no longer have the minimum variance among the class of unbiased estimators (Koop, 2006). The White heteroscedasticity test can be used to detect heteroscedasticity. When heteroscedasticity is detected in the sample, the researcher can make use of heteroscedasticity-robust standard errors to cope with this problem (Brooks, 2008). In the dataset used in this study, the OLS errors turn out to be heteroscedastic and thus heteroscedasticity-robust standard errors are used. Secondly, the OLS estimators rely on an assumption of normally distributed model residuals (Brooks, 2008). One of the most commonly applied tests for normality is the Jarque-Bera test. This test uses the property of a normally distributed random variable, that the entire distribution is characterized by the first two moments – the mean and variance. The standardized third and fourth moments of a distribution are known as its skewness and kurtosis. Skewness measures the extent to which a distribution symmetric about its mean value and kurtosis measures how fat the tails of the distribution are. A normal distribution is not skewed and is defined to have a coefficient of kurtosis of 3 (Brooks, 2008). When evidence of non-normality in the residuals is found, it is not obvious what should be done. The data can be checked on extreme residuals, which may cause a rejection of the normality assumption. The corresponding data points can be removed so that the residuals follow a normal distribution more closely. However, this means that a data point which represents a useful piece of information is removed, and thus it is decided that in some cases it is best to not remove outliers (Koop, 2006). It turns out that the residuals from the market model regression are not normally distributed. After checking that extreme outliers are not due to errors in the data, I decide to not remove any data points.

Nonparametric test

(16)

MacKinlay, 1997). When abnormal returns are non-normally distributed, parametric tests are not well specified. Non-parametric tests do not rely on an assumption of normal distribution and are more powerful at detecting a false null hypothesis of no abnormal returns (Serra, 2002).

Corrado (1989) introduced a popular nonparametric test for use in event studies. His rank-test dominates most other nonparametric tests, such the sign test, in detecting abnormal performance (Corrado and Zivney, 1992). Specifically, Corrado (1989) claims that his nonparametric rank test for abnormal security-price performance is preferable to the parametric t-test for a broad spectrum of fat-tailed security returns distributions. He shows that the requirement that the excess-return distributions are symmetrical for correct test specification is often problematic for parametric tests. The method Corrado (1989) outlines in his study is correctly specified no matter how skewed the cross-sectional distribution of abnormal returns.

To apply the nonparametric test, first the daily market-model based abnormal returns are ranked for each firm in the sample. The abnormal returns used in ranking are those from day t=-244 to t=5, that is, those daily returns realized in the estimation period plus the event period (see figure 1). Each firm’s abnormal returns are ranked from the lowest (250) to the highest (1) over this 250-day period. The ranking procedure transforms the distribution of security excess returns into a uniform distribution across the possible rank values regardless of any asymmetry in the original distribution (Corrado, 1989). The average rank equals one-half plus the half the number of observed returns, which is 125.5. After the ranking procedure, the following day 0 test statistic can be calculated:

n it i=1 +5 n 2 it t=-244 i=1

1 (K -125.5)

n

T =

1

1 (

(K -125.5))

250 n

(12) Multivariate regressions

(17)

multivariate cross-sectional regressions on the firm abnormal returns on event day 0, day 1, day 2 as well as on the cumulative event returns on day 0-2 and day 1-5. The determinants used in this multivariate regression are motivated on the literature discussed section 2. Specifically, I research the relation between abnormal returns after a firm’s sponsorship announcement and the firm market value, business type, sponsorship type, home country wealth and firm location.

The size of a firm, as proxied by multiplying the number of outstanding shares by the stock price of the firm seems to be related to whether a sponsoring firm is located in the home country of a sports event or not. In the earlier papers, small companies are often domestic sponsors (e.g. Clark, Cornwell and Pruit, 2002. To take account of the possibility of a size-home country interaction effect, a dummy taking a value of 1 for a domestic company will be included in the multivariate regression. Since the study by Mahar, Paul and Stone (2004) indicates that the effect of a sponsorship announcement is larger when the sponsor is a business-to-consumer (B2C) company in comparison to the effect of a business-to-business (B2B) company, a dummy taking a value of 1 for consumer-related production will be included as well. Another independent variable included in the regression is the GDP of the sponsor’s home country. Since wealthier companies have more money to spend - we assume a positive correlation between home country wealth and company wealth - it is expected that sponsors from wealthier countries have a more successful sponsorship advertising campaign and are therefore able to generate higher abnormal returns after a sponsorship announcement (Walliser, 2003). This discussion leads to the following regression model:

(C)ARit = β1Mi,t + β2Di + β3Ci + β4Gi,t + β5Pi + β6Si + eit (13)

Where:

Mit is the market capitalization of firm i at time (date) t

Di is a dummy taking a value of 1 for a domestic company and 0

for a foreign company

Ci is a dummy taking a value of 1 for a B2C company and 2 for a

B2B company

(18)

Pi a dummy for the type of sponsor, taking a value of 1 when the

company is a partner type of sponsor

Si a dummy for the type of sponsor, taking a value of 1 when the

company is a supporter type of sponsor

, β1, β2, β3, β4 ,β5,β6 are the coefficients of the model

This regression is run on both the simple abnormal returns and the cumulative (in this case, calculated over different event windows) abnormal returns in the event window. To test whether there exists a difference in the power of the determinants among the different events, this regression is also run on a split sample, based on the timing of the event (Summer-Winter) and the type of event (Olympics-Football).

4. Data

This paper covers the sponsors of some of the major sporting events that occurred during the past six years. The six-year time interval is chosen because of data availability issues. Specifically, information regarding sponsors of the Olympic Games that took place over four years ago cannot easily be recovered, and information regarding sponsors and announcement dates of the large football tournaments beyond six years back is hard to find. The sample therefore includes sponsors of five Olympic Games, as well as those of four large football tournaments from 2006 on. Unfortunately, the sponsors of the Olympics of Torino 2006 could not be included in the sample, since not enough information could be recovered on the internet. The sponsors of the Olympics of Athens 2004 are included because the sponsorship announcement dates regarding this version of the games are given in the study of Samitas, Kenourgios and Zounis (2008).

(19)

Supporters’ and ‘Official Suppliers’. These sponsors are a mix of national and international companies. The data used in this study comes from Partners, Supporters and Suppliers for the Olympics in Sochi, London, Vancouver, Beijing, and Athens. The ’Worldwide Partner’ type Olympic sponsors are not included in the sample for this research, because of a lack of available data in this regard.

Table 2 Overview of the Sponsors This table shows how the 100 sponsors in the sample relate to the different sports events. In column 3 through 9 other characteristics of these sponsors are displayed. The third and fourth column present respectivel y the number of domestic and foreign sponsors in the sample.

The fifth trough seventh column show the amount of Partner, Supplier and Supporter sponsors in the selection. The last two columns illustrate the number of respectively Business -to-Consumer and

Business-to-Business companies in the sample.

Event Sponsors Domestic Foreign Partner Supplier Supporter B2C B2B Summer Olympics

Athens 2004

15 6 9 4 6 5 8 7

FIFA World Cup Germany 2006

9 2 7 9 0 0 9 0

UEFA European Cup Austria-Switzerland 2008 6 0 6 5 1 0 6 0 Summer Olympics Beijing 2008 6 1 5 6 0 0 6 0 Winter Olympics Vancouver 2010 26 14 12 3 5 18 17 9

FIFA World Cup South Africa 2010

10 3 7 7 3 0 8 2

(20)

The FIFA and the UEFA use a similar system for categorizing their sponsors as that for the Olympics, but they group Suppliers and Supporters together under ‘Sponsor’. The Partners of the UEFA and FIFA are generally long-time sponsors, but their contracts are renewed regularly (www.fifa.com). The specific football tournaments whose sponsors are included in the sample are the World Championships in South-Africa and Germany, and the European Championships in Austria-Switzerland and Poland-Ukraine.

I manage to identify 164 available sponsorship announcement dates for use in this study. These announcement dates were found on the homepage of the Olympics and that of the F IFA and UEFA, as well as in previous studies and local newspapers. Unfortunately, 64 of these cases had to be excluded from further analysis because they involved privately held companies or firms for which stock price data was unavailable. A list of the selected sponsors that are included in the final sample is displayed in appendix 1. A summary of the included sponsors, categorized by event, is shown in table 2.

To determine whether a company was a domestic or foreign sponsor, the ISIN-codes of the companies were compared to the country codes given by Thomson Datastream. The ISIN-codes in Datastream, which are unique for every company, always start with the country code. This code shows the country in which the company and the accompanying stock are based. The data needed to determine the type of sponsor could be found via the event’s home page. Information about the type of business was found by checking the company’s home page, while cross-checking this with the core-business of the company as described in its mission statement. An overview of all the variables and summary statistics can be found in appendix 5 and 6.

The stock prices needed to generate abnormal returns and as inputs in the multivariate regressions were collected via Datastream. The data consists of the beginning-of-the-day total return index values around the date of sponsorship announcement of each of the 100 sponsors in the sample.

(21)

indices of the sponsors. All index values, which are used to calculate daily log-returns, are retrieved from Datastream.

Table 3 Overview of stock returns This table shows the average security return, along with key descriptive statistics. In the tenth column, average of the accompanying market returns as calculated

using the FTSE world index is displayed for each respective event day.

Security Returns Market

Return Day Mean Maximum Minimum

Standard

Deviation Skewness Kurtosis

Jarque-Bera Observations Mean -5 -0.58% 5.1% -13.9% 3.2 -1.894 7.810 156.222 100 0.02% -4 -0.43% 7.6% -12.0% 2.8 -0.512 6.751 63.031 100 -0.30% -3 0.21% 16.7% -10.9% 3.3 1.082 1.078 271.913 100 -0.20% -2 -0.27% 5.4% -22.3% 3.1 -3.607 2.580 2384.747 100 0.06% -1 -0.00% 9.7% -12.5% 2.8 -0.822 9.692 197.888 100 -0.01% 0 0.19% 14.4% -6.9% 2.6 1.317 1.063 271.966 100 0.01% 1 0.49% 10.9% -8.1% 2.7 0.595 5.707 36.457 100 0.00% 2 0.51% 12.4% -5.5% 2.8 1.755 8.494 177.183 100 0.05% 3 -0.09% 7.4% -6.9% 2.1 -0.057 4.834 14.07.430 100 -0.04% 4 0.68% 40.5% -13.3% 5.0 5.098 4.140 6579.211 100 0.12% 5 0.48% 11.9% -15.8% 3.7 -0.448 8.130 113.049 100 -0.01%

(22)

5. Empirical Results

This section first presents and discusses the empirical results for the parametric test developed by Brown and Warner (1985) and Corrado’s (1989) rank-test. After this, I discuss the results of the multivariate regression model.

In applying the event study methodology, first the average and cumulative abnormal returns are calculated. The sponsoring firms’ returns are matched to the return on a market proxy and equation (2) is used to obtain expected returns. The abnormal returns are found as the difference between the actual returns and the expected returns.

Table 4 The average abnormal returns and the test statistics on the days in the event window. The second, thi rd

and fourth column represent the Avera ge Abnormal Returns (AARs ) cal cula ted wi th the ma rket model and thei r a ccompanyi ng pa ra metri c and nonpa rametri c test s ta tisti cs , where the return on the FTSE world i ndex is used to proxy for the ma rket return. The fifth, si xth and seventh column displa y the sa me tes ts , but here the domes ti c ma rket

indi ces a re used to proxy for the ma rket. The da taset consists of dail y returns of 100 sponsors of la rge sports events.

Day AAR

(Ma rket proxy = world index) Parametric Test statistic Corrado test statistic AAR

(Ma rket proxy = domes ti c i ndex) Parametric Test statistic Corrado test statistic -5 -0.52% -1.760 0.029 -0.66% -2.282 -0.535 -4 -0.16% -0.664 -1.893 -0.38% -1.463 -2.064 -3 0.40% 1.320 0.430 0.12% 0.365 0.367 -2 -0.41% -1.310 -0.349 -0.18% -0.579 1.111 -1 -0.10% -0.383 -0.052 0.00% 0.022 -0.141 0 0.22% 0.949 0.122 0.16% 0.676 0.409 1 0.61% 2.780*** 1.487* 0.41% 1.654** 1.948** 2 0.48% 1.773** 0.649 0.54% 1.889** 1.278 3 0.07% 0.360 0.156 -0.03% -0.169 -0.602 4 0.40% 0.975 0.402 0.26% 0.631 0.500 5 0.52% 1.392 2.909*** 0.55% 1.559* 1.656**

Note: ***, ** and * denote significance one-sided testing at 1, 5 and 10 % respectively

(23)

cumulative average abnormal returns (CAARs) and the accompanying test statistics for varying event windows. Both tables also show the results when the domestic market index of each sponsor is used as a proxy for the market portfolio instead of the FTSE world index. The abnormal returns reported are calculated with log-returns, and are subsequently multiplied by 100.

On the basis of earlier literature, I expect to find a positive abnormal return on the event day or on the few days directly following the event day. The results displayed in table 4 show a significantly positive average abnormal return on both day 1 and 2 (.61% and .48%, respectively) for the returns calculated using the world index. These results are the same as those obtained when domestic indices are used as the market proxy instead of the world index. When the domestic indices are used as market proxies, a significantly positive average abnormal return is found on both day 1 and 2 (.41% and .54%, respectively). The rank-test statistics, however, lead to different conclusions. The rank-test shows a positive average abnormal return on the first day of the event period, since the average rank on this day (when using the world index as the market proxy) is 139.96, which is above the expected 125.5 (the highest abnormal returns have 250 as rank). Possible explanations for the finding of no significant abnormal return on the event day itself (day 0) are that sponsorship announcements are not published on the same day, or that sponsorship announcements (sometimes) take place when the stock-markets are closed (Agrawal and Kamakura, 1995). Looking at the inconsistency in the findings according to the different tests, I can conclude that the abnormal return on the singular day 1 following the event day has an above average abnormal return. Therefore, the first hypothesis presented in the introduction is not rejected.

(24)

Table 5 Overview of the cumulative average abnormal returns (CAARs) and the test statistics for various event windows

In this table, the CAARs a re displa yed for va rious event windows . The fi rs t column shows the event window on whi ch the CAAR is calculated. The second, thi rd and fourth column represent the cumula ti ve a verage abnormal returns whi ch a re cal cula ted using the ma rket model , wi th the a ccompanyi ng pa ra metri c and nonpa rametri c test s ta tisti cs , where the FTSE world index is used as a ma rket proxy. The fifth, si xth and seventh column display the same tests , but

here the domes ti c ma rket indi ces a re used as ma rket proxies . The returns shown a re the cumulati ve returns for the da ys in the event window. The da ta consists of 100 sponsors of la rge sports events.

Event Window CAAR (Market proxy = world Index) Parametric Test statistic Corrado Test statistic CAAR (Market proxy = domestic index) Parametric Test statistic Corrado Test statistic [-5, 5] 1.50% 1.656** 1.047 1.21% 1.249 _1.184 [0, 5] 2.31% 2.855*** 1.881** 2.16% 2.554*** _2.118** [1, 5] 2.18% 2.840*** 2.006** 1.97% 2.592*** _1.934** [0, 2] 1.26% 2.913*** 1.304* 1.15% 2.582*** _2.098** [1, 3] 1.09% 2.754*** 1.143 0.96% 2.423*** _1.514* [2, 4] 1.06% 1.848** 0.052 0.93% 1.682** _0.678 [3, 5] 1.12% 1.660** 1.356* 1.00% 1.553* _0.897 [2, 3] 0.43% 1.283* 0.348 0.53% 1.625* _0.477 [1, 2] 1.04% 2.932*** 1.510* 0.98% 2.815*** _2.281** [0, 1] 0.82% 2.848*** 1.138 0.61% 1.745* _1.666** [-1, 0] 0.11% 0.293 0.049 0.17% 0.437 _0.189

Note: ***, ** and * denote one-sided significance at 1, 5 and 10%, respectively

(25)

respectively, an average return of around 0.43% a day. This result is thus consistent with the expected results of this study (hypothesis 2).

Now that the AARs have been calculated for the entire sample, it is interesting to see whether there is a difference in AARs between the various events. Returns following sponsorship announcements of Olympics’ sponsors have been researched before, but those of sponsors of Football tournaments have not. Therefore, splitting the sample may bring to the fore differences in AARs. For both subsamples, the AARs are calculated using the Brown and Warner (1985) method with the world-index as a proxy for the market index. The results of this splitting are shown in Appendix 2, table 8. Some of the results of the split sample are similar to the results obtained over the entire sample. Again, the AAR on the first day of the event period is significantly positive, 0.5% (t=1.909) for the Olympics and 0.8% (t=2.147) for the Football sponsors. However, the subsample of Football Tournaments’ sponsors shows a negative, insignificant AAR on day 2 (-0.2%), where the Olympics’ sponsors show a strongly significant positive AAR of 0.7% (t=2.083). On day 5 the AARs of the Football tournament sponsors are significant and positive (1.9%, t=2.635), where the Olympics’ sponsors show a negative but insignificant AAR of -0.0%. The AARs of the Olympics’ sponsors are consistent with the expectations based on the earlier literature, however, the results of the Football Tournaments on event day 2 and 5 are unexpected.

The subsample of the Olympics can be further split up into two parts, the Summer and the Winter games. When the tests are conducted separately for the Summer and Winter Olympics, some interesting results appear. These are reported in appendix 2, table 9. It turns out that a significant positive AAR on the second day of the event period is only present at the Winter Olympics (1.7%, t=2.972), whereas a significant positive AAR on day 1 is only present among the Summer sponsors (0.7%, t=1.821). When checking the data again, no large outliers are found among the Winter sponsors, and therefore no reasonable explanation exists for the strong significant AAR on day 2. These results cannot be explained by the earlier literature.

Multivariate regression

(26)

insight into the market’s perception of sport event sponsorship, a multivariate regression analysis is conducted with firm-specific factors as independent variables and the AARs and CAARs of the firms as dependent variables. This kind of analysis may lead to new knowledge about which firms are likely to benefit the most from being a sponsor.

Table 6 presents the results of the regression analysis as defined by equation (13). As dependent variables, the AARs on day 1 and 2 of the event period and the CAARs over the interval [0,2] and [1,5] are used, since the strongest results were found for these days and periods in the previous subsection. Day 0 is included, since this is the actual date on which a sponsorship announcement took place. The adjusted R-squared values of the multivariate regressions are all below 0.284, indicating that the regression model can explain 28.4 percent of the variance in de dependent variable. This implies that the explanatory power of the model is relatively weak, which is confirmed by the values of the model F-statistic, which is insignificant almost all cases.

The independent variables used in the regressions are grouped into firm specific, contract related and socio-economic factors. To test whether the relationship between abnormal returns and the independent variables differs among the various type of events, the same tests are also conducted on the subsamples Olympics, Football, Summer Olympics and Winter Olympics. The results of these tests can be found in appendix 3, tables 10-13. In the following paragraphs, the results of the multivariate regressions are discussed per factor type.

(27)

Table 6 Overview of the results of the multivariate regressions

This table represents the results of multivariate regressions as defined by equation (13). The first column presents the independent model variables, and each row represents respectively (from left to right) the

model coefficients when the r egressions are run with the abnormal returns on day 0, 1 or 2 as the dependent variable or the cumulative abnormal return in the time intervals [0,2] and [1,5]. T-statistics are

reported in parentheses. The independent variables are a domestic -dummy, a B2C dummy, firm market capitalization, home country gross domestic product, a dummy for the partner-type sponsors and a dummy for the supporter-type sponsors as the independent variables. The regr essions are run for 100

sponsors of large sports events.

Variable AR (0) AR (1) AR (2) CAR [0,2] CAR [1,5] Constant 0.010 (1.270) 0.001 (0.141) 0.022** (2.410) 0.037** (2.379) 0.059** (2.266) Domestic -0.009 (-1.512) 0.010* (1.871) -0.003 (-0.407) -0.002 (-0.148) 0.008 (0.400) B2C -0.002 (-0.275) 0.005 (0.907) -0.012 (-1.555) -0.005 (-0.407) -0.017 (-0.786) Market Cap (* 10-12) GDP (*10-15) 0.007 (0.681) -0.638 (-1.186) -0.017* (-1.743) 0.622 (1.280) -0.007 (-0.534) -0.637 (-1.019) -0.018 (-0.813) -0.660 (-0.616) -0.038 (-1.022) -0.996 (-0.550) Partner -0.004 (-0.585) -0.008 (-1.385) -0.005 (-0.732) -0.028** (-2.218) -0.025 (-1.158) Supporter 0.006 (0.876) -0.006 (-0.869) -0.007 (0.813) -0.013 (-0.911) -0.052** (-2.169) Adj. R-squared -0.002 0.063 -0.004 0.020 0.023 F-statistic 0.975 2.11 0.930 1.331 1.395 Probability (F) 0.447 0.059 0.477 0.251 0.225 N 100 100 100 100 100

(28)

On the basis of earlier literature, it was expected that the announcement of a sponsorship would have a larger impact on the smaller firms than on the larger firms (Samitas, Kenourgios and Zounis, 2008). The finding of a majority of negative coefficients is therefore in line with the literature and the expectation as formulated in hypothesis 3, however, since the evidence is not definite, the hypothesis needs to be rejected. The coefficients on the dummy for firm type (B2C versus B2B), when estimated for the entire sample, take on both positive and negative values in the interval of -0.039 to 0.037, depending on the time period studied. A value of -0.039 implies a lower abnormal return of 0.039 percent point when a company is a B2C type of company. When the sample is split up, the majority of the results show a negative relationship between the B2C-type and the abnormal return. However, in most cases the coefficients are insignificantly different from zero and thus the evidence is not strong enough to conclude the existence of any relationship between company type and abnormal returns following a sponsorship announcement. The earlier literature suggests that the B2C companies have - on average - higher abnormal returns in comparison to B2B companies (Mahar, Paul and Stone, 2004). Thus, the results are not in line with the expectations and hypothesis 4 is rejected.

(29)

suggesting that the abnormal return after a sponsorship announcement by a local firm is significantly higher (Clark, Cornwell and Pruitt, 2002 and 2008). The results provide no evidence of a significant relationship between the location of the sponsoring firm and the abnormal returns after a sponsorship announcement. Therefore, hypothesis 7 is rejected.

The contract related independent variable used in the regression analysis is the contract type. Specifically, the UEFA, the FIFIA and the Olympics have their sponsors grouped into categories. They all make use of the partner- and supporter types (the UEFA and FIFA refer to supporter as sponsor), but only the Olympics add the supplier type to this list. Earlier research resulted in a suggestion that the type of contract is related to the degree of abnormal return after a sponsorship announcement. It is expected that the partner-type sponsors have the highest abnormal return relative to the other type of sponsors. (Samitas, Kenourgios and Zounis, 2008) When comparing the results for the different types of sponsors over the whole sample and the subsamples, it can be concluded that the supplier-type sponsors have the highest abnormal return on average. That is, in most cases, the coefficients of the partner and supporter type sponsors are negative, implying that the supplier-type which is the ’base case’ has a more positive or less negative coefficient than the other two types of sponsors do. These findings are not consistent with the suggestions of the literature and thus not supportive of hypothesis 5.

Overall, most of the results from the multivariate regressions are not consistent with the expectations based on the earlier literature. The next section summarizes the findings and further discusses the implications and limitations of this study.

6. Conclusion, Implications and Limitations

Paper summary

(30)

sponsoring of new athletic events. This article ads new empirical findings to the growing literature on sponsoring firms’ stock returns.

First, an analysis of the average abnormal returns (AARs) and the cumulative average abnormal returns (CAARs) in the event window showed that a firm’s decision of becoming a sponsor for the Olympic Games or for a Football Championship is treated as a sign of good news by shareholders, providing them with positive abnormal returns. For the events tested, sponsorship announcements were expected to have a positive impact on firms’ stock returns. The empirical results turned out positive and significant for each of the events. Positive and significant CAARs are found for all the time intervals covering the first couple of days after a sponsorship announcement. The most significant and positive CAAR was found on the event window [0,2] and showed an average return of 0.42% per day. However, the results according to the parametric test statistic developed by Brown and Warner (1985) do not always correspond with those according to Corrado’s rank-test (1989); part of the results of the two test types conflict amongst the different events. In some cases, the average abnormal return is significant only on the second day in the event window and in other cases only on the first day. These results are in line with the findings of Miyazaki and Morgan (2001) and Clark et al. (2002).

On the basis of the earlier literature, it is expected that smaller, domestic, partner, and business-to-consumer companies would have a higher degree of abnormal return in comparison to, respectively, larger, non-domestic, supplier or supporter, and business-to-business companies. In this article, a sponsorship announcement appears to lead to higher average abnormal stock returns in smaller firms. However, this finding is insignificant in most cases, and thus the hypothesis needs to be rejected. Most other determinants turned out to be insignificantly related to the event-window abnormal returns. One surprising result was the effect of the type of sponsor. It was expected that the partner-type would have the highest return. Instead, the supplier-type was shown to have the highest abnormal return in most cases, and therefore the expectations were inconsistent with the findings of Samitas, Kenourgios and Zounis (2008).

Limitations and suggestions for future research

(31)

ago. The inclusion of more events would have increased the reliability of the research findings. Another limitation lies in the number of determinants that could be included in the multivariate regression model specification. More information regarding possible determinants for the degree of abnormal returns was unavailable via the internet or Thomson Datastream. Therefore, it was impossible to include more explanatory factors for the degree of abnormal return, which may have caused omitted variable bias in the results (Brooks, 2008). The inclusion of more variables could possibly have lead to better regression specifications. The inclusion of more variables is also a suggestion for future research.

(32)

References

Agrawal, J. and Kamakura, W.A., 1995, The economic worth of celebrity endorsers: an event study analysis, Journal of Marketing, 59, 56-62.

Apostolopoulou, A. and Papadimitriou, D., 2004, ‘Welcome Home’: Motivations and Objectives of the 2004 Grand National Olympic Sponsors, Sport Marketing Quarterly, 13, 180-192.

Books, C., 2008, Introductory Econometrics for Finance, Cambridge University Press, second edition Brown, S.J., and Warner, B.J., 1985, Using daily stock returns: the case of event studies, Journal of

Financial Economics, 14, 3-31.

Brown, S.J., and Warner, B.J., 1980, Measuring security price performance, Journal of Financial Econo mics, 8:3, 205-258.

Clark, J.M., Cornwell, T.B., and Pruitt, S.W., 2002, Corporate Stadium Sponsorships, Signaling Theory, Agency Conflicts, and Shareholder Wealth, Journal of Advertising Research November/Dec ember, 16-32. Clark, J.M., Cornwell, T.B., and Pruitt, S.W ., 2008, The impact of title event sponsorship announcements on shareholder wealth, Markel Let, 20, 169-182.

Cornwell, T.B., Pruitt, S.W., and Clark, J.M., 2005, The Relationship Between Major -League Sports’ Official Sponsorship Announcements and the Stock Prices of Sponsoring Firms, Journal of the Academy of

Marketing Science, 33, 401-412.

Corrado, C.J., 1989, A nonparametric test for abnormal security returns, Journal of financial econo mics, 23, 385 – 395.

Corrado, C.J. and Zivney, T.L., 1992, The specification and power of the sign test in event study hypothesis tests using daily stock returns, Journal of Financial and Quantitative Analysis, 27:3, 465-478.

Crompton, J., 2004, Conceptualization and alternate operationalizations of the measurement of sponsorship effectiveness in sport, Leisure studies, 23: 3, 267-281.

Dyckman, T., Philbrick, D., and Stephan J., 1984, A comparison of event study methodologies using daily stock returns: a simulation approach, Journal of Accounting Research, 22, 1-30.

Fama, E.F., 1970, Efficient Capital Markets, Journal of Finance, 25:1, 283–417.

Fama, E.F., 1970, Efficient capital markets: a review of theory and empirical work. Journal of Finance 25:2, 383-417.

Fannin, R., 1988, Gold Rings of Smoke rings?, Marketing and Media Decisions September, 64-70.

Farrel, F.A. and Frame, W.S., 1997, The value of Olympic Sponsorships: Who is capturing the gold?, Journal

of Market Focused Management,2, 171-182.

Gardner, M.P., and Schuman, P.L., 1987, sponsorship: an important component of the promotions mix,

(33)

Gardner, M.P., and Shuman, P.L., 1988, sponsorships and small businesses, Journal of Small Business

Management, 26:4, 44-52.

Gillis, R., 2005, FIFA turns the screw, Marketing, 17.

IEG, Inc., 2005, 06 Outlook: s ponsorship growth back to double digits, IEG Sponsorship Report, 24, 4–5. IEG, Inc., 2007, Projection: sponsorship growth to increase for fifth straight year, IEG Sponsorship, Report 26, 1,4.

Javalgi, R., Traylor, M., Gross, A., et al., 1994, Awareness of Sponsorship an Corporate image: An Empirical Investigation, Journal of Advertising Dec ember, 47-58.

Kinney, W.R. Jr., and McDaniel, L.S., 1996, How to improve effectiveness of substantive analytical procedures, CPA Journal, 66:4, 54-55.

Koop, G., 2006, Analysis of financial data, John Wiley and Sons, first edition

MacKinlay A.C., 1997, Event studies in Economics and Finance, Journal of Econo mic Litera ture, 35, 13-39. Mahar, J., Paul, R. and Stone, L., 2004, An Examination of Stock Market Response to NASCAR Race Performance, Working Paper.

Mahony, D.F. and Howard, D.R., 2001, Sport business in the next decade: A general overview of expected trends, Journal of Sport Management, 15, 275-296.

Meenaghan, T., 1991, The role of sponsorship in the marketing communications mix, Interna tional Journal

of Advertising, 10:1, 35-47.

Meenaghan, T., 1999, Current developments & future directions in sponsorship, International Journal of

Advertising, 17:1, 3-28.

Mishra, D.P., Bobinski Jr., G.S., and Bhabra H.S., 1997, Assessing the economic worth of corporate event sponsorships: a stock market perspective, Journal of Market Fo cused Managemen t, 2, 149-169.

Miyazaki, A.D., and Morgan, A.G., 2001, As sessing Market Value of Event Sponsoring: Corporate Olympic Sponsorships, Journal of Advertising Resea rch , 41:1, 9-15.

Nichols, D., 1988, Do Olympic Tic-Inn’s Really Work?, Incentive, September, 127-130.

Olson, E.T.H.M., 2009, Sponsorship effect metric: assessing the financial value of sponsoring by comparisons to television advertising, Journal of the Academy of Marketing Scien ce, 37:4, 504-515. Park, N.K., 2004, A guide using event study methods in multi -country settings, Strategic Management

Journal, 25, 655-668.

Payne, M., 1998, Ambush Marketing: The Undeserved Advantage, Psychology and Marketing, 15:4, 323-331.

(34)

Samitas, A., Kenourgios D., and Zounis, P., 2008, Athens’ Olympic Games 2004 impact on sponsors’ stock returns, Applied Financial Economics, 18, 1569-1580.

Sandler, D. and Shani, D., 1989, Olympic sponsorship versus ambush marketing: Who gets the gold?,

Journal of Advertising Research , August/September, 9-14.

Serra, A.P., 2002, Event study tests: a brief survey, Working Paper, University of Porto; faculty of economics.

Shani, D., and Sandler, D., 1996, Climbing the sports event pyramid, Sports ma rketing, August 26, 6. Smiths, G., 2004, Brand Image Transfer Through Sponsorship: A Consumer Learning Perspective, Journal

of Marketing Management, 20, 457-474.

The Economist, 2008, Fun, Games and money. A special report on the sports business, The Economist, August.

Tripodi, J.A., and Hirons, M., 2009, Sponsorship Leveraging Case Studies –Sydney 2000 Olympic Games,

Journal of Promotion Managemen t, 15, 118–136.

Verity, J., 2002, Maximizing the marketing potential of sponsorship for global Brands, European Business

Journal, 4:4, 161-173.

Walliser, B., 2003, An international review of sponsorship research: extension and update, International

Journal of Advertising, 22:1 , 5-40.

Internet sources

(35)

Appendix 1

Table 7 Overview of the sponsors included in the sample. For every company, the announcement date, event, home market, industry, nature of the company and type of contract are shown. The sponsors are ordered based on the announcement date.

No.

Company Announcement date Event Home Market Origin Industry B2B/B2C Sponsor

Type 1 Coca Cola 30-1-1998 World Cup Football Germany 2006 United States of America Foreign Consumer Food B2C Partner

2 Heineken 8-2-2001 Summer Olympics Athens 2002 Netherlands Foreign Brewing B2C Partner

3 Alpha Bank 8-2-2001 Summer Olympics Athens 2002 Greece Domestic Banking B2C Partner

4 Toshiba 7-6-2001 World Cup Football Germany 2006 Japan Foreign Consumer electronics B2C Partner

5 Hyundai 3-7-2001 Summer Olympics Athens 2002 South-Korea Foreign Automobile B2C Partner

6 Deutsche Telekom 6-12-2001 World Cup Football Germany 2006 Germany Domestic Telecommunications B2C Partner

7 Ticketmaster 1-3-2002 Summer Olympics Athens 2002 United States of America Foreign Ticket services B2C Partner

8 McDonald's 17-4-2002 European Cup Football Austria-Switzerland 2008 United States of America Foreign Consumer Fast Food B2C Partner

9 Shell Hellas SA 25-7-2002 Summer Olympics Athens 2002 Great-Britain Foreign Petroleum B2C Partner

10 ABB 1-9-2002 Summer Olympics Athens 2002 China Foreign Electric power distribution B2B Partner

11 Alstom 1-9-2002 Summer Olympics Athens 2002 France Foreign Electric power distribution B2B Partner

12 Siemens 1-9-2002 Summer Olympics Athens 2002 India Foreign Electric power distribution B2B Partner

13 Philips 14-11-2002 World Cup Football Germany 2006 Netherlands Foreign Technology products B2C Supporter

14 Hyundai 6-12-2002 World Cup Football Germany 2006 South Korea Foreign Automobile B2C Supporter

15 Continental 21-1-2003 World Cup Football Germany 2006 Germany Domestic Tier development B2C Supporter

16 Fujifilm 5-2-2003 World Cup Football Germany 2006 Japan Foreign Electronic imaging B2C Supporter

17 Altec 1-4-2003 Summer Olympics Athens 2002 Greece Domestic Computer technology B2B Supporter

(36)

19 Intracom 1-4-2003 Summer Olympics Athens 2002 Greece Domestic Computer technology B2B Supplier

20 PC Systems 1-4-2003 Summer Olympics Athens 2002 Greece Domestic Computer technology B2C Supplier

21 Mizuno Corporation 1-7-2003 Summer Olympics Athens 2002 Japan Foreign Sports equipment B2C Supplier

22 Yahoo 16-7-2003 World Cup Football Germany 2006 United States of America Foreign Web technology B2C Supplier

23 PPC S.A. 1-3-2004 Summer Olympics Athens 2002 Greece Domestic Public Power B2B Supplier

24 Adidas 18-3-2004 Summer Olympics Athens 2002 Germany Foreign Sportswear B2C Supplier

25 Volkswagen 10-6-2004 Summer Olympics Beijing 2008 Germany Foreign Automobile B2C Partner

26 China Mobile 21-7-2004 Summer Olympics Beijing 2008 Hong Kong Foreign Telecom services B2C Partner

27 Sinopec 11-10-2004 Summer Olympics Beijing 2008 Hong Kong Foreign Petroleum B2C Partner

28 Bell 18-10-2004 Winter Olympics Vancouver 2010 United States of America Foreign communications B2C Supporter

29 Adidas 24-1-2005 Summer Olympics Beijing 2008 Germany Foreign Sportswear B2C Supporter

30 Rona 3-5-2005 Winter Olympics Vancouver 2010 Canada Domestic Consumer Products B2C Supporter

31 Petro-Canada 5-6-2005 Winter Olympics Vancouver 2010 CAnada Domestic Petroleum B2C Supporter

32 Johnson& Johnson 26-7-2005 Summer Olympics Beijing 2008 United States of America Foreign Consumer Products B2C Supporter

33 Picc 15-9-2005 Summer Olympics Beijing 2008 China Domestic Property and Casualty B2C Supplier

34 Budweiser 27-4-2006 World Cup Football South Africa 2010 Belgium Foreign Brewing B2C Supplier

35 Adidas 28-6-2006 European Cup Football Austria-Switzerland 2008 Germany Foreign Sportswear B2C Supplier

36 FNB 6-7-2006 World Cup Football South Africa 2010 South-Africa Domestic Banking B2C Supplier

37 McDonald's 8-7-2006 World Cup Football South Africa 2010 United States of America Foreign Consumer Fast Food B2C Supplier

38 McDonald's 8-7-2006 World Cup Football Germany 2006 United States of America Foreign Consumer Fast Food B2C Supplier

39 MTN 13-7-2006 World Cup Football South Africa 2010 South-Africa Domestic Telecommunications B2C Supplier

40 Carlsberg 24-8-2006 European Cup Football Austria-Switzerland 2008 Denmark Foreign Brewing B2C Supplier

41 Dow 23-10-2006 Winter Olympics Vancouver 2010 United States of America Foreign Chemicals B2B Supplier

42 Teck 1-12-2006 Winter Olympics Vancouver 2010 Canada Domestic Mining B2B Supplier

43 Ricoh 18-1-2007 Winter Olympics Vancouver 2010 Japan Foreign Automation solutions B2B Supplier

44 Canadian Pacific Railway 23-1-2007 Winter Olympics Vancouver 2010 Canada Domestic Railway B2C Supplier

45 Birks 7-2-2007 Winter Olympics Vancouver 2010 Canada Domestic Diamond jewelry B2C Supplier

46 Jackson-triggs 7-2-2007 Winter Olympics Vancouver 2010 United States of America Foreign Wining B2C Supplier