How do corporate sports sponsorships affect the
value of the sponsoring firm?
An empirical analysis on the effects of firm and sponsorship specific factors in sports
sponsorships on the stock returns of the sponsoring firm.
Cian Garry
February 2009
Master Thesis
University of Groningen
Preface
This thesis marks the ending of my ‘study career’, from which I have spent the final 2,5 years at the University of Groningen. During the past months I have written this thesis as the final graduation project of the specialization finance within the MSc Business Administration. I would like to use this opportunity to give my thanks to my supervisor, for both my bachelor and master theses, prof. dr. L.J.R. Scholtens. The constructive support and advice he gave me during the course have been a great motivation and also of great use in writing my thesis. Furthermore, I would like to thank everyone who has helped and supported me during my ‘study career’.
How do corporate sports sponsorships affect the value of
the sponsoring firm?
An empirical analysis on the effects of firm and sponsorship specific factors in sports
sponsorships on the returns of the sponsoring firm.
Cian Garry
ABSTRACT
JEL Classification: G14, G39, M39
Keywords: Corporate sport sponsorship, firm value, firm and sponsorship specific factors
I.
Introduction
IN RECENT YEARS SPORTS MARKETING AND SPONSORSHIPS have evolved significantly. Consequently, a large number of firms have increased their spending on sports sponsorships. In fact, IEG1 expects sponsorship spending by European firms to increase by 10.4% in 2008 adding up to a total of $11.7 billion. On average, this accounts for 15% of European firms’ total spending on marketing activities (Lester 2005). The majority of these sponsorship expenses are invested in sports sponsorship activities (Thjømøe et al. 2002), making sports sponsorships an important and valuable aspect of many firms’ marketing and communication strategies (Javalgi et al. 1994). Moreover, several authors (i.e. Cornwell et al. (2001), and Farrelly et al. (2006)) argue that
1
This paper examines the effect of sports sponsorship announcements on the value of the sponsoring firm through multiple regression analyses for the period 1997-2008. In
these analyses the effect of both firm and sponsorship specific factors on firm value, measured by the sponsoring firms’ stock returns, are taken into account for European
and US firms. The effect of these factors varies across EU and US firms and also across industries. Especially firm size and a good fit between the sponsoring firm and
eventually sponsorships can become a source of competitive advantage to the sponsoring firm. On the other hand, for sports entities the benefits (right fees) that arise from sponsorships are an important source of income (Farrelly et al. 2006). With sports clubs developing into professional organizations, clubs need to be creative in generating adequate funding. For example, sponsorship revenues consist of approximately 25% of total income for professional soccer clubs in Europe, whereas in European basketball and rugby this percentage is higher (Andreff and Staudohar 2000). Pruitt et al. (2004) add to this that for teams in the National Association for Stock Car Auto Racing (NASCAR) sponsorship income is their lifeblood.
In general, sponsorship can be defined as ‘the provision of assistance either financial or in kind to an activity by a commercial organization for the purpose of achieving commercial objectives’ (Meenaghan 1983, p. 10). In the case of a sports sponsorship firms thus provide a fee to a specific sports entity such as, a sports club (e.g. a soccer club), sports association (e.g. national soccer association or Olympic Committee), a sporting event/tournament (e.g. a world cup) or a sports stadium. The main objectives of firms’ sponsorship activities nowadays are, improving image, brand awareness, and sales (Mishra et al. (1997), Grohs et al. (2004)). Whereas in the past media coverage was the main objective for many firms (Grohs et al. 2004). Important however, is that if a sponsorship is to add value to the firm the benefits of the sponsorship should at some point exceed the initial investment. Clark et al. (2002) argue that a sponsorship should increase sales which in turn should enhance the sponsoring firms’ stock price. In addition, Gwinner (1997) found that with the increasing amounts invested in sports sponsorships firms do expect a reasonable return on these investments. Whereas in early days sponsorships were mainly seen as ‘charitable donations'.
the sponsoring firm will serve as the dependent variable to measure the effect on the value of both European and US firms. As independent variables I will include firm specific (including firm size and firm age) and sponsorship specific (including contract length and the sponsor-entity fit) factors. Finally, with the results of this study a comparison can be made whether differences exist in the effects of a sports sponsorship on the market value of European and US firms. The remainder of this paper is organized as follows: Section II discusses prior studies on this topic, the methodology and data is discussed in sections III and IV. Next, section V presents and discusses the results. Finally, section VI contains the main conclusions and implications.
II.
Literature review
In the literature on sponsorships Meenaghans’ (1983) definition of a sponsorship is generally accepted. Using this definition a sponsorship can be defined as ‘the provision of assistance either financial or in kind to an activity by a commercial organization for the purpose of achieving commercial objectives’ (Meenaghan 1983, p. 10). Cornwell et al. (2005) add an important aspect to this definition by expressing that a sponsorship involves paying a fee in advance in exchange for potential future communication values. This supports the findings of Speed and Thompson (2000) that it can take years before a firm generates value from a sports sponsorship. Compared to regular
advertising sponsorships have a number of major differences. First, related to the aspect that a
In recent year’s firms’ sponsorship objectives have shifted towards general matters like improving brand awareness and company image (Grohs et al. (2004), Farrelly et al. (2006)). In fact, Farrelly et al. (2006) find that sponsors use the image and characteristics of the sports entity to enhance or change their own image. Because of this, sponsorship activities now are an integral and valuable part of a large number of firms’ strategic communication (Javalgi et al. (1994), Farrelly et al. (2006)). Additionally, in many cases sponsorships provide firms with the opportunity of corporate hospitality in the public relation function (Meenaghan 1991).
With the increasing amount of money spent on sponsorships2, research has turned its attention to the effects of sponsorships on firm value. The main question to be asked from an investors’ point of view is: does a sports sponsorship add value to the sponsoring firm? If a sports sponsorship is to add value to the firm the benefits of it should, at some point, exceed the initial investment. Clark et al. (2002) summarize this by stating that a sponsorship ‘should enhance sales over some economically relevant time horizon, these expected increases in sales are subsequently reflected in higher valuations of the sponsoring firms’ stocks’. With firms becoming more professional in their approach towards sponsorships, a firms’ management will thus expect a return on their sponsorship investments
(Thjømøe et al. (2002), Gwinner (1997)). A number of studies that have addressed this issue (Clark et al. (2002), Cornwell et al. (2005), Mishra et al. (1997), Miyazaki and Morgan (2001), Pruitt et al. (2004), Tsiotsou and Lalountas (2005)) show that sport sponsorships have a positive effect on firm value. In all studies the effect on firm value is measured through the event-study methodology
following the first public announcement of the sponsorship deal by the sponsoring firm3. These studies all use the first announcement of a sponsorship since at this point the information is new to the market. Consequently, in an efficient market the sponsorship investment will be directly reflected in the sponsoring firms’ stock price following the announcement (MacKinlay 1997). Therefore the effect of a sponsorship announcement on firm value can be measured through the abnormal stock returns. A striking finding on the results of the event studies is the small sample size in all studies. In addition, since daily stock returns were used in all studies the results might suffer from problems with non-normality. As noted by Brown and Warner (1985), daily stock returns exhibit substantial departures from normality. A second remarkable finding is that all studies focus on the US market (i.e. the effect of sponsorship announcements on the market value of US based firms) and no study has addressed the European market. Despite the fact that the European market is the second largest market in sports sponsorships4. Thus, little is known about the effect of sports sponsorships on the value of European firms.
2 Lester (2005) expects sponsorship spending by European firms to double in the period 2006-2010. 3 Table IX in the appendix provides a detailed analysis of the applied methodology and results of these event studies.
Table I
Summary of previous studies conducting regression analyses to assess the effect of sports sponsorship announcements on the value of the sponsoring firm.
*CAR = Cumulative Abnormal Return, AR= Abnormal return.
Study Period Model (Market index) Dependent variable*
Independent variables Results
Clark et al. (2002)
1985-2000 - Market model, (CRSP Value-Weighted index) - OLS regression
CAR 3-day event window ( t=-1, to t=+1
Firm type, firm size, fee paid, contract length, local firm, population, sports league.
The factors high tech, local firm, and contract length are significantly positive related to firm value. Firm size is negatively related. Cornwell et al. (2005) 1990-2003 - Market model, (CRSP Value-Weighted index) - OLS regression
CAR 21-day event window (t=-10, to t=+10)
Market share, market value, agency problem, high tech (industry), congruence (fit), sports league.
High tech firms and a good fit increase the sponsoring firms’ stock returns following a sponsorship announcement. Market share is significantly negative related to firm value.
Mishra et al. (1997) 1986-1995 - Market model, (CRSP Value-Weighted index) - OLS regression AR on the event date (t=0)
Firm size, advertising expenditure, return on assets.
Return on assets is significantly positive related to firm value at a 10% confidence level. Pruitt et al. (2004) 1995-2000 - Market model, (CRSP Value-Weighted index) - OLS regression
CAR 2-day event window ( t=-1 and t=0)
Firm size, agency problems, fit, results sports entity, sponsoring type.
Firm specific factors
Firm size
Only the study of Clark et al. (2002) has found a significant relation between firm value and firm size. The theoretical explanation for the relation between firm value and firm size states that: the percentage value of a sponsorship investment declines as firm size (measured by the market value of equity) increases. Consequently any benefit of a sponsorship represents a relatively smaller percentage of firm value as firm size increases (Clark et al. (2002), Pruitt et al. (2004), Cornwell et al. (2005)). In the studies of Mishra et al. (1997), Pruitt et al. (2004), and Cornwell et al. (2005) the variable was included but no significant relation was found. Mishra et al. (1997) argue that there is no significant relation between firm size and firm value because the majority of firms in their sample are large multinational firms. Qualitative results by Thjømøe et al. (2002) support this explanation, as this study shows that large firms are more active in sports sponsorships compared to smaller firms. Based on the preceding discussion I suggest the following hypothesis:
:
0
H The abnormal stock returns of the sponsoring firm are not related to firm size.
:
1
a
H The abnormal stock returns of the sponsoring firm are negatively related to firm size.
Market share
In relation to the factor market share, Cornwell et al. (2005) use Weber’s Law (Miller 1962) as an explanation for their results. Weber’s law states that stimulus changes which are expected to make a noticeable difference are a constant proportion of the initial starting point from the stimulus. Thus related to market share, firms with a large market share announcing a sports sponsorship will be expected to gain less from the sponsorship since the starting point for comparison will be the large market share. Because of data constraints this factor will not be further examined in this study.
Industry
future. Meenaghan (1983) and Javalgi et al. (1994) show more general results. Based on a survey they conclude that image transfer through a sponsorship differs across firms and industries. Therefore, I will not only examine whether high tech firms gain more from sponsorships but also include other industries. This is done by dividing my sample into portfolios based on the industry in which the firms are active. The preceding discussion suggests a second hypothesis:
:
0
H The abnormal stock returns of the sponsoring firm are equal across industries.
:
2
a
H The abnormal stock returns of the sponsoring firm differ across industries.
Sponsorship decision
A common problem in sports sponsorship decisions is that managers prefer to invest in a sports entity of their own preference (i.e. favorite sports club) rather than investing in a sponsorship opportunity that maximizes firm value (Pruitt et al. 2004). Qualitative results of Gwinner (1997) and Thjømøe et al. (2002) also indicate a conflict of interest in sponsorship decisions. Gwinner (1997) finds that the sponsorship decision is often influenced by the managers’ current preferences. The study of Thjømøe et al. (2002) shows that, in many cases, sponsorship choices are based on personal connections of the management. As a consequence these sponsorship decisions are not always in line with the firms’ ‘official’ communication strategy. Pruitt et al. (2004) use the agency problem theory as the main explanation for this conflict of interest. Agency problems exist whenever managers (who are non-owners of the firm) prefer their own welfare and interests above those of shareholders (Jensen and Meckling 1976). In the case of sports sponsorships the agency problem can thus be specified to a conflict of interest between the manager of the firm and investors. This occurs especially in firms which generate a substantial free cash flow (Jensen 1986). In addition, Clark et al. (2002) and
Cornwell et al. (2005) note that sports sponsorships carry significant potential for agency conflicts (i.e. free tickets for big sporting events or free sky-box seating). Based on these findings it can be argued that managers’ preferences have a negative impact on the generated value from a sports sponsorship, this should especially be present in firms with a large free cash flow. Therefore, I expect the relation between the stock returns of the sponsoring firm and the measure for this factor (measured as the cash flow per share) to be negative. Thus,
:
0
H The abnormal stock returns of the sponsoring firm are not related to the sponsorship decision.
:
3
a
Local firm
Results by Clark et al. (2002) indicate that ‘local sponsorships’ increase stock returns in comparison with ‘non-local sponsorships’. Clark et al. (2002) explain this finding as a sort of “home field advantage” of the firm. A sponsorship is labeled as ‘local’ when a firm sponsors a sports entity in the same province or state as the headquarters of the sponsoring firm. Results from a survey by Meenaghan (2001) support the excess returns of local sponsorships, as it shows that the respondents rate local sponsorships higher than sponsorships by an ’outside’ firm. The main argument is that local sponsorships generate more goodwill. Thus, in relation to firm value I expect local sponsorships to have a positive effect on the stock returns of the sponsoring firm:
:
0
H The abnormal stock returns of the sponsoring firm are not related to local sponsorships.
:
4
a
H The abnormal stock returns of the sponsoring firm are positively related to local sponsorships.
Firm age
Using the explanation of Clark et al. (2002) and Cornwell et al. (2005) for the higher returns of high tech firms I will add another firm specific factor namely: the age of the sponsoring firm. Up till now no previous study has examined this factor. The constructed theory states that: high tech firms, which in general, are relatively small and young use sponsorships as a tool to increase brand
awareness and profile themselves. Therefore it could be argued that young firms benefit more from a sports sponsorship compared to older and established firms. Thus, I expect a negative relation with firm value:
:
0
H The abnormal stock returns of the sponsoring firm are not related to firm age.
:
5
a
H The abnormal stock returns of the sponsoring firm are negatively related to firm age.
Sponsorship specific factors
Fee paid
whether a sports sponsorship is successful. It is argued that a larger fee contributes to brand perception and image (Cornwell et al. 2001). Alternatively, it could also be argued that the fee paid is negatively related to firm value: with a large initial investment a firm needs to generate more benefits from a sponsorship before it is value enhancing for the sponsoring firm. Only for a limited number of
sponsorships in this study the fee was disclosed. Therefore I will not examine this factor since this will not provide representative results.
Contract length
Long term sponsorships are valued more positively than short term deals (Clark et al. 2002). Here again, the signaling theory can be used as an argument. By entering into a long-term sponsorship the sponsoring firms’ management can signal their confidence in future profitability. Investors’ preference for long-term sponsorships is also supported by Speed and Thompson (2000), who find that it can take several years before the sponsoring firm generates value from a sponsorship. Amis et al. (1999) and Cornwell et al. (2001) support this by arguing that a long-term relationship can eventually lead to a competitive advantage for the sponsoring firm. Thus:
:
0
H The abnormal stock returns of the sponsoring firm are not related to contract length.
:
6
a
H The abnormal stock returns of the sponsoring firm are positively related to contract length.
Sponsor –Entity fit (congruence)
As shown by Pruitt et al. (2004) and Cornwell et al. (2005) a good fit between the sponsoring firm and the sponsored sports entity has a positive effect on firm value. In addition, qualitative research by Grohs et al. (2004) shows that sponsor-entity fit is a crucial factor in generating sponsor awareness, since in many cases the image of a sponsoring firm is associated with the characteristics and image of the sponsored event (Keller 1993). Cornwell et al. (2005) and Pruitt et al. (2004) label a sponsorship deal as a ‘good fit’ when the sponsoring firm has a direct tie to the sponsored sport (i.e. an automotive brand sponsoring a racing team). Cornwell et al. (2005) present two competing theories regarding congruence. First, congruence suggests that the ability of people to remember information is influenced by factors such as similarity or relatedness and when it is congruent with prior
expectations. Contrary to this, it is argued that incongruent information results in greater recall of sponsors since incongruence requires more elaborate processing by a person. However, Cornwell et al. (2005) add to this that for the latter theory no support has been found. Therefore, I suggest the
following hypotheses:
:
0
H The abnormal stock returns of the sponsoring firm are not related to sponsor-entity fit.
:
7
a
Sponsoring type
The factor sponsoring type indicates whether a sponsorship is engaged by a single brand or division of a firm or by the total firm (i.e. the total firm name is used primarily in the sponsorship). Pruitt et al. (2004) show that ‘total firm’ sponsorships increase the sponsoring firms’ stock returns compared to those involving a single division. Meenaghan (1991) argues that this is caused by the fact that a sponsorship can not carry a detailed product message. Consequently, using a sponsorship for general targets like brand and corporate image is seen as more effective and valuable. Thus:
:
0
H The abnormal stock returns of the sponsoring firm are not related the type of sponsoring.
:
8
a
H The abnormal stock returns of the sponsoring firm are positively related the type of sponsoring.
Contrary to most previous studies I will focus my research on the European market, since little is known about the effect of sports sponsorships on the value of European firms. However, I will also include US firms to serve as a peer group for the European firms. Furthermore, I will apply a larger sample compared to previous work which should provide more reliable results. In line with the study of Mishra et al. (1997) I will use a sample which consists of a wide variety of events. This is in contrast with most other studies on this topic where one specific type of sponsorship is examined (i.e. sponsoring of sports stadiums). Consequently, this study will provide investors with a clear guideline to value sports sponsorships that are undertaken by firms. With the characteristics of a new
sponsorship investors can assess whether this will be beneficial to the firm. In addition, it should provide firms with an overview of those factors which are important in making a successful sponsorship decision.
Table II
Firm specific factors in relation with the stock returns of the sponsoring firm, following a sports sponsorship announcement.
Factor Effect Hypothesis
Firm size (measured as the market value of the firm)
Negatively related to stock returns. Since the percentage value of the initial sponsorship investment declines when firm size increases (Clark et al. 2002). Mishra et al. (1997), Pruitt et al. (2004), and Cornwell et al. (2005) did not find evidence supporting the firm size hypothesis.
Negatively related to firm value.
Market share Investors perceive firms with a relatively small market share to gain more from a sports sponsorship compared to firms already holding a large market share (Cornwell et al. 2005).
Not examined.
Industry Results of Clark et al. (2002) and Cornwell et al. (2005) showed that high tech firms (firms in the computer and internet business) gain significantly more from sports sponsorships compared to companies in more traditional industries.
The effect of sports sponsorships on firm value differs across industries.
Sponsorship decision In the decision making process of sports sponsorships agency problems are present, especially in firms with a large free cash flow (Pruitt et al. 2004). Cornwell et al. (2005) did not find evidence for agency problems in relation with sports sponsorships.
Negatively related to firm value.
Local firm Clark et al (2002) show a positive relation between stock returns and
sponsorships announced by local firms (thus the firm sponsoring a local sports entity).
Positively related to firm value.
Firm age High tech firms, which are characterized by being relatively small and young, Generate more from sponsorships compared to firms in more established industries. Especially young firms use sponsorships to increase brand awareness (Clark et al. (2002) and Cornwell et al. (2005)).
Table III
Sponsorship specific factors in relation with the stock returns of the sponsoring firm, following a sports sponsorship announcement.
Factor Effect Hypothesis
Fee paid The relative fee paid by firms has no significant impact on firm value (Clark et al. (2002). In other studies (i.e. Cornwell et al. (2005)) this characteristic was not taken into account due to data constraints.
Not examined.
Contract length Contract length is positively related to firm value. Investors value long-term contracts more positive than short-term contracts (Clark et al. (2002).
Positively related to firm value.
Sponsor –Entity fit (congruency)
Sponsorships where a good fit exists between the sponsoring firm and sports entity are more valuable to the sponsoring firm compared to sponsorships deals without a good fit (Pruitt et al. (2004), Cornwell et al. (2005)).
Positively related to firm value.
Sponsoring type Sponsorships that cover a total firm are judged more favorably by investors compared to sponsorships involving a single brand or product. A ‘total firm sponsorship’ can thus be regarded as more value enhancing Pruitt et al. (2004).
III. Methodology
I will examine how a sports sponsorship affects firm value from an investors' perspective. The effect of a sports sponsorship announcement on firm value will be measured by the stock returns of the sponsoring firm. In order to assess this topic I have formulated two main research questions: Which type of firm maximizes firm value through a sports sponsorship? And, what are the characteristics of a value maximizing sports sponsorship? I will answer these research questions by conducting a number of cross-sectional regression analyses using the Ordinary Least Squares (OLS) methodology. This is in line with previous studies on this topic which all use the OLS model. In the analyses the firm specific and sponsorship specific factors will be analyzed separately. In both cases the cumulative abnormal stock returns following a sports sponsorship announcement will serve as the dependent variable. The firm and sponsorship specific factors are specified as the independent variables.
I will use daily closing prices to calculate returns. The stock prices of the sponsoring firms are retrieved from Yahoo Finance5 and Datastream and transformed into returns using:
1 1 − − − = it it it it S S S R (1)
Where Rit is the return of security i at time t, Sit is the price of security i at time t, and Sit−1 is the price of security i at time t -1. I will use the simple return formula instead of continuously
compounded returns, which are generally used in finance studies. Continuously compounded returns are not additive across a portfolio when the weighted average of a portfolio is calculated (Brooks 2002). Since I will use industry based portfolios in this study I therefore use simple returns.
Abnormal returns are calculated using the event study methodology. Market efficiency is one of the main assumptions in event studies, which implies that all information available to the market is reflected in a firms’ stock price. Thus, the effects of an event will also be immediately reflected in a firms’ stock price (MacKinlay 1997). The presence of any systematic abnormal returns would thus violate the assumption of market efficiency since only normal returns are expected (Grinblatt and Titman 2002). Brown and Warner (1980) define the abnormal returns of a security as ‘the difference between its actual ex-post return and that which is predicted under the assumed return-generating process’. Thus, abnormal returns are present when the securities’ ex-post return in the event window deviates from the assumed return which is calculated through the estimation window. Consistent with the previous studies on sports sponsorship I will use the market model to calculate abnormal returns. In the market model the returns of a security are related to the return on a relevant market index, both market wide factors and the securities’ systematic risk are taken into account. The model, which is estimated through the OLS methodology, has the following linear form:
it t m i i i
R
R
τ=
α
+
β
,+
ε
^ ^ (2) Where Riτ is the return of security i at time t, Rm,t is the return on the market at time t, andε
it is thezero-mean disturbance term. i
^
α
and i ^β
are the parameters of the model and are estimated using:m i i i ^ ^ ^
µ
β
µ
α
=
−
(3)∑
∑
+ = + =−
−
−
=
1 0 1 1 2 ^ 1 0 ^ ^ ^)
(
)
)(
(
T T m m T T m m i i iR
R
R
τ τ τ τ τµ
µ
µ
β
(4)With equation 2 abnormal returns can be calculated using:
t m i i i i
R
R
AR
, ^ ^β
α
τ τ=
−
−
(5)Where ARiτ is the abnormal return of security i at time t. In calculating abnormal returns I use an estimation period of 150 days ranging from t = -160 to t = -10 relative to the event date (t = 0). This is in line with most previous studies on this topic, where the estimation windows vary from 126 days to 250 days. According to MacKinlay (1997) a large estimation window is necessary because of the variance of the abnormal returns. This variance consists of two components: the disturbance variance, and the additional variance as a consequence of the sampling error in
α
i andβ
i which is common for all event window observations. Using a large estimation window will cause the contribution of the sampling error to the variance to become zero (MacKinlay 1997). I use the S&P 500 and S&P Europe 350 as the market indexes for the US and EU market respectively. Both indexes consist of a large number of firms which minimizes the effect of an event (i.e. the sponsoring firms’ returns) on the performance of the index6. In addition, the S&P Europe 350 covers all markets from which the EU firms in the sample are included.6
The maximum firm weight in the index is 2.48% for the S&P Europe 350 index and 3.96% for the S&P 500. (source: http://www2.standardandpoors.com/spf/pdf/index/SP_Europe_350_Factsheet.pdf and
In order to draw inferences for event windows larger than 1 day it is necessary to use cumulative abnormal returns (CAR), which are calculated using:
∑
==
2 1)
,
(
1 2 τ τ τ ττ
τ
i iAR
CAR
(6)∑
==
N i iN
CAR
1 2 1 2 2 2 1(
,
)
1
))
,
(
var(
τ
τ
σ
τ
τ
(7)Corresponding with the studies of Mishra et al. (1997), Clark et al. (2002), and Pruitt et al.(2004) I use an event window of 3 days (ranging from t = -1 to t = +1). In all three studies the effect of a
sponsorship announcement was mainly present on the event date and the days surrounding the event. In some cases, a sponsorship deal was leaked before the official announcement was made which could affect the returns in the estimation window. Therefore, in line with Mishra et al. (1997) and Cornwell et al. (2005), I include a number of days (t = -9 to t = -2) between the estimation period and the event window to minimize any possible effects of the sponsorship announcements in the estimation period.
Regression analyses
As a next step I will perform multiple regression analyses, using the OLS model, to examine the effect of the firm and sponsorship specific factors on the market value of the firm. I will analyze the firm and sponsorship specific factors separately to provide clear answers on the two main research questions. In the regression analyses the total, EU, and US samples will be examined. Corresponding with previous studies on this topic the regression containing firm specific factors has the following general form: it i i i i i
i
SIZE
DECISION
LOCAL
AGE
CAR
τ=
α
+
β
1+
β
2+
β
3+
β
4+
ε
(8)Where SIZE is the market value of the sponsoring firm, DECISION is a measure of potential agency problems in the sponsorship decision, LOCAL is a dummy variable indicating whether the firm is sponsoring a local sports entity, and AGE is the age (in years) of the sponsoring firm. Finally,
τ
i
The effect of the sponsorship specific factors on firm value is measured using: it i i i i
i
LENGTH
FIT
TYPE
CAR
τ=
α
+
β
1+
β
2+
β
3+
ε
(9)Where, LENGTH is the length (in years) of a sponsorship deal, FIT is a dummy variable which indicating whether a fit exists between the sponsoring firm and the sports entity (i.e. the sponsoring firm is directly related to the sport or the product is likely to be used while attending or watching a sports event), and TYPE is a dummy variable indicating whether the sponsorship covers the total firm or just a single product, brand, or division of the sponsoring firm. Only for 12 events the fee paid in the sponsorship deal was announced, because of this I have excluded this factor since using only 12 observations will not provide representative results.
The hypothesis regarding the factor industry is tested by forming industry-based portfolios of the firms in the total sample. I test these portfolios using:
it i i i i i
i
SIZE
DECISION
LOCAL
AGE
CAR
τ=
α
+
β
1+
β
2+
β
3+
β
4+
ε
(10) it i i i ii
LENGTH
FIT
TYPE
CAR
τ=
α
+
β
1+
β
2+
β
3+
ε
(11)In this case, the dependent variable is the CAR of the relevant portfolio. The industry based portfolios are only examined for the total sample (thus all EU and US firms) to ensure the portfolios are
sufficiently large, which should provide reliable results. Finally, I will test whether extensions of sponsorships have a different effect on firm value compared to the announcements of ‘new’
sponsorships. In this case an event is labeled as an ‘extension’ when the sponsoring firm announces to extend an existing or expiring sports sponsorship. In other studies, on the topic of sports sponsorships, the effect of extensions on firm value is not examined. To assess the effect of sponsorship extensions I will also run two regression analyses using a portfolio of all sponsorship extensions:
it i i i i i
i
SIZE
DECISION
LOCAL
AGE
CAR
τ=
α
+
β
1+
β
2+
β
3+
β
4+
ε
(12) it i i i ii
LENGTH
FIT
TYPE
CAR
τ=
α
+
β
1+
β
2+
β
3+
ε
(13)Corresponding with the studies of Mishra et al. (1997), Clark et al. (2002), Pruitt et al. (2004), and Cornwell et al. (2005) the betas of all regression equations (equations 8-13) will be tested for significance using the following test statistic:
^ * ^
)
(
i i iSE
t
β
β
β
−
=
(14) Where i ^β
is the observed value,β
i* is the value ofβ
under the null hypothesis, and ^ ) ( i SEβ
is the standard error of i ^β
.Following the regression analyses I will perform a number of additional tests in order to assess whether the equations (8-11) satisfy the assumptions of the OLS-model. First, I will test whether the error-term in the equations (
ε
it) is normally distributed, therefore, I apply the Bera-Jarque test. However, Brooks (2002) notes that when using a sufficiently large sample the influence of non-normality should have virtually no consequences on the results. Second, I will test the error terms for any patterns of heteroscedasticity using an ARCH test. Since the ARCH test also accounts for ‘volatility pooling’ and ‘volatility clustering’ I prefer this test over White’s test. If any ARCH effects are present in my sample I will use the GARCH-model instead. The main difference, between White’s test and the ARCH test, is that the GARCH-model does not have the assumption of a constant variance in the error term. Finally, by applying the Durbin-Watson test I will test for any autocorrelation in the error terms.As a final test I will examine whether, for the firm and sponsorship specific factors, any significant differences are present between the EU and US samples. This will be done by comparing the betas of these samples for all factors using an independent samples t-test:
s s r r s r
n
s
n
s
x
x
t
2 2+
−
=
(15)IV. Data
The constructed sample consists of 178 events selected from the period 1997-20087.
Corresponding to previous studies on this topic (i.e. Cornwell et al. (2005) and Mishra et al. (1997)) an event is defined as the press release of the sponsoring firm announcing that they are entering into a sports sponsorship or extending an existing or expiring sponsorship. The date on which these press releases were made is defined as the event date (t=0)8. Events are selected using the news archives on (sports) sponsorships of IEG’s Sponsorship.com, SportBusiness and Sports-City9, these archives list major sponsorship announcements made by firms throughout the world. All announcements made by publicly listed firms that are headquartered in either Europe or the US are selected. In line with the methodology of Mishra et al. (1997) sponsorship deals with only local exposure are excluded, resulting in a sample that consists of sponsorships with a national or international character.
Furthermore, the sample consists of a wide variety of sports entities, whereas most other studies (i.e. Clark et al. (2002) and Pruitt et al. (2004)) focus on one specific type of sports sponsorship. This should provide more general results on sports sponsorships and lead to a representative answer on the main research questions. From the initial sample of 215 events, 16 events are excluded due to data unavailability, 16 events are excluded to control for event clustering, and 5 US events are excluded to avoid effects of 9/11 on the returns. Table IV presents descriptive statistics of the final sample. In addition, descriptive statistics of the CAR of the events are presented in table V.
7 A detailed overview of all events is provided in appendix A.
Table IV
Descriptive statistics of the events in the total, EU, and US samples10, and the distribution of the events across industries11. The row ‘extensions relates to the number of events where an extension of the sponsorship was announced. These extensions are also included in the total, EU, and US samples.
Number of events
Total EU US
Events 178 96 82
Firms 105 54 51
Number of events per firm
Total EU US 1 64 30 34 2 20 12 8 3 12 7 5 4 7 4 3 5 2 1 1
Distribution of EU firms & events
Firms Events Austria 1 4 Denmark 1 3 France 6 8 Germany 15 24 Italy 2 2 Netherlands 8 17 Spain 4 6 Sweden 2 2 Switzerland 1 2 United Kingdom 14 28 Portfolios Total EU US Agro-Food 38 16 22 Finance 45 32 13 IT & Telecom 25 12 13 Leisure 18 13 5
Oil & Chemicals 11 5 6
Services & Health 18 8 10
Transport & Equipment 23 10 13
Extension 26 12 14
10 A majority of the firms are multinationals with operations and stock listings in both Europe and the US. The distribution of firms is based on the country where the firms are headquartered.
Table V
Descriptive statistics of the events’ CAR and the CAR of the constructed portfolios. In all cases the CAR is the 3-day event window ranging from t = -1 to t = +1 relative to the event date.
CAR N Mean Median Minimum Maximum St. Dev. Skewness Kurtosis
Sample Total 178 -0.0020 -0.0028 -0.2015 0.0856 0.0359 0.122 5.701 EU 96 -0.0094 -0.0073 -0.2015 0.0745 0.0391 0.190 5.121 US 82 0.0067 0.0048 -0.0722 0.0856 0.0295 0.043 6.381 Portfolios Agro-Food 38 0.0017 0.0025 -0.0407 0.0461 0.0222 0.395 5.078 Finance 45 -0.0087 -0.0052 -0.2015 0.0745 0.0477 0.085 4.080 IT & Telecom 25 -0.0085 -0.0136 -0.0722 0.0856 0.0414 -0.078 5.085 Leisure 18 -0.0067 -0.0055 -0.0850 0.0511 0.0321 0.459 4.878 Oil & Chemicals 11 0.0108 0.0047 -0.0411 0.0810 0.0366 0.209 3.588 Services & Health 18 0.0144 0.0175 -0.0411 0.0810 0.0352 0.253 3.206 Transport & Equipment 23 0.0058 0.0060 -0.0679 0.0656 0.0302 -0.225 9.382 Extension 26 0.0032 0.0053 -0.0529 0.0625 0.0239 0.3600 4.4407
sponsoring firms12. This is in line with the studies of Clark et al. (2002), who use the free cash flow of the firm, and Pruitt et al. (2004) who use the cash flow per share. Both studies argue that the free cash flow is a relevant proxy for measuring agency problems. By dividing the free cash flow by the number of outstanding shares (on day t = -10 relative to the event date) I calculate the free cash flow per share. Data on the firms’ free cash flow and outstanding shares is collected from Datastream. Finally, firm age (measured in years) is retrieved from Yahoo Finance13. Table VI presents descriptive statistics on the firm and sponsorship specific factors.
12 The free cash flow of the firm provided by Datastream is calculated by subtracting the costs of operating activities from the net cash flow of the firm. Since the data is only available on an annual basis I use the most recent figure prior to the event date.
Table VI
Descriptive statistics of the firm and sponsorship specific factors. The factor ‘size’ is measured in millions of Euro’s. ‘Decision’ represents the free-cash flow per share for each event. The factors age
and length are both measured in years. The factors ‘local’, ‘fit’ and ‘type’ show how the events are distributed for these dummy variables14.
Average Median Minimum Maximum St. Dev.
Size Total 41159 26515 207 177829 42075 EU 38046 26343 207 144735 37059 US 44803 26620 450 177829 47252 Decision Total 0.00487 0.00296 -0.02275 0.05458 0.00783 EU 0.00583 0.00292 -0.02275 0.05458 0.01003 US 0.00375 0.00296 -0.01332 0.01731 0.00368 Age Total 65.5 43.0 3.0 281.0 59.6 EU 65.7 35.5 6.0 281.0 66.9 US 65.3 45.0 3.0 206.0 50.2 Length Total 3.8 3.0 1.0 25.0 2.9 EU 3.4 3.0 1.0 10.0 2.0 US 4.4 3.5 1.0 25.0 3.7 Dummy variables N N-0 N-1 Local Total 178 106 72 EU 96 69 27 US 82 37 45 Fit Total 178 128 50 EU 96 70 26 US 82 58 24 Type Total 178 53 125 EU 96 28 68 US 82 25 57
V.
Results
In this section I analyze the results of the regression analyses on the firm and sponsorship specific factors. In general the results of this study differ from the results of most previous studies. For the total and US samples these main differences seem to be caused by the applied sample size and the variety of the sample. Furthermore, the results show that the effect of both the firm and sponsorship specific factors differs for EU and US firms. First, I will discuss the results of the firm specific factors, followed by the results of the sponsorship specific factors.
Firm specific factors
The results of the regression analyses on the firm specific factors size, decision, local, and age are presented in table VII. For the total, EU, and US samples, as for all portfolios, the betas for the firm specific factors are given together with their corresponding level of significance (p-value). Furthermore, additional statistics on all regressions are presented including: the adjusted R2, the test statistics of the Bera-Jarque and Durbin-Watson tests, and the significance level of the ARCH tests. The additional statistics all relate to the regression analysis named in the first column of table VII.
Firm size
Firm size is expected to be negatively related to firm value following a sports sponsorship announcement. In general, this hypothesis is confirmed by the results. For the total, EU, and US samples a negative relation is found which is significant for both the total and US samples. The extension portfolio is the only portfolio which shows a significant relation between firm size and firm value, again the sign of the beta is negative. For the industry based portfolios the majority of
industries also show a negative beta, however, for none of the portfolios a significant relation was found. A general remark concerning firm size is the small size of all betas, indicating that this factor has a limited effect on firm value following a sports sponsorship announcement. As noted before the results of this study confirm the hypothesis that firm size is negatively related to firm value. This is in line with the results of Clark et al. (2002) who argue that; for large (US) firms sponsorship spending represents a relatively smaller percentage value of firm value, which in turn has the effect that investments in sports sponsorships have more impact on the firm value of small firms. Nevertheless, contrary to the findings of this study and the study of Clark et al. (2002), results of Mishra et al. (1997) and Thjømøe et al. (2002) show that especially large multinational firms are active in sports
Table VII
Results of the regression analyses on the firm specific factors using: CARiτ =αi+β1iSIZE+β2iDECISION+β3iLOCAL+β4iAGE+εit
15
. For α and all β’s the
p-values are given in parentheses. The Bera-Jarque and Durbin-Watson statistics are the actual test statistics, whereas for the ARCH test the level of significance is given. These additional statistics all relate to the regression for the CAR of the sample/portfolio named in the first column.
CAR N α β1 β2 β3 β4 Adj. R
2
Bera-Jarque ARCH Durbin-Watson
Sample
Total 178 -0.004 -7.8E-08 0.260 0.002 4.3E-05 0.017 471.902** 0.940 1.947
(0.567) (0.078)* (0.430) (0.675) (0.621) EU 96 -0.008 -5.8E-08 0.480 -0.005 -7.0E-06 0.023 236.804** 0.949 2.068 (0.314) (0.457) (0.230) (0.492) (0.951) US 82 0.005 -1.1E-07 -0.208 0.000 1.1E-04 0.018 2.861 0.075* 2.085 (0.621) (0.058)* (0.752) (0.964) (0.146) Portfolios
Agro-Food 38 0.004 -1.1E-07 0.958 -0.012 2.5E-05 0.044 0.461 0.531 1.535
(0.670) (0.451) (0.323) (0.192) (0.738)
Finance 45 -0.013 6.8E-08 0.373 0.002 -4.0E-05 -0.084 67.201** 0.898 1.887
(0.478) (0.619) (0.513) (0.869) (0.803)
IT & Telecom 25 -0.009 -8.8E-08 -0.776 0.015 -1.4E-04 -0.130 0.279 0.970 2.205
(0.768) (0.566) (0.839) (0.434) (0.754)
Leisure 18 -0.042 1.1E-06 0.122 0.034 5.5E-04 0.281 0.356 0.820 1.909
(0.035)** (0.100) (0.954) (0.041)** (0.197)
Oil & Chemicals 11 -0.024 3.7E-08 3.070 -0.002 2.9E-04 -0.245 3.552 1.909
(0.307) (0.910) (0.402) (0.915) (0.302)
Services & Health 18 0.005 -1.9E-07 0.835 0.007 4.6E-05 0.031 1.245 0.211 1.850
(0.667) (0.133) (0.643) (0.691) (0.621)
Transport & Equipment 23 0.012 -2.6E-07 0.054 -0.012 2.9E-05 -0.169 0.158 0.952 2.118
(0.634) (0.612) (0.881) (0.429) (0.877)
Extension 26 2.5E-04 -1.5E-07 -0.522 0.015 1.2E-04 0.150 0.494 0.604 2.049
(0.983) (0.039)** (0.418) (0.120) (0.216)
*=significant at 10% confidence level, **=significant at 5% confidence level.
Sponsorship decision
The majority of results on the sponsorship decision are opposite to the expected negative relation. Only the US sample, and the IT & Telecom, and extension portfolios show a negative relation between potential agency problems in the sponsorship decision and firm value following a sponsorship announcement, for all other regressions a positive beta was found. However, for none of the betas this relation is statistically significant. Results of the independent samples t-test did show a highly
significant difference between the betas of the EU and US samples16. Indicating that the effect of agency problems in sports sponsorship announcements differs for EU and US firms. In general, no conclusion can be drawn for the hypothesis on agency problems in the sponsorship decision process. This is in line with the results of Cornwell et al. (2005) where also no clear relation was found. However, Pruitt et al. (2004) did find a significant relation between agency problems and firm value following a sponsorship announcement. It is possible that these differing results are caused by the data used to measure agency problems, as, for sports sponsorship decisions, there is no generally accepted measure to identify potential agency problems. Consequently, different data is used in this study (free cash flow per share), and the studies of Cornwell et al. (2005) (cash flow) and Pruitt et al. (2004) (cash flow per share).
Local firm
The factor local, which indicates whether a firm is sponsoring a local or non-local sports entity, is expected to have a positive effect on firm value. In general this is confirmed by the results where the majority of betas have a positive sign. Only the EU sample, and the Oil & Chemicals, and Transport & Equipment portfolios show a negative sign. For the latter portfolio this could be explained by the fact that the majority of firms in this portfolio operate especially on a global basis. The leisure portfolio is the only portfolio which shows a significant positive relation for the local factor.
Therefore, for this portfolio the hypothesis that the sponsoring of a local sports entity enhances the effect of the sponsorship on firm value is accepted. The fact that the betas of most regressions show a positive sign supports the results of Clark et al. (2002), who also found a positive relation between the stock returns of a firm announcing the sponsoring of a local sports entity. This also supports
Meenaghan’s (2001) theory that sponsoring a local sports entity generates more goodwill.
Firm age
Finally, regarding firm age I expect a negative relation with firm value, as young firms use sponsorships as a tool to increase brand awareness and profile themselves. Only the EU sample, and the Finance, and IT & Telecom portfolios have a negative beta which is in line with the hypothesis. Contrary to the hypothesis all other regressions show a positive relation with firm age. However, first
it should be noted that none of the betas is significant. Second, as with firm size, all betas are relatively small which indicates a limited effect of firm age on the value of the sponsoring firm following a sponsorship announcement. In general, these results do not support the hypothesis that young firms generate more value from sponsorship announcements.
The fact that the relationship between the firm specific factors and firm value differs across portfolios confirms the hypothesis that industrial differences not only exist for high tech firms (as argued by Clark et al. (2002) and Cornwell et al. (2005)) but the effect of sports sponsorships on firm value is industry dependent. Concerning these results I should add however that that the sample size of most portfolios was relatively small which could affect the reliability of the results. Compared to previous studies, a general finding on the firm specific factors is that most results are less powerful. Especially for the total, EU, and US samples this could be explained by the applied sample size. In previous studies examining sports sponsorship announcements sample size varies from 24 events (Pruitt et al. 2004) to a maximum of 53 (Cornwell et al. 2005), compared to 178 events for the total sample, 96 EU events, and 82 US events in this study. A second general result is the low R2 of all regression analyses. In the studies of Clark et al. (2002), Pruitt et al. (2004), and Cornwell et al. (2005) R2 values vary from 25% to 71%, compared to a maximum of 28% for the leisure portfolio in this study. A possible explanation for these differences could be the wide variety of sponsored sports entities in the sample of this study. Since Clark et al. (2002), Cornwell et al. (2005), and Pruitt et al. (2004) all focus on one specific type of sports sponsorship. The study of Mishra et al. (1997) is the only study which also uses a wide range of sports entities, in this study the R2 of 7% is also significantly lower.
Additional tests on the assumptions of the OLS-model show that the total and EU samples and the finance portfolio suffer from non-normality. This could be caused by using the simple returns formula instead of continuously compounded returns. For all performed regression analyses the Durbin-Watson statistics show no evidence of autocorrelation in the residual terms. Only for the agro-food portfolio is the statistic relatively low; however this statistic is not sufficiently low to reject the hypothesis of no autocorrelation. The ARCH test showed that ARCH effects are present in the US sample. Therefore a GARCH test was performed on the US sample, however, this test did not provide any differing results.
Finally, to assess the robustness of all results on the firm specific factors I have performed the same regression analyses using a 7-day event window (ranging from t = -3 to t = +3). These
differing results17. The main difference is found for firm size, where contrary to the results of the 3-day event window and the hypothesis the majority of betas are positive. In addition, for the IT & Telecom portfolio the positive relation between firm size and firm value is significant. The results on the sponsorship decision are strengthened by the 7-day event window. Again the majority of betas show, contrary to the hypothesis, a positive sign. Moreover, the betas of the total and EU samples are both statistically significant. For the local factor the positive and significant beta of the leisure portfolio is consistent with the 7-day event window. However, the other regressions show slightly more negative betas which is contrary to the hypothesis. In fact, for the EU sample this negative beta is significant. Finally, for firm age the results are consistent with the 3-day event window. The same accounts for R2 and the results of the ARCH and Durbin-Watson tests. The results of the Bera-Jarque test show that with the 7-day event window the samples and portfolios suffer more from
non-normality.
Sponsorship specific factors
As a next step I will discuss the results of the regression analyses on the sponsorship specific factors contract length, fit, and sponsoring type. These results are presented in table VIII. For the total, EU, and US samples, as for all portfolios, the betas for all sponsorship specific factors are given together with their corresponding level of significance (p-value). Furthermore, statistics on all regressions are presented. These include the adjusted R2, the test statistics of the Bera-Jarque and Durbin-Watson tests, and the significance level of the ARCH tests. The additional statistics all relate to the regression analysis named in the first column of table VIII.
Contract Length
Based on the results from Clark et al. (2002) I expect contract length to be positively related to firm value following a sports sponsorship announcement. However, the results do not provide an unambiguous answer; From the 11 betas on contract length 6 show a negative sign and 5 a positive sign. The portfolios with a statistically significant beta also show differing results. The services & health portfolio has a significant and positive beta, whereas the transport & equipment, and extension portfolios both have a negative beta. For the latter portfolio this negative sign might signal that sponsorship extensions are not always judged as value enhancing by investors. Based on these results no distinctive answer can be given for the hypothesis on contract length. On the other hand, it should be noted that the effect of contract length seems limited since the values of all betas are relatively small.
Table VIII
Results of the regression analyses on the sponsorship specific factors using: CARiτ =αi+β1iLENGTH+β2iFIT+β3iTYPE+εit
18
. For α and all β’s the
p-values are given in parentheses. The Bera-Jarque and Durbin-Watson statistics are the actual test statistics, whereas for the ARCH test the level of significance is given. These additional statistics all relate to the regression for the CAR of the sample/portfolio named in the first column.
CAR N α β1 β2 β3 Adj. R2 Bera-Jarque ARCH Durbin-Watson
Sample
Total 178 -4.9E-04 3.0E-04 2.4E-03 -4.8E-03 -0.011 321.976** 0.943 1.948
(0.936) (0.763) (0.639) (0.354)
EU 96 -2.3E-03 -1.0E-03 8.7E-03 -8.7E-03 -0.007 182.869** 0.944 2.037
(0.781) (0.573) (0.262) (0.265)
US 82 8.1E-03 1.6E-04 -6.3E-03 4.4E-04 -0.026 2.109 0.122 1.936
(0.350) (0.884) (0.324) (0.948)
Portfolios
Agro-Food 38 -7.5E-03 -1.1E-03 0.010 0.012 0.002 0.234 0.334 1.391
(0.449) (0.421) (0.297) (0.133)
Finance 45 9.0E-03 -4.5E-03 0.022 1.5E-04 -0.032 109.686** 0.936 1.962
(0.555) (0.104) (0.007)** (0.991)
IT & Telecom 25 -0.038 8.3E-03 0.027 5.4E-03 0.011 1.650 0.975 2.355
(0.109) (0.137) (0.015)** (0.767)
Leisure 18 -0.010 1.8E-03 -0.019 0.013 -0.164 0.295 0.747 1.325
(0.241) (0.610) (0.337) (0.234)
Oil & Chemicals 11 0.034 -2.9E-03 2.4E-03 -0.038 -0.017 0.670 1.356
(0.298) (0.584) (0.907) (0.141)
Services & Health 18 -0.016 1.7E-03 -1.6E-03 0.026 0.487 1.549 0.442 1.911
(0.015) (0.003)** (0.838) (0.014)**
Transport & Equipment 23 0.031 -2.5E-03 -2.5E-03 -0.023 0.077 0.401 0.907 1.867
(0.021)** (0.089)* (0.800) (0.136)
Extension 26 0.018 -2.5E-03 3.3E-03 -7.0E-03 -0.054 1.398 0.915 1.699
(0.213) (0.086)* (0.784) (0.601)
*=significant at 10% confidence level, **=significant at 5% confidence level.
Sponsor-entity fit
The factor sponsor-entity fit, which indicates whether a fit exists between the sponsoring firm and the sponsored sports entity, is expected to be positively related to firm value (i.e. the announcement of a sponsorship where a ‘good fit’ exists is expected to be more value enhancing for the sponsoring firm compared to sponsorships without a good fit). In general the results support this hypothesis; for the majority of regressions a positive relation between fit and firm value is found. In addition, for two portfolios, finance and IT & Telecom, this positive relation is significant. However, the results for these two portfolios should be taken with care since in these industries only a small number of sponsorships have a good fit. Nevertheless, for these portfolios we can accept the hypothesis that firm value of the sponsoring firm is positively related to the factor sponsor-entity fit. This is also in line with previous results of Pruitt et al. (2004) and Cornwell et al. (2005). In addition, results of the t-test showed that the betas of the EU and US samples differ significantly19. Especially for EU firms fit seems to be an important factor in sports sponsorships.
Sponsoring type
The hypothesis concerning sponsoring type states that: sponsorships which cover a total firm increase the sponsoring firms’ stock returns compared to sponsorships involving a single brand or division of the firm. The results do not provide a clear answer on this hypothesis. The beta of the services & health portfolio is the only significant beta and, in line with the hypothesis, has a positive sign. This also supports the previous findings of Pruitt et al. (2004). However, from the other
regressions five betas show a negative sign. Nevertheless, the results of the services & health portfolio are important since especially in these industries and in the agro-food industry the majority of sports sponsorships are undertaken by a single brand or product. Thus, most firms in these industries can increase their benefits from a sports sponsorship by using the main firm name in the sponsorship.
The effects of the sponsorship specific factors on firm value also differ across the different portfolios. Hence, for these factors this confirms the hypothesis that the effect of sports sponsorships on firm value is industry dependent. This is in line with previous findings of Meenaghan (1983) and Javalgi et al. (1994). As with the firm specific factors, the results of the sponsorship specific factors are less powerful compared to previous studies. In fact, only for the factor fit can a clear answer be provided. Again this could be the result of the applied sample size which, as noted before, is larger compared to previous studies on sports sponsorships. Furthermore, in line with the firm specific factors and the results of Mishra et al. (1997) the R2 values of all regressions on the sponsorship specific factors are also relatively low. The results of the ARCH and Durbin-Watson tests indicate that there is no sign of heteroscedasticity and autocorrelation in the residual terms. However, the
Watson values for the portfolios Agro-Food, Leisure, and Oil & Chemicals are relatively low which makes the test inconclusive for these portfolios.
Finally, for the sponsorship specific factors I have also examined the robustness of the results by performing the same regression analyses using a 7-day event window (ranging from t = -3 to t = +3)20. These regressions are performed on the total, EU and US samples and on all portfolios. In general, these results are consistent with the results of the 3-day event window. Concerning contract length more support is found for the hypothesis that firm value is enhanced by announcing long-term sponsorships. The majority of betas are positive and for the IT & Telecom and Services & Health portfolios the positive betas are significant. These results correspond with previous results of Clark et al. (2002). For the factors fit and sponsoring type no striking differences were found with the 7-day event window compared to the 3-day window. In addition, R2 and the results of the ARCH and Durbin-Watson tests are also consistent with 3-day event window. Finally, in line with the firm specific factors the Bera-Jarque test showed that the samples and portfolios suffer more from non-normality.
VI. Conclusion
In this paper I examined how firm value of European and US firms is affected following the announcement of a sports sponsorship. The effect of the firm and sponsorship specific factors on firm value differs across the total, EU, and US samples and also across industries. This is in contrast with previous studies which show more uniform results. In addition, the studies of Clark et al. (2002), Pruitt et al. (2004), and Cornwell et al. (2005) also show more powerful results where the firm and
sponsorship specific factors have more impact on firm value compared to this study. An explanation for these differences can be the applied sample size and the variety of sports entities in this study. Compared to the previous studies I apply a larger sample and a wider range of sports entities.
Two main questions were examined in this study. First, which type of firm maximizes firm value through a sports sponsorship? Especially small firms can generate value from a sports sponsorship, a conclusion that is in line with the findings of Clark et al. (2002). Furthermore, local sponsorships have a positive effect on the value of the sponsoring firm. The firm specific factors decision and age do not show a significant relation with firm value following a sports sponsorship announcement. Second, what are the characteristics of a value maximizing sports sponsorship? The most important factor for a successful sponsorship is a good fit between the sponsoring firm and the sports entity. The factor fit shows a positive relation with firm value, this confirms previous results of
Pruitt et al. (2004) and Cornwell et al. (2005). The effect of a sponsorships’ contract length is limited and also industry dependent. In addition, sponsorship extensions are not always perceived as value enhancing for the sponsoring firm, as contract length is significantly negative related to the extensions portfolio.
The results of this study are useful for both firms and investors. Firms can enhance their benefits from a sports sponsorship by making a deliberate choice for a sports entity in which the effects of both firm and sponsorship specific factors are taken into account. Accordingly, investors can use this framework to judge whether an investment in sports sponsorships is value enhancing for the sponsoring firm. In addition, this study has provided a first general insight into the effect of sports sponsorships on firm value of EU firms. The results indicate that, compared to US firms, sports sponsorships are valued in a different way for EU firms. For both firm and sponsorship specific factors the effect on firm value differs for EU and US firms. This is especially true for the factors sponsorship decision and fit. Further research on the European market is required to provide more detailed
information on the value of sports sponsorships for EU firms and also to generate new general insights in the relation between sports sponsorships and firm value.
References
Amis, John, Trevor Slack, and Tim Berrett, 1999, Sport Sponsorship as distinctive competence, European Journal of Marketing, 33(3/4): 250-272.
Andreff, Wladimir, and Paul D. Staudohar, 2000, The evolving European model of professional sports finance, Journal of Sports Economics, 1(3): 257-276.
Brooks, Chris, 2002, Introductory Econometrics for Finance, Cambridge: Cambridge University Press.
Brown, Stephen J., and Jerold B. Warner, 1980, Measuring security price performance, Journal of Financial Economics, 8(3): 205-258.
Clark, John M., T. Bettina Cornwell, and Stephen W. Pruitt, 2002, Corporate stadium sponsorships, signaling theory, agency conflicts and shareholder wealth, Journal of Advertising Research, 42(6): 16-32.
Cornwell, T. Bettina, Donald P. Roy, and Edward A. Steinard II, 2001, Exploring managers’ perceptions of the impact of sponsorship on brand equity, Journal of Advertising, 30(2): 41-51.
Cornwell, T. Bettina, Stephen W. Pruitt, and John M. Clark, 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(4): 401-412.
Cornwell, T. Bettina, Clinton S. Weeks, and Donald P. Roy, 2005, Sponsorship-linked marketing: opening the black box, Journal of Advertising, 34(2): 21-42.
Farrelly, Francis, Pascale Quester, and Rick Burton, 2006, Changes in sponsorship value: competencies and capabilities of successful sponsorship relationships, Industrial Marketing Management, 35(8): 1016-1026.
Grinblatt, Mark, and Sheridan Titman, 2002, Financial Markets and Corporate Strategy, 2nd ed., New York: McGraw-Hill.
Grohs, Reinhard, Udo Wagner, and Sabine Vsetecka, 2004, Assessing the effectiveness of sports sponsorships – an empirical examination, Schmalenbach Business Review, 56(2): 119-138.
Gwinner, Kevin, 1997, A model of image creation and image transfer in event sponsorship, International Marketing Review, 14(3): 145-158.
Javalgi, Rajshekhar G, Mark B. Traylor, Andrew C. Gross, and Edward Lampman, 1994, Awareness of sponsorship and corporate image: an empirical investigation, Journal of Advertising, 23(4): 47-58.
Jensen, Michael C., 1986, Agency costs of free cash flow, corporate finance, and takeovers, American Economic Review, 76(2): 323-329.
