Author/Graduate: Anniek de Vries Student number: S1550969
Title:
The Value of Formula One Sponsorship: The Effect
of Sponsorship Announcements on the Stock Price
Master’s degree programme: International Economics and Business Name of supervisors: M. Koetter, R.H. Koning
ABSTRACT
This paper analyses the effects of the announcements of Formula One sponsorship on the stock prices of the firms. New announcements, renewals of contracts as well as sponsorship withdrawals are researched. The findings shows that there is a positive and significant reaction for new announcement while the reaction is negative and significant for renewals of sponsorship contracts. The reaction to withdrawals is insignificant. Furthermore, the factors that are of influence are found to be the title sponsorship, cash flow and the size of the firm where the first has a positive effect while that of the latter two is mainly negative.
CONTENTS
1. INTRODUCTION
Worldwide sponsorship is expected to continue to grow over years to come as it already did over the last few years (IEG Report, 2007, 2008). In 2007 the total sponsorship spending worldwide was estimated to be 37.9 billion dollar, while that figure grew with over 14% to 43.1 billon dollar in 2008. The estimates for 2009 are an increase of 3.9% to 44.8 billion dollar. The majority of this is spent on sponsorships of sports.
As in most sports, sponsorship is of great importance is Formula One. Formula One racing is one of the most popular sports in the world. It is a sport which has a highly international profile, great group of followers, and large audiences. The number of television viewers shows an increasing number over the last few years with the 2008 season attracting a television audience of 600 million worldwide (www.f1-live.com). The sport is an expensive one for the Formula One teams involved, with the teams spending up to 400 million dollar a year. And although the aim is to reduce this spending in future years, the costs will still remain high. It is thus necessary for the Formula One teams to find the funds to be able to compete. Where 15-20% comes from TV revenue and prize money, the majority of around 80-85% of this money comes from corporate sponsorship (BBC, 2006).
Firms invest large sums of money to be a sponsor of a Formula One team with the aim of increasing their brand awareness. Traditional research on the effectiveness of sponsorship mainly focused on this increasing awareness for customers, for example by measuring television exposure (Arthur, Dolan and Cole, 1998) and the effect on customers (Quester, 1997). Over recent years, however, event studies that look at the impact of sponsorship announcements on the stock prices of the sponsors have become more popular to examine the sponsorship effectiveness (Clark, Cornwell and Pruitt, 2008; Cornwell, Pruitt and Clark, 2005; Farrell and Frame, 1997 Miyazaki and Morgan, 2001; Pruitt, Cornwell and Clark, 2004;). Stock prices immediately reflect the investors’ reaction to new information in the market place, in this case the sponsorship announcement. Thus when the sponsorship is considered to be favourable to a company, the stock prices should rise.
previous literature is that announcements of the withdrawal of sponsorship from Formula One are also analysed. As benefits of a sponsorship are expected to decline as time goes on the expectation is that this leads to a positive reaction. Firms no longer invest in a sponsorship that is no longer beneficial. Furthermore, the factors that influence the possible change in stock prices are analysed. First, title sponsorship is hypothesized to have a larger effect than main sponsorship. Second, congruent sponsors (thus firms active in the automotive industry and related industries) are hypothesized to have higher stock prices. Third, sponsorship of a successful Formula One team is hypothesized to lead to higher stock prices. Fourth, cash flow is hypothesized to be negatively related to the stock price as it is an indicator of agency problems. Finally, the size of the firms is hypothesized to be negatively related to stock prices as smaller firms are better able to capture the synergistic marketing benefits.
The structure of the remainder of this paper is a follows. First, a section that describes of Formula One and Formula One sponsorship. In the third section the previous research on effects of sponsorship announcements on stock markets is described. Section 4 describes the data and the methodology used in thus study. Section 5 then presents the results. Section 6 concludes.
2. FORMULA ONE RACING
2.1 History of Formula One Sponsorship
The Fédération International Automobile (FIA) is a governing body for many racing series and Formula One is the highest class. Formula One cars are single seaters that reach speeds up to 360km/h. A racing season consists of a series of races (Grand Prixs) that take placxe all around the world. The results of all races combined determine a World Drivers Champion and a World Constructors Champion. The first Formula One Championship took place in 1950 and the sport has grown since. And with it the importance of sponsorship.
thus is willing to pay more for the sponsorship, there is no exact control over the coverage. Message means that in traditional advertising the message can be created by the advertiser while that of sponsorship is delivered by association. Implementation has to do with the fact that the sponsor needs to recognise that extra money is needed to exploit the rights of the sponsorship. Commercials with the Formula One team and drivers, or print advertisement for example may be required and are an additional investment into the sponsorship. The reaction of the audience to traditional advertising is expected to be different to that to sponsorship activities. It is possible that the sponsorship is more appreciated due to ‘goodwill’ towards the sponsored Formula One team. Lastly, there are personal motives which means that sponsorship may be chosen for personal benefits that the decision makers perceive. An example of this are the hospitality benefits offered by the sponsored Formula One team (Meenaghan, 1991).
Sponsorship in Formula One was first introduced not long after the introduction of commercial sponsorship. In 1968 Lotus introduced Imperial Tobacco as a sponsor and named the team Gold Leaf Team Lotus. This was the start of the Formula One sponsorship. Soon all the teams realised the benefits and tried to attract sponsors (CNN, 2007). For a long time Tobacco sponsorship was one of the main sources of sponsorship for many Formula One teams. However, with the EU announcement in 2002 that Tobacco sponsorship would be banned for the 2006 season teams were forced to look for other options. Although the ban is in place for a few years now Ferrari still has a main sponsor from the tobacco industry, the Marlboro brand of Philip Morris. However, Philip Morris does not appear on the car, only a barcode (Gibson, 2002).
1.2 Sponsorship in Formula One and the costs and benefits
Before defining the specific sorts of sponsorship in Formula One a more general definition of sponsorship is proposed. Many definitions of sponsorship were used in previous literature, but the one most generally used in recent research is from Meenaghan (1983, p57). Meenaghan states that “sponsorship can be regarded as the provision of assistance either financial or in-kind to an activity by a commercial organization for the purpose of achieving commercial objectives” and this definition can be said to apply to all sorts of auto racing sponsorships (Cornwell, Pruitt and van Ness, 2001).
generally officially announced to the press and have the highest costs. We can roughly distinguish between three types of Formula One team sponsors which each face different costs (BBC, 2006):
1. Title sponsorship: These sponsors are part of the name of a Formula One team (e.g. ING Renault or Vodafone McLaren Mercedes) and the sponsor name is positioned on the most visible parts of the car. The costs of such sponsorship are between 15 to 50 million dollar, depending on the performance of the team in years before.
2. Main sponsor: These sponsors are interested in being linked with a sport such as Formula One and are also on clearly visible places on the car. The costs of such sponsorship are between 3 and 15 million dollar, again depending on the performance of the team.
3. Trade link-ups (or “official supplier”): The companies have a less visible place on the car and also not as large. These firms are usually involved in the high-tech or engineering industry and want to be associated with the sport. The costs of such sponsorship are between 1 to 3 million dollar, again depending on the performance of the team.
However, one must keep in mind that these costs are average estimates, and could also be higher. The Ferrari-Vodafone sponsorship for example was rumoured to involve around 80 million dollar (Carter, 2005). Vodafone switched their sponsorship to McLaren in 2007 and the estimated spending on the sponsorship their is around 50 million dollar a year (Lewis, 2009).
Some of the drivers also have individuals sponsors, who help them get a seat in Formula One. Sponsorship of this kind is hard to measure separately, as most of these sponsors will get a place on the car and are considered a team sponsor. Medion was announced as a sponsor of the Spyker Formula One team in 2006, where they followed their driver Adrian Sutil whom they had been sponsoring in lower racing classes for years (Medion, 2006). A similar situation was seen when Trust acted as a personal sponsor of Jos Verstappen to try and buy a seat at Jordan F1 in 2004 (Henry, 2004).
of Telefónica is prominently placed at the sides of various parts of the circuit (Telefónica, 2008). Second, there is also the possibility to be a regular sponsor whose name appears on the boarding at the side of the circuit. Sponsorship of a Formula One circuit is costly, just as the sponsorship of a Formula One team. The costs of Santander being the title sponsor of the British GP of 2007 for example, were estimated to be 8 million dollar. (Noble, 2007). For 2009 there are 8 different sponsors, who are the title sponsor of one or more tracks on the Formula One calendar. Most sponsorship deals are long term, with some sponsorship deals exceeding ten year. An overview for 1994-2009 is given in table A1 in the appendix. However, it is difficult to do research on the effects of the circuit sponsorship as many of the sponsorship deals are not announced officially to the press.
2. LITERATURE REVIEW
2.1 Previous literature
The types of research on sponsorship effectiveness can be roughly divided into a few streams; studies about the exposure in the media, studies linked to the awareness of customers of the sponsor/sponsorship, and measuring sponsorship returns by looking at stock price changes. Media exposure studies focus on how often a sponsor appears in the media and the duration of this exposure. Arthur, Dolan and Cole (1998) investigate the relationship between the expected amount of television exposure and the position of the motorcyclist. They conclude that more competitive bikes gain more television exposure. Looking at the most visible sponsor in the 2007 season, the first position is held by McLaren’s title sponsor Vodafone while second position was taken by Ferrari’s title sponsor Philip Morris (Marlboro). McLaren and Ferrari were also the first two teams in the 2007 championship showing the link between success and exposure appears to be present in Formula One as well (Margaux Matrix, 2008). Awareness studies focus on the awareness that is created for sponsors by the sponsorship in the mind of potential customers. Quester (1997) investigates the Adelaide Formula One Grand Prix to establish the impact of sponsorship on brand and/or company awareness and the occurrence of ambush effects. The event increases customer awareness, mainly among those who attended the event. However, ambush effects occur which means that there are firms which tried to associate themselves with the Formula One Grand Prix to obtain the same benefits as an official sponsor at less costs. More recently a stream of research focuses on the reaction of the stock prices of a sponsoring firm to a sponsorship announcement. As this study follows that stream of research, the previous research on stock price reactions will be more thoroughly discussed.
No previous event studies on Formula One have been done to establish the effects of a sponsorship announcement for the firm. Furthermore, studies that focus solely on new sponsorship announcements of auto and/or motor racing are rare. An exception is Cornwell, Pruitt and Van Ness (2001), who study the IRL1 and specifically the effects of winning the Indianapolis 500 on the stock prices of the sponsoring firms. Their results show that there were no increases in stock prices for ‘winning firms’ in general, but if the sponsor has direct ties to the automotive industry they experience almost 3% increases in stock prices around winning the Indianapolis 500.
NASCAR2 is the most researched auto racing investigated with regards to sponsor announcement effects on stock prices however this is mostly in combination with other sports or events. Pruitt, Cornwell and Clark (2004) conduct research which focuses solely on NASCAR. They find an increase in stock prices for sponsoring firms following a sponsorship announcement of 1.13%. Looking at the total net present value (NPV) this means a 334 million dollar increase of market value for all firms on the day of the sponsorship announcement. In other studies which have a combined dataset of NASCAR and other sports the data solely from NASCAR again shows positive returns. The effect is approximately a 2% increases in share price (Clark, Cornwell and Pruitt, 2008).
Clark, Cornwell and Pruitt (2008) also include PGA, LPGA, professional tennis tours and NCAA bowl games3 in the above mentioned study. The overall sample gives no evidence of a stock market reaction. The results for NASCAR are positive while those for NCAA bowl sponsorships are clearly negative. The other sports show neither positive or negative significant results. An earlier study of Cornwell, Pruitt and Clark (2005) looks at the Major League Sports NLF, MLB, NBA, NHL and PGA4 finds an overall positive significant effect of official sponsorships on the stock market. Shi and Ghosh (2005) combine NLF, MLB, NBA, NHL and PGA together with NASCAR finding that the overall returns are positive with an effect of around 1%.
Farrell and Frame (1997) investigate the sponsorship of the 1996 Atlanta Summer Olympics and find that the sponsorship has a significant negative effect on the stock prices. However, Mishra, Bobinski and Bhabra (1997) who also conduct research on sponsorship of the Olympics, as well as other sports and events, find a positive return for the stock market. Miyazaki and Morgan (2001) also look at the 1996 Olympic Summer Games to research the effects of sponsorship announcement and find a positive return for the stock market as well. The studies by Farrell and Frame (1997) and Miyazaki and Morgan (2001) cover largely the same firms, 24 out of 26 and 27 respectively, but use different estimations periods as well as different event windows. This in combination with the small variation in investigated firms gives different results, showing the influence of the used estimation period and event windows on the results.
Next to first-time sponsorship announcements there are also announcements of the extension of existing sponsorship agreements. This was not thoroughly discussed in research
2 NASCAR: National Association for Stock Car Auto Racing
3 PGA: Professional Golfers Association, LPGA: Ladies Professional Gold Association, NCAA: National Collegiate Athletic Association 4
so far, but there is research on re-announcements for NASCAR, PGA and NCAA bowl games. These show that both the original announcement and the re-announcements are found to be positive for NASCAR (albeit insignificant). NCAA announcements are originally negative while the renewal is neutral. The PGA originally has insignificant results while the re-announcements are found to be negative. These last two results are explained by the authors as possible consequences of the dramatic rise in sponsorship fees for the PGA tour whereas the costs of NCAA did not significantly change (Clark, Cornwell and Pruitt, 2008).
The studies on stock price responses of sponsorship announcements thus shows different results. Whereas the studies on NASCAR display an overall positive results, studies on other sport do not have such clear-cut results. Some studies show a clear positive result on the stock prices, others find a negative result and some no results at all. Research on re-announcements is scarce and does not display a clear pattern.
Previous studies also investigate different factors that can potentially affect the increase or decrease in stock prices. First, there is the matter of congruent sponsorship. Congruent sponsorship is the sponsorship by a firm in an industry related to the sponsored object. This congruent sponsorship is included in practically every study on the subject. The assumption made is that announcements of congruent sponsors are more positively perceived by the market then non-congruent sponsors. Meenaghan (1983) and Otker and Hayes (1987) note that the stronger the link between a sponsor and a sponsored event, the larger the positive effect of the sponsorship. Findings for NASCAR show that this is clearly true as a sponsor that has direct ties to the automotive industry has an increase in share price that is 2.8% higher than that of a company which has no ties to the automotive industry (Pruitt, Cornwell and Clark, 2004) For NASCAR, tennis, golf and college bowl games combined this percentage is actually as high as 3.4% (Clark, Cornwell and Pruitt, 2008).
Some studies also include measures to control for the size of the firms. Cornwell, Pruitt and Clark (2005) use market value of equity and find that the effect is negative, but it is not of a significant influence to impede with a successful major league sponsorship. Pruitt, Cornwell and Clark (2004) find that the effect for total corporate assets is positive, but again it is not significant. Other studies that control for firm size with total assets also find it to be not significant (Farrell and Frame, 1997; Mishra, Bobinski and Bhabra, 1997).
Cash flow is included in some studies to control for the fact that the sponsorship deal may be concluded for the personal gain of the managers involved. The sponsorship deal may suffer from agency problems. In Formula One this could be of particular interest because of the hospitality benefits that come with sponsoring a Formula One team. Cash flow is a useful proxy as agency problems are more likely in firms with high cash flows. The reason for this is that as the effectiveness of shareholder monitoring of corporate expenses decreases as cash flow increases. For a combined study on NASCAR, PGA, LPGA, professional tennis tours and NCAA bowl games cash flow does not display a significant effect (Clark, Cornwell and Pruitt, 2008). For a study solely on NASCAR cash flow is negative and significant, indicating that agency problems are present and the agents thus act from the opportunity of personal gain (Pruitt, Cornwell and Clark, 2004).
An overview of the most important literature can be found in table 1.
Table 1. Main findings in event studies on sports sponsorship Author(s) Sports Event Key findings
Event study Regression analysis Clark, Cornwell, and Pruitt (2008) NASCAR, PGA, LPGA, professional tennis tours, NCAA bowl games Sponsorship announcement Total sample insignificant
Positive for NASCAR Negative for NCAA bowl games
Congruent
sponsorship, market value, high tech firm,
Sponsorship
re-announcement
Positive for NASCAR Insignificant for NCAA
Negative for PGA Cornwell,
Pruitt and van Ness (2001)
IRL Winning the
Indy 500 Insignificant Congruent sponsorship Cornwell, Pruitt and Clark (2005) Major League Sports (NLF, MLB, NBA, NHL, PGA) Sponsorship announcement
Total sample positive Congruent
Farrell and Frame (1997) Atlanta Summer Olympics 1996 Sponsorship announcement
Negative Total assets
Mishra, Bobinski and Bhabra (1997) Sports, Olympics and miscellaneous Sponsorship announcement
Positive Return on assets
Miyazaki and Morgan (2001) Atlanta Summer Olympics 1996 Sponsorship announcement Positive Pruitt, Cornwell and Clark (2004) NASCAR Sponsorship announcement Positive Congruent sponsorship, success of the sponsored team, cash flow Shi and Ghosh
(2005) NLF, MLB, NBA, NHL, PGA, NASCAR Sponsorship announcement
Total sample positive Return on investment, return on equity, price-to-book ratio total assets, institutional
ownership
2.2 Hypotheses
Firms engage in sponsorship to reach certain objectives. These objectives can be classified as corporate objectives (such as image and goodwill), product-related objectives (linking a brand to the sponsored object), increased sales, coverage in media, hospitality (such as being a guest at a Formula One race), and personal objectives. These objectives can overlap and are not mutually exclusive (Meenaghan, 1983). Reaching these objectives makes a sponsorship beneficial to the firm which is of course the objective. The effectiveness of sponsorship as a whole is increasingly looked at by investigating the impact on stock prices as stock prices immediately reflect the investors’ reaction to the new information in the market place. When the sponsorship is considered to be favourable to a company, the stock prices should rise. When the sponsorship is considered to be negative, the stock price should go down. This is measured by investigating the abnormal returns of the stock price. The abnormal returns is the difference between the expected return and the actual return, where the difference is triggered by a certain event, in this case the sponsorship announcement.
previous research with automotive sports (NASCAR). Furthermore, the continued growth in sponsorship investment signals that firms perceive benefits.
Hypothesis 1: The announcement of sponsorship of a Formula One team is positively associated with abnormal stock market returns
Past studies on re-announcements of sponsorship are scarce, a study on NASCAR shows that a re-announcement of sponsorship has positive effects, although not always significant (Clark, Cornwell and Pruitt, 2008). The contract extension could signal that the sponsoring firm perceives there are benefits gained from the sponsorship and that therefore the sponsorship is renewed. A re-announcement is therefore expected to have a positive effect.
Hypothesis 2: The re-announcement of an existing sponsorship of a Formula One team is positively associated with abnormal stock market returns.
In recent years a few large sponsor have left Formula One, such as ING (Noble, 2009). On the other hand, no new large sponsorship deals were made (Margaux Matrix, 2008). This could signal that Formula One sponsorship is not considered to be favourable as before. The withdrawal from a money consuming sponsorship deal could thus actually be perceived to be positive. Furthermore, while the renewal of a sponsorship contract could be an indicator that the firm perceives the sponsorship to be beneficial the withdrawal of a sponsorship could signal that the firm no longer perceives any benefits from a sponsorship involvement. Meenaghan (1983) notes that a gradual decrease in interest may occur which means that the benefits of the sponsorship gradually decreases as well. The withdrawal from a sponsorship may be a confirmation by the firm that this is indeed the case, and that a level is reached where benefits are no longer achieved. Therefore, the sponsorship withdrawal is hypothesized to lead to positive abnormal stock returns.
Hypothesis 3: The withdrawal of a sponsor from the sponsorship of Formula One team is positively associated with abnormal stock market returns.
Title sponsors pay the largest sums of money, and in return get the most visible place on the car. Furthermore, they appear in the official team name, such as AT&T Williams or Marlboro Ferrari. Main sponsors on the other hand, have a prominent place on the car but they do not appear named in every press release. One would thus expect that title sponsorship has a higher return than being a main sponsor.
Hypothesis 4a:The abnormal stock returns of firms announcing Formula 1 sponsorship will be larger for title sponsors than for main sponsors
Hypothesis 4c: The abnormal stock returns of firms announcing a withdrawal from Formula 1 sponsorship will be larger for title sponsors than for main sponsors
Sponsors and products which are linked with the automotive industry may have the sponsorship viewed as more positive. Cornwell, Pruitt and van Ness (2001) and Pruitt, Cornwell and Clark (2004) show this to be true for NASCAR. Meenaghan (1983) and Otker and Hayes (1987) argue that he stronger the link between a sponsor and a sponsored event, the larger the positive effect of the sponsorship.
Hypothesis 5a: The abnormal stock returns of firms announcing Formula 1 sponsorship will be positively associated with congruent sponsorship
Hypothesis 5b: The abnormal stock returns of firms re-announcing Formula 1 sponsorship will be positively associated with congruent sponsorship
Hypothesis 5c: The abnormal stock returns of firms announcing a withdrawal from Formula 1 sponsorship will be positively associated with congruent sponsorship
More successful Formula One teams will have more television exposure (Margaux Matrix, 2008), hence the sponsors will be more noticed. The sponsored firms name will more often be displayed on television. The increased exposure increase the value of the sponsorship (Arthur, Dolan and Cole, 1998). This could indicate that the stock returns are higher for the more successful Formula One teams.
Hypothesis 6a: The abnormal stock returns of firms announcing Formula 1 sponsorship will be positively associated with the success of the Formula 1 team
Hypothesis 6b:The abnormal stock returns of firms re-announcing Formula 1 sponsorship will be positively associated with the success of the Formula 1 team
Hypothesis 6c:The abnormal stock returns of firms announcing a withdrawal from Formula 1 sponsorship will be positively associated with the success of the Formula 1 team
Hypothesis 7a: The abnormal stock returns of firms announcing Formula 1 sponsorship will be negatively related to cash flow
Hypothesis 7b: The abnormal stock returns of firms re-announcing Formula 1 sponsorship will be negatively related to cash flow
Hypothesis 7c: The abnormal stock returns of firms announcing a withdrawal from Formula 1 sponsorship will be negatively related to cash flow
The sponsors of the Formula One teams differ in size and total assets can be the indicator to control for size. The expectation is that the sign is negative because for a given fixed level of sponsorship net present value must decline as the size of the firm goes up (Pruitt, Cornwell and Clark, 2004). Cornwell, Pruitt and van Ness (2001) argue that smaller firms generally offer fewer products and operate in fewer markets than larger firms. Total value of the sponsorship perceived by investors, the percentage gain in abnormal returns, is larger for these smaller firms. Any given dollar of benefit of the sponsorship displays a smaller percentage of the total firm when a firms is larger. Furthermore, smaller firms may be better able to capture synergistic marketing benefits (Clark, Cornwell and Pruitt, 2002).
Hypothesis 8a: The abnormal stock returns of firms announcing Formula 1 sponsorship will be negatively related to total assets
Hypothesis 8b: The abnormal stock returns of firms re-announcing Formula 1 sponsorship will be negatively related to total assets
Hypothesis 8c: The abnormal stock returns of firms announcing a withdrawal from Formula 1 sponsorship will be negatively related to total assets
3. DATA AND METHODS
3.1 Data Collection
The sample of this research involves the title and main sponsors of Formula One teams. Title sponsor are first identified as the sponsors which are included in the official team name of the Formula One team. Then main sponsors are identified as the sponsors which are on the car prominently. That includes the side pods, rear wing and/or front wing. Smaller sponsor are excluded from this study as these not as clearly displayed on the car, and thus do not get as much notice. In addition, they are mostly not as prominently announced.
websites of either the sponsor of the Formula One team in the form of a press release, or in the news archives of the websites autosport.com and grandprix.com. The websites have an extensive news archive on Formula One news. For some sponsorships an announcement date could not be identified, thus these sponsorships are excluded from this study. After identifying the sponsorship announcement dates the Lexis Nexis database is used to check if there were no other announcements around the time of the sponsorship announcement which could affect the outcome. This leads to the removal of three of the identified announcements. In the end, the sample eventually consists of 40 useable sponsorship announcements. Table 2 provides an overview of the sponsors that are used in this study. This sample size can be said to hold the middle compared to other studies. However looking at studies that focus on only one sport or event the sample size is relatively large. A combined studies on the MLB, NBA, NFL, NHL and PGA covers 53 sponsorship announcements (Cornwell, Pruitt and Clark, 2005) while a study on 76 firms includes sponsoring of 30 sporting events, 20 Olympic games and 26 miscellaneous events (Mishra, Bobinski and Bhabra, 1997). A study solely focusing on NASCAR has 24 announcements in their study (Pruitt, Cornwell and Clark, 2004). A study solely focusing on the IRL has 28 announcements (Cornwell, Pruitt and van Ness, 2001), while 2 studies on the 1996 Summer Olympics in Atlanta include 26 (Farrell and Frame, 1997) and 27 announcements (Miyazaki and Morgan, 2001).
Table 2: Formula One team sponsors
Sponsorname Team Event date (mm-dd-yyyy)
1 Marlboro - Philip Morris Ferrari 12/09/1996
2 Vodafone Ferrari 05/25/2001
3 Shell Ferrari 09/06/1995
4 Vodafone McLaren 12/14/2005
5 Johnnie Walker - Diageo McLaren 02/22/2005
6 Banco Santander McLaren 11/02/2006
7 Mobil1 - Exxon Mobil McLaren 03/08/2006
8 ING Renault 10/16/2006
9 Elf – TotalFinaElf Renault 02/26/2001
10 Telefónica Renault 02/20/2004
11 Akai Renault 10/20/1997
12 Panasonic Toyota 07/03/2001
13 Denso Toyota 01/16/2004
14 Esso - Exxon Mobil Toyota 03/01/2001
15 KDDI Toyota 03/03/2003
16 Petronas BMW 11/24/2005
17 Intel BMW 12/15/2005
18 Credit Suisse BMW 10/14/2005
19 Dell BMW 05/08/2006
20 Credit Suisse Sauber 01/29/2001
21 Parmalat Sauber 01/25/1999
22 Winfield - British American Tobacco Williams 01/03/1997
24 AT&T Williams 10/20/2006
25 Allianz Williams 05/19/2000
26 Royal Bank of Scotland Williams 01/06/2005
27 Lenovo Williams 02/02/2007
28 Philps Shavers - Philips Williams 12/06/2005
29 AT&T Jaguar 01/09/2001
30 HSBC Stewart GP 09/17/1996
31 Seita Prost GP 02/13/1997
32 Acer Prost GP 02/22/2001
33 Yahoo Prost GP 01/31/2000
34 Kingfisher Airlines Force India 11/12/2007
35 ICICI Force India 02/07/2008
36 Medion SpykerF1 01/17/2007
37 Spyker SpykerF1 09/11/2006
38 Repsol Jordan 02/16/1998
39 Orange - Vodafone Arrows 03/08/2000
40 Repsol Arrows 01/22/1999
Following this list of initial sponsors active on the stock market, the re-announcements and/or withdrawals of these contracts are indentified. After identifying these sponsorship announcement dates, again the Lexis Nexis database is used to check if there were no other announcements around the time of the re-announcements and withdrawals which could affect the outcome. In the end there is a sample of 24 re-announcements (table 3) and 11 sponsorship withdrawals (table 4). To keep the research findings relevant only sponsorship announcements of the last 15 years are included, that includes January 1993-December 2008. All data on the stock prices are obtained from Datastream.
Table 3 Sponsorship reannouncements
Sponsorname Team Event date (mm-dd-yyyy)
1 Marlboro - Philip Morris Ferrari 9/5/2005
2 Marlboro – Philip Morris Ferrari 1/29/2007
3 Vodafone Ferrari 12/16/2004
4 Shell Ferrari 4/20/2000
5 Shell Ferrari 4/25/2005
6 Johnnie Walker - Diageo McLaren 2/13/2006
7 Johnnie Walker - Diageo McLaren 10/8/2007
8 Hugo Boss McLaren 5/25/2001
9 Mild Seven - Japan Tobacco Benetton 7/24/2002
10 Mild Seven - Japan Tobacco Benetton 1/20/2003
11 Panasonic Toyota 10/7/2005
12 Denso Toyota 1/10/2007
13 Allianz Williams 11/13/2008
14 Allianz Williams 2/7/2007
15 Royal Bank of Scotland Williams 5/1/2007
16 Hamleys Williams 5/5/2008
17 Philips Shavers – Philips Williams 11/23/2006
18 Philips Shavers – Philips Williams 12/5/2007
20 HSBC Jaguar 7/10/2001
21 Gauloises - Seita Prost 8/25/1997
22 Benson & Hedghes – Gallaher Group Jordan 2/5/2003
23 Benson & Hedghes – Gallaher Group Jordan 2/3/2004
24 DHL – Deutsche Post Jordan 2/21/2002
Table 4: Sponsorship withdrawals
Sponsorname Team Event date (mm-dd-yyyy)
1 Mild Seven – Japan Tobacco Renault 2/1/2006
2 ING Renault 2/16/2009
3 Credit Suisse BMW 1/19/2009
4 Winfield - British American Tobacco Williams 7/2/1999
5 Compaq/HP Williams 9/15/2005
6 Royal Bank of Scotland Williams 2/25/2009
7 Gauloises – Seita Prost 9/26/2000
8 British American Tobacco BAR 10/19/2006
9 Spyker Spyker 10/5/2007
10 Orange – France Telecom Arrows 10/11/2002
11 Telefónica Renault 1/25/2007
Data on the factors that affect the value of the sponsorship announcements are obtained from different sources, namely the official Formula One website, the official website of the Formula One Team, the websites of the sponsors and Datastream.
Whether the firms is a main sponsor or title sponsor is determined by looking at the official websites of the Formula One teams and the Official Formula One website. When the sponsor name is used in the official teamname, the sponsor is determined to be a title sponsor. If the sponsor name is not used in the official teamname, but still prominently displayed on the car, then the sponsor is determined to be a main sponsor.
Whether the sponsor is a congruent sponsor is determined by looking at the industry in which the firm is active. When this industry is related to the automotive industry the sponsor is determined to be a congruent sponsor. In this study that means car manufacturers and oil and gas companies.
The determinant for agency problems is cash flow. Cash flow data is obtained from Datastream and is the cash flow per share on the date of the announcement..
The determinant for firm size, total assets, is chosen to reflect the differences in corporate scale. Total assets on the date of the announcement is taken from Datastream.
3.2 Concepts and methods
3.1.1 Event study methodology
The paper that first introduced the event study methodology to look at stock price reactions is by Fama, Fisher, Jensen and Roll (FFJR) (1969). Since then the methodology is increasingly popular to investigate the reaction of stock prices to the announcement of events. The two major reasons to use this methodology are: 1. testing whether the market efficiently incorporates information and 2. examining the exact impact of events on stock prices (Binder, 1998). The advantages of measuring sponsorship returns by looking at stock price changes with the use of event studies is that in an efficient market the stock prices should incorporate all the news available immediately. Thus when an announcement is made this should be immediately incorporated into the stock price therefore looking at the abnormal returns with an event study gives a good measure of the response to such an announcement.
There are different alternative models to calculate the abnormal returns on stock prices. The abnormal return can occur because of a certain event influencing the return. This causes the observed return to be different from the expected return. Here the event is the announcement of a sponsorship contract with a Formula One team. The standard market model is the one mostly used in event studies today. The model is based on the original FFJR (1969) model which examines the effect of the announcement of stock splits on stock prices. Event studies can be conducted with monthly or daily data, for this study daily data is chosen as the event dates can be precisely determined in this study and using monthly data would thus not capture the event as well. Furthermore, the power is also greater as Brown and Warner (1985) proved for the market model and market adjusted returns model.
For the standard market model the announcement returns are calculated following the analysis on event studies by Brown and Warner (1985). The basic steps in performing an event study are described by Henderson (1989) and can be applied as follows:
here as the event dates can be precisely determined in this study and using monthly data would thus not capture the event as well. Furthermore, the power is also greater as Brown and Warner (1985) prove for the market model and market adjusted returns model.
2. Define the returns of the individuals firms in the absence of this news.
The change in market value of each firm announcing a sponsorship is measured by comparing the actual return on the event day to the predicted normal return if the event had not taken place. Using the standard market model as described above, the expected return, ri,t, for firm i on day t can be expressed as a linear function of the returns on the market portfolio, rm,t, just as in (1) t i t m i i t i r r, =α +β , +ε , 3. Measure the abnormal return
Then, just as in (2) the abnormal return, ARi,t, for firm i on day t is given by
) ( , , ,t it i i mt i r a br AR = − +
where ri,tt is the expected return for firm i on day t, and ai and bii are the estimates of
the market model parameters αi and βi. These parameters are obtained by using an OLS regression estimation using an interval of 225 days starting at t=-250 and running to t=-25. This end of the estimation window is chosen so that the estimation window does not include the event window. Biased estimates can occur when the normal return is estimated including the period were the abnormal returns are expected to occur. Furthermore, t=0 is set as being the announcement date.
4. Define the aggregated returns
Now that the individual abnormal returns are determined for each firm, one can look at the average of all sponsoring firms that this study will address (an overview is in the next section). The average abnormal return for all the firms in the sample on day t is given by
∑
= = N i t i t i AR N AR 1 , , 1 (9)shorter estimation window is also chosen to check if the results change significantly. This robustness check is with an estimation window of 149 days, namely over [-170,-21].
Next the returns can be aggregated across time. The cumulative abnormal returns (CARs) can be calculated as
∑
= = 2 1 , t t t t i AR CAR (10)where CAR denotes the average cumulative abnormal return for the event period used. As in previous studies the interval data [0,1] is used. In addition the intervals 2,2], [-5,5], and [-10,10] are also used. As the estimation window, this event window is chosen based on previous studies on the effect of sponsorship announcements. The event windows of [0,1] is the standard window in many studies as it includes the event days and the day after, this account for the fact that the event may have happened when the trading day had already closed. The longer event windows are included to incorporate that the event may be anticipated by the market due to rumours of the announcement before it was made official and that the information of the event may need some time to be absorbed by the market (for example Cornwell, Pruitt and Clark, 2005; Farrell and Frame, 1997). Some studies (for example Pruitt, Cornwell and Clark, 2004) include windows starting 50 or 100 days before the event and 50 to 100 days after, but by using these long event windows it is very well possible that another event is measured than the intended event as the data is contaminated by other important news for the firms
Studies on the US stock markets (NYSE, AMEX, NASDAQ) use the CRSP5 database to find a market return with which to calculate the abnormal returns. (for example Clark, Cornwell and Pruitt, 2008; Cornwell, Pruitt and Clark, 2005; Farrell and Frame, 1997; Pruitt, Cornwell and Clark, 2004) As the data from this study are global an international market index is selected, namely the S&P Global 1006. This index is chosen because most of the sponsors of Formula One teams are large global companies and the S&P Global 100 measures the performance of 100 multinational global companies. An alternative index could be the Grand Prix index offered by Stoxx which consists of the 15 main players in the Formula One industry. This index includes the six engine manufacturers, the tire supplier, 4 oil and fuel suppliers as well as 5 title sponsors. However, data is only available from the first of January 1997 which is a disadvantage as the sample in this study is already small. Furthermore, the index is weighted and 50 percent is made up by the engine manufacturers. Most weight thus
5
CRSP: Center for Research in Security Prices
lies on firms active in the automotive industry itself while the sponsors in this study are active in different large industries. Performance of firm in this study can better be compared to that of different multinationals therefore making the S&P Global 100 a better performance indicator.
5. Statistically test the cumulative abnormal returns for significance.
The last step is to test whether the cumulative abnormal returns are significant. The cumulative abnormal returns are regressed with robust standard errors and then tested for significance.
There are different alternative regression models to the standard market model in (1). The methods to calculate the normal returns can be roughly divided into two categories, namely statistical and economical models (MacKinley, 1997). Statistical models follow solely statistical assumptions while economic models use assumptions on investment behaviour which are not solely based on statistical assumptions, although they are necessary. The market model is a statistical model, and an alternative statistical model could be the constant mean return model. In the constant mean return model the abnormal return is given by
i t i t i r r AR, = , − (5)
where ri,t is given by the expected return on day t and ri,t is the mean-average return
on stock i over the estimation period. This method means that there is no explicit control for the risk of the stock or the market portfolio return during the estimation period (Binder, 1998). Furthermore, this is considered the simplest model by Brown and Warner (1985) as only one parameter is estimated, namely ri,t, and there is no need for market returns.
However, while this will make the model more simple to use it does have a disadvantage. The event period market returns are not controlled for, only the previous stock returns. This means that there will be a greater variance in the abnormal returns that are calculated (Binder, 1998). Another alternative statistical models is the market adjusted returns model. In this model the abnormal return is given by
t m t i t i r r AR, = , − , (6)
assumption is made that each firm has the same average return and risk characteristics as the market. These restrictions make it that the market model is preferred, especially when information to use an estimation period is present.
An example of an economic model is the Capital Asset Pricing Model (CAPM). The CAPM model gives the abnormal returns as
)] ( [ , , , , ,t it ft i mt ft i r r r r AR = − +β − (7)
where ri,t is the expected return, rf,t the risk free interest rate rm,t the market return and βi
measures the market returns in access to the risk-free interest rate. The CAPM was mostly used in event studies in the 1970s, but the restrictions imposed by the CAPM have since been questioned. Fama and French (1996) discuss the pricing anomalies, which are the basis of this. The pricing anomalies are defined as the patterns in average stock prices, which are usually related to firm characteristics, but not explained by the CAPM model. Thus the CAPM model is not capable of explaining the anomalies, and therefore CAPM model may bias the results. Since the market model can avoid these problems the use of the CAPM model has reduced (MacKinley, 1997).
Overall, the market model used in this study is generally considered to be the best alternative (Binder, 1998). Brenner (1977) researched different specifications of the model, such as described above, and finds that the market model in (1) does as well as the alternatives. Brown and Warner (1985) confirm that the market model performs very well under a variety of conditions. Furthermore, the standard market model has the advantage that it takes both market trends and firm’s risk into account. The constant mean return model which does not take into account the market trend whereas the market adjust model does not take into account the firm’s risk individually. The CAPM model problems with pricing anomalies again speaks in favor of using the market model for this event study.
the return of a market model co-varies with the residual7. Scholes and Williams (1977) introduce an approach to correct for the bias in the beta. The Scholes-Williams approach means that lead, lagged, and contemporaneous estimates of beta are estimated separately, and then the weighted average of these estimates is used as the beta8.
Brown and Warner (1985) look at the Scholes-Williams approach to correct for non-synchronous trading for different levels of abnormal performance. They conclude that using this methodology does not lead to a significant difference in the results compared to using the OLS market model. To see whether the Scholes-Williams approach leads to different results in the case of Formula One sponsorship lies outside of the scope of this current study.
Another potential problem is autocorrelation. Autocorrelation means that the stock price on a certain day is related to the stock price on the previous day. When autocorrelation exists, this thus means that the assumption from an OLS regression that the error terms are uncorrelated is violated, and this leads to overestimated t-values. However, introducing correction methods does not appear to have an effect on the results. Brown and Warner (1985) study the properties of daily stock returns and how their characteristics were of an influence to event studies. In this study, Brown and Warner (1985) employ an adjustment for autocorrelation. The conducted research does not show significant changes in the results of the event study, leading Brown and Warner to conclude that “the benefits from autocorrelation adjustments appear to be limited” (1985, p20). This study includes Durbin’s alternative test for autocorrelation9 to check for autocorrelation in the stock prices. Durbin’s alternative test is used as it is a more general test for serial correlation which does not require all regressors to be strictly exogenous.
First, Durbin’s alternative test for autocorrelation is performed for new announcements. The p-values of the Durbin test can be found in table A.3 in the appendix and it shows that at the 5 percent significant level autocorrelation is detected for 13 firms10. Employing a correction for autocorrelation lies outside the scope of this study. Therefore, the firms where autocorrelation is detected have been deleted from the original data set and the event study is employed to the reduced sample for new announcements. This is not the most ideal option for dealing with autocorrelation but it offers a good comparison within the scope of this study. For the sample of renewals, Durbins’ test for autocorrelation is also performed
7 A detailed description can be found in Scholes and Williams (1977; 315-316) 8 See Scholes and Williams (1977) for an explanation of the procedure 9 See Durbin (1970)
and it detects 10 firms where autocorrelation is present at the 5 percent significance level (see table A.4 in the appendix for the p-values)11. These firms are deleted from the original data set as well, and the regression will be run again. Although the sample size of sponsorship withdrawals is already small, the same procedure with regards to correlation is still followed to see if there are any alterations in the results. Thus Durbins’ test for autocorrelation is performed and 4 firms were found to have autocorrelation problems at the 5 percent significance level (see table A.5 in the appendix for the p-values)12. These firms are deleted from the original data set and the regression was run again. The sample from which firms with stocks with displayed autocorrelation are deleted are referred to as the reduced sample.
3.1.2 Multiple regression
Next to researching whether there is an effect when a sponsorship announcement is made, this study also aims to explain which factors influence the reaction. The individual cumulative abnormal returns found in the event study for each firm act as the dependent variable. The independent variables are variables that can explain the cumulative abnormal returns. The regression is as follow:
CARi = α + β1 TITLESPONSORi + β2 AUTOMOTIVEi + β3 POINTSSCOREDi + β4 CASHFLOWi + β5 TOTALASSETSi + εi
Where i denotes the firm and
CAR: The cumulative abnormal return of a firm over the specified event period
TITLESPONSOR: Dummy variable, 1 if firms is a title sponsor and 0 if firm is a main sponsor
AUTOMOTIVE: Dummy variable, 1 if sponsoring firm is active in the automotive industry, 0 if the sponsoring firm is not
POINTSSCORED: The points scored in the Formula One championship by the sponsored team in the year before the sponsorship was announced
CASHFLOW: Denotes the cash flow per share reported by the firm on the day of the announcement
TOTALASSETS: Denotes total assets reported by the firm on the day of the announcement. All results are reported for the announcements that display significant abnormal returns. Only these returns are used as the regression means to explain the abnormal returns. When these are not present the regression has nothing to explain.
11 1. Philip Morris – Ferrari, 2. Philip Morris – Ferrari, 5. Shell – Ferrari, 7. Diageo – McLaren, 9. Japan Tobacco – Benetton, 11 Panasonic – Toyota, 13. Allianz – Williams, 14. Allianz – Williams, 17. Philips – Williams, 18. Philips – Williams
One possible violation of the assumptions of the OLS model is heteroskedasticity. To find out whether heteroskedasticity is present in the regression a test can be performed. The most used test for heteroskedastisticy is the Breusch Pagan test. However, as this is generally considered to be a large sample test and the samples here are only small this is not considered to be the best method. Instead, the more general White test is employed13. The White test does not assume a particular relation between the variance of the error terms and the variables. The White test tests whether the model is homoskedastic, thus whether the residual variance is constant. The results of performing the White test are shown in table A.6 in the appendix, these are the p-values of the test. According to the White test the regression with full sample re-announcements suffers from heteroskedasicity at the 10 percent significant level as the p-values for cash flow and total assets. To correct for this White’s robust standard errors are used for these regressions. That way the test statistics for the estimates are no longer incorrect in the presence of heteroskedasticity. The other samples do not suffer from heteroskedasticity and thus do no need the robust standard errors.
Another possible violation of the assumptions of the OLS model can come from collinearity in the model. Collinearity means that there is a linear relationship between two of the independent variables. To test for this a correlation matrix is produced. This matrix (table A.7 in the appendix) shows that the correlation is no higher than 0.5 for the explanatory variables thus overall the correlation is not considered to be very high. The model thus does not suffer from a collinearity problem. Collinearity measures the relationship between two variables, but there may also be a problem with more than two variables. Therefore a test for multicollinearity was also performed. The variance inflation factor (VIF) is an indicator of multicollinearity as it measures how much the variance of a coefficient increases due to the multicollinearity. The VIFs are displayed in the appendix in table A.8. A value above 2.5 for VIF might indicate a multicollinearity problem. The values for this sample are all below 2,5 thus the results show that there is no problem with multicollinearity in any of the regressions.
Next, the sample is tested for outliers. An outlier is an observation which has an extreme value of x or y. If an outlier is very large it can have a strong influence on the results. To check whether any outliers are present scatter plots with a fitted line are created for the variables points scored, cash flow and total assets with regards to the cumulative abnormal
13 See White (1980)
returns. The graphs shows that there could be a potential problem with outliers for cash flow and total assets as these display extreme values on the right (appendix graphs A.9). For the reduced sample of new announcements there are two extreme outliers for cash flow, namely Philip Morris-Ferrari and KDDI-Toyota. For the full sample of re-announcement there is an outlier in total assets, namely RBS-Williams . The reduced sample also has this outlier for total assets and another one, namely Japan-Tobacco Benetton, for cash flow. Removing them is a sensitive case as the samples are already very small and the outliers represent normally observed variables. However, compared to the other data points the observations are very extreme and are thus bound to influence the results for cash flow and total assets.
Lastly, the normality assumption is tested. The Jarque-Bera test measures deviations from normality by looking at skewness and kurtosis. Skewness measures how symmetric the residuals are around zero and kurtosis looks at whether the data is peaked or flat. The results of the Jarque-Bera test, the chi-squared results, are displayed in table A.10 in the appendix. The rejection region at 95% with 2 degrees of freedom is 5.991. For values smaller than 5.991 there is thus not enough evidence to reject the normality assumption at the 5% level of significance. This indicates that there is normality for the new announcements and for the re-announcements over the [-5,5] event window. The only sample for which the normality assumption can be rejected is the [-2,2] window for re-announcements. The results of these tests should thus be treated with caution.
4. RESULTS
4.1 Event Study
4.2.1 New Formula One sponsorship announcements
Table 5 reports the results of the event study over the 4 different specified time frames for the total sample of 40 firms reporting new sponsorship announcements. The results show that the abnormal returns are all positive, indicating the sponsorship announcement was considered to be positive but the results are not significant for any of the time frames specified. This indicates that the announcement of a Formula One sponsorship deal does not influence the stock prices significantly. The hypothesis that the announcement of a Formula One sponsorship deal leads to positive abnormal stock returns is thus rejected14.
Table 5: Cumulative abnormal returns
New announcement Re-announcement Sponsorship withdrawal
N Full sample (40) Reduced sample (27) Full sample (24) Reduced sample (14) Full sample (11) Reduced sample (7)
β p-value β p-value β p-value β p-value β p-value β p-value
[0,1] 0.0023 0.523 0.0069 0.113 -0.0002 0.958 0.0031 0.428 0.003 0.766 0.0017 0.854
[-2,2] 0.0064 0.318 0.0151 0.032** -0.0117 0.051* -0.0071 0.259 0.0287 0.275 0.0133 0.480 [-5,5] 0.0063 0.416 0.0118 0.157 -0.0175 0.012** -0.0209 0.025** 0.0271 0.461 0.0352 0.288 [-10,10] 0.0052 0.606 0.0081 0.462 -0.0079 0.263 -0.0138 0.201 0.0381 0.260 0.0382 0.470
A possible reasons for the insignificant results could be that the sponsorship deals involving Formula One teams are often subject to rumours long before the actual announcement takes place. The Spanish bank Banco Santander, for example, which is currently a major sponsor for the McLaren team has been rumoured to be moving their sponsorship to McLaren rival Ferrari for nearly a year while an official announcement is yet to be made (Sportweek, 2008). This means that as the sponsorship is expected it could be incorporated into the price. As a result, the announcement data has no significant abnormal returns as the announcement is already incorporated into the stock price when the rumours became strong.
These results are different from previous research on automotive sports such as NASCAR. A reason might be that he number of sponsors in NASCAR that have ties to the automotive industry are clearly higher than in Formula One, where sponsoring mainly came from Tobacco companies in the 1990s and early 2000s and then moved to banking. While there are some tobacco companies that sponsor NASCAR, no banking firms are involved in sponsorship. As argued in previous studies, the stronger the link between a sponsor and sponsored event the more positively the sponsorship is perceived (Meenaghan, 1983; Otker and Hayes, 1987. Furthermore, previous studies (Clark, Cornwell and Pruitt, 2008; Pruitt, Cornwell and Clark, 2004) proved that congruent sponsorship leads to higher stock returns. The low amount of congruent sponsors in Formula One could thus explain why no significant returns are found.
Next, the test is repeated for the reduced sample of 27. This is the sample that remains after the stocks which showed autocorrelation with Durbin’s alternative test are removed. With new announcements the results remain positive over all event windows but there is now a significant result over the [-2,2] event period at the 5 percentage level. Autocorrelation violated the OLS assumption of uncorrelated error terms which means that standard errors are underestimated and t-statistics overestimated, this is thus an opposite results of what was expected. However, as has been discussed previously, autocorrelation appears to give little problems in these sort of event studies (Henderson, 1989). The change from no significance to significance may just simply be due to the change in sample.
team has a positive effect on the value of the firm and the hypothesis that the announcement of a sponsorship agreement with a Formula One teams is positively associated with abnormal stock market returns can not be rejected for this sample.
4.2.2 Renewals of existing Formula One sponsorship contracts
Next the 24 re-announcements of existing sponsorship agreements are also investigated. The results of this are also presented in table 5 over the same specified time frames as for the new announcements. The results show that the re-announcement of existing sponsorship agreements exhibit negative abnormal returns that are significant for the time period [-2,2] at the 10 percent significance level and for the time period [-5,5] at the 5 percent significance level. This means that a renewal of an existing sponsorship agreement is thus viewed as clearly negative by the company’s shareholders and the hypothesis that there is a positive association between an announcement of a sponsorship renewal and abnormal returns is thus rejected. Previous research from Clark, Cornwell and Pruitt (2008) finds insignificant results, apart from significant negative results for PGA sponsorship. These results are in line with Shi and Gosh (2005) who find negative and significant results for NASCAR.
The renewal of a sponsorship agreement is thus viewed as not being a good investment, i.e. not worth the invested money. A reason for this could be the gradual decrease in interest for a sponsorship, as noted by Meenaghan (1983). This decreasing interest leads to decreased exposure for the sponsor. While the firm may still perceive benefits, the investors may be worried about the decreasing interest in the sponsorship. Another reason could be that the shareholders view is that the benefits from the sponsorship have not reached them. This could be because of agency problems. The investors see managers profiting from the hospitality benefits that Formula One teams offer while they may not see the same benefits for the firm as a while.
Next, the event study is employed to the reduced sample of re-announcements. The results are also in table 5, and they still show a significant negative abnormal returns for the [-5,5] event window. The hypothesis that there is a positive association between an announcement of a sponsorship renewal and abnormal returns is thus again rejected.
4.2.3 Withdrawal from Formula One sponsorship
are not significant rejecting the hypothesis that the withdrawal from Formula One sponsorship leads to positive abnormal returns. This thus indicates that the withdrawal from a sponsorship with a Formula One teams does not influence the share prices significantly.
Next, the event study is employed to the reduced sample for withdrawal announcements. The results are also in table 1, and they again show positive but not significant results. The hypothesis that there is a positive association between an announcement of a sponsorship withdrawal and abnormal returns is again rejected.
4.2 Regression Analysis
4.2.1 Summary statistics
Table 6 gives a the summary statistics for each variable over the different event windows. This shows that the sample has approximately as many title as main sponsors, but that congruent sponsors do not appear to have a large presence in Formula One sponsorship. Indeed, Formula One title sponsorship was dominated by Tobacco sponsorship till the EU introduced a ban on the sponsorship and since then the banking industry has played a dominant role. The lack of congruent sponsorship in this sample of Formula One sponsorship might also be due to the fact that there are more congruent sponsors present than in the sample but these are not main or title sponsor but take a less prominent sponsoring role.
Table 6: Summary statistics
Table 7: Regression results
New announcement Re-announcement Re-announcement
Reduced sample (25) Full sample (22) Reduced sample (11)
CAR-2,+2 CAR-2,+2 CAR-5,+5 CAR-5,+5
Coefficient p-value Coefficient p-value Coefficient p-value Coefficient p-value
constant 0.0234 0.078 -0.0019 0.795 -0.0319 0.028 -0.1078 0.030 Titlesponsor 0.0247 0.117 0.1075 0.299 0.0391 0.011 0.0304 0.140 Automotive -0.0196 0.246 -0.0002 0.958 0.0207 0.101 -0.0038 0.866 Pointsscored -0.0001 0.394 -0.0000 0.222 -0.0000 0.691 0.0002 0.231 Cashflow 0.0037 0.356 -0.0002 0.038 -0.0002 0.011 0.0198 0.072 Totalassets -0.0000 0.051 -0.0000 0.434 -0.0000 0.984 0.0000 0.298 R-squared 0.3309 0.3119 0.3888 07975 Adjusted R-squared 0.1549 0.0969 0.1978 0.5950
New announcement Re-announcement Re-announcement
Reduced sample (25) Full sample (22) Reduced sample (11)
CAR-2,+2 CAR-2,+2 and CAR-5,+5 CAR-5,+5
Mean Minimum Maximum Mean Minimum Maximum Mean Minimum Maximum
Titlesponsor 0.3600 0 1 0.5455 0 1 0.5455 0 1
Automotive 0.2800 0 1 0.1364 0 1 0.1818 0 1
Pointsscored 42.9200 0 182 89.7727 0 262 75.4546 0 262
Cashflow 2.1485 -1.328 8.484 31.3380 0.328 224.658 2.8372 0.328 5.518
4.2.1 New announcements
The final results of the multiple regression analysis are presented in table 7. The coefficient for title sponsor is not significant thus rejecting the hypotheses that returns will be larger for title sponsors than for main sponsors. The abnormal returns confirmed that sponsorship is beneficial as a whole and whether the sponsor enters Formula One as a title or main sponsor does not have any effect on the returns from the sponsorship.
Opposed to prior literature (from example Pruitt, Cornwell and Clark, 2004) congruent sponsorship is not a significant factor in explaining the abnormal returns. The hypothesis that the abnormal returns are positively related to congruent sponsorship is thus rejected. It has been established before that congruent sponsors does not seems to have a large presence as main or title sponsors in Formula One when compared to for example IRL and NASCAR Shell, one of Ferrari’s main sponsor, reported benefits of 2 to 3 times the sponsorship costs indicated that there is a gain from being a prominent sponsor (Marketing, 2001). Furthermore, it could be that for firms that have ties with the automotive industry that the benefits of small sponsorship are already sufficient. Hence they do not see the value of becoming a main or title sponsor.
Points scored is not relevant in explaining the abnormal returns at conventional levels of significance. The hypothesis that the success of the Formula One team is positively related to abnormal returns is rejected. Previous results show mixed results as there are no significant results in a study on IRL (Cornwell, Pruitt and van Ness, 2001), but there are positive significant results in a study on NASCAR (Pruitt, Cornwell and Clark, 2004). Successful teams have more television exposure (Arthur, Dolan and Cole, 1998) but the costs of sponsoring a successful team are also higher than sponsoring an less successful team. It could be the case that investors thus do not perceive the benefits as the costs are also larger.
