<<

‘We’ Are the Champions: Examining the Effects of Same-League Rival Clubs in the European Football Industry by Mohamed Kerbek

A Thesis presented to The University of Guelph

In partial fulfillment of requirements for the degree of Master of Science in Marketing and Consumer Studies

Guelph, Ontario, Canada © Mohamed Kerbek, July, 2015 ABSTRACT

‘WE’ ARE THE CHAMPIONS: EXAMINING THE NETWORK EFFECTS OF SAME- LEAGUE RIVAL CLUBS IN THE EUROPEAN FOOTBALL INDUSTRY

Mohamed Kerbek Advisor: University of Guelph, 2015 Sergio Meza

In business, is associated with exchanging blows with a competitor in order to win customers. Rarely does a rival profit from the presence of another, and in such instances the benefits are based on a production side benefit, rather than increased marketability. In the sporting industry, the business between rivals seems to defy that logic. Rivals seem to profit from the presence of another without any locational antecedent. This paper aims to examine the impact of sports rivalry on club financial performance from a spatial and general rivalry perspective. It utilizes multi-linear regressions to determine the impact of having a rival on a team’s year-end finances. The results found significant support for rivalry in general and competitive balance. No support was found for geographic . Important implications for competitive strategy theories in the sporting industry, and decision-making criteria for practitioners. Dedication To my dad, May God Bless his soul, who I wish was here for my defense and my graduation. You will never be forgotten. To my mom and aunt, the strongest most enduring humans I have ever known, thank you for being so patient, understanding, and forever giving. To my friends who stuck by me in what I can only describe as one of the most challenging periods of my life, I cannot sufficiently voice my appreciation. This is for all of you. Acknowledgements I would like to thank my advisor, Professor Sergio Meza for his support and help in shaping my passion for this beautiful game into a discernable thesis topic. Thank you for providing me with the direction and know-how to proceed with this paper, and thank you for your patience with and belief in me. Also I would like to take this opportunity to express my sincere gratitude towards Professor Paul Anglin and Professor Vinay Kanetkar, the esteemed members of my advisory committee who helped me infinitely in structuring this paper and refining my thought to make this paper what it is today. I thank you for your passion, your valuable insight and your thought provoking conversation. I would like to thank Professor Towhidul Islam and Professor Timothy Dewhirst for their role in guiding me before I set foot in the University of Guelph, and their valuable advice and belief in my ability to fulfill the requirements of this program. Equally, I want to extend my gratitude to Professor Karen Finlay and Professor Theodore Noseworthy, both who encouraged me when I felt the challenge was too great, and for their constant advice and guidance. Thank you to soon to be Dr. Aseel Al-Ghamdi, who offered me valuable insight into my work as a fellow sports researcher and football enthusiast, and an ever-supportive friend. To Professor Tanya Mark, who chaired my Seminar and contributed valuable insight into the conversation at both the seminar and the Brown Bag presentation where I first presented this idea at the University of Guelph, thank you very much for your efforts, they are genuinely and wholeheartedly appreciated. To the esteemed staff in the Marketing and Consumer Studies Office, Rita Raso, Dominica Alderton and Cori Wells, I would not have gotten here without your constant help in my time here. I cannot tell you how much I appreciate it, or how much I am indebted to you. It has been a long ride, and I have been very fortunate to encounter so many remarkable people along the way. For those who are not mentioned in this section, pardon my spatial constraints but believe me when I say that I am every bit as grateful, and I definitely would not have been able to do this without you. iii Table of Contents

Chapter 1.0 Introduction...... 1 1.1 Preface and Motivation………………………………………………………...……………...1 1.2 Introduction to Football...... 7 Chapter 2.0 Literature Review ...... 12 2.1 Competition & Competitive Cooperation in Business ...... 12 2.2 The Curious Case of the Sporting Industry ...... 17 2.3 The Impact of Sports Rivalries on Demand ...... 19 2.4 Bridging the Gap ...... 22 2.5 How to Measure the Impact of Rivalry on Club Financial Performance ...... 23 Chapter 3.0 Empirical Framework ...... 25 3.1 Financial Success Metrics ...... 25 3.2 Revenue Streams ...... 26 3.2.1 Classifying Revenue Streams ...... 26 3.2.2 The Scope of Revenue Streams ...... 27 3.3 Demand Drivers ...... 29 3.3.1 Classifying Demand Drivers ...... 29 3.3.2 Scope of Demand Drivers ...... 35 Chapter 4.0 Methodology ...... 38 Chapter 5.0 Results ...... 48 Chapter 6: Conclusion & Discussion ...... 78 6.1 Summary of Findings ...... 78 Geographic Rivalry Hypotheses ...... 78 6.2 Contributions/Implications ...... 79 Managerial ...... 80 6.2 Limitations ...... 81 6.3 Future Research Considerations ...... 83 Bibliography ...... 85 APPENDICES ...... 89 Appendix A: Stakeholder Analysis: Consumers of the Football Product ...... 90 Appendix B: On Sources of Revenue ...... 91 Appendix C: Descriptive Statistics ...... 93 Appendix D: Study Results ...... 94 iv List of Tables

Table 1.1 Same city clubs represented in Market Value and Operating Revenue Ranks………………………………………………………………………………………………….P. 2

Table 3.1 Revenue Stream Scope ………………………………………………… P.27 Table 3.2 Classification of Demand Drivers……………………………………….P.23 Table 3.3 Demand Drivers & Their Corresponding Revenue Streams…………….P.36 Table 4.1 Variable Class & Operational IV………………………………………..P.48 Table 5.1 Numerical Impact of Each IV on DVs…………………………………..P.38 Table 5.2: Adjusted R-Square Values across models for all DVs…………………P.50 Table 5.3: Impact of 1,000 change in population total & match-day revenues……P.51 Table 5.4: Impact of Change of £1,000 in per capita GDP on total & match-day revenue …………………………………………………………………….………………..P.52 Table 5.5: League Rank Coefficient values & impact levels across models for all DVs ……………………………………………………………..……………………….P.54 Table 5.6: Impact of League Rank Coefficients on Commercial Revenue………..P.54 Table 5.7: UCL Participation Coefficient values & impact levels across models for all DVs…………………………………………………………………………………P.55 Table 5.8: EL participation coefficient values & impact levels across models for all DVs ………………………………………………………………………………………P.57 Table 5.9: Games played coefficient values across models for all DVs…………...P.57

Table 5.10: Games Played Coefficient Values and impact levels across Total & Match-day Revenue DVs ……………………………………………………………………….P.58 Table 5.11: Games Played Coefficient Values and impact levels across Television Revenue DV………………………………………………………………………...... P.59 Table 5.12: Transfer Expenditure Coefficient Values………………………………P.60

Table 5.13: Impact of a £1 mil change in transfer expenditure on total revenue…....P.60 Table 5.14: Transfer expenditure impact on Television and Commercial Revenue..P.61

Table 5.15: Impact of 1,000 change in average attendance on total and match-day revenue…………………………………………………………………………….....P.62

Table 5.16: Impact of 1,000 change in stadium capacity on commercial revenue….P.63 Table 5.17: Presence of a Local Rival coefficient value on all DVs……………..…P.64 v

Table 5.18 Distance to Nearest Rival coefficient value on all DVs…………………….....P.65 Table 5.19 Presence of a FIFA-determined Rival coefficient value on all DVs….…….....P.66 Table 5.20 Presence of a FIFA-determined Rival impact on Total and Match-day Revenue …………………………………………………………………………………..………….P.68 Table 5.21 Presence of a FIFA-determined Rival impact on Television & Commercial Revenue…………………………………………………………………………………...... P.79 Table 5.22 3Yr Avg Goal Difference with nearest competitor from 1 coefficient value on all DVs………………………………………………………………………………...... ,,,,,,,,,,,P.70 Table 5.23 3-Yr Avg Goal Difference with FIFA-determined rival from 1 coefficient value on all DVs………………………………………………………………………………...... P.71 Table 5.24: Rank Difference from FIFA-determined Rival coefficient on all DVs……...... P.72 vi List of Figures

Figure 3.1 Comprehensive Empirical Model…………………………………….P.37 1

Chapter 1.0 Introduction

1.1 Preface and Motivation

How should we feel about rivalry, and just how healthy is a little competition? Well, the

answer to that depends on what you are competing for. Traditional business literature

emphasized that in markets where competing companies offer the same products to the same

consumers, the more intense the rivalry between competitors, the greater the constrain on market

share and profitability (Porter, 1985). Companies competing for the same customers are in a

fixed ‘pie’ scenario, where the market size remains relatively the same, and consumer choice will

deviate towards one or the other.

Interestingly enough, the same intuition does not seem to apply to the sporting industry.

Worth an estimated $80 billion per year (Kearney, 2012), the dynamics this industry subscribes

to appear to be different. As a matter of fact, the financial metrics data for top European soccer

(football) clubs shows an overrepresentation of clubs with same-city competitors (or rivals):

Table 1.1 Same city clubs represented in Market Value and Operating Revenue Ranks

Year % Club in Same City % Club in Same City in % Club in the Same City (UEFA) Forbes Market Value in Deloitte Operational Rank Rev. Rank 2013 21% 35% 35% 2012 22% 35% 35% 2011 24% 45% 47% 2010 19% 35% 43% 2009 18% 36% 35% 2008 15% 40% 35% 2007 21% 44% 30% 2006 20% 36% 40% 2005 17% 40% 55% 2004 20% 35% 40% AVG. 20% 38% 41% 2

The factors affecting this disparity were not found to be demographic in nature, with cities like Manchester and Dortmund (populations of sub 1,000,000) are overrepresented, while cities like Paris and Berlin rarely feature over the past 10 years. It is also not a function of income that cities with rivals tend to thrive; Milan’s GDP per capita is lower than Rome’s and yet Milan consistently features two clubs, while Rome features one. The effect is also independent of production-side benefits, where limited numbers of clubs on the list share facilities that maybe allow them to focus elsewhere and thus generate higher ROIs.

In sports, the presence of same-city competitor is known as a cross-town rivalry. In the absence of exogenous explanations for this phenomenon, the data summarized in the table above seems to suggest that the presence of a sporting rival actually improves the financial performance of teams. It seems to be the case then, that rivalry increases the total ‘pie’ in a market that is not a zero-sum game. That the size of the total market is increasing in itself is not subject to doubt,

Deloitte’s data shows significant growth year after year in total revenues earned by football clubs, and not just because of the increase in ticket prices; match-day revenues have been comprising a declining portion of a club’s average total revenue in the past ten years (Deloitte,

2013). How significant then, is the role of rivalry on the income of a sporting institution, and how much does it contribute to this growth? Is this effect unique to cross-town rivals, or is it applicable to all forms or rivalry?

When the characteristics of the sporting industry are analyzed, it seems that traditional business views are not necessarily applicable (Neale, 1964). In the unique situation where two competing firms must compete in order to offer the end product to the customer, the demand for that product will impact both said firms (Neale, 1964). Known as the inverted product joint, this influenced academics to argue for the treatment sports leagues (formalized competitions between 3

sporting firms or clubs) as a singular firm rather than a set of separate firms (in keeping with the

NBA, NFL and NHL franchising models, for example), albeit that these companies have

different appeals to consumers and largely acquire their own customers or fans (Sloane, 1971). The

quality of the product on offer to the customer is contingent on the performance of both teams.

Sporting literature characterises the quality of the sporting product through examining

consumer motivations to watch sports. The uncertainty of the outcome, and the excitement that

comes with that uncertainty are significant factors, as is the amount that is at stake; it is logical to suggest that a championship game will generate greater interest than a regular season encounter, and data supports that intuition (Nielsen, 2011). Adding to the stakes and increasing affect is

rivalry, which in sporting literature is defined as a state of intense “affective disposition” from

the players, staff and/or fans of one sporting institution towards another (Szymanski, 2008). With

more emotion hanging on the outcome, a rivalry game is more likely to generate consumer

demand (Mason, 1999) (Szymanski, 2008) (Brugging and Roosma, 2011).

However, while examples have demonstrated that rivalry games are in greater demand,

the data presented in the table above suggests a greater; that the significance of these games

(which are traditionally two games at most out of a total 34 games at the very least) is enough to

impact financial performance across the entire year. Do top performing teams with rivals then

draw greater attention that carries forward across the season, or is it purely a coincidence that

teams with rivals are overrepresented in ’s football elite? The research questions that this

begets are more specifically posited below:

Do clubs derive financial benefits from having a rival club in the same city?

4

In itself, this question posits interesting follow-ups. Rivalries vary in type, as there are

various antecedents to a rivalry (those include historical, political and religious – for more information, refer to section 2.3 in the literature review). The questions posited are whether this effect, if existent at all, is unique to cross-town rivals:

Is this effect present for cross-town rivals, or is it generalizable across rivalries as a whole?

The existing literature pointed the virtues of rivalry’s impact on clubs in terms of

marketability and ticket sales for rivalry games, and highlighted the impact of rivalry on fan

enjoyment and involvement from a consumer standpoint. Much of what makes the rivalry what it

is also embedded in a state of competitive balance; the two rivals have to be performing within

proximity to each other, at least in rivalry matches (Mason, 1999) (Szymanski, 2008). Two

critical factors were lacking: 1. The impact demonstrated was for one particular encounter; in

other words there was no literature pointing to the fact that clubs who have rivalries are perform

better financially 2. There was no specific reference given to cross-town rivalry that leads us to

believe that there is an effect unique to same city rival clubs as opposed to rival clubs as a whole.

One particular challenge to this notion is that the there has not been a method to quantify

rivalry in sports; there is no rivalry metric or measurement, rather it is a state present in fans’

mindset where no measurable correlation between any particular antecedent, proxy or outcome

and that mindset has been found. Therefore, defining what a rivalry is in terms of football would

either require extensive ethnographic research that is beyond the time and budgetary

considerations for this research, and therefore, there was a reliable, albeit limited measure that

can still operationalize this research to test for the impact of rivalry on year end financial

performance. The world governing football body, FIFA, presents rivalries within each country’s

league (for more information, refer to section 1.1). This is limiting from the perspective of

5

identifying particular types of antecedents, as the research is now restricted to testing for

geographic and cross-town rivalries, and other forms of rivalry bundled as one. If the type of

rivalry were indeed significant, then other rivalry forms would run the risk of diluting the results.

However, since there is little from the literature to suggest that one particular form of rivalry is

the one that can have this sort of impact on financial performance, the research methodology still remains acceptable.

If the observations in Table 1.1 indeed pointed to a greater picture, and there was a correlation between the presence of a rival (or the presence of a closely performing rival), then the impact certainly extends beyond what is two games in a season that would consist of no less than 38 games for most clubs (for more information on leagues and fixtures, refer to section 1.1).

To address the aspect of competitive balance, it would also be expected that how the rivals perform relative to the other rival (rather than just overall) would be a factor that must be incorporated into the analysis. This points to an obvious gap that this research is trying to fill, and the hypotheses pertain to the states of rivalry and competitive balance:

H1. Geographic rivals will have a positive impact on year-end financial performance

H2. FIFA classified rivals will have a positive impact on year-end financial performance.

The hypotheses above break down into more specific aspects pertaining to competitive balance (refer to Section 4), and geographic rivalry takes on two forms: cross-town rivalry where teams are located in the same city, and rivalry by distance (nearest team, another rivalry antecedent – refer to section 2.3), and the rivals as pre-classified by an authority on the matter.

In order to assess the impact of sporting rivalry on a club’s financial performance, it is important to put rivalry within context to other factors proven to impact a club’s financial

6

success. Literature has identified a group of factors that include socio-demographic elements,

aspects of a clubs operational strategy including stadium capacity, ticket pricing, the type of

personnel recruited as players or staff, and aspects of the club’s on field performance (for more information, refer to the empirical framework in section 3). These factors can only corroborate to particular revenue streams (refer to empirical framework in section 3), with the scope of socio- demographics for example only relevant to revenue streams accrued from local sources. When direct variables could not be found, proxies were developed to measure aspects fundamental to a club’s financial success.

The methodology of research selected was a multiple linear regression (refer to section

4), which has been used previously in sports to measure the impact of aspects pertaining to clubs’ financial success on year-end metrics such as ticket sales or revenue (Sloane, 1971)

(Staudahar, 1986). The regression would have to incorporate the aforementioned success factors and a club financial performance metric (year-end) as a dependent variable.

The measurement of impact on financial performance was restricted to a club’s operating revenue, with data provided from Deloitte’s Money League, an annual survey that presents

Europe’s top 20 clubs by revenue from football related operations. The selection of what is a non-random dataset (i.e. The elite institutions or the top performers in the area) posits a restriction on the generalizability of the results; should the data show an effect on revenue, it will be only applicable to the elite performing clubs of the European football industry rather than across all conditions for all countries. The way Deloitte breaks down revenue streams (refer to section 4) means that the impact of rivalry can be tested on every operational facet of clubs’ revenue generating activities: commercial income, television income and match-day income

(Deloitte, 2003).

7

There were six regressions deployed for each revenue stream DV; one that included FIFA

FIFA-determined rivalry elements, another that incorporated geographic rivalry, and one that

incorporated both. Those three combinations were deployed with or without their respective competitive balance variables (the results between the rival teams – for more information, refer

to section 4).

The results found that FIFA appointed rivalries (or thus, rivalry in general), had a

significant effect on revenue. The competitive balance variables signified a significant effect of

past rivalry games with a closely performing rival on revenue, where the greater the proximity of

the result between the teams, the greater the year-end revenue accrued.

What these results point towards is that rivalry does indeed have an impact on clubs’

financial performance, at least with respect to Europe’s top performing football clubs. The

competitive balance variable in particular suggests that investing in a rival’s well being could

return better results than investing in one’s own if the disparity in the performances is too great,

given that competitive balance is a healthy contributor to revenue.

Theoretically, these results show a greater need to investigate rivalry in sports on much

greater scale. While it supports the intuition that rivalry does indeed contribute to market growth

of the sporting industry, more research is required to validate the fact and ascertain conditions in

which this is applicable. They point to indicators that rivalry in sports can lead to counter-

intuitive results that are in contrast to business strategy literature, and further demonstrate an

extension of rivalry literature into the sporting arena. More than anything however, this points to

the urgent need to measure or quantify sporting rivalry in order to best operationalize it.

8

Pragmatically, such a finding would leverage club investment decisions and operational

policy, as well as governance policy from the league – where disproportionate fixture allocation can be viewed as favourable in revenue-sharing leagues (refer to section 1.1). It also leverages investment decisions on the behalf of consortium takeovers (a common feature in the football business landscape in recent years), and further informs sponsor and broadcaster investment decisions.

The section below introduces football as a sport in the context of regulations and governing bodies, and provides further insight into its league structures and regulations, which the empirical framework for this research considered in its application. The literature review

(Chapter 2) previews the relevant academic work on the subject matter, and extends into the empirical framework (Chapter 3) that previews factors affecting the financial success of football clubs. The theoretical framework also explicates the inclusion of rivalry as a testable hypothesis, and qualifies the methodology used in Chapter 4. Chapter 5 previews and explicates the study results, while Chapter 6 analyzes the contributions and limitations of this dissertation, along with recommendations for future research.

1.2 Introduction to Football

Football, or soccer as it is known in North America, is a team sport that is played with two competing teams with eleven on-field players that play against one another over the course of ninety minutes with the target of scoring more goals into the opponents’ net in order to win the game (FIFA, 2013). It is the most popular sport in the world, with 3.3-3.5 billion fans worldwide (Stølen, Chamari, Castagna, & Wisløff, 2005) (Deloitte, 2014).

9

Football events regularly feature on the top of the list of viewed television broadcasts

(Deloitte, 2010), soccer clubs and players frequent the most valuable sporting brands and highest

earning athletes lists (Forbes, 2014) (Forbes, 2013), and the FIFA World Cup, a once every four

years competition where national soccer teams compete against each other for what is the

world’s most renowned and prestigious team soccer award is the most anticipated and popular

sporting event in the world (FIFA.com, 2014). Football’s major competitions are comprised of

national competitions, which include global competitions such as the World Cup and the

Confederations Cup and continental competitions such as Europe’s EURO competition and

Africa’s AFCON (FIFA, 2014). However, soccer also has club competitions that are exclusive to

the domestic area, and continental club competitions that group top performing club teams from

each country to face their continental counterparts (UEFA, 2014).

Brief History and Background

The set of laws that the most defining in the formation of as we see it

today were primarily influenced through the efforts of the -based Football Association in

1863 (FA, 2007). The formation of association football, where clubs would play one another in

either one-off exhibition games, or compete in with diverse competitive regulations

and structures ensued (FA, 2007).

The first structured local league was developed in England, widely known as the home of

football. The league format entailed that teams would play one another home and away with

varying points awarded to wins, draws, and typically none to losses. The rankings of the clubs

and the league winners would be determined by the points accrued at season’s end.

10

Today, almost every country or region has its own professional soccer league

competition. They typically operate under the umbrella of a continental governing body and the

international football governing body, Fédération Internationale de Football Association

(commonly known as FIFA) (FIFA, 2009). Each competition has its own governing regulations,

varying tiers and structures, but most conform to the international standards of awarding points

based on results and measuring rankings subsequently.

League Structures and Models

Most sporting associations or leagues utilize a league structure or model in which teams

compete against each every opponent in the same league multiple times (at ‘home’; whereby the

one team plays host to the other and the event is held in the stadium in which the team is based

and ‘away’ which is held at the opposing team’s turf) and rankings are made as a function of the

wins and draws accrued, with differentiating factors on points levels including head to head

records and goal differentials (goals scored minus goals conceded). Various forms stem from that

basic model, where teams either proceed to a knockout round format of varying types; opponents

can either play a best-of series, where the first team to accrue a pre-agreed number of wins

proceeds to the next round, a one-time game with the winner proceeding, or a home and away

format where the aggregate score-line is computed to determine who progresses – and knockout

formats are typically ultimately determined by the final game(s), where the last remaining two

teams in the competition are paired together with the opportunity of winning the honors in

question (Sloane, 1971).

The relegation and non-relegation structures are another point of difference; in relegation

league models bottom ranking teams at each season end are relegated to a lower tier of the

league (the England based Football Association, for example, has 4 tiers of professionally

11

competitive football) and top performing teams in the lower tier are promoted to take their place.

This model is more popular in European sports leagues and less utilized in North America

(Zimbalist, 2003).

12

Chapter 2.0 Literature Review

2.1 Competition & Competitive Cooperation in Business

Competition in an industry continually works to drive down the rate of

return on invested capital toward the competitive floor rate of return, or the

return that would be earned by the economist's "perfectly competitive"

industry. (Michael E. Porter,1985)

The majority of business strategy literature that touches on competition suggests that

competition is a negative influence on company’s financial performance. In contrast to the

situation of a monopoly, where a supplier can sell a product at a much higher price than their

marginal cost (Colander, 2004), competitive businesses will comparatively have their

profitability. Competition is one of Porter’s five forces that influence strategy, with Porter

postulating that competitive intensity varies negatively with profitability (Porter, 1979). This

claim is upheld and treated as an assumption (Schmidt, 1996). Competitors are viewed as a

constraint on the revenue of a business and its probability of survival, and a dynamic equilibrium

between competitors is constrained for the individual competitor by the presence of competition

(Henderson, 1983). Competitors typically ‘jockey’ and venture into exchanging strategic blows

to one another, with strategies and tactics such as price wars, entry barriers and advertising

campaigns deployed in order to gain an edge over the competition (Porter, 1979).

The state of the competitive system has it such that an individual competitor thrives by

creating and sustaining advantages over the competition (Henderson, 1983). The principle of

sustainable competitive advantage, introduced and forwarded by Michael J. Porter, is based on

the ability to provide a unique and sustainable offering utilizing a variety of resources. The

13

objective of this offering is to ensure that the competitor cannot match it, and therefore it

provides a unique market access to a customer segment that desires that very offering.

Competitive business practices are generally geared with the aim of nullifying the relative

weaknesses of the firm to its competitor and creating and sustaining comparative strengths that

provide unique value to the customer, or by providing lowest possible cost (Porter, 1985).

While literature on competition in business is abundant, there is also a significant stream

of literature on cooperation between business rivals. Facility sharing and cooperation along the

value chain was found to assist companies from a cost cutting perspective. Sharing facilities for manufacturing, for example, reduces overhead and step costs in the manufacturing process, thereby increasing the profit margin (Ghoshal, 1987). Cooperating between competitors along the value chain has been proven to improve revenue generation. Co-opetition, a term coined to describe competitive cooperation, is a developing concept that encompasses competing business entities sharing their resources in order to generate a better bottom line. Airline companies have formed alliances centered on customer loyalty programs, where rewards points are transferrable from one airline to another; this is done with the purpose of expanding the reach of flight offerings and incentivizing consumer loyalty towards the participating companies when selecting a service provider for a particular destination (Zea & Feldman, 1998) (Lederman, 2007). In similar vein, hotel chains and restaurants aligned their interests through the usage of joint reward systems that incentivize consumer choice towards a particular set rather than the entirety of the market (Al Khattab, 2007).

14

Spatial Competition in Business

The existence of a same-market competitor has been found to constrain profit by virtue of

increasing competitive intensity (Porter, 1979). It is viewed that a competitor that is providing

the same offering within a given market will negatively impact profitability because it provides

consumers with multiple alternatives at similar transit costs for the producer (Henderson, 1983).

Early spatial models in business, such as the Hoteling model (Hoteling, 1929), introduce the idea

that businesses will take competitive measures in terms of location when they are catering to a

market in which a competitor exists. Traditionally the Hoteling model is a 2-player model, but

extensions have been made to simulate Hoteling’s model with more than two players. The result

is the same, with an equilibrium price point impossible. Porter’s logic on the competitive forces

and subsequent work with regards to business competition is in good agreement with these

findings (Porter, 1979).

However, an entire stream of research focuses on the cooperative and network benefits of

businesses in the same industry locating within proximity of one another. The concept of

companies in the same industry locating within proximity was first discussed by Alfred Marshall

in his book Principles of Economics (Marshall, 1890). Marshall demonstrates that businesses that

are involved in the production of a similar product were located in geographic clusters (Marshall,

1890). He argued that in conditions where companies are then able to optimize efficiency and

reduce transaction costs, with the recruitment of skilled labor and the exchange of knowledge

becoming significantly easier and faster (Marshall, 1890). The underlying assumption behind

this was that companies were small, produced minimally differentiated products, and were highly

competitive (Marshall, 1890). The principal of minimal differentiation was forwarded by

Boulding (Boulding, 1963), but the context focused on competitive intensity rather than the

15

network benefits gained. However, a subsequent stream of literature in geographic economics

and business strategy were in good agreement with Marshall’s findings.

From a surplus theory perspective, agglomeration literature finds that there is significant

network output merit to industry players locating within proximity (Papageorgiou, 1979).

Advantages of agglomeration include lower transaction costs and an increased rate of knowledge transfer, with externalities external to the firm including an increased production-side network benefit from locating within proximity to same-industry players (Henderson J. V., 2001)

(Gordon & McCann, 2000). The scale externalities vary directly with the number of industry

players within the area (Krugman, 1993).

Key to the assumptions behind agglomeration economies and cluster theories is that the

markets being competed for extend beyond the immediate market place in question – which is

similar in the sporting context (Baimbridge, Cameron and Dawson, 1996). With the growth of international economies, and the transit costs decreasing rapidly, the location of a business has become less impactful its profitability (Porter, 1990). In another sense, spatial competition’s

downward pressure on profit was playing a less prominent role for bigger businesses in particular

(Porter, 1990). Businesses along the same industry cluster together (Porter, 1990) and derive

substantial benefits in doing so. Clusters, or agglomerations in geographic economics tend to

profit from the end of the production function. The network effect of adding a company to the

cluster implies a reduction in transit costs, a downward pressure on other companies along the

value chain and the gravitation of other areas of the production function (Porter, 1990).

Porter cites the California wine industry as an example of a cluster, where companies

along the value chain of production locate within proximity and extrapolate an increased value

from being located within proximity together (Porter, 1990). Krugman demonstrates the network

16

effects gained from the US manufacturing belt, which was concentrated in the Northeast and the

Midwest, and still maintained approximately 65% of manufacturing employment in the United

States (Krugman, 1991).

Given that they would be competing for the same markets irrespective of where they

locate, locating the same area allows businesses to exploit pre-existing infrastructure that better

allows for the production function (Porter, 1990). Consider the example of Silicon Valley, where

the tech industry personnel are gathered in close conjunction along with several companies,

investors, relevant media outlets and research institutions. Technology giants will gravitate

towards the area given the increased access to production inputs and collective cooperation with

their competitors over the value chain (i.e. applying pressure on value chain producers).

The extensive work demonstrating the benefits of competitors locating within a same

market space (especially when the market they are catering to expands beyond the spatial area in

which they are located) is focused on value chain exploitation and production side network

benefits. The sport industry it appears, subscribes to different network dynamics. The

disproportionate representation of same city clubs in the Deloitte Money League Revenue rank

indicates that the benefit gained is not from a cost saving or transactional efficiency perspective.

Clubs recruit players and scout competitively, sell tickets at different price points through their

own vendor systems, have separate endorsement deals with global sportswear manufacturers and

recruit personnel from a global marketplace. While there is some evidence of facility sharing, it

does not account for revenue increase (it would primarily point towards cost savings, and while it

can be argued that the increased efficiency allows for greater investment in revenue generating

activities, but less than 10% of Deloitte’s money league clubs with same-market competitors

actually share facilities with that competitor).

17

The positive network effect of having an additional supplier, therefore, appears to be

independent of value chain gains; the cluster benefit derived in this industry is not one derived

from the presence of infrastructure that supports the production function. If there exists a positive

network effect from having a club operate in the same market, it would be derived from the

demand side. This becomes a more feasible explanation upon examining how sporting rivals

operate.

2.2 The Curious Case of the Sporting Industry

Unlike most industries, the product offered by the sporting firm is co-dependent on its

competitors. Given that most sporting events involve a match-up between teams or individuals,

the quality of the product on offer is also dependent on what the rival offers. It is for this reason

that the principle of the inverted joint production, where two separate processes are combined in

order to produce a singular product (Neale, 1964). The nature of this intertwined connection

means that a club cannot deliver a quality offering without its rival, and implies that demand

would then be contingent on the joint success of the clubs in their respective processes. In

leagues that size from 16-20 teams (in addition to competition in continental competition), there

exists the possibility that every two competing teams are treated as rivals. It is important to

distinguish rivals in this context, where a special animosity is derived between two teams and fans feel a particular hatred towards the opposing team and its fans (Lee, 1985).

It follows intuition then, that the presence of rivalries, whether in a spatial context or a

rivalry context in general, would increase consumer utility. The following sections detail what

the sporting institution offers by way of value to customers, and the role that rivalry plays in that

value offering.

18

The Value Offering of the Sports Organization

Like other forms of entertainment, sport offers a utopia, a world where everything is

simple, dramatic and exciting, and euphoria is always a possibility ... Sport

entertains, but can also frustrate, annoy and depress. But it is this very uncertainty

that gives its unpredictable joy their characteristic intensity. (Whannel, 1992

p.134)

The unique value proposition of the sporting product is entertainment (Whannel, 1992), which is derived from the unpredictability of the outcome of the events, and fans derive utility from the excitement that is produced from watching the competitive matches (Euchner, 1993).

The stream of psychological literature on the matter suggests that utility can be derived from

identification with a team, sport, and an affiliation with the consumption community that is

comprised of the other fans (Mason, 1999) (Guschwan, 2011) (Sanford & Scott, 2014). The

value offering of a sports club can also apply to those who identify with a nationalistic or

regional loyalty, whereby one derives utility from the support of their local club or national team

in sporting events (Guschwan, 2011). Given that the intensity of emotional responses is an

important component of how fans derive utility from sports, games of substantial importance

increase the utility of the viewing consumer (Sanford & Scott, 2014).

From the above, it can be inferred that the greater the intensity of the competitive state or

rivalry between the two teams (firms), the greater the consumer utility, and existing literature

support this conclusion (Whannel, 1992) (Osborne, 2008). The role of rivalry then, is to add to

the excitement and the uncertainty of the outcome, and what is at stake.

19

2.3 The Impact of Sports Rivalries on Demand

There is a vast body of literature that attempts to capture the role of rivalry in the experiential utility derived by the consumer. A major part of the literature attempts to examine the rivalry from a demand perspective, and argues that rivalry stimulates local demand. Sloane argues that rivalries have a positive impact on the demand of the product, and cites empirical evidence that at least suggests that football clubs in multi-rival cities are “at least well accounted for in terms of population heads” (Sloane, 1971). It has also been suggested that firms invest in what is known as “rivalry capital” of their athletes, in an attempt to nurture and organizational culture and subsequent behaviors that would increase the intensity of sporting rivalries, which is believed to increase demand for the sporting product (Osborne, 2008). Research assessing the intensity of rivalries has also been conducted with independent variables comprising of ticket sales (indicating that rivalries would stimulate demand) (Sanford & Scott, 2014), and has found that rivalries have a positive impact on ticket sales (Furtonato, 2006) (Paul & Weinbach, 2013)

(Lemke, Leonard, & Tlhokwane, 2010) (Bruggink & Roosma, 2011) .

There is a significant stream of literature aiming to understand, measure and quantify sporting rivalries. The reasons behind the creation of rivalry vary significantly; rivalries have

formed with respect to performance success (Dreyer, 2013). The following section aims to

capture this literature in the context relevant to what this paper seeks to contribute; the role of

rivalry in sports enhances a firm’s financial performance.

The utility derived by spectators from rivalry is further increased by media amplification, the point where the matter becomes a ‘chicken and egg’ scenario; fans identify with rivalries without apportioning significant attention to their causes or nature. Consider the example of Real

Madrid and Barcelona FC, where the two football clubs share a deep ideological split; Barcelona

20

is a club synonymous with the region of Catalunya, an area seeking independence from Spain, while Real (Spanish for Royal) Madrid is synonymous with General Franco’s regime and derives much of its historical successes to his preference for the team. El Clasico, the famous

between these two teams, is now centered on the fact that Real Madrid and Barcelona are age old

enemies in constant competition for titles (Szymanski et al, 2008). This is particularly important in the context of the modern sports era, as the antecedent to a century old rivalry may be an important component for a broadcaster’s decision to allocate air time (Refer to Appendix A), but the affiliations of the international fan base – which comprise the bulk of television revenue

(Baimbridge, Cameron & Dawson, 1996) are made independently of their sympathies to such antecedents (Weis, 1986). Nevertheless, the presence of a rivalry, and a network effect of the fan

base that projects its intensity, is of significance in determining the broadcaster’s decision as a consumer (Refer to Appendix A).

It is found that motivations for team rivalry vary from geographic/spatial to political and historical, to performance and competitiveness for honors and even playing styles. The literature classifying rivalry antecedents or conditions that facilitate a rivalry can be classified into three major streams.

Types of Rivalries

Political

Rivalries have formed between teams based on the historic identity of what they represent (Raspaud & Lachheb, 2014). Deeply imbued rivalries have emerged as a result of economic competition between their regions (Dreyer, 2013), a class-based comparison (Raspaud

& Lachheb, 2014), and even conflicting religious affiliations for the team and its respective fan

21

base (Baimbridge, Cameron, & Dawson, 1996). Older clubs tend to be at the center of rivalries

due to the respective histories and encounters (FIFA, 2014).

Performance Rivalries

Contextualizing rivalries in comparative performance contexts has been a prevalent issue

in sports literature. However, the literature focuses on the comparative performance between the

two sides. A gulf in quality between one competitor and the other does not constitute an absence

of rivalry, but since quality is a central construct to the utility derived from the sporting product

it is important to understand that comparative quality is an endogenous construct to the intensity

of a rivalry. Results that are closer to each other, running on the assumption that fans seek a

slim/last-gasp victory where the outcome is uncertain (Whannel, 1992) (Euchner, 1993) (Lee,

1985), between teams ranked closer together are measures of quality, which is critical to the

relevance of a rivalry. The presence of a rivalry as FIFA and fans identify them, however (FIFA,

2015) is still central to this construct as it appears on the Deloitte observations noted in the

introduction.

Spatial Rivalries

Spatial rivalries are of particular interest to European football due to the differences in

fan preferences between Europe and North America. While North American sporting models

exhibit limited regional segments, European fans are more prone to regional affiliations and

loyalty as a result of their competitive leagues being divided by country (Fort & Fizel, 2004). As

a result, literature applying to rivalries in a spatial context is particularly relevant to European

sporting models.

22

There is a growing body of literature identifying the role of rivalries in the construction

of self-identity and the positioning of what sports consumers call ‘home’ (Kraszewski, 2008)

(Guschwan, 2011). The embedded sense of regional loyalties and inter-regional discourse is

evident in sporting literature (Lee, 1985) (Mason, 1999) (Euchner, 1993). The exposition of fans

and identification with one team or the other is central to intensifying hostility between rivals, and therefore it follows intuition (and FIFA’s evaluation) that football clubs operating within the

same league are more subject to develop a rivalry (given the frequency of encounters).

Furthermore, there is another stream of literature found to demonstrate the role of spatial rivalries in the context of fan network effects. The greater the proximity of fans and the higher the potential for their interaction, the likelier it is for a rivalry to emerge (Krazewski, 2008).

2.4 Bridging the Gap

This research is attempting to fill an obvious gap. Extending Porter’s clusters and geo- economic models to the sporting industry, we find that there is limited production side benefit that could be gained by a sporting club (and while it is possible to increase that, the status quo is that the benefit derived is not a production side one). A demand-side network externality that positively impacts suppliers would be identified and would quantify the value of industry clustering for sports. Furthermore, the role of a rivalry in stimulating demand beyond its geographic scope would also be identified, which would be of particular value in conjunction with emerging economic literature around sport and its globalization. A measurement of the impact of club rivalry on financial performance would significantly contribute to sports and business strategy literature in that regard.

23

2.5 How to Measure the Impact of Rivalry on Club Financial Performance

It therefore follows that a club having a rival will have a positive impact on its financial

performance. Given the breakdowns of rivalry constructs above, it also follows that spatial rivals

(by definition: teams competing in the same league within the same local market) would have a positive effect on financial performance. That this effect is independent of a value chain benefit, it would have to be on revenue rather than on profitability.

H1: Spatial rivals will have a positive effect on year-end revenue.

It also follows intuition that rivalries with precedent antecedents besides space (which

can be any collection of causes ranging from historical, economic, political, etc…) should have a

positive effect on economic performance. Therefore, a rivalry with a greater proximity in

performance over a past period (in more detail in Methodology) should have a positive effect on

club financial performance as well:

H2: Performance based rivals will have a positive effect on year-end revenue.

Basic Methodology

To predict the role of rivalries on financial performance, running regressions that model

the financial performance of clubs would be central to identify predictors of club success. For the

purposes of this paper, variables will be incorporated to measure the role of spatial rivalries as a predictor, and variables to measure the role of performance based rivalry. To isolate performance

from quality, we would compare Team A to its rival Team B and then to the most proximate on-

field performer on the list (Team C). Running variables simultaneously will satisfy two criteria:

24

The quality construct will still be significant to rivalry. This is important because literature

identifies that a rivalry with a huge gulf in quality would be less salient to fans (especially global fans).

The empirical framework below models the financial success of a football club based on pre-existing literature and highlights predictor variables for revenues in sporting institutions.

25

Chapter 3.0 Empirical Framework

This section builds a conceptual model based on the literature review that links the

different consumers of the sporting product to the revenue streams generated through the club’s

operations, and to the demand drivers that impact the revenue streams. The model specifies the

scope of each revenue stream (whether local or global) and its corresponding demand drivers.

This is done with the aim of isolating impact demand drivers into the specific revenue streams

that they impact, which consequently creates a framework that tests for presence of a positive

network effect derived from an existing local competitor, and the specific mechanism through

which it operates (i.e. which revenue stream it impacts and how it can overcome the revenue

constraints typically imposed by the presence of a competitor in the same local market).

3.1 Financial Success Metrics

Most major soccer clubs operate as either private or public limited companies (Section

1.1: Football as a Business), with the major notable exceptions being the MLS, which utilizes a

franchising system similar to that adopted by the NBA, MLB and NFL (Zimbalist, 2003).

Therefore, typical valuation methods such as Market Value, Operating Revenue and

Profitability are applicable when measuring the financial success of soccer institutions. A

number of corporations release reports measuring such variables; Forbes annually releases a

ranking of the top soccer clubs by market value measured in US Dollars (Forbes, 2013), Deloitte

ranks the top clubs by ‘Revenue from Football Operations’ (Deloitte, 2013) and Brand Finance

ranks clubs through Brand Value (BrandFinance, 2013).

26

3.2 Revenue Streams

3.2.1 Classifying Revenue Streams

The metrics listed in the previous section are a reflection of business activities that feed

into cost and revenue, including combined off and on-field activities that moderate exposure and,

subsequently, brand value, merchandising strategies and agreements with affiliates.

Revenues amassed by sports clubs can be categorized into four major streams (Mason,

1999)

Fans: This is in reference to match-day revenue, which is amassed from ticket sales, the

on-site purchase of memorabilia and other complimentary products and club memberships.

Television: Comprises of the individual broadcasting agreements, in-house club

channels, and revenues accrued from collective broadcasting agreements (Mason, 1999).

Sponsorship: The endorsement and advertising agreements in place between the club

and entities that offer financial compensation in exchange for association with the club and

exposure through the club’s customer-facing operations – this includes and is not limited to: in-

stadium advertising, advertisements on jerseys, the standing position as the manufacturer of the

club’s training and match-day uniforms, advertising in press conferences and television.

Government/Community: The revenue accrued from government grant, tax credits and

co-operative projects to improve upon the local community and develop grass-roots sporting

projects.

There has been a call to incorporate modern media channels (specifically, the online

space) into a separate revenue stream (Ioakimidis, 2010). Raymond Boyle and Richard Haynes’ findings suggest that the facilitation of viewership in online spaces requires an incorporation of

27

an online medium into revenue streams generated from differing activities such as using the

online store to purchase memorabilia and fan-wear, online membership and access to exclusive

club content, viewing competitive events through online mediums, and the deployment of

sponsorship initiatives on club websites and fan pages (Boyle & Haynes, 2002). However, given

that these activities are ultimately sub-components of what is traditionally known as a revenue

aggregation process, adding the online space as a revenue stream of its own adds little to the

purpose of this paper, and is arguably unproductive. The activities will classify under the revenue

streams mentioned above, and subsequently are constituents of the overall empirical framework,

with the applicable scope (see below).

3.2.2 The Scope of Revenue Streams

As mentioned in the literature review, the revenue streams of football clubs are no longer

reliant on a local fan-base. The growth in media dispersion and the global nature of soccer as a

sport has created entirely new financing models for European soccer clubs (Andreff &

Staudohar, 2000). In fact, it was found that the emergence of satellite broadcasting has created

large revenue potential from international broadcasting that typically overshadows the role of the

local fan base (Baimbridge, Cameron, & Dawson, 1996). Deloitte’s research found that

sponsorship demands increase with televised exposure, and that the growing international

exposure has highly correlated with an increase of continental and global sponsors (Deloitte,

2010). Forbes reports similar findings in its analysis of the market value of soccer clubs, whereby it links the role of televised revenue and sponsorship closely to the market value of the sporting institution at hand (Forbes, 2011).

It is therefore important to classify revenue streams by the scope at which their contributing activities can achieve; fans’ purchases, for example, is a predominantly local

28

activity (with match-day tickets, on-site purchases constituting a need for physical presence on

the location, at which point, international purchases are restricted to tourists and visitors, which

would not compose a significant portion of the revenue stream in question). The same applies to

government and community revenues; the governments that would grant the club are generally

of a local nature (in addition, that revenue stream would not be significant enough to validate the

materiality component of the Generally Accepted Accounting Practices). Therefore, the scope of

the above revenue streams is local in nature, and is comprised in the

However, with substantial research emphasizing the role of television revenue as a factor

in club income (Baimbridge, Cameron, & Dawson, 1996), and the subsequent global nature of

sponsorships that ensue, the scope of broadcasting revenue is global, as it translates into revenue

from global audience, and is mostly dependent on populations that reside outside the local area in

which clubs are located (Mason, 1999) (Baimbridge, Cameron and Dawson, 1996). Pragmatic examples include the fact that Fox Sports, a US-based sports channel that is part of the FOX conglomerate, broadcasts games from the Spanish La Liga, the European Champions League and the Italian Serie A and Al Jazeera, which broadcasts matches from all major European soccer competitions to subscribers in the Middle East (FOX) (AlJazeera BeIN Sports).

The same applies to sponsorships, which as postulated by Deloitte and Forbes’ findings, are increasingly global in their scope; Manchester United is affiliated with US-based sporting goods manufacturer Nike, Swiss watchmakers Hublot, Turkish Airlines, DHL, Chevrolet and

Aon Insurance to name a few, Barcelona advertise Qatar Airways on their jersey, Arsenal’s home stadium is known as the Emirates (the UAE based airliner also have their logo at the front of the jersey of Italian side AC Milan), to name a few examples (Deloitte, 2013). Therefore, it becomes evident that television and sponsorships are capable of generating revenue that is

29

beyond the local fan-base. Table 1 outlines the classification of revenue streams and the

subsequent scope under which they fall.

Table 3.1: Revenue Stream Scope Revenue Driver Market Scope 1. Fans Local 2. Broadcasting Local, National, Global 3. Corporate Sponsors Local, National, Global 4. Government, Community Local, National

3.3 Demand Drivers

As demonstrated by the literature review and the section above, the sporting product on offer for the customer can be classified into four major streams of revenue. Revenue is a function of demand (Colander, 2004) , and thus drawing a relationship where demand drivers (be it drivers to attract the consumer or to meet the demand in itself) are important constituents in determining revenue. Classifying demand drivers is important in providing a structure to the club revenue streams, and also in determining the role that is played by each demand driver and its corresponding revenue stream.

3.3.1 Classifying Demand Drivers

Not all demand drivers can be determined by the club. There are other influences at play that may impact demand that are not relevant to the club presenting the product/service; and identifying which ones are within the control of the club and which ones are not helps not only identify what actionable insight can be developed from this paper, but further determines what of can be ruled out as a cause behind the success of clubs that operate within the same local market.

Demand factors can be exogenous to the control of the institution, whereby the institution

has very little to no say in determining them. These factors can be a byproduct of the area in

30

which the club operates or due to the rules and regulations imposed upon the club by the

governing bodies of the competition(s) in which it competes, or contingent on the general

behavioral attitude towards the product or service offering.

External Demand Drivers

We define external demand drivers as givens that are outside the control of the club itself,

and cannot be directly changes as a result of management preferences. External demand drivers

are divided into several categories:

Psychographic/Demographic Demand Drivers

The demand drivers that can be classified under the psychographic and demographic

category are ones that pertain to the relevant area surrounding the club’s main offering, the

match-day events that are in the local area. Those drivers characterize the local market and

provide pointers to the presence of a demand for the product on offer by the club. The regional

demographic and psychographic demand drivers are: the population of the city in which the club

is located and the city’s average income.

The above drivers that act as determinants on demand for the club’s offering. These

factors cannot possibly be determined by the club, and while a big club in a small city can

possibly impact the average income (Siegfried & Zimbalist, 2000), and potentially skew it depending on the size of the city, it is impossible for a club to control for these factors, albeit that it can impact them. The support for the sport can be partially impacted by marketing strategies and over time (James & Ross, 2004), however once again this is not a simplistic decision mechanism available to the club’s directors in order to so, and thus cannot be within the control of the club.

31

It furthermore stands to sense that a club would locate in an area with pre-existing

support for the sport as well as the population and income level that would uphold its facilities

and generate revenue that would comfortably offset the cost of running an institution of its size

(Kelly, Hoffman, & Carter, 1987), and therefore while these factors may change drastically due

to economic forces outside the control of the organization, this model will be tested in a uniform

manner that would account for such changes in the overall data set.

Competition-Dependent Demand Drivers

This section of the model refers to the actions undertaken by the clubs in competition

with the club and how they accordingly influence demand for the product or service offered by

the club in question. Clubs are granted the capacity to undertake executive actions that can

influence the well-being of the club, and yet it has an impact on the competition as well

(Whannel, 1992)

Operational Strategy of Competing Clubs

The operational strategy of the competing clubs will subsequently impact the level of on-

field success that they enjoy, and thus the level of competitiveness of the league overall (which

therefore partially determines the demand for the product, as it hinges on the uncertainty of the

outcome (Whannel, 1992)). The marketability of competing clubs will also determine how

desirable the end product (which is not only contingent on the offering of the individual club in

question, but the opponent(s)), and thus will also act as an according driver for demand.

League Dependent Demand Drivers

The last of the exogenous drivers of demand, demand drivers dependent on the league in

which the club participates in are factors that would influence overall demand for the individual

32

club’s offering and yet also outside the control of the club (Sloane, 1971). As outlined in the

literature review, clubs participate in league/ style competitions that are governed by

an association which sets out rules and regulations for the competition, determines fixture dates,

and, depending on the level of involvement, can disburse revenue based on placement in the

competition or to incentivize clubs (Mason, 1999).

The literature review examined different types of league structures and the differences in

guidelines that can be imposed upon the participating clubs, which range from stipulating

competitive rules and arranging match fixtures, to a more comprehensive management strategy that governs mass broadcasting agreements and revenue allocation to the participating teams

(Refer to Appendix B). These factors are also outside the club’s individual scope of control and

are decided on by the participating club in the league’s tiered structure (Mason, 1999).

Demand factors pertaining to the clubs’ own activities can be viewed as endogenous

demand drivers. Such drivers are fully within the control of the club, and can be manipulated as

seen fit by the directors or managers. These drivers will subsequently influence the value of the product or service of the club to the consumer, and subsequently the willingness to buy the product.

Club-Controlled Demand Drivers

Demand drivers that are considered endogenous are within control of the club’s executive setup, and can be changed in order to influence the club’s well-being. Some of those factors have subsequent influences on competing clubs within the same league (refer to section 1.3.1.1.2).

These drivers are what the club essentially does in order to stimulate demand and create revenue streams. The endogenous demand drivers are the club’s operational strategy, merchandising and

33

memorabilia, the ticket prices charged by the club, and the stadium capacity (which arises from

the club’s decisions).

Operational strategy refers to expenditure on players and facilities, sporting and business

decisions that influence the well-being of the club. Merchandising and memorabilia refers to the

decisions on the marketing mix executed on the merchandising aspect of the club’s business,

which includes but is not limited to official jerseys and club wear, official match programs and

souvenirs, print and media items. Controlling the pricing of season and match-day tickets will

influence demand depending on the price elasticity of demand for the product, which is in part a

function of disposable income and support for the sport (Colander, 2004).

Co-Dependant Demand Drivers

There also exist demand drivers that are dependent on both the club and external forces

such as the clubs with which it competes and the rules and regulations of the

league/tournament/association under which it offers the products and services. These factors can

be partially influenced by the decisions of the individual clubs and yet are also contingent on the

decision making of the competitive clubs and the league association as a whole. The Co-

Dependent demand factors are as follows: On-field performance (where the club’s ranking and

achievments depend on both the club and its competition), the competitiveness of the league as a

whole, and the rivalries that emerge which are also a byproduct not just of the club’s affect, but

that of the competition.

While sporting decisions within the club’s control will have a direct influence on on-field

performance, similar decisions utilized by the opposition will influence the effectiveness of the said club’s strategies. This in turn has an impact on the competitiveness of the league, but the

34

competitiveness in itself (not just how well the club in question performs) is an important

constituent in the demand for the overall product, which as mentioned in the literature review,

hinges in large part on the uncertainty of the outcome (Whannel, 1992). Another important factor in the impact of demand for the sporting product, in conjunction with competitiveness and club performance, is rivalries with competing opponents (refer to literature review). This is important in investing an emotional output that combines with the uncertainty of the outcome in order to create a value offering (Whannel, 1992).

Table 1.2 summarizes the demand drivers incorporated into the model and their classification within the realm of the club’s control.

Table 3.2 Classification of Demand Drivers

Demand Driver Category Classification

Population Regional Demographic/Psychographic Socio-Demographic

Disposable Income Regional Demographic/Psychographic Socio-Demographic

Support for Sport Regional Demographic/Psychographic Socio-Demographic

Competitor Competition Dependent External Operational Strategy

Rules and Fixtures League Determined External Pre-Existing League Determined External Agreements Membership League Determined External Conditions Club Operational Club-Dependent External Strategy Matchday and Club-Dependent External Season Ticket Prices Stadium Capacity Club-Dependent Club-Dependent Club Performance Cross and Semi-Dependent Co-Dependent Competitiveness Cross and Semi-Dependent Co-Dependent Rivalry Cross and Semi-Dependent Co-Dependent

35

3.3.2 Scope of Demand Drivers

The revenue streams generates by the sporting institutions are dependent on demand, and

thus subsequently the demand drivers or determinants. Therefore, it follows that demand drivers

would have their own scope, meaning that that each driver of demand has a sphere of influence

defined by the type of purchase that arises from the demand.

The demographic and psychographic demand drivers (refer to table 1.2) impact local

demand for the sporting product. Factors such as population of the region in which the club is

located, the average income of the residents of the region, and their support for the sport in

question is stipulated to impact revenue streams generated from the local population.

Competitor-dependent factors are likely to impact both local and global revenue streams,

considering that the demand for the product here is a function of willingness rather than ability –

and subsequently impacts a broader audience that is reached through televised and online media.

Logical deduction and the stipulation that corporations will attempt to act in their best interest

would imply that this would impact subsequent demand from sponsors, advertisers and

broadcasters. Thus, this category combines drivers that drive both local and global consumers in

terms of reach and attractiveness.

The same line of reasoning applies to the league regulations; fixtures and rules will

quantify the supply of the product available to meet the existing demand both in terms of on-site

operations and televised events. Membership conditions and revenue disbursement will impact

the quality of the facilities, the operational strategy deployed by the club and its marketability,

and consequently have impacts on both local global revenue streams such as match-day sales,

sponsorship and broadcasting revenue.

36

Endogenous demand drivers such as ticket prices, merchandising and stadium capacity

only are only tied to the local revenues of the club, while operational strategy and marketability

impact both local and global revenue streams.

Co-Dependent demand drivers mostly impact both local and global revenue streams, with

aspects such as performance impacting the support for the club both within the local area and

from a global perspective, rivalry generating attention on a local and global level. The competitiveness of the league is likely to have a greater impact on global revenue streams than it would on local (given that local revenue streams would be faced with a lack of an alternative for this particular driver).

Table 3.3 categorizes the demand drivers in terms of the level of control available to the

club (club-controlled vs external vs Co-Dependent), classifies them into the according market scope (local vs global) and lists the revenue streams that they impact.

37

Table 3.3 Demand Drivers and their Corresponding Revenue Streams

Demand Driver Scope Corresponding Revenue Streams Population Local Fans

Disposable Income Local Fans

Competitor Local and Global Fans, Television Operational Strategy Rules and Fixtures Local and Global Fans, Television

Membership Local and Global Fans, Television, Conditions Sponsorship,

Government

Club Operational Local and Global Fans, Television, Strategy Sponsorship, Government Matchday and Season Local Fans Ticket Prices Stadium Capacity Local Fans Club Performance Local and Global Fans, Television, Sponsorship Competitiveness Local and Global Fans, Television, Sponsorship Rivalry Local and Global Fans, Television, Sponsorship

38

A diagrammatic representation of the conceptual model is presented below. It illustrates the categorization of demand drivers (endogenous vs exogenous vs Co-Dependent), the classification of demand drivers and revenue streams (local vs global) and the relationship between demand drivers and revenue streams and the financial success metrics. Figure 3.1 Comprehensive Empirical Model

External Demand Drivers Club – Dependent Demand Drivers

Co‐Dependent Demand Drivers

39

Chapter 4.0 Methodology

The conceptual model explicated the relationships found between demand drivers (as

found by the literature) and total revenue. It further explained the potential role that rivalry can

play into determining club financial performance. Most importantly, developing market scopes

ensures that demand drivers correspond to the relevant revenue streams.

A regression equation on total revenue would incorporate the factors that speak to club

success; Table 4.1 summarizes the findings of the conceptual model below in a list of demand

drivers (variable classes) and the relevant IV/proxy IV to be utilized in a regression.

Table 4.1: Variable Class and Operational IV Variable Class Variable, Coefficient Socio-Demographic Population, β1 Income, β2 Performance League Rank, β3 UCL/EL Participation, β4, β5 Games Played, β6 Operational Strategy Stadium Capacity, β7 Average Attendance (000), β8 Transfer Expenditure (000), β9 Geographic Rivalry Presence of Local Rival 0/1, β10 Distance to Nearest Rival, β11 FIFA Rivalry Presence of App Rival 0/1, β12 3 Year Goal Difference with Rival distance from 1 (geographic and FIFA-determined), Competitive Balance β13 β14* Rank difference between rivals, (geographic and FIFA-determined), β15, β16 ** *The 3 year period was identified as optimal by literature (Talarico & Moore, 2012) ** FIFA experts classify rivalries based on all antecedents (FIFA, 2015) Isolating Comparative Quality Aspects in Rivalry from Quality

In section 2.1 we identified the importance of separating quality as a construct from the

comparative quality that is endogenous to the intensity of a rivalry. In order to ensure that there

would be no confusing the role of quality and rivalry (and since rivalry in part is a factor of

competitive quality) – the results between the club and its nearest national competitor (i.e. goal

40

difference over 3 years and rank difference) would be incorporated into the model as well. That

way, the comparative impact of quality and comparative quality in a rivalry would be measured;

if the significance of the impact only came from quality then it would render the variables of

goal difference insignificant.

Incorporating FIFA Rivalry Classifier

Literature has demonstrated that little by the way of successful attempts have been made

to measure rivalry empirically. Variables that capture rivalry from the on-the-pitch perspective

were not present (Tyler & Cobbs, 2009). The usage of proxies to measure what a rivalry yielded

no such success, and the majority of work to understand rivalry from the fan or personnel side

has been qualitative in nature.

Adding to that challenge is the antecedents to rivalry; the fact that rivalry or the unified

set of feelings that emerge from rivalry (Dalakas & Mancon, 2012) are based on a variety of

historical, political, religious and identity reasons that were mentioned in the literature review.

As a result, it would follow logic that it would be difficult to capture rivalry from data derived

from on-field performances (albeit that comparative performance would be a measure of the

intensity of a rivalry). Therefore, in order to avoid measuring a rivalry (given that no measure has

been developed), the utilization of an expert panel, amassing antecedents to determine rivals

becomes necessary. Referring to the section above, we can measure the intensity of this rivalry

and its impact on overall financial performance by isolating from most proximate performers

(whether they are rivals or not).

41

The Dependent Variables

The Deloitte Money League and the Annual Review of Football Finance (to be utilized in

studies 1 and 2, respectively) present total revenue and break it down into components:

 Match-day revenue: Ticket Sales and Hospitality Packages

 Television Revenue: Revenue from Broadcasting Agreements

 Commercial Revenue: Merchandising and Sponsorship

As a result, by tracing each club activity (and thus demand driver) to its corresponding revenue stream on the list, a more coherent set of regressions becomes apparent:

 Regressing on television revenue would negate the need to incorporate local

demographics (given that it is a global revenue source)

 Regressing on commercial income would incorporate stadium capacity (since sponsors

can be local and will be motivated by stadium size) but not include aspects such as

population or disposable income

 Match-day revenue as a DV would only incorporate local demographic factors from

demand drivers.

The Selection of Revenue from Football Operations as the Dependent Variable

The cause behind selecting the Revenue from Football Operations as the Dependent

Variable is two-fold. Firstly, it is certainly available through the Deloitte Annual Review of

Football Finance for at least the top two tiers of clubs in English football for no less than the past five years, as aspect that was not readily available when concerning any of the other financial metrics at any of their respective sources. Secondly, Revenue from Football Operations is sufficiently volatile in order to accommodate the effects at hand, and yet sufficiently stable not to

42

be substantially swayed by other uncertain metrics that cannot be immediately measured or

approximated with sufficient precision such as brand value. It also eliminates cost from the

equation which, for the sake of simplicity, conveniently circumvents the problems in evaluating cost structures, debt refinancing, and operates with the principle that the football club is a utility

maximizing organization that acts in its financial best interest.

The study is a linear regression model that tests the impact of independent variables that approximate demand for the football product on the operating revenue for the sporting institution in question. This is done for 20 clubs over a 10 year time period (2005-2014).

The study utilizes the Deloitte Money League as its primary data source. Demographic

information was obtained from Eurostat, the official European database that deals with

demographic data, and official league results to compute goal differences were obtained from the

respective league institutions. The Deloitte Money League is a publicly available annual ranking

of the 20 top performing European clubs by revenue from football operations (operating

revenue). The report breaks revenue down into three components: Match-day, Television and

Commercial.

The Regression Equations

Each revenue stream would have six regression equations:

1. Model 1: To study the impact of adding a FIFA-determined rival binary on the

revenue stream.

2. Model 2: To study the impact of adding a FIFA-determined rivalries on the

revenue stream, with variables of competitive balance to control for quality.

43

3. Model 3: To study the impact of adding geographic (spatial and local) rivalries on

the revenue stream.

4. Model 4: To study the impact of adding geographic (spatial and local) rivalries on

the revenue stream, with variables of competitive balance to control for quality.

5. Model 5: To study the impact of adding both FIFA-determined and geographic

rivalries on the revenue stream.

6. Model 6: To study the impact of adding both FIFA-determined and geographic

rivalries on the revenue stream, with variables of competitive balance to control

for quality.

Models for Total and Match-day Revenue (y1,y2):

Model 1: y1, y2 = β 0 + β1 log10 (Population) + β2 log10 (GDP per capita) + β8log10 (Average

attendance) + β9log10 (transfer expenditure) + β3(League Rank) + β4 , β5(European Participation)+

β6(Number of Matches – Regular Season Matches) + , β12 (Presence of a FIFA appointed rival)

Model 2: y1, y2 = β 0 + β1 log10 (Population) + β2 log10 (GDP per capita) + β8log10 (Average

attendance) + β9log10 (transfer expenditure) + β3(League Rank) + β4 , β5(European Participation)+

β6(Number of Matches – Regular Season Matches) + β10 (Presence of a FIFA appointed rival) +

β14 (3 year GD with FIFA appointed rival)+ β16 (Rank difference with FIFA appointed rival)

Model 3: y1, y2 = β 0 + β1 log10 (Population) + β2 log10 (GDP per capita) + β8log10 (Average

attendance) + β9log10 (transfer expenditure) + β3(League Rank) + β4 , β5(European Participation)+

β6(Number of Matches – Regular Season Matches) + β10 (Presence of a local rival)+ β11 (Distance to nearest rival)

44

Model 4: y1, y2 = β 0 + β1 log10 (Population) + β2 log10 (GDP per capita) + β8log10 (Average

attendance) + β9log10 (transfer expenditure) + β3(League Rank) + β4 , β5(European Participation)+

β6(Number of Matches – Regular Season Matches) + β10 (Presence of a local rival)+ β11 (Distance to nearest rival) + β13 (3 year GD with closest distance rival) + β15 (rank difference with closest

distance rival)

Model 5: y1, y2 = β 0 + β1 log10 (Population) + β2 log10 (GDP per capita) + β8log10 (Average

attendance) + β9log10 (transfer expenditure) + β3(League Rank) + β4 , β5(European Participation)+

β6(Number of Matches – Regular Season Matches) + β10 (Presence of a local rival)+ β11 (Distance to nearest rival)+ β12(Presence of a FIFA-appointed rival)

Model 6: y1, y2 = β 0 + β1 log10 (Population) + β2 log10 (GDP per capita) + β8log10 (Average

attendance) + β9log10 (transfer expenditure) + β3(League Rank) + β4 , β5(European Participation)+

β6(Number of Matches – Regular Season Matches) + β10 (Presence of a local rival)+ β11 (Distance to nearest rival) + β13 (3 year GD with closest distance rival) + β15 (rank difference with closest

distance rival) β10 (Presence of a FIFA appointed rival) + β14 (3 year GD with FIFA appointed

rival)+ β16 (Rank difference with FIFA appointed rival)

The regression models that employ television income (y3) as a dependent variable will

differ in the removal of three variables:

1. The socio-demographic variable of population will be removed given that it is not

relevant to the global scope of television income.

2. The socio-demographic variable of GDP per capita will be removed given that it is

not relevant to the global scope of television income.

45

3. The match-day based variable of average attendance will be removed given that it

does not contribute to the revenue stream and is not relevant to the scope of television

income.

In similar vein to the regression models with television income as a DV, regressions with commercial income as a DV will omit the above variables. However, they will include stadium capacity across all six models in order to account for on-site merchandising activity and provide a proxy for real estate size and viewership numbers, which explain sponsorship and advertising revenue.

Hypotheses

Following the hypotheses developed in Section 2.5, it would be expected that:

H1. The presence of a local rival will positively impact total revenue.

H1.1 The presence of a local rival will positively impact television revenue.

The interest from broadcasters (refer to Appendix A: Stakeholder Analysis) and

subsequent willingness to pay is based on the attractiveness of the product to the market segment

in question. Therefore, given the identified role of rivalry in demand, television revenue is

expected to grow. In similar vein, the interest from sponsors should yield similar results, and thus

the expectation is that commercial revenue will be positively impacted by local rivalries:

H1.2 The presence of a local rival will positively impact commercial revenue.

Given the network effects of rivalry (refer to literature review), the expectation is that

there will be a negative relationship between revenue forms and the distance between rivals:

46

H1.3 The distance between one rival and the other will have a negative relationship with

total revenue.

H1.3.1: The distance between one rival and the other will have a negative relationship

with television revenue.

H1.3.2: The distance between one rival and the other will have a negative relationship

with commercial revenue.

As demonstrated by the literature review, the rivalry intensity will increase with a greater

state of competitive balance (i.e. more at stake). Therefore, it is expected that variables that

denote competitive balance between rivals will have an effect on revenue:

H1.4 The closer the 3 year average goal difference between rivals to 1, the greater the impact on total revenue.

H1.4.1 The closer the 3 year average goal difference between rivals to 1, the greater the impact on television revenue.

H1.4.2 The closer the 3 year average goal difference between rivals to 1, the greater the impact on commercial revenue.

Based on the above, the an increase in the x value denoting how far a 3 year average goal difference is from 1 is expected to decrease revenue. Similar outcomes are expected for the difference in rank between the rivals (the closer the distance, the greater the competitive balance between the teams):

H1.5 The closer the rank difference between the rival teams, the greater the impact on total revenue.

47

H1.5.1 The closer the rank difference between the rival teams, the greater the impact on

television revenue.

H1.5.2 The closer the rank difference between the rival teams, the greater the impact on

commercial revenue.

The cause behind having a negative relationship between rank distance and goal

difference revenue is because the smaller distance implies a greater rivalry intensity (Talarico &

Moore, 2012). The intensity impact of a result is also measured by its proximity to an absolute

value of 1 based on literature stipulating that proximity with a defined winner will escalate

rivalry from a positive and negative affect (Talarico & Moore, 2012) (Euchner, 1993).

Similarly to the geographic rivalry hypotheses, the hypotheses for FIFA-determined

rivalries are made with the expectation that rivalry and competitive balance will impact revenue:

H2. The presence of an appointed/ FIFA-determined rival will have a positive impact on total revenue.

H2.1 The presence of an appointed/FIFA-determined rival will have a positive impact on television revenue.

H2.2 The presence of an appointed/FIFA-determined rival will have a positive impact on commercial revenue.

H2.3 The closer the 3 year average goal difference distance with the nearest ranked rival is to 1, the greater the impact on total revenue. H2.3.1 The closer the average 3-year goal difference distance with the nearest ranked rival is to 1, the greater the impact on television

48

revenue. H2.3.2 The closer the average 3-year goal difference distance with the FIFA-

determined rival is to 1, the greater the impact on commercial revenue.

H2.4 The closer the rank difference between the FIFA-determined rivals, the greater the

impact on total revenue.

H2.4.1 The closer the rank difference between the FIFA-determined rivals, the greater the

impact on television revenue.

H2.4.2 The closer the rank difference between the FIFA-determined rivals, the greater the

impact on commercial revenue.

No formal hypotheses are made for match-day revenue streams, which stems from a lack

of existing literature concerning the continual effect of two matches per season on attendance

figures for the remainder of the season. This is supplemented by the pragmatic finding that

attendance rates and season ticket holders are consistently high across the board for the clubs

involved in Deloitte’s Money League, and therefore, especially given the size of the sample, it

would be difficult to find differences for match-day revenue.

49

Chapter 5.0 Results

The study contained a total of 24 regressions: six regressions were done for each dependent variable (total revenue, match-day revenue, television revenue and commercial revenue) with the independent variables changed according to what would have an impact on the revenue stream. The results are available in Appendix D, while the descriptive variables are in

Appendix C.

Of the six regressions conducted for each DV, the one selected is the sixth. The selection

of the regression as an emphasis for the analysis is due to the fact that it incorporates rivalry and

competitive balance predictors for both rivalry forms identified in this thesis, and thus it tests the

relative importance of all the rivalry forms in question and further controls for comparative

quality through the competitive balance variables. However, there is some important insight to

be generated through looking at the other regressions individually, and particularly when it

comes to assessing the role of geographic rivalries as a stand-alone (see study results below).

To smooth for sharp effects and normalize significant variability in the data, logarithms

were taken for all dependent variables, and for several predictor variables: population, average

income, stadium capacity, average attendance, transfer expenditure, wages and distance to

nearest rival. Therefore, the scale of the impact of each variable in dollar terms is quantified by

eliminating the logarithms from the equation. The table below summarizes the impact that each

variable would have on the regression:

50

Table 5.1: Numerical impact of each IV on DVs

Variable Class Variable, Coefficient* Impact on DVs Socio- Population, β1 y1,2,3,4= Y1- Y0 where: Y1/Y0*** = β Demographic (x1/x0)

Income, β2 y1,2,3,4= Y1- Y0 where: Y1/Y0*** = β (x1/x0)

Performance League Rank, β3 y1,2,3,4 = Y1 – Y0 where: Y1/Y0 = β (x1/x0)

β4,5 UCL/EL Participation, β4, β5** y1,2,3,4 = 10

Games Played, β6 y1,2,3,4 = Y1 – Y0 where: Y1/Y0 = β (x1/x0)

Operational Stadium Capacity, β7 y1,2,3,4= Y1- Y0 where: Y1/Y0*** = β Strategy (x1/x0)

Average Attendance (000), β8 y1,2,3,4= Y1- Y0 where: Y1/Y0*** = β (x1/x0)

Transfer Expenditure (000), β9 y1,2,3,4= Y1- Y0 where: Y1/Y0*** = β (x1/x0)

β11 Geographic Presence of Local Rival 0/1, β11 y1,2,3,4 = 10 Rivalry Distance to Nearest Rival, β12 y1,2,3,4= Y1- Y0 where: Y1/Y0*** = β (x1/x0)

β13 FIFA-determined Presence of App Rival 0/1, β13 y1,2,3,4 = 10 rival

3 Year Goal Difference with Rival y = 10 β14 1,2,3,4 (geographic and appointed), β ** Competitive 14 β15, 16 Balance Rank difference between rivals, y1,2,3,4 = 10 (geographic and appointed), β15, β16 ** *The coefficient denotes coefficients assigned to exemplify the impact of a variable on the DV (column 3 of Table X). The order will differ depending on assigned IVs in each regression. **Denotes multiple variables that are operated in the same way. ***Y0 is the base mean, and Y1 is the new value with a change of x by one unit.

Data Fit

The R2 value was higher for both Total and Match-day revenue than it was for the other revenue streams. The difference can be attributed to the lack of information that would more accurately model revenue streams from television and commercial revenue, such as the number of televised matches on given days and the number of sponsors, respectively.

51

Table 5.2: Adjusted R-Square Values across models for all DVs Model Total Revenue Match-day Television Commercial Revenue Revenue Revenue 1 0.711 (0.245) 0.594 (0.172) 0.536 (0.383) 0.405 (0.439) 2 0.721 (0.241) 0.613 (0.168) 0.574 (0.368) 0.402 (0.440) 3 0.635 (0.286) 0.675 (0.159) 0.459 (0.413) 0.426 (0.448) 4 0.639 (0.284) 0.711 (0.150) 0.458 (0.413) 0.442 (0.441) 5 0.729 (0.246) 0.709 (0.151) 0.527 (0.385) 0.441 (0.442) 6 0.736 (0.243) 0.745 (0.141) 0.579 (0.364) 0.459 (0.434)

* Number between brackets is the standard error.

There was a significant pairwise correlation (0.685) between stadium capacity and

average attendance, meaning that only one could have been used as a predictor variable for

revenue streams. Independent of rivalry variables, the best line of best fit results came from

average attendance for the total and match-day revenue streams (which is what average attendance is relevant to), and therefore average attendance was the choice predictor variable for said regressions. For commercial revenue, stadium capacity continued to be utilized (as it both acts as a predictor for commercial purchases made on the stadium grounds as well as the number of sponsor displays).

Socio-Demographic Variables

Using total revenue as a dependent variable meant that the socio-demographic variables that literature found to be significant for ticket sales (i.e. match-day revenue and revenues that are local in scope) would be used. The socio-demographic variables, population and average income, had differing effects on the dependent variables. Table 5.3 summarizes the findings for population:

52

Table 5.3: Impact of 1,000 change in population total & match-day revenues Model Total Revenue Matchday Revenue Impact of 1,000 unit change on y (£) Total Rev Match-day Rev 1 0.044 (0.020) 0.064 (0.014) 5,283.10 1325.93 2 0.046 (0.019) 0.063 (0.013) 5,200.55 1305.21 3 0.084 (0.026) 0.045 (0.014) 3,714.66 932.29 4 0.076 (0.026) 0.037 (0.014) 3,054.27 766.55 5 0.037 (0.023) 0.042 (0.014) 3,467.02 870.13 6 0.037 (0.022) 0.026 (0.014)* 2,146.24 538.65* -Number between brackets is standard error -Bolded values are statistically significant (at 95% confidence interval) **Starred values are within the 90% confidence interval

For regressions that utilized FIFA-determined rivalries exclusively, population was

statistically significant. As can be seen for total revenue, population was significant in Models 1 through 4. The lack of significance in regressions 5 and 6 can be attributed to the addition of multiple variables that diminished the predictive power of population. In models that applied both geographic and spatial rivalries, population was not found to be statistically significant

(even at the 10% level).

For the statistically significant results, the impact a change in population would have would be the coefficient multiplied by 1,000 and then multiplied by y (given the notations). As can be seen above, a change of 1,000 in population would induce a change of up to £5,283.10 (or

£5.28 per citizen) for total revenue in Model 1 (refer to Table 5.1 for mathematical operation guide).

In the case of match-day revenue in particular, the effects were almost significant across

all six models; the notable exception being model 6 (that incorporates both rivalries and their

respective competitive balance predictors), where the result was significant at the p=0.1 level

(p=0.065). This is a clear indicator that population is a significant constituent in local revenue

streams, and the greater the population the greater the impact on finances. This is consistent with

53

several findings in literature, which imply that socio-demographics are important for local

revenue streams.

Average income, the other socio-demographic predictor, did not have as significant an impact on the total and match-day revenue models. The table below demonstrates the coefficient values found for match-day income and sheds light on its lack of substance in this model:

Table 5.4: Impact of Change of £1,000 in per capita GDP on total & match-day revenue

Regression Total Revenue Matchday Impact of 1,000 unit change (£) on y Revenue Total Rev Match-day Rev 1 0.094 (0.056)* 0.030 (0.039) 168,726.51* 42,345.38 2 0.094 (0.055)* 0.031 (0.039) 174,353.17* 43,757.50 3 0.031 (0.068) 0.042 (0.038) 236,256.95 59,293.53 4 0.043 (0.068) 0.060 (0.098)* 337,595.34 84,726.47* 5 0.186 (0.065) 0.018 (0.040) 101,218.83 25,402.94 6 0.133 (0.060) 0.069 (0.035) 388,283.76 97,447.77 -Number between brackets is standard error -Bolded values are statistically significant (at 95% confidence interval) **Starred values are within the 90% confidence interval

As can be seen in Table 5.4, per capita GDP is only statistically significant for two models with total revenue as a dependent variable. The models that incorporate both rivalry forms are the ones where per capita GDP is statistically significant for total revenue, while it is only significant for model 6 when match-day revenue is the dependent variable (albeit that it is within the 10% confidence interval for model 4, which incorporates geographic/local rivalry with competitive balance). Without further analysis to rule out phenomena such as intercorrelation

(which is unnecessary, given that the variables that lend or detract support from the hypotheses are unaffected), the impact of per capita GDP cannot be verified, as the absence of parsimonious effects lends suggestion that there could be other causes behind the statistical significance in

Models 5 and 6 for total revenue, and Model 6 for match-day revenue.

54

The absence of unanimous statistical significance on match-day revenue (and thus, total

revenue; there is no other stream where average income can have an impact) is not entirely

surprising (the reason for inclusion was that literature had conflicting views on the role of

income), given the fact that stadiums in this category of club almost always operate at near-

capacity (and the intuition that the club as a firm would offer seats and merchandise at rates that are affordable to the consumer).

What is surprising is the significant disparity in the coefficient values between total and match-day; while it is expected that the contribution to match-day income would not be greater than that towards total revenue, the fact that the contribution towards total revenue can be more than double that towards match-day income indicates that either there is another revenue stream that is impacted by average income (the likeliest being commercial income), or that there is a significant flaw within the GDP per capita variable as a proxy for average income (or that the requirement for an accurate proxy would be an average income within a particular radius, rather than the entire NUTS 3 region). Adding GDP per capita to commercial regressions would also run the risk of over-explanation, as there is a degree of correlation between the revenue streams and thus variables that are not relevant to how the revenue stream operates can be found significant (ex. Average attendance on television revenue) simply due to their accuracy as proxies of how the club would succeed across different revenue stream scopes.

Performance Variables

Aimed at quantifying the role of the club’s on-field performance on its revenue streams, several performance variables were deployed in order to most accurately capture this particular construct. A total of four variables that proxy how well the club has performed during that

55

season. The results were mixed, but were largely within line of expectations for most variables operationalized.

League Rank: The league rank variable was only significant for commercial revenue, which is to be expected, given the rewards structure that each league disburses (refer to Figure A2.1 in

Appendix B), which Deloitte classes as commercial revenue (Deloitte, 2014).

Table 5.5: League Rank Coefficient values & impact levels across models for all DVs Model Total Revenue Matchday Television Commercial Revenue Revenue Revenue 1 -0.011 (0.006)* 0.001 (0.004) -0.001 (0.009) -0.034 (0.010) 2 -0.011 (0.006)* 0.001 (0.004) -0.002 (0.008) -0.034 (0.010) 3 -0.012 (0.007)* -0.000 (0.004) -0.004 (0.010) -0.039 (0.011) 4 -0.012 (0.008) -0.002 (0.004) 0.002 (0.011) -0.034 (0.012) 5 -0.012 (0.006)* 0.002 (0.004) 0.003 (0.009) -0.036 (0.011) 6 -0.007 (0.007) -0.001 (0.004) 0.011 (0.010) -0.030 (0.012) -Number between brackets is standard error -Bolded values are statistically significant (at 95% confidence interval) *Starred values are within the 90% confidence interval

Table 5.6: Impact of League Rank Coefficients on Commercial Revenue

Model Commercial Impact of 1 unit Revenue change on y (%) 1 -0.034 (0.010) 14.4% 2 -0.034 (0.010) 14.4% 3 -0.039 (0.011) 16.4 % 4 -0.034 (0.012) 14.4% 5 -0.036 (0.011) 15.3% 6 -0.030 (0.012) 12.9%

-Number between brackets is standard error -Bolded values are statistically significant (at 95% confidence interval) **Starred values are within the 90% confidence interval

The table above demonstrates the average impact of a change in the team’s league ranking on their financial standing. Given that the significance was only found for Commercial

Revenue, the values were only computed for that particular stream. Given the negative sign, the smaller the league position (i.e. the higher), and the greater the revenue. Given merit portion of

56

television rights disbursement (which is actually classified by Deloitte as commercial income) -

these values stand to sense, with an average commercial income of £68.9 million, a change in

league position should impact revenue by approximately £9.5-11 million.

The differences between the values across the 6 models imply that rivalry variables

impact the relationship between league placement and commercial revenue; a reference to Figure

A2.1 in Appendix B will demonstrate that while the compensation method for each differs, the

number of televised matches variability (with the idea that rivalry games are more likely to be

televised) is what explicates the variation in the impact of commercial revenue once rivalry

variables are introduced.

UCL Participation: Participation in the UEFA Champions’ League, Europe’s elite club

competition, yields significant financial rewards upon the participant. UEFA utilizes a tiered

reward system that remunerates participants based on qualifying to the competition and

subsequent accomplishments within it (UEFA, 2015).

Interestingly, the binary variable denoting participation in the UEFA Champions’ League

yielded no significant effect across any of the models; and was only within the 10% confidence

interval once.

Table 5.7: UCL Participation Coefficient values & impact levels across models for all DVs Model Total Revenue Match-day Television Commercial Revenue Revenue Revenue 1 0.062 (0.071) -0.078 (0.119) 0.155 (0.157) -0.033 (0.126) 2 0.077 (0.274) -0.092 (0.049) 0.191 (0.105)* -0.041 (0.128) 3 0.114 (0.207) 0.017 (0.050) 0.132 (0.128) -0.003 (0.139) 4 0.108 (0.090) 0.003 (0.047) 0.149 (0.129) -0.001 (0.138) 5 0.052 (0.078) 0.000 (0.048) 0.079 (0.120) -0.030 (0.138) 6 0.084 (0.077) -0.025 (0.045) 0.139 (0.115) -0.005 (0.137) -Number between brackets is standard error *Starred values are within the 90% confidence interval

57

A deeper look at the data explicates this phenomenon: the descriptive statistics (refer to

Appendix C) indicate that the mean of the binary variable denoting UCL participation is 0.70

(meaning that 140 of the 200 clubs across the 10 years of data have participated in the

Champions’ League. Therefore, this large representation implies that participation in Europe’s

elite (and one its most financially lucrative) club competitions was not exclusive to the best

earners on the list, but it was almost a pre-requisite to be on the list (given that logic,

participation in Europe’s secondary competition, the Europa League, should create a negative

coefficient). The signs, for the vast majority of the coefficients, are positive (indicating a

positive correlation between participation in the UCL and revenue); the notable exception is

commercial revenue, where the signs are largely negative but not statistically significant.

Europa League Participation: Europe’s second tier continental competition is less viewed,

esteemed, and financially rewarded than the UEFA Champions’ League. It is therefore entirely plausible that, in a list of Europe’s top earning clubs, participation in the EL will actually cause a negative impact on revenue (i.e. when compared to clubs who play in the Champions’ League, the revenue for UEFA league participants should be smaller). While participation in the Europa

League would still be better than not participating in continental competition whatsoever, the assumption that Europe’s top financially performing clubs are most likely to be in European

competition (which is supported by data, a total of 172 clubs from the 200 strong sample have

participated either in the UCL or the EL, 140 of which were in the UCL – for more details refer

to Appendix C), and thus it is which competition the club participates in; in other words, if a club

is in the EL that means they are missing out on the extra revenue from the UCL, and thus being in the EL should have a negative impact on your earnings.

58

Table 5.8: EL participation coefficient values & impact levels across models for all DVs Regression Total Revenue Match-day Television Commercial Revenue Revenue Revenue 1 -0.084 (0.076) -0.118 (0.053) -0.173 (0.116) 0.040 (0.133) 2 -0.072 (0.075) -0.128 (0.052) -0.140 (0.111) 0.034 (0.133) 3 -0.155 (0.097) -0.078 (0.054) -0.318 (0.138) -0.046 (0.150) 4 -0.166 (0.097)* -0.095 (0.051)* -0.305 (0.138) -0.054 (0.148) 5 -0.046 (0.085) -0.034 (0.052) -0.170 (0.132) 0.030 (0.151) 6 -0.049 (0.084) -0.051 (0.049) -0.128 (0.308) 0.033 (0.149) -Number between brackets is standard error -Bolded values are statistically significant (at 95% confidence interval) **Starred values are within the 90% confidence interval

The data scarcely yields statistically significant results for EL participation, and in large part this is due to the small size of EL participants (32; the bottom 25% of the clubs alone comprise 50 clubs). Thus, any outlier (i.e. one of the big teams having an off-season on the field but still maintaining good financial performance for that year) that happened to participate in the

EL, or the many smaller-revenue clubs that made it to the UCL for the season (but are on unequal ground due to other factors) will mean that the impact of the EL variable is significantly diminished. However, for the few times that the variable did attain significance (Models 1 and 2

in Match-day revenue, Models 3 and 4 in Television Revenue), the coefficient effect was

substantial: for Model 1 in match-day revenue, participation in the EL would decrease income by

41.9%, for Model 2 participation in the EL would decrease match-day income by 44.5%.

In Models 3 and 4 for Television Revenue, participation in the EL would decrease

revenue by 76.8% and 75.5%, respectively. Of course, these numbers are not implying that not

participating in the EL would eliminate this decrease; rather they are signifying that participating

in the UCL (the overwhelming majority alternative in this dataset) is what would eliminate it.

Games Played over Regular Season Games: This variable was made to denote the progression of

each team into knockout competitions (including European competitions and domestic cups).

The subtraction of regular season games was put in place in order to ensure that there is equity

59

across the teams (otherwise, German clubs, who play 34 games to the 38 that are played in

English, Spain, Italy – would be at a 2 game handicap, for example). The number of games played in a season was statistically significant across all DVs besides commercial income.

Table 5.9: Games played coefficient values across models for all DVs

Model Total Revenue Match-day Television Commercial Revenue Revenue Revenue 1 0.015 (0.004) 0.014 (0.003) 0.021 (0.006) 0.007 (0.007) 2 0.014 (0.004) 0.015 (0.003) 0.018 (0.006) 0.007 (0.007) 3 0.021 (0.005) 0.011 (0.003)* 0.032 (0.007) 0.013 (0.008) 4 0.023 (0.005) 0.014 (0.003) 0.030 (0.008) 0.015 (0.008)* 5 0.014 (0.005) 0.008 (0.003) 0.022 (0.007) 0.008 (0.008) 6 0.016 (0.005) 0.011 (0.003) 0.019 (0.007) 0.009 (0.008) -Number between brackets is standard error - Bolded values are statistically significant (at 95% confidence interval) **Starred values are within the 90% confidence interval

Table 5.10: Games Played Coefficient Values and impact levels across Total & Match-day Revenue DVs Model Total Revenue Impact of 1 unit Match-day Impact of 1 unit change on y Revenue change on y 1 0.015 (0.004) 7.15% 0.014 (0.003) 6.66% 2 0.014 (0.004) 6.66% 0.015 (0.003) 7.15% 3 0.021 (0.005) 10.15% 0.011 (0.003)* 5.20%* 4 0.023 (0.005) 11.17% 0.014 (0.003) 6.66% 5 0.014 (0.005) 6.66% 0.008 (0.003) 3.75% 6 0.016 (0.005) 7.64% 0.011 (0.003) 5.20% -Number between brackets is standard error - Bolded values are statistically significant (at 95% confidence interval) **Starred values are within the 90% confidence interval

As evident from table 5.10, the coefficient values for Total Revenue are statistically significant across all six models. The impact level of an extra game played varies from 6.66% to

11.17%, with Model 4 (local rivalry with competitive balance variables) demonstrating the greatest impact, while both models 2 and 5 (FIFA-determined rivalry, both rivalries) show the lesser impact of a unit change on overall revenue.

60

Similarly, the coefficient values for Match-day Revenue are significant at p=0.05 for 5 of the 6 models; model 3 (geographic rivalry) is significant at the p=0.10 level. The effects on match-day revenue are less pronounced per unit change in games played, however, with the

range of coefficient impact going from 3.75% (Model 5) to 7.15% (Model 2).

Table 5.11: Games Played Coefficient Values and impact levels across Television Revenue DV Model Television Impact of 1 unit Revenue change on y 1 0.021 (0.006) 10.15% 2 0.018 (0.006) 8.64% 3 0.032 (0.007) 15.88% 4 0.030 (0.008) 14.82% 5 0.022 (0.007) 10.66% 6 0.019 (0.007) 8.64% -Number between brackets is standard error -Bolded values are statistically significant (at 95% confidence interval) **Starred values are within the 90% confidence interval

The impact of games played on television revenue (as a percentage) is the greatest out of

all the statistically significant coefficients for this particular variable. This range varies from

8.64% (Models 2 and 6 – signifying that competitive balance for FIFA-determined rivalries has a downward effect on the impact of games played) to 15.88% (Model 3). The reason behind this impact is that the clubs in question are likelier to have their matches televised (Deloitte, 2005).

Therefore, with the greater number of games that they play, the greater number of televised games (and hence, the greater the impact on television revenue).

Operational Strategy Variables

Transfer Expenditure: Net transfer expenditure is best available proxy to estimate a club’s

willingness to recruit and retain talent (wages paid data was unavailable for most clubs). There is

a significant correlation between transfer expenditure and wages paid, with the amount paid to

extract a player from their contract usually an indicator of their worth to the selling team (and

61

thus the wage that they would command). With the exception of models where IVs were regressed on match-day revenue (where transfer expenditure was not significant for any of the 6 models), transfer expenditure was significant across every model for all DVs.

Table 5.12: Transfer Expenditure Coefficient Values

Model Total Revenue Match-day Television Revenue Commercial Revenue Revenue 1 0.104 (0.016) 0.010 (0.011) 0.147 (0.025) 0.135 (0.029) 2 0.107 (0.016) 0.007 (0.011) 0.156 (0.024) 0.134 (0.029) 3 0.126 (0.019) 0.018 (0.011)* 0.171 (0.026) 0.164 (0.030) 4 0.124 (0.019) 0.016 (0.010) 0.177 (0.028) 0.158 (0.030) 5 0.120 (0.017) 0.007 (0.011) 0.151 (0.026) 0.148 (0.030) 6 0.109 (0.017) 0.005 (0.010) 0.155 (0.025) 0.146 (0.030) -Number between brackets is standard error -Bolded values are statistically significant (at 95% confidence interval) **Starred values are within the 90% confidence interval

Table 5.13: Impact of a £1 mil change in transfer expenditure on total revenue Model Total Revenue Impact of £1 mil. change on y 1 0.104 (0.016) 1,457,670.45 2 0.107 (0.016) 1,470,888.01 3 0.126 (0.019) 1,554,619.46 4 0.124 (0.019) 1,545,803.98 5 0.120 (0.017) 1,528,174.18 6 0.109 (0.017) 1,479,700.20 -Number between brackets is standard error -Bolded values are statistically significant (at 95% confidence interval) **Starred values are within the 90% confidence interval

Transfer expenditure is at its most impactful on total revenue in Model 3, and at its least impactful in Model 1. The differences are minor (max to min range is £100,000) as an absolute, but in relative terms, that is a 10% difference in ROI. As a general trend, geographic rivalry variables tend to increase the impact of transfer expenditure on the overall model, and FIFA- determined rivalry variables decrease it. An investment of £1 million in transfers will increase

62

revenue by no less than £1,457,670, and up to £1,554,619 (meaning a return that ranges from

45.7% to 55.5%).

Table 5.14: Transfer expenditure impact on Television and Commercial Revenue

Model Television Impact of £1 mil. Commercial Impact of £1 mil. Revenue change on y Revenue change on y 1 0.147 (0.025) 1,261,681.33 0.135 (0.029) 1,204,855.79 2 0.156 (0.024) 1,277,730.12 0.134 (0.029) 1,203,336.11 3 0.171 (0.026) 1,304,485.15 0.164 (0.030) 1,248,941.21 4 0.177 (0.028) 1,315,189.63 0.158 (0.030) 1,239,817.78 5 0.151 (0.026) 1,268,813.74 0.148 (0.030) 1,224,614.75 6 0.155 (0.025) 1,275,946.76 0.146 (0.030) 1,221,574.54 -Number between brackets is standard error - Bolded values are statistically significant (at 95% confidence interval) **Starred values are within the 90% confidence interval

The impact of transfer expenditure on television and commercial revenue is higher than it

is on total revenue (which, as previously seen with population, is the case due to other variables

having varying degrees of significance across models, and due to unobserved variables in

television and commercial – such as number of televised matches and number of sponsors – that

would have more accurately limited the scale its impact).

For television revenue, the impact ranged from £1,261,681 to £1,315,189 per million

spent. This constitutes a return on investment from 26.1 to 31.5%. As was the case with total

revenue, the lowest values are associated with Models 1 and 2 (where FIFA-determined rivalry

variables are in play), and the higher values were with geographic rivalry variables (Models 3

and 4).

While the effect of transfer expenditure on commercial revenue was not as high as that of

television, the range was still higher than that of total revenue; the lowest figure was a return of

£1,204,855.79 (20.5%) and the highest was £1,248,941.21 (24.9%). As was the case with total

63

revenue and television revenue DVs, the highest impact transfer expenditure had on commercial

revenue came from geographic rivalry models 3 and 4, and the lowest impact came from FIFA-

determined rivalry models 1 and 2.

Average Attendance: This variable, which is the average attendance per game for the home

games of the season, is used in total revenue and match-day revenue models. Average attendance

captures the match-day revenue portion of results and has a large, statistically significant

explanatory impact (p value range from 0.00 to 0.02) on both total and match-day revenue DVs.

Table 5.15 in the following page summarizes the coefficient values and financial impact that

average attendance has on both total and match-day revenue.

Table 5.15: Impact of 1,000 change in average attendance on total and match-day revenue Model Total Revenue Impact of 1,000 Match-day Impact of 1,000 unit unit change on y Revenue change on y 1 1.032 (0.153) £4,053,767.99 1.096 (0.108) £1,081,144.05 2 1.063 (0.152) £4,176,800.89 1.070 (0.106) £1,055,228.80 3 1.289 (0.175) £5,075,994.80 1.199 (0.098) £1,183,937.14 4 1.260 (0.176) £4,960,390.43 1.178 (0.093) £1,162,962.61 5 1.303 (0.124) £5,131,827.13 1.350 (0.145) £1,335,006.37 6 1.017 (0.158) £3,994,262.55 1.093 (0.092) £1,078,153.16 -Number between brackets is standard error - Bolded values are statistically significant (at 95% confidence interval) **Starred values are within the 90% confidence interval

Statistically significant for every model across both total and match-day revenue DVs,

average attendance has the largest statistically significant monetary impact on those models. For

total revenue, the range of impact of a 1,000-person change in average attendance is £3.994

million to £5.075 million (a revenue change of £3,994 – £5,075 per attendee per season). For

match-day revenue, the average attendance change of 1,000 will impact from £1.055 million to

£1.335 million (£1,055 to £1,335 per attendee per season).

64

Stadium Capacity: Due to its significant correlation with average attendance, stadium capacity

was not used in the regressions on total and match-day revenue. However, as previously

mentioned, it is a useful proxy for commercial revenue as it estimates amount of advertising real

estate and exposure (quantity of advertising space, number of viewers: both functions of how

much a sponsor would be willing to pay). In the regressions on commercial income as a

dependent variable, stadium capacity was statistically significant for all 6 models.

Table 5.16: Impact of 1,000 change in stadium capacity on commercial revenue

Model Commercial Impact of 1,000 Revenue unit change on y 1 0.498 (0.120) £557,943.81 2 0.510 (0.123) £687,047,99 3 0.829 (0.139) £1,120,265.81 4 0.832 (0.138) £1,124,352.72 5 0.759 (0.140) £1,024,972.10 6 0.696 (0.155) £939,318.58 -Number between brackets is standard error - Bolded values are statistically significant (at 95% confidence interval) **Starred values are within the 90% confidence interval

The range of impact that a change of 1,000 in stadium capacity has on commercial

revenue is from £0.558 million to £1.124 million (a return of £558 – £1,124 per seat). Stadium

capacity’s coefficient is visibly lower for Models 1 and 2, and higher for models 3 and 4. As

previously mentioned, this impact is most probably due to the fact that the variation in

explanatory power (geographic rivalry has more limited explanatory power than FIFA-

determined rivalry, meaning variables will take on more significant roles in the absence of others

that would have statistical significance).

65

Rivalry Variables

Geographic Rivalry: These variables (presence of a local rival, distance to nearest rival) were

deployed to test hypotheses 1 to 1.3. Across the board, the results are inconclusive, due to an

inconsistency in the significance and directional impact of the variables across all DVs.

Table 5.17: Presence of a Local Rival coefficient value on all DVs Model Total Revenue Matchday Television Commercial Revenue Revenue Revenue 1 n/a n/a n/a n/a 2 n/a n/a n/a n/a 3 -0.284 (0.127) 0.151 (0.071) -0.359 (0.160) -0.178 (0.174) 4 -0.326 (0.015) 0.072 (0.070) -0.295 (0.171)* -0.278 (0.183) 5 -0.014 (0.118) 0.260 (0.072) -0.071 (0.160) -0.028 (0.184) 6 0.026 (0.125) 0.197 (0.072) 0.099 (0.163) -0.107 (0.195) -Number between brackets is standard error -Bolded values are statistically significant (at 95% confidence interval) **Starred values are within the 90% confidence interval

For total revenue, geographic rivalry is only significant in models that only incorporate

geographic rivalry (Models 3 and 4). In those models, the impact of geographic rivalry is

statistically significant and the presence of a geographic rival decreases total revenue by 48% to

52.8%. This result contradicts Hypothesis 1, and in combination with the lack of significance

across the other conditions, means that the results do not lend support. When FIFA-determined

rivalry is inserted in conjunction with geographic rivalry, the presence of a local rival variable

loses its explanatory power.

For match-day revenue, the combination of geographic and FIFA-determined rivalries

gives the geographic rival significance that it does not attain on its own. The impact is positive,

where having a geographic rival can increase match-day revenue by 57.40% (Model 6) to

81.97% (Model 5). However, the results as stand-alones are not reliable, and would not be in

66

support of any hypotheses regardless; the absence of a significant effect for geographic rivalry

without the FIFA-determined rivalry variable means that the results are inconclusive.

The only statistically significant coefficient for the presence of a local rival variable on

television revenue is in Model 3 (which is only geographic rivalry), it is at -0.359; the negative

sign casts doubt on Hypothesis 1.1, and the absence of significant effects in conjuncture with

competitive balance variables (Model 4 is only significant at p=0.10, and denotes that geographic

rivalry has a negative effect on revenue) and with FIFA-determined rivalry variables (Models

5,6) means that the television revenue models for geographic rivalry do not lend support to

Hypothesis 1.1.

There is no statistically significant coefficient for the geographic rivalry on commercial

revenue (at p=0.05 or p=0.10) and the negative signs of the coefficients indicate an opposing

relationship to revenue than the one stipulated in Hypothesis 1.2. Therefore, the results from the

commercial revenue models do not lend support to Hypothesis 1.2.

Distance to the Nearest Competitor: Deployed to test the network rivalry effect (refer to

Literature Review for more information), the geographic distance to the nearest competitor

(measured in km) should have an inverse relationship with revenue to support Hypotheses 1.3,

1.3.1 and 1.3.2.

Table 5.18 Distance to Nearest Rival coefficient value on all DVs Regression Total Revenue Matchday Television Commercial Revenue Revenue Revenue 1 n/a n/a n/a n/a 2 n/a n/a n/a n/a 3 -0.151 (0.070) 0.007 (0.039) -0.185 (0.092) -0.135 (0.101) 4 -0.180 (0.017) -0.044 (0.039) -0.143 (0.100) -0.195 (0.108)* 5 -0.065 (0.066) 0.083 (0.042) -0.021 (0.092) -0.040 (0.107) 6 0.068 (0.072) 0.022 (0.042) 0.150 (0.098) -0.055 (0.124) *Number between brackets is standard error -Bolded values are statistically significant (at 95% confidence interval)

67

-Starred values are within the 90% confidence interval

As can be seen from Table 5.18, the distance variable is statistically significant on total

revenue only in models that exclusively incorporate geographic rivalry variables. The addition of

FIFA-determined rivalry variables causes the distance variable to lose statistical significance and

explanatory power. For the models that it is significant (Models 3 and 4), every 100 km increase

in distance causes total revenue to drop from 15.1% to 18.0%. However, this result is not

reliable, and overall, the results from the Total Revenue regressions do not lend support to

Hypothesis 1.3.

There is only one statistically significant distance variable in the Match-day Revenue

regressions (Model 5). The coefficient is positive, implying that an increase in distance to the

nearest rival by 100 km will lead to 8.3% increase in revenue. This can be explained by the fact

that local fans will choose to support to a club closer to them, and the further away a club is from

the area in question, the more likely the fan is to support/attend games with a more proximate

club (in essence, it limits consumer choice and provides a geographic monopoly).

For television revenue, Model 3 (geographic rivalry) was the only model in which the

distance to nearest rival variable was statistically significant. The coefficient of -0.185 implies a

positive change of 18.5% in commercial revenue with every 100 km nearer the nearest

competing team is located. As the only significant outcome for the nearest competitor variable

for television revenue, this result does not lend support to Hypothesis 1.3.1.

The distance to nearest competitor variable does not have any statistically significant

effects on commercial revenue, albeit that Model 3 displays a negative coefficient (which would impact revenue by 19.5%) within the p = 0.10 interval. However, the results cannot lend support to Hypothesis 1.3.2.

68

While there is no sufficient evidence to support any of the hypotheses pertaining to

distance to the nearest competitor, there is sufficient evidence to warrant further investigation

into this issue. The varying levels of significance and the consistent general vector of the

coefficients implies that there might be an effect pertaining to competitor proximity, but that perhaps the distance in km is not the best proxy to measure it. Alternative suggestions include commute time and cost.

FIFA-determined Rivalry: The variable operationalized to test for the significance of FIFA-

determined rivalry is the binary variable that addresses the presence of a rival as determined by

FIFA on the Deloitte Money League Top 20 for that year. The mean value is 0.5, meaning half

the teams have a FIFA-determined rival. Similar to the local rival variable, this variable is

deployed to test Hypotheses 2 – 2.3.

Table 5.19 Presence of a FIFA-determined Rival coefficient value on all DVs Model Total Revenue Match-day Television Commercial Revenue Revenue Revenue 1 0.276 (0.043) 0.107 (0.030) 0.302 (0.066) 0.290 (0.076) 2 0.266 (0.042) 0.115 (0.029) 0.279 (0.064) 0.293 (0.076) 3 n/a n/a n/a n/a 4 n/a n/a n/a n/a 5 0.254 (0.052) 0.138 (0.032) 0.390 (0.079) 0.215 (0.092) 6 0.258 (0.051) 0.138 (0.030) 0.319 (0.076) 0.195 (0.091) -Number between brackets is standard error - Bolded values are statistically significant (at 95% confidence interval) **Starred values are within the 90% confidence interval

As can be seen in Table 5.19, the FIFA-determined rivalry variable is statistically

significant for every across all DVs. The coefficient signs are consistently positive, meaning that

the presence of a rival on the Deloitte Money League ranking positively impacts revenue. This

lends support to Hypotheses 2 – 2.3.

69

Table 5.20 Presence of a FIFA-determined Rival impact on Total and Match-day Revenue Model Total Revenue Impact of change Match-day Revenue Impact of change on on y y 1 0.276 (0.043) 27.6% 0.107 (0.030) 10.7% 2 0.266 (0.042) 26.6% 0.115 (0.029) 11.5% 3 n/a n/a n/a n/a 4 n/a n/a n/a n/a 5 0.254 (0.052) 25.4% 0.138 (0.032) 13.8% 6 0.258 (0.051) 13.8% 0.138 (0.030) 13.8% -Number between brackets is standard error -Bolded values are statistically significant (at 95% confidence interval) **Starred values are within the 90% confidence interval

As can be seen from table 5.20, the most conservative estimate for the impact of having a

rival on the Deloitte Money League ranking is 13.8% on total revenue (Model 6). Without

competitive balance elements, it can go as high as 26.6% (Model 5), and without local rivalry the

impact can range from 26.6% to 27.6% (Models 2 and 1, respectively). With a mean total

revenue value of £200 million (Refer to Appendix C for descriptive statistics), having a FIFA-

determined rival can increase revenue within an average range of £27.6 million to £55.2 million.

The competitive balance variables seem to diminish the magnitude of the FIFA-determined

rivalry coefficient’s impact on total revenue, and there is a significant discrepancy between the figures quoted in Model 6 and the other models. On the side of conservatism (assuming the competitive balance variables are appropriate), and due to the fact that Model 6 does have the highest adjusted R-square value across all DVs, the 13.8% figure is the one that is most appropriate. The results, namely the consistent statistical significance, lend strong support to

Hypothesis 2.

The impact of havingFIFA-determined rival on match-day revenue is smaller for match- day revenue than it is for total revenue, with the figure ranging from 10.7% (Model 1) to 13.8%

70

(Models 5, 6). The addition of geographic rivalry and competitive balance variables seems to

increase the effect of the FIFA-determined rivalry variable on match-day revenue.

Table 5.21 Presence of a FIFA-determined Rival impact on Television & Commercial Revenue Model Television Impact of Commercial Impact of change Revenue change on y Revenue on y 1 0.302 (0.066) 30.2% 0.290 (0.076) 29.0% 2 0.279 (0.064) 27.9% 0.293 (0.076) 29.3% 3 n/a n/a n/a n/a 4 n/a n/a n/a n/a 5 0.390 (0.079) 39.0% 0.215 (0.092) 21.5% 6 0.319 (0.076) 31.9% 0.195 (0.091) 19.5% -Number between brackets is standard error -Bolded values are statistically significant (at 95% confidence interval) **Starred values are within the 90% confidence interval

The impact of the FIFA-determined rivalry variable is greatest on television revenue. The

range of impact is from 27.9% (Model 2) to 39.0% (Model 5). With a mean value of £80 million

for television income, the range of impact is £22.32 million to £31.2 million. The introduction of

competitive balance variables seems to diminish the magnitude of the FIFA-determined rivalry

coefficient, thereby decreasing its effect on television revenue. Overall, while the 31.9% (having

a rival would increase commercial revenue in Model 6 by £25.52 million) obtained in Model 6

(best R-square, all variables included) may potentially be overstated (the adjusted R-square for

television income in Model 6 is at 0.579, and there are some unobserved variables such as

number of televised games that can mitigate the impact of the FIFA-determined rivalry variable),

the consistent significance of the rivalry variable across all Models regressed on television

income lends strong support to Hypothesis 2.1.

The impact of the FIFA-determined rivalry variable on commercial revenue ranges from

19.5% (Model 6) to 29.3% (Model 1). The addition of geographic rivalry variables and their

corresponding competitive balance predictors diminishes the impact of the FIFA-determined

71

rivalry variable; with a mean value of £69 million, the average impact of having a rival ranges from £13.455 million to £20.27 million. The low value for the commercial revenue adjusted r- square (0.459 in Model 6) implies that the effect may be overstated, but its statistical significance across all 6 models still lends strong support to Hypothesis 2.2.

Overall, the impact of the FIFA-determined rivalry variable on all six models across all four dependent variables indicates that there is a strong relationship between rivalry and financial performance. The presence of a rival on the Deloitte rankings influences financial performance positively and significantly. It is important to note that the average combined monetary impacts of the sub-revenue streams (match-day, television and commercial) sums up to more than the impact on total revenue. This does not take away from the strength of the finding, and can be attributed to several reasons, including:

1. The intercorrelation of revenue streams: Revenue streams are correlated, and thus the

impact that is seen on each is not entirely down to that particular DV, rather it involves

endogeneity from other DVs.

2. The r-square for television and commercial income indicate that there are unobserved

variables that can be deployed to obtain better models. The inclusion of these variables

can mitigate the impact of the FIFA-determined rivalry variable to more accurate levels.

72

Competitive Balance Variables

For Geographic Rivalry (Models 4 &6)

Table 5.22 3Yr Avg Goal Difference with nearest competitor from 1 coefficient value on all DVs Regression Total Revenue Match-day Television Commercial Revenue Revenue Revenue 1 n/a n/a n/a n/a 2 n/a n/a n/a n/a 3 n/a n/a n/a n/a 4 0.115 (0.062)* 0.142 (0.033) -0.020 (0.088) 0.240 (0.094) 5 n/a n/a n/a n/a 6 0.159 (0.055) 0.003 (0.003) -0.003 (0.080) 0.264 (0.096) -Number between brackets is standard error - Bolded values are statistically significant (at 95% confidence interval) **Starred values are within the 90% confidence interval

The 3-year average goal difference distance (absolute value) from 1 is supposed to denote

the recent state of competitive balance between two teams. In the case of Geographic rivalries,

the two teams are the two teams that are closest together from the Deloitte Money League Rank.

In the event that there is no nearby team, the value assigned was significantly higher than the

highest distance (almost twice as high) in order to ensure that it would not skew the results

favorably. This variable is deployed with the aim of supporting Hypotheses 1.4, 1.4.1 and 1.4.2.

As is evident from the results, there is no support for any of the aforementioned hypotheses. In total revenue, the only time the goal difference variable attains significance is in

Model 6, and the coefficient is positive (implying that the greater the goal difference distance from 1 or the less competitive the recent run of head to head fixtures, the greater the impact on revenue. As a matter of fact, for every 3-year average goal more than 1, total revenue would increase by 44.2%. The results on the total revenue regressions do not lend support to

Hypothesis 1.4. Model 4 is significant at p=0.10, but also has a positive sign and implies a relationship opposite to the one postulated by Hypothesis 1.4.

73

Similarly, Model 4 for match-day revenue is statistically significant and suggests that an

increase in the average goal difference would increase revenue by 38.6%. No other models are

significant at p=0.05 or p=0.10. Television revenue models do not have any significance for the

goal difference variable, albeit that they carry the expected negative sign. The results lend no

support to Hypothesis 1.4.1

Commercial revenue regressions yield a significant result for goal difference in models 4 and 6, but both are positive and imply an inverse relationship between competitive balance and financial performance. Therefore, the results do not support Hypothesis 1.4.2.

In similar vein, the rank difference between a team and their nearest competitor yields no significant results for any of the dependent variables, and therefore offer no support to

Hypotheses 1.5, 1.5.1 and 1.5.2.

To suggest that competitive balance is unfavorable in sports based on the results of an inductively constructed variable would be ill-advised (and in disagreement with all sports literature on the matter as far as this author’s knowledge); a possible explanation is that the variable should operationalized over a longer time span (albeit that literature was in support of the 3-year time period). Another likely explanation, which is supported by the negative results generated from the geographic rivalry variable, is that a geographic rival is actually detrimental to demand (income), and thus a disparity between the rivals can have significant influences on the decisions of sponsors and fans. This explanation becomes likelier when evaluating that television income produced negative (albeit statistically insignificant) results; sponsors may want

to pick a clear winner, match-going fans may want to see blow-outs, but broadcasters will want

to appeal to the excitement that consumers feel (Whannel, 1993) (Baimbridge, Cameron &

Dawson, 1996).

74

However, unlike the goal difference variable, the signs are consistently negative for all

four dependent variables across all the applicable models. A possible explanation for the absence

of a significant effect is that the rank number is static for season end, and does not carry into the

next season. Lagging the variable could help the rank difference variable attain explanatory

power, and measuring the effect of rank difference over the previous years could make it

statistically significant as well.

For FIFA-determined Rivalry (Models 2 & 6)

The goal difference variable for FIFA-determined rivalry is operationalized in the same

manner as the geographic rivalry goal difference variable, with the difference being that the 3

year goal difference average is measured against the rival FIFA-determined by FIFA (if

applicable). A large goal difference was assigned for teams that did not have a FIFA-determined

rival. This variable is deployed with the aim of testing Hypotheses 2.3, 2.3.1 and 2.3.2.

The variable behaved largely in the manner it was expected to, which is in contrast to the

geographic goal difference variable.

Table 5.23 3-Yr Avg Goal Difference with FIFA-determined rival from 1 coefficient value on all DVs Regression Total Revenue Match-day Revenue Television Commercial Revenue Revenue 1 n/a n/a n/a n/a 2 -0.083 (0.032) 0.069 (0.022) -0.199 (0.048) 0.028 (0.059) 3 n/a n/a n/a n/a 4 n/a n/a n/a n/a 5 n/a n/a n/a n/a 6 -0.149 (0.039) 0.040 (0.022)* -0.174 (0.054) -0.108 (0.070) -Number between brackets is standard error - Bolded values are statistically significant (at 95% confidence interval) **Starred values are within the 90% confidence interval

75

The negative coefficient of the goal difference variable for both models 2 and 6 implies a

positive relationship between competitive balance and revenue for determined rivalries. Model 2

(FIFA-determined rivalry with competitive balance) implies that a change in the 3-year average

goal difference distance from 1 between FIFA-determined rivals would impact total revenue by

17.39%, while Model 6 stipulates that the impact is actually at 29.04%. The significant disparity between the two is caused by the introduction of the geographic rivalry variables (and their

respective competitive balance proxies), but the consistent significance means that the results

lend support to Hypothesis 2.3.

The match-day results show a negative relationship between competitive balance and

match-day revenue, which does not contradict any hypotheses but also implies that match going

fans (i.e. the higher involvement consumers) prefer to see games where rivals are heavily

defeated. One variable is statistically significant (Model 2), while Model 6 is significant at

p=0.10.

Television revenue was significant for both models 2 and 6, and signifies a positive

relationship between competitive balance and television revenue; Model 2 stipulates that the

impact of a change in the distance of the 3-year goal difference average from 1 on television

revenue would be 36.76%, while the coefficient in model 6 indicates that the difference would be

33.01% of television revenue. The statistical significance of both lends support to Hypothesis

2.3.1.

The average goal difference variable was not found to be statistically significant (or

within the 10% confidence interval) for the commercial revenue DVs; this means that the results

do not lend support to Hypothesis 2.3.2.

76

In summary, the goal difference variable was found to be supportive for FIFA-

determined rivalries. Similar to the binary variables, the results are contrasting with geographic

rivalry results, and this suggests that it is not the geographic aspect of rivalry that is financially beneficial.

For total, television and commercial revenue, the values were not statistically significant for the rank difference from FIFA-determined rival variable. Therefore, rank difference variable for FIFA-determined rivals did not lend support to Hypotheses 2.4, 2.4.1 or 2.4.2.

77

Table 5.24: Rank Difference from FIFA-determined Rival coefficient on all DVs

Regression Total Revenue Match-day Television Commercial Revenue Revenue Revenue 1 n/a n/a n/a n/a 2 -0.002 (0.003) -0.005 (0.002) -0.002 (0.005) -0.006 (0.006) 3 n/a n/a n/a n/a 4 n/a n/a n/a n/a 5 n/a n/a n/a n/a 6 -0.001 (0.004) 0.131 (0.032) -0.005 (0.006) 0.004 (0.007) -Number between brackets is standard error -Bolded values are statistically significant (at 95% confidence interval) **Starred values are within the 90% confidence interval

Match-day revenue yielded statistically significant results for both models 2 and 6,

however while Model 2 yielded a negative coefficient (indicating a positive relationship between competitive balance and revenue), Model 6 yielded a positive coefficient. The insertion of geographic rivalry variables (and/or their competitive balance predictors is what is responsible

for this result, and thus no postulations can be made about the rank difference variable for FIFA- determined rivals and its influence on match-day revenue.

78

Chapter 6: Conclusion & Discussion 6.1 Summary of Findings

Geographic Rivalry Hypotheses

Ultimately, the Study results did not lend support to Hypothesis 1 or any of the sub-

hypotheses falling under Hypothesis 1. On the contrary, the findings, while inconclusive,

suggested that local rivalry had a negative impact on revenue (refer to Table 5.17), and the

findings were inconclusive on the spatial competition effects (the distance variable – refer to

Table 5.18). In similar fashion, the 3-year goal difference distance from 1 for geographic rivals,

one of the competitive balance variables, behaved in a manner opposite to Hypotheses 1.4, 1.41

and 1.42 – and the results suggest that a state of competitive imbalance between local rivals is

more favorable towards revenue. In connection to the literature, the presence of a rivalry and the

state of competitive balance’s effects on revenue are as Porter put them from a Geographic

perspective: the competitive balance harms revenue, and thus, the findings are concurrent with

what the literature has to say on other industries. The distance variable was not significant and

did not lend support to Hypotheses 1.3, 1.31, 1.32, but that is primarily due to the

operationalization of the variable rather than what the results modeled.

FIFA-determined Rivalry Hypotheses

The results lend strong support to Hypotheses 2 – 2.31 and no support towards

hypotheses 2.32, 2.4, 2.41 and 2.42. It was found that the presence of a FIFA-determined Rival

on the Deloitte Money League rankings in the same year positively influences financial

performance across all four dependent variables (refer to Appendix D), and there was partial

evidence to support that a state of equal competitive balance between rival clubs would

positively impact revenues as well (refer to Table 5.23). The goal difference variable was

79

statistically significant for the total, match-day and television revenue, but was not significant for

commercial revenue.

6.2 Contributions/Implications

Theory

In similar vein to the remainder of industries, geographic market dynamics seem

to operate in the same manner in sports as they do in other industries. The presence of a

local competitor will constrain profitability, and a state of equal competitive balance will

also limit revenue. This is in line with how rivalry operates in business (Porter, 1985),

and there is no theoretical contribution in economics as far as geographic rivalry is

concerned.

However, interestingly enough, it seems like a state of competitive imbalance

between local rivals or geographically proximate competitors in Deloitte’s Money

League rankings can generate additional revenue. This is a counter-intuitive finding; the

premise of sport, as put by Whannel, is the very uncertainty of outcome and excitement

(Whannel, 1993).

From the general (FIFA classified) rivalry perspective, the results suggest that

having rivalries (just not local ones) makes a football club thrive financially, and that

implies that the business view on rivalry is fundamentally different from the sporting

industry perspective on rivalry: in sports, rivalry is good for business. This opens the door

to many possibilities with regards to rivalry literature extension into the sporting industry,

and an outright exception to Porter’s 5 Forces Model (Porter, 1979).

Furthermore, there is partial support towards competitive balance being positive

for rivals (refer to FIFA-determined Rivalry Hypotheses section), and that is also in

80

contrast to how business strategy views competition, and a unique extension atop

literature concerning consumer and stakeholder utility from sports.

Managerial

This paper can potentially have several contributions that apply to broadcasters,

policymakers in leagues and sports governing bodies, executives in clubs, sponsors.

For Broadcasters: Though behavior already indicates that broadcasters are aware of this,

broadcasters may be now more inclined to negotiate broadcasting rights with individual clubs

rather than with leagues as a whole (thereby able to optimize cost through discarding or underpaying for games of lesser attractiveness to the fans), and further develop a more integrated role in the marketing activities of the clubs to which they are contractually tied to showcase the rivalry effects in order to guarantee greater fan pull.

For Policy Makers: There are several policy makers that would benefit from this information.

League policy makers would be more likely to consider imbalanced fixture schedules that favor

local rivalries (meaning that local rivals would play against each other more so than other clubs),

and emphasize points of sale in their negotiation agreements with broadcasters, and develop a

more cautious anti-trust policy when dealing with financials of multi-rival clubs. European

Football’s governing body, UEFA, is currently deploying Financial FairPlay rules, which are

rules governing the expenditure of clubs and restricting it to set proportions of their income in

order to prevent large clubs from utilizing their better credit to incur great amounts of debts that

sustain their competitive advantage and give everyone an opportunity to increase their off and on

field performance. This would be an aspect to consider in order to level the playing field; greater

allowances should be made for clubs who would opt to invest in their rivals (should this prove as

financially beneficial, as addressed by one of the research questions) as that would reduce the

81

rivals spending to income proportion and elongate the period through which the investor club

would recoup their gains.

Another host of European clubs, an association known as the G18 have been aiming to break

away from their local leagues and UEFA by creating a league that combines Europe’s top teams;

it would be advisable that they consider local rivalries and their impact – especially with teams

performing at a similar level.

Furthermore, for policymakers aiming to stimulate and develop local leagues and grow the

industry in the country, the potential findings of this paper would suggest to create more than

once club per market, in order to ensure that the development of rivalry can generate the

attention of the fans, and consequently broadcasters and sponsors.

For Executives: Club executives may look at conditions where they can increase revenue streams

they generate by stimulating a rival, and may find a better ROI in doing so than they would had

they spent it on themselves. In 2013, German club Bayern Munich famously refunded their fans’

season tickets after a season where they were peerless winners in the Germany league, with

controversial club president Uli Hoeness declaring that ‘the entertainment factor was absent for

the fans because there was no competition’. Investing to make a rival more competitive could be

a route that improves the attractiveness of the product that the club itself is offering, and

consequently improves the bottom line.

6.2 Limitations

This research suffers from several limitations, including: 1. The data set is not representative: This dataset measures the financial performance of

Europe’s top 20 clubs by financial performance (heteroskedasticity). The findings

therefore cannot be generalized until more studies find similar results with other club

82

distributions, particularly across entire leagues (so that the biggest and smallest clubs are

put to the same test).

2. Unobserved Variables impacting coefficients: In many instances, especially when

concerning television and commercial income (refer to Appendix D), there are some

unobserved variables that are influencing the magnitude of statistically significant

variables. For example, if the coefficients are to be taken at face value, average

attendance can financially influence match-day revenue more than it can total revenue (in

absolute terms). Given that match-day revenue is a sub-constituent of total revenue, that

is pragmatically impossible (assuming average attendance does not negatively influence a

revenue stream, which it does not), there are lacking variables that would give more

accurate coefficient values for the other variables. Variables such as the number of

televised games and number of sponsors would improve television and commercial

revenue respectively, and further put the weight of the other coefficients in a more

accurate scale. While the results show that rivalries positively impact profitability, the

coefficients are not necessarily the best indicators of that magnitude of that influence.

Furthermore, if data was available on travel costs or commute times to the nearest

geographic rival, it would have posited a better test of the spatial dimension of rivalry

(i.e. how close, near, conveniently accessible is a club’s nearest competitor). The number

of trophies won, a win ratio against a rival are better alternatives to league rank difference

in terms of what constitutes competitive balance (Neale, 1964).

3. Method does not allow for examining the influence of previous years: The MLR is static,

and thus cannot measure the impact of a previous year’s league standing, for example. A

method that allows for examining the impact of an event on subsequent years (lagging,

83

EFA) may produce more accurate results on the exact role of rivalry, and shed more

insight into the role of competitive balance.

6.3 Future Research Considerations

The findings of this thesis open up some exciting research avenues when concerning rivalry in soccer, or sports in general. Most importantly however, the most important avenue for research to explore is the uniformity of this finding.

Given the concerns cited with regards to whether rivalry is positively related to income for all clubs, or just top performing clubs, is an important question to answer. The first future research consideration must be about testing the generalizability of this principle. Not only would this lend significant implications for investors, broadcasters and sponsors, but it will also

explicate the scale of the finding found, and whether it’s a unique case of elite clubs or

applicable irrespective of what pre-conditions are there.

Another important future research consideration is regarding rivalry antecedents. This

study took rivals as determined by FIFA and grouped them together, albeit that classifying with

antecedents (historical, political, religious, performance-based – see literature review) in order to

determine if there is one (or a group) that is driving this effect.

More research must be conducted with regards to geographic rivalry in order to

determine an impact on revenue streams. While the findings of this study lent strong support to

ruling out local rivalries as beneficial to revenue, spatial rivalries (nearest competitor) may be a

potential rivalry dimension that positively impact income (especially given the negative – albeit

statistically insignificant – coefficients produced by the distance to nearest competitor variable).

84

Research that measures the impact of rivalry on subsequent years of financial

performance is also important to accurately capture the role of rivalry on financial results. A

more dynamic model than an MLR is required.

Measuring rivalry intensity should be an objective; literature currently says that rivalry is

a negative affective disposition towards a particular team. But the measurement of this

disposition on the playing staff is important in understanding what truly constitutes a rivalry. In-

game statistics should be used to proxy this: events such as shots on target, yellow and red cards,

help capture the intensity of a rivalry, and thus develop a better idea on the effect of rivalry

intensity, and determine an empirical measure for rivalry intensity.

85

Bibliography

Al Khattab, S. A. (2007). Marketing Strategic Alliances: The Hotel Sector in Jordan. International Journal of Business and Management , 2012. Andreff, W., & Staudohar, P. D. (2000). The Evolving European Model of Professional Sports Finance. Journal of Sports Economics , 257-276. Baimbridge, M., Cameron, S., & Dawson, P. (1996). Satellite Television and the Demand for Football: A Whole New Ball Game? Scottish Journal of Political Economy , 317-333. Boulding, K. (1963). Microeconomics. New York: Harpers. Boyle, R., & Haynes, R. (2002). New Sport Media. Sport, Culture and Society , 96-114. Bruggink, T. H., & Roosma, C. (2011). Interleague play and the Big Mac attack: Estimating the within-season demand for Major League Baseball. International Business and Economics Research Journal . Colander, D. C. (2004). Economics Fith Edition. New York: McGraw Hill/Irwin. Dalakas, V., & Mancon, J. P. (2012). Fan identification, Schadenfreude toward hated rivals, and the mediating effects of Importance of Winning Index (IWIN). Journal of Services Marketing , 51-59. Deloitte. (2005). Annual Review of Football Finance. London: Deloitte. Deloitte. (2006). Annual Review of Football Finance. London: Deloitte. Deloitte. (2007). Annual Review of Football Finance. London: Deloitte. Deloitte. (2008). Annual Review of Football Finance. London: Deloitte. Deloitte. (2009). Annual Review of Football Finance. London: Deloitte. Deloitte. (2010). Annual Review of Football Finance. London: Deloitte. Deloitte. (2011). Annual Review of Football Finance. London: Deloitte. Deloitte. (2012). Annual Review of Football Finance. London: Deloitte. Deloitte. (2013). Annual Review of Football Finance. London: Deloitte. Deloitte. (2014). Annual Review of Football Finance. London: Deloitte. Deloitte. (2005). Deloitte Money League London: Deloitte. Deloitte. (2006). Deloitte Money League London: Deloitte. Deloitte. (2007). Deloitte Money League London: Deloitte.

86

Deloitte. (2008). Deloitte Money League London: Deloitte. Deloitte. (2009). Deloitte Money League London: Deloitte. Deloitte. (2010). Deloitte Money League London: Deloitte. Deloitte. (2011). Deloitte Money League London: Deloitte. Deloitte. (2012). Deloitte Money League London: Deloitte. Deloitte. (2013). Deloitte Money League London: Deloitte. Deloitte. (2014). Deloitte Money League London: Deloitte. Dreyer, D. R. (2013). Exploring the Concept of Rivalry: From and to the Yankees and Red Sox. Journal of Political Science Education , 308-319. FIFA. (2013). Laws of the Game. Retrieved from FIFA: http://www.fifa.com/mm/document/footballdevelopment/refereeing/81/42/36/log2013en_neutral. pdf FIFA, 2015. Classic Football Club Rivalries. Retrieved from FIFA: http://www.fifa.com/classicfootball/clubs/rivalries/ Fort, R., & Fizel, J. (2004). International Sports Economics Comparisons. Westport: Praeger Publishers. Furtonato, J. A. (2006). Scheduling promotional events in Major League Baseball: examining team and sponsor desires. International Journal of Sports Marketing & Sponsorship . Ghoshal, S. (1987). Global Strategy: An Organizing Framework. Strategic Management Journal , 425-440. Gordon, I. R., & McCann, P. (2000). Industrial Clusters: Complexes, Agglomerations, and or Social Networks. Urban Studies , 513-532. Guschwan, M. (2011). Fans, Romans, Countrymen: Soccer Fandom and Civic Identity in Contemporary Rome. International Journal of Communication . Henderson, D. B. (1983). The Anatomy of Competition. Journal of Marketing , 7-11. Henderson, J. V. (2001). Marshall's Scale Economies. Journal of Urban Economics , 1-28. Ioakimidis, M. (2010). Online marketing of professional sports clubs: engaging fans on a new playing field. International Journal of Sports Marketing & Sponsorship , 271-282. James, J. D., & Ross, S. D. (2004). Comparing Consumer Motivations Across Multiple Sports. Sports Marketing Quarterly , 17-25. Kelly, S., Hoffman, K. D., & Carter, S. (1987). Franchise relocation and sport introduction: a sports marketing case study of the Carolina ’ fan adoption plan. Journal of Services Marketing , 469-480.

87

Kraszewski, J. (2008). Pittsburgh in Fort Worth: Football bars, sports television, sports fandom, and the management of home. Journal of Sport and Social Issues , 139-157. Krugman, P. R. (1993). On the Relationship Between Trade Theory and Location Theory. Review of Economics , 110-122. Krugman, P. R. (1991). The Case of the US Manufacturing Belt. In P. R. Krugman, Geography and Trade (pp. 11-25). Palatino: MIT Press. Lederman, M. (2007). Do enhancements to loyalty programs affect demand? The impact of international frequent flyer partnerships on domestic airline demand. The RAND Journal of Economics , 1134-1158. Lemke, R. J., Leonard, M., & Tlhokwane, K. (2010). Estimating Attendance at Major League Baseball Games for the 2007 Season. Journal of Sports Economics , 316-348. Mason, D. S. (1999). What is the Sports Product and who Buys it? The Marketing of Professional Sports Leagues. European Journal of Marketing , 402-419. Neale, W. C. (1964). The Peculiar Economics of Professional Sports. The Quarterly Journal of Economics , 1-14. Osborne, E. (2008). Rivalries. 83rd Annual Western Economic Association. Honolulu. Papageorgiou, G. (1979). Agglomeration. Regional Science and Urban Economics , 41-59. Paul, R. J., & Weinbach, A. P. (2013). Determinants of Dynamic Pricing Premiums in Major League Baseball. Sport Marketing Quarterly . Porter, M. E. (1979). How competitive forces shape strategy. Harvard Business Review , 137- 145. Raspaud, M., & Lachheb, M. (2014). A Centennial Rivalry, Ahly vs. Zamalek: Identity and Society in Modern Egypt. In Identity and Nation in African Football: Fans, Community and Clubs (pp. 99-115). Palgrave MacMillan. Sanford, K., & Scott, F. (2014). Assessing the Intensity of Sports Rivalries Using Data from Secondary Market Transactions. SSRN . Siegfried, J., & Zimbalist, A. (2000). The Economics of Sports Facilities and Their Communities. Journal of Economic Perspectives , 95-114. Sloane, P. J. (1971). The Economics of Professional Football: The Football Club as a Utility Maximizer. Scottish Journal of Political Economy , 121-146. Stølen, T., Chamari, K., Castagna, C., & Wisløff, U. (2005). Physiology of Soccer. Sports Medicine , 501-536.

88

Talarico, J. M., & Moore, K. M. (2012). Memories of 'The Rivalry': Differences in How Fans of Winning and Losing Teams Remember the Same Game. Applied Cognitive Psychology , 746- 756. Tyler, B., & Cobbs, J. (2009). Advancing toward and Understanding of Sport Rivalry. Sport Marketing Association. Cleveland: SSRN. Wann, D. L., Royalty, J., & Roberts, A. (2000). The Self-Presentation of Sport Fans: Investigating the Importance of Team Identification and Self-Esteem. Journal of Sport Behvior . Weis, K. (1986). How Print Media Affects Sports and Violence: The Problem of Sports Journalism. International Review for the Sociology of Sport , 239-258. Whannel, G. (1992). Fields in Vision: Television Sport and Cultural Transformation. London: Routledge. Zea, M., & Feldman, D. (1998). Going Global: The Risks and Rewards of Airline Alliance- Based Network Strategies. In F. G. Butler, & R. M. Keller, Handbook of Airline Marketing (pp. 545-549). Evanston, IL: Northwest Airlines. Zimbalist, A. (2003). Sport as a Business. Oxford Review of Economic Policy , 503-511.

89

APPENDICES

90

Appendix A: Stakeholder Analysis: Consumers of the Football Product Customer Description Motivation Fans The end-users who actually watch the The excitement that is derived from games and support the respective the uncertainty of the outcome. teams. Social construction and the Purchase memorabilia and club wear fulfillment of nationalistic/regionalist and various items that demonstrate affiliations. their affiliation and loyalty to the club/institution. Identity construction; that can either be tied to geographic spheres, the success of the institution or the “underdog” tag. Networks Television broadcasters who purchase The objective is to maximize revenue the rights to televise the competitive by gaining maximum possible games and broadcast them to a viewership, which subsequently specific region or lease them to others allows for the broadcasters to charge for a set period of time. premium pricing for the advertisers, sponsors and sub-leasers of the Can either negotiate deals with clubs product. individually or the league as a whole. They will therefore be attracted to the Can act as middlemen who sub-lease same features that attract the fans, as the broadcasting rights to regional per their efforts to generate fan broadcasters. viewership. Sponsors Corporations who aim to be affiliated The objective is to maximize with the clubs in order to gain exposure and therefore clubs with the exposure through the club’s activities highest marketability and attention . from the customers in certain Their range of agreements can vary geographic scopes will be targeted. from being an official sponsor or partner to having their logo within the They will be looking for some stadium or on the jersey(s), can also attributes that are similar to what the be the kit manufacturer and official fans look for: on-pitch performance transport partner, and can offer and brand ambassadors being critical products and services through the factors in their decision. club. Government Looking for clubs whose business activities can serve objectives. Their search will be based on a set of criteria that varies depending on the Can offer payment in the form of tax objective at hand. cuts, subsidies and contractual agreements.

91

Appendix B: On Sources of Revenue

Gate Receipts and Match-Day Sales Historically a major source of income for all sporting clubs, gate receipts and fan commercial expenditures within the premises comprised as much as 85% of average club income annually in the 1970’s (Andreff & Staudohar, 2000). This particular source of revenue is comprised of ticket sales (that can either be done on a per-season basis at a discounted per-match rate or individually) and supplementary services offered within the site during sporting events (this includes food and beverage offerings, on-site item sales, catering and so forth). The tickets themselves adhere to a tiered pricing system that is typically determined by the proximity to the event and the ‘class’ of the ticket (with executive class tickets offering a more exclusive experience that also includes catering services, being in proximity to club personnel and brand ambassadors and ‘giveaways’). Memorabilia and Merchandising This includes items such as club official replica sportswear, branded items, media, collectibles and whatever a club chooses to dispense through its retail channels and on-site megastore. This source of revenue comprised an average 10% of club income in the late 1990’s (Andreff & Staudohar, 2000). Memorabilia sales can be done through what is typically known as the club megastore (on-site), through locations outside that belong to the club, or through vendors and retail channels that are within the club’s supply chain. Merchandising is directly tied to the marketability of the club and the present playing staff that act as the face of the organization; demand increases when marketable players play for the club, with income from jersey sales alone reaching up to 50 million sterling pounds per player per season at times (Rodriguez, 2014). Sponsorship and Advertising In the earlier days sponsorship and advertising revenue did not comprise a large portion of the club’s spending. In contrast, it is now one of the two most dominant sources of income for a top flight football club competing in one of Europe’s top five leagues (Baimbridge, Cameron & Dawson, 1996). Sponsorship and advertising revenue will utilize an integrated marketing communication strategy that utilizes the club events to present products or increase visibility and credibility through communication channels that capitalize on the clubs existing audience. Sponsorship revenue is comprised of several facets: 1) sponsors who appear on club wear (such as team jerseys; this typically includes one major sponsor whose logo is visible at the front of the team jersey and the sporting company that manufactures the jersey), 2) sponsors who appear on the clubs advertising billboards at match events, 3) partners that appear through public relations and media relations. The size of sponsorship packages is influenced by a multitude of factors, including the visibility of the sponsor on the club’s ‘real estate’ (be it the training ground, stadium, conference rooms, or even club attire), the role that the sponsor plays with the club and

92

the nature of the exposure (local, continental or global).There exists a direct relationship, therefore, between the club’s brand appeal and the amount of sponsorship revenue that they incur. Television and Broadcaster Revenue Initially, television and revenue obtained from broadcasters in exchange for televising club matches was a small, if not non-existent source of income for European football clubs (Jenrennaud & Kesenne, 2006). In 1976, French Football Club Olympique De Marseilles famously turned down an offer of $50,000 per match in order to televise its matches (Baimbridge, Cameron & Dawson, 1996). However, this would not persist for long, and since the late 1980’s, television revenue has been comprising an increasing proportion of total club receipts, rising to comprise over 33% of the English Premier League’s turnover in 2004-2005 (Deloitte, 2005). There are two primary models that clubs employ when conducting transactions with television broadcasters. The first model, employed in Spain, implies an individual bargaining process for television rights – meaning that the each member club of the league would negotiate the rights to broadcast single or multiple games with the television providers. The other model employed is collective. In this model, the league association negotiates a television contract individually prior to the discussion and implementation of revenue distribution amongst the member clubs based on multiple criteria (including on-field merit, number of matches televised and home/away fixture allocation). Figure AB.1 below shows the differing redistribution of income agreements within the respective European football leagues:

Figure AB.1 Television Rights Redistribution in European Football Leagues (Jenrennaud & Kesenne, 2006)

93

Appendix C: Descriptive Statistics

Descriptive Statistics N Minimum Maximum Mean Std. Deviation City Pop 199 256652.0 13854740.0 2422594.465 2948407.2575 log of population 199 12.46 16.44 14.0947 1.07610 Per Capita GDP (000s) 200 .00 85100.00 35082.0250 12416.27594 log of GDP per capita 199 9.14 11.35 10.4109 .35246 League Position 200 1 21 3.95 3.623 Champions League 200 0 1 .70 .459 Participation Europa League Participation 200 0 1 .16 .372 number of games played 200 1.00 31.00 14.6000 5.82750 minus regular season games Average Attendance (000s) 200 20.872 80.521 50.93747 14.569763 Log(attendance) 200 1.32 1.91 1.6883 .13110 Stadium Capacity 200 32115 115000 61291.34 17538.992 log of stadium capacity 200 10.38 11.65 10.9829 .28592 Transfer Expense (mil eur) 200 .00 257.40 45.0067 39.45130 log of transfer expenditure 195 -4.02 5.55 3.3882 1.17430

Presence of App Rival 0/1 200 0 1 .50 .501 Presence of Local Rival 0/1 200 0 1 .41 .494 distance to nearest team 200 .00 1000.00 222.2720 277.76379 rank difference from app 200 .00 20.00 7.7050 7.08725 rank difference from nearest 200 .00 20.00 4.1200 4.65445 Total Rev (mil eur) 200 63.10 518.90 200.0235 96.85934 log of total revenue 200 4.14 6.25 5.1932 .45375 Matchday Income (mil eur) 200 7.70 137.50 50.2000 32.83990 TV Income (mil eur) 200 11.30 199.20 80.8745 40.74102 log of telivision income 200 2.42 5.29 4.2516 .56379 Commercial Income (mil 200 16.00 254.70 68.9490 44.37580 eur) log of commercial income 200 2.77 5.54 4.0644 .56983 Valid N (listwise) 193

94

Appendix D: Study Results DV: Total Revenue

Variable Dimension Variable Model 1 Model 2 Model 3 Model 4 Model 5 Model 6 Socio-Demographic Population 0.044 (0.020) 0.046 (0.019) 0.084 (0.026) 0.076 (0.026) 0.037 (0.023) 0.037 (0.022) Income 0.094 (0.056)* 0.094 (0.055)* 0.031 (0.068) 0.043 (0.068) 0.186 (0.065) 0.133 (0.060)

Performance League Rank -0.011 (0.006)* -0.011 (0.006)* -0.012 (0.007)* -0.012 (0.008) -0.012 (0.006)* -0.007 (0.007) UCL (0/1) 0.062 (0.071) 0.077 (0.274) 0.114 (0.207) 0.108 (0.090) 0.052 (0.078) 0.084 (0.077) EL (0/1) -0.084 (0.076) -0.072 (0.075) -0.155 (0.097) -0.166 (0.097)* -0.046 (0.085) -0.049 (0.084) GamesPlayed 0.015 (0.004) 0.014 (0.004) 0.021 (0.005) 0.023 (0.005) 0.014 (0.005) 0.016 (0.005)

Transfer Exp. 0.104 (0.016) 0.107 (0.016) 0.126 (0.019) 0.124 (0.019) 0.120 (0.017) 0.109 (0.017) Avg Attendance 1.032 (0.153) 1.063 (0.152) 1.289 (0.175) 1.260 (0.176) 0.303 (0.124) 1.017 (0.158) Geographic Rivalry Local Rival (0/1) n/a n/a -0.284 (0.127) -0.326 (0.015) -0.014 (0.118) 0.026 (0.125) Dist. Rival (km) n/a n/a -0.151 (0.070) -0.180 (0.017) -0.065 (0.066) 0.068 (0.072)

Determined Rivalry FIFA Rival (0/1) 0.276 (0.043) 0.266 (0.042) n/a n/a 0.254 (0.052) 0.258 (0.051)

Competitive Balance Local Comp Rank Diff Local n/a n/a n/a -0.003 (0.006) n/a -0.005 (0.310) 3 Year GD Local n/a n/a n/a 0.115 (0.062)* n/a 0.159 (0.055) FIFA Comp Rank Diff FIFA n/a -0.002 (0.003) n/a n/a n/a -0.001 (0.004) 3 Year GD FIFA n/a -0.083 (0.032) n/a n/a n/a -0.149 (0.039)

Constant 1.210 (0.585) 1.181 (0.576) 1.153 (0.711) 1.193 (0.712)* -4.803 (1.329) 0.786 (0.202) Adjusted R Square 0.711 (0.245) 0.721 (0.241) 0.635 (0.286) 0.639 (0.284) 0.729 (0.246) 0.736 (0.243)

95

DV: Match-day Revenue

Variable Dimension Variable Model 1 Model 2 Model 3 Model 4 Model 5 Model 6 Socio-Demographic Population 0.064 (0.014) 0.063 (0.013) 0.045 (0.014) 0.037 (0.014) 0.042 (0.014) 0.026 (0.014)* Income 0.030 (0.039) 0.031 (0.039) 0.042 (0.038) 0.060 (0.098)* 0.018 (0.040) 0.069 (0.035)

Performance League Rank 0.001 (0.004) 0.001 (0.004) -0.000 (0.004) -0.002 (0.004) 0.002 (0.004) -0.001 (0.004) UCL (0/1) -0.078 (0.119) -0.092 (0.049) 0.017 (0.050) 0.003 (0.047) 0.000 (0.048) -0.025 (0.045) EL (0/1) -0.118 (0.053) -0.128 (0.052) -0.078 (0.054) -0.095 (0.051)* -0.034 (0.052) -0.051 (0.049) GamesPlayed 0.014 (0.003) 0.015 (0.003) 0.011 (0.003) 0.014 (0.003) 0.008 (0.003) 0.011 (0.003)

Operational Strategy Transfer Exp. 0.010 (0.011) 0.007 (0.011) 0.018 (0.011)* 0.016 (0.010) 0.007 (0.011) 0.005 (0.010) Avg Attendance 1.096 (0.108) 1.070 (0.106) 1.199 (0.098) 1.178 (0.093) 1.350 (0.145) 1.093 (0.092) Geographic Rivalry Local Rival (0/1) n/a n/a 0.151 (0.071) 0.072 (0.070) 0.260 (0.072) 0.197 (0.072) Dist. Rival (km) n/a n/a 0.007 (0.039) -0.044 (0.039) 0.083 (0.042) 0.022 (0.042)

Determined Rivalry FIFA Rival (0/1) 0.107 (0.030) 0.115 (0.029) n/a n/a 0.138 (0.032) 0.138 (0.030)

Competitive Balance Local Comp Rank Diff Local n/a n/a n/a 0.001 (0.003) n/a 3 Year GD Local n/a n/a n/a 0.142 (0.033) n/a 0.003 (0.003) FIFA Comp Rank Diff FIFA n/a -0.005 (0.002) n/a n/a n/a 0.131 (0.032) 3 Year GD FIFA n/a 0.069 (0.022) n/a n/a n/a 0.040 (0.022)*

Constant -1.668 (0.411) -1.644 (0.401) -1.747 (0.396) -1.757 (0.377) -0.331 (0.852) -1.715 (0.356) Adjusted R Square 0.594 (0.172) 0.613 (0.168) 0.675 (0.159) 0.711 (0.150) 0.709 (0.151) 0.745 (0.141)

96

DV: Television Revenue

Variable Dimension Variable Model 1 Model 2 Model 3 Model 4 Model 5 Model 6 Socio- Demographic Population n/a n/a n/a n/a n/a n/a Income n/a n/a n/a n/a n/a n/a

Performance League Rank -0.001 (0.009) -0.002 (0.008) -0.004 (0.010) 0.002 (0.011) 0.003 (0.009) 0.011 (0.010) UCL (0/1) 0.155 (0.157) 0.191 (0.105)* 0.132 (0.128) 0.149 (0.129) 0.079 (0.120) 0.139 (0.115) EL (0/1) -0.173 (0.116) -0.140 (0.111) -0.318 (0.138) -0.305 (0.138) -0.170 (0.132) -0.128 (0.308) GamesPlayed 0.021 (0.006) 0.018 (0.006) 0.032 (0.007) 0.030 (0.008) 0.022 (0.007) 0.019 (0.007)

Transfer Exp. 0.147 (0.025) 0.156 (0.024) 0.177 (0.028) 0.151 (0.026) 0.155 (0.025) Avg Attendance n/a n/a n/a n/a n/a n/a Geographic Rivalry Local Rival (0/1) n/a n/a -0.359 (0.160) -0.295 (0.171)* -0.071 (0.160) 0.099 (0.163) Dist. Rival (km) n/a n/a -0.185 (0.092) -0.143 (0.100) -0.021 (0.092) 0.150 (0.098)

Determined Rivalry FIFA Rival (0/1) 0.302 (0.066) 0.279 (0.064) n/a n/a 0.390 (0.079) 0.319 (0.076)

Competitive Balance Local Comp Rank Diff Local n/a n/a n/a -0.010 (0.008) n/a -0.009 (0.007) 3 Year GD Local n/a n/a n/a -0.020 (0.088) -0.003 (0.080) FIFA Comp Rank Diff FIFA n/a -0.002 (0.005) n/a n/a -0.005 (0.006) 3 Year GD FIFA n/a -0.199 (0.048) n/a n/a n/a -0.174 (0.054)

Constant 3.341 (0.137) 3.443 (0.134) 3.586 (0.281) 3.535 (0.284) 3.213 (0.273) 3.095 (0.261) Adjusted R Square 0.536 (0.383) 0.574 (0.368) 0.459 (0.413) 0.458 (0.413) 0.527 (0.385) 0.579 (0.364)

97

DV: Commercial Revenue

Variable Dimension Variable Model 1 Model 2 Model 3 Model 4 Model 5 Model 6 Socio-Demographic Population n/a n/a n/a n/a n/a n/a Income n/a n/a n/a n/a n/a n/a

Performance League Rank -0.034 (0.010) -0.034 (0.010) -0.039 (0.011) -0.034 (0.012) -0.036 (0.011) -0.030 (0.012) UCL (0/1) -0.033 (0.126) -0.041 (0.128) -0.003 (0.139) -0.001 (0.138) -0.030 (0.138) -0.005 (0.137) EL (0/1) 0.040 (0.133) 0.034 (0.133) -0.046 (0.150) -0.054 (0.148) 0.030 (0.151) 0.033 (0.149) GamesPlayed 0.007 (0.007) 0.007 (0.007) 0.013 (0.008) 0.015 (0.008)* 0.008 (0.008) 0.009 (0.008)

Operational Strategy Transfer Exp. 0.135 (0.029) 0.134 (0.029) 0.164 (0.030) 0.158 (0.030) 0.148 (0.030) 0.146 (0.030) Stadium Capacity 0.498 (0.120) 0.510 (0.123) 0.829 (0.139) 0.832 (0.138) 0.759 (0.140) 0.696 (0.155) Geographic Rivalry Local Rival (0/1) n/a n/a -0.178 (0.174) -0.278 (0.183) -0.028 (0.184) -0.107 (0.195) Dist. Rival (km) n/a n/a -0.135 (0.101) -0.195 (0.108)* -0.040 (0.107) -0.055 (0.124)

Determined Rivalry FIFA Rival (0/1) 0.290 (0.076) 0.293 (0.076) n/a n/a 0.215 (0.092) 0.195 (0.091)

Competitive Balance Local Comp Rank Diff Local n/a n/a n/a -0.009 (0.008) n/a -0.013 (0.009) 3 Year GD Local n/a n/a n/a 0.240 (0.094) n/a 0.264 (0.096) FIFA Comp Rank Diff FIFA n/a -0.006 (0.006) n/a n/a n/a 0.004 (0.007) 3 Year GD FIFA n/a 0.028 (0.059) n/a n/a n/a -0.108 (0.070)

Constant -1.906 (1.337) -2.060 (1.378) -5.243 (1.533) -5.270 (1.515) -4.701 (1.530) -4.023 (1.672) Adjusted R Square 0.405 (0.439) 0.402 (0.440) 0.426 (0.448) 0.442 (0.441) 0.441 (0.442) 0.459 (0.434)