Tese de Doutoramento 2017

Empirical Analysis of Broadcast

Demand

Demand, Competitive Balance, Demand for Tickets and Revenue Generation in Brazilian Football Market

Doutoramento.

Brazilian Football Market Brazilian

in Thadeu Miranda Gasparetto

Generation Generation

“Mención Internacional”

Thadeu Miranda Tese de Thadeu Gasparetto. Analysis Demand, Competitive Balance, Empirical of Broadcast Revenue for Tickets and

Escola Internacional de Doutoramento

Thadeu Miranda Gasparetto

DOCTORAL DISSERTATION

EMPIRICAL ANALYSIS OF BROADCAST DEMAND, COMPETITIVE BALANCE, DEMAND FOR TICKETS AND REVENUE GENERATION IN BRAZILIAN FOOTBALL MARKET

Supervised by: Ángel Barajas

Year: 2017

“International Mention”

ACKNOWLEDGEMENTS

I wish to express my sincere thanks to my supervisor Dr. Ángel Barajas. I am very grateful for all the effort he has made for my academic development. He was responsible for the several academic opportunities I had in over this period and I thank him very much for that. His valuable advices and feedbacks in our researches, as well as his pursuit for excellence in all activities he does, taught me the true value of this degree.

I would like to show appreciation of all IDLab members. That unexpected opportunity to visit Russia in 2015 has become a place of work with really great people. I am very proud to be a part of this team and thank you all for the constructive suggestions I have been receiving from you.

I am very thankful to the Professor Sean Hamil. He opened the doors of Birkbeck, University of London, for me, where I have lived a remarkable academic and personal experience over three months. Living in London had always been a dream and studying there made possible to improve my English skills and obtain the International Mention.

I am grateful for all members from the PhD Defence Committee. I would like to thank Dr. Antonio López Castedo regarding the academic support he offered me over all PhD program. I would also like to thank Vinicius Paiva and Dr. Ary Rocco for the data set provided, which allowed me the development of the broadcast demand chapter.

I am thankful for Universidad de Vigo for the grant Axudas Predoutorais that I have been receiving since 2016, which contributed with my full dedication in the PhD program.

I would like to thank all my friends and relatives in . It would be very unfair to quote names of some because I could forget others who also contributed in their own way. Regardless of the huge distance between Brazil and , I spent amazing moments with them all the few times I came back home. At the same time, I appreciate all people I met at Facultad de Ciencias Empresariales y Turismo and the new friends I made in Spain and UK. I enjoyed a lot every time we played football or went out for some beers.

Finally, I would like to thank my parents. Without their love and support over the last years this work would not have been possible. Although the Atlantic Ocean was between us, they have always been there for me. They will always be my examples and I must express gratitude to them for all the human values that I have. Thank you for everything.

INDEX

RESUMEN EN ESPAÑOL ...... 17

CAPÍTULO 2: EL MERCADO BRASILEÑO DE FÚTBOL ...... 19 CAPÍTULO 3: EL IMPACTO DE CADA TORNEO EN LA GENERACIÓN DE INGRESOS ...... 19 CAPÍTULO 4: INFLUENCIA DEL FORMATO COMPETITIVO ...... 20 CAPÍTULO 5 : EL BALANCE COMPETITIVO ...... 20 CAPÍTULO 6: DEMANDA DE ENTRADAS EN LA PRIMERA DIVISIÓN ...... 21 CAPÍTULO 7: DIFERENCIA EN LA DEMANDA DE ENTRADAS EN CADA UNA DE LAS DIVISIONES DE BRASILEÑA ...... 22 CAPÍTULO 8: DEMANDA DE ENTRADAS: EFECTO REDISTRIBUTIVO DE INGRESOS EN LOS CAMPEONATOS ESTATALES ...... 23 CAPÍTULO 9: DEMANDA DE FÚTBOL EN TELEVISIÓN ...... 24 CONCLUSIONES Y FUTURAS INVESTIGACIONES ...... 25 1. INTRODUCTION ...... 29

1.1. RESEARCH QUESTIONS AND METHODS ...... 30 1.2. CONTRIBUTIONS ...... 33 2. THE BRAZILIAN FOOTBALL MARKET ...... 37

2.1. INSTITUTIONAL STRUCTURE ...... 37 2.2. SPORTING CALENDAR AND DOMESTIC TOURNAMENTS ...... 38 2.2.1. Brazilian State Championships ...... 40 2.2.2. Regional Championships ...... 41 2.2.3. Brazilian Cup ...... 42 2.2.4. Brazilian League ...... 43 2.3. ECONOMIC CONTEXT: LABOUR MARKET AND FINANCIAL ASPECTS ...... 44 2.3.1. Labour Market ...... 44 2.3.2. Financial Aspects ...... 46 3. THE IMPACT OF DIFFERENT TOURNAMENTS ON THE REVENUE GENERATION ...... 51

3.1. PROBLEM STATEMENT ...... 51 3.2. METHODS ...... 52 3.3. RESULTS AND DISCUSSION ...... 55 3.4. FINAL REMARKS ...... 57 4. THE INFLUENCE OF COMPETITIVE DESIGN ...... 59

4.1. PROBLEM STATEMENT ...... 59 4.2. THEORETICAL BACKGROUND ...... 59 4.3. METHODS ...... 62 4.4. RESULTS ...... 64 4.4.1. Competitive Balance ...... 64

4.4.2. Fans’ Interest ...... 68 4.4.3. Considerations about Brazilian Football ...... 70 4.5. FINAL REMARKS ...... 72 5. COMPETITIVE BALANCE ...... 73

5.1. PROBLEM STATEMENT ...... 73 5.2. THEORETICAL BACKGROUND ...... 74 5.3. METHODS ...... 76 5.3.1. The Model ...... 76 5.3.2. Sample ...... 79 5.3.3. Statistical Analysis ...... 79 5.3.4. The Dispersion of Positions ...... 80 5.3.5. Confrontation with Traditional Models ...... 80 5.4. RESULTS AND DISCUSSION ...... 81 5.4.1. Index Validation ...... 90 5.4.2. Competitive Balance and Average Attendance ...... 91 5.4.3. Limitations and Further Research ...... 92 5.5. FINAL REMARKS ...... 93 6. THE DEMAND FOR TICKETS IN THE BRAZILIAN LEAGUE’S FIRST DIVISION ...... 95

6.1. PROBLEM STATEMENT ...... 95 6.2. FOOTBALL STADIUMS IN BRAZIL ...... 97 6.3. THEORETICAL BACKGROUND ...... 100 6.4. METHODS ...... 103 6.4.1. Data ...... 103 6.4.2. Models and Variables ...... 104 6.4.2.1. Match Characteristics ...... 105 6.4.2.2. Uncertainty of Outcome ...... 106 6.4.2.3. Stadium Characteristics ...... 107 6.5. RESULTS AND DISCUSSION ...... 108 6.5.1. Match Characteristics ...... 108 6.5.2. Uncertainty of Outcome...... 112 6.5.3. Stadiums Characteristics ...... 112 6.5.4. Limitations and Further Research ...... 114 6.5.5. Practical Implications ...... 115 6.6. FINAL REMARKS ...... 116 7. DIFFERENCES IN THE DEMAND FOR TICKETS IN EACH DIVISION OF THE BRAZILIAN LEAGUE ...... 119

7.1. PROBLEM STATEMENT ...... 119 7.2. THEORETICAL BACKGROUND ...... 120 7.3. METHODS ...... 121 7.3.1. Model and Variables ...... 121

7.4. RESULTS AND DISCUSSION ...... 124 7.5. FINAL REMARKS ...... 128 8. DEMAND FOR TICKETS: DISTRIBUTION OF WEALTH IN THE BRAZILIAN STATE CHAMPIONSHIPS ...... 129

8.1. PROBLEM STATEMENT ...... 129 8.2. THEORETICAL BACKGROUND ...... 131 8.3. METHODS ...... 133 8.3.1. Method and Hypotheses ...... 133 8.3.2. Sample ...... 133 8.3.2.1. Campeonato ...... 134 8.3.2.2. ...... 134 8.3.2.3. ...... 136 8.3.3. The Model and Variables ...... 136 8.4. DESCRIPTIVE ANALYSIS OF MINEIRO, CARIOCA AND PAULISTA STATE CHAMPIONSHIPS ...... 137 8.5. RESULTS AND DISCUSSION ...... 141 8.6. FINAL REMARKS ...... 144 9. BROADCAST DEMAND ...... 147

9.1. PROBLEM STATEMENT ...... 147 9.2. BROADCAST ...... 148 9.3. THEORETICAL BACKGROUND ...... 151 9.4. METHODS ...... 154 9.4.1. Data ...... 154 9.4.2. Model and Variables ...... 155 9.4.3. Model Selection ...... 161 9.5. RESULTS AND DISCUSSION ...... 161 9.5.1. Similarities ...... 163 9.5.2. Differences ...... 164 9.5.3. Limitations and Further Research ...... 167 9.6. FINAL REMARKS ...... 168 10. CONCLUSIONS ...... 171 REFERENCES ...... 179

INDEX OF TABLES

TABLE 1. FEATURES OF THE TIERS IN THE BRAZILIAN LEAGUE ...... 44 TABLE 2. SALARIES IN THE BRAZILIAN FOOTBALL MARKET (2015) ...... 45 TABLE 3. SPORTING PERFORMANCE DIAGRAM ...... 53 TABLE 4. VIF OUTPUTS ...... 54 TABLE 5. INDIVIDUAL EFFECTS FROM DIFFERENT TOURNAMENTS ON REVENUE GENERATION: BRAZILIAN FOOTBALL MARKET 2010-2014 ...... 55 TABLE 6. SEASONAL HICB, C4ICB AND AVERAGE ATTENDANCE DURING THE CAMPEONATO BRASILEIRO 1991-2014 SEASONS ...... 64 TABLE 7. MANN-WHITNEY U TEST FOR THE HICB: PLAY-OFFS VS. ROUND-ROBIN ...... 67 TABLE 8. MANN-WHITNEY U TEST FOR THE C4ICB: PLAY-OFFS VS. ROUND-ROBIN ...... 67 TABLE 9. MANN-WHITNEY U TEST FOR AVERAGE ATTENDANCE: PLAY-OFFS VS. ROUND- ROBIN ...... 70 TABLE 10. MAXIMUM IMBALANCE IN A DOUBLE ROUND-ROBIN LEAGUE WITH 20 PARTICIPANTS ...... 78 TABLE 11. APD OF THE FOOTBALL LEAGUES IN GERMANY, BRAZIL, SPAIN, FRANCE, NETHERLANDS, , ITALY, AND RUSSIA ...... 81 TABLE 12. SHAPIRO-WILK TEST...... 84 TABLE 13. LEVENE TEST ...... 84 TABLE 14. OUTPUT FROM THE ONE-WAY ANOVA ...... 84 TABLE 15. TUKEY POST HOC TEST ...... 85 TABLE 16. OF TITLES AND AVERAGE POSITION BETWEEN 2006/07 AND 2013/14 ...... 88 TABLE 17. HICB, C4ICB AND THE APD IN THE BRAZILIAN LEAGUE FROM THE 2006 TO THE 2013 SEASON ...... 90 TABLE 18. SUMMARY STATISTICS ...... 105 TABLE 19. 3SLS REGRESSIONS REGARDING ATTENDANCE AND TICKET PRICES (PART 1) ...... 109 TABLE 20. 3SLS REGRESSIONS REGARDING ATTENDANCE AND TICKET PRICES (PART 2) ...... 110 TABLE 21. DEMAND FOR TICKETS: ALL BRAZILIAN LEAGUE TIERS (2012, 2013 AND 2014) ...... 125 TABLE 22. TOP 10 BRAZILIAN CITIES BY POPULATION AND CLUBS IN THE MAIN BRAZILIAN CITIES (2017) ...... 127 TABLE 23. BRAND-TEAMS THAT HAVE QUALIFIED FOR PLAY-OFFS ...... 130 TABLE 24. SUMMARY STATISTICS: STATE CHAMPIONSHIP (MINEIRO) . 138 TABLE 25. SUMMARY STATISTICS: STATE CHAMPIONSHIP (CARIOCA) 138 TABLE 26. SUMMARY STATISTICS: STATE CHAMPIONSHIP (PAULISTA) ..... 139 TABLE 27. NET INCOME (PROFIT AND LOSS) BY CHAMPIONSHIP AND YEAR: NUMBER OF MATCHES ...... 139 TABLE 28. NET INCOME (PROFIT AND LOSS): NORMAL MATCHES VS. MATCHES AGAINST BRAND TEAMS ...... 140

TABLE 29. NET LOSS/INCOME BY PLAY-OFF ...... 140 TABLE 30. AVERAGE VALUES: ATTENDANCE, MATCH DAY REVENUES AND NET INCOME ...... 140 TABLE 31. EMPIRICAL ESTIMATIONS (3SLS): ATTENDANCE AND MATCH DAY REVENUES ...... 142 TABLE 32. TOTAL NET INCOME BY CHAMPIONSHIP: NORMAL MATCHES VS. MATCHES AGAINST BRAND TEAMS ...... 144 TABLE 33. VARIABLES DESCRIPTION ...... 156 TABLE 34. SUMMARY STATISTICS ...... 157 TABLE 35. DETERMINANTS OF TELEVISION AUDIENCES IN RIO DE JANEIRO AND SÃO PAULO ...... 162 TABLE 36. ALTERNATIVE REGRESSION ANALYSING UNCERTAINTY OF OUTCOME IN SÃO PAULO ...... 166

INDEX OF FIGURES

FIGURE 1. 2017 BRAZILIAN COMPETITION SCHEDULE ...... 39 FIGURE 2. TOTAL REVENUE IN BRAZILIAN FOOTBALL: TOP 27 CLUBS ...... 47 FIGURE 3. TOTAL REVENUE BY SOURCE: TOP 27 CLUBS (R$ MILLIONS) ...... 47 FIGURE 4. NET REVENUES, OPERATIONAL COSTS AND EBITDA: BRAZILIAN CLUBS ...... 48 FIGURE 5. TOTAL DEBT: BRAZILIAN CLUBS ...... 49 FIGURE 6. HICB FOR THE CAMPEONATO BRASILEIRO ...... 65 FIGURE 7. C4ICB FOR THE CAMPEONATO BRASILEIRO...... 65 FIGURE 8. DIFFERENCES IN THE HICB IN THE CAMPEONATO BRASILEIRO: PLAY-OFFS VS. LEAGUE ...... 66 FIGURE 9. DIFFERENCES IN THE C4ICB IN THE CAMPEONATO BRASILEIRO: PLAY-OFFS VS. LEAGUE ...... 66 FIGURE 10. EVOLUTION OF AVERAGE ATTENDANCE IN THE CAMPEONATO BRASILEIRO: 1991-2014 ...... 69 FIGURE 11. COMPARISON OF AVERAGE ATTENDANCE IN THE CAMPEONATO BRASILEIRO: PLAY-OFFS VS. LEAGUE ...... 69 FIGURE 12. SEASONAL AVERAGE ATTENDANCE IN THE CAMPEONATO BRASILEIRO: 1967 TO 2016 ...... 73 FIGURE 13. BOX PLOT OF THE APD IN THE DOMESTIC LEAGUES IN GERMANY, BRAZIL, SPAIN, FRANCE, NETHERLANDS, ENGLAND, ITALY, PORTUGAL AND RUSSIA ...... 82 FIGURE 14. LINE CHART OF THE APD IN THE BRAZILIAN, GERMAN AND DUTCH LEAGUES ...... 83 FIGURE 15. DISPERSION OF THE AVERAGE POSITION OF THE CLUBS FROM 2006/07 TO 2013/14 ...... 87 FIGURE 16. NUMBER OF BRAZILIAN STADIUMS BY TOTAL CAPACITY ...... 98 FIGURE 17. BRAZILIAN STADIUMS BY QUALITY ...... 99 FIGURE 18. NUMBER OF MATCHES PLAYED AT NON-USUAL STADIUMS BY QUALITY ..... 100 FIGURE 19. FIRST, SECOND, THIRD AND FOURTH DIVISIONS OF THE BRAZILIAN LEAGUE: SEASONAL AVERAGE ATTENDANCE 2013-2015...... 119 FIGURE 20. MATCH DAY REVENUES (R$): ALL BRAZILIAN LEAGUE TIERS (2012, 2013 AND 2014) ...... 123

Empirical analysis of broadcast demand, competitive balance, demand for tickets and revenue generation in Brazilian football market

RESUMEN EN ESPAÑOL

La presente investigación analiza el mercado brasileño de fútbol profesional. A pesar de que Brasil es uno de los países más exitosos en fútbol a nivel internacional, su mercado interno no está desarrollado al mismo nivel. El Campeonato Brasileiro – la Liga

Brasileña – no figura entre las más destacadas del mundo. Los demás torneos brasileños, como la Copa de Brasil y los Campeonatos Estatales, tampoco son conocidos a nivel internacional. Al mismo tiempo, en las últimas décadas el fútbol brasileño se ha convertido en una fábrica de jugadores, produciendo mano de obra para los mejores clubes Europeos. En ese sentido, el presente estudio tiene por objeto analizar el mercado brasileño de fútbol desde una perspectiva amplia, buscando la explicación de su situación y plantear medidas que favorezcan su desarrollo.

Esta tesis está basada en siete investigaciones elaboradas en los últimos tres años.

Todos estos estudios han sido presentados en los congresos más relevantes sobre la gestión y la economía del deporte como la European Conference on Sports Economics en

2015 y 2016, la European Association for Sport Management Conference en 2014 y 2015, la Gijón Conference on Sports Economics en 2015, 2016 y 2017, la IASE International

Conference in Sports Economics en 2014, el Congreso Iberoamericano de Economía del

Deporte en 2014, 2016 y 2017, así como el Congresso Brasileiro de Gestão do Esporte en 2015. Algunos estudios también han sido presentados en conferencias de carácter más general dentro del área del economía y gestión como la April International Academic

Conference on Economic and Social Development en 2016 y 2017 y la International

Conference in Applied Research in Economics en 2015. El hecho de haber presentado esos trabajos que componen la tesis ha facilitado que se recibieran críticas y comentarios de algunos de los mayores expertos en la materia.

17

Resumen en Español

Además, algunas de las investigaciones ya han sido publicadas en revistas científicas. La primera fue el artículo nombrado “ or Just League: A Debate in

Brazilian Football”, que se publicó en 2015 en la revista The Open Sports Sciences

Journal, indexada en Scopus. El segundo estudio, titulado “Reanalizando la

Competitividad en la Industria del Fútbol”, fue publicado en 2016 en la Revista de

Administração de Empresas, indexada en Scopus y Web of Science. El tercer artículo,

“Influencia de los torneos en la generación de ingresos”, fue publicado en 2016 en la

Revista de Psicología del Deporte, también indexada en Scopus y Web of Science. Dos artículos más, “Brand-teams and distribution of wealth in Brazilian State

Championships” y “Fan preferences: One country, two markets and different behaviours”, están en proceso de revisión en revistas especializadas – la segunda investigación ha pasado por la primera ronda de evaluación y se está llevando a cabo las sugerencias de los revisores. Todas las demás investigaciones se enviarán para pasar el proceso de revisión para que sean publicadas.

Esta tesis está estructurada en diez capítulos: (1) Introducción; (2) El Mercado

Brasileño de Fútbol; (3) El impacto de cada torneo en la generación de ingresos; (4)

Influencia del formato competitivo; (5) El Balance Competitivo; (6) Demanda de entradas en la Primera División; (7) Diferencia en la demanda de entradas en cada una de las divisiones de la Liga Brasileña; (8) Demanda de entradas: efecto redistributivo de ingresos en los Campeonatos Estatales; (9) Demanda de fútbol en televisión; y (10)

Conclusiones. Una vez que cada capítulo ha analizado un tema específico, cada uno presenta su correspondiente base teórica, metodología y resultados. Las preguntas de investigación, los métodos empleados y los principales hallazgos de los Capítulos 2 a 9 se presentan a continuación.

18

Empirical analysis of broadcast demand, competitive balance, demand for tickets and revenue generation in Brazilian football market

Capítulo 2: El Mercado Brasileño de Fútbol

Este capítulo tiene como principal objetivo presentar una contextualización del fútbol brasileño que sirva de soporte al resto de la tesis. Se plantea dos preguntas de investigación: ¿Cómo se organiza el mercado Brasileño de fútbol? y ¿Cómo está la situación económico-financiera del mercado Brasileño de fútbol? Se ha empleado una metodología descriptiva, explorando las características peculiares del fútbol brasileño. Se explican las principales características como su organización a través de la Confederación

Brasileña de Fútbol a nivel nacional que, a su vez, integra veintisiete Federaciones

Estatales. Cabe señalar la existencia de diferentes torneos. Algunos no tienen parangón en otros países como es el caso de los Campeonatos Estatales y Regionales. El calendario deportivo empieza en enero y termina en diciembre, siendo así distinto del fútbol

Europeo. Se observa que los clubes orientan su gestión hacia la maximización de victorias

– aunque los ingresos hayan aumentado en los últimos años, los costes también han incrementado en similar o incluso mayor proporción. Esto además ha sido el origen de las grandes deudas que los clubes presentan a día de hoy.

Capítulo 3: El impacto de cada torneo en la generación de ingresos

Puesto que en el fútbol brasileño los equipos participan en muchos y distintos torneos, la pregunta de investigación de este capítulo es: ¿Cómo afecta cada torneo a la generación de ingresos de los clubes de fútbol? Para responder esta cuestión se ha empleado una regresión lineal en datos de panel con efectos fijos de los clubes. Se ha evidenciado que solo tres torneos afectan positivamente la generación de ingresos de los clubes brasileños: la Liga, la Copa y la Copa de América (el más importante torneo de clubes en Sudamérica). Sin embargo, se observó que el formato competitivo juega un papel importante. Los buenos resultados en torneos eliminatorios – ambas Copa del Brasil y – se traducen en ingresos en el actual año fiscal. Por otro

19

Resumen en Español lado, una buena actuación en la Liga repercute en la generación de ingresos de la temporada siguiente. Esto se debe a que ese buen desempeño en Liga habilita para participar en torneos internacionales, proporciona mayores ingresos televisivos y quizás facilite la venta de jugadores a Europa por precios más altos.

Capítulo 4: Influencia del formato competitivo

Este capítulo tiene como pregunta de investigación si el cambio de formato competitivo ha sido ventajoso. En la temporada 2003, el Campeonato Brasileño adoptó como formato competitivo el doble Round-Robin, siguiendo el modelo de las principales ligas Europeas. Así, ese capítulo ha buscado analizar los efectos proporcionados por ese cambio. En la primera etapa del estudio se emplea un análisis descriptivo, comparando la competitividad y el interés de los aficionados durante veinticuatro temporadas: doce disputadas en sistema de play-offs (1991-2002) y doce con formato liga (2003-2014). Para analizar la competitividad se han empleado dos indicadores – Herfindahl Index of

Competitive Balance (HICB) y C4 Index of Competitive Balance (C4IC) –. El interés de los aficionados se ha medido por la asistencia media de la liga en cada temporada. En la segunda etapa dichos valores han sido comparados estadísticamente a través de la prueba de Mann-Whitney U para muestras independientes. Como resultados, se ha evidenciado que el cambio en el formato competitivo proporcionó una mayor competitividad al

Campeonato Brasileiro. Sin embargo, no se han encontrado cambios significativos estadísticamente en el interés de los aficionados – aunque se haya observado una pequeña tendencia de mayores asistencias medias.

Capítulo 5: El Balance Competitivo

Hay un sentir común en Brasil de que su liga nacional es la más equilibrada del mundo, una vez que a priori nunca se sabe quién la ganará. Así, la pregunta de investigación de este capítulo es si el Campeonato Brasileiro es más equilibrado que otras

20

Empirical analysis of broadcast demand, competitive balance, demand for tickets and revenue generation in Brazilian football market ligas de fútbol. Para contestar esta pregunta, se han analizado nueve ligas de fútbol:

Bundesliga (Alemania), Campeonato Brasileiro (Brasil), La Liga BBVA (España), Ligue

1 (Francia), Eredivise (Holanda), (Inglaterra), (Italia), Primeira

Liga (Portugal) y Premier League (Rusia). Se ha elaborado por primera vez un indicador denominado Accumulated Points Difference (APD) basado en la diferencia acumulada de puntos entre los participantes y ajustado por el máximo desequilibrio posible. Esto es necesario dado que diversos torneos poseen diferente números de participantes. Además del análisis descriptivo, se ha llevado a cabo un análisis de varianza ANOVA one way con post hoc de Tukey para compararlas estadísticamente. Además, se ha confrontado el nuevo indicador APD con otros dos modelos tradicionales – Herfindahl Index of

Competitive Balance (HICB) y C4 Index of Competitive Balance (C4ICB). Los resultados han evidenciado que la Liga Brasileña resulta la más equilibrada entre la muestra investigada. A su vez, no se ha apreciado diferencias estadísticas entre las ligas Europeas.

Capítulo 6: Demanda de entradas en la Primera División

Este capítulo se centra en la demanda de entradas en el fútbol brasileño. Aunque la demanda sea uno de los temas más investigados dentro de la Economía del Deporte, hasta el presente momento solo hay un artículo publicado teniendo el fútbol brasileño como objeto de investigación. El objeto general de ese capítulo ha sido reanalizar la demanda de entradas en la Primera División de la Liga Brasileña. Cabe destacar que en los últimos años los equipos brasileños han jugado muchos partidos en estadio no- habituales, tanto en Arenas construidas para el Mundial de 2014 como también en estadios de baja calidad. Así, se supone que eso podría generar diferentes efectos en la asistencia a los partidos de los clubes. En ese sentido, este capítulo plantea dos preguntas de investigación: ¿Cómo afecta jugar en un estadio no-habitual a la demanda de entradas? y ¿influye la calidad de los estadios en los precios de las entradas y en su demanda? El

21

Resumen en Español método usado ha sido un sistema de ecuaciones en datos de panel con estimador en

Mínimos Cuadrados en Tres Etapas (MC3E). Se han empleado los efectos fijos de los clubes locales. Siete modelos han sido llevados a cabo buscando evaluar los diferentes efectos de las distintas características de los estadios.

Los resultados han confirmado el efecto positivo de la calidad del partido y de los clásicos para aumentar la demanda de entradas. Sin embargo, se ha evidenciado una relación cuadrática entre probabilidad de victoria y asistencia en forma de “U”, rechazando así la teoría clásica de la Incertidumbre del Resultado. No obstante, los principales hallazgos del capítulo están relacionados con las características de los estadios. Se ha evidenciado que los estadios no-habituales disminuyen la asistencia, pero los clubes suelen cobrar entradas más caras. Por otro lado, si los equipos deciden jugar en buenos estadios – como los construidos o reformados para el Mundial 2014 y los categorizados como 5 estrellas – se compensa el efecto negativo de jugar en un estadio no-habitual. Además, los niveles de seguridad, comodidad e higiene son factores importantes para incrementar la demanda de entradas – aunque estos elementos supongan un encarecimiento de las entradas.

Capítulo 7: Diferencia en la demanda de entradas en cada una de las divisiones de la

Liga Brasileña

Pocos estudios han investigado la demanda de entradas en divisiones inferiores del fútbol profesional. Los dos que se propusieron hacerlo han evidenciado diferencias y similitudes con respeto a la máxima categoría del fútbol. Por eso, el presente capítulo busca analizar por primera vez los factores que influyen en la demanda de entradas en las cuatro divisiones del Campeonato Brasileiro. Así, la pregunta de investigación ha sido:

¿es igual la demanda de entradas para todas las divisiones del fútbol brasileño? Seis regresiones lineales en datos de panel han sido llevadas a cabo, controlándolas por

22

Empirical analysis of broadcast demand, competitive balance, demand for tickets and revenue generation in Brazilian football market temporada, región del país y división. Los resultados han demostrado que el éxito histórico y el desempeño actual son los factores más importantes para atraer el interés de los aficionados. A su vez, se ha verificado que factores socio-económicos juegan un papel importante en las divisiones inferiores.

Capítulo 8: Demanda de entradas: efecto redistributivo de ingresos en los Campeonatos

Estatales

Este capítulo es pionero en el estudio de los torneos más peculiares del fútbol brasileño: los Campeonatos Estatales. Dichos torneos son jugados solo por clubes de mismo Estado, aunque tengan diferentes niveles competitivos. De hecho, participan equipos de Primera, Segunda, Tercera, Cuarta e incluso equipos sin división nacional.

Además, una característica particular de estos campeonatos es que siempre se deciden en partidos eliminatorios. Así, las preguntas de investigación han sido: ¿Cuáles son los determinantes de la demanda de entradas en los Campeonatos Estatales? y ¿Cómo influyen los partidos eliminatorios y los Brand-teams en la demanda de entradas y en los ingresos de taquillas?

La primera etapa del estudio se ha efectuado un análisis descriptivo buscando explicar las peculiaridades de estos torneos. Se ha definido el concepto de Brand-teams que no ha sido utilizado previamente en la literatura. A continuación, se ha llevado a cabo un sistema de ecuaciones con el estimador en Mínimos Cuadrados en Tres Etapas (MC3E) para contrastar las hipótesis planteadas. Los resultados han demostrado mayor demanda de entradas durante los fines de semana y que los Brand-teams y los partidos eliminatorios incrementan de forma significativa estadísticamente la asistencia. Además, se ha evidenciado un efecto redistributivo de los ingresos en dichos campeonatos. La presencia de los Brand-teams es ventajoso para los clubes pequeños, puesto que los partidos contra

23

Resumen en Español estos grandes equipos generan enormes ingresos de taquillas, compensando así las pérdidas que esos clubes pequeños suelen tener en el resto de los partidos.

Capítulo 9: Demanda de fútbol en televisión

La demanda televisiva de los deportes es un tema de interés actualmente en la

Economía del Deporte. Sin embargo, lo que se presenta en este capítulo es el primer estudio que investiga este asunto en el fútbol brasileño. Además que se plantea si diferentes mercados tienen diferentes comportamientos. En ese sentido, el objetivo general del capítulo ha sido descubrir los determinantes de la demanda televisiva del fútbol brasileño. Además, los partidos de fútbol en Brasil son regionalizados– la cadena de cada estado decide los partidos que va a transmitir. Así, se supone que podrán existir diferencias entre los aficionados de cada región. Por lo tanto, la pregunta de investigación de este capítulo se formula de la siguiente manera: ¿Son distintos los determinantes de la demanda televisiva en diferentes Estados?

El método econométrico empleado consiste en regresiones lineales en datos de panel con efectos fijos de los clubes televisados en cada estado. Este estimador ha considerado los siguientes equipos: Corinthians, Palmeiras, São Paulo y Santos en el estado de São Paulo; y Botafogo, Flamengo, Fluminense y Vasco en el estado de Rio de

Janeiro. Tres modelos han sido elaborados buscando encontrar el respectivo comportamiento de los aficionados con respecto a los siguientes factores: incertidumbre del resultado, preferencia por la victoria y aversión a la pérdida. Los resultados han corroborado los recientes estudios en otros mercados. Se ha confirmado que la calidad de los partidos y los clásicos incrementan la audiencia en televisión. Por otro lado, la demanda televisiva de fútbol es mayor en días laborales, al revés de lo que se observó en

Europa. El principal hallazgo de la investigación ha sido las diferencias observadas entre los mercados. Los aficionados en Rio de Janeiro presentan aversión a la pérdida, mientras

24

Empirical analysis of broadcast demand, competitive balance, demand for tickets and revenue generation in Brazilian football market en São Paulo les atrae partidos con reducida incertidumbre – un análisis adicional ha confirmado que se aumenta la demanda en partidos contra equipos más débiles o muy superiores al local.

Conclusiones y Futuras Investigaciones

El mercado brasileño de fútbol ha sido investigado en detalle en la presente tesis doctoral. En líneas generales, aspectos como los efectos del cambio de formato competitivo y la demanda de entradas y de fútbol en televisión han sido analizados. De esta manera, se presentan contribuciones teóricas, metodológicas y empíricas. Los hallazgos evidencian que el Campeonato Brasileiro es uno de los más equilibrados del mundo. Sin embargo, esto no se traduce en elevadas asistencias medias a los estadios para ver a los equipos.

Los estudios de demanda manifiestan que el factor más importante para atraer los aficionados es la calidad del partido. También se ha observado que la calidad de los estadios juega un papel importante en la demanda. En ese sentido, la Confederação

Brasileira de Futebol podría perfeccionar la recién elaborada Licencia de Clubes solicitando a los equipos mejores infraestructuras deportivas, puesto que eso incrementaría la asistencia a los estadios. Al mismo tiempo, estudios llevados a cabo en el fútbol europeo y en los deportes estadounidenses evidencian que jugadores estrellas incrementan la asistencia a los estadios. Por lo tanto, nuevas investigaciones podrían analizar dichos efectos en el fútbol brasileño. Si se confirma que la presencia del talento tiene impacto en el aumento de la demanda de ingresos, la Confederação Brasileira de

Futebol y sus clubes necesitarían desarrollar métodos para retener los mejores jugadores en el mercado interno.

25

Resumen en Español

Las diversas partes de la tesis vienen a poner de manifiesto que el fútbol brasileño resulta un interesante objeto de investigación. Las peculiaridades del mercado interno brasileño posibilitan los más variados estudios dentro de la economía del deporte y los futuros hallazgos podrían contribuir al desarrollo de la literatura. Se estimulan nuevas investigaciones que analicen empíricamente la situación económico-financiera de los clubes brasileños. Actualmente los equipos dependen en gran medida de los ingresos de televisión y de los traspasos de sus mejores jugadores. Ambas fuentes de recursos forman un significativo elemento en la recaudación total. Por lo tanto, fuentes alternativas de financiación podrían ayudarles a mantener grandes talentos en el mercado local, así como la disminución de sus deudas.

Se ha demostrado que el Campeonato Brasileiro es una de las ligas más equilibradas del mundo. Comprender los factores que contribuyen a ello podría ser un tema de interés para futuras investigaciones. Elementos como el número de jugadores y entrenadores locales, cómo funcionan las canteras en Brasil, los aspectos económicos de los clubes, los traspasos y fichajes, los diferentes torneos y los aspectos culturales, entre muchos otros, podrían explicar el elevado nivel de competitividad encontrada. Así, esos hallazgos podrían ser útiles para las ligas del mundo más importantes, puesto que la competitividad en dichas ligas ha disminuido notablemente en las últimas décadas tal y como se ha probado en la tesis.

Los diferentes torneos – Regionales y Estatales – también se presentan como atractivos objetos de investigación. La distribución regional de los clubes, los equipos de diferentes categorías jugando entre ellos todas las temporadas y los efectos en todo el mercado laboral podrían generar importantes consideraciones sobre el mercado laboral del fútbol brasileño. Estos torneos posibilitan a un inmenso número de jugadores profesionales la oportunidad de jugar contra los más importantes clubes brasileños al

26

Empirical analysis of broadcast demand, competitive balance, demand for tickets and revenue generation in Brazilian football market menos una vez a la temporada. De hecho, muchos jugadores que no han logrado estar en las canteras de los grandes clubes brasileños utilizan los Campeonatos Estaduais como la ocasión de enfrentarles y posiblemente ser fichados por dichos clubes. Aunque resulte raro suponer que un club de Primera División pueda interesarse por un jugador de un equipo de Cuarta División, esto ocurre frecuentemente en el fútbol brasileño – algunos incluso logran de alcanzar la Selección Nacional. Esa mayor movilidad de jugadores entre las categorías podría ayudar a explicar la alta competitividad en el Campeonato

Brasileiro, como también el éxito de Brasil en el fútbol internacional.

Pocas investigaciones han evaluado la y la Copa Libertadores de

América a pesar de la relevancia de ambas en el mercado brasileño. Las demandas de entradas y televisión, el balance competitivo de dichos torneos, la distribución geográfica en estos campeonatos – por estados y países, respectivamente – son interesantes temas a ser investigados en el futuro. La Copa do Brasil ha tenido diversos cambios en el formato competitivo desde su primera edición. Así, buscar un formato óptimo puede resultar en una importante contribución al fútbol local. Por otro lado, la Confederación

Sudamericana de Fútbol (Conmebol) ha aumentado el número de participantes en su principal torneo, la Copa Libertadores de América. Con eso, el Campeonato Brasileiro que antes aseguraba plazas a los cuatro mejores clasificados, ha aumentado ese número para seis en la temporada 2016. Esto seguramente reflejará en cambios del comportamiento de los clubes y en la competitividad de la Liga. En esto sentido, se estimulan nuevas investigaciones en las próximas temporadas.

El fútbol brasileño ha sido examinado profundamente en esa tesis doctoral.

Algunos de los hallazgos de los estudios aquí presentados contribuyen a la literatura y aportan consideraciones que pueden ser útiles para a la Confederação Brasileira de

Futebol, las cadenas de televisión y los clubes de fútbol en Brasil a la hora de tomar 27

Resumen en Español decisiones para intentar perfeccionar el mercado interno. Sin embargo, se considera esencial el desarrollo de nuevas investigaciones para que el fútbol brasileño logre la evolución deseada por todos sus aficionados en Brasil.

28

Empirical analysis of broadcast demand, competitive balance, demand for tickets and revenue generation in Brazilian football market

1. INTRODUCTION

Brazil could be considered to be the most successful country at international football. The Canarinha, the national team, has been the most successful team in the

Fédération Internationale de Football Association (FIFA) World Cup, having won the trophy on five occasions: 1958, 1962, 1970, 1994 and 2002. In addition, the team has won the FIFA Confederations Cup four times, the ’ Cup eight times and a Gold

Medal at the Summer Olympic Games.

Several well-known players were responsible for achieving these titles. The most famous among them is Pelé, the FIFA Player of the 20th century. However, players such as Zagallo, Didi, Nilton Santos, , Bellini, , , ,

Gérson, Tostão, , Cafú, , , Ronaldo and

Ronaldinho were equally crucial in securing these triumphs and helping to make Brazil

“The Country of Football”, as remarked by Kittleson (2014).

Nevertheless, this cognomen may only be partially true. While the Brazilian national team is highly recognizable around the world, its domestic football market has not enjoyed the same level of development. The Brazilian League is not usually considered as one of the top professional football leagues in the world, mostly because the best Brazilian players are playing for European clubs. The higher salaries paid by

European teams, as well as their opportunity to face other great world players, might explain the numerous outgoing transfers. Moreover, Brazilian is the most transferred nationality in the football market, while the Brazilian football market is the leader in both incoming and outgoing transfers (FIFA TMS, 2017).

Nowadays, the Brazilian football market is facing some economic and financial problems. Despite the growth in total revenues over recent years, the clubs’ expenses have

29

Introduction been increasing at similar proportion. Additionally, the top clubs have been burdened by higher amounts of debt. These factors, among many others, may explain the limited interest among fans in attending matches. In other words, they contribute to creating a vicious circle, where each aspect has been negatively influencing each other and hence led to a decline in the Brazilian football market.

While the present work carries out an economic analysis of the Brazilian football market, each chapter focus on a different issue within the scope of the sports economics literature. In this sense, each one has its particular aim as well as a specific theoretical background, methodology and results. The dissertation is structured around 10 chapters as follows: (1) Introduction, (2) The Brazilian Football Market, (3) The Impact of

Different Tournaments on Revenue Generation, (4) The Influence of Competitive Design,

(5) Competitive Balance, (6) The Demand for Tickets in the Brazilian League’s First

Division, (7) Differences in the Demand for Tickets in Each Division of the Brazilian

League, (8) The Demand for Tickets: The Distribution of Wealth in the Brazilian State

Championships, (9) Broadcast Demand in Brazilian Football, and (10) Conclusions.

1.1. Research Questions and Methods

Chapter 2 introduces two research questions: (1) How is the Brazilian football market structured? (2) What is the current economic situation faced by the Brazilian football market? Through a descriptive analysis, the domestic background is explained, with reference to certain features such as institutional structure, the sporting calendar, domestic tournaments, and economic and financial issues.

Once it has been shown that the Brazilian clubs play different tournaments every season, Chapter 3 aims to answer the following research question: How does each tournament affect revenue generation among Brazilian professional football clubs? The

30

Empirical analysis of broadcast demand, competitive balance, demand for tickets and revenue generation in Brazilian football market econometric approach consists of panel data linear regression involving fixed effects with regard to the clubs.

The Brazilian League changed its competition design in 2003. Against this backdrop, Chapter 4 considers the following research question: Did the change from play- offs to a round-robin design enhance both the competitive balance (CB) and fans’ interest? An empirical analysis is used to answer this question. The first step uses a descriptive approach for comparing competitiveness, using the Herfindahl Index of

Competitive Balance (HICB) and the C4 Index of Competitive Balance (C4ICB), and fans’ interest (measured by seasonal average attendance) over a period of 24 years: 12 consecutive play-off seasons followed by 12 editions of the round-robin competition design. Afterwards, the Mann-Whitney U test for independent samples is employed in order to evaluate whether there are statistical differences between the periods.

The fifth chapter is also interested in the level of competitiveness within the

Campeonato Brasileiro (Brazilian League). In Brazil, there is a common understanding that league is one of the most balanced tournaments in the world. As such, the following research question is posed: Is the Brazilian League statistically more balanced than the eight most important European Leagues? A new index, known as the Accumulated Points

Difference (APD), is elaborated in this chapter. Firstly, the outputs are presented and descriptively discussed. Afterwards, one-way ANOVA with the Tukey post hoc test, is employed in order to compare the Brazilian League with other domestic tournaments.

Furthermore, the new index is compared with two traditional indices (HICB and C4ICB) to certify its quality.

The demand for tickets is the topic of Chapter 6. Although it is one of the most researched issues with the sports economics literature, only one paper has been published,

31

Introduction which focuses on Brazilian football. Thus, the general aim of this chapter is to reanalyse the determinants of attendance in the First Division of the Brazilian League.

Notwithstanding, Brazilian football involves a peculiarity: clubs play some matches in non-usual stadiums. Therefore, two research questions are posed: 1) How does a non- usual stadium affect the demand for tickets? 2) Does the quality of the stadium have any impact on attendance as well as ticket prices? An econometric analysis is performed using a panel data Three-Stage Least Squares (3SLS) estimator with fixed effects relating to the home club. Seven models are created in order to evaluate the impact of different stadiums’ characteristics.

The demand for tickets in all tiers of the Brazilian League is the topic of Chapter

7. It is noteworthy that few papers about the lower divisions on professional football have been published to date. Given that minor divisions may present different determinants, the research question in this chapter concerns whether the demand for tickets is similar for all Brazilian League tiers. Six panel data linear regressions are carried out, taking into account regional differences among the clubs and divisions.

Chapter 8 involves novel research into a most peculiar kind of tournament in

Brazil: the State Championships. Various types of clubs, from Brand-teams and Non-

League teams, play such competitions. Moreover, another peculiarity among them is that all tournaments are decided in play-off matches. Thus, the aim of this chapter is to evaluate how decisive stages in a tournament and the participation of Brand-teams influence the demand for tickets, as well as match day revenues of all participating teams.

Firstly, a descriptive analysis is performed in order to show some particularities about these tournaments. After that, a 3SLS equations system is employed to test the hypotheses.

32

Empirical analysis of broadcast demand, competitive balance, demand for tickets and revenue generation in Brazilian football market

Chapter 9 represents the first attempt to investigate the broadcast demand in the context of Brazilian football. In this sense, the general aim of the chapter is to discover the determinants of television audiences with regard to the Brazilian League. That said, as football matches are broadcast regionally in Brazil, there may be some differences with the findings of previous papers in other contexts. Therefore, the research question in this chapter concerns whether fans from different states display different behaviours. An empirical analysis is performed using panel data linear regressions with clubs as fixed effects. This estimator considers the following clubs: Corinthians, Palmeiras, São Paulo and Santos in São Paulo State; and Botafogo, Flamengo, Fluminense and Vasco in Rio de Janeiro State. Only matches involving these clubs are broadcast in every state. Three models are created in order to identify three different fan preferences: uncertainty of outcome, win preference and loss aversion.

1.2. Contributions

All the work carried out over the course of the PhD period were presented at specialized conferences about sports economics and management, including the European

Conference on Sports Economics (2015 and 2016), the European Association for Sport

Management Conference (2014 and 2015), the Gijón Conference on Sports Economics

(2015, 2016 and 2017), the IASE International Conference in Sports Economics (2014), the Congreso Iberoamericano de Economía del Deporte (2014, 2016 and 2017) and the

Congresso Brasileiro de Gestão do Esporte (2015). Some of the research was also presented at general economics conferences, such as the April International Academic

Conference on Economic and Social Development (2016 and 2017) and the International

Conference on Applied Research in Economics (2015). Therefore, all research included in the present dissertation has received comments and feedback from the most relevant specialists in this field, which should have helped to improve the quality of the thesis.

33

Introduction

When analysing the importance of various tournaments with regard to revenue generation in Brazil, differences have been among the tournaments within the Brazilian football market. The outputs shown that only three competitions have a statistical impact on revenue generation: the Brazilian Cup, the Copa Libertadores and the Brazilian

League. However, the findings also show that competition design plays a major role. Both cups have an impact on the current economic year, while a strong performance in the

Brazilian League is reflected in the following season. This chapter is an adaptation of the paper entitled as ‘Influencia de los torneos en la generación de ingresos’, published in

Revista de Psicología del Deporte in 2016.

The chapter entitled ‘The influence of competitive design’ empirically confirms that the level of competitiveness in the Brazilian League has improved following the change in the design of the competition to double a Round-Robin format. In addition, the findings show that fans’ interest has involved no statistical variation since the change in

2003. That chapter is based on the paper called ‘Playoffs or Just League: A Debate in

Brazilian Football’, published in The Open Sports Science Journal in 2016.

A methodological contribution is presented in the chapter on Competitive

Balance. The APD index is presented for the first time and used to compare and analyse the level of competitiveness with regard to nine professional football leagues. Moreover, the results confirming the Brazilian League as the most competitive of all the leagues in the sample is the empirical contribution from that work. This chapter is an adaptation from the paper ‘Reanalizando la competitividad en la industria del fúlbol: Diferencia acumulada de puntos’ published in Revista de Administração de Empresas (RAE-FGV) in 2016.

34

Empirical analysis of broadcast demand, competitive balance, demand for tickets and revenue generation in Brazilian football market

Some empirical evidence is given in the chapter entitled ‘The demand for tickets in the Brazilian League First Division’, which is in line with previous findings on

Brazilian football, as well as showing some differences. Nevertheless, the most important contribution concerns the extent to which stadium features affect the demand for tickets and ticket prices. The fact is that playing at non-usual stadiums reduces the demand for tickets, but allows clubs managers to charge expensive ticket prices for games held there.

On the other hand, the higher the overall quality of the stadium and its specific aspects

(security, comfort and hygiene), the greater the offset regarding the negative effects of playing at non-usual stadiums. Thus, it is confirmed that good facilities not only attract fans, they can help to increase the demand for tickets where they are offered.

The results of the chapter on the differences in the demand for tickets in each division of the Brazilian League confirm that current sport performance is a similar driver for every divisions, which in turn increases the demand for tickets in all of them.

However, historical success increases attendance rates in the top divisions, but has no impact on the lower levels. On the other hand, higher regional socio-economic aspects are related to a higher demand for tickets in the third and fourth tiers.

The Brazilian state championships are researched for the first time in the chapter entitled ‘The demand for tickets: the distribution of wealth in the Brazilian State

Championships’. The analysis of the demand for tickets demonstrates that decisive matches increase demand, while weekends also lead to higher attendances. However, the most important empirical contribution of this study is that brand-teams strongly increase demand for tickets. Furthermore, the presence of these clubs in the State Championships is essential, as they provide a redistributive effect on revenue among all participants.

35

Introduction

The chapter on broadcast demand in Brazilian football presents both a theoretical contribution and empirical evidence. The behavioural differences between two football markets inside the same country are analysed for the first time in this work, as previous studies have only considered the country as a whole in their estimations. The empirical analysis evidences that Rio de Janeiro and São Paulo fans have some dissimilar determinants in respect of television audiences. Moreover, the determinants of broadcast demand were examined for the first time in Brazil and therefore the outputs may be interesting to clubs, television channels and policymakers.

36

Empirical analysis of broadcast demand, competitive balance, demand for tickets and revenue generation in Brazilian football market

2. THE BRAZILIAN FOOTBALL MARKET

The Brazilian football market is different in many ways to the professional

European ones in terms of institutional structure, the sporting calendar, domestic tournaments and several other peculiarities. The purpose of this first chapter is to explain in detail how the Brazilian football market is organized, as well as look at some current economic and financial features with the use of a descriptive analysis.

2.1. Institutional Structure

FIFA is the governing body for football on a worldwide scale, with 211 national affiliated associations, which is even more than the United Nations (193). They are divided into six continental Confederations: African (CAF), Asian (AFC), European

(UEFA), North American and Caribbean (CONCACAF), Oceania (OFC) and South

American (CONMEBOL). Founded in 1904 and based in Switzerland, FIFA is responsible for organizing and structuring the major international tournaments of football, and beach soccer.

According to FIFA’s official website (www..com): “FIFA supports the associations financially and logistically through various programmes. But they also have obligations. As representatives of FIFA in their countries, they must respect the statutes, aims and ideals of football’s governing body and promote and manage our sport accordingly.” This means that any national association has to follow FIFA’s rules, while enjoying some degree of autonomy in terms of organize its own domestic market. This is the starting point of the Brazilian market’s peculiarities.

While, in the most part of the countries, football is organized and controlled by

Football Federations, the Confederação Brasileira de Futebol (CBF), Brazil’s Football

Confederation, is the main governing body in the Brazilian context. This organization is 37

The Brazilian Football Market called a confederation because it comprises 27 State Football Federations, each of which covers a specific Brazilian state. In this sense, the CBF shares its command with state bodies, giving them some responsibilities, such as club and player registration, as well as organizing their own competitions

However, there is another difference worth noting: although, in the most important football markets, Football Federations are mainly responsible for formulating rules and controlling national teams, their domestic tournaments are organized by proper Leagues, such as La Liga in Spain, the Premier League in UK and the in Germany. On the other hand, the CBF controls the national team, as well as organizes all domestic national competitions in Brazil. Therefore, the CBF determines the calendar, the design of the competitions, the promotion and relegations systems, and all financial aspects of the national tournaments, including sponsorship, broadcast rights and revenue sharing.

2.2. Sporting Calendar and Domestic Tournaments

The Brazilian football calendar runs alongside the calendar year, that it, it begins every January and finishes every December. This may be understandable given that Brazil is located in the southern hemisphere, meaning that summer occurs between December and March. Nevertheless, Brazilian domestic tournaments are structured in an unusual way. Figure 1 depicts the 2017 Brazilian competition schedule.

38

Empirical analysis of broadcast demand, competitive balance, demand for tickets and revenue generation in Brazilian football market

Figure 1. 2017 Brazilian competition schedule

Source: Available at http://cdn.cbf.com.br/content/201701/20170112102000_0.pdf

Different kinds of tournament can be observed in Figure 1. Indeed, there are two more competitions than other countries habitually have. Despite the National League,

National Cup and International Competitions, the Brazilian clubs must also play at state and regional levels. This is a consequence of the vastness of Brazilian territory and the way in which football has developed in the country. However, the distribution of tournaments over the year is the least common characteristic. The clubs opening the season play at state and regional levels, as well as in the Brazilian Cup (Copa do Brasil) and Copa Libertadores de América (Libertadores Cup), which is the most important continental competition in . On the other hand, the Brazilian League itself only starts in May, three months after the beginning of the season.

The high number of tournaments may result in too many matches. Indeed, the top

Brazilian clubs play many more games than European ones. Among the European top

Leagues, the Premier League in the UK may involve clubs playing an even higher number

39

The Brazilian Football Market of matches. In a hypothetical situation, where a club is successful in all the tournaments in which it participates, its team could play a total of 74 official matches in the Premier

League, the FA Cup, an international competition and the Capital One Cup. However, that would be a rarity. In fact, Barcelona was the European club that played the highest number of matches in the 2015 season, with 65 games. On the other hand, 10 Brazilian clubs played more the 65 matches in that same season1. Furthermore, the most successful

Brazilian club may end up playing as many as 91 official football matches in the 2017 season2.

Another oddity is that domestic tournaments do not stop for official international fixtures organized by FIFA. Although the best Brazilian players play abroad, there are still some international players who play for Brazilian clubs. Hence, these clubs need to play some domestic matches without their (theoretically) best players.

2.2.1. Brazilian State Championships

The Campeonatos Estaduais, Brazil’s State Championships, are the oldest football tournaments in Brazil. The oldest of them is the Campeonato Paulista (São Paulo

State Championship), created in 1902. These kinds of tournament were organized given the vastness of the Brazilian territory, as well as the weak transport system in the country at the beginning of the last century. Thus, as it was impossible to structure a national competition, the clubs started to play among themselves at state levels. The States

Championships have been played uninterrupted since its first edition.

The State Championships take place in every state in Brazil. The country is a federal republic formed by 26 states and a federal district, meaning that there are 27 state

1 For more information, see http://www.lance.com.br/futebol-nacional/clubes-brasileiros-jogaram-mais- 2015-que-elite-europa.html. 2 For more information, see http://espn.uol.com.br/noticia/636492_veja-como-um-clube-brasileiro-pode- fazer-91-jogos-so-oficiais-em-2017. 40

Empirical analysis of broadcast demand, competitive balance, demand for tickets and revenue generation in Brazilian football market championships. This geographic phenomenon also results in diverse features within these championships. Only clubs from the same state can play in each State Championship, while there are regular matches between teams with enormous competitive differences, such as the current Libertadores Cup holder playing against a Non-Division team.

Compare this situation with that of Spanish football: if Spain’s Comunidades

Autónomas (autonomous communities) were regarded as states, for example, clubs like

Real , Atlético de Madrid, Getafe, Rayo Vallecano, Alcorcón, Las Rozas, Torrejón and Vallecas would all been playing in the Madrid State Championship, despite the gap in their sporting talent and finances. At the same time, the Cataluña State Championship would be disputed by Barcelona, Espanyol, Palamós, La Cava, Balaguer and Terrassa, among others.

The State Football Federations are responsible for defining the rules of their respective tournaments. Hence, tournaments could involve different competition designs and a varying number of participants. However, one characteristic that is common to all is that every State Championship has play-off stages. In addition, most of the State

Championships have more than one division, which allows the State Football Federations to decide upon the promotion and relegation system.

Undoubtedly, the economic and technical gaps among the clubs play a crucial role in these tournaments. The champions in every State Championship are frequently from the First Division of the Brazilian League. However, some clubs from lower tiers and even Non-Division clubs have been victorious in the State Championships in the past.

2.2.2. Regional Championships

The oldest regional, or interstate, championship is the Taça Rio-São Paulo (Rio-

São Paulo Cup), which was played for the first time in 1933. However, in 1966, when

41

The Brazilian Football Market clubs from other states were included, it lost its regional character. Meanwhile, in the

1990s, the Taça Rio-São Paulo returned, with other regional championships created, such as the Copa Sul-Minas (South and Minas Gerais Cup) and the (North- east Cup), before ending again at the beginning of the 2000s. Recently, the Copa do

Nordeste returned, while two other tournaments were created, namely, the

(Green Cup) and the (Premier League). Notwithstanding, their current relevance into Brazilian football market remains unclear for both clubs and fans.

2.2.3. Brazilian Cup

The Brazilian Cup is similar to some national cups in the world, such as the Copa del Rey, the FA Cup and the Taça de Portugal. The first edition of the tournament, which has a knockout system, took place in 1989; it has been played without interruption since then. The biggest return from winning the Brazilian Cup is that the winner is entered into the next edition of the Libertadores Cup edition. Thus, it represents a shortcut to the most important South American competition and even an opportunity for clubs from lower divisions to participate. Indeed, in recent years, two Second Tier teams have been champions of the competition: Santo André in 2004 and Paulista de Jundiaí in 2005.

The number of participants in this championship has changed over time. The first edition of the Brazilian Cup involved 32 teams, with the number increasing to 40 clubs in 1996 and 64 participants in 2001. The current edition (2017) has 91 participants, with at least one team from each Brazilian state. More information about the classification criteria, previous champions, and the current rules and schedules can be found at http://www.cbf.com.br/competicoes/copa-brasil-masculino/informacoes/2017.

42

Empirical analysis of broadcast demand, competitive balance, demand for tickets and revenue generation in Brazilian football market

2.2.4. Brazilian League

The Campeonato Brasileiro is the most important professional championship in the Brazilian football market. The tournament, which was set up in 1959, is fairly new compared to tournaments in some European countries. As explained before, only State and Regional Championships were played in Brazil before that time. Then, with the aim of securing Brazil’s participation in the first Copa Libertadores de América, the

Confederação Brasileira de Desportos (CBD)3 decided to organize a national championship.

The Brazilian League had various names, including Taça Brasil, Torneio Roberto

Gomes Pedrosa, Taça de Prata, Troféu Copa Brasil, Copa Brasil, Copa União and Copa

João Havelange, and various competitions designs. The first edition in 1956 was played by 16 clubs in a knockout format. Nevertheless, several other formats were used between

1956 and 2002, such as the regionalization and division into groups of the Brazilian

Leagues, a tournament with 96 participants in 1979, and even a championship where a team from the Second Division finished as runner-up in First Division in the same season

(2000). Besides, there was a common element in that every season involved play-off stages to determine the outcome. Notwithstanding, after 44 seasons, the Brazilian League adopted a double Round Robin design in 2003.

Currently, the Brazilian League is divided in four tiers: First, Second, Third and

Fourth Divisions. Naturally, the first edition of the Brazilian League represents the beginning of its First Division, which started in 1959. The Second Division played for the first time in 1971. However, only in the 1980s did this tier start play on an annual basis.

The first edition of the Third Division was played in 1981, while the Fourth Division was

3 The governing body at that time was the CBD, a precursor to the CBF. 43

The Brazilian Football Market created in 2009. All of these lower tiers have featured different competitive designs, as well as varied numbers of participants over time, in the same way as the First Division.

The Brazilian League is an open professional football tournament, with a promotion and relegation system between tiers. However, there are some differences among its competitive categories. Indeed, the Fourth Division aims to be a proper national tournament as it involves clubs from all Brazilian states. Notwithstanding, once the promotion to the upper tiers are achieved, based on sporting prowess, some economic aspects come into play. Thus, teams in the First Division are often clubs from the wealthier cities and states. Table 1 presents some characteristics from every club in the current season (2017).

Table 1. Features of the tiers in the Brazilian League

1st Division 2nd Division 3rd Division 4th Division Nº Participants 20 20 20 68 Design Double Round- Double Round- Group Stage Group Stage robin robin (2 of 10) (17 of 4) Play-offs Play-offs (Quarterfinals) (Round of 32) Libertadores 64 - - - Promotion - 4 4 4 Relegation 4 4 4 - States 8 12 13 27 Source: Self-elaboration from Confederação Brasileira de Futebol official website - www.cbf.com.br

2.3. Economic Context: Labour Market and Financial Aspects

2.3.1. Labour Market

Brazilian domestic football comprises a huge labour market. According to the

CBF5, in 2016, there were 766 professional football clubs and 21,743 professional contracts in Brazil. In addition, the CBF has also reported the existence of 313 amateur

4 Frequently are four clubs classified to Libertadores Cup by Brazilian League. However, the CONMEBOL has changed the number of participants in the 2017 edition, consequently increasing the number of clubs by country. On the other hand, this number may change over the season yet. 5 See http://www.cbf.com.br/noticias/a-cbf/raio-x-do-futebol-contratos-e-valores#.WN31qm997IU. 44

Empirical analysis of broadcast demand, competitive balance, demand for tickets and revenue generation in Brazilian football market clubs and 26,966 non-professional contracts in the same season. While the CBF did not scrutinize the non-professional contracts, these could cover both youth and amateur players6. However, the sporting calendar may be harmful to the domestic labour market.

Most of the clubs (around 83%) only play in the States Championships. As these tournaments take place in the first half of the season, several clubs and players has no involvement in professional activities in the second half.

Despite the high number of professional players, only a small proportion of them earn good salaries in Brazilian football. In 2015, approximately 82% of them earned an amount close to the minimum salary in Brazil (R$ 788.00); indeed, in the same year, professional players typically earned more than 200% less than the average salary (R$

2,227.50). The distribution may be understandable, given that, in the football industry, a slight increase in talent represents significant growth in revenue, as highlighted by Rosen

(1981) and Rodríguez (2012). However, taking into account the findings from FIFPro

(2017), 97.43 % of all players in the Brazilian football market come with the 74.4% least paid in the world. The following Table 2 presents some information about salaries in the

Brazilian football market for the 2015 season.

Table 2. Salaries in the Brazilian football market (2015)

Monthly Salary (R$) Monthly Salary (€)* Number of Players % Up to 1,000 Up to 297.15 23,238 82.40 1,000.01 to 5,000 297.16 to 1,485.75 3,859 13.68 5,000.01 to 10,000 1,485.76 to 2,971.50 381 1.35 10,000.01 to 50,000 2,971.51 to 14,857.48 499 1.77 50,000.01 to 100,000 14,857.49 to 29,714.96 112 0,40 100,000.01 to 200,000 29,714.97 to 59,429.92 78 0,28 200,000.01 to 500,000 59,429.93 to 148,574.81 35 0,12 More than 500,000.01 More than 148,574.82 1 0,00 Total 28,203 100 Source: Self-elaboration from http://www.cbf.com.br/noticias/a-cbf/raio-x-do-futebol-salario-dos- jogadores#.WN32xm997IV) *Exchange rate: € 1.00 = R$ 0,29715 on 31st March 2017.

45

The Brazilian Football Market

On the other hand, Brazilian players are highly regarded around the world. Brazil was the most active market in terms of both incoming and outgoing transfers in 2016

(FIFA TMS, 2017), as well as being the most transferred nationality since 2011, with a total of 8,614 players (FIFA TMS, 2016). This represents more than twice the second most transferred nationality (Argentinian) and almost three times the third (English). As there is no information about individual players’ salaries, it is difficult to infer why so many players move outside the country. Nonetheless, the fact that many of them are badly paid, along with around 83% of all clubs stopping their activities in the second half of the season (June-December), may explain the large number of outgoing transfers.

2.3.2. Financial Aspects

There has been an increasing trend in Brazilian clubs’ revenue. The total revenue for the top 27 clubs in Brazil has more than doubled since 2009, by around 18% each year. The only decrease took place in the 2014 season, when it fell by 2%, although the positive trend returned the following year, as can be seen in Figure 2.

However, growth is mainly due to one factor: broadcast rights. Although other factors have been increasingly important in the last six years, only broadcast rights have had a drastically positive impact. Moreover, clubs usually consider the transfer fees as part of their revenue, despite the level of uncertainty regarding this source on annual basis, not to mention the (theoretical) reduction in the quality of their squads, as highlighted in

Figure 3.

46

Empirical analysis of broadcast demand, competitive balance, demand for tickets and revenue generation in Brazilian football market

Figure 2. Total revenue in Brazilian football: top 27 clubs

4000 35% 3640 3500 31% 30% 29% 3128 3154 2893 3000 24% 25% 2500 2216 20%

2000 1723 15% 15% 1389 1500 11% 10%

1000 5%

500 0% -2% 0 -5% 2009 2010 2011 2012 2013 2014 2015

Revenues (R$ MM) Variation

Source: Self-elaboration from Análise Econômico-Financeira dos Clubes de Futebol Brasileiros (ITAÚ

BBA, 2016).

Figure 3. Total revenue by source: top 27 clubs (R$ millions)

4000 3640

3500 3218 3154 2893 3000

2500 2216

2000 1723

1500

1000

500

0 2010 2011 2012 2013 2014 2015

TV Sponsors Transfers Matchday Stadium Others Total

Source: Self-elaboration from Análise Econômico-Financeira dos Clubes de Futebol Brasileiros

(ITAÚ BBA, 2016).

47

The Brazilian Football Market

Sloane (1971) developed a classical theory, which regards European football clubs as ‘utility maximizers’7. The theory states that these football clubs essentially spend as much money as they can in order to hire talent needed to achieve sporting success. This phenomenon may be observable in the Brazilian football as well. Even though Brazilian clubs have been increasing their revenue, their expenses have been increasing at a similar rate, except for 2015, when expenses saw a slight decrease while there was significant growth in revenue, as can be seen in Figure 4.

Figure 4. Net revenues, operational costs and EBITDA: Brazilian clubs

4000

3500

3000

2500

2000

1500

1000

500

0 2010 2011 2012 2013 2014 2015

Revenues (R$ MM) Costs (R$ MM) EBITDA (R$ MM)

Source: Self-elaboration from Análise Econômico-Financeira dos Clubes de Futebol Brasileiros (ITAÚ

BBA, 2016).

Utility maximization behaviour may be understandable in the Brazilian context, as the clubs are non-profit companies and do not need to share profits with stakeholders.

In this sense, their main objective is winning matches and trophies. On the other hand, they could be better managed, as the clubs have been reporting increased levels of debt,

7 It is also defined as ‘win maximization’; see, e.g., Késenne (1996) and Garcia-del-Barrio and Szymanski (2009). 48

Empirical analysis of broadcast demand, competitive balance, demand for tickets and revenue generation in Brazilian football market given that, every year, the clubs are investing all of their financial resources in buying and remunerating players. Club debt is presented in Figure 5.

Figure 5. Total debt: Brazilian clubs

7000 5977 5721 6000 4789 5000 1499 1666 3796 4000 1119 1174 1106 860 3000 1142 1170 2000 3048 3205 2528 1000 1766 0 1 2 3 4

Taxes (R$ MM) Operating Debts (R$ MM) Bank Debts (R$ MM) Total Debt (R$ MM)

Source: Self-elaboration from Análise Econômico-Financeira dos Clubes de Futebol Brasileiros

(ITAÚ BBA, 2016).

One way to secure a more sustainable financial footing could be to cut expenses and find new capital injections in order to pay off debts, as suggested by Barajas and

Rodríguez (2014) in the case of Spanish football. Nevertheless, Brazilian football is an under-researched topic nowadays. In this sense, there needs to be much more focus on trying to better understand and enhance the resilience of Brazil’s football market.

49

The Brazilian Football Market

50

Empirical analysis of broadcast demand, competitive balance, demand for tickets and revenue generation in Brazilian football market

3. THE IMPACT OF DIFFERENT TOURNAMENTS ON THE

REVENUE GENERATION

3.1.Problem statement

The Brazilian football clubs have been participating in several competitions over the seasons, as explained in the previous chapter. On the one hand, their sporting influence and prowess are clear at the state, regional, national and international levels. On the other hand, there may be a lack of parity regarding their economic impact, as the relevant tournaments are organized by different federations, meaning that the values of the championship prizes may also differ.

Szymanski (2015) highlights two essential relationships in professional football: the more a club invests in players, the greater its long-term sporting performance needs to be; at the same time, the greater the levels of its sporting success, the more revenue should be generated. Sass (2016) agrees with these assumptions, adding that continued success increases a club’s market size and in turn provides higher levels of revenue.

Soriano (2009), meanwhile, has defined this relationship as the virtuous circle of football.

Szymanski and Kuypers (1999) and Barajas, Fernández-Jardón and Crolley (2005), have empirically evidenced this relationship in England and Spain, respectively.

As Szymanski and Kuypers (1999) only analysed the performance of clubs in the

English Premier League, the authors neglected the possibility of evaluating the impact of success in other tournaments, such as continental and international cups. On the other hand, Barajas, Fernández-Jardón and Crolley (2005) have estimated the impact of every tournament in combination over a season. Conducting such a comprehensive analysis makes it impossible to examine how each championship affects revenue generation, since

51

The impact of different tournaments on the revenue generation there are different sporting, media and economic factors at play. With these previous observations in mind, the aim of the present chapter is to examine how each tournament impacts on revenue generation among professional football clubs.

3.2. Methods

The five most important tournaments in the Brazilian football market are analysed: namely, the Brazilian League, the Brazilian Cup, the State Championships, the

Copa Libertadores de América and the . The data set comprises an unbalanced panel with 28 professional football clubs over five seasons: 2010, 2011, 2012,

2013 and 2014. As the competition designs are different within the sample, two measures of sporting performance are employed.

The sporting performance in the Brazilian League (i.e., double round-robin format) is examined through a ranking created by Szymanski and Kuypers (1999). The formula is as follows:

푅푎푛푘𝑖푛푔 = − log (푝/(푛 + 1 − 푝)) where p is the final position of each club and n is the number of participants. This formula is used due to the existence of a promotion and relegation system in the Brazilian League.

Thus, in the present case, the value of n is 40, in light of the number of participants in the first (20) and second (20) tiers during these seasons. Besides Szymanski and Kuypers

(1999), this raking is also used in other works by Barajas, Fernández-Jardón and Crolley

(2005), Kuper and Szymanski (2010), and Szymanski (2015).

The knockout tournaments are examined through a diagram elaborated by Barajas,

Fernández-Jardón and Crolley (2005). This method transforms the final position into points, as shown in Table 3 below:

52

Empirical analysis of broadcast demand, competitive balance, demand for tickets and revenue generation in Brazilian football market

Table 3. Sporting performance diagram

Final Classification Position Points Champion 1 37 Runner up 2 33 Semi-final 3-4 27 Quarterfinal 5-6-7-8 21 Round of 16 9-10-11-12-13-14-15-16 15 Round of 32 17…32 9 Round of 64 33…64 6 Round of 218 65…128 3 Other rounds 129+ 0 Source: Self-elaboration based on Barajas, Fernández-Jardón y Crolley (2005).

In an open football market, a strong sporting performance can lead to participation in more prestigious tournaments. The best teams in the Second Division are promoted to the first tier, while the top four clubs in the Brazilian League qualify for the Libertadores de América and the Brazilian Cup titles, which in turn leads to qualification for the best

South American tournament as well. Thus, lagged seasons may impact on current revenue generation. Moreover, as players’ transfer fees are a great source of revenue in Brazilian football, with around 14.82% of total revenue generated by this source (Itaú BBA, 2015), a strong sporting performance could increase the probability of outgoing transfers.

Therefore, a lagged sporting performance is employed with the aim of capturing these effects.

The econometric approach consists of panel data linear regression with clubs as fixed effects. The fixed effects estimator is employed in order to capture non-observable characteristics for each club, which could impact on revenue generation. Season control variables are also used. The model is as follows:

53

The impact of different tournaments on the revenue generation

revenuesit = αit + β1leagueit + β2leagueit-1+ β3libertadoresit + β4libertadoresit-1+

β5sudamericanait + β6sudamericanait-1+ β7cupit + β8cupit-1+ β9stateit

+ β10stateit-1+ β11season11it + β12season12it+ β13season13it +

β14season14it + μit

The quality of the model and the stability of its coefficients are tested with the

Variance Inflation Factor (VIF). This approach analyses the multicollinearity level between the dependent variable and the explanatory factors. The outputs are presented in

Table 4.

Table 4. VIF outputs

Variable VIF VIF² Tolerance R²

Revenues 2.31 1.52 0.4330 0.5670

League 1.63 1.28 0.6139 0.3861

Lag_League 1.58 1.26 0.6328 0.3672

Libertadores 1.85 1.36 0.5392 0.4608

Lag_Libertadores 1.49 1.22 0.6704 0.3296

Sudamericana 1.19 1.09 0.8428 0.1572

Lag_Sudamericana 1.16 1.08 0.8643 0.1357

Cup 1.32 1.15 0.7591 0.2409

Lag_Cup 1.21 1.10 0.8247 0.1753

State 1.18 1.09 0.8446 0.1554

Lag_State 1.25 1.12 0.7998 0.2002

VIF mean 1.47

Source: Self-elaboration

Kennedy (1992) and Neter, Wasserman and Kutner (1989) recommend a VIF lower than 10. However, Pan and Jackson (2008) propose that this value be lower than 4.

54

Empirical analysis of broadcast demand, competitive balance, demand for tickets and revenue generation in Brazilian football market

Therefore, the outputs confirm the quality of the model and its coefficients when taking into account both works.

3.3. Results and Discussion

Table 5 below shows the outputs from the linear regression.

Table 5. Individual effects from different tournaments on revenue generation: Brazilian football market 2010-2014

Variables Fixed Effects Model

League 4.766e+06 (2.875e+06) Lag_League 8.313e+06*** (2.665e+06) Libertadores 1.011e+06*** (355,163) Lag_Libertadores 67,769 (326,714) Sudamericana 192,702 (350,004) Lag_Sudamericana -260,318 (325,709) Cup 881,786*** (280,334) Lag_Cup 344,771 (266,554) State 626,831 (454,784) Lag_State -53,555 (354,416) Constant 2.311e+07 (1.999e+07)

Season Fixed Effects Yes Club Fixed Effects Yes Observations 124 Number of clubs 28 R² 0.651 Standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1

55

The impact of different tournaments on the revenue generation

The regression explains around 65.1 % of the total revenue from Brazilian clubs.

The season dummy variables have captured the statistical increase revenues across the sample. Three tournaments have a statistical impact on revenue generation, as shown in the table above: Brazilian League (p = 0.002), Brazilian Cup (p = 0.006) and Copa

Libertadores (p = 0.002). Nonetheless, in the Brazilian League, its lagged term provides higher revenues, while, in the Brazilian Cup and Copa Libertadores, it is the current season that enhance revenues.

The State Championships have no statistical impact on revenue generation.

Although there are prizes for final classification, these values have little influence on total income. The Copa Sudamericana did not show any statistical effect either. Indeed,

Brazilian clubs did not give much importance to this tournament during these seasons.

Meanwhile, the CBF has been changing the criteria for classification on a regular basis.

Between 2010 and 2012, the clubs placed among the 5th to 12th places in the Brazilian

League qualified for the Copa Sudamericana. However, in the 2013 and 2014 editions, only the clubs that had already been eliminated from the Brazilian Cup were able to participate in this international competition, which meant that clubs placed in lower positions could have played for the Copa Sudamericana. Although it seems paradoxical, in the 2015 edition, a Brazilian club that had not competed in any of the national divisions

(i.e., a Non-) could have played in the Copa Sudamericana. Therefore, the relative unimportance of that tournament in the Brazilian football market is confirmed.

The competition designs in the Brazilian League, the Brazilian Cup and the Copa

Libertadores may explain their differences in terms of revenue generation. In knockout tournaments, clubs increase their incomes at every stage thanks to prizes or match day revenue. In this sense, such revenue can impact on current economic performance, as confirmed by a report on the Brazilian Cup, issued by the CBF to State Federations and

56

Empirical analysis of broadcast demand, competitive balance, demand for tickets and revenue generation in Brazilian football market their clubs (CBF, 2015). Moreover, in Article 16.7 of the General Rules of CONMEBOL, published in 2014, that a club can earn more “according to the matches it plays at a local level and depending on the rounds it passes” (CONMEBOL, 2014).

On the other hand, a lagged Brazilian League affects revenue generation, in the sense that, although the best clubs in the league receive more valuable prizes, most of the benefits from final qualification are likely to be enjoyed in the next season, when the top four clubs play for the Copa Libertadores de América. In this tournament, teams can expect higher levels of television-related income based on their final position, and even enhance their sponsorship and outgoing transfer fees. In this sense, the effects provided by the lagged season are understandable.

There is no similar paper, which has analysed the individual effects from tournaments, in the literature. Notwithstanding, some works may concur with the present findings. Krautmann and Ciecka (2009) confirm that participation in the Major League

Baseball (MLB) play-offs provides an increase of US$ 11 million in revenue for the clubs.

Moreover, Andreff and Bourg (2006), Pawlowski, Breuer and Hovemann (2010), and

Peeters (2001) highlight the reduction in competitiveness within domestic European

Leagues as a consequence of more valuable UEFA Champions League prizes. Thus, the aforementioned studies may confirm the importance of knockout tournaments in revenue generation. Furthermore, Sass (2016) remarks that sporting performance can increase both attendance and revenue. In this sense, it is reasonable to assume that a lagged performance can have an impact on the next season.

3.4. Final Remarks

The findings confirm that the design of a competition and its importance impact on revenue generation, while other results confirms that, in the Brazilian football market,

57

The impact of different tournaments on the revenue generation three tournaments are statistically important to the enhancement of revenue: the Brazilian

League, the Brazilian Cup and the Copa Libertadores. However, success in both domestic and international cups have a financial impact on the current season, while a lagged sporting performance in the league provides higher income levels. Further research could analyse whether these effects also occur in European football. Nonetheless, the financial impact of international competitions, such as the UEFA Champions League, tends to be different in other countries.

58

Empirical analysis of broadcast demand, competitive balance, demand for tickets and revenue generation in Brazilian football market

4. THE INFLUENCE OF COMPETITIVE DESIGN

4.1.Problem Statement

As described in Chapter 2, the Brazilian League has involved several different competition designs throughout its history. Despite the differences among the seasons, between 1959 and 2002, all editions were decided in play-off stages. In 2003, however, the CBF employed the double round-robin format in line with the top European Leagues.

There are many supporters of the previous design, while there are others who prefer the round-robin format. Moreover, there have been some discussions about which tournament style is more balanced.

Drummond, Araujo Jr. and Shikida (2010) indicate that the CB may be improved following a design change. Nevertheless, the influence of a format change on fan preferences remains unclear. Therefore, the aim of this chapter is to reanalyse the effects provided by this design change on CB, as well as examine fans’ interest in both periods.

4.2. Theoretical Background

Neale (1964) notes the “peculiar economics” of the sports industry. The author states that a sporting product is a joint product, in that the presence of at least two companies (e.g., clubs) is mandated in order to create a sporting product (e.g., a match, a fight, a race or a tournament). Moreover, using the Louis-Schmelling paradox, he also explains that competitiveness is the key driver in increasing fans’ interest and, consequently, revenue.

The earliest sports economics paper developed by Rottenberg (1956) emphasizes the importance of the uncertainty of outcome to attract fans. According to the author, ”[a] team that wins 80 per cent of its games will attract fewer patrons than a pennant-winning

59

The Influence of competitive design team that wins 55 per cent of them” (Rottenberg, 1956, p. 246). This concept is defined as the classical Uncertainty of Outcome Hypothesis (UOH).

Under these assumptions, many studies have been devoted to analyse the CB in sports leagues. Indeed, the CB is defined by Levin et al. (2000) in terms of when all clubs enjoy a reasonable probability to qualify for the play-offs. Rodríguez (2012) has adjusted this concept for European football, defining it as when all participants have the possibility of qualifying for the UEFA Champions League. According to Forrest and Simmons

(2002), high levels of competitiveness ought to provide better levels of television audience size, average attendance and general interest.

Zimbalist (2002) states that CB can be analysed from two perspectives: level of dominance and level of concentration. However, CB could also be examined by short-, medium and long-term dimensions. Indeed, there are several methods for analysing CB in sports leagues. The most used methods include: the Herfindahl-Hirschman Index (HHI) and C5 Index of Competitive Balance (C5ICB), as employed by Michie and Oughton

(2004); the Gini coefficient, as used by Schimidt (2001); the Competitive Balance Ratio

(CBR), as developed by Humphreys (2002); and the National Measure of Seasonal

Imbalance (NAMSI), as created by Goossens (2006). Evans (2014) provides an overarching literature review on this topic.

There are various competitive designs for organizing a football tournament, with the most common being the round-robin format, which can be defined as a tournament where the all the teams face each other, with the champion being the team with the most points at the end. The Top European Leagues employ a variant of this kind of this format known as Double Round-Robin, which means that all clubs must play against each opponent twice, i.e., a match at home and a match as the away team. This format can be

60

Empirical analysis of broadcast demand, competitive balance, demand for tickets and revenue generation in Brazilian football market seen in England (Kendall, 2008), Germany (Bartsch, Drexl & Kroger, 2006), Italy (Della

Croce & Olivieri, 2006), Spain, France and other countries (Griggs & Rosa, 1996;

Goossens & Spieksma, 2012). Nevertheless, leagues such as those in , ,

Belgium, , Denmark, and the US are organized around different formats, as reported by Griggs and Rosa (1996) and Goossens and Spieksma (2012). On the other hand, the impacts generated by design changes have hardly been investigated. Nevertheless, three papers can be found about this topic.

Goossens, Beliën and Spieksma (2010) researched the possible changes in the competitive model in the Belgian League, finding that the design suggested by the Royal

Belgian Football Association could improve the level of competitiveness in its league.

However, they stressed that an increase in the total number of matches could jeopardize the performance of the teams qualifying for European-level championships.

Lee and Fort (2011) conducted a longitudinal investigation into English football and observed changes in competitiveness after historical events, such as pre- and post- war comparisons. Other than all the changes in CB in earlier seasons, this study identified a significant increase in imbalance following the 1997 season. According to the authors, this fact came as a consequence of the financial inequality among participating clubs in the UEFA Champions League.

In the third paper, from Drummond, Araujo Jr. and Shikida (2010), the CB in the

Brazilian League was analysed over the period from 1971 to 2009 using four indices: C4,

Gini, Herfindahl and Top 4. The results graphically show that competitiveness has increased slightly over the years, while the regression models confirm an increase in CB after the design change. Notwithstanding, as the focus of CB theory is the level of fans’ interest, this topic could be better analysed.

61

The Influence of competitive design

4.3. Methods

A natural experiment that happened to Brazilian football is analysed. A balanced period, involving 12 seasons with the play-off format (1991-2002) and the following 12 seasons with the Round-Robin design (2003-2014), is selected in order to examine the effects on CB and fans’ level of interest. Two hypotheses are tested, which take into account recent complaints from the CBF, clubs and fans who prefer the previous design:

H1: Tournaments involving the play-off design are more balanced than ound-

Robin ones.

H2: Fans prefer a tournament with play-offs.

As the number of participants varies over the seasons analysed, the CB is analysed using the the Herfindahl Index of Competitive Balance (HICB), as developed by Michie and Oughton (2004). This formula is an adaptation of the Herfindahl-Hirschman Index

(HHI) which eliminates the errors caused by the different numbers of teams. The index is presented as follows:

∑푁 푠2 퐻퐼퐶퐵 = ( 𝑖=1 𝑖 ) ∗ 100 1⁄푁

where si is the share of team i of the total points in the league, and N is the number of teams in the league.

The C4ICB, which is a formula used to provide robustness to the results, is an adaptation proposed by Drummond, Araújo Jr. and Shikida (2010) of the C5ICB, as developed by Michie and Oughton (2004). The distance between the four top teams and the other participants is related to the number of teams qualifying every year for the main

62

Empirical analysis of broadcast demand, competitive balance, demand for tickets and revenue generation in Brazilian football market

South American tournament, the Copa Libertadores de América. The index is shown below:

퐶4 퐶4퐼퐶퐵 = ( 푟푎푡𝑖표 ) ∗ 100 4 푁 where

푆푢푚 표푓 푝표𝑖푛푡푠 푓푟표푚 푡ℎ푒 푡표푝 푓표푢푟 푐푙푢푏푠 퐶4 = 푟푎푡𝑖표 푆푢푚 표푓 푝표𝑖푛푡푠 푓푟표푚 푎푙푙 푝푎푟푡𝑖푐𝑖푝푎푛푡푠

The maximum balance in both indices has a value of 100. Thus, if the value increases, CB decreases. The estimations are made using the points obtained at the end of the season. The results of the play-off matches are considered to estimate the final classification in those seasons (1991-2002). The data have been collected from the

Rec.Sport.Soccer Statistics Foundation website (www.rsssfbrasil.com).

Average attendances are used as a proxy of interest, following the work by

Szymanski (2001). These data are collected from the Futdados website

(http://futdados.com). The average attendance is calculated by dividing the total attendance by the number of matches. The occupancy rate is not appropriate to Brazilian football due to the fact that football clubs in Brazil often play as the home team in different stadiums.

Firstly, the level of competitiveness and fans’ interest in both periods of the

Campeonato Brasileiro is observed graphically. Subsequently, the values are compared statistically using the Mann-Whitney U test for independent samples, where p-values

63

The Influence of competitive design lower than 0.05 are considered significant. The seasonal HHICB, C4ICB and average attendances are presented in Table 6 below.

Table 6. Seasonal HICB, C4ICB and average attendance during the Campeonato Brasileiro 1991-2014 seasons

Season HICB C4ICB Average Attendance

1991 131.18 128.1534 13,760 1992 128.70 127.6813 16,814 1993 195.28 193.7008 10,914 1994 127.74 144.1935 10,222 1995 156.05 139.6619 10,332 1996 136.63 132.1782 10,913 1997 159.92 164.0509 10,497 1998 150.18 155.3374 13,487 1999 155.56 150.7937 17,018 2000 154.13 149.1388 11,546 2001 164.86 151.7757 11,401 2002 142.77 137.2756 12,866 2003 137.64 134.0804 10,468 2004 138.34 135.1206 7,556 2005 133.64 129.7897 13,600 2006 132.31 133.2694 12,300 2007 118.10 124.2857 17,471 2008 137.93 133.6207 16,992 2009 125.24 124.7592 17,807 2010 135.43 132.9735 14,839 2011 133.33 127.5362 14,976 2012 139.13 138.1643 13,013 2013 126.95 129.8828 14,951 2014 133.59 137.4046 16,555 Source: Self-elaboration

4.4. Results

4.4.1. Competitive Balance

The CB in the Brazilian League, using both HHICB and C4ICB, is presented in

Figure 6 and Figure 7 below. A simple graphical analysis suggests that competitiveness has increased since the change to the Round-Robin format. This comparison may be more persuasively emphasized by Figure 8 and Figure 9 below.

64

Empirical analysis of broadcast demand, competitive balance, demand for tickets and revenue generation in Brazilian football market

Figure 6. HICB for the Campeonato Brasileiro

Source: Self-elaboration

Figure 7. C4ICB for the Campeonato Brasileiro

Source: Self-elaboration

65

The Influence of competitive design

Figure 8. Differences in the HICB in the Campeonato Brasileiro: play-offs vs. league

Source: Self-elaboration

Figure 9. Differences in the C4ICB in the Campeonato Brasileiro: play-offs vs. league

Source: Self-elaboration

66

Empirical analysis of broadcast demand, competitive balance, demand for tickets and revenue generation in Brazilian football market

Two particular seasons reported a significant competitive imbalance: 1993 and

2005. The method does not allow for any inference regarding this finding, but there is a possible explained. In 1993, the CBF eliminated the second national division, including the two teams from the Campeonato Brasileiro that were relegated from the 1992 season and another 12 clubs that had played in the lower division in the previous year. This resulted in a tournament with 32 teams, which could have created a major technical imbalance among them. On the other hand, in 2005, the Campeonato Brasileiro was marked by the manipulation of results. As a consequence, one referee went to jail, with the 11 games refereed by him having to be played again. Thus, the imbalance may be a result of the change in the qualification process after the new games. However, this imbalanced situation was only observable when using the HHICB; no relevant findings were obtained by the C4ICB measure.

Although the graphs show an improvement in competitiveness after the changes in 2003, the Mann-Whitney U test for independent samples is carried out to check whether there was any significant difference between the groups. The results are presented in

Table 7 and Table 8 below.

Table 7. Mann-Whitney U test for the HICB: play-offs vs. Round-Robin

Null Hypothesis Test Sig. Decision Distribution in Play- Mann-Whitney U Test ,017* Reject the null offs is the same than in for independent hypothesis League samples Source: Self-elaboration. (* p < 0,05)

Table 8. Mann-Whitney U test for the C4ICB: play-offs vs. Round-Robin

Null Hypothesis Test Sig. Decision Distribution in Play- Mann-Whitney U Test ,008* Reject the null offs is the same than in for independent hypothesis League samples Source: Self-elaboration. (* p < 0,05)

67

The Influence of competitive design

The test confirms statistical difference between the formats, which means there is evidence that the Round-Robin design has increased the level of CB in the Brazilian

League. The finding corroborates the results from Drummond, Araujo Jr. and Shikida

(2010), whose evaluation of the historical levels of competitiveness in the Campeonato

Brasileiro in the seasons from 1971 and 2009, using four indices, emphasized an increase in competitiveness in the more recent years. Thus, their conclusion is that, after the format changed, the main Brazilian football tournament became more competitive. Therefore, the present findings support the previous results, which include some recent seasons.

4.4.2. Fans’ Interest

The preference among fans, as measured by the average attendance at stadiums for each competition design, is evaluated. Figure 10 and Figure 11 show the evolution of average attendance in the Campeonato Brasileiro and the differences between the periods under the league and play-off systems.

The interest in the competition seems to have increased after the design change in the 2003 season, even though the 2004 season had the lowest average attendance. This could be explained in light of the many criticisms about the round-robin design from the press and fans during that season. Those criticisms emerged once the champion of the

Brazilian League in 2003 also won the Campeonato Estadual and the Copa do Brasil.

Thus, the press and the fans imagined that Brazilian football would start to become imbalanced. However, this did not happen: the 2003 season is the only one where a team won all three tournaments in Brazil.

68

Empirical analysis of broadcast demand, competitive balance, demand for tickets and revenue generation in Brazilian football market

Figure 10. Evolution of average attendance in the Campeonato Brasileiro: 1991-2014

Source: Self-elaboration

Figure 11. Comparison of average attendance in the Campeonato Brasileiro: play-offs vs. league

Source: Self-elaboration

69

The Influence of competitive design

Even though the graphs apparently shown an improvement in fans’ interest, the

Mann-Whitney U test has been employed again in order to identify any significant difference between groups. Nevertheless, no statistical difference for both periods is found. The output is presented in Table 9 below.

Table 9. Mann-Whitney U test for average attendance: play-offs vs. Round-Robin

Null Hypothesis Test Sig. Decision Distribution in Play- Mann-Whitney U Test ,101 Accept the null offs is the same than in for independent hypothesis League samples Source: Self-elaboration

This finding confirms that the design change had no statistical impact on fans’ interest. However, a slight growth trend regarding average attendance in the Campeonato

Brasileiro is noted. The average attendance in the period involving the league system was

14% higher than in the seasons involving play-offs. In addition, comparing the 1991 and

2002 championships (the 12 seasons with the play-off format), the attendance decreased by 6.50%. However, in the league period, average attendance grew by 58.15% between the 2003 and the 2014 seasons. It is worth mentioning that the average attendance in the

Campeonato Brasileiro has never been high. The historical peak was 22,953 persons in the 1983 season. Therefore, the decrease from 12,866 spectators in the 2002 season (the last involving play-offs) to 16,555 spectators in the 2014 season (involving the league system) is remarkable.

4.4.3. Considerations about Brazilian Football

Violence is one of the main problems in Brazilian football: 234 football-related deaths were recorded in Brazil between 1988 and 20138. Although the number of deaths

8 ‘Lancenet! Violência entre torcidas já matou 234 pessoas no Brasil, sendo 30 este ano.’ Available from http://www.lancenet.com.br/minuto/Violencia-torcidas_organizadas_0_1044495544.html. 70

Empirical analysis of broadcast demand, competitive balance, demand for tickets and revenue generation in Brazilian football market decreased in 20149, it remains at unacceptable levels. However, Murad (2013) states that the problem is not specifically linked to football, but is a social reflection of Brazil overall, with an intolerable number of victims of automobile accidents, drugs and other killings.

However, until now, not much has been done in Brazil to restrain violence in football, which could be the main reason why certain fans are not attending games in stadiums

(Guterman, 2009).

Brazilian clubs have enjoyed a unique opportunity to attract fans in recent seasons.

Brazil hosted the FIFA World Cup 2014, for which it constructed new stadiums and refurbished many others. Feddersen, Maenning and Borcherding (2006) observe a positive “novelty effect” in the German Bundesliga after the construction of new stadiums. This kind of effect should have applied to Brazilian football as well, as teams should have increased their average attendance levels once they began exploiting the full potential of new sports venues.

The number of members (sócios) of Brazilian clubs is still low, however, when considering the Brazilian population as a whole and the estimated number of fans of each team. Nevertheless, clubs could still improve several aspects of the club-fan relationship to increase their revenue. Solberg (2001) highlights the price bundling strategy to increase revenue, while Howard and Crompton (2004) propose the creation of various season ticket models to attract different groups of fans. Specifically, in the case of Brazil,

Madalozzo and Villar (2009) propose that Brazilian clubs sell season tickets to their fans, which is not common in Brazilian football. Sordi and Bello (2014) corroborate this idea, while also suggesting a strategy to include other products in these ticket packages

9 ‘Castro CO. Brasil é o recordista de mortes por causa do futebol.’ Available from http://oglobo.globo.com/esportes/brasil-o-recordista-de-mortes-por-causa-do-futebol-14923352. 71

The Influence of competitive design

Another interesting fact should be mentioned: one of the most competitive and attractive leagues in the whole sample happened in the 2009 season, which may not have involved mere causality. In their paper, Ribeiro and Urrutia (2012) explain a program to optimize the scheduling of the Campeonato Brasileiro, involving a model based on the

2005 and 2006 seasons, which the CBF used for the first time for the 2009 tournament.

As a result, one of the most balanced leagues was observed. These authors also report that, while they used the same program to generate the 2010 calendar, it had some limitations because the FIFA World Cup 2010 and a higher ratio of teams from the same state negatively influenced the season. However, although competitiveness had declined from one year to the next, the 2010 tournament was also decided by the last match, a fact that did not occur in some of the previous seasons. Thus, there are other ways to increase competitiveness in a football tournament and make it more attractive.

4.5. Final Remarks

The findings of this chapter cannot be interpreted in terms of the Round-Robin being better than the play-off design. The outputs simply evidence the fact that, in the

Brazilian case, it was beneficial from a competitiveness point of view, but has no impact on fans’ interest. On the other hand, the success of the North American Major Leagues, as well as tournaments such as the UEFA Champions League and the FIFA World Cup, confirm the feasibility and importance of this system of competition.

The results presented here confirm that the Brazilian League became more balanced after the design change to the Double Round-Robin format. On the other hand, the interest among fans has remained consistent over the researched period, despite a slight positive trend. The pursuit of progress in the Brazilian League needs to focus on other aspects, such as reducing the prevalence of violence, optimizing the ticket sales for games and improving the relationship between clubs and fans.

72

Empirical analysis of broadcast demand, competitive balance, demand for tickets and revenue generation in Brazilian football market

5. COMPETITIVE BALANCE

5.1.Problem Statement

The previous chapter evidenced the fact that competitiveness in Brazilian football has been increasing, while fans’ interest has not statistically risen following the design change, despite a slight positive trend. Indeed, the seasonal average attendance has never been high in the Brazilian League, as indicated in Figure 12 below.

Figure 12. Seasonal average attendance in the Campeonato Brasileiro: 1967 to 2016

25000

20000

15000

10000

5000

0

Source: Self-elaboration from Rec.Sport.Soccer Statistics Foundation

According to the seminal papers from Rottenberg (1956) and Neale (1964), the uncertainty of outcome is the key driver for attracting fans. Thus, higher levels of CB should result in better average attendances, as noted by Dobson and Goddard (2011) and

Forrest and Simmons (2002).

The study of CB is relevant, given that related findings can provide some practical implications for policymakers in terms of salary caps (Fort & Quirk, 1995; Szymanski,

2003), revenue sharing (Budzinski, 2012; Szymanski, 2003) and Financial Fair Play

73

Competitive Balance legislation (Franck, 2014; Müller, Lammert & Hovemann, 2012; Peeters & Szymanski,

2014; Preuss, Haugen & Schubert, 2014; Szymanski, 2014). Thus, an analysis of competitiveness could be relevant for professional football leagues.

There is a commonly held view in Brazil that its domestic league is one of the most competitive football tournaments in the world. Indeed, Drummond, Araújo Jr. and

Shikida (2010) have analysed competitiveness in the Brazilian League. However, they did not compare it with any other tournaments. Meanwhile, Levy (2011) has shown that the Brazilian League was more balanced than some other tournaments from 1997 to 2010, but no statistical analysis was performed to compare them. Therefore, the general objective of the present chapter is to make a statistical comparison of CB between the

Brazilian League and some European Leagues.

5.2. Theoretical Background

El-Hodiri and Quirk (1971) developed the first method for analysing CB in sports leagues. Since then, several scholars have created new indices to examine CB, such as

Vrooman (1995), Hausmand and Leonard (1997), and Késenne (2000). However, these aforementioned authors did not take into account the possibility of player transfers, while some even claimed that there was equal revenue capacity among clubs, which is a very different situation to that in professional football nowadays.

On the other hand, numerous others indices, as well as economic and managerial formulas, have been employed over the last decades in reference to different characteristics found in other professional sports. In general, within the sports economics literature, the most popular are: the C5 Index from Michie and Oughton (2004); the Top

3, as developed by Goossens (2006); the Competitive Balance Ratio created by

Humphreys (2002); the Gini Coefficient; and the Herfindahl and Hirschman Index (HHI).

74

Empirical analysis of broadcast demand, competitive balance, demand for tickets and revenue generation in Brazilian football market

Nonetheless, Fort and Maxcy (2003) note that there is no ideal method to analyse competitiveness in sports leagues, as each formula may evaluate CB from different perspectives.

For Zimbalist (2002), CB can be measured from two aspects: level of concentration and level of dominance; meanwhile, Cairns, Jennett and Sloane (1986) state that competitiveness can be analysed from three temporal dimensions: short-, medium-, and long-term. Short-term CB is based on match-by-match data; medium-term CB is examined within a season, such as in relation to a championship race or fighting relegation; and long-term CB concerns the dominance of certain clubs over several seasons.

Short-term CB was first analysed by Peel and Thomas (1988) through betting odds. Moreover, Peel and Thomas (1992), Czarnitzki and Stadtmann (2002), and

Buraimo and Simmons (2008) have employed the Theil Index (Theil, 1967) to measure it. The UCS formula, as developed by Janssens and Késenne (1987) and modified by

Pawlowski and Anders (2012), can measure medium-term CB as well. On the other hand, several papers have analysed long-term CB, such as those by Booth (2004), Eckard

(1998), Goossens (2006), Humphreys (2002), Michie and Oughton (2004) and

Pawlowski, Breuer and Hovemann (2010).

Different ways to measure clubs’ performance for the purpose of analysing competitiveness have also been reported in the literature. The most traditional approach can involve the winning percentage. Noll (1988) first used this approach, followed by

Scully (1989) Fort (2007), Quirk and Fort (1991), Trandel and Maxcy (2011), Vrooman

(1996), and several other authors. Nevertheless, it may be more appropriate to North

American Major Leagues because their competitions have no draws.

75

Competitive Balance

As professional football has three possible outcomes (win, draw or lose), other measurements may be more suitable. Koning (2000) and Michie and Oughton (2004) have employed the ratio of the points of a group of clubs to those of the other participants.

Curran, Jennings and Sedgwick (2009), Goossens (2006), and Szymanski and Kuypers

(1999) have used the dominance of clubs by taking into account the final classification over several seasons.

The analysis of various national leagues may pose some problems due to each country having a different number of participants. Thus, some papers, such as those by

Koning (2000), Michie and Oughton (2004), and Owen, Ryan and Weatherspoon (2007), have controlled for this factor through adjustments to their models.

As explained in the previous chapter, the studies by Levin et al. (2000) and

Rodríguez (2012) defined pure competitiveness in terms of when all clubs have the chance to qualify for the play-offs and the UEFA Champions League, respectively. In this sense, the most balanced tournament should be the one where there is the least difference in points between all participating clubs. Based on this assumption, the present chapter aims to elaborate a new index in order to analyse the level of competitiveness in football leagues and compare it using the accumulated difference of points among the competitors as a tool.

5.3. Methods

5.3.1. The Model

The present paper develops a mathematical index to analyse and compare CB in football leagues with different numbers of participants. The model is based on the accumulated difference of points among the clubs, which is calculated by the subtraction of the total points of the champion with those of the runner up. This is repeated

76

Empirical analysis of broadcast demand, competitive balance, demand for tickets and revenue generation in Brazilian football market successively until the difference is calculated from the penultimate to the last team.

However, the index is adjusted according to the maximum imbalance in order to compare leagues with different numbers of participants, as Goossens (2006) and Owen (2010) have done previously. The index is developed for football leagues where a win is equal to three points, a draw is one point and a loss is zero points.

A hypothetical double round-robin league involving two clubs, A and B, can be used as an example in this context. This design has been chosen as it is the most used in professional football leagues, as confirmed by Goossens and Spieksma (2012). The maximum imbalance is generated when one of the teams (e.g., A) wins all the matches against its opponent (B). Thus, A obtains the maximum number of points (six), while B receives no points.

A league with one more participant (A, B and C) is simulated as well. The maximum imbalance happens when a club (A) wins all of its matches, thus obtaining 12 points. The runner-up (B) wins its other matches and receives six points, while the last placed team (C) loses all its matches, thus receiving no points. The difference in points between the clubs is six, as per the previous illustration. Table 10 shows the maximum imbalance in a double round-robin league with 20 participants.

77

Competitive Balance

Table 10. Maximum imbalance in a double round-Robin League with 20 participants

Position Club W D L Points Difference 1 A 38 0 0 114 2 B 36 0 2 108 6 3 C 34 0 4 102 6 4 D 32 0 6 96 6 5 E 30 0 8 90 6 6 F 28 0 10 84 6 7 G 26 0 12 78 6 8 H 24 0 14 72 6 9 I 22 0 16 66 6 10 J 20 0 18 60 6 11 K 18 0 20 54 6 12 L 16 0 22 48 6 13 M 14 0 24 42 6 14 N 12 0 26 36 6 15 O 10 0 28 30 6 16 P 8 0 30 24 6 17 Q 6 0 32 18 6 18 R 4 0 34 12 6 19 S 2 0 36 6 6 20 T 0 0 38 0 6 114 Note: W = win, D = draw, L = loss. Source: Self-elaboration.

The ‘maximum imbalance’ is applicable in every double round-robin league, independent of the number of participants. Thus, every final classification, which is different from that shown in the Table 10, confirms a higher degree of competitiveness than presented there. Hence, the formula for maximum imbalance is as follows:

푈푛푏푎푙푎푛푐푒푚á푥 = 6 ∗ (푁 − 1)

Thus, considering the previous points, the Accumulated Points Difference (APD) index is as follows:

78

Empirical analysis of broadcast demand, competitive balance, demand for tickets and revenue generation in Brazilian football market

∑푁 (푇푃 − 푇푃 ) 퐴푃퐷 = ( 𝑖=1 𝑖 𝑖+1 ) ∗ 100 푈푛푏푎푙푎푛푐푒푚á푥

Where N is the number of participants and TP is the total points of every club i at the end of the season.

5.3.2. Sample

The following nine football leagues are examined over eight seasons (from

2006/07 to 2013/14): the Brazilian League, the Dutch , the English Premier

League, the French , the German Bundesliga, the Italian Serie A, the Portuguese

Primeira Liga, the and the Spanish La Liga. These tournaments are selected due to having similar versions of the double Round-Robin design. The data are collected from the Rec.Sport.Soccer Statistics Foundation

(http://www.rsssf.com/histdom.html) website.

5.3.3. Statistical Analysis

As explained before, the APD index is measured by the ratio of the accumulated points difference in a league to its maximum imbalance. Thus, the outputs mean is the percentage of imbalance of every football league, which is comparable to other tournaments.

Furthermore, Hann, Koning and Wittellostuijn (2007) note the potential for comparing CB through standard statistical methods. Therefore, after establishing the

APD, the results are compared with the one-way ANOVA with the Tukey post hoc test, in order to find any statistical differences among the leagues. Previously, the Shapiro-

Wilk and Levene tests were carried out to examine the sample.

79

Competitive Balance

5.3.4. The Dispersion of Positions

The dispersion of positions in the final league tables is also analysed. The aim of this approach is to inspect the dominance of certain clubs in the researched football leagues. As each of them has a promotion and relegation system, the clubs that have played in the lower tiers take N+1 as their final position.

5.3.5. Confrontation with Traditional Models

As the APD index is employed for the first time, its comparison with traditional indices may be necessary. Thus, the APD is confronted by the Herfindahl Index of

Competitive Balance (HICB) as well as C4 Index of Competitive Balance (C4ICB), which is an adaption by Drummond, Araújo Jr. and Shikida (2010) of the C5ICB, as presented by Michie and Oughton (2004). The Campeonato Brasileiro is chosen for the purposes of evaluation.

The HICB is an adjustment of the Herfindahl-Hirschmand Index (HHI), which is a formula that is often used in business to measure market concentration (Barajas,

2010). Its outcome is given between 0 and 1, where 1 represents a monopoly and 0 is an infinite number of companies sharing the market. According to Michie and Oughton

(2004), the HHI is sensitive to the number of clubs in every league; the authors also adjusted it to compare different tournaments.

The C4ICB has been developed across a number of papers. Firstly, Koning

(2000) created the CRk, which is the ratio of the total points of k teams to the maximum possible points, which are achievable by those clubs. After that, Michie and Oughton

(2004) elaborated the C5 Ratio, which is the ratio of the sum of points from the top five clubs to the total points within the league. Nonetheless, the index was adjusted by Michie and Oughton (2004) in order to compare tournaments with different numbers of

80

Empirical analysis of broadcast demand, competitive balance, demand for tickets and revenue generation in Brazilian football market participants, thereby creating the C4ICB. Both the HICB and C4ICB formulas are presented below:

∑푁 푠2 퐻퐼퐶퐵 = ( 𝑖=1 𝑖 ) ∗ 100 1⁄푁 where si is the total points of club i by the sum of points in the league, while N is the number of participants.

퐶4 푟푎푡𝑖표 퐶4퐼퐶퐵 = ( ) ∗ 100 4⁄푁 where C4 is the ratio of the sum of points of the top four clubs to the total points in the whole league, while N is the number of participants.

In both indices, the maximum balance is represented by 100. Thus, the further away from 100, the more unbalanced has been the league.

5.4. Results and Discussion

Table 11 below shows the ADP from the researched leagues.

Table 11. APD of the football leagues in Germany, Brazil, Spain, France, Netherlands, England, Italy, Portugal and Russia

Accumulated Points Difference (APD) Season Germany Brazil Spain France Netherlands England Italy Portugal Russia 2006 43,14 43,86 42,11 41,23 56,86 53,51 62,28 52,22 52,22 2007 46,08 52,63 51,75 48,25 44,12 66,67 48,25 58,89 47,78 2008 40,20 35,09 47,37 47,37 50,00 50,88 47,37 52,22 43,33 2009 45,10 31,58 57,02 48,25 69,61 58,77 46,49 61,11 58,89 2010 45,10 37,72 57,89 49,12 56,86 41,23 50,88 67,78 53,33 2011 56,86 35,09 73,68 42,11 55,88 56,14 54,39 62,22 45,56 2012 68,63 41,23 57,89 47,37 51,96 56,14 57,02 61,11 50,00 2013 63,73 49,12 57,02 57,89 41,18 49,12 67,54 55,56 48,89

APDaverage 51,10 40,79 55,59 47,70 53,31 54,06 54,28 58,89 50,00 Source: Self-elaboration.

81

Competitive Balance

The Brazilian League has the best average APD among the researched leagues, as it is the most competitive tournament within the sample. A graphical analysis is conducted by the box plot method and shows similar results, as can be seen in Figure 13.

Figure 13. Box plot of the APD in the domestic leagues in Germany, Brazil, Spain, France, Netherlands, England, Italy, Portugal and Russia

Source: Self-elaboration

The previous results provide evidence of the fact that the Brazilian League is the most competitive league within the sample. On the other hand, there are similar results among the European Leagues, apart from France and Russia, which have slight better levels of competitiveness. At the same time, two outliers are found in the Graph 1. In the

Spanish case, this represents the 2011-12 season, when the imbalance was remarkable: the champion won the tournament with 100 points, while the difference between the runner-up and the team in third place was 30 points. In the French case, the outlier

82

Empirical analysis of broadcast demand, competitive balance, demand for tickets and revenue generation in Brazilian football market represents the 2013-14 season, when Paris Saint-Germain won its second title in a row, confirming its sporting and financial supremacy in French football following the arrival of foreign investors at the club.

A strong fluctuation among the ADP within every league can be noted in both

Table 11 and Figure 13. This dynamic behaviour is presented below in Figure 14, using

Brazil, Germany and the Netherlands as examples.

Figure 14. Line chart of the APD in the Brazilian, German and Dutch Leagues

Source: Self-Elaboration

The dynamic behaviour of CB is evidenced in Figure 14. Although the Brazilian

League is the most competitive league among them, the imbalance has been increasing recently. Moreover, the German Bundesliga was somewhat balanced between 2006 and

2010, but the competitiveness has decreased intensely in recent years. On the other hand, the Dutch Eredivisie has been improving its CB of late.

83

Competitive Balance

The APD in all leagues is statistically compared as well. In the first step, the distribution of the data is examined using the Shapiro-Wilk test, while the homogeneity of variances is confirmed using the Levene test. The results are presented in Table 12 and

Table 13, respectively.

Table 12. Shapiro-Wilk test

Shapiro-Wilk

Statistic df Sig. APD Germany 0,856 8 0,110 Brazil 0,945 8 0,658 Spain 0,912 8 0,367 France 0,869 8 0,148 Netherlands 0,947 8 0,683 England 0,978 8 0,954 Italy 0,912 8 0,366 Portugal 0,935 8 0,561 Russia 0,977 8 0,947 Source: Self-elaboration.

Table 13. Levene test

Levene Test df1 df2 Sig. 1,303 8 63 0,258 Source: Self-elaboration.

All leagues have normal distribution, while the Levene test confirms the homogeneity of variances. Hence, these results allow us to carry out the one-way

ANOVA with the Tukey post hoc test, with the outputs presented in Table 14 and Table

15 below.

Table 14. Output from the one-way ANOVA

Sum of squares df Mean Square Z Sig. Between Groups 1759,184 8 219,898 3,801 0,001 Within Groups 3644,709 63 57,853 Total 5403,893 71 Source: Self-elaboration.

84

Empirical analysis of broadcast demand, competitive balance, demand for tickets and revenue generation in Brazilian football market

The significance (0.001) evidences statistical differences in competitiveness among the researched leagues. The outputs of the Tukey post hoc test are presented as follows.

Table 15. Tukey post hoc test

99% Confidential Interval (I) League (J) League Mean Difference (I-J) Std. Error Sig. Lower Bound Upper Bound Brazil Spain -14,80263*** 3,80304 0,007 -29,1684 -0,4368 England -13,26754** 3,80304 0,023 -27,6333 1,0982 Italy -13,48684** 3,80304 0,020 -27,8526 0,8789 France -6,90789 3,80304 0,671 -21,2737 7,4579 Germany -10,31347 3,80304 0,165 -24,6793 4,0523 Netherlands -12,51935** 3,80304 0,040 -26,8851 1,8464 Portugal -18,09942*** 3,80304 0,000 -32,4652 -3,7336 Russia -9,21053 3,80304 0,291 -23,5763 5,1553 Spain Brazil 14,80263*** 3,80304 0,007 ,4368 29,1684 England 1,53509 3,80304 1,000 -12,8307 15,9009 Italy 1,31579 3,80304 1,000 -13,0500 15,6816 France 7,89474 3,80304 0,499 -6,4711 22,2605 Germany 4,48916 3,80304 0,958 -9,8766 18,8550 Netherlands 2,28328 3,80304 1,000 -12,0825 16,6491 Portugal -3,29678 3,80304 0,994 -17,6626 11,0690 Russia 5,59211 3,80304 0,865 -8,7737 19,9579 England Brazil 13,26754** 3,80304 0,023 -1,0982 27,6333 Spain -1,53509 3,80304 1,000 -15,9009 12,8307 Italy -,21930 3,80304 1,000 -14,5851 14,1465 France 6,35965 3,80304 0,761 -8,0061 20,7254 Germany 2,95408 3,80304 0,997 -11,4117 17,3199 Netherlands ,74819 3,80304 1,000 -13,6176 15,1140 Portugal -4,83187 3,80304 0,936 -19,1977 9,5339 Russia 4,05702 3,80304 0,977 -10,3088 18,4228 Italy Brazil 13,48684** 3,80304 0,020 -,8789 27,8526 Spain -1,31579 3,80304 1,000 -15,6816 13,0500 England ,21930 3,80304 1,000 -14,1465 14,5851 France 6,57895 3,80304 0,726 -7,7868 20,9447 Germany 3,17337 3,80304 0,995 -11,1924 17,5392 Netherlands ,96749 3,80304 1,000 -13,3983 15,3333 Portugal -4,61257 3,80304 0,951 -18,9784 9,7532 Russia 4,27632 3,80304 0,968 -10,0895 18,6421 France Brazil 6,90789 3,80304 0,671 -7,4579 21,2737 Spain -7,89474 3,80304 0,499 -22,2605 6,4711 England -6,35965 3,80304 0,761 -20,7254 8,0061 Italy -6,57895 3,80304 0,726 -20,9447 7,7868 Germany -3,40557 3,80304 0,992 -17,7714 10,9602 Netherlands -5,61146 3,80304 0,862 -19,9772 8,7543 Portugal -11,19152* 3,80304 0,098 -25,5573 3,1743

85

Competitive Balance

Russia -2,30263 3,80304 1,000 -16,6684 12,0632 Germany Brazil 10,31347 3,80304 0,165 -4,0523 24,6793 Spain -4,48916 3,80304 0,958 -18,8550 9,8766 England -2,95408 3,80304 0,997 -17,3199 11,4117 Italy -3,17337 3,80304 0,995 -17,5392 11,1924 France 3,40557 3,80304 0,992 -10,9602 17,7714 Netherlands -2,20588 3,80304 1,000 -16,5717 12,1599 Portugal -7,78595 3,80304 0,518 -22,1517 6,5798 Russia 1,10294 3,80304 1,000 -13,2628 15,4687 Netherlands Brazil 12,51935** 3,80304 0,040 -1,8464 26,8851 Spain -2,28328 3,80304 1,000 -16,6491 12,0825 England -,74819 3,80304 1,000 -15,1140 13,6176 Italy -,96749 3,80304 1,000 -15,3333 13,3983 France 5,61146 3,80304 0,862 -8,7543 19,9772 Germany 2,20588 3,80304 1,000 -12,1599 16,5717 Portugal -5,58006 3,80304 0,866 -19,9459 8,7857 Russia 3,30882 3,80304 0,994 -11,0570 17,6746 Portugal Brazil 18,09942*** 3,80304 0,000 3,7336 32,4652 Spain 3,29678 3,80304 0,994 -11,0690 17,6626 England 4,83187 3,80304 0,936 -9,5339 19,1977 Italy 4,61257 3,80304 0,951 -9,7532 18,9784 France 11,19152* 3,80304 0,098 -3,1743 25,5573 Germany 7,78595 3,80304 0,518 -6,5798 22,1517 Netherlands 5,58006 3,80304 0,866 -8,7857 19,9459 Russia 8,88889 3,80304 0,336 -5,4769 23,2547 Russia Brazil 9,21053 3,80304 0,291 -5,1553 23,5763 Spain -5,59211 3,80304 0,865 -19,9579 8,7737 England -4,05702 3,80304 0,977 -18,4228 10,3088 Italy -4,27632 3,80304 0,968 -18,6421 10,0895 France 2,30263 3,80304 1,000 -12,0632 16,6684 Germany -1,10294 3,80304 1,000 -15,4687 13,2628 Netherlands -3,30882 3,80304 0,994 -17,6746 11,0570 Portugal -8,88889 3,80304 0,336 -23,2547 5,4769 (***) The mean difference is significant at the 0.01 level. (**) The mean difference is significant at the 0.05 level. (*)The mean difference is significant at the 0.10 level. Source: Self-elaboration.

The outputs presented in Table 15 confirm that the Brazilian League is the most competitive tournament within the sample. This tournament differs by 1% from the

Spanish La Liga and the Portuguese Primeira Liga, and 5% from the English, Italian and

Dutch leagues. However, its competitiveness is comparable to other domestic leagues in

Germany, France and Russia. The European tournaments indicate similar CB levels, with only the French Ligue 1 being slightly more competitive than the Portuguese Primeira

Liga (p < 0.10).

86

Empirical analysis of broadcast demand, competitive balance, demand for tickets and revenue generation in Brazilian football market

A complementary analysis is made in relation to the average position of all clubs that participated in those leagues between 2006/07 and 2013/14. The results are presented in Figure 15. As explained before, the clubs that had played on the lower tiers took the

N+1 position during those particular seasons. Notwithstanding, this graphical analysis aims to observe the higher positions (those closest to the first place) in order to examine the possible dominance of certain clubs in their respective domestic competitions.

Figure 15. Dispersion of the average position of the clubs from 2006/07 to 2013/14

Source: Self-elaboration.

Inferences cannot be made when using this graphical analysis approach.

Notwithstanding, the behaviour of the leagues and their clubs may be observed. The

Brazilian League has the closest ‘points’ within its graph, which can be interpreted to mean that there are more fluctuations in relation to the teams’ final positions. The Spanish

La Liga has two clubs that are much more superior than any of the others. There are also

87

Competitive Balance two ‘big groups’ in Europe, namely, the English Premier League and the Italian Serie A, in which there are seven teams that are regularly in the running to win the championship or qualify for the UEFA Champions League, for example. The German Bundesliga comprises a club that is totally dominant, while there are two others that, on occasion, have a chance to win the championship title. In the other domestic tournaments (France,

Netherland, Portugal and Russia), there are no big ‘groups’ of teams, but the presence of clubs whose average position is close to the top spot points to their dominance during the researched period. Table 16 shows the average position (AP) of those clubs that have won at least one title (T) between 2006/07 and 2013/14.

Table 16. Number of titles and average position between 2006/07 and 2013/14

Germany Brazil Spain Club T AP Club T AP Club T AP

Bayern M. 4 1,88 São Paulo 3 4,25 Barcelona 4 1,63 Borussia D. 2 4,88 Cruzeiro 1 6,25 R.Madrid 3 1,75 Stuttgart 1 7,63 Flamengo 1 8,13 Atlético 1 5,00 Wolfsburg 1 8,5 Fluminense 2 8,63 Corinthians 1 9,63

France Netherland England Club T AP Club T AP Club T AP

Lyon 2 2,75 Ajax 4 1,63 Man.Utd 5 2 Marseille 1 3,5 PSV 2 2,63 Chelsea 1 2,75 Lille 1 4,88 Twente 1 3,5 Man.City 2 5,63 Bordeaux 1 5,13 AZ 1 5,75 PSG 2 7,25 Montpellier 1 13,38

Italy Portugal Russia Club T AP Club T AP Club T AP

Inter 4 3,25 6 1,50 CSKA 3 2,13 Milan 1 3,63 Benfica 2 2,25 Zenit 3 2,38 Juventus 3 5,38 Rubin 2 5,25 Source: Self-elaboration.

88

Empirical analysis of broadcast demand, competitive balance, demand for tickets and revenue generation in Brazilian football market

Table 16 confirms the dominance of some clubs in their respective domestic leagues. When a club’s average position is close to 1 or 2, this means that the team was always in the running for the championship, either as the winner or quite close behind.

Thus, this may confirm the dominance of clubs such as FC Bayern München in Germany,

FC Barcelona and Real Madrid in Spain, AFC Ajax in the Netherlands, Manchester

United in England, and FC Porto in Portugal.

The findings of this chapter are in line with the results from Levy (2011), who compared the competitiveness in the Brazilian League and nine European domestic tournaments and found a more CB in Brazil between 1997/98 and 2010/11. The APD values also validate the statement from Drummond et al. (2010) that the Brazilian League has significant degree of competitiveness. Furthermore, the findings on European

Leagues are in agreement with several works, including those by Goossens (2006), Groot

(2008), Flores, Forrest and Tena (2010), Pawlowski et al. (2010), and Criado, García,

Pedroche and Romance (2013), while Montes, Sala-Garrido and Usai (2014) also confirmed a reduction in CB in recent seasons.

Notwithstanding, there is a difference between the findings of the present chapter and Criado et al. (2013), who examined competitiveness using complex networks and found an increase in CB in the German Bundesliga between 2011/12 and 2012/13.

However, an opposite result is found with the APD. If the champion wins the title with

25 points more than the runner-up, it may be difficult for this situation to be considered as a balanced tournament. However, the difference between findings could result from the measurement of sporting performance, as was the case when using the final points in the league table. Meanwhile, Criado et al. (2013) analysed the final position of the clubs and their interactions over the season.

89

Competitive Balance

5.4.1. Index Validation

The index elaborated in the present chapter is compared with two traditional measures (HICB and C4ICB) in order to validate it. Although the APD is based on maximum imbalance and the other two measures are based on maximum equilibrium, all of them have similar interpretations with regard to an increase/decrease in competitiveness. The Brazilian League has been chosen for examination, with the outputs presented in Table 17 below.

Table 17. HICB, C4ICB and the APD in the Brazilian League from the 2006 season to the 2013 season

Season HICB C4ICB APD 2006 105,21 133,27 43,86 2007 104,79 124,29 52,63 2008 104,39 133,62 35,09 2009 103,23 124,76 31,58 2010 105,34 132,97 37,72 2011 104,11 127,54 35,09 2012 105,94 138,16 41,23 2013 105,07 129,88 49,12 Source: Self-elaboration.

Similar results were expected for all indices as only one league is examined.

However, some differences are found. On the one hand, the 2007 season was the most competitive according to the C4ICB; on the other, the HICB shows a regular degree competitiveness. Meanwhile, the ADP indicates that 2007 was the most unbalanced season, which means that the APD may better reflect what happened in that season. The champion (São Paulo) stayed in first place over 27 rounds until it won the title, which constitutes a record during the researched period. Moreover, São Paulo is also the club that won the title with the most rounds in advance (34th round) prior to the design change to the double Round-Robin format (Cruzeiro also won the Brazilian League in 2013 in the 34th round). In addition, the difference between the champion and the runner-up was

90

Empirical analysis of broadcast demand, competitive balance, demand for tickets and revenue generation in Brazilian football market

15 points, which is sufficient for the tournament to be referred to as ‘balanced’.

Furthermore, the worst participant ever in the Brazilian League (since 2003) has played in the 2007 edition, obtaining a mere 17 points at the end of the season. In this sense, the

2007 edition should be considered as an unbalanced season as confirmed by the ADP.

A similar finding is observable for 2013. As mentioned previously, the champion

(Cruzeiro) won the title with some rounds in advance in the 34th round, while the second worst team following the introduction of the round-robin design played in this season

(i.e., a club that obtained 20 points in 38 rounds). Although the HICB and C4ICB show a moderate degree of competitiveness in that season, the outputs from the APD seem to be more appropriate.

The 2009 season was considered the most balanced according to both the APD and the HICB, which corroborates with the findings of Ribeiro and Urrutia (2012), who structured the Brazilian League schedule for the 2009 season based on the 2005 and 2006 editions. According to the authors, the consequence of their efforts was that it resulted in the most competitive Brazilian League, as four clubs were in the championship race in the last round.

5.4.2. Competitive Balance and Average Attendance

The findings of this chapter confirm that the Brazilian League is the most balanced tournament within the sample. Notwithstanding, the classical concept from Rottenberg

(1956) and Neale (1964), which states that competitiveness attracts fans, does not work for Brazilian football. The average attendance in the Brazilian League was around 15,893 in 2013, while unbalanced leagues in Italy, the Netherlands and Spain have reported better attendance levels: 23,385, 19,557 and 36,631, respectively10.

10 These data are available from http://www.worldfootball.net/. 91

Competitive Balance

On the one hand, Koenigstorfer, Groeppel-Klein and Kunkel (2010) note that CB is the most important determinant for attracting fans in both the English and German domestic leagues. However, they also highlight that clubs’ stadium ambience and international success increases interest among fans. On the other hand, several factors influence the demand for tickets, including a club’s image (Beccarini & Ferrand, 2006), the novelty effect (Coates & Humphreys, 2005), the presence of superstars (DeSchriver,

2007), ticket selling strategies (Howard & Crompton, 2004), sporting performance

(Theodorakis, Alexandris, Tsigilis & Karvounis, 2013), championship races (Pawlowski

& Nalbantis, 2015) and numerous other variables, as highlighted by Borland and

Macdonald (2003).

Thus, CB may remain as an important factor in order to attract fans. However, it does not work alone, given that attendance can be explained by a range of factors in combination. Therefore, although competitiveness offers significant advantage for teams in the Brazilian League, several other elements should be developed in order to increase live attendance rates at matches. In this sense, only proper demand-focused studies can help general managers and policymakers to enhance the Brazilian League.

5.4.3. Limitations and Further Research

The main limitation of the APD index is that it was elaborated for football leagues using a double Round-Robin design, where a win equals three points, a draw one point and a loss zero points. Thus, it should be adjusted in order to analyse other sports, as well as football leagues with different designs.

The present chapter has simply compared the APD with two other well-known indices: HICB and C4ICB. Although the method has been validated, the ADP may be confronted with other formulas, as well as be tested in relation to other football leagues.

92

Empirical analysis of broadcast demand, competitive balance, demand for tickets and revenue generation in Brazilian football market

The dynamic behaviour of CB is an interesting topic that is worthy of further development. Moreover, the relationship between competitiveness and league policies could also be researched, as proposed by Fort and Maxcy (2003). In this sense, a better understanding of the factors that affect the increase and decrease CB may help football managers to make positive adjustments to the league. Hence, better policies should be developed in order to maintain, or even enhance, fans’ interest in football tournaments.

The problems that Brazilian football has been experiencing are all too evident.

Several topics have been discussed, such as the financial and economic problems of clubs, the competition design of domestic tournaments, the maintenance of the States

Championships, the creation of the Primeira Liga, Brazil’s Financial Fair Play legislation and the need for more equal revenue distribution within the league. Therefore, in light of this chapter, further research into the Brazilian football industry is anticipated.

5.5. Final Remarks

The APD index, which is presented for the first time, is confirmed as a useful index for analysing football leagues incorporating a double round-robin design.

Moreover, the findings reveal the Brazilian League to be the most balanced tournament within the sample. On the other hand, the Spanish La Liga, the Dutch Eredivisie, the

English Premier League and the Portuguese Primeira Liga have shown lower levels of competitiveness compared to the Brazilian League. Notwithstanding, there are no statistical differences in the CB among the European leagues.

The present work does not seek to analyse the quality of the tournaments, the clubs or their players, only the degree of competitiveness within the researched competitions.

Although there are several other indices for analysing CB, the APD index has been developed to examine CB from a different perspective. Thus, this chapter has made both a methodological and a theoretical contribution.

93

Competitive Balance

94

Empirical analysis of broadcast demand, competitive balance, demand for tickets and revenue generation in Brazilian football market

6. THE DEMAND FOR TICKETS IN THE BRAZILIAN LEAGUE’S

FIRST DIVISION

6.1.Problem Statement

Chapters 4 and 5 have shown that the average attendance in the Brazilian League has never been significant. However, only a few scholars have shown any interest in research this topic, with Madalozzo and Villar (2009) having authored the only published paper about this issue to date. Moreover, a peculiar phenomenon can be observed in

Brazilian football: namely, that teams playing some matches in non-usual stadiums.

Between 2013 and 2015, 21 out of 26 clubs that took part in the First Division played at least one match in a non-usual stadium. Thus, 14.3% of all Brazilian League matches were played in different stadiums.

The term non-usual stadium cannot be interpreted as a neutral site. While neutral refers to a place where both teams are playing under theoretically equal competitive conditions, non-usual refers to stadiums where there are specific home and away clubs.

The FA Cup Final, habitually played at Wembley Stadium, is as an example of a neutral site, as the stadium has no club owner. Hence, that stadium is neutral for both teams during any FA Cup Final. However, a non-usual stadium could be assumed to be a place where a club has chosen to play due to some particular conditions.

A hypothetical situation can be illustrated as follows. Imagine that there are some problems with Old Trafford’s stands, which must be renovated over the next two months.

Thus, Manchester United, whose home is Old Trafford, needs to play on other field for the next four home matches. Nevertheless, the club managers have decided to play these matches at different stadiums: they will play against Arsenal at the Etihad Stadium; against Chelsea at the Anfield Stadium; and the other two matches at the DW Stadium.

95

The Demand for Tickets in the Brazilian League’s First Division

Although the owners of these stadiums are Manchester City, Liverpool and Wigan, respectively, Manchester United will be the home team in all those four matches.

Manchester United fans will have priority to buy tickets, while the club will sell its official merchandise, as well as carry out other pre-matching services. In this sense, those non- usual stadiums can be considered as Manchester United’s home stadium during those particular matches.

There are sundry reasons why some Brazilian clubs do not always play on the same field. Firstly, several clubs do not have their own stadium. For example, regarding the participants in the Brazilian League between 2013 and 2015, 38% of them rented stadiums. Moreover, other clubs are compelled to play in different venues if the owners of the scheduled stadiums need to use them for organizing other events.

Meanwhile, prior to Brazil hosting the FIFA World Cup in 2014, 12 stadiums were constructed or refurbished in order to meet the quality requirements of this mega-event.

Thus, in the years prior to this World Cup, some clubs had to play in different locations while their usual stadiums were under construction. On the other hand, some of these new stadiums were built in cities without any successful football clubs. As those stadiums are not in use throughout the season, some First Division clubs have decided to play certain matches there due to the quality of these new facilities.

The clubs have also been employing the following strategy: playing at high-level stadiums, usually selling expensive tickets and seeking to engage fans from other regions.

Notwithstanding, the clubs cannot always play every round in those stadiums because, in some cases, they are too far from their home cities, while other clubs may also be playing there. Thus, the clubs should consider playing in other stadiums across the country.

96

Empirical analysis of broadcast demand, competitive balance, demand for tickets and revenue generation in Brazilian football market

Considering these atypical characteristics, the aim of this paper is to analyse the effect of non-usual stadiums on the demand for tickets in the First Division of the

Brazilian League. The results of the present work may help football managers to decide where their clubs should play certain matches if they do not have their own stadium.

Furthermore, as it involves empirical demand-focused research, the findings could provide other practical implications in order to increase live match attendance rates, which in turn will increase revenue.

6.2. Football Stadiums in Brazil

According to the Cadastro Nacional de Estádios de Futebol (CNEF), or the

National Register of Football Stadiums, there are 789 football fields that are suitable to hold a professional match in Brazil. However, this may not be quite as it seems. The sample comprises high-level facilities, such as World Cup Arenas, as well as very small stadiums (for example, 36% of them have no floodlights). Thus, there are considerable differences in size among the stadiums in Brazil, as highlighted in Figure 16 below.

Although there are several stadiums in the country, only a small portion of them can be used for the First Division of the Brazilian League. Competition regulations require a minimum capacity of 15,000 seats, while only 110 stadiums (or 13.9% of the sample) meet this requirement. In any case, capacity does not always mean quality.

After the 2014 FIFA World Cup, the Brazilian Ministry of Sport developed the

Guia de Classificação dos Estádios (Guide to Stadium Classification)11. According to this guide, “the intention is not to create a ranking of football stadiums. We just want to

11 Available from http://www.esporte.gov.br/arquivos/snfut/Sisbrace/SISBRACE_LIVRETO.pdf 97

The Demand for Tickets in the Brazilian League’s First Division show the quality of each one and establish criteria to improve those that need improvement and to develop even more those that are already good”.

Figure 16. Number of Brazilian stadiums by total capacity

Number of Stadiums by Capacity

450 420 400

350

300

250 190 200

150

100 69 52 50 18 6 9 1 9 4 11 0 <5000 ≥ 5000 ≥ 10000 ≥ 15000 ≥ 20000 ≥ 25000 ≥ 30000 ≥ 35000 ≥ 40000 ≥ 45000 ≥ 50000 <10000 <15000 <20000 <25000 <30000 <35000 <40000 <45000 <50000

Source: Self-elaboratiom. Data from Cadastro Nacional de Estádios de Futebol12. N = 789.

The ministry evaluated 155 football stadiums in order to develop the guide. This represents around 19.65% of the total stadiums on the CNEF. The guide also consists of a technical report produced by experts in Security, Comfort, and Hygiene.

Some engineering aspects, such as structural issues, fire protection and evacuation plans, relate to security. Comfort is concerned with “thermal, acoustic, lighting, infrastructure and visual issues”, as well as accessibility for people with reduced mobility, including obesity, and the elderly. Parking, public areas and accessibility to the stands are considered as comfort issues as well. Hygiene is assessed according to the quality and cleanliness of food service areas, toilets and health facilities.

12 Available at http://cdn.cbf.com.br/content/201601/20160121152439_0.pdf. 98

Empirical analysis of broadcast demand, competitive balance, demand for tickets and revenue generation in Brazilian football market

Despite the fact that Brazil has hosted the World Cup and has a considerable number of stadiums, there are few first-class arenas, as confirmed by Figure 17. Indeed, only 10.33% of them have a four- or five- rating. In this sense, although several stadiums can be used for First Division matches (on account of having more than 15,000 seats), the majority are medium- or low-quality facilities.

Figure 17. Brazilian stadiums by quality

Stadiums by Quality 70 59 60 51 50

40 29 30

20 13 10 3 0 * ** *** **** *****

Source: Self-elaboration from Guia de Classificação dos Estádios. N = 155

As previously stated, 14.3% of all Brazilian League matches took place in a non- usual stadium between the 2013 and the 2015 seasons. Indeed, 63 different fields were used as non-usual stadiums. Figure 18 shows how many matches were played in every category of stadium. As evidenced by the graph below, clubs have decided to play in a variety of different places. Sometimes, teams play in a small stadium in their home city, while, in some situations, they may select a five-star arena. Unfortunately, there is no information to explain their choice of stadium in which to play.

99

The Demand for Tickets in the Brazilian League’s First Division

As this peculiar situation is a regular occurrence in Brazilian football, while there is no paper on this issue, the present research aims to analyse the effect that playing a non-usual stadium and its features may have on the demand for tickets and ticket prices.

Figure 18. Number of matches played at non-usual stadiums by quality

Number of matches at non-usual stadiums by quality

100 94 90 80 70 60 50 40 30 30 20 15 11 13 10 0 * ** *** **** *****

Source: Self-elaboration. N = 163.

6.3. Theoretical Background

The demand for tickets is a widely researched topic within sports economics. The articles from Borland and Macdonald (2003) and García and Rodríguez (2009) provide a thorough background to this issue. The attendance function is often modelled as a single equation, using inputs such as consumer preferences, stadium characteristics, uncertainty of outcome and economic factors. Indeed, most papers on sport demand have employed an econometric approach involving Ordinary Least Squares (OLS), as García and

Rodríguez (2009) observe. However, some studies have carried out their analysis by other

100

Empirical analysis of broadcast demand, competitive balance, demand for tickets and revenue generation in Brazilian football market estimation methods, such as Factor Analysis, Generalized Least Squares (GLS),

Instrumental Variables (IV), Tobit, Three-Stage Least Squares (3SLS) and Two-Stage

Least Squares (2SLS). Certainly, the model selection must take into account the distribution of the dependent variable, as well as its relationship with the explanatory factors.

An important subject related to sports stadiums and ticket demand is the so-called

“novelty effect” or “honeymoon effect”, which refers to the fact that interest among fans in attending matches increases after the construction of new sports arenas. The findings of Coates and Humphreys (2005), for example, observed a novelty effect for around nine seasons of the Major League Baseball (MLB) and the National Basketball Association

(NBA) as well as five years in the National Football League (NFL). Moreover, several other papers have confirmed similar behaviour, such as those by Quirk and Fort (1997),

Howard and Crompton (2003), Zygmont and Leadley (2005), and Feddersen, Maennig and Borcherding (2006).

Some papers have analysed the effects resulting from stadium characteristics.

Ulrich and Benkenstein (2010) found that several physical characteristics increase the stadium atmosphere, which in turn enhances the fans experience. Greenwell, Fink and

Pastore (2002) discovered that facilities’ attributes increase consumer satisfaction in minor league ice hockey. On the other hand, Boded and Bernarche-Assollant (2009) observed that some aspects are important to certain groups of supporters, but not for others. Furthermore, Watanabe, Matsumotoa and Nogawa (2013) confirmed that both hospitality and services management are key elements in the keeping spectators at a professional golf tournament.

101

The Demand for Tickets in the Brazilian League’s First Division

The effects of stadium features on the demand for tickets have been researched as well. Wakefield and Sloan (1995) have demonstrated that stadium services, such as parking, cleanliness and food services, positively affect fans’ desire to stay and attend further matches at the stadium. Moreover, Feddersen and Maennig (2009) have compared spectators’ preference in pure football arenas and multipurpose facilities, finding that monofunctional stadiums increase the demand for tickets by around 10% in the German

Bundesliga. In addition, DeSchriver, Rascher and Shapiro (2016) and Gómez-González et al. (2016) found a similar effect in (MLS).

Furthermore, Mirabile (2015) has analysed the demand for tickets at neutral sites in Division I-A college football, demonstrating that the determinants of attendance at neutral sites are in accordance with other traditional sport demand models. Thus, he confirmed the positive impact of the current and historical success of teams, the size of the clubs and their historical levels of attendance. Moreover, new stadiums and larger metropolitan areas could increase demand as well.

Madalozzo and Villar (2009) are the authors of the only published paper about the demand for tickets in the Brazilian League. Their data cover around 1,851 matches played between 2003 and 2006, while their econometric approach comprises two panel data linear regressions (OLS) with random and fixed effects (the Hausman statistic has indicated the Random Effects Model to be the more appropriate model). In the process, they found that the Nestlé promotion13 was the most powerful determinant in attracting fans, while derby matches, well-known visiting teams and current home team

13 The Nestlé promotion consisted of a free football ticket after purchasing a certain number of Nestlé products. 102

Empirical analysis of broadcast demand, competitive balance, demand for tickets and revenue generation in Brazilian football market performance also have a strong impact of increasing match attendance rates. Additionally, a reduction in the number of available tickets also increases demand.

Notwithstanding, some points should be reanalysed. An endogeneity problem between ticket prices and match attendance may exist, while all explanatory factors can influence price and attendance rates as well. Moreover, as there are no season tickets in the Brazilian context14, ticket prices may be strongly affected by match characteristics, with the willingness to attend a match possibly depending heavily on the price of admission. Furthermore, although the Hausman Test confirmed that the random effects model is the best regression model, fixed effects regarding the home team should be an assumption, as they take into account unobserved home team characteristics when estimating unbiased coefficients. Indeed, several recent papers has employed this approach. Therefore, the purpose of the present paper is to reanalyse the demand for tickets in the Brazilian League, while addressing the effect of non-usual stadiums and stadium features for the first time.

6.4. Methods

6.4.1. Data

The data set under analysis comprises all 1,140 Brazilian League matches in the

2013, 2014 and 2015 seasons. However, 11 games were excluded because they were played behind closed gates. There were no sell-out games in these three seasons, while the average attendance for the whole period was around 16,178 spectators. Nonetheless, interest in attending matches did increase: 14,951 spectators in 2013, 16,556 in 2014, and

14 There is no proper season ticket scheme in Brazil. Notwithstanding, a few clubs have socio programmes, which allows fans to attend a certain number of matches per season. 103

The Demand for Tickets in the Brazilian League’s First Division

17,045 in 2015. The highest attendance occurred in 2015 (67,011 attendees), while the lowest was in 2014 (766). The data were collected from the official website of the CBF.

6.4.2. Models and Variables

The econometric approach consists of seven panel data Three-Stage Least Squares

(3SLS) models using match-by-match information. This methodology has been chosen on the assumption of the endogeneity between ticket prices and the demand for live matches, as the 3SLS model avoids this. Furthermore, the model comprises the home team fixed effect estimator by taking into account unobserved club-specific effects. The general simultaneous equation model for any home club i and season t is presented below.

Ln(퐴푡푡푒푛푑푎푛푐푒) = 푓 (푀퐶 , 푈푂 , 푆퐶 ) { 𝑖푡 𝑖푡 𝑖푡 𝑖푡 푅푒푎푙 푇𝑖푐푘푒푡 푃푟𝑖푐푒𝑖푡 = 푓 (푀퐶𝑖푡, 푈푂𝑖푡, 푆퐶𝑖푡)

The first dependent variable is the logarithm for the number of attendees at each match. The original attendance is transformed due to the non-linear distribution of this variable. Hence, normalization allows for carrying out linear regressions. The second dependent variable is the real ticket price. The real price is measured by the average ticket price of each match by taking into account the Brazilian inflation rate every year.

Meanwhile, MC, UO and SC are the groups of explanatory variables related to match characteristics, uncertainty of outcome and stadium characteristics, respectively. Table

18 presents summary statistics for the variables employed in the empirical analysis.

104

Empirical analysis of broadcast demand, competitive balance, demand for tickets and revenue generation in Brazilian football market

Table 18. Summary statistics

Standard Variable Observations Mean Minimum Maximum Deviation Attendance 1129 16177.19 11556.73 766 67011 Real Ticket Price (R$) 1129 26.67704 14.64659 2.7928 128.5456 Match Quality 1140 0.446266 0.129208 0 1 Round 1140 19.5 10.97067 1 38 Weekdays 1140 0.282456 0.450392 0 1 Saturday 1140 0.22193 0.415727 0 1 Sunday 1140 0.495614 0.5002 0 1 Derby 1140 0.082456 0.275179 0 1 Home Win 1140 0.503804 0.128807 0.184502 0.854701 Probability Relative Stadium 1140 1.001151 0.385952 0.190264 3.349347 Capacity Non-usual Stadium 1140 0.142983 0.350209 0 1 Non-usual Stadium 1140 0.032456 0.177286 0 1 WC Year Construction 1140 46.09912 27.54423 0 99 Gen. Classification 1140 3.750877 1.084348 1 5 Security Level 1140 3.945614 1.054539 1 5 Comfort Level 1140 3.185965 0.835229 1 5 Hygiene Level 1140 3.449123 1.420423 1 5 Source: Self-elaboration.

6.4.2.1. Match Characteristics

The Match Quality variable is measured by the sum of both teams’ points before the match and relativizing it by the maximum possible points in each round. Buraimo and

Simmons (2015) have previously used this variable.

Pawlowski and Nalbantis (2015) have verified the levels of interest among fans at the beginning and end of championships. Hence, Round and Round variables2 were employed in order to capture the quadratic behaviour in the preferences of Brazilian supporters.

105

The Demand for Tickets in the Brazilian League’s First Division

Fan preference related to the day of the week is explored by dummy variables of matches played on Saturdays and Weekdays (matches on Sundays are used as a base).

Madalozzo and Villar (2009) have reported that higher attendances occur on weekends.

However, this phenomenon can also reflect the possible difference in the demand for tickets and ticket prices on Saturdays and Sundays.

A Derby dummy variable is employed, which takes ‘1’ when the match is a derby and ‘0’ otherwise. This kind of match is often more crowded, as evidenced by Buraimo and Simmons (2008), Madalozzo and Villar (2009), and Martins and Cró (2016).

6.4.2.2. Uncertainty of Outcome

Rottenberg (1956) has remarked that balanced games lead to higher attendances –

Uncertainty of Outcome Hypothesis (UOH). Notwithstanding, recent papers have been rejecting this theoretical claim. The findings reported by Buraimo and Simmons (2008) suggest that fans prefer to attend a match against a weaker team than an uncertainty game.

Buraimo and Simmons (2015) also showed that television spectators are more interested in the quality of the players than in the UOH. Coates, Humphreys and Zhou (2014), meanwhile, observed loss aversion behaviour in the MLB, with Humphreys and Zhou

(2015) corroborating this finding, while also identifying home win preference behaviour among MLB fans. Thus, Home Win Probability and its squared value are employed in order to analyse the Uncertainty of Outcome based on the works of Coates, Humphreys and Zhou (2014) and Humphreys and Zhou (2015). This method can show how short- term CB affects the demand for tickets in Brazilian football. Home win probabilities are derived from the betting odds, while the data are collected from the website

OddsPortal.com.

106

Empirical analysis of broadcast demand, competitive balance, demand for tickets and revenue generation in Brazilian football market

6.4.2.3. Stadium Characteristics

All the equation systems analyse the impact of non-usual stadiums in the demand for tickets. A dummy variable takes the value ‘1’ when a match was played in a non-usual stadium, and ‘0’ otherwise. The first model assesses its impact separately, while the dummy variables for non-usual stadiums are the specifics of each team.

In the subsequent models, different features related to non-usual stadiums are included. Model 2 includes the dummy variable of World Cup Arenas when the clubs played a match at one of these kinds of non-usual stadium. The World Cup arenas dummy considers the football fields used in that mega-event.

Model 3 includes the Year of stadium Construction and its squared value in order to determine whether fans are interested in attending matches at traditional or new stadiums. Model 4 evaluates the general classification of the stadium as established by the Brazilian Ministry of Sport. The classification of those stadiums represents a categorical variable from one to five stars. In this sense, each category takes a specific dummy in order to check their effects on match attendance and ticket price.

Models 5, 6 and 7 substitute the general classification variables for a group of dummy variables related to security, comfort and hygiene levels, respectively. These variables can determine how each aspect affects the demand for tickets and ticket prices.

Once the clubs have played some matches at non-usual stadiums, the relative stadium capacity is employed in the same way as a control variable. This variable is calculated by the ratio of the total capacity of the used stadium in a match to the total capacity of the traditional arena (it takes ‘1’ when the club plays in its traditional stadium).

As average attendance and occupancy rates are very low in the First Division of the

107

The Demand for Tickets in the Brazilian League’s First Division

Brazilian League, the clubs may choose to play in certain non-usual stadiums, based on the expected demand for each game.

6.5. Results and Discussion

The outputs of the seven models are shown in Table 19 and Table 20. The regressions explain around 58% of the demand for tickets as well as the real ticket price.

The results are consistent, as there are only slight variations between them. Most of the variables show significant and robust effects.

6.5.1. Match Characteristics

Match quality has a strong impact in terms of attracting supporters to a stadium, while strong performances by both teams can generate high expectations about a match, which in turn lead to better attendance rates. Czarnitzki and Stadtmann (2002), Pawlowski and Nalbantis (2015), and Reilly (2015) confirm the importance of the recent performance of teams to increasing the demand for tickets. At the same time, more high-profile matches can lead to increases in ticket prices as well. This result is understandable, given that clubs usually sell expensive tickets for important matches. Moreover, as only a few

Brazilian clubs have “season tickets”15, they will need to adjust prices even more according to the attractiveness of the match.

15 This is the socio programme referred to in footnote 2. 108

Empirical analysis of broadcast demand, competitive balance, demand for tickets and revenue generation in Brazilian football market

Table 19. 3SLS regressions regarding attendance and ticket prices (part 1)

(General) (General) (WorldC) (WorldC) (Age) (Age) (Class) (Class) VARIABLES Attendance Price Attendance Price Attendance Price Attendance Price

Match Quality 0.593*** 7.374*** 0.632*** 9.020*** 0.588*** 7.007*** 0.532*** 5.911** (0.138) (2.748) (0.136) (2.477) (0.138) (2.541) (0.135) (2.512) Round 0.0147** -0.0860 0.0130** -0.159 0.0144** -0.105 0.0135** -0.123 (0.00622) (0.123) (0.00609) (0.111) (0.00619) (0.114) (0.00607) (0.113) Round2 -0.000268* 0.000103 -0.000229 0.00176 -0.000269* 2.96e-05 -0.000255* 0.000256 (0.000152) (0.00302) (0.000149) (0.00272) (0.000151) (0.00279) (0.000148) (0.00276) Weekdays -0.190*** -2.067*** -0.192*** -2.127*** -0.189*** -2.004*** -0.197*** -2.398*** (0.0376) (0.746) (0.0368) (0.672) (0.0374) (0.690) (0.0366) (0.681) Saturday -0.125*** -1.402* -0.137*** -1.917*** -0.123*** -1.280* -0.133*** -1.606** (0.0407) (0.808) (0.0399) (0.728) (0.0405) (0.748) (0.0397) (0.738) Derby 0.294*** 6.075*** 0.254*** 4.397*** 0.280*** 4.982*** 0.252*** 4.521*** (0.0583) (1.158) (0.0574) (1.048) (0.0582) (1.074) (0.0571) (1.062) Home Win Probability -2.954*** -125.4*** -2.404*** -102.3*** -2.735*** -108.0*** -2.435*** -106.1*** (0.859) (17.04) (0.844) (15.42) (0.858) (15.82) (0.840) (15.62) Home Win Probability2 2.741*** 107.3*** 2.290*** 88.37*** 2.543*** 91.66*** 2.232*** 89.21*** (0.828) (16.43) (0.813) (14.85) (0.827) (15.25) (0.810) (15.06) Relative Stadium 0.613*** 7.435*** 0.499*** 2.653*** 0.609*** 7.175*** 0.292*** -4.996*** Capacity (0.0488) (0.969) (0.0504) (0.921) (0.0561) (1.035) (0.0651) (1.210) Non-Usual Stadium -0.383*** 4.544*** -0.587*** -4.003*** -0.397*** 3.558*** -0.213*** 10.44*** (0.0518) (1.028) (0.0583) (1.066) (0.0584) (1.078) (0.0580) (1.078) Non-Usual Stadium 0.754*** 31.67*** WC (0.107) (1.954) Year Construction -0.226 -18.44*** (0.182) (3.354) Year Construction2 5.79e-05 0.00472*** (4.60e-05) (0.000849) 2.Gen Classification 0.0550 15.22*** (0.185) (3.437) 3.Gen Classification 0.0209 9.418*** (0.167) (3.099) 4.Gen Classification 0.211 26.14*** (0.209) (3.887) 5.Gen Classification 0.621*** 31.20*** (0.190) (3.533)

Constant 9.245*** 52.56*** 9.238*** 52.28*** 229.9 18,059*** 9.513*** 54.09*** (0.262) (5.209) (0.257) (4.691) (179.6) (3,312) (0.308) (5.722)

Season FE Yes Yes Yes Yes Yes Yes Yes Yes Home Club FE Yes Yes Yes Yes Yes Yes Yes Yes Observations 1,129 1,129 1,129 1,129 1,129 1,129 1,129 1,129 R-squared 0.552 0.512 0.571 0.604 0.556 0.583 0.575 0.595 Standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1

109

The Demand for Tickets in the Brazilian League’s First Division

Table 20. 3SLS regressions regarding attendance and ticket prices (part 2)

(Security) (Security) (Comfort) (Comfort) (Hygiene) (Hygiene) VARIABLES Attendance Price Attendance Price Attendance Price

Match Quality 0.587*** 6.960*** 0.553*** 5.720** 0.578*** 6.941*** (0.135) (2.531) (0.133) (2.502) (0.135) (2.490) Round 0.0136** -0.143 0.0119** -0.135 0.0119* -0.140 (0.00608) (0.114) (0.00601) (0.113) (0.00609) (0.112) Round2 -0.000258* 0.000831 -0.000223 0.000586 -0.000221 0.000914 (0.000148) (0.00277) (0.000147) (0.00275) (0.000149) (0.00273) Weekdays -0.202*** -2.420*** -0.199*** -2.437*** -0.196*** -2.240*** (0.0367) (0.685) (0.0363) (0.679) (0.0366) (0.673) Saturday -0.127*** -1.597** -0.138*** -1.511** -0.128*** -1.595** (0.0397) (0.742) (0.0393) (0.736) (0.0397) (0.730) Derby 0.268*** 4.964*** 0.242*** 4.253*** 0.256*** 4.641*** (0.0571) (1.066) (0.0565) (1.060) (0.0571) (1.050) Home Win Probability -2.558*** -108.7*** -2.327*** -104.7*** -2.541*** -111.5*** (0.839) (15.68) (0.832) (15.60) (0.839) (15.43) Home Win Probability2 2.377*** 91.54*** 2.174*** 87.53*** 2.316*** 92.76*** (0.810) (15.13) (0.802) (15.02) (0.810) (14.89) Relative Stadium Capacity 0.403*** -1.098 0.295*** -3.670*** 0.430*** 2.507** (0.0685) (1.281) (0.0614) (1.150) (0.0653) (1.201) Non-Usual Stadium -0.230*** 11.00*** -0.201*** 8.747*** -0.348*** 4.906*** (0.0577) (1.078) (0.0550) (1.030) (0.0628) Non-Usual Stadium WC

2.Security -0.345* 1.408 (0.183) (3.421) 3.Security -0.0570 -2.318 (0.179) (3.362) 4.Security 0.442** 19.17*** (0.186) (3.491) 5.Security 0.856 34.63*** (0.532) (9.971) 2.Comfort -0.135 3.895* (0.122) (2.235) 3.Comfort -0.105 0.647 (0.110) (2.026) 4.Comfort 0.421*** 19.61*** (0.111) (2.038) 5.Comfort 0.231* 7.894*** (0.132) (2.424) 2.Hygiene -0.345* 3.895* (0.183) (2.235) 3.Hygiene -0.0570 0.647 (0.179) (2.026) 4.Hygiene 0.442** 19.61*** (0.186) (2.038) 5.Hygiene 0.856 7.894*** (0.532) (2.424)

Constant 9.251*** 44.83*** 9.579*** 64.09*** 9.417*** 56.43*** (0.319) (5.967) (0.321) (6.023) (0.260) (4.775)

Season FE Yes Yes Yes Yes Yes Yes Home Club FE Yes Yes Yes Yes Yes Yes Observations 1,129 1,129 1,129 1,129 1,129 1,129 R-squared 0.575 0.590 0.584 0.597 0.575 0.603 Standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1

110

Empirical analysis of broadcast demand, competitive balance, demand for tickets and revenue generation in Brazilian football market

The fixtures variables, Round and Round2, involve inverted U-shaped behaviour.

This result is in opposition to Pawlowski and Nalbantis’ (2015) findings. The peculiarities of Brazilian football may explain why this is the case. In Brazil, season begins in January and ends in December, while the Brazilian League only starts in May. In the first months of the year, clubs play in the State Championships, while the best of them also appear in the most important South American tournament, the Copa Libertadores. Thus, the expectations of fans with regard to their teams at the beginning of the League do not exist.

In addition, interest may be lower if a club does not play well in the aforementioned tournaments or if it is playing in the decisive stages of the Copa Libertadores. Meanwhile, the final rounds may have a negative effect, given that, in the 2013, 2014 and 2015 seasons, the champion won the league in the 34th, 36th and 35th rounds, respectively. Thus, this kind of scenario could reduce the interest levels among supporters at the end of a championship.

Sunday matches experience higher attendance rates as expected. Madalozzo and

Villar (2009) previously reported that weekdays reduce the demand for tickets in the

Brazilian League. However, the current findings shown that there is limited demand for tickets to matches on Saturdays as well. At the same time, it should be noted that ticket prices are lower on weekdays and Saturdays.

The derby dummy has a robust and strong effect, which increases attendance rates and ticket price. The rivalry between the clubs may explain the higher levels of interest among fans, which in turn presents a great opportunity to clubs to sell expensive tickets.

Buraimo and Simmons (2008) and Martins and Cró (2016) have found similar results in terms of the demand for tickets to English Premier League and the Portuguese Primeira

Liga, respectively. Moreover, even broadcast demand is higher for derby matches, as evidenced by Forrest, Simmons and Buraimo (2005) and Buraimo and Simmons (2015). 111

The Demand for Tickets in the Brazilian League’s First Division

6.5.2. Uncertainty of Outcome

The findings reject the classic Uncertainty of Outcome Hypothesis, which is in line with recent papers from Coates, Humphreys and Zhou (2014), Cox (2015),

Humphreys and Zhou (2015), and Jena and Reilly (2016). The significant quadratic behaviour of home win probability in all models confirms that more certain matches increase the demand for tickets to Brazilian football. The tipping point of home win probability is around 0.46. Specifically, this means that supporters are highly interest in their home team winning as well as playing against strong opponents. Similar behaviour is found in relation to ticket price, which may reflect clubs’ decision to charge higher prices for specific matches, whether against weak or strong teams, based on the level of interest among fans for these games.

6.5.3. Stadiums Characteristics

The most important goal of the present paper is to understand the effects of non- usual stadiums on the demand for tickets. The seven models confirm that matches played in different stadiums result in lower attendance rates. However, the second regression suggests an offset opportunity. When clubs play at World Cup arenas, as non-usual stadiums, attendances are higher, which equipoises the negative impact. On the other hand, ticket prices are higher for non-usual stadiums, which may be explained by the fact that matches at non-usual stadiums tend to be uncommon.

The possible novelty effect of playing in a World Cup stadium is checked after showing an increasing in the demand for tickets. Nonetheless, the third model indicates that the age of the stadium does not matter to supporters. Both variables have no statistical effect. Hence, as the World Cup arenas are known for their high quality facilities16, the

16 During the World Cup, people in Brazil used the expression “Padrão FIFA” (“In FIFA style”) to refer to high-quality products. 112

Empirical analysis of broadcast demand, competitive balance, demand for tickets and revenue generation in Brazilian football market subsequent models seek to analyse whether this is an important factor when it comes to attracting fans. The fourth model confirms that not only World Cup arenas, but also other good stadiums (those with five stars) can increase the demand for tickets. Therefore, it can be stated that fans consider the general quality of the football ground when deciding whether to attend a match. On the other hand, ticket prices are affected by every improvement in the stadium, given that the entrance fee at two-, three-, four- and five- star stadiums is more expensive than at one-star stadium.

The fifth model confirms that only five star for security increases the demand for tickets. On the other hand, all categories reflect higher ticket prices. The sixth model shows that two stars for comfort reduces the attendance, while four stars increase demand.

However, there is no impact on ticket demand with five stars. This reduction may be the result of higher prices and minor differences in security from one to two stars. Moreover, the non-effect on the demand for five-star arenas may be due to the significant hike in ticket prices when teams play at these venues. Although security matters to fans (as evidenced in respect of four-star stadiums), they may not be willing to pay higher ticket prices to attend games at a five-star arena. Meanwhile, four and five stars for hygiene do increase attendance rates and ticket prices.

A negative effect of non-usual stadiums is plausible. Although the biggest

Brazilian clubs have supporters from all over the country, loyalty among remote supporters may be different to that among fans from their home cities. Playing once in a distant town might attract some attendees, but playing there on several occasions could deplete demand. Clearly, in the researched period, most matches in non-usual stadiums were not played out of choice by the clubs but because some of them did not have their own arena, while other clubs were constructing or reforming their stadium. Thus, they may have expected a reduction in attendance. 113

The Demand for Tickets in the Brazilian League’s First Division

Nevertheless, the results may involve some practical implications. Clubs must consider positive factors in order to maximize the demand for tickets to their matches in non-usual stadiums. It is essential to take into account the quality of the arena when considering whether it will attract more fans and offset the negative impact of play away from the home venue. Moreover, if possible, they should play their more popular fixtures, derby games or Sunday matches in non-usual stadiums in order to increase the demand for tickets there. Notwithstanding, the most important strategy that clubs need to pursue is to construct, refurbish or obtain concessions with better conditions to allow them to play the entire Brazilian League at the same stadium.

6.5.4. Limitations and Further Research

The main aim of the present paper was to analyse the effects of non-usual stadiums and quality aspects on the demand for tickets and ticket prices. Nevertheless, some other factors might have been missed. Future research could explore whether playing at non- usual stadiums reduces the home advantage, increases the sales of merchandise in that region and brings about changes in the number of socios, as well as whether or not the behaviour of their fans on the social networks indicates an endorsement of the strategy to play in other cities. Some managerial decisions could also minimize the negative effects of playing in non-usual stadiums in Brazil.

On the other hand, in the European football context, it is observable that, when the bigger teams are playing national cup matches in small cities, the consequence is often sell-out fixtures. Clubs such as FC Barcelona, FC Bayern München, Manchester United and Real Madrid always attract fans due to their star players. Hence, a study of the impact of these kinds of club playing a small number of matches at non-usual stadiums could be interesting.

114

Empirical analysis of broadcast demand, competitive balance, demand for tickets and revenue generation in Brazilian football market

6.5.5. Practical Implications

The findings of the present chapter suggest some practical implications to football managers in Brazil. Firstly, it has been robustly proven that playing in non-usual stadiums reduces the demand for tickets. In this sense, the most important decision that clubs should make is to effectively construct or renegotiate current contracts with their stadium’s owners.

Obviously, the constructions of new stadiums can take a long time before matches can be played in them. Therefore, within this construction period, football managers ought to take into account the quality of the opposing team, the current performance of both clubs, the day of the week, and the quality aspects of the stadium in order to control the negative effects when playing on a non-usual field. Thus, through a better understanding of how each factor affects the demand for tickets, teams may be able to choose the best stadium in which to play in every situation.

The outputs also demonstrated that match attendance rates on weekdays are statistically lower than on weekends, a finding that is in line with Madalozzo and Villar

(2009). In the Brazilian League, there are often two rounds per week; in other words, clubs play a match at the weekend and another on a weekday. Moreover, as a result of its atypical sporting calendar, the Brazilian League does not stop its matches on FIFA dates.

Hence, the clubs that have players on international duty must play those games without their best players. Thus, the CBF needs to fully reorganize fixtures by holding only one

Brazilian League match each week and on weekends.

The average attendance in the Brazilian League is undoubtedly small when taking certain factors into account, such as the international success of the Brazilian national team, the sporting performance of Brazilian clubs in South American championships, and

115

The Demand for Tickets in the Brazilian League’s First Division the country’s population. Therefore, several policies should be adopted in order to improve the organization of the Brazilian League.

6.6. Final Remarks

This is the first paper that aims to research the effects of non-usual stadiums on the demand for tickets. Meanwhile, the determinants of attendance in the First Division of the Brazilian League were reanalysed, while stadium features were examined for the first time. The determinants of ticket prices have also never been researched before in the

Brazilian football context.

The results confirm the positive effect of match quality and derby games on attracting attendees to games. The outputs also show that matches on Sundays increase demand. In contrast to European fans, supporters in Brazil have little interest in matches in the first and final rounds of the tournament, which may be the result of the peculiar sports calendar in the country. The classic UOH is rejected, as it has been in other recent papers. This means that Brazilian fans prefer more certain matches against weak or strong opponents. Moreover, the effects related to non-usual stadiums represent the most important finding. The outputs of the seven regressions confirmed that such stadiums reduce the demand for tickets. However, good-quality stadiums can offset this effect and increase the number of attendees. Furthermore, the strategy to sell expensive tickets at non-usual stadiums is evidenced.

Some practical implications are suggested, such as the construction of new stadiums by Brazilian clubs or, at least, better contract renegotiations with stadium owners. However, the quality of stadium facilities is a key factor when attracting fans, which should be taken into account. In addition, changing the fixture schedule in the

Campeonato Brasileiro, where matches take place on weekends, is also suggested.

116

Empirical analysis of broadcast demand, competitive balance, demand for tickets and revenue generation in Brazilian football market

Therefore, if the clubs need to play at a non-usual stadium, they should bear in mind the quality aspects of the stadium, along with other factors that could attract fans.

117

The Demand for Tickets in the Brazilian League’s First Division

118

Empirical analysis of broadcast demand, competitive balance, demand for tickets and revenue generation in Brazilian football market

7. DIFFERENCES IN THE DEMAND FOR TICKETS IN EACH

DIVISION OF THE BRAZILIAN LEAGUE

7.1.Problem Statement

The reduced demand for tickets is not specific to the First Division of the Brazilian

League. Indeed, all tiers report lower average attendances, as shown in Figure 19 below.

Figure 19. First, Second, Third and Fourth Divisions of the Brazilian League: seasonal average attendance 2013-2015

18000 16000 14000 12000 10000 8000 6000 4000 2000 0 First Second Third Fourth

2013 2014 2015

Source: Self-elaboration from Confederação Brasileira de Futebol.

As mentioned above, the only published paper on the demand for tickets in

Brazilian football is by Madalozzo and Villar (2009). Although the previous chapter reanalysed this study and provided some new findings, both works simply examine the demand for tickets at the top flight of Brazilian football. Given that certain factors may work differently at various levels, the aim of this chapter is to analyse the demand for tickets in all Brazilian League tiers for the first time.

119

Differences in the demand for tickets in each division of the Brazilian League

7.2. Theoretical Background

Most of the literature on the demand for tickets in professional football is focused on the main leagues, such as the domestic top tiers in Europe. On the other hand, in the

North American context, some papers have been published on minor or college leagues, such as those by Siegfried and Eisenerg (1980), Kaempfer and Pacey (1986), Fizel and

Bennett (1989), Branvold and Bowers (1992), Price and Sen (2003), Paul, Humphreys and Weinbach (2012), and Mirabile (2015). Nevertheless, only two papers can be found on the determinants of attendances at lower divisions in professional football: Buraimo

(2008) and Jena and Reilly (2016).

In the first paper, Buraimo (2008) examined both ticket and broadcast demand in the second tier of English football, finding that some factors have a similar impact on both types of demand, such as players’ talent and derby matches. Nonetheless, he confirmed that broadcast matches could reduce live attendances at this competitive level.

In Jena and Reilly’s (2006) work, the UOH was tested in the context of the Irish second tier. As a result, their findings confirm the positive effect resulting from the quality of home team performance and fixtures, as well as a negative impact resulting from the distance between clubs. However, the most important goal involves an inverted U-shaped form with regard to the uncertainty of outcome, which is in line with the classical theory from Rottenberg (1956) and Neale (1964). Meanwhile, the authors found that this outcome was in opposition to that observed in the Irish Premier League. Thus, in their estimation, the preferences of fans in the top tier relate to their home teams, while, in the second tier, fans are more interested in balanced contests.

120

Empirical analysis of broadcast demand, competitive balance, demand for tickets and revenue generation in Brazilian football market

As there is no paper on the determinants of attendance in the lower tiers of the

Brazilian League, this chapter aims to examine the phenomenon and identify any differing factors among them.

7.3. Methods

The data set comprises all 302 football clubs that have participated at least once in one of the four Brazilian League divisions during the 2012, 2013 and 2014 seasons.

The data were collected from the official website of the CBF.

7.3.1. Model and Variables

The econometric approach consists of panel data OLS regression for every division d, region r and season t. The dependent variable is the logarithm of the seasonal average attendance of every club. Although the previous chapter analysed the demand, as observed in match-by-match data, seasonal average attendance of the clubs has been chosen due to data availability on the lower tiers. Papers such as those by Demmert

(1973), Scully (1989), Dobson and Goddard (1996), Szymanski and Smith (1997), and

Coates and Humphreys (2007) have used similar measures with regard to their dependent variable. The general model is as follows:

ln(푎푣푒푟푎푔푒 푎푡푡푒푛푑푎푛푐푒)푑푟푡

= 훽0 + 훽1푎푣푒푟푎푔푒 푡𝑖푐푘푒푡 푝푟𝑖푐푒푑푟푡 + 훽2푐푢푟푟푒푛푡 푝푒푟푓표푟푚푎푛푐푒푑푟푡

+ 훽3ℎ𝑖푠푡표푟𝑖푐푎푙 푠푢푐푐푒푠푠푑푟푡 + 훽4푝표푝푢푙푎푡𝑖표푛 (푡표푡푎푙, 푟푒푙푎푡𝑖푣푒)푑푟푡

+ 훽5퐻퐷퐼 (표푟𝑖푔𝑖푛푎푙, 푟푒푙푎푡𝑖푣푒)푑푟푡 + 훽6푛푢푚푏푒푟 표푓 푐푙푢푏푠푑푟푡

Six regressions are estimated. The general model presented above is carried out in the first two regressions. In the first, population and HDI variables are taken by their original values, although population is determined by the logarithm transformation.

121

Differences in the demand for tickets in each division of the Brazilian League

However, in the second regression, they take relative values. The four subsequent models analyse every Brazilian League division separately.

The first explanatory variable is Average Tickets Price, which is calculated by dividing match day revenues by average attendance. The second explanatory variable is a proxy of current performance, which is measured using the ranking developed by

Szymanski and Kuypers (1999). The formula, as presented in Chapter 3, is reproduced here:

푅푎푛푘𝑖푛푔 = − log (푝/(푛 + 1 − 푝)) where p is the final position of each club and n is the number of participants in all tiers.

In this case, n takes the value of 100 for 2012 and 101 for 2013 and 2014. As noted in

Chapter 3, previous studies have used this formula, such as those by Szymanski and

Kuypers (1999), Barajas, Fernández-Jardón and Crolley (2005), Kuper and Szymanski

(2010), Szymanski (2015), and Gasparetto and Barajas (2016).

Historical success is the third explanatory variable, which is calculated using an index developed for the first time in this work. The index considers all titles in the First,

Second, Third and Fourth Divisions in the Brazilian League won by the clubs, while employing specific weights for each tier. The differences in total match day revenues among the tiers over the sample (2012-14) are used to create the weights (Figure 20).

Indeed, the differences are similar to those observed by Anderson and Sally (2013) with regard to salaries among various categories in English football. The formula is as follows:

퐻𝑖푠푡표푟𝑖푐푎푙 푆푢푐푐푒푠푠 퐼푛푑푒푥

= (푇𝑖푡푙푒1푠푡 ∗ 0.7686) + (푇𝑖푡푙푒2푛푑 ∗ 0.1458) + (푇𝑖푡푙푒3푟푑 ∗ 0.0639)

+ (푇𝑖푡푙푒4푡ℎ ∗ 0.0217)

122

Empirical analysis of broadcast demand, competitive balance, demand for tickets and revenue generation in Brazilian football market

Figure 20. Match day revenues (R$): all Brazilian League tiers (2012, 2013 and 2014)

700,000,000.00 0.9000

0.8000 600,000,000.00 0.7686 0.7000 500,000,000.00 0.6000

400,000,000.00 0.5000

300,000,000.00 0.4000

0.3000 200,000,000.00 0.2000 100,000,000.00 0.1458 0.1000 0.0639 0.0217 0.00 0.0000 First Second Third Fourth

2012 2013 2014 Percentage

Source: Self-elaboration from Confederação Brasileira de Futebol.

The other explanatory variables are related to the cities where the clubs are located, namely, the city’s population, the city’s Human Development Index (HDI), and the number of clubs from the same city playing in one of the Brazilian League tiers. In the first regressions, the city population is taken by its logarithm and HDI of its original value. Notwithstanding, Brazil is a large country with huge socio-economic inequality.

Thus, both population and HDI are taken by their relative values in the second regression, which may capture some effects relevant to bigger cities and higher standards of living.

Mourão (2010) found that, in the European football context, clubs located in highly urbanized areas with high levels of Gross Domestic Product (GDP) display superior sporting performance. Thus, this variable seeks to determine whether it has an effect on the demand for tickets as well. The number of clubs in the same city is examined, as more

123

Differences in the demand for tickets in each division of the Brazilian League clubs in the same city could reduce local demand. The data regarding population are available from the IBGE Cidades website17, HDI from the Atlas do Desenvolvimento

Humano do Brasil website18, and the number of clubs per city from the Brazilian League tables.

The demand for tickets was not controlled by stadium capacity, as these data are not wholly available. Notwithstanding, this may not be a problem because average attendance does not exceed 40% of the total capacity in the First Division, while it is even smaller in lower divisions.

7.4. Results and Discussion

The outputs of the six regressions are presented in Table 21 below. The general models explain around 70% of the demand for tickets in the entire Brazilian League. The second model is better than the first, based on its R2. The specific models explain around

40% (Fourth Division) and 70% (Second and Third Divisions). The VIFs reject the multicollinearity among the variables.

Ticket price has no statistical effect in all models. While this may seems unusual,

Borland and Macdonald (2003) have noted that some papers have found similar results.

Moreover, this may also be a peculiarity of South American football. As ticket prices are inexpensive and their variations are minor, other factors may influence more in the willingness to attend a game. Notwithstanding, this phenomenon could be reanalysed using other econometric approaches, such as equations systems involving a Three-Stage

Least Squares estimator and Instrumental Variables as used by García and Rodríguez

(2002).

17 Available at http://www.cidades.ibge.gov.br/. 18 Available from http://www.atlasbrasil.org.br/2013. 124

Empirical analysis of broadcast demand, competitive balance, demand for tickets and revenue generation in Brazilian football market

Table 21. Demand for tickets: all Brazilian League tiers (2012, 2013 and 2014)

VARIABLES General General 1st 2nd 3rd 4th

Tickets Price 0.0111 0.00782 0.00534 -0.0126 0.0158 -0.00522 (0.00817) (0.00813) (0.00518) (0.0152) (0.0244) (0.0283) Current Performance 0.445*** 0.425*** 0.258*** 1.268*** 1.441*** 0.375*** (0.0553) (0.0551) (0.0613) (0.285) (0.325) (0.0796) Historical Success 0.180*** 0.192*** 0.101*** 0.242* 1.400** 5.058** (0.0601) (0.0644) (0.0330) (0.128) (0.636) (2.283) Ln (Population) 0.0684 (0.0611) City HDI 0.738 (1.549) Relative Population 6.039 3.338 -1.476 -20.71 3.480 (7.187) (4.911) (12.19) (31.51) (26.57) Relative HIDI 3.255*** 0.169 -1.512 5.220* 2.469 (1.045) (2.484) (2.318) (2.617) (1.790) Nº of Clubs 0.0578 0.0412 0.0758 0.158 0.0958 0.105 (0.0766) (0.0782) (0.0582) (0.138) (0.236) (0.230)

Constant 6.036*** 4.024*** 8.453*** 10.58*** 2.559 4.858*** (1.063) (1.131) (2.799) (2.717) (2.725) (1.814)

Division FE Yes Yes - - - - Region FE Yes Yes Yes Yes Yes Yes Season FE Yes Yes Yes Yes Yes Yes

VIF 3.62 3.53 3.40 6.94 2.64 1.55 Observations 302 302 60 60 61 121 R-squared 0.696 0.703 0.560 0.703 0.707 0.400 Standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1

The results show that current performance and historical success are the main drivers for attracting fans to stadiums in the entire Brazilian League. This variable is also significant in all specific models. These findings are in line with the literature, such as those reported by Madalozzo and Villar (2009), Cox (2015), and Martins and Cró (2016), despite each of these studies employing different measures. This is a somewhat plausible scenario in a context where there are no season ticket schemes. Thus, the interest among fans is strongly impacted by the success of their team.

125

Differences in the demand for tickets in each division of the Brazilian League

Historical success is critical to attracting fans in all models too. Sass (2016) remarks that historical and ongoing success can increase clubs’ market size. In this sense, clubs with a successful history have larger fan bases, resulting in greater demand for tickets. A noteworthy observation is that, in the First and Second Divisions, current performance is much more important than historical success in attracting fans. On the other hand, in the Third Division, both factors have a slightly similar impact, while, in the Fourth Division, historical success has substantially higher influence in terms of attracting fans.

The relative HDI indicates a positive impact on attracting fans, which means that clubs located in regional cities with higher standards of living may experience a higher demand for tickets than other cities with a poorer quality of life. Thus, the higher the socio-economic levels experienced by residents in cities, the greater the likelihood that they will attend a football match more often. On the other hand, the specific outputs confirm that higher relative HDI may increase the average attendance in the lower tiers, particularly the Third Division. Although the model is unable to clarify the reasons for this finding, in a country with significant social imbalance, it is possible that residents in cities with better socio-economic situations could afford to attend a football match more often. Moreover, Mourão (2010) has shown that clubs located in cities with higher GDP display superior levels of performance, while the present findings show that, at least in the Brazilian case, better standards of living may also increase the demand for tickets.

Nevertheless, this variable does not have impact the Fourth Division, which could be explained by its complex and distinctive structure, involving a significant number of participants and a competitive design. In this sense, as some omitted variables may have helped to explain the demand for tickets in Brazil, further analysis should be conducted to address this matter.

126

Empirical analysis of broadcast demand, competitive balance, demand for tickets and revenue generation in Brazilian football market

In addition, city population, both as a logarithm and in relative terms, does not have any statistical impact. Even though a larger population could result in larger markets, all clubs have their own fans. Moreover, the number of clubs in each city has no statistical impact either. Thus, as these city features may not fully represent their respective market size, the absence of effect may be a factor19.

The number of clubs variable has no impact on attracting fans to a stadium.

Feddersen and Maennig (2009) found that, in the German Bundesliga, the greater the number of clubs within a radius of 100 km radius, the bigger the reduction in the demand for tickets. On the other hand, the present findings show that, in the Brazilian case, there is no influence on the demand for tickets, which may be a peculiarity of that market.

While, in European countries, the most important cities usually have only one club

(exceptions include London and Madrid), the main cities in Brazil often have at least two big clubs, as indicated in Table 22 below. Thus, the concentration of clubs in the same city does not negatively affect demand for tickets in Brazil.

Table 22. Top 10 Brazilian cities by population and clubs in the main Brazilian cities (2017)

City (population) Club (Brazilian League titles) São Paulo (12,038,175) Corinthians (6), Palmeiras (9), São Paulo (6), Santos (8) Rio de Janeiro (6,498,837) Botafogo (2), Fluminense (4), Flamengo (5), Vasco (4) Brasília (2,977,216) - Salvador (2,938,092) (2), Vitória (0) Fortaleza (2,609,716) Ceará (0), Fortaleza (0) (2,513,451) Atlético-MG (1), Cruzeiro (3) (2,094,391) - Curitiba (1,893,997) Atlético-PR (1), Coritiba (1) Recife (1,625,583) Náutico (0), Sport (1) (1,481,019) Grêmio (2), Internacional (3) Source: Self-elaboration from IBGE Cidades and Brazilian Football Confederation.

19 Unfortunately, there is no information on the fan communities of all of the 302 clubs in the analysis. Thus, city population could better represent the available proxy of market size. 127

Differences in the demand for tickets in each division of the Brazilian League

7.5. Final Remarks

This chapter confirms that the determinants of the demand for tickets within the

Brazilian League tiers are slightly similar. The findings indicate that current performance and historical success are the main drivers for attracting fans in all divisions. On the other hand, better socio-economic indicators play a major role in attracting fans in the lower levels. Further research could reanalyse the Fourth Division, as the specific regression presents a lower explanatory power. Moreover, the absence of any statistical impact on price could be reanalysed using other econometric approaches, such as Instrumental

Variables, or equations systems involving the Three-Stage Least Squares estimator.

128

Empirical analysis of broadcast demand, competitive balance, demand for tickets and revenue generation in Brazilian football market

8. DEMAND FOR TICKETS: DISTRIBUTION OF WEALTH IN

THE BRAZILIAN STATE CHAMPIONSHIPS

8.1.Problem Statement

Although Brazilian clubs now play in national and continental tournaments, for many years, the State Championships were the most important football tournaments in

Brazil, as well as an important factor in promoting this sport in the country, as noted by

Holanda (2014).

A peculiar characteristic of the State Championships is the competition between clubs from same geographical area and different national levels, even non-division clubs.

Thus, it is often the case that there are matches where ‘big’ teams, with millionaire budgets and experience of international competitions, are pitted against poor clubs with players that earn very low wages.

The State Championships have been progressively losing their relevance, despite their historical reputation. Instead, national and international tournaments have become more attractive to fans, as their teams can play against clubs at a supposedly similar level.

Consequently, in recent seasons, some teams have been playing a number of matches in the state championships with lower-profile and youth team players.

In recent years, the mass media and several fans have been complaining about the

States Championships, while a popular movement called Bom Senso FC has been established by professional players to raise related concerns. All these groups have proposed changes to competition formats, including a reduction in the number of participants and the exclusion of the ‘Brand teams’ in certain tournaments.

129

Demand for Tickets: Brazilian State Championships

Nevertheless, despite the effort to arrive at an appropriate solution for those tournaments, no scientific investigation has interrogated these championships until now.

Therefore, the aim of the present chapter is to analyse how the play-off stages and the presence of ‘Brand-teams’ affect the demand for tickets and, subsequently, match day revenues of all participants teams.

At this point, it is necessary to define what is meant by a ‘Brand-team’. This is a team that, by dint of its history, sporting track record and typically significant budgetary resources, has the potential to attract a high number of fans, in turn becoming a valuable

‘brand’. Obviously, each team can claim that to be a brand, but here the meaning relates to the differential potential to attract supporters or consumers of their products, i.e., live and broadcast matches, merchandise, etc.

Undoubtedly, the economic and technical gaps between clubs play a major role in these tournaments. The winners in every State Championship are frequently from the First

Division of the Brazilian League. However, some clubs from the lower tiers and even

Non-Division clubs have been State Championships winners throughout history. In addition, besides the financial differences between them, qualification to the play-off stages is not always assured for Brand-teams. Table 23 shows that, even in recent seasons, some big clubs have been eliminated in the first rounds.

Table 23. Brand-teams that have qualified for play-offs

Mineiro Paulista Carioca-TG Carioca-TR Carioca 2015 100%(n=2) 100%(n=4) (n=4)- (n=4)- 100%(n=4) 2014 100% 75% - - 100% 2013 100% 100% 100% 50% - 2012 100% 100% 100% 75% - 2011 100% 100% 75% 75% - 2010 100% 50% 100% 100% - 2009 100% 100% 75% 100% - Source: Self-elaboration.

130

Empirical analysis of broadcast demand, competitive balance, demand for tickets and revenue generation in Brazilian football market

8.2. Theoretical Background

Recently, the effects caused by brand-teams have become a relevant research topic. Blumrodt, Bryson and Flanagan (2012) show how football clubs’ corporate social responsibility can influence their brand equity via engagement with fans. Feldmann

(2007) considers two components of this effect: perceived brand image and perceived sporting success. Moreover, Beccarini and Ferrand (2006) assert that sporting performance, efficient management and satisfaction of their fans can influence the image of a club.

Biner (2014) reports an increase in the demand for tickets in the NFL when the visiting team was a major club, i.e., a Brand-team. The author also finds an increase in television’s demand for matches among clubs in the same city, while observing that

Arizona fans are more interested in the visiting team than the local one.

Similar effects have been observed in professional football. Koenigstorfer,

Groeppel-Klein and Kunkel (2010) suggest that the international success of FC Bayern

München and Manchester United has been essential and significant in increasing the attractiveness of the German Bundesliga and the English Premier League, respectively.

Czarnitzki and Stadmann (2002), meanwhile, report that fans are more interested in games against high-reputation opponents than balanced matches. Furthermore,

Pawlowski and Anders (2012) confirm a significant increase in the demand for tickets when club visitors present a ‘strong brand’.

Madalozzo and Villar (2009) also observe this effect in the Brazilian League, in turn confirming an increase in the demand for tickets if the visiting teams are from the

São Paulo and Rio de Janeiro States. At the same time, other researchers corroborate

131

Demand for Tickets: Brazilian State Championships these results, such as Santana and Da Silva (2009) and Bortoluzzo, Iaropoli and Machado

(2011).

The effects resulting from competition design have also been investigated. Lee and Fort (2008) note that play-offs increase the level of fans’ interest, which in turn increments the revenues generated by decisive matches in the MLB. Krautmann and

Ciecka (2009), meanwhile, state that clubs generate approximately US4 11 million extra in income due to their participation in the MLB play-offs.

According to Fernandes (2000), match day tickets were the only source of revenue for football clubs over many years. However, Gaffney (2013) comments that the revenue structure in Brazilian football has changed in response to the globalization of sports.

Notwithstanding, tickets still represent a significant percentage of teams’ total revenue: around 17.42% of total revenue in the case of the top Brazilian football clubs, according to Itaú BBA (2016).

The aforementioned papers have only analysed the impact of visiting Brand-teams on attendance levels at stadiums. Nevertheless, the present paper aims to determine how clubs could increase match day revenues as well. This matter is relevant to the Brazilian

State Championship, as this effect could financially help the smaller clubs playing in those tournaments. In parallel, the impact of the play-off stages is also researched, taking into account the findings reported by Lee and Fort (2008) and Krautmann and Ciecka

(2009). Therefore, this chapter seeks to discover the existence of a redistributive effect of wealth, as well as qualify the impact caused by decisive matches in revenue generation.

132

Empirical analysis of broadcast demand, competitive balance, demand for tickets and revenue generation in Brazilian football market

8.3. Methods

8.3.1. Method and Hypotheses

The present study aims to present pioneering economic research on the Brazilian

State Championships. Thus, it has an exploratory character, with an equations system used for testing the hypothesis. The 3SLS estimator is the chosen estimation method in order to exclude the possible endogeneity of the variables (Greene, 2002).

The following hypotheses are proposed in this work:

H1. The presence of Brand-teams in a competition increases match day revenues for all participating teams.

Going further, this hypothesis is divided into two sub-hypotheses in order to consider the process by which brand image affects revenue:

H1a. The presence of Brand-teams in competitions increases attendance in the championships.

H1b. Attendance increases match day revenues.

8.3.2. Sample

The sample comprises three of the most important Brazilian States

Championships, Carioca, Mineiro and Paulista, which relate to Rio de Janeiro, Minas

Gerais and São Paulo, respectively. The financial data were gathered from the websites of the state federations.

The classification of clubs as Brand-teams was accomplished in two parts. Firstly, we selected all of the founders of the , a group created in 1987 by Brazil’s elite teams. The objective of this group was to develop the Brazilian football market (Silva

Jr., Salazar & Feitosa, 2014). A second criterion was included: only teams with more than

133

Demand for Tickets: Brazilian State Championships

20 State Championships titles and at least one championship in the Campeonato

Brasileiro were selected. Thus, the following clubs can be considered as Brand-teams:

Botafogo, Flamengo, Fluminense and Vasco in the Carioca State Championship;

Atlético-MG and Cruzeiro in the Mineiro State Championship; and Corinthians,

Palmeiras, Santos and São Paulo in the Paulista State Championship.

The data set consists of all 1,114 matches over the 2012, 2013 and 2014 seasons:

216 from Mineiro, 380 from Carioca and 518 from Paulista. As explained before, each

State Federation has autonomy in defining the competitive format of its championship.

The number of competitors, the number of matches and the promotion and relegation system are determined by the State Federations and the associated clubs. In the following sub-sections, the competition design adopted in each state is briefly explained.

8.3.2.1.

This tournament takes place in Minas Gerais State and involves 12 clubs every season. In the group stage, each team plays against all opponents once. The top four teams then qualify for the semi-finals. The semi-final stage is played over two matches, with each team playing once at home and once away (first-placed vs. fourth-placed team; and second-placed team vs. third-placed team). The final follows the same model with two matches, i.e., a home and an away leg. The Campeonato Mineiro had the same competition design during the 2012, 2013 and 2014 seasons. Further information is available from http://fmf.com.br/default.aspx.

8.3.2.2. Campeonato Carioca

The Rio de Janeiro Championship (Campeonato Carioca) has had two different competition designs. In 2012, the group stage comprised 16 teams, divided into two groups of eight, while the tournament had two parts: Taça Guanabara (TG) (Guanabara

134

Empirical analysis of broadcast demand, competitive balance, demand for tickets and revenue generation in Brazilian football market

Cup) and Taça Rio (TR) (Rio Cup). First, the TG was played, following by the TR. In the

TG, the teams played against opponents in their own group, resulting in a total of seven games for each club. The top two clubs in each group reached the semi-finals (first-placed team in Group A vs. second-placed team in Group B; and first-placed team in Group B vs. second-placed team in Group A). The semi-finals were contested as two single matches, with the winners progressing to a final, which was also played as a single match.

In the TR, Group A teams played against teams from Group B, resulting in each team playing eight games. As in the former case, the top two in each group met in the semi- finals and the winners in the final, all contested in a single match schema. In the end, the winner of the TG faced the winner of the TR in the championship final, which was played over two games, one in each stadium.

The 2013 season has similar competitive format, but with a slight difference. In the TG the teams from Group A played against opponents from the Group B and in the

TR the clubs played against opponents in their own group. In addition, in the 2013 season the same team won both the TG and TR, which meant that they automatically became champion without the need to play in a final.

On the other hand, in the 2014 season, the Carioca Championship involved a first stage with 16 clubs, where all the clubs played each other once, with the top four clubs qualifying for the semi-finals (first place vs. fourth place; and second place vs. third place). The semi-final stage was played over two legs, although the top two (first- and second-placed teams in the preliminary phase) enjoyed a ‘draw advantage’: in the case of an equal sum of goals in both matches, the qualified club was the top placed team from the first phase. The final was played over two legs as well, where the best qualifying team from the first phase also enjoying a ‘draw advantage’. All this information can be found at http://www.fferj.com.br/. 135

Demand for Tickets: Brazilian State Championships

8.3.2.3. Campeonato Paulista

The Paulista Championship (in São Paulo State) has had two different designs over the relevant period as well. In 2012 and 2013 editions, 20 teams participated, where all the teams played with each other once. The top eight clubs qualified for the quarter- finals. This stage was decided in a single match, where the best four played as home teams. The confrontation was organized as follows: first place vs. eighth place; second place vs. seventh place; third place vs. sixth place; and fourth place vs. fifth place. The four winners of their respective matches progressed to the semi-finals, which were also settled in a single match. On the other hand, the final was played over two legs, with the second leg held at the stadium of the best qualifying team in the first phase. Besides the traditional play-offs, in 2012 and 2013 editions of the Paulista Championship, the four small teams that lost in the quarter-finals classified to the ‘B’ semi-final. The rules were the same as the Paulista Championship, that is, one match for the ‘B’ semi-final and a two-legged ‘B’ final.

The 2014 season involved opening rounds in which 16 clubs played against each other once. The top eight qualified for the quarter-finals, which were played in a single match. The winners of these matches then progressed to the semi-finals (once again involving a single game), with two best performing teams playing the final over two legs.

More information is available at http://www.futebolpaulista.com.br/Home/.

8.3.3. The Model and Variables

The econometric approach consists of a 3SLS equations system to test the hypotheses presented above for every home-team i, State Championship k and date t. The model is as follows:

Ln(푅푒푣푒푛푢푒푠) = 푓 (퐴푡푡푒푛푑푎푛푐푒, 푃푙푎푦표푓푓, 퐵푟푎푛푑푇푒푎푚) { 푘푡 푘푡 퐿푛(퐴푡푡푒푛푑푎푛푐푒)𝑖푘푡 = 푓 (푊푒푒푘푒푛푑, 푃푙푎푦표푓푓, 퐵푟푎푛푑푇푒푎푚)𝑖푘푡

136

Empirical analysis of broadcast demand, competitive balance, demand for tickets and revenue generation in Brazilian football market

Attendance and Revenue are continuous variables expressed by their natural logarithm, while the play-off expresses two dummy variables, Semi-final and Final, which take the value ‘1’ if the match corresponds to that particular phase, and ‘0’ otherwise. The Brand-team variable assumes a value of ‘1’ when a Brand-team played as a visiting team, whereas the Weekend variable takes the value ‘1’ when a match occurred at weekend (either Saturday or Sunday), and ‘0’ otherwise. The home team fixed effect, whose control variable is i, is used to account for non-observed factors that influence ticket demand.

8.4. Descriptive Analysis of Mineiro, Carioca and Paulista State Championships

Table 24, Table 25 and Table 26 present the summary statistics for the three researched State Championships over the whole period. Firstly, the average attendances are evidently low in all of the tournaments. Moreover, there is no trend that is growing or decreasing, as there are slight similarities over the seasons between each championship.

On the other hand, the significant standard deviations indicate heterogeneous samples with almost empty stadiums, as well as others that are rather crowded; the minimum and maximum values also confirm this observation.

137

Demand for Tickets: Brazilian State Championships

Table 24. Summary statistics: Minas Gerais State Championship (Mineiro)

Variable Campeonato Mineiro (n=216)

Mean Std. Dev. Minimum Maximum

Attendance 2013-15 5,351.92 9,023.48 57 53,585

2013 6,451.96 10,641.33 57 52,989

2014 4,257.25 6,941.45 357 48,818

2015 5,346.56 9,093.91 189 53,585

Revenues (R$) 177,718.70 425,011.30 810 3,677,635

Semi-Final 0.055556 0.229594 0 1

Final 0.027778 0.164717 0 1

BrandTeam 0.203704 0.403687 0 1

Weekend 0.800926 0.400231 0 1 Source: Self-elaboration.

Table 25. Summary statistics: Rio de Janeiro State Championship (Carioca)

Variable Campeonato Carioca (n=380) Mean Std. Dev. Minimum Maximum Attendance 2013-15 3,558.25 7,420.66 150 58,446 2013 2,428.23 4,498.46 150 32,770 2014 2,827.51 5,428.32 150 42,697 2015 5,372.30 10,477.10 243 58,446 Revenues (R$) 127,647.60 375,909.30 0 3,286,580 Semi-Final 0.031579 0.175107 0 1 Final 0.01579 0.124825 0 1 BrandTeam 0.260526 0.439501 0 1 Weekend 0.652632 0.476762 0 1 Source: Self-elaboration.

138

Empirical analysis of broadcast demand, competitive balance, demand for tickets and revenue generation in Brazilian football market

Table 26. Summary statistics: São Paulo State Championship (Paulista)

Variable Campeonato Paulista (n=518) Mean Std. Dev. Minimum Maximum Attendance 2013-15 6,500.02 7,641.85 32 39,479 2013 6,270.55 7,012.68 81 36,306 2014 5,686.22 6,466.25 32 34,964 2015 7,607.17 9,250.07 213 39,479 Revenues (R$) 233,003.50 452,849.00 767 4,181,281 Semi-Final 0.011583 0.107103 0 1 Final 0.011583 0.107103 0 1 BrandTeam 0.198842 0.399515 0 1 Weekend 0.604247 0.489485 0 1 Source: Self-elaboration.

If a reduction in interest among fans cannot be proven across the sample, the critical financial situation can be verified: Table 27 shows a high percentage of matches with net losses. However, comparing the net income (loss or profit) of matches against brand teams and that of other games (referred to as normal matches) shows that latter kind of match can increase clubs profit (Table 28). Likewise, play-off matches could have a similar positive effect on increasing club profits, as shown in Table 29.

Table 27. Net income (profit and loss) by championship and year: number of matches

Mineiro Carioca Paulista 2013 2014 2015 2013 2014 2015 2013 2014 2015 Loss 36 45 40 77 105 93 56 55 65 Income 36 27 32 49 21 35 146 103 93 %Loss 50 62.5 55.56 61.11 83.33 72.66 27.72 34.81 41.14 Source: Self-elaboration.

139

Demand for Tickets: Brazilian State Championships

Table 28. Net income (profit and loss): normal matches vs. matches against brand teams

Normal match Against Visiting Brand-Teams Matches % Matches % Loss 499 57.49 73 29.67 Profit 369 42.51 173 70.33 Source: Self-elaboration.

Table 29. Net loss/income by play-off

Group Stage Matches Play-off Matches Matches % Matches % Loss 568 54.10 4 6.25 Income 482 45.90 60 93.75 Source: Self-elaboration.

Table 30 confirms that fans are more interested in play-off matches than group stage ones, as higher values were observed than for the regular stages. Indeed, average attendance rates are significantly higher, while their consequences include greater match day revenue levels and higher net profits.

Table 30. Average values: attendance, match day revenues and net income

Championship Stage Group Stage Play-offs Attendance Revenues NetIncome Attendance Revenues NetIncome (R$) (R$) (R$) (R$) MG/13 5,685.38 198,972.00 69,560.04 14,884.33 558,773.33 189,719.31 MG/14 3,021.55 69,947.34 4,360.44 17,850.00 675,414.67 151,802.74 MG/15 3,853.44 110,944.77 21,329.83 21,770.83 985,181.67 483,285.88 RJ/13 1,813.10 40,117.86 -9,291.44 14,525.67 594,647.50 206,646.77 RJ/14 2,143.52 73,686.08 -4,816.16 16,507.33 954,272.50 307,888.65 RJ/15 4,032.27 126,008.03 14,708.47 32,619.50 1,682,595.00 663,465.20 SP/13 5,733.04 156,094.44 85,491.80 14,781.25 465,433.92 278,466.31 SP/14 4,892.02 135,586.86 62,901.65 20,577.50 867,257.69 504,637.08 SP/15 6,657.26 285,077.10 145,316.31 25,418.00 1,926,878.72 1,167,778.97 Source: Self-elaboration.

140

Empirical analysis of broadcast demand, competitive balance, demand for tickets and revenue generation in Brazilian football market

8.5. Results and Discussion

Table 31 below presents the empirical outputs of the equations system. The model explains 93.52% of the match day revenues and 79.29% of the attendance. Both hypotheses, H1a and H1b, can be accepted in light of the obtained results.

All the explanatory variables have a positive impact on both match day revenues and live attendance rates. For matches played on weekends, there is a higher demand for tickets than on weekdays. This finding is in line with the results reported by Madalozzo and Villar (2009) in their analysis of Brazilian League matches. Either semi-final or final matches increase live attendance rates. The decisive character of those matches may explain the higher level of interest among fans for such games. However, the most powerful determinant of attendance at State Championship matches is playing against a

Brand-team. This is consistent with the findings by Czarnitzki and Stadmann (2002) and

Pawlowski and Anders (2012) in the German Bundesliga.

On the other hand, attendance is the greatest explanatory factor in relation to growth in match day revenue generation. This is to be expected, due to the fact that this source of revenue strongly depends on the number of tickets sold. It can also be stated that the play-off stages (semi-finals and final) also increase match day revenue, which concurs with the findings of Lee and Fort (2008) and Krautmann and Ciecka (2009), who found similar behaviour in the finances of the MLB clubs when involved in play-offs.

Furthermore, matches against Brand-teams also enhance match day revenue generation.

Indeed, the impact resulting from play against major teams is even more significant than that resulting from play-off games.

141

Demand for Tickets: Brazilian State Championships

Table 31. Empirical estimations (3SLS): attendance and match day revenues

Variables (All Sample) (All Sample) (Only Small Clubs) (Only Small Clubs) Ln(Revenues) Ln(Attendance) Ln(Revenues) Ln(Attendance)

Ln(Attendance) 1.247*** - 1.043*** - (0.0148) - (0.0339) - Semi-Final 0.171** 0.204* 0.294* 0.447** (0.0849) (0.118) (0.1520) (0.2149) Final 0.207* 0.401*** 0.560* 1.040** (0.110) (0.153) (0.301) (0.424) Brand-Team 0.251*** 1.389*** 0.610*** 1.524*** (0.0380) (0.0459) (0.0631) (0.0501) Weekend - 0.135*** - 0.123*** - (0.0372) - (0.0430)

Constant 0.825*** 7.101*** 2.216*** 6.830*** (0.114) (0.258) (0.238) (0.253)

Home-Team FE - Yes - Yes Championship FE Yes Yes Yes Yes Season FE Yes Yes Yes Yes

Observations 1,112 1,112 832 832 R-squared 0.935 0.793 0.906 0.726 Standard errors in parentheses

*** p<0.01, ** p<0.05, * p<0.1

An alternative regression analysis only finds matches with small clubs, as home teams has been analysed already. Every explanatory factor has a similar effect.

Nevertheless, the impact of playing against Brand-teams is even higher for both attendance and match day revenues, which confirms that the presence of Brand-teams in the Brazilian State Championships is definitely an important provider of financial aid to small clubs.

142

Empirical analysis of broadcast demand, competitive balance, demand for tickets and revenue generation in Brazilian football market

Therefore, it can be stated that the main factor that increases match attendance rates is the presence of Brand-teams in these tournaments, while, in this case, the championship stages are less significant. Considering this point, the presence of the

Brand-teams in the State Championships is imperative, as they help to increase attendance and, indirectly, provide a positive financial effect, caused by greater sales of tickets. Table

32 may validate this finding, as it shows that, despite a reduced number of matches against

Brand-teams, compared with normal ones, such matches provide much higher net incomes.

Thus, it can be emphasized that fans are more interested in attending these kinds of matches, which in turn generates larger match day revenues. In other words, the presence of Brand-teams in the Brazilian State Championships provides a redistributive effect of wealth among all the participants, resulting in more financial security for small clubs. In this sense, the proposal from the mass media to exclude Brand-teams from the

State Championships, as well as exclusively apply the play-off format to competition designs, should not be implemented, as doing so would reduce match day revenues for small clubs even more.

143

Demand for Tickets: Brazilian State Championships

Table 32. Total net income by championship: normal matches vs. matches against brand teams

Championship Total Net Income Nº of Matches

Normal match Against Brand-Team

MG/13 R$ 2,043,650.54 R$ 9,163,895.00 57 / 15

MG/14 R$ 122,319.43 R$ 1,076,286.04 58 / 14

MG/15 R$ 1,390,655.03 R$ 2,916,829.24 57 / 15

RJ/13 -R$ 1,058,047.66 R$ 1,182,955.36 94 / 32

RJ/14 -R$ 698,491.57 R$ 1,967,883.78 93 / 33

RJ/15 -R$ 498,458.78 R$ 6,273,683.15 94 / 34

SP/13 R$ 8,053,279.86 R$ 11,531,757.33 160 / 42

SP/14 R$ 6,783,734.09 R$ 6,688,609.53 129 / 29

SP/15 R$ 15,775,938.57 R$ 15,363,739.12 126 / 32 Source: Self-elaboration.

8.6. Final Remarks

The present chapter has analysed, for the first time, the Brazilian State

Championships. The impact of Brand-teams on growing attendance and, consequently, enhancing match day revenues has been established, especially in terms of a redistributive effect of revenue among all participants.

While higher levels of match day revenue in the play-offs are confirmed, the effect from the presence of Brand-teams as opponents is even higher than the stage at which the match is played (by more than 300%). Thus, although decisive matches should logically enjoy high rates of attendance, the presence of Brand-teams is an essential factor to increment the demand for tickets. In this way, changes to the competitive format involving a tournament exclusively based around play-offs is not justified, as this could reduce the probability of enhancing match day revenues for small clubs.

All the three studied tournaments have reported poor average attendance rates, even though the Paulista tournament exhibited a slightly better average number of

144

Empirical analysis of broadcast demand, competitive balance, demand for tickets and revenue generation in Brazilian football market attendees. Thus, additional research is necessary in order to analyse several other factors that might affect the demand for matches in the State Championships, as well as generate solutions for attracting supporters to stadiums. Further investigation is also needed on other Brazilian State Championships, as there may be specific characteristics in each area and its teams, as well as highlight some positive implications for those regions.

The impact of Brand-teams on attendance and match day revenues is evidenced.

When a Brand-team plays as a visitor, it attracts a larger number of fans to the stadiums, which in turn generates greater revenue for the local club. In the context where big and small clubs are playing among themselves, a redistributive effect is created, with the latter aiding the former indirectly. This finding has special relevance in Brazil, considering the weak financial situation faced by several football clubs there (Proni & Zaia, 2014), as well as that fact that 85% of all Brazilian professional football clubs only play in the State

Championships over a season. Therefore, maintaining the Brazilian State Championships is essential for small teams’ financial survival.

145

Demand for Tickets: Brazilian State Championships

146

Empirical analysis of broadcast demand, competitive balance, demand for tickets and revenue generation in Brazilian football market

9. BROADCAST DEMAND

9.1.Problem Statement

Sports demand is one of the most broadly researched topics in the sports economics literature. Borland and Macdonald (2003) and García and Rodríguez (2009) have analysed more than 80 papers exploring this issue. Despite a wide range of papers on the demand for tickets, an interest in the demand for broadcast sports is only a recent phenomenon. From a financial perspective, the revenue from broadcasting represents the most important source of sports-related income. Thus, it is crucial to identify and analyse the factors affecting the demand for televised sports.

Meanwhile, Borland and Macdonald (2003) signalled several possibilities in which to conduct research on this issue. They also noted that most of the literature focuses on the United States and Europe, while the results can only be regarded as a ‘generality’.

Although their paper was written more than a decade ago, only a small number of studies about African, Asian and Latin American professional sports has developed since then.

Hence, much more exploration in this area is needed.

Some recent papers have rejected the classical theory approach from Rottenberg

(1956) and Neale (1964) regarding the importance of UOH in attracting fans. This theory is derived from these seminal papers, which note that fans are more interested in balanced matches involving uncertain sporting results. However, Coates, Humphreys and Zhou

(2014) and Humphreys and Zhou (2015) have referred to two other important consumer choices: loss aversion and win preference. As these findings are in conflict with the classical assumption, this opens up a fundamental issue for further research.

Notwithstanding, Andreff (2014) assumed that television viewers are attracted by the uncertainty of outcome, not by the specific teams that are playing. As Coates, Humphreys

147

Broadcast Demand and Zhou (2014) and Humphreys and Zhou (2015) have found in relation to fan preferences in live game attendances in the MLB, the present study aims to explore the relationship between television audiences and demand for Brazilian League matches.

9.2. Broadcast Football in Brazil

Broadcast rights represent an important source of revenue in the Brazilian football market. According to a report by Itaú BBA (2016), broadcast rights provide 42.25% of the total revenue for the 24 most important Brazilian football clubs. While, in 2010, television rights represented €175.35 million, in 2014, this amount more than doubled to

€462.58 million20. Thus, an analysis of the determinants of broadcast demand in this particular market may be required.

The most important tournament in Brazil (Brazilian League) is similar to other domestic football leagues around the world. In this competition, 20 clubs play in a double round-robin competition. The tournament also has a promotion and relegation system, where the club with the highest number of points at the end of the championship becomes the champion, while the four worst teams are relegated to the Second Division. Moreover, the top four teams qualify for the most important South American cup, the Copa

Libertadores de América21.

The Campeonato Brasileiro has always been broadcast by the same television channel, namely, the Rede Globo, the leading free-to-air channel in Brazil, which is the holder of the broadcasting rights for the Brazilian League. Over a long period, Rede Globo obtained the rights for all media platforms in the Brazilian League. However, according to Mattos (2012), in 2010, the Conselho Administrativo de Defesa Economica (CADE),

20 1. R$583 million and R$ 1538 million. Exchange Rate €1.00 = R$ 3.3248 in 07/03/2017 21 2. The number of classified teams has increased to 7 in the 2016 edition. However, there is no confirmation about the number of clubs in the further seasons 148

Empirical analysis of broadcast demand, competitive balance, demand for tickets and revenue generation in Brazilian football market i.e., the Administrative Council of Economic Defence, intervened in the sales of broadcasting rights and established the unbundled selling of each media platform: free- to-air, Pay television, Pay-Per-View, mobile and the internet. Moreover, the CADE pointed out that the free-to-air channel could sub-license its broadcasting rights to other channels.

Nevertheless, the current situation is completely akin to what was the case several years ago, given that Rede Globo continues to hold the broadcasting rights. Although unbundled selling is mandatory, the channels that bought the rights to the Brazilian

League for Pay television and Pay-Per-View platforms form part of Rede Globo: SporTV and PremierFC, respectively. Obviously, these channels have participated in a competitive bidding procedure, yet Rede Globo still maintains a sort of monopoly. Mattos

(2012) reports that Brazilian clubs started negotiating broadcasting rights for the period

2012-14, along with another free-to-air television channel, because they could sign individual contracts. Nonetheless, no fragmentation took place, while all the clubs separately negotiated their contracts with Rede Globo. Thus, this channel has continued to be the holder of all Brazilian League broadcasting rights.

In recent seasons, Brazilian football fans have been able to watch football on four alternative television channels: Rede Globo and Rede Bandeirantes22 (known as Band) on a free-to-air platform; SporTV on Pay television; and PremierFC on aas Pay-Per-View channel. The number of broadcast matches varies from channel to channel. Rede Globo and Band broadcast twice a week on Wednesdays at 10 pm and Sundays at 4 pm; meanwhile, SporTV broadcasts matches three times a week on Wednesdays, Thursdays

22 3. On 3rd May 2016 Band confirms that it will not broadcast the Brazilian League for that season. 149

Broadcast Demand and Saturdays at 7.30 pm, 9 pm and 6.30 pm, respectively. However, fans can watch all

380 matches of the Brazilian League on PremierFC.

Another peculiar characteristic of televised football in Brazil is that free-to-air matches are regionally broadcast. Although Rede Globo is a national channel, its television schedule is specific to each state. Some programmes, such as soap operas and national news, are the same throughout the country. However, in the particular case of football games, the television channels in each of the states can determine which matches they will broadcast.

Under these circumstances, the matches are broadcast according to which teams are playing. If a state has at least one club playing in the First Division of the Brazilian

League, the channel will usually broadcast their matches. Likewise, the channels in the states having historically broadcast the matches of the best Brazilian clubs. For example, in the Rio de Janeiro State, the Rede Globo only broadcasts matches when Botafogo,

Flamengo, Fluminense or Vasco are playing; in this sense, in the Brazilian League, this television channel will not broadcast matches involving clubs from Goiás against Bahia or against Paraná, for example. This same situation happens in states such as São Paulo, Minas Gerais and . However, some states have no clubs in the First Division. Thus, the television channels adjust their schedules by selecting either matches played by famous clubs or important games.

An odd situation is related to the matches broadcast by Band. This channel sub- licenses the free-to-air football broadcasting rights in Brazil. Nevertheless, in its contract with Rede Globo, Band is merely allowed to broadcast the same matches that Globo broadcasts. Thus, Brazilian fans have the chance to watch the same game on two different free-to-air channels by simply choosing their favourite channel.

150

Empirical analysis of broadcast demand, competitive balance, demand for tickets and revenue generation in Brazilian football market

Thus, in light of the economic importance of broadcasting rights in Brazilian football and the lack of study in the literature on this topic, this paper aims to analyse the determinants of broadcast football in Brazil. Furthermore, taking these peculiarities into account, this paper specifically seeks to compare the two markets and determine whether the demand for broadcast football between them differs.

9.3. Theoretical Background

Sports consumption experienced a revolution about 25 years ago. Several television channels emerged in the broadcast sports market, while the number of televised matches has increased. This has in turn changed fan behaviour. At that time, clubs questioned whether broadcast games might decrease ticket demand, given that television was capable of substituting match attendance at stadiums. Thus, some scholars analysed the impact of broadcast sports on match attendance.

Several papers have been presented in different sports, such as rugby (Baimbridge et al., 1995; Carmichael et al., 1999), Australian football (Borland, 1987), MLB

(Bruggink & Eaton, 1996; Humphreys, 2002) and professional Football (Allan, 2004;

Allan & Roy, 2008; Baimbridge, Cameron & Dawson, 1996; García & Rodríguez, 2002).

However, these authors revealed mixed results. Some papers have found reductions in match attendance, while many others have discovered no effect. Meanwhile, Borland and

Macdonald (2003, p. 488), while analysing some of these studies, concluded that the ‘live broadcast of a match may decrease attendance at the match, but nevertheless stimulate interest in the sporting competition in a way that increases total attendance’.

Although the impact of broadcast matches on stadium attendance remains unclear, it has provided substantial ongoing increases in revenue for the clubs. Hoehn and

Lancefield (2003) reflected the enormous growth of broadcasting rights in the NFL, the

151

Broadcast Demand

FIFA World Cup, the English Premier League, the UEFA Champions League and the

Summer Olympic Games. Moreover, according to Deloitte (2017), 39% of the total revenues of the top 20 clubs in Europe is derived from broadcasting rights during the previous season. In light of this financial rise, some researchers have conducted studies to identify the determinants of broadcast sports in order to provide relevant insight into clubs, leagues, fans and television channels.

Broadcast demand has been comprehensively analysed in numerous sports and different countries. The North American Major Leagues were studied by Coates,

Humphreys and Zhou (2014), Foster, O’Reilly, Shimizu, Khosla and Murray (2014),

Grimshaw and Burwell (2014), Hausman and Leonard (1997), Kanazawa and Funk

(2001), Paul and Weinbach (2007), Salaga and Tainsky (2015a), Tainsky and McEvoy

(2012), Tainsky (2010), and Mongeon and Winfree (2012). Moreover, Dang, Booth,

Brooks and Schnytzer (2015) have researched Australian football, Rodríguez, Pérez,

Puente and Rodríguez (2015) have analysed professional cycling, and Tainsky, Salaga and Santos (2012) have examined the Ultimate Fighting Championship (UFC).

Likewise, the broadcast demand in professional football has been profoundly researched as well. From an international perspective, Nüesch and Franck (2009) analysed the World Cup and the European Football Championship. These authors showed that the expected game quality is a powerful determinant to increase television ratings, while patriotism also plays a big role. In addition, Feddersen and Rott (2011) examined television audiences at German national team matches, finding that the type of the match, the importance of the tournament, the star players’ presence and the high quality of opponents increase the number of viewers.

152

Empirical analysis of broadcast demand, competitive balance, demand for tickets and revenue generation in Brazilian football market

However, some papers have been conducted from a domestic league viewpoint.

Pérez, Puente and Rodríguez (2015) analysed the Spanish La Liga, Johnsen and Solvoll

(2007) researched the Danish and Norwegian top leagues, and Forrest, Simmons and

Buraimo (2005), Alavy, Gaskell, Leach and Szymanski (2010), and Buraimo and

Simmons (2015) explored broadcast matches in the English Premier League.

In the Spanish case, Pérez, Puente and Rodríguez (2015) found that television audience levels decrease when channels broadcast matches with no clubs from the region in question. On the other hand, they showed that Real Madrid and FC Barcelona can independently increase television ratings by 74% when their matches are broadcast.

Johnsen and Solvoll (2007) reported interesting findings related to fan preferences, finding that strategic scheduling, related to time slots, is an important way in which public channels can increase their audiences. However, the authors remarked that, for private broadcasters, content is the key factor. Notwithstanding, they verified that the uncertainty of outcome has no important role in increasing audience numbers. They suppose that this happens due to the presence of several casual viewers.

Working with a minute-by-minute data set, Alavy et al. (2010) present an important consideration about the effects of UOH on broadcast demand. Although they affirm that the uncertainty of outcomes could matter to viewers, their findings confirm that, during a match, fans are likely to switch channel when the draw look likes the final outcome. Thus, they found that matches that end in victories have higher average television ratings.

Furthermore, Buraimo and Simmons (2015) and Forrest, Simmons and Buraimo

(2005) demonstrate that match quality is the most important element in increasing television ratings in the English Premier League. However, according to these studies,

153

Broadcast Demand the impact of the uncertainty of outcome has been changing. In the earliest paper, the authors noted that the UOH affects the interest level among viewers, albeit modestly. On the other hand, the recent paper confirms the fact that the importance of UOH has been decreasing over time to the extent that it has no impact nowadays. In this sense, fans are more interested in watching the best matches with several star players than balanced contests.

While most researchers have only explored the effects of UOH on sports demand,

Coates, Humphreys and Zhou (2014) have developed a theoretical consumer choice model to understand fans’ preference in terms of attending a sports match. Their model, based on Koszegi and Rabin’s (2006) paper, was also empirically tested by a structural econometric method. Their results were in opposition to those of the classical UOH, thus confirming the influence of loss aversion behaviour on live attendances at MLB matches.

Humphreys and Zhou (2015) corroborate those findings, remarking that win preference and loss aversion are factors that affect consumer choice and positively influence the demand for tickets. Nevertheless, these papers have only analysed live attendance.

Therefore, the phenomenon may have an effect on television audiences as well.

9.4. Methods

9.4.1. Data

The sample consists of 228 broadcast football matches from the First Division of the Brazilian League over the 2013, 2014 and 2015 seasons. Each state in Brazil broadcasts different football games as previously mentioned. Thus, the data set is divided into two sub-samples, including 115 matches broadcast in Rio de Janeiro State and 113 in São Paulo State, which may highlight distinct determinants according to their fan behaviour. These 228 matches all represent broadcast matches from the Brazilian League in both states during the researched period. 154

Empirical analysis of broadcast demand, competitive balance, demand for tickets and revenue generation in Brazilian football market

This paper analyses two well-known markets in Brazil, Rio de Janeiro and São

Paulo, which are the wealthiest states according to GDP. Moreover, São Paulo and Rio de Janeiro are respectively the first and third most populated states in Brazil23. In sports terms, they have the greatest share of the most successful football clubs throughout the whole history of the Brazilian League. The clubs from these two states have won the league 45 times or 75% of the total.

9.4.2. Model and Variables

A model to investigate the determinants of television audiences in Brazilian football is developed. Several works related to the demand for tickets, as well as broadcast demand, were considered when creating the model. Table 33 shows the variable description, while Table 34 summarizes the descriptive statistics. The designed model for each date t and team i is presented as follows:

ln (푇푉푎푢푑𝑖푒푛푐푒𝑖푡)

= 훽0 + 훽1푤푒푒푘푒푛푑𝑖푡 + 훽2푚푎푡푐ℎ_푞푢푎푙𝑖푡푦𝑖푡 + 훽3표푝푝표푛푒푛푡_푡𝑖푡푙푒푠𝑖푡

+ 훽4푑푒푟푏푦𝑖푡 + 훽5푐ℎ푎푚푝𝑖표푛_푑푒푓𝑖푛푒푑𝑖푡 + 훽6푟푒푙푒푔푎푡𝑖표푛𝑖푡

+ 훽7푟𝑖푣푎푙_푝푙푎푦𝑖푛푔𝑖푡 + 훽8푑𝑖푠푡푎푛푐푒𝑖푡

+ 훽9(푡ℎ푒𝑖푙, 푤𝑖푛_푝푟푒푓푒푟푒푛푐푒, 푙표푠푠_푎푣푒푟푠𝑖표푛)𝑖푡

+ 훽10푚표푛푡ℎ_푑푢푚푚𝑖푒푠𝑖푡 + 훽11푠푒푎푠표푛_2014𝑖푡 + 훽12푠푒푎푠표푛_2015𝑖푡

+ 휀𝑖푡

The model consists of a panel data linear regression with team fixed effects. As explained before, the broadcast matches in Brazil are regionalized and take into account the presence of local teams playing. Therefore, this methodology takes the specific and

23 Instituto Brasileiro de Geografia e Estatística (Brazilian Institute of Geography and Statistics). Available at http://www.ibge.gov.br/home/. 155

Broadcast Demand non-observable effects of each football club into account to explain the dependent variable. Hence, it generates unbiased and more consistent estimators. This estimator considers the clubs Botafogo, Flamengo, Fluminense and Vasco in the Rio de Janeiro

State, and Corinthians, Palmeiras, São Paulo and Santos in the São Paulo State. Only matches involving the presence of these clubs were broadcast in both markets, confirming the need for fixed effects.

Table 33. Variables description

Variable Description Dependent Variable TV Audience This is the number of people that watched each game. Match Characteristics Weekend This is a dummy variable that takes value equals 1 if the match was played on weekend and 0 otherwise. Match Quality This is the sum of both teams’ points prior to the match. It is relativized by the maximum possible sum in each round. Opponent Titles Opponent number of titles in the Brazilian League First Division. Champion This is a dummy variable that takes value equals 1 if the match was broadcast after Defined the Champion was defined Relegation This is a dummy variable that takes value equals 1 if the broadcast match has a club fighting against relegation after the Champion was defined Substitute Factors Rival Playing This is a dummy variable that takes value equals 1 if at least one rival was playing at the same time and 0 otherwise. Distance This variable represent the distance in kilometres between the city where the club is placed and where the match is taking place Uncertainty of Outcome Theil Index This variable represents the level of Uncertainty of Outcome measured by Theil Index. Win Preference This is a dummy variable that takes value equals 1 if the probability according to the odds to win was bigger than the sum of probability to draw or lose. Loss Aversion This is a dummy variable that takes value equals 1 if the sum of the probability to win and draw was bigger than the probability to lose. Seasonal Effect Month Dummies These variables represent the month dummies according to the month each match took place. Source: Self-elaboration.

The dependent variable is the average number of people who watched each broadcast match on Rede Globo. These values are estimated from the IBOPE points, which represent the most important measure of television audiences in Brazil. IBOPE is the abbreviation of Instituto Brasileiro de Opinião Pública e Estatística (Brazilian

156

Empirical analysis of broadcast demand, competitive balance, demand for tickets and revenue generation in Brazilian football market

Institute of Statistic and Public Opinion), one of the biggest market research companies

in South America and the most well-known of this kind of company in Brazil. The

dependent variable assumes its original value due to its normal distribution, as well as the

better interpretation of the estimated coefficients.

Table 34. Summary statistics

RIO DE JANEIRO SÃO PAULO Variable Obs Mean Std. Dev. Min Max Obs Mean Std. Dev. Min Max

TV audience 115 2,126,977 438,483.3 1,209,802 3,139,533 113 3,313,210 569,729.3 2,229,768 4,755,888 Weekend 115 0.7217391 0.4501038 0 1 113 0.7168142 0.4525528 0 1 Match Quality 115 0.437017 0.1291896 0 0.7154471 113 0.4908546 0.1334802 0 1 Opponent 115 2.53913 2.514125 0 8 113 2.345133 2.012328 0 8 Titles Derby 115 0.0521739 0.2233508 0 1 113 0.1238938 0.3309279 0 1 Champion 115 0.0782609 0..2697571 0 1 113 0.0707965 0.257627 0 1 Defined Relegation 115 0.0353982 0.1856073 0 1 113 0.0608696 0.2401373 0 1 Rival Playing 115 0.5217391 0.5017133 0 1 113 0.619469 0.4876801 0 1 Distance 115 834.4609 615.5949 0 2307 113 647.4814 615.7693 0 2644 Theil 115 1.030975 0.0780401 0.6338242 1.097555 113 1.064377 0.0524462 0.7677848 1.097948 Win Pref. 115 0.2434783 0.4310596 0 1 113 0.4690265 0.5012626 0 1 Loss Aversion 115 0.6782609 0.4691879 0 1 113 0.9026549 0.2977475 0 1 Source: Self-elaboration.

For a long period, the IBOPE was the only corporation responsible for analysing

television audiences in Brazil. However, in 2015, a German company called GfK entered

that market. Nevertheless, the holder of the broadcasting rights (Rede Globo) has made

no use of the information provided by GfK. Thus, Brazilian football audiences are

exclusively measured by the IBOPE.

The IBOPE points are measured by a Local People Meter (LPM), a type of

software installed in certain television, which represents a statistical portion of the

population. When a person turns on the television set, the LPM sends this information to

157

Broadcast Demand a central station where the data are collected and used to calculate the total audience for every television programme. The market size depends on the number of televisions in every state, as well as the number of people in each area. Moreover, the market size experiences slight changes every year, on account of the demographic development.

Thus, the two researched markets in this paper differ with each other. In 2013, one IBOPE point in São Paulo represented 185,814 people; meanwhile, in Rio de Janeiro, it represented 108,587. In São Paulo, one point signified 193,281 in 2014 and 198,162 in

2015, whereas, in Rio de Janeiro, it represented 109,982 and 116,279, respectively.

The explanatory variables are divided in three groups: Match Characteristics

(MC), Substitute Factors (SF) and Uncertainty of Outcome (UO). In the first group, the

Weekend dummy variable is employed in order to analyse how the fixtures affect the preferences among Brazilian supporters. Previous papers have also used variables, such as those by Forrest, Simmons and Buraimo (2005), Buraimo and Simmons (2015)24, and

Salaga and Tainsky (2015b). Through this variable, one can observe whether fans are more interested in watching football matches on weekends or weekdays (to be precise, on Sundays or Wednesdays25).

Five other variables associated with contest quality form part of the MC group:

Match Quality, Opponent Titles, Derby, Championship Defined dummy and Relegation dummy. The first variable is a proxy for the quality of the match, which is calculated by the sum of both teams’ points prior to the game, before relativizing it by the maximum possible points in each round. Buraimo and Simmons (2015) have employed this variable in their work. The second variable is a proxy for the historical success of the opponents.

24 Buraimo and Simmons (2015) have evaluated that by the Weekday variable. 25 The time slots are not researched due to the broadcast matches on free-to-air TV always take place at Wednesday 10pm and Sunday 4pm. 158

Empirical analysis of broadcast demand, competitive balance, demand for tickets and revenue generation in Brazilian football market

Matches against successful opponents should attract larger television audiences.

Feddersen and Rott (2011) confirmed the positive association between higher levels of contest quality and interest among fans. The third contest quality is a dummy variable for analysing derby matches, which typically attract more fans, due to club rivalry: Koning and Achterhof (2015) and Madalozzo and Villar (2009) empirically found that derby matches actually increase stadium attendance rates. The Championship Defined variable is a dummy variable, which takes ‘1’ when the match is broadcast once the championship is already defined. The outcome of the Brazilian League, in the period of interest, has been defined early on: i.e., rounds 34, 36 and 35 in 2013, 2014 and 2015, respectively.

Thus, the final rounds may result in smaller television audiences. Notwithstanding, after the title definition, television companies may decide to broadcast matches where there are clubs fighting against relegation, which could in turn increase the size of the television audiences; as such, the Relegation dummy variable should capture this effect.

Two different substitute factors are employed in the models: Rivals Playing at the

Same Time and Distance. The first variable could control a situation where the majority of the television audience for a match are supporters of each team. Evidently, in a free- to-air channel, some casual viewers are able to watch football matches free of charge.

Thus, if the television audience does not reduce in size when at least one rival is playing on two channels at the same time, this could mean that the audiences are loyal to either club. The Distance variable is measured by the distance between the home city of the respective club and the place where it is playing, in kilometres. Free-to-air television channels often broadcast matches when the local club is playing as an away team.

However, some matches are broadcast in the same city where the match takes place.

Therefore, the Distance variable seeks to determine whether television audiences decrease when a team plays at home and whether larger distances lead to higher ratings.

159

Broadcast Demand

The last group of explanatory variables is related to Uncertainty of Outcome. As the literature shows, mixed evidence has been presented on this matter, with three different regressions estimated in order to analyse the behaviour of fans in each market.

The betting odds are used to create all the variables concerning Uncertainty of Outcome.

The data were collected from http://www.oddsportal.com/. Firstly, the uncertainty of outcome is measured by the Theil index, which was developed by Theil (1967) and first employed in a football context by Peel and Thomas (1992). It takes into account the three possible football outcomes (win, lose and draw), with higher values reflecting more balanced matches.

The second analysed behaviour, Win Preference, is measured by a dummy variable equal to ‘1’ when the probability of winning is higher than the sum of the probabilities of ending in a draw or losing. This should represent a clear possibility of winning. If this variable has a positive impact (i.e., increasing television audience size), this confirms that the supporters display a win preference behaviour.

The last Uncertainty of Outcome variable represents Loss Aversion, which is measured by a dummy variable that takes the value ‘1’ when the sum of the probabilities to win or draw is higher than the probability of losing. If there is a positive significant relationship between this variable and broadcast demand, this means that the real interest among fans concerns not losing the game. Thus, it can reveal fans’ behaviour in relation to loss aversion.

Furthermore, month dummies are employed in order to capture seasonal effects over the tournament. Pawlowski and Anders (2012) and Pawlowski and Nalbantis (2015) found a U-shaped relationship between attendance and the league fixtures in Austria and

Switzerland. This finding confirms the fact that supporters in these countries usually

160

Empirical analysis of broadcast demand, competitive balance, demand for tickets and revenue generation in Brazilian football market prefer to attend matches at the beginning and the end of the season. As the Brazilian

League starts in April, this month is used as a base. At the same time, Pawlowski and

Anders (2012) observe that the lowest match attendance occurs in the winter season.

Thus, these month dummies could capture whether this effect, or its opposite, applies to broadcast matches as well.

9.4.3. Model Selection

As three different models are estimated in each market, changing the variables related to the Uncertainty of Outcome, the Akaike Information Criterion (AIC) and the

Bayesian Information Criterion (BIC) are carried out in order to analyse the goodness of fit. The AIC and BIC were presented by Akaike (1974) and Schwarz (1978), respectively, both of which consider the maximum value of the likelihood function of the models and their estimated parameters. The relative best model is the one with lower values, which in turn can determine the best regression among them.

9.5. Results and Discussion

Table 35 presents the outputs of the regressions. The first three models refer to

Rio de Janeiro State and the other three refer to São Paulo. The models explain around

68% of the broadcast demand in Rio de Janeiro and about 74% in São Paulo. As expected, the markets present similarities as well as some differences. The results are consistent for

Rio de Janeiro and São Paulo, as the coefficients for their models are similar.

The third model is the best regression for Rio de Janeiro, while the first regression is the best for São Paulo, when taking into account the AIC and BIC values. As each model estimates the presence of different fan behaviour, only the most appropriated in each market is explained below.

161

Broadcast Demand

Table 35. Determinants of television audiences in Rio de Janeiro and São Paulo

Rio de Janeiro São Paulo VARIABLES Theil Win Pref Loss Aversion Theil Win Pref Loss Aversion

Weekend -524,676*** -528,920*** -521,993*** -762,945*** -770,285*** -766,132*** (32,651) (31,359) (29,390) (63,249) (70,942) (71,821) Match Quality 420,355 407,476 484,742* 701,997** 873,468** 654,887** (189,059) (222,531) (201,943) (212,051) (182,861) (146,883) Opponent Titles 23,504** 22,354 28,924** 33,045 38,807 33,629 (4,989) (10,626) (5,935) (30,687) (26,740) (30,021) Derby 555,239*** 572,887** 515,719*** 515,692** 361,976** 487,968*** (78,307) (111,624) (80,162) (118,023) (70,027) (54,364) Champion Defined 105,720 132,082 122,994 -57,584 -88,474 -75,642 (75,929) (85,506) (80,663) (100,974) (128,744) (65,268) Relegation 314,679* 307,372** 273,961* -2,754 97,568 -7,844 (108,415) (69,847) (113,783) (170,985) (199,996) (230,208) Rival Playing -102,616 -99,633 -102,833 -5,354 -518.7 14,248 (86,568) (86,672) (89,588) (64,611) (58,228) (58,001) Distance (km) 45.93* 56.58** 48.89 -1.139 -21.54 -37.43 (19.29) (17.68) (25.90) (32.98) (30.13) (41.81) Theil 234,090 -1.558e+06** (239,100) (394,457) Win Preference 5,068 110,882 (89,888) (60,170) Loss Aversion 91,397** -205,857 (25,218) (149,047) May -58,264 -62,434 -74,661 -8,275 -20,727 20,952 (121,417) (104,065) (88,236) (160,224) (152,162) (195,130) June -271,881 -280,967 -257,510 -535,847** -547,716** -528,556* (209,332) (191,936) (173,099) (142,550) (104,305) (167,016) July -182,693 -201,030 -183,126 -279,321** -281,878 -232,710** (110,866) (90,445) (86,728) (51,615) (119,876) (45,669) August -126,145 -134,584 -139,809 -181,291 -197,248 -164,981* (207,742) (193,506) (177,986) (79,586) (149,437) (66,214) September -16,492 -30,395 -12,624 -58,593 -88,198 -38,383 (183,674) (179,262) (170,530) (59,292) (41,987) (71,905) October -91,378 -102,276 -97,180 -73,575 -55,285 -34,541 (116,176) (100,875) (94,039) (42,309) (38,566) (37,315) November -69,425 -81,825 -87,850 -15,695 -45,164 11,464 (209,999) (200,931) (181,244) (55,799) (36,020) (35,154) December 22,721 -20,567 -4,092 -103,811 -65,434 47,434 (224,152) (243,792) (214,513) (420,131) (422,116) (516,933)

Constant 1.973e+06*** 2.228e+06*** 2.112e+06*** 5.041e+06*** 3.285e+06*** 3.571e+06*** (167,892) (196,351) (158,169) (383,686) (90,055) (38,870)

AIC 3160.465 3160.917 3158.329 3158.308 3162.673 3161.149 BIC 3168.699 3169.151 3166.564 3166.49 3170.855 3169.331

Season FE Yes Yes Yes Yes Yes Yes Club FE Yes Yes Yes Yes Yes Yes Observations 115 115 115 113 113 113 R-squared 0.680 0.679 0.686 0.741 0.730 0.734 Number of id 4 4 4 4 4 4 Robust standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1

162

Empirical analysis of broadcast demand, competitive balance, demand for tickets and revenue generation in Brazilian football market

9.5.1. Similarities

The Weekend dummy confirms that broadcast football audiences in Brazil are bigger on weekdays for both Rio de Janeiro and São Paulo States. This result contradicts the findings of Forrest, Simmons and Buraimo (2005). However, these authors conducted their research on the English Premier League, which suggest there are differences in the behaviour among Brazilian and English supporters. On the other hand, this finding may complement the results of Madalozzo and Villar (2009), who confirmed that stadium attendance in the Brazilian League is higher at weekends. The opportunity cost could explain these differences, in that it may be more convenient for audiences to watch these games on television, given that the free-to-air broadcast matches are on Wednesdays at

10 pm. Therefore, football fans in Brazil may be more likely to watch broadcast matches on weekdays. Indeed, a weekend match diminishes on average to around 525,000 and

763,000 viewers in Rio de Janeiro and São Paulo, respectively.

The Match Quality variable has a positive impact on audiences in Rio de Janeiro and São Paulo, which means that the current performance of both teams in the Brazilian

League increases the television audience size. The findings highlight that an increase of

10% in match quality results in a growth of 48,000 spectators in Rio de Janeiro and

70,000 in São Paulo. This result is in line with Borland and Macdonald (2003) who identified greater demand for matches with superior contest quality. Additionally, it is noteworthy that both live and broadcast match fans display common behaviour in Brazil because, according to the results of Madalozzo and Villar (2009), current performance increases the demand for Brazilian League tickets as well.

The derby dummy has shown that rivalry between clubs has a strong positive effect on increasing the number of viewers. This result corroborates the findings of

Koning and Achterhof (2015) and Madalozzo and Villar (2009), while confirming the 163

Broadcast Demand effect on broadcast games. As shown in Table 3, a derby match increases the audience size by more than half a million viewers in both states.

The Distance variable, as well as the Rival Playing and Champion Defined dummies, has no impact on both markets. There are two reasons why there is a non- statistical impact in terms of the distance aspect. The first hypothesis is that broadcast fans usually have no wish to go to the stadium, preferring instead to watch the match on television. The second possibility is that, even though many people in major markets, such as Rio de Janeiro and São Paulo, change their consumption patterns from watching a match on television to going to the stadium, this change has no significant impact on audience levels26. The non-statistical effect of the dummy variable, when a rival is playing at the same time, suggests that the television audience usual comprises actual supporters of either club. Hence, the results contradict the assumptions of Andreff (2014) and

Johnsen and Solvoll (2007), due to the possibility that free-to-air viewers are highly interested in specific football teams, at least in the context of Brazilian football. However, matches played after the champion is already decided do not affect broadcast demand.

The fans may still show interest in watching those games on television because their clubs are also playing, while free-to-air broadcasts, by definition, do not involve any fee charged to the viewers.

9.5.2. Differences

Two variables indicate a positive impact on Rio de Janeiro viewers only: the

Opponent Title and the Relegation dummy. Although the Opponent Title variable has a statistical impact, the increment provided by this variable is slight. On the other hand,

26 The largest football stadium by capacity in those markets is Maracanã (Rio de Janeiro): 78,838 people. Considering the average TV audience in Rio de Janeiro, perhaps a sold out in Maracanã will not affect the broadcast demand there. 164

Empirical analysis of broadcast demand, competitive balance, demand for tickets and revenue generation in Brazilian football market despite a 10% significance level, matches played by clubs fighting against relegation led to an average increase of 274,000 viewers. This may be understandable, given that three clubs in the 2013 season, one in 2014 and another one in 2015 from Rio de Janeiro were in this awkward position: Vasco was relegated in the 2013 and 2015 seasons, and

Botafogo in the 2014 edition.

No seasonal impact is found for Rio de Janeiro. Although the audiences were larger in April and decreased over the course of the tournament, no statistical effect was found in respect of the Month Dummies. On the other hand, in São Paulo, a statistical negative decrease was noted in June and July. These months could point to a negative reaction from viewers due to the onset of winter. However, this is unlikely, given that the football is ‘consumed at home’, which probably means the weather does not matter. Thus, two reasons other may be applicable here. The first is related to the winter holidays, when the number of spectators could be reduced by people finding other leisure activities to pursue. The second reason assumes that interest among fans decreases because these matches are in the middle of a tournament, which means that they tend not to be decisive in terms of the final outcome.

The most important goal of the paper is to understand the effect resulting from the

Uncertainty of Outcome. Firstly, the impacts it creates are in line with those reported in the literature on broadcast demand in professional sports, such as by Alavy et al. (2010),

Buraimo and Simmons (2015), and Forrest, Simmons and Buraimo (2005).

165

Broadcast Demand

Table 36. Alternative regression analysing Uncertainty of Outcome in São Paulo

São Paulo VARIABLES UOH

Weekend -761,801*** (62,218) Match Quality 744,910** (150,636) Opponent Titles 34,631 (28,371) Derby 492,534** (89,744) Champion Defined -51,042 (116,743) Relegation 490.8 (184,120) Rival Playing -5,708 (60,576) Distance (km) 3.399 (22.60) Win Probability -2.681e+06** (808,771) Win Probability2 3.619e+06** (804,714) May -18,066 (165,692) June -539,226** (149,322) July -281,919*** (45,389) August -180,299 (87,248) September -63,415 (71,665) October -73,282 (41,645) November -22,969 (68,659) December -127,376 (448,670)

Constant 3.817e+06*** (89,504)

AIC 3158.344 BIC 3166.526

Season FE Yes Club FE Yes Observations 113 R-squared 0.741 Number of id 4 Robust standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1

Furthermore, although Rio de Janeiro and São Paulo are in the same country, the fans in each state have different reasons for watching a televised football match. As Table

35 shows, São Paulo supporters prefer more certain sporting results (a negative Theil

166

Empirical analysis of broadcast demand, competitive balance, demand for tickets and revenue generation in Brazilian football market index), while Rio de Janeiro fans display loss aversion behaviour. The first finding is consistent with the results presented by Coates and Humphreys (2010), who found, in the

NFL, fans expect no balanced games, and Martins and Cró (2016), who found a similar result in the Portuguese Football League (a negative effect of the Theil index). However, as the Theil Index only assumes preferences for balanced or unbalanced games, an alternative regression, which analyses the UOH by win probability and its squared value, is carried out, with its outputs presented in Table 36 above. The alternative regression shows similar impacts in all variables to the best previous model. Moreover, the outputs confirm that fans are interested in matches against weak opponents, as well as strong clubs, with the tipping point being a 67% probability of winning. Thus, this result corroborates with the findings of the aforementioned papers. The second finding, meanwhile, is in line with the findings of Coates, Humphreys and Zhou (2014). By analysing the MLB, these scholars confirmed the prevalence of Loss Aversion behaviour among fans. The results therefore support these two possible fan behaviours and also confirm important differences in consumer preferences between football markets within the same country.

9.5.3. Limitations and Further Research

The method employed here does not allow for inferring the differences between the markets. The historical behaviour of fans may provide an explanation, but only further research can confirm this hypothesis. However, better performances from São Paulo teams, compared with those of Rio de Janeiro, in those seasons could explain why their win preference is higher than their loss aversion.

In the present work, fans’ preferences were only analysed in the two biggest football markets in Brazil. The respective teams are successful and have an enormous fan base. Nevertheless, even matches are broadcast nationwide on free-to-air channels every 167

Broadcast Demand week, this vast country has some states without clubs in the First Division. Thus, some new papers could explore this phenomenon, in the same way that Tainsky and McEvoy

(2012) approached the NFL.

The Brazilian League is one of several tournaments that take place in Brazilian football. Football clubs play the National Cup and two international competitions, as well as State Championships. Thus, analysing the determinants of broadcast demand in these tournaments would be appropriate, with specific focus on their differences in terms of sporting prowess, competition design, relevance and participants. These elements may affect fans’ impulse towards watching a broadcast match.

Moreover, several differences between cities and regions may exist in other contexts, such as in European countries. Although matches are usually broadcast to a nationwide audience, the determinants may differ by area according to supporter preferences. Further research could consider Major Leagues or college sports in the

United States and on European professional football in the European context.

Another issue of note regarding broadcast sports is related to pay-per-view channels. The determinants of their demand may differ from those of free-to-air channels.

It is perhaps the case that any uncertainty of outcome measure, win preference or loss aversion does not work, due to the fans having paid for a subscription to watch all the matches in the whole tournament played by their team. Furthermore, differences between regions may also exist, which could be explored in the future.

9.6. Final Remarks

This chapter is the first to analyse differences between the determinants of broadcast demand in two regions within the same country. Nowadays, broadcast revenues represent the most important source of revenue for football clubs. Hence, this kind of

168

Empirical analysis of broadcast demand, competitive balance, demand for tickets and revenue generation in Brazilian football market research is crucial in understanding supporter preferences. As expected, both similarities and differences between the markets were evidenced. This research demonstrates that

Brazilian fans prefer to watch football matches on television on weekdays. Moreover, the

Uncertainty of Outcome Hypothesis was rejected, given that Brazilian supporters’ preferences are more related to contest quality, such as current performance, the historical success of the opponent or team rivalry, which is in line with the findings of recent papers.

Furthermore, Loss Aversion behaviour among Rio de Janeiro fans and the negative impact of the Uncertainty of Outcome among São Paulo supporters, who are likely to watch matches against week or strong opponents, are confirmed.

Although broadcast matches divided by regions are a peculiarity of the Brazilian football market, the findings could pose implications for other countries. As consumer preference may differ in some areas, television channels and sport clubs should analyse the specific characteristics of each area in order to better understand the wishes of their supporters. By understanding the diversity of fan behaviour, television channels should be able to more effectively decide which matches to broadcast. Indeed, doing so could even increase the level of fans’ interest in these games, as well as generate more revenue for both channels and teams. Further studies along these lines are encouraged within other contexts, such as the North American Major Leagues and European football.

169

Broadcast Demand

170

Empirical analysis of broadcast demand, competitive balance, demand for tickets and revenue generation in Brazilian football market

10. CONCLUSIONS

The present study, which has analysed certain economic aspects related to the

Brazilian League, may represent a novel piece of exploratory research, as only a few papers about sports economics in the Brazilian context have been previously published.

Several findings have been reported and practical implications suggested, as well as issues for further research in the coming years.

Some peculiarities of the Brazilian football market are described in the Chapter 2.

The Brazilian professional football market has an unusual institutional structure, in which

State Federations have autonomy over organizing certain tournaments and registering players. Moreover, the sporting calendar is quite different to that of most other countries.

The existence of various kinds of domestic tournament, such as rRegional and State

Championships, is another peculiarity. Further papers could analyse how the institutional structure influences the organization and nature of these tournaments.

The second chapter also discussed some of the economic and financial problems affecting Brazilian football clubs. The pursuit of utility maximization can be observed, as revenues have been increasing along with costs at similar levels. The significant amount of debts is a major financial problem, as well as the dependence on television revenues, which needs to be addressed in the next few years. On the other hand, given that the information presented in this chapter is merely descriptive, more robust analysis should be performed to better understand the entire economic situation faced by Brazilian teams, in the same way that Barajas and Rodríguez (2014) analysed the situation with Spanish football.

The findings from Chapter 3 confirm that each tournament has a different impact on revenue generation. Indeed, in the Brazilian football market, only the Brazilian 171

Conclusions

League, the Brazilian Cup and the Copa Libertadores show a positive statistical influence in terms of increasing total revenues. Nonetheless, the structure of those tournaments plays a big role. While strong performances in cup championships may raise current revenue levels, success in the league results in income growth in the following economic year. In knockout tournaments, clubs earn more money every round for which they qualify. Thus, these kinds of tournament have a financial impact during the current season. On the other hand, excellent results in a league tournament can affect the financial situation in the years that follow, as successful clubs are able to participate in other tournaments. In turn, these clubs should receive a higher percentage of television revenues, sign better contracts with sponsors and possibly transfer their most outstanding players to other countries for a lucrative fee, which is a current behaviour found among the Brazilian clubs. New studies could analyse whether this is a common factor in other

South American countries.

The impacts in terms of the Competitive Balance and fans’ interest following changes to the competition design in the Brazilian League were examined in Chapter 4.

Although Drummond, Araújo Jr. and Shikida (2010) have previously commented about the increase in CB, it was statistically confirmed for the first time in this work. On the other hand, no statistical significant change was observed regarding fans’ interest levels after 2003: despite a slight positive trend, average attendance rates have remained low.

Papers analysing fans’ interest in other play-off tournaments have reported similar findings. The Brazilian Cup could be an interesting subject for study, as well as the Copa

Libertadores de América. Indeed, similar methodology as used by Szymanski (2001), when comparing the interest levels of English fans in FA Cup and English Premier

League matches, could be used to examine this topic in Brazilian football.

172

Empirical analysis of broadcast demand, competitive balance, demand for tickets and revenue generation in Brazilian football market

The concept of CB was interrogated in Chapter 5. Firstly, a new index was created, based on the accumulated difference of points among the clubs, which represents the main methodological contribution of this work. In turn, this approach was used to compare competitiveness levels in the Brazilian League with those in another eight domestic leagues, namely, England, France, Germany, Italy, the Netherlands, Portugal, Russia and

Spain. The one-way ANOVA with the Tukey post hoc test confirmed that the Brazilian

League was the most balanced tournament among the sample. Moreover, no statistical differences were observed among the European leagues. Next, the ADP index was validated, through a comparison involving the HICB and C4ICB, with a view to analysing the Campeonato Brasileiro. The results confirmed the quality of the new index. Further papers could seek to discover which factors make the Brazilian League more balanced than others.

The demand for tickets in the First Division of the Brazilian League was reanalysed in Chapter 6. In addition, the impacts resulting from non-usual stadium features and certain quality aspects were researched for the first time, with the findings being in line with previous results, such as the general quality of the match, derbies and matches on weekends increasing the demand for tickets. Scheduling Brazilian League matches only at weekends, as some European domestic leagues do, is suggested as practical measure supported by the results to easily increase average attendance, albeit slightly. On the other hand, previous findings reported by Madalozzo and Villar (2009) on the Uncertainty of Outcome is rejected: U-shaped behaviour was found, indicating that

Brazilian fans are more interested in matches with lower uncertainty levels than those against weak or strong teams.

Meanwhile, the main goals from the sixth chapter are related to stadium characteristics. Firstly, non-usual stadiums decrease the demand for tickets. On the other 173

Conclusions hand, clubs managers have been charging expensive ticket prices when their club play matches outside their home city. Furthermore, the positive impact resulting from the general quality, security, comfort and hygiene levels felt during match attendance is evidenced as well. Notwithstanding, all upper categories (using one star as a base) have higher ticket prices. Some issues of significance could be researched in future papers, such as the effects on sporting performance when matches are played at non-usual stadiums. Moreover, as high-quality stadiums increase the demand for tickets as well as ticket prices, the Brazilian Football Confederation could employ a more severe licensing system for clubs in order to improve facilities for First Division participants.

The demand for tickets in all Brazilian League tiers was investigated in Chapter

7. While previous studies have researched the second tiers in England and Ireland, this is the first analysis to consider a football market at a fully nationwide level. The findings confirm that current performance is a key driver in attracting fans to stadiums in Brazil.

Moreover, differences between the tiers are also found: in the two top tiers (First and

Second Divisions), the historical success of the clubs is an important factor in increasing seasonal attendance; on the other hand, in the lower tiers (Third and Fourth Divisions), higher socio-economic aspects (measured by the relative HDI) could increase the demand for tickets. More research into the lower tiers is encouraged, as their tournaments have different competitive designs and several distinctive aspects, which provide an interesting context for novel analysis.

The eighth chapter examined the most peculiar tournament in the Brazilian football market: the State Championships. The demand for tickets to those tournaments, on a match-by-match basis, is analysed for the first time. The findings show that matches on weekends have higher attendances, with play-off matches and the presence of Brand- teams increasing the demand for tickets. Furthermore, this chapter confirmed the

174

Empirical analysis of broadcast demand, competitive balance, demand for tickets and revenue generation in Brazilian football market redistributive effect of revenue among clubs, with the presence of Brand-teams in those tournaments being beneficial to the smallest ones, as matches against Brand-teams enhance revenue potential for the latter, which in turn can offset their net loss. Further studies could examine other State Championships in Brazil in order to determine whether this effect is similar across the entire country.

At the same time, both Regional and State Championships should be examined in more detail, given that they represent an important part of the Brazilian football labour market. Indeed, these tournaments provide several players with an excellent opportunity to face the best Brazilian clubs every season. Although this could be unusual in other contexts, in the Brazilian case, several players are transferred from the third and fourth tiers, as well as Non-League teams, to First Division clubs. In some cases, players are eventually transferred internationally a few years after moving from one domestic team to another. As such, the peculiar nature of “player movements” within the Brazilian football market could have a positive influence on competitiveness in the Brazilian

League, as well as for the Brazil’s national team in international games.

Chapter 9 represents the first work to analyse the determinants of broadcast matches in Brazil. Some findings, such as the importance of match quality and derby games, related to increases in television audience size are in line with those reported in recent papers. However, midweek games indicate a higher degree of broadcast demand than weekend ones, which is the opposite to what has been observed in the European context. Relegation is another important point in terms of attracting viewers in the

Brazilian case. At the same time, the behaviour of fans from two different states (Rio de

Janeiro and São Paulo) were compared, with our findings highlighting differences between them, which represent an empirical contribution to the field. Rio de Janeiro fans display loss aversion behaviour, while São Paulo fans prefer more certain matches: a 175

Conclusions negative coefficient from the Theil index was found and confirmed by a U-shaped form with regard to the probability of winning and its squared term. In light of these reported differences, further papers could analyse fans’ behaviour in other Brazilian states in order to determine whether there are other dissimilarities. Furthermore, previous papers on both

European and North American sports leagues have analysed their national markets in combination. In this sense, as differences between states, autonomous communities and regions may exist, this matter should be examined further.

Two important tournaments were left out of this dissertation: the Copa do Brasil and the Copa Libertadores da América. Indeed, these tournaments are underexplored in the sports economics literature. Thus, topics such as the demand for tickets, broadcast demand, CB and the effects related to the geographical distribution of participants could be analysed in future. Meanwhile, as the Brazilian Cup has changed its competitive format several times, finding an optimal design could be a worthy pursuit. Moreover,

CONMEBOL, South America’s football confederation, has increased the number of participants in the Copa Libertadores, such that six clubs, instead of the previous four, from the Brazilian League now qualify to play for the Copa Libertadores. Further research could explore whether this situation influences club behaviour and the level of competitiveness in the Brazilian League.

The Brazilian football market has been comprehensively examined in the present dissertation. To summarize, the effects resulting from design changes, CB, and both live and broadcast demand have been analysed, along with the presentation of a number of theoretical, methodological and empirical contributions. The findings confirm that, while the Brazilian League should expect strong long-term competitiveness, this will not result in higher attendances. In particular, however, the interest among fans regarding high- profile matches and excellent stadium facilities should be considered by both the CBF

176

Empirical analysis of broadcast demand, competitive balance, demand for tickets and revenue generation in Brazilian football market and Brazilian clubs when creating policies to enhance the domestic market. They should also find ways to keep the best Brazilian players in the domestic market or, at least, transfer them to “second-level” championship clubs, instead of those in the Brazilian

League. In this sense, the CBF’s Club Licensing straight should be better aligned with encouraging clubs to offer the best facilities by the clubs and retain their best players.

Furthermore, changes in the schedule are suggested, such that some matches in the

Brazilian League are played across the year, rather than concentrated in certain months, as well as only at weekends, in the same way of some of the European domestic leagues.

Finally, it is expected that the present dissertation will inspiration further research into the Brazilian football market. The peculiarities of this particular institutional structure offer a vast range of research avenues, which should be explored in order to broaden the scope of the sports economics literature.

177

Conclusions

178

Empirical analysis of broadcast demand, competitive balance, demand for tickets and revenue generation in Brazilian football market

REFERENCES

Akaike, H. (1974). A New Look at the Statistical Model Identification. IEEE Transactions on Automatic Control, 19 (6), 716-23.

Alavy, K., Gaskell, A., Leach, S., & Szymanski, S. (2010). On the Edge of Your Seat: Demand for Football on Television and the Uncertainty of Outcome Hypothesis. International Journal of Sport Finance, 5 (2), 75-95.

Allan, G., & Roy, G. (2008). Does Television Crowd Out Spectators? New Evidence from the . Journal of Sports Economics, 9 (6), 592-605.

Allan, S. (2004). Satellite television and football attendance: The not so super effect. Applied Economics Letters, 11, 123-25.

Anderson, C., & Sally, D. (2013). Os Números do Jogo - Porque Tudo o Que Você Sabe Sobre Futebol Está Errado. São Paulo: Paralela.

Andreff, W. (2014). Building Blocks for a Disequilibrium Model of European Team Sports Leagues. International Journal of Sport Finance, 9, 20-38.

Andreff, W., & Bourg, J. (2006). Broadcasting rights and competition in European football. In.: C. Jeanrenaud y S. Késenne (eds.), The economics of sport and the media. Edward Elgar, Cheltenham.

Andreff, W., & Bourg, J.-F. (2006). Broadcasting rights and competition in European football. In C. Jeanrenaud & S. Kesenne (Eds.), The economics of sport and the media (pp. 37-70). Cheltenham, UK: Edward Elgar.

Araújo Jr, A. F., Shikida, C., & Monasterio, L. M. (2005). Uma análise econométrica do futebol brasileiro. Análise Econômica, 23 (44), 217-240.

Baimbridge, M., Cameron, S., & Dawson, P. (1995). Satellite Broadcasting and Match Attendance: The Case of Rugby League. Applied Economics Letters, 2, 343-6.

Baimbridge, M., Cameron, S., & Dawson, P. (1996). Satellite Television and the Demand for Football: A Whole New Ball Game? Scottish Journal of Political Economy, 43 (3), 317-33.

179

References

Barajas, A. (2010). Competitive balance and concentration ratios in economic terms in Liga BBVA. Working Paper SSRN. Available at: http://ssrn.com/abstract=2283411.

Barajas, A., & Rodríguez, P. (2014). Spanish Football in Need of Financial Therapy: Cut Expenses and Inject Capital. International Journal of Sport Finance, 9 (1), 73- 90.

Barajas, A., Fernández-Jardón, C., & Crolley, L. (2005). Does sports performance influence revenues and economics results in Spanish football? Munich Personal RePEc Archive Paper No. 3234, University Library of Munich, German.

Barros, C. P., & García-del-Barrio, P. (2011). Productivity drivers and markets dynamics in the Spanish first division football league. Journal of Productivity Analysis, 35 (1), 5-13.

Barros, C. P., & Leach, S. (2006). Performance evaluation of the English Premier Football League with data envelopment analysis. Applied Economics, 38, 1449-1458.

Barros, C. P., & , G. (2014). A Bayesian stochastic frontier of Italian football. Applied Economics, 46 (20), 2398-2407.

Barros, C. P., Assaf, A. G., & Araújo Jr., A. F. (2011). Cost performance of Brazilian soccer clubs: A Bayesian varying efficiency distribution model. Economic Modelling, 28, 2730-2735.

Barros, C. P., Assaf, A., & Sá-Earp, F. (2010). Brazilian Football League Technical Efficiency: A Simar and Wilson Approach. Journal of Sports Economics, 11 (6), 641-651.

Barros, C. P., Peypoch, N., & Tainsky, S. (2014). Cost efficiency of French soccer league teams. Applied Economics, 46 (8), 781-789.

Barros, C. P., Wanke, P., & Figueiredo, C. (2015). The Brazilian Soccer Championship: an efficiency analysis. Applied Economics, 47 (9), 906-915.

Bartsch, T., Drexl, A., & Kroger, S. (2006). Scheduling the professional soccer leagues of Austria and Germany. Computational Operations Research, 33 (7), 1907- 1937.

180

Empirical analysis of broadcast demand, competitive balance, demand for tickets and revenue generation in Brazilian football market

Beccarini, C., & Ferrand, A. (2006). Factors affecting soccer clubs season tickets holder’s satisfaction: The influence of club image and fan’s motives. European Sport Management Quarterly, 6 (1), 1-22.

Bernard, A. B., & Busse, M. R. (2004). Who wins the Olympic Games: Economic Resources and Medals Totals. The Review of Economics and Statistics, 86 (1), 413-417.

Bernard, A. B., & Busse, M. R. (2004). Who wins the Olympic Games: Economic Resources and Medal Totals. The Review of Economics and Statistics, 86 (1), 413-417.

Biner, B. (2014), “Parity in professional sports when revenues are maximized”, Economic Modelling, 40, 12-20.

Blumrodt, J., Bryson, D., & Flanagan, J. (2012), “European football teams' CSR engagement impacts on customer-based brand equity”, Journal of Consumer Marketing, 29 (7), 482-493.

Boded., G., & Bernache-Assollant, I. (2009). Do fans care about hot dogs? A satisfaction analysis of French ice hockey spectators. International Journal of Sport Management and Marketing, 5 (1/2), 15-37.

Booth, R. (2004). The economics of achieving competitive balance in the Australian Football League, 1897-2004. Economic Society of Australia, 23 (4), 325-344.

Borland, F., & Macdonald, R. (2003). Demand for Sport. Oxford Review of Economic Policy, 19 (4), 478-502.

Borland, J. (1987). The Demand for Australian Rules Football. Economic Record, 63 (3), 220-30.

Bortoluzzo, A. B., Iaropoli, P. T., & Machado, S. J. (2011), “Demand for Brazilian Football: A Censored Model Approach”, working paper, Insper Working Paper, No. 237, pp. 1-18.

Branvold, S. E., & Bowers, R. (1992). The use of promotions in college baseball. Sport Marketing Quarterly, 1 (1), 19-24.

181

References

Bruggink, T.H., & Eaton, J.W. (1996). Rebuilding Attendance in Major League Baseball: The Demand for Individual Games`. In J. Fizel, E. Gustafsen, & L. Hadley (Eds.), Baseball Economics: Current Research. Westport, CT: Praeger.

Budzinski, O. (2012). The institutional framework for doing sports business: Principles of EU competition policy in sports markets. International Journal of Sport Management and Marketing, 11 (1-2), 44-72.

Buraimo, B., & Simmons, R. (2008). Do Sport Fans Really Value Uncertainty of Outcome? Evidence from the English Premier League. International Journal of Sport Finance, 3, 146-155.

Buraimo, B., & Simmons, R. (2015). Uncertainty of Outcome or Star Quality? Television Audience Demand for English Premier League Football. International Journal of the Economics of Business, 22 (3), 449-69.

Buts, C., Du Bois, C., Heyndels, B., & Jegers, M. (2011). Socioeconomic Determinants of Success at the Summer Paralympics. Journal of Sports Economics, 14 (2), 133- 147.

Cairns, J. P., Jennett, N., & Sloane, P. (1986). The economics of professional teams sports: A survey of theory and evidence. Journal of Economic Studies, 13 (1), 3- 80.

Carmichael, F., Millington, J., & Simmons, R. (1999). Elasticity of Demand for Rugby League Attendance and the Impact of SkyB. Applied Economics Letters, 6 (12), 797-800.

Castro, C. O. (2014). Brasil é o recordista de mortes por causa do futebol. Available at: http://oglobo.globo.com/esportes/brasil-o-recordista-de-mortes-por-causa-do- futebol-14923352

CBF. (2015). Tabela de Cotas para a Copa do Brasil / 2015. Ofício DFI – 001/2015 (Circular). Confederação Brasileira de Futebol, Rio de Janeiro. Available at http://www.futebolcearense.com.br/2010/downloads/arquivos/arquivo_3719.pdf.

Coates, D., & Humphreys, B. R. (2005). Novelty effect of new facilities on attendance at professional sporting events. Contemporary Economic Policy, 23 (3), 436-455.

182

Empirical analysis of broadcast demand, competitive balance, demand for tickets and revenue generation in Brazilian football market

Coates, D., & Humphreys, B. R. (2007). Ticket Prices, Concessions and Attendance at Professional Sporting Events. International Journal of Sport Finance, 2 (3), 161- 170.

Coates, D., & Humphreys, B. R. (2010). Week to Week Attendance and Competitive Balance in the National Football League, International Journal of Sport Finance, 5 (4), 239-252.

Coates, D., Humphreys, B. R., & Zhou, L. (2014). Reference-Dependent Preferences, Loss Aversion, and Live Game Attendance. Economic Inquiry, 52 (3), 959-73.

Condon, E. M., Golden, B. L., & Wasil, E. A. (1999). Predicting the success of nations at the Summer Olympic Games using neural networks. Computers & Operations Research, 26, 1243-1265.

CONMEBOL. (2014). Copa Bridgestone Libertadores de América 2014 – Reglamento. Confederación Sudamericana de Fútbol, Luque. Recuperado de http://www.conmebol.com/sites/default/files/reglamento_copa_bridgestone_libe rtadores_2014_1.pdf

Cox, A. (2015). Spectator Demand, Uncertainty of Results, and Public Interest: Evidence from the English Premier League. Journal of Sports Economics, 1-28. DOI: 10.1177/1527002515619655.

Criado, R., García, E., Pedroche, F., & Romance, M. (2013). A new method for comparing rankings through complex networks: Model and analysis of competitiveness of major European soccer leagues. CHAOS, 23 (043114), 1-10.

Curran, J., Jennings, I., & Sedgwick, J. (2009). ‘Competitive balance’ in the top level of English football, 1948-2009: An absent principle and a forgotten ideal. The International Journal of the History of Sport, 26 (11), 1735-1747.

Czarnitzki, D., & Stadmann, G. (2002). Uncertainty of outcome versus reputation: Empirical evidence for the First German Football Division. Empirical Economics, 27 (1), 101-112.

183

References

Dang, T.M., Booth, R., Brooks, R., & Schnytzer, A. (2015). Do TV Viewers Value Uncertainty of Outcome? Evidence from the Australian Football League. Economic Record, 91 (295), 523-35.

De Bosscher, V., De Knop, P., Bottenburg, M., Shibli, S., & Bingham, J. (2009). Explaining international sporting success: An international comparison of elite sport systems and policies in six countries. Sport Management Review, 12, 113- 136.

De Bosscher, V., Shibli, S., Bottenburg, M., & Truyens, P. N. J. (2010). Developing a Method for Comparing the Elite Sports Systems and Policies of Nations: A Mixed Research Methods Approach. Journal of Sports Management, 24, 567-600.

Della Croce, F., & Olivieri, D. (2006). Scheduling the Italian Football League: and ILP- based approach. Computational Operations Research, 33 (7), 1963-1974.

Deloitte. (2017). Football Money League: Planet Football. Sport Business Group.

Demmert, H. G. (1973), The economics of professional team sports, Lexington Books, Lexington, MA.

DeSchriver, T. D. (2007). Economics and finance much adieu about Freddy: Freddy Adu and attendance in Major League Soccer. Journal of Sport Management, 21 (3), 438-451.

DeShriver, T. D., Rascher, D. A., & Shapiro, S. L. (2016). If we build it, will they come? Examining the effect of expansion teams and soccer-specific stadium on Major League Soccer attendance. Sport, Business and Management: An International Journal, 6 (2) 205-227.

Dobson, S. M., & Goddard, J. A. (1996). The Demand for Football in the Regions of England and Wales. Regional Studies, 30, 443-453.

Dobson, S., & Goddard, J. (2011). The economics of football. Cambridge: Cambridge University Press.

Drummond, L., Araújo Jr, A. F., & Shikida, C. D. (2010). Campeonato Brasileiro de Futebol e Balanço Competitivo: uma análise do período 1971-2009. Revista Brasileira de Futebol, 3 (2), 73-87.

184

Empirical analysis of broadcast demand, competitive balance, demand for tickets and revenue generation in Brazilian football market

Eckard, E. W. (1998). The NCAA cartel and competitive balance in college football. Review of Industrial Organization, 13 (3), 347-369.

El-Hodiri, M., & Quirk, J. (1971). An economic model of a professional sports league. Journal of Political Economy, 79 (6), 1302-1319.

Espitia-Escuer, M., & García-Cebrián, L. I. (2006). Performance in sports teams: Results and potential in the professional soccer league in Spain. Management Decision, 44 (8), 1020-1030.

Evans, R. (2014). A review of measures of competitive balance in the ‘analysis of competitive balance’ literature [Research Paper Series, Vol. 7, N.º 2]. University of London, Birkbeck Sport Business Centre, London.

Feddersen, A., & Maennig, W. (2009). Arenas Versus Multifunctional Stadiums: Which Do Spectators Prefer? Journal of Sports Economics, 10 (2), 180-191.

Feddersen, A., & Rott, A. (2011). Determinants of Demand for Televised Live Football: Features of the German National Team. Journal of Sports Economics, 12, 352- 69.

Feddersen, A., Maennig, W., & Borcherding, M. (2006). The novelty effect of new soccer stadia: the case of Germany. International Journal of Sport Finance, 1 (3), 174- 188.

Feldmann, S. (2007), Bewertung von Sportmarken: Messung und Wirkungen der Markenstärke von Fuβballbundesligavereinen (Marktorientierte Unternehmensführung). Frankfur am Main: Peter Lang.

Fernandes, L. F. F. (2000), “A gestão dos clubes de futebol como clube empresa: estratégias de negócio”, MBA dissertation, Faculdade de Administração da Universidade Federal do Rio Grande do Sul (UFRS), Porto Alegre, Brazil.

FIFA TMS. (2017). Global Transfers Market Report 2017. Available from: https://www.fifatms.com/data-reports/reports/.

FIFA TMS. (2016). Players Nationality in the Transfer Market. Available from: https://www.fifatms.com/data-reports/reports/.

185

References

Fizel, J. L., & Bennett, R. W. (1989). The Impact of College Football Telecasts on College Football Attendance. Social Science Quarterly, 70 (4), 980-988.

Flores, R., Forrest, D., & Tena, J. D. (2010). Impact on competitive balance from allowing foreign players in a sports leagues: Evidence from European soccer. Kylos, 63 (4), 546-557.

Forrest, D., & Simmons, R. (2002) Outcome Uncertainty and Attendance Demand in Sport: The Case of English Soccer. Statistician, 51 (2), 229-241.

Forrest, D., Simmons, R., & Buraimo, B. (2005). Outcome Uncertainty and the Couch Potato Audience. Scottish Journal of Political Economy, 52 (4), 641-61.

Fort, R. (2007). Comments on ‘Measuring Parity’. Journal of Sports Economics, 8 (6), 642-651.

Fort, R., & Maxcy, J. (2003). Competitive balance in sports leagues: An introduction. Journal of Sports Economics, 4 (2), 154-160.

Fort, R., & Quirk, J. (1995). Cross-subsidizations, incentives, outcome in professional teams sports leagues. Journal of Economic Literature, 33 (3), 1265-1299.

Foster, G., O`Reilly, N., Shimizu, C., Khosla, N., & Murray, R. (2014). Determinants of Regional Sport Network Television Ratings in MLB, NBA, and NHL. Journal of Sport Management, 28, 356-75.

Franck, E. (2014). Financial fair play in European football clubs: What is it all about? International Journal of Sports Finance, 9 (3), 193-217.

Gaffney, C. (2013), “Virando o jogo: The Challenges and Possibilities for Social Mobilization in Brazilian Football”, Journal of Sport and Social Issues, 37, 1-20.

García, J., & Rodríguez, P. (2002). The Determinants of Football Match Attendance Revisited: Empirical Evidence from the Spanish Football League. Journal of Sports Economics, 3 (1), 18-38.

García, J., & Rodríguez, P. (2009). Sports Attendance: A Survey of the Literature 1973- 2007. Rivista di Diritto ed Economia Dello Sport, 5 (2), 111-51.

186

Empirical analysis of broadcast demand, competitive balance, demand for tickets and revenue generation in Brazilian football market

Gasparetto, T., & Barajas, A. (2015). Análisis del Balance Competitivo por la Diferencia de Puntos Acumulada (DPA). Proceedings of XXV Jornadas Hispanolusas, Ourense, España.

Gasparetto, T., & Barajas, Á. (2016). Influencia de los torneos en la generación de ingresos. Revista de Psicología del Deporte, 25 (Supp. 1), 33-36.

Gasparetto, T., Fernández-Jardón, C., & Barajas, A. (2014). Brand teams and distribution of wealth in Brazilian State Championships. Working paper BRP 30/MAN/2014, Higher School of Economics, National Research University.

Gómez-González, C., García-Unanue, J., Sánchez-Sánchez, J., Ubago, E. G., & del Corral, J. (2016). Revista de Psicología del Deporte, 25 (Suppl. 1), 19-22.

Goossens, D. R., & Spieksma, F. C. R. (2012). Soccer schedules in Europe: An overview. Journal of Scheduling, 15 (5), 641-651.

Goossens, D. R., Beliën, J., & Spieksma, F. C. R. (2010). Comparing league formats with respect to match importance in Belgium football. Annals of Operations Research, 194, 223-240.

Goossens, K. (2006). Competitive balance in European football: Comparison by adapting measures: National measure of Seasonal Imbalance and Top 3. Rivista di Diritto ed Economia Dello Sport, 2 (2), 77-122.

Greene, W. H. (2002), Econometric Analysis. Upper Saddle River, New Jersey: Prentice Hall.

Greenwell, T. C., Fink, J. S., & Pastore, D. L. (2002). Assessing the Influence of the Physical Sports Facility on Customer Satisfaction within the Context of the Service Experience. Sport Management Review, 5, 129-148.

Griggs, T. S., & Rosa, A. (1996). A tour of European soccer schedules, or testing the popularity of GK2n. Bulletin of the Institute of Combinatorics and its Applications, 18, 65-68.

Grimshaw, S.D, & Burwell, S.J. (2014). Choosing the Most Popular NFL Games in a Local TV Market. Journal of Quantitative Analysis in Sports, 10 (3), 329-43.

187

References

Groot, L. F. M. (2008). Economics, uncertainty and European football: Treds in competitive balance. Cheltenham: Edward Elgar.

Guterman, M. (2009). O futebol explica o Brasil – Uma história da maior expressão popular do país. São Paulo: Editora Contexto.

Guzmán, I., & Morrow, S. (2007). Measuring efficiency and productivity in professional football leagues: evidence from English Premier League. Central European Journal of Operations Research, 15, 309-328.

Haan, M., Koning, R. H., & Witteloostuijn, A. (2007). Competitive balance in national European soccer competitions. In J. Albert, & R. H. Koning (Eds.). Statistical thinking in sports (Chap. 4, pp. 63-76). London: Taylor and Francis.

Hausman, J., & Leonard, G. (1997). Superstars in the National Basketball Association: Economic value and policy. Journal of Labor Economics, 15 (4), 586-624.

Hoehn, T., & Lancefield, D. (2003). Broadcasting and Sport. Oxford Review of Economic Policy, 19 (4), 552-68.

Hoffmann, R., Ging, L. C., & Ramasamy, B. (2002). The Socio-economic determinants of International Soccer Performance. Journal of Applied Economics, 2 (2), 253- 272.

Holanda, B. B. B. (2014), “The fan as actor: the popularization of football and Brasil’s sports audience”, Football & Society, 15 (1), 8-18.

Houston, R., & Wilson, D. (2002). Income, leisure and proficiency: an economic study of football performance. Applied Economics Letters, 9 (14), 939-943.

Howard, D. R., & Crompton, J. L. (2003). An empirical review of the stadium novelty effect. Sport Marketing Quarterly, 12 (2), 111-117.

Howard, D., & Crompton, J. (2004). Tactics used by sports organizations in the United States to increase ticket sales. Managing Leisure, 9 (2), 87-95.

Humphreys, B. (2002). Alternative measures of competitive balance in sports leagues. Journal of Sports Economics, 3 (2), 133-148.

188

Empirical analysis of broadcast demand, competitive balance, demand for tickets and revenue generation in Brazilian football market

Humphreys, B. R., & Zhou, L. (2015). The Louis-Schmelling Paradox and the League Standing Effect Reconsidered. Journal of Sports Economics, 16 (8), 835-852.

International Monetary Fund. (2014). World Economic Outlook Database 2014. Available from: http://www.imf.org/external/pubs/ft/weo/2014/01/weodata/index.aspx

Itaú BBA. (2015). Análise Econômico-Financeira dos Clubes de Futebol Brasileiros de 2015: Dados Financeiros de 2014. Informe 2015. São Paulo, Brasil.

Itaú BBA. (2016). Análise Econômico-Financeira dos Clubes de Futebol Brasileiros de 2016: Dados Financeiros de 2015. São Paulo: Itaú BBA.

Janssens, P., & Késenne, S. (1987). Belgium football attendances. Tijdschrift voor Economie en Management, 32 (3), 305-315.

Jena, F., & Reilly, B. (2016). Testing the uncertainty outcome hypothesis using data from second tier soccer in Ireland. Applied Economics Letters, 1-4, DOI: http://dx.doi.org/10.1080/13504851.2016.1148252.

Johnsen, H., & Solvoll, M. (2007). The Demand for Televised Football. European Sport Management Quarterly, 7 (4), 311-35.

Kaempfer, W. H., & Pacey, P. L. (1986). Televising college football: The complementarity of attendance and viewing. Social Science Quarterly, 67 (1), 176-185.

Kanazawa, M.T., & Funk, J.P. (2001). Racial Discrimination in Pro Basketball. Economic Inquiry, 39 (4), 599-608.

Kendall, G. (2008). English football fixtures over holiday periods. Journal of Operational Research Society, 59, 743-755.

Kennedy, P. (1992). A Guide to Econometrics. Oxford: Blackwell.

Késenne, S. (2000). Revenue sharing and competitive balance in professional team sports. Journal of Sports Economics, 1 (1), 56-65.

Kittleson, R. (2014). The Country of Football: Soccer and the Making of Modern Brazil. Oakland, CA: University of California Press Books.

189

References

Koenigstorfer, J., Groeppel-Klein, A., & Kunkel, T. (2010). The attractiveness of national and international football leagues: Perspectives of fans of “star clubs” and “underdogs”. European Sport Management Quarterly, 10 (2), 127-163.

Koning, R. H. (2000). Balance in competition in Dutch soccer. The Statistician, 49 (3), 419-431.

Koning, R., & Achterhof, J. (2015). Live football demand. In P. Rodríguez, S. Késenne, & R. Koning (Eds.), The Economics of Competitive Sports. Cheltenham: Edward Elgar.

Koszegi, B., & Rabin, M. (2006). A model of reference-dependent preferences. The Quarterly Journal of Economics, 121 (4), 1133-1165.

Krautmann, A. C., & Ciecka, J. (2009). The Postseason Value of an Elite Player to a Contending Team. Journal of Sports Economics, 10 (2), 168-179.

Kuper, S., & Szymanski, S. (2010). ¡El fútbol es así! Una explicación económica sobre los mitos y verdades del deporte. Empresa Activa, Barcelona.

Lancenet! (2014). Violência entre torcidas já matou 234 pessoas no Brasil, sendo 30 este ano. Available at: http://www.lancenet.com.br/minuto/Violencia- torcidas_organizadas_0_1044495544.html.

Lee, Y. H., & Fort, R. (2008), “Attendance and the Uncertain-of-Outcome Hypothesis in Baseball”, Review of Industrial Organization Vol. 33, No. 4, pp. 281-295.

Lee, Y. H., & Fort, R. (2011). Competitive Balance: Times Series Lessons from English Premier League. Working paper nº 2011-02, Research Institute for Market Economy, Sogang University.

Leeds, M. A., & Leeds, E. M. (2009). International Soccer Success and National Institutions. Journal of Sports Economics, 10 (4), 369-390.

Leoncini, M. P., & Silva, M. T. (2005). Entendendo o futebol como um negócio: Um estudo exploratório. Gestão & Produção, 12 (1), 11-23.

190

Empirical analysis of broadcast demand, competitive balance, demand for tickets and revenue generation in Brazilian football market

Levin, R., Mitchell, G., Volcker, P., & Will, G. (2000). The report of the independent members of the Comissioner's Blue Ribbon Panel on baseball economics. Available at: http://www.mlb.com/mlb/downloads/blue_ribbon.pdf.

Levy, P. (2011). Sports Administration: An examination of the competitive balance concept through European and Brazilian domestic soccer leagues comparison. M.Sc. thesis, Escola de Administração de Empresas de São Paulo, Fundação Getúlio Vargas, Brasil.

Luiz, J. M., & Fadal, R. (2011). An economic analysis of sports performance in Africa. International Journal of Social Economics, 38 (10), 869-883.

Macmillan, P., & Smith, I. (2007). Explaining International Soccer Rankings. Journal of Sports Economics, 8 (2), 202-213.

Madalozzo, R., & Villar, R. B. (2009). Brazilian Football: What Brings Fans to the Game? Journal of Sports Economics, 10 (6), 639-50.

Martins, A. M., & Cró, S. (2016). The Demand for Football in Portugal: New Insights on Outcome Uncertainty. Journal of Sports Economics, 1-25. DOI: 10.1177/1527002516661602.

Mattos, C. (2012). Broadcasting Football Rights in Brazil: The Case of Globo and ‘Club of 13’ in the Antitrust Perspective. Estudos Econômicos, 42 (2), 337-362.

Michie, J., & Oughton, C. (2004). Competitive balance in football: Trends and effects. Research paper 20014 nº 2, Football Governance Research Center, Birkbeck College, University of London.

Milan, B. (1998). O país do futebol. Rio de Janeiro: Record.

Mirabile, M. P. (2015) The Determinants of Attendance at Neutral Site College Football Games. Managerial and Decision Economics, 36 (3), 191-204.

Mongeon, K., & Winfree, J. (2012). Comparison of Television and Gate Demand in the National Basketball Association. Sport Management Review, 15, 72-79.

191

References

Montes, F., Sala-Garrido, R., & Usai, A. (2014). The lack of balance in the Spanish first division football league. European Sport Management Quartely, 14 (3), 282- 298.

Mourão, P. J. R. (2010). Regional Determinants of Competitiveness: The Case of European Soccer Teams. International Journal of Sports Finance, 5, 222-234.

Müller, C., Lammert, J., & Hovemann, G. (2012). The financial fair play regulations of UEFA: An adequate concept to ensure the long-term viability and sustainability of European club football? International Journal of Sport Finance, 7, 117-140.

Murad, M. (2013). Práticas de violência e mortes de torcedores no futebol brasileiro. Revista USP, 99, 139-152.

Neale, W.C. (1964). The Peculiar Economics of Professional Sports. Quarterly Journal of Economics, 78 (1), 1-14.

Neter, J., Wasserman, W., & Kutner, M. H. (1989). Applied Linear Regression Models. Homewood: Irwin.

Noll, R. (1988). Professional basketball [Working Paper Nº 144]. Stanford University, Studies in Industrial Economics, Stanford.

Nüesch, S., & Franck, E. (2009). The Role of Patriotism in Explaining the TV Audience of National Team Games – Evidence from Four International Tournaments. Journal of Media Economics, 22, 6-19.

Owen, P. D. (2010). Limitations of the relative standard deviation of win percentages for measuring competitive balance in sports leagues. Economics Letters, 109 (1), 38-41.

Owen, P. D., Ryan, M., & Weatherspoon, C. R. (2007). Measuring competitive balance in professional team sports using the Herfindahl-Hirschman index. Review of Industrial Organization, 31 (4), 289-302.

Pan, Y., & Jackson, R. T. (2008). Ethnic difference in the relationship between acute inflammation and serum ferritin in US adult males. Epidemiology and Infection, 136, 421-431.

192

Empirical analysis of broadcast demand, competitive balance, demand for tickets and revenue generation in Brazilian football market

Paul, R., Humphreys, B. R., & Weinbach, A. (2012). Uncertainty of outcome and attendance in college football: Evidence from four conferences. The Economic and Labour Relations Review, 23 (2), 69-82.

Paul, R.J., & Weinbach, A.P. (2007). The Uncertainty of Outcome and Scoring Effects on Nielsen Ratings for Monday Night Football. Journal of Economics and Business, 59, 199-211.

Pawlowski, T., & Anders, C. (2012). Stadium attendance in German professional football: The (un)importance of uncertainty of outcome reconsidered. Applied Economics Letters, 19 (16), 1553-1556.

Pawlowski, T., & Nalbantis, G. (2015). Competition Format, Championship Uncertainty and Stadium Attendance in European Football – A Small League Perspective. Applied Economics, 47 (38), 4128-39.

Pawlowski, T., Breuer, C., & Hovemann, A. (2010). Top club's performance and the competitive situation in European domestic football competitions. Journal of Sports Economics, 11 (2), 186-202.

Peel, D. A., & Thomas, D. A. (1988). Outcome uncertainty and the demand for football: An analysis of match attendance in the . Scottish Journal of Political Economy, 35 (3), 242-249.

Peel, D. A., & Thomas, D. A. (1992). The demand for football: Some evidence on outcome uncertainty. Empirical Economics, 17 (2), 323-331.

Peeters, T. (2011). Broadcast Rights and Competitive Balance in European Soccer. International Journal of Sport Finance, 6, 23-39.

Peeters, T. (2011). Broadcast rights and competitive balance in European soccer. International Journal of Sport Finance, 6 (1), 23-39.

Peeters, T., & Szymanski, S. (2014). Financial fair play in European football. Economic Policy, 29 (78), 343-390.

Pérez, L., Puente, V., & Rodríguez, P. (2015). Are Broadcast Sporting Events of “General Interest”? A Regional Panel Data Analysis of TV Ratings for Spain`s La Liga. Journal of Media Economics, 28 (1), 7-19.

193

References

Pitts, J. D. (2016). Determinants of Success in the National Football League’s Postseason: How Important is Previous Playoff Experience? Journal of Sports Economics, 27 (1), 86-111..

Pluri. (2014). 3º ranking mundial de público nos estádios 2013/14. São Paulo: Fernando .Available at: http://www.pluriconsultoria.com.br/uploads/relatorios/PLURI%20especial%20 - %20ranking%20mundial%20de%20publico%20nos%20estadios%202014%20- %20campeonatos.pdf.

Preuss, H., Haugen, K., & Schubert, M. (2014). UEFA financial fair play: The curse of regulation. European Journal of Sport Studies, 2 (1), 33-51.

Price, D. I., & Sen, K. C. (2003). The demand for game day attendance in college football: an analysis of the 1997 Division 1-A season. Managerial and Decision Economics, 24 (1), 35-46.

Proni, M. W., & Zaia, F. H. (2014). Financial condition of Brazilian soccer clubs: an overview. Soccer & Society, 15 (1), 108-122.

Quirk, J., & Fort, R. D. (1991). Pay dirty: The business of professional teams sports. Princeton, New Jersey: Princeton University Press.

Quirk, J., & Fort, R. D. (1997). Pay Dirt: The Business of Professional Team Sports. 2nd ed. Princeton, NJ: Princeton University Press.

Reilly, B. (2015). The Demand for League of Ireland Football. The Economic and Social Review, 46 (4), 485-509.

Ribeiro, A. S., & , F. (2012). Portuguese football league efficiency and players’ wage. Applied Economics Letters, 19, 599-602.

Ribeiro, C. C., & Urrutia, S. (2012). Scheduling the Brazilian soccer tournament: Solution approach and practice. Interfaces, 42 (3), 260-272.

Rodríguez, C., Pérez, L., Puente, V., & Rodríguez, P. (2015). The Determinants of Television Audience for Professional Cycling: The Case of Spain. Journal of Sports Economics, 16 (1), 26-58.

194

Empirical analysis of broadcast demand, competitive balance, demand for tickets and revenue generation in Brazilian football market

Rodríguez, P. (2012). La economía del deporte. Estudios de Economía Aplicada, 30 (2), 1-32.

Rohde, M., & Breuer, C. (2016). Europe’s Elite Football: Financial Growth, Sporting Success, Transfer Investment, and Private Majority Investors. International Journal of Financial Studies, 4 (12), 1-20.

Rosen, S. (1981). The Economics of Superstars. American Economic Review. 71, 845- 858.

Rottenberg, S. (1956). The baseball players` labor market. Journal of Political Economy, 64, 242.-258.

Salaga, S., & Tainsky, S. (2015a). Betting Lines and College Football Television Ratings. Economics Letters, 132, 112-16.

Salaga, S., & Tainsky, S. (2015b). The Effects of Outcome Uncertainty, Scoring, and Pregame Expectations on Nielsen Ratings for Bowl Championship Series Games. Journal of Sports Economics, 16 (5), 439-459.

Santana, S. K. S., & Da Silva, A. S. (2009), ‘The determinants of demand in football matches during the 2007 Brazilian Championship”, IASE/NAASE Working Paper Series, No. 09-06, pp. 1-18.

Sass, M. (2016). Glory Hunters, Sugar Daddies, and Long-Term Competitive Balance under UEFA Financial Fair Play. Journal of Sports Economics, 17 (2), 148-158.

Schimidt, M. B. (2001). Competition in Major League Baseball: The impact of expansion. Applied Economics Letters, 8 (1), 21-26.

Schwarz, G. (1978). Estimating the Dimension of a Model. The Annals of Statistics, 6 (2), 461-64.

Scully, G. W. (1989). The business of major league baseball. Chicago, IL: University of Chicago Press.

Scully, G. W. (1989). The Business of Major League Baseball. Chicago: University of Chicago Press.

195

References

Siegfried, J. T., & Eisenberg, J. D. (1980). The Demand for Minor League Baseball. Atlantic Economic Journal, 8 (2), 59-69.

Silva Jr., A. S., Salazar, V. S., & Feitosa, M. G. G. (2014), “O Clube dos 13 e o novo cenário do futebol brasileiro: Uma análise a partir dos campeonatos baiano, goiano, paranaense e pernambucano”, Revista Brasileira de Ciências do Esporte, 36 (1), 103-122.

Sloane, P. J. (1971). The economics of professional football: the football club as a utility maximiser. Scottish Journal of Political Economy, 18 (2), 121-146.

Solberg, H. (2001). Finding a price policy that maximises the ticket revenues at sporting events: A focus on price bundling. European Sport Management Quarterly, 1 (4), 300-319.

Sordi, J. D., & Bello, E. C. (2014). Contribuições para o desenvolvimento dos pacotes de ingressos no futebol brasileiro: uma abordagem exploratória. Revista Intercontinental de Gestão do Desporto, 4 (2), 229-249.

Soriano, F. (2009). La pelota no entra por azar: Ideias de management desde el mundo del fútbol. Barcelona: Leqtor.

SportTV. (2015). Pesquisa: violência é a principal causa que afasta torcedores dos estádios. Available at: http://sportv.globo.com/site/programas/redacao- /noticia/2015/02/pesquisa-violencia-e-principal-causa-que-afasta- torcedores-dos-estadios.html.

Szymanski, S. (2001). Income Inequality, Competitive Balance and the Attractiveness of Teams Sports: Some Evidence and Natural Experiment from English Soccer. The Economic Journal, 111, F69-F84.

Szymanski, S. (2003). The economic design of sporting contests. Journal of Economic Literature, 41 (4), 1137-1187.

Szymanski, S. (2014). Fair is foul: A critical analysis of UEFA financial fair play. International Journal of Sports Finance, 9 (3), 218-229.

Szymanski, S. (2015). Money and Football: A Soccernomics Guide. Nation Books, New York.

196

Empirical analysis of broadcast demand, competitive balance, demand for tickets and revenue generation in Brazilian football market

Szymanski, S., & Kuypers, T. (1999). Winners and losers: The business strategy of football. London: Viking.

Szymanski, S., & Smith, R. (1997). The English Football Industry: profit, performance and industrial structure. International Review of Applied Economics, 11 (1), 135- 153.

Tainsky, S. (2010). Television Broadcast Ratings for National Football League Contests. Journal of Sports Economics, 11, 629-40.

Tainsky, S., & McEvoy, C. D. (2012). Television Broadcast Demand in Markets without Local Teams. Journal of Sports Economics, 13 (3), 250-65.

Tainsky, S., Salaga, S., & Santos, C.A. (2012). Determinants of Pay-Per-View Broadcast Viewership in Sports: The Case of the Ultimate Fighting Championship. Journal of Sport Management, 27, 43-58.

Theil, H. (1967). Economics and information theory. Chicago: McGraw-Hill.

Theodorakis, N. D., Alexandris, K., Tsigilis, N., & Karvounis, S. (2013). Predicting spectators’ behavioural intentions in professional football: The role of satisfaction and service quality. Sport Management Review, 16 (1), 85-96.

Trandel, G. A., & Maxcy, J. G. (2011). Adjusting winning percentage standard deviations and a measure of competitive balance for home advantage. Journal of Quantitative Analysis in Sports, 7 (1), 1-17.

Uhrich, S., & Benkenstein, M. (2010). Sport Stadium Atmosphere: Formative and Reflective Indicators for Operationalizing the Construct. Journal of Sport Management, 24, 211-237.

UNDP. (2014). Human Development Report 2014, Sustaining Human Progress: Reducing Vulnerabilities and Building Resilience, UN Plaza, New York, NY 10017, USA. Available from: http://hdr.undp.org/sites/default/files/hdr14- summary-en.pdf

Vrooman, J. (1995). A general theory of professional sports leagues. Southern Economic Journal, 61 (4), 971-990.

197

References

Vrooman, J. (1996). The baseball players labour market reconsidered. Southern Economic Journal, 63 (2), 339-360.

Wakefield, K. L., & Sloan, H. J. (1995). The effect of team loyalty and selected stadium factors on spectator attendance, Journal of Sport Management, 9 (3), 153-172.

Watanabe, Y., Matsumoto, K., & Nogawa, H. (2013). Variables Influencing Spectators’ Desire to Stay at A Professional Golf Tournament In . Contemporary Management Research, 9 (3), 283-298.

Zimbalist, A. (2002). Competitive balance in sports leagues: An introduction. Journal of Sports Economics, 3 (2), 111-121.

Zygmont, Z. X., & Leadley, J. C. (2005). When Is the Honeymoon Over? Major League Baseball Attendance 1970–2000. Journal of Sport Management, 19, 278-299.

198