Economics of Football, The - Stephen Dobson & John Goddard
Economics of Football, The - Stephen Dobson & John Goddard
The Pitt Building, Trumpington Street, Cambridge, United Kingdom
The Edinburgh Building, Cambridge CB2 2RU, UK
40 West 20th Street, New York, NY 10011-4211, USA
477 Williamstown Road, Port Melbourne, VIC 3207, Australia
Ruiz de Alarcn 13, 28014 Madrid, Spain
Dock House, The Waterfront, Cape Town 8001, South Africa
http://www.cambridge.org
Stephen Dobson and John Goddard 2004
First published in printed format 2001
ISBN 0-511-03353-2 eBook (Adobe Reader)
ISBN 0-521-66158-7 hardback
SD
JG
Contents
List of gures
List of tables
Preface
Acknowledgements
1 Introduction
1.1 The economics of team sports: three seminal contributions and the
subsequent development of the subject
1.2 Outline of this volume
1.3 The use of econometrics in this volume
page ix
xi
xv
xviii
1
3
11
16
22
24
38
56
71
80
87
90
101
109
123
125
126
150
167
187
189
191
203
209
221
227
236
vii
viii
Contents
List of references
Index
240
242
245
252
262
264
265
280
284
286
296
303
306
309
314
317
318
335
342
371
373
375
381
397
416
418
419
425
439
440
452
Figures
List of gures
278
279
288
356
395
396
398
399
Tables
2.1
2.2
2.3
2.4
2.5
2.6
2.7
2.8
2.9
2.10
2.11
2.12
2.13
2.14
2.15
2.16
2.17
2.18
xii
3.1
3.2
3.3
3.4
3.5
3.6
3.7
3.8
3.9
3.10
3.11
3.12
3.13
3.14
3.15
3.16
3.17
List of tables
151
152
154
154
155
156
157
159
160
161
162
163
168
180
183
184
184
185
186
192
193
194
List of tables
4.4
4.5
4.6
4.7
4.8
4.9
4.10
4.11
4.12
4.13
4.14
4.15
4.16
4.17
5.1
5.2
5.3
6.1
6.2
6.3
6.4
6.5
6.6
6.7
xiii
194
195
196
197
198
199
200
201
202
207
233
235
236
237
257
258
260
266
268
270
274
275
276
290
xiv
6.8
6.9
6.10
6.11
6.12
6.13
7.1
7.2
7.3
8.1
8.2
8.3
8.4
8.5
8.6
8.7
8.8
8.9
8.10
8.11
List of tables
293
294
299
300
308
312
322
348
369
378
390
392
406
406
408
411
412
412
412
415
Preface
xvi
Preface
team sports in general) represents an important sub-eld within economics. In the USA economists have been writing and publishing books and
scholarly articles on baseball, basketball and (American) football since
the mid-1950s. Consequently, the academic literature is dominated by
studies of the economics of North American sports. These writings shed
light on a wide range of important issues, including the determinants
of the compensation received by professional sports players, the nature of
joint production in team sports, competitive balance, uncertainty of
outcome and the distribution of playing talent in sports leagues and the
contribution of the coach or manager to team performance.
The common thread linking research into all of these topics is the
development and testing of economic hypotheses, using sports as a laboratory within which suitable data can be collected. A major attraction of
sports to empirical economists, indeed, is that the availability of data
permits investigation of a large number of economic propositions that
would be dicult to test in other areas, because of a lack of suitable data.
The professional football sector therefore oers opportunities for empirical research in areas such as consumer behaviour, labour economics and
industrial organisation, as well as more specialised topics of particular relevance to the economics of sports.
At this point, the readers attention is drawn to a small number of stylistic conventions that are adopted throughout the present volume.
Currently, the English league football season runs from August to May,
although in the past there have been minor variations; for many years the
league season was scheduled to nish at the end of April, but often overran owing to postponements caused by bad weather. Often the season
which began in August 1998 and nished in May 1999 is referred to as
the 19981999 season. In order to be concise, however, in this volume all
football seasons are referred to by their end-year only. The season which
ran from August 1998 to May 1999 is therefore known as the 1999
season.
Readers familiar with the recent history of English football will be
aware that an important organisational change took place during the
summer of 1992 (or, in present terminology, at the start of the 1993
season), with the creation of the Premier League. Previously, the Football
League was responsible for the organisation of league competition involving all 92 professional clubs in England andWales.The clubs were divided
into four divisions, with promotion and relegation of a few clubs (usually
between two and four) taking place between each division at the end of
each season. Between the 1959 and 1992 seasons, the divisions had
simply been numbered from 1 to 4. The decision of the Division 1 clubs
to withdraw from the Football League resulted in the formation of their
Preface
xvii
Acknowledgements
xviii
Introduction
Introduction
1.1 The economics of team sports: three seminal contributions and the subsequent development of the subject
Rottenberg:The baseball players labor market, Journal of Political
Economy, 1956
Rottenberg is widely credited with writing the rst academic analysis of
the economics of professional team sports.The paper was written at a time
when US professional baseball players contracts included a reserve
clause. Once he signed his rst one-year contract with a team in organised
baseball, a player ceased to be a free agent. On expiry of his present contract, his team retained the option to renew his contract for another year.1
This served to limit players freedom of movement, by binding them to
their present employers. Eectively, the reserve clause created a monopsony in the players labour market: each contracted player could negotiate with only one potential buyer of his services.The baseball authorities
defended the reserve clause on the grounds that it was necessary to ensure
an equal distribution of playing talent among opposing teams.Without the
reserve clause, the rich teams (with the largest potential markets) would
outbid the poorer ones for the best available players. This would tend to
reduce uncertainty of outcome and spectator interest in the league competition as a whole, and depress the attendances and revenues of all teams.
Rottenbergs contribution was to argue that free agency in the players
labour market would not necessarily lead to a concentration of the best
players in the richest teams. In other words, a reserve clause was not a
necessary condition to ensure competitive balance. Professional team
sports are intrinsically dierent from other businesses, in which a rm is
likely to prosper if it can eliminate competition and establish a position as
a monopoly supplier. In sports, it does not pay a rich team to accumulate
star players to the extent that (sporting) competition is greatly diminished, because of the joint nature of production in sports. Consequently,
a team that attempts to accumulate all of the best available playing talent
will nd at some stage that diminishing returns begin to set in: In baseball no team can be successful unless its competitors also survive and
prosper suciently so that the dierences in the quality of play among
teams are not too great. . . . At some point, therefore, a rst star player is
worth more to poor team B than, say, a third star to rich team A. At this
point, B is in a position to bid players away from A in the market
(Rottenberg, 1956, pp. 254, 255).
11
This was subject only to a rule that prevented the wage being cut by more than 25 per
cent in any one year if the team was in one of the major leagues.
Introduction
mobility and the entry and exit of clubs. In short, the league rather than
the individual team is the rm in the economists sense. A sports league
should be regarded as analogous to a multi-plant rm, in which the individual teams are plants, subject to decisions which are taken and implemented collectively at league level.
If the sports league is the rm in the economic sense, this raises the
question as to why it is unusual to observe direct competition between
rival leagues operating within the same sport. Although the National
League and American League do operate simultaneously in baseball,
analytically they should be regarded as one larger multi-league rm,
since they come together at the end of each season to produce the World
Series. Geographical division is a more common form of segmentation
though, according to Neale, one that is inherently unstable.Where direct
competition is prevented by geography, prot incentives tend to promote
enlargement and the elimination of geographical boundaries.2
Competition between dierent sports is more common than competition
between rival leagues within the same sport, though segmentation based
on nation, region, season of the year or even social class are still
common.3 All such forms of segmentation tend to inhibit direct
head-to-head competition.
Neale suggests that the general lack of competition between sports
leagues arises because the cost and demand characteristics of the market
for professional team sports tend to create conditions of natural monopoly, making it ecient for a single league to supply the entire market. On
the cost side, Neale suggests that the long-run average cost curve is probably horizontal. Although an increase in the scale of production might
entail the use of less ecient playing inputs, raising average costs, this
tends to be oset by an enthusiasm eect. If the sport operates on a
larger scale, public enthusiasm encourages more people to take up
playing, eventually raising the supply of players at the highest level. To
some extent, the enthusiasm eect makes supply and demand interdependent: if more people play the sport, more will also want to attend
matches at the professional level. Finally, the existence of rival leagues
would eectively break the monopsony power of teams as buyers of
12
13
This observation seems to be borne out by recent changes in the structure of competition in European football.
Traditionally, English cricket and rugby union were upper- or middle-class sports, while
football and rugby league were working-class sports. More recently, the class divisions
may have become more blurred, but have not disappeared altogether. Cricket is played
only in summer; rugby league switched from a winter to a summer calendar in the late
1990s; football and rugby union are still played in winter. Rugby league still has a strong
regional identity centred on Lancashire and Yorkshire. Cricket is popular throughout
England, but very much a minority sport in Scotland.
Introduction
playing services, enhancing players bargaining power in wage negotiations. This would tend to make costs higher than they are when one
league operates as a monopoly supplier.
On the demand side, Neale suggests that baseball teams produce a
number of streams of utility: directly for spectators who buy tickets for
seats in the stadium and for television viewers who watch the match at
home; and indirectly for everyone who enjoys following the championship race as it unfolds.The closer the competition, the greater is the indirect eect. For newspapers and television companies in particular, the
indirect eect is a marketable commodity that helps sell more of their
product. The size of the indirect eect depends on the scale and universality of the championship, and is therefore maximised when the league is
a monopoly supplier. Overall it is clear that professional sports are a
natural monopoly, marked by denite peculiarities both in the structure
and in the functioning of their markets (Neale, 1964, p. 14).
An important implication is that the peculiar economic characteristics
of professional sports leagues and their constituent teams should be recognised by legislatures, by the courts and by the general public, whenever
practices such as collective decision making or other (apparently) anticompetitive modes of behaviour come under scrutiny.
Sloane:The economics of professional football: the football club as a
utility maximiser, Scottish Journal of Political Economy, 1971
Sloanes paper questions Neales conclusion that the league rather than
the individual team or club is the relevant rm (or decision making unit)
in professional team sports. In the case of English football, for example,
the sports governing bodies merely set the rules within which clubs can
freely operate. Most economic decisions, such as how much money to
spend on stadium development and how many players to employ, are
made by the clubs. Although the total quantity of output (the number of
matches played by each club) is regulated, this clearly reects the clubs
common interest. In cartels, it is not unusual for rms to reach joint decisions concerning price or production, but this does not imply that the
cartel should be elevated to the theoretical status of a rm. In short,
Sloane suggests that Neales argument tends to overemphasise mutual
interdependence. The fact that clubs together produce a joint product is
neither a necessary nor a sucient condition for analysing the industry as
though the league was a rm (Sloane, 1971, p. 128).
Having argued that the club is the relevant economic decision maker,
Sloane goes on to raise a number of important questions concerning the
objectives of sports clubs. Implicit in the reasoning of both Rottenberg
and Neale is an assumption of prot maximisation. Despite the peculiarities of sports economics elucidated by Neale, the behaviour of professional sports teams is analysed within a very conventional analytical
framework. While this may be reasonable in the case of US professional
team sports, where many teams do have an established track record of
protability, Sloane suggests that it may not be universally applicable.
Through most of the history of English football, prot making clubs have
been very much the exception and not the rule. Chairmen and directors
with a controlling interest in football clubs are usually individuals who
have achieved success in business in other elds.Their motives for investing may include a desire for power or prestige, or simple sporting enthusiasm: a wish to see the local club succeed on the eld of play. In many
cases, prot or pecuniary gain seems unlikely to be a signicant motivating factor. If so, it may be sensible to view the objective of the football club
as one of utility maximisation subject to a nancial solvency constraint.
The nancial solvency constraint recognises that the benevolence of any
chairman or director must reach its limit at some point.
Non-prot maximising models of the rm had received considerable
attention in the economics literature during the decade prior to the publication of Sloanes article. A major drawback to the general introduction
of the utility maximisation assumption in the theory of the rm is that it
may be rationalised so that it is consistent with almost any type of behaviour and therefore tends to lack operational signicance (Sloane, 1971, p.
133).
In the case of football clubs, however, it is not too dicult to identify
several plausible and easily quantiable objectives. Sloane suggests the
following:
Prot. Sloanes expectation that prot is not the sole or even the most
important objective does not preclude its inclusion as one of a number
of arguments in the utility function.
Security. Simple survival may be a major objective for many clubs.
Decisions (concerning, for example, sales of players) may aim more at
ensuring security than at maximising playing success.
Attendance or revenue. A capacity crowd enhances atmosphere and a
sense of occasion, and may in itself be seen as a measure of success.
Recently, an increasing willingness to charge whatever ticket prices the
market will bear suggests that revenue (or prot) carry a heavier weight
in the utility function than in earlier periods, when it was usual to charge
the same price for all matches, irrespective of the level of demand.
Playing success. This is probably the most important objective of all, and
one to which chairmen, directors, managers, players and spectators can
all subscribe.
Introduction
subject to R 0 T
(1.1)
10
Nearly a quarter of a century later, and with the benet of hindsight after
witnessing a number of attempts at introducing such changes by the
North American sports governing bodies in the interim, Fort and Quirk
(1995) and Vrooman (1995) provided a compelling demonstration of the
usefulness of this framework for policy analysis in team sports.
Sports economists have often claimed that an abundance of easily
accessible data on the characteristics and performance of individual personnel makes professional team sports an ideal laboratory for the empirical scrutiny of economic theories and hypotheses that might prove
dicult to test elsewhere. An article by Scully (1974), which investigated
the relationship between the performance of individual players in US
Major League Baseball and their compensation, provides a classic
example of a contribution of this kind. Scullys motivation was to demonstrate that severe restrictions on the rights of the players to sell their services in the labour market (in the form of the reserve clause) led to
exploitation perpetrated by the teams as employers, in the sense that individual players wages were systematically below their contributions to
their teams revenue-earning capability. But the applicability of Scullys
methods and the relevance of his ndings went much further than this.
Identifying the links between pay, productivity and human capital is a task
of central importance towards achieving a much broader understanding
of how labour markets work, as well as insights into specic issues such as
inequality and discrimination in opportunity and compensation.
Developments elsewhere in the eld of economics are also capable of
providing valuable insights into the economics of sports.Two such examples are articles by Rosen (1981) on the economics of superstars, and by
Lazear and Rosen (1981) on the earnings of chief executives in large corporations. Both provide insights into an issue of major controversy in the
public debate about the economics of sports: the massively inated earnings of the leading stars. Rosen demonstrated how in certain professions
the technological conditions of supply create markets in which the top
performers can service very large audiences at little or no incremental
cost to themselves as the audience size increases.The outcome is that the
extra compensation of the top performers (in comparison with slightly
less talented rivals) is hugely disproportionate to their extra talent. Film
stars and musicians are obvious examples, but the model is relevant to
sports stars as well. Lazear and Rosen likened the high earnings of chief
executives to a lavish rst prize in a tournament.The possibility of emerging as a future prize winner provides incentives for others to strive for
success early in their careers, with obvious benets for the productivity of
the workforce as a whole. The internal labour market of a sports team,
with its squad of aspiring youngsters competing for rst-team places that
Introduction
11
12
minimum admission price and gate and television revenue sharing are
some of the most prominent measures of this kind, all of which are
described in chapter 2. The chapter concludes by reporting an empirical
investigation in which a number of standard economic measures of convergence, divergence and inequality are applied to an English club-level
league gate revenues data set.
Competitive balance and uncertainty of outcome are the main themes
of chapter 3. Following Rottenberg, a number of US economists have
developed theoretical models of competitive balance in sports league
competition, which formalise some of Rottenbergs original insights, and
permit exploration of a number of other policy issues. Naturally, the US
literature on this topic has developed primarily with North American professional team sports in mind. Chapter 3 begins with a description of the
North American model for the organisation and regulation of professional team sports. A number of features which dier signicantly from
their counterparts in the British or European model are highlighted.The
chapter then reviews the main ndings of the US literature, and considers
their relevance for the case of English football. Two key assumptions of
the US literature in particular seem questionable if these models are
applied directly to European football. Following Sloane, the rst is the
assumption that team owners are motivated solely by prot.The second is
the assumption that playing talent is drawn from a pool whose total size is
xed. Especially since the Bosman ruling in 1995 (see chapter 2),
European football labour markets have been open rather than closed,
with players frequently hired from outside the domestic league. In
chapter 3 the competitive balance model is adapted for English football
by introducing behavioural assumptions other than prot maximisation,
and by treating the labour market as open.
Closely balanced competition is important because it maximises
uncertainty of outcome or unpredictability: an essential characteristic of
any sports contest. The empirical sections of chapter 3 examine the
extent to which individual match results in English football are predictable or unpredictable. Patterns and trends in English league match
results and goal scoring records since the 1970s are identied and compared with equivalent data from a number of other European leagues.
The main long-term trend in the data is a progressive improvement in the
performance of away teams, and a corresponding decline in the importance of home advantage. Analysis of sequences of match results reveals
evidence of a negative persistence eect: a recent run of good results
appears to create either pressure or complacency, increasing the risk that
the next result will be bad. Chapter 3 concludes by developing a statistical model that processes the patterns in sequences of past match results,
Introduction
13
14
Introduction
15
world of highly lucrative television contracts, merchandising and sponsorship, gate revenues still constitute a signicant proportion of the total
revenues of most football clubs. Furthermore, spectators who attend
matches in person are not just passive consumers.Their presence contributes in a fundamental way to the quality of the product, by generating
atmosphere and a sense of occasion. Most spectators also seek to
inuence match outcomes though the eects of vocal encouragement and
criticism of players, managers and match ocials. For these reasons, as
well as the enthusiasm eect suggested by Neale, supply and demand
are really interdependent.
Identifying the causes of variations in attendances, from season to
season and from match to match, is therefore the main objective of
chapter 7.The chapter reviews the empirical literature on the demand for
football attendance, highlighting some of the main problems of model
specication, estimation and interpretation encountered by researchers
in this area. Sociological case studies from Spain and Scotland are also
used to illustrate the impact of broader social or cultural inuences on
patterns of demand for football attendance. Chapter 7 concludes by presenting an empirical analysis of English club-level attendances during the
post-Second World War period. The analysis shows the eects on attendances of factors such as short-term loyalty, team performance and
admission prices. It also investigates the relationship between various
socio-economic, demographic and football-related characteristics of each
club and its home town, and average attendances over the long term.
Neoclassical economics predicts that unco-ordinated interactions
between large numbers of self-interested buyers and sellers operating in
free, unregulated markets can normally be relied upon to deliver an
ecient allocation of scarce resources. A key assumption is that all relevant information about the prices and characteristics of goods and services is transmitted rapidly and accurately among all participants in the
relevant market. Chapter 8 presents two case studies, which focus on the
eciency with which relevant information about the performance of
football teams on the eld of play is impounded into the prices at which
trade takes place in particular markets.
The rst case study concerns the secondary market for trade in the
shares of football clubs quoted on the London Stock Exchange (LSE).
Following a description of the historical evolution of present-day structures of ownership and control in English club football, the empirical
investigation seeks to quantify the extent to which major events that take
place on the eld of play, such as European qualication and elimination,
or domestic promotion and relegation, carry implications for the share
prices of football clubs that are oated on the LSE. If new information
16
Introduction
17
(1.2)
18
to assume that the error term follows some specic probability distribution, such as the normal distribution.
The i-subscripts on the variables yi, x2i, x3i, . . . , xki and ui in (1.2) indicate that the relationship between the independent variables and the
dependent variable holds over a number of observations contained within
a sample of data.The observations are indexed over i1 . . . N, where N is
the total number of observations.5 The sample therefore comprises N
complete sets of observed numerical values of x2i, x3i, . . . , xki and yi. The
numerical values of the coecients 1,2, 3, . . . , k are all unknown, and
the purpose of the regression analysis is to process the data in order to
obtain the best possible numerical estimates of these coecients. The
choice of estimation method depends on the specication of the model.
Estimation methods used in this volume (and referred to in the text)
include Ordinary Least Squares (OLS), Generalised Least Squares
(GLS), Weighted Least Squares (WLS), Fixed Eects and Random
Eects (FE and RE), Maximum Likelihood (ML) and Seemingly
Unrelated Regressions estimation (SUR). Estimated regression
coecients are usually denoted b 1, b 2, b 3, . . . , b k, to distinguish them
from the corresponding true (but unknown) coecients 1, 2, 3, . . . ,
k.
Once a regression model has been estimated, the signs and numerical
magnitudes of the estimated coecients convey information about the
direction and strength of the relationship between each independent variable and the dependent variable. Because the estimated coecients are
based on a limited sample of data, there is always a suspicion of imprecision or unreliability attached to them. For this reason, estimation
methods for regression coecients provide a standard error with each
coecient estimate. The standard error reects the reliability of the estimate: the smaller the standard error, the greater the reliability. It can also
be used to construct a condence interval for the true value of the
coecient. A condence interval is simply a range of likely values for the
true coecient located on either side of the estimated coecient.We can
claim that the true coecient should lie somewhere within this range,
subject only to a small (and specied) probability of being wrong.
The ratio of an estimated coecient to its standard error, usually
known as a z-statistic or t-statistic (depending on which type of regression
model is being used), provides a convenient method for assessing whether
the estimation has succeeded in identifying a relationship that is reliable
15
In cases where data are observed over a number of dierent time periods, t-subscripts
are used instead of i-subscripts in order to denote time. For data with both a cross-sectional and a time series dimension, variables may be indexed using both i - and t-subscripts.
Introduction
19
20
Introduction
21
to econometrics, while Kennedy explains the terminology of econometrics in a clear, succinct and non-technical manner. Greene, meanwhile,
oers a more advanced treatment, including full technical details of a
large number of more highly specialised econometric models and estimation techniques.
23
fees received, provide the main contributions to the revenue side of the
football clubs prot and loss account. Historical trends in the size and
composition of English footballs revenue base are described in sections
2.4 and 2.5.
The main component of a football clubs costs is expenditure on
players, through both wages and transfer fee expenditure. Sections 2.6
and 2.7 provide a brief account of the historical development of English
footballs labour and transfer markets. Historically, restrictions on the
mobility of players, together with a net ow of transfer expenditure from
the stronger to the weaker clubs, were perceived to have played an important role in maintaining a certain degree of equality between the richer
and the poorer clubs, in both sporting and nancial terms. But, more
recently, the capacity of these mechanisms to function as they have in the
past has come under considerable strain. Particularly challenging to the
traditional organisation of the transfer market has been the landmark
Jean-Marc Bosman ruling in the European Court of Justice (ECJ) in
1995 (see below).
Together, sections 2.22.7 provide a comprehensive account of the historical development and present-day condition of English footballs commercial structure. Section 2.8 widens the perspective again, by providing
a brief description of the historical evolution of football into its presentday form in two other countries: France and Japan. The histories of professional football in these two countries provide some fascinating
contrasts, both with each other and with the English story that has been
told in previous sections of chapter 2.
The distribution of nancial resources between clubs is the main theme
of the nal, empirical part of chapter 2, in section 2.9. A number of economic measures of convergence and divergence are applied to an English
league revenues data set covering all seasons between the mid-1920s and
the late 1990s. The objective is to gauge the extent to which disparities
between the revenue bases of clubs operating at dierent levels within the
league have widened over time. Unsurprisingly, the convergence analysis
shows that inequalities within English football were considerably greater
by the end of the 1990s than they were in the mid-1920s. More surprising, perhaps, is the nding that almost all of the increase in inequality can
be attributed to two much shorter periods within the last 75 years. The
rst big rise in inequality happened over a period of less than 20 years
duration, between the late 1950s and the mid-1970s. The second period
of rising inequality began at the start of the 1990s, and is still underway at
the time of writing. These ndings tend to reinforce the interpretation of
English footballs economic history advanced in the earlier sections of
chapter 2.The 1960s and the 1990s in particular appear to be transitional
24
Figure 2.1 Average home attendances, rst-division clubs, ve European leagues, 2000 season
Source: <http://www.football.sportsites.co.uk/>.
England
Pa
ris
Sa
in
tG
er
ai
10,000
20,000
30,000
40,000
50,000
60,000
France
Germany
In
te
om
10,000
20,000
30,000
40,000
50,000
60,000
70,000
ila
La
zi
v
Ju
en
tu
Fi
e
or
nt
in
a
l
Bo
og
na
R
eg
in
a
r
To
in
Ba
Italy
ri
Pa
rm
a
U
di
ne
se
r
Ve
on
a
Le
cc
e
C
ag
lia
ri
Pe
ru
gi
a
a
Pi
ce
nz
a
n
Ve
ez
ia
30
31
revenues or budgets, the French league turns out to be the most egalitarian.The values of this statistic were 0.34 (France), 0.39 (Germany), 0.49
(Italy), 0.52 (England) and 0.79 (Spain).
By allowing out-of-contract players complete freedom of movement
within EU boundaries, the 1995 ECJ ruling in the Jean-Marc Bosman
case has had major implications for both mobility and levels of compensation within the footballers labour market. The case arose when Bosman
was oered new terms inferior to those of his previous contract by his
Belgian club, RFC Lige. Having declined the new contract, Bosman was
refused permission by Lige to join the French club Dunkerque. Bosman
sued Lige citing restraint of trade, took his case to the ECJ, and won. At
the same time the Court outlawed all restrictions on the eligibility of
European Union (EU) nationals to play elsewhere within the EU; previously the three-foreigner rule had prevented clubs from elding more
than three foreign nationals in the same team at the same time. This
change has enabled clubs to recruit players from beyond national boundaries to a far greater extent than ever before.Table 2.1 shows data on the
nationalities of all squad players with rst-division clubs in the big ve
European leagues at the end of the 1999 season.The same data are shown
in summary form in gure 2.2.
For France, Germany, Italy and Spain, gure 2.2 groups players into
ve broad categories by nationality: players who were home nationals of
each country; and players from the rest of Europe; Sub-Saharan Africa;
Central or South America; and the rest of the world. Because English
recruitment patterns are signicantly dierent, a dierent set of categories is used in this case. Of the four continental leagues, Germanys was
the most cosmopolitan, with German nationals accounting for only 58
per cent of all German club squad players. The corresponding percentages for Spain, Italy and France were 61 per cent, 68 per cent and 74 per
cent, respectively. Only 57 per cent of English squad players were from
England and Wales; but if players from Scotland and Northern Ireland
are also counted, the proportion of British Premier League squad players
rises to 64 per cent.
German and English clubs employed higher proportions of foreign
Europeans than clubs in the other three countries. Poland (16 players)
and Croatia (15) were the most heavily represented European countries
in Division 1 of the Bundesliga, but in general German clubs seem willing
to employ non-German Europeans from a very wide range of countries.
In contrast, employment of non-British Europeans by English clubs was
more highly concentrated by nationality: apart from the Irish Republic
(with 29 players), Norway (23), France (22), Italy (17) and Holland (15)
were all strongly represented. The Italian and Spanish proportions of
604
TOTAL
481
Italy
327
France
18
Croatia
11
Yugoslavia
8
Scandinavia
14
Rest of Europe
27
Central/South America 52
Sub-Saharan Africa
19
Rest of World
5
ITALY
494 TOTAL
Germany
289
Poland
16
Croatia
15
Holland
12
Hungary
10
Former USSR
10
Rest of Europe
81
Sub-Saharan Africa
23
Central/South America 15
Rest of World
23
GERMANY
419 TOTAL
France
311
Belgium
7
Yugoslavia
5
Rest of Europe
24
Sub-Saharan Africa
43
Central/South America 19
North Africa
9
Rest of World
1
346
44
29
22
17
15
41
44
12
12
22
England
Rest of UK
Ireland
France
Italy
Holland
Scandinavia
Rest of Europe
Sub-Saharan Africa
Central/South America
Rest of World
TOTAL
FRANCE
ENGLAND
Table 2.1. Nationalities of squad players at rst-division clubs, ve European leagues, summer 1999
TOTAL
527
Spain
323
Argentina
47
Brazil
21
Uruguay
15
Rest of Central/South 13
America
France
13
Yugoslavia
12
Portugal
10
Rest of Europe
58
Sub-Saharan Africa
12
Rest of World
3
SPAIN
33
Rest of World
8
England
Rest of Europe
16
Scandinavia
7
Ireland
5
Rest of UK
7
France
Rest of World
2
Central/South America
5
Sub-Saharan Africa
10
Rest of Europe
9
France
74
34
Germany
Central/South America
3
Sub-Saharan Africa
5
Rest of Europe
29
Germany
58
Italy
Central/South America
11
Rest of World
1
Sub-Saharan Africa
4
Rest of Europe
16
Italy
68
35
Rest of World
1
Central/South America
18
Sub-Saharan Africa
2
Rest of Europe
18
Spain
61
36
ENGLAND
Alan Shearer
Newcastle
11United
Michael Owen
Liverpool
Duncan Ferguson Newcastle
11United
Chris Sutton
Chelsea
Marcel Desailly
Chelsea
Kanu
Arsenal
Patrick Viera
Arsenal
Emmanuel Petit
Arsenal
FRANCE
Fabien Barthez
Sonny Anderson
Robert Pires
Marco Simone
Fabrizio Ravinelli
Ali Bernarbia
Monaco
Lyon
Marseille
Monaco
Marseille
Paris St
Germain
28,000
24,000
21,000
21,000
20,000
20,000
20,000
20,000
42,307
26,923
26,923
23,076
23,076
23,076
GERMANY
Stefan Eenberg
Bayern
11Mnchen
Krassimir Balakov Stuttgart
Giovane Elber
Bayern
11Mnchen
Andreas Moller
Dortmund
Ze Roberto
Leverkusen
Fredi Bobic
Dortmund
Ciriaco Sforza
Kaiserslauten
48,076
48,076
32,692
30,769
28,846
25,000
25,000
ITALY
Alessandro del
11Piero
Christian Vieri
Ronaldo
Juan Veron
Filippo Inzaghi
Alessandro Nesta
Gabriel Batistuta
Roberto Baggio
George Weah
Pavel Nedved
Oliver Bierho
Francesco Totti
Edgar Davids
SPAIN
Steve
11McManaman
Patrick Kluivert
Nicolas Anelka
Jimmy Floyd
11Hasselbaink
Roberto Carlos
Raul
Jari Litmanen
Ronald De Boer
Frank De Boer
Rivaldo
Luis Figo
Juventus
69,230
Internazionale
Internazionale
Lazio
Juventus
Lazio
Fiorentina
Internazionale
AC Milan
Lazio
AC Milan
Roma
Juventus
55,769
50,000
48,076
41,208
40,384
34,615
34,340
34,340
32,692
26,923
26,923
26,923
Real Madrid
65,384
Barcelona
57,692
Real Madrid
55,769
Atltico Madrid 44,230
Real Madrid
Real Madrid
Barcelona
Barcelona
Barcelona
Barcelona
Barcelona
40,000
40,000
38,461
25,000
25,000
25,000
25,000
Note:
Data are weekly earnings after tax converted into sterling. All gures are approximate
owing to uctuating exchange rates. They do not include match bonuses, signing-on fees
or endorsements.
Source: World Soccer (November 1999).
Newcastle United). Table 2.3 charts the upward progress of the world
record transfer fee since the rst 1 million transfer of Guiseppe Savoldi
from Bologna to Napoli in 1975.Table 2.3 also lists the 10 highest transfer fees paid worldwide, as of summer 2000. Only once since 1975 has an
English club broken the ItalianSpanish monopoly of the world record
37
Table 2.3. World record transfers and 12 highest transfers, summer 2000
Player
Selling club
Buying club
Fee
(million)
Year
Bologna
Juventus
Boca Juniors
Barcelona
PSV Eindhoven
Fiorentina
Marseille
Sampdoria
Torino
Blackburn Rovers
Barcelona
So Paulo
Lazio
Parma
Barcelona
Napoli
Vicenza
Barcelona
Napoli
Milan
Juventus
Milan
Juventus
Milan
Newcastle United
Internazionale
Real Betis
Internazionale
Lazio
Real Madrid
11.2
11.75
13
15
16
18
10
12
13
15
19
22
31
36
37
1975
1978
1982
1984
1987
1990
1992
1992
1996
1996
1998
1999
1999
2000
2000
Real Madrid
Lazio
Internazionale
Barcelona
Real Madrid
Paris St Germain
Roma
Real Betis
Parma
Lazio
Internazionale
Atltico Madrid
37
36
31
25
23.5
22
22
22
21
20
19
19
2000
2000
1999
2000
1999
2000
2000
1998
1999
1999
1997
1998
38
39
a landmark meeting of representatives of a number of London and suburban clubs in October 1863, English footballs present-day governing
body, the Football Association (FA), was established (Walvin, 1994).
English footballs two most durable and important club competitions
both emerged soon afterwards. Fifteen clubs rst contested the FA Cup
during the winter of 18712. The Wanderers, a team comprising players
who had attended the leading public schools and Oxford or Cambridge
Universities, defeated the Royal Engineers in the rst nal, watched by
2,000 spectators. During the 1880s, footballs geographical centre of
power shifted away from the predominantly southern ex-public school
and ex-university clubs, towards teams based in manufacturing towns
and cities in the midlands and the north west. Enlightened factory
owners and employers, many of whom were themselves graduates of the
public school system, began to see the benets for workplace morale and
productivity of regular Saturday afternoon holidays, which created the
opportunity for the development of organised forms of working-class
leisure activity. New football clubs began to form throughout England,
often at the instigation of local church leaders, during the late 1870s and
1880s. A northern team, Blackburn Olympic, won the FA Cup for the
rst time in 1883, followed by Blackburn Rovers, three times winners
between 1884 and 1886.
The principle of professionalism was accepted by the Football
Association in July 1885. Its recognition was a key development in the
processes leading to the eventual formation of the Football League in
1888. Twelve teams were members of the league during its rst three
seasons, with Preston North End the inaugural champions in the 1889
and 1890 seasons, followed by Everton in 1891. During the next three
decades, membership of the Football League expanded progressively. In
the 1893 season, a second division of 12 teams was created, while membership of the top division was expanded to 16. Several of the leading
names of the modern era made their debut appearances in the 1890s,
including Manchester United (then Newton Heath) in the 1893 season
and Liverpool in the 1894 season. Woolwich Arsenal was the rst southern club admitted, also in the 1894 season. Total membership of the two
divisions was expanded to 32 teams in the 1895 season, then 36 in the
1899 season and 40 in the 1906 season.
The league nally achieved a membership of magnitude comparable to
the combined strength of the present Premier League and Football
League (92 teams) shortly after the First World War. In the 1921 season,
22 teams from the Southern League were added to the existing twodivision membership (which had by then grown to 44 teams) to form the
new Division 3 (South). In the 1922 season a new Division 3 (North) was
40
41
to keep their best players fresh for the more important league and
European xtures.1
Regular competitive football at European level was rst introduced in
1955. Birmingham City and a combined London team both competed in
the rst Fairs Cup tournament (the predecessor of the UEFA Cup)
staged between 1955 and 1958. Meanwhile, Manchester United was the
rst English team to enter the European Cup in the 1957 season. By the
1961 season, both the Fairs Cup and a new European CupWinners Cup
were also operating on an annual basis. From then until 1985, when
English clubs were excluded from European tournaments following the
hooligan-related Heysel stadium disaster at the European Cup Final in
Brussels, the three European competitions attracted a combined quota of
up to eight leading English clubs per season. Englands ban was rescinded
in the 1991 season. Since then, European competition has acquired an
increasingly prominent position in the xture lists of the leading clubs.
The importance of regular European participation has been enhanced by
rapid growth in the nancial rewards available from increasingly lucrative
television contracts. During the 1990s, a league format was introduced for
the early stages of the European Cup, which was renamed the Champions
League. This removed the threat of early elimination and guaranteed all
entrants a certain minimum number of xtures. In the 1999 season, 32
teams, including two or three from each of the leading European domestic
leagues (rather than just the domestic champions) gained entry to the
Champions League. Some teams (the champions and runners-up from
the major football powers) enter the league direct, while others (thirdplaced teams from the major powers, and champions from lesser powers)
participate in a qualifying tournament, which provides a means of eliminating some of the weaker aspirants before the main competition gets
underway. The 1999 season also saw the consolidation of the UEFA and
CupWinners Cups into a single UEFA Cup tournament.
Team performance
Although the main focus of this chapter is on the nancial rather than the
sporting performance of football clubs, it is obvious that there are direct
linkages between the two. At the micro level, a clubs capacity to generate
revenue depends on its teams success on the eld of play. On the other
11
Manchester Uniteds withdrawal from the 2000 seasons FA Cup, in order to take part in
the inaugural FIFA World Club Championship in January 2000, is a clear sign that the
older and traditionally more prestigious FA Cup competition is subject to similar pressures.
42
Davies, Downward and Jackson (1995) and Dobson and Goddard (1998a) report statistical analyses of the two-way link between team performance and nancial performance,
for rugby league and for football, respectively.
43
Arsenal 5
Everton 2
Man City 1
She Wed 1
Sunderland 1
19309
Championships
Hudderseld 3
Liverpool 2
Burnley 1
Everton 1
Newcastle 1
She Wed 1
WBA 1
19209
Man Utd 19
Wolves 16
Arsenal 8
Tottenham 8
Portsmouth 7
Preston 5
Blackpool 3
Chelsea 3
Liverpool 3
WBA 2
Burnley 1
Derby 1
Hudderseld 1
Sunderland 1
Man Utd 3
Wolves 3
Arsenal 2
Portsmouth 2
Chelsea 1
Liverpool 1
Tottenham 1
194759
Man Utd 10
Liverpool 9
Tottenham 8
Burnley 7
Leeds 7
Everton 5
Ipswich 3
Man City 3
Wolves 3
Nottm Forest 2
She Wed 2
Chelsea 1
Liverpool 2
Man Utd 2
Burnley 1
Everton 1
Ipswich 1
Leeds 1
Man City 1
Tottenham 1
19609
Liverpool 19
Leeds 10
Derby 7
Arsenal 5
Nottm Forest 5
Everton 4
Ipswich 2
Man City 2
QPR 2
Chelsea 1
Man Utd 1
Tottenham 1
WBA 1
Liverpool 4
Derby 2
Arsenal 1
Everton 1
Leeds 1
Nottm Forest 1
19709
Table 2.4. Most successful English league teams in league and cup competition, 19201999
Liverpool 24
Everton 8
Man Utd 6
Ipswich 5
Arsenal 4
Aston Villa 3
Nottm Forest 3
Southampton 2
Tottenham 2
Watford 2
West Ham 1
Liverpool 6
Everton 2
Arsenal 1
Aston Villa 1
19809
Man Utd 21
Arsenal 9
Liverpool 7
Blackburn 5
Newcastle 5
Aston Villa 4
Leeds 3
Chelsea 1
Crystal Pal 1
Norwich 1
Nottm Forest 1
She Wed 1
Tottenham 1
Man Utd 5
Arsenal 2
Blackburn 1
Leeds 1
Liverpool 1
19909
Arsenal 2
Everton 1
Man City 1
Newcastle 1
Portsmouth 1
Preston 1
She Wed 1
Sunderland 1
WBA 1
Newcastle 3
Arsenal 1
Aston Villa 1
Bolton 1
Blackpool 1
Charlton 1
Derby 1
Man City 1
Man Utd 1
Nottm Forest 1
WBA 1
Wolves 1
FA Cup Wins
Bolton 3
Aston Villa 1
Blackburn 1
Cardi 1
Hudderseld 1
Newcastle 1
She Utd 1
Tottenham 1
Leeds 1
Man Utd 1
Newcastle 1
Tottenham 1
West Ham 1
Aston Villa 1
Birmingham 1
Chelsea 1
Leeds 1
Leicester 1
Norwich 1
QPR 1
Swindon 1
WBA 1
Tottenham 3
Everton 1
Liverpool 1
West Ham 1
Man City 1
Man Utd 1
WBA 1
Wolves 1
Liverpool 4
Chelsea 1
Leeds 1
Man City 1
Nottm Forest 1
Tottenham 1
Aston Villa 2
Man City 2
Nottm Forest 2
Tottenham 2
Stoke 1
Wolves 1
Arsenal 2
Chelsea 1
Ipswich 1
Leeds 1
Liverpool 1
Man Utd 1
Southampton 1
Sunderland 1
West Ham 1
Liverpool 2
Aston Villa 1
Everton 1
Ipswich 1
Nottm Forest 1
Tottenham 1
Liverpool 4
Arsenal 1
Luton 1
Norwich 1
Nottm Forest 1
Oxford 1
Wolves 1
Man Utd 2
Liverpool 2
Tottenham 2
Coventry 1
Everton 1
West Ham 1
Wimbledon 1
Man Utd 2
Arsenal 1
Chelsea 1
Aston Villa 2
Arsenal 1
Chelsea 1
Leicester 1
Liverpool 1
Man Utd 1
Nottm Forest 1
She Wed 1
Tottenham 1
Man Utd 4
Arsenal 2
Chelsea 1
Everton 1
Liverpool 1
Tottenham 1
46
19609
19709
19809
19909
St Etienne 4
Monaco 2
Nantes 2
Stade Reims 2
St Etienne 4
Marseille 2
Nantes 2
Monaco 1
Strasbourg 1
Bordeaux 3
Monaco 2
Nantes 2
Marseille 1
Paris SG 1
St Etienne 1
Marseille 4
Auxerre 1
Bordeaux 1
Lens 1
Monaco 1
Nantes 1
Paris SG 1
Bordeaux 13
Nantes 12
Monaco 10
Paris SG 6
St Etienne 6
Marseille 5
Sochaux 3
Roubaix 2
Auxerre 1
Montpellier 1
Toulouse 1
Marseille 16
Monaco 11
Paris SG 11
Auxerre 5
Bordeaux 5
Lens 3
Lyon 3
Metz 2
Nantes 4
47
19709
19809
19909
B Mnchengbach 5
B Mnchen 3
Hamburg 1
Kln 1
B Mnchen 6
Hamburg 2
Stuttgart 1
Werder Bremen 1
B Mnchen 4
B Dortmund 2
Kaiserslauten 2
Stuttgart 1
Werder Bremen 1
B Mnchen 21
Hamburg 14
Werder Bremen 10
Kln 6
Stuttgart 6
B Mnchengbach 2
Bayer dingen 1
B Mnchen 20
B Dortmund 9
Kaiserslauten 8
B Leverkusen 6
Werder Bremen 6
Eintr. Frankfurt 3
Kln 2
Stuttgart 2
Freiburg 1
Hertha Berlin 1
Schalke 04 1
19609
19709
19809
19909
Championships
AC Milan 4
Juventus 3
Torino 3
Internazionale 2
Fiorentina 1
Internazionale 3
Juventus 3
AC Milan 2
Bologna 1
Fiorentina 1
Juventus 5
AC Milan 1
Cagliari 1
Internazionale 1
Lazio 1
Torino 1
Juventus 4
AC Milan 2
Internazionale 2
Napoli 1
Roma 1
Verona 1
AC Milan 4
Juventus 3
Napoli 1
Sampdoria 1
Juventus 18
Roma 11
Internazionale 9
Napoli 9
AC Milan 5
Fiorentina 3
Verona 3
Torino 2
AC Milan 19
Juventus 15
Internazionale 7
Lazio 5
Parma 4
Sampdoria 4
Napoli 3
Fiorentina 1
Torino 1
Udinese 1
48
19609
19709
19809
19909
Championships
Rangers 7
Hibernian 3
Aberdeen 1
Celtic 1
Hearts 1
Celtic 4
Rangers 3
Dundee 1
Hearts 1
Kilmarnock 1
Celtic 7
Rangers 3
Celtic 4
Aberdeen 3
Rangers 2
Dundee Utd 1
Rangers 9
Celtic 1
Celtic 23
Aberdeen 16
Rangers 9
Dundee Utd 7
Hearts 4
St Mirren 1
Rangers 29
Celtic 12
Aberdeen 9
Hearts 4
Motherwell 3
Dundee Utd 1
Hibernian 1
St Johnstone 1
19609
19709
19809
19909
Real Madrid 8
Atltico
11Madrid 1
Barcelona 1
Real Madrid 5
Atltico
11Madrid 3
Barcelona 1
Valencia 1
Real Madrid 5
Athletic
11Bilbao 2
Real Sociedad 2
Barcelona 1
Barcelona 6
Real Madrid 3
Atltico
11Madrid 1
Real Madrid 22
Barcelona 12
Real Sociedad 10
Athletic
11Bilbao 8
Atltico
11Madrid 5
Espaol 1
Sporting Gjon 1
Valencia 1
Barcelona 22
Real Madrid 16
Atltico
11Madrid 6
D La Corua 6
Valencia 4
Athletic
11Bilbao 2
Real Betis 1
Real Mallorca 1
Real Sociedad 1
Real Zaragoza 1
49
teams overall for the entire post-1945 period, obtained by aggregating the
points awarded for top three nishes. Because there were variations
between countries in the date of resumption of the national league competition after 1945, the points scores in table 2.6 have been converted to
percentages, to permit meaningful comparisons of championship dominance between countries.
By the criterion of the number of teams that have achieved a top three
nish during the post-1945 period, the English league turns out to be the
most democratic of the six that are analysed in tables 2.42.6. In total, 29
dierent English teams have achieved top three nishes. Several of these
appeared in the top three for the rst time during the 1980s
(Southampton, Watford, West Ham) or 1990s (Crystal Palace and
Norwich). France with 25 dierent top three nishers, and Germany
with 20 are the next highest on this criterion. The German total is based
on a signicantly smaller number of seasons, however, the West German
Bundesliga having only resumed on a national basis in the 1964 season.
Italy, Scotland and Spain, with 18, 17 and 15 teams, respectively, had the
smallest numbers of top three nishers.
In terms of the success achieved by the most successful teams during
the post-1945 period, the English league has clearly been dominated to a
greater extent by Liverpool and Manchester United than has France by
its most successful teams. France, indeed, seems to be something of a
special case, with no team apparently capable of maintaining a dominant
position for more than a few years at a time. Stade Reims, the most successful team of the 1950s and early 1960s and twice European Cup
runners up to Real Madrid, were eventually relegated in 1967 and have
not returned to the First Division. The dominance of St Etienne, champions 10 times between 1957 and 1981 and a major force in European
competition during the 1970s, had also run its course by the start of the
1980s, although the club did manage to preserve First Division status
until 1996. Spells of success for Bordeaux and Marseille in the 1980s and
1990s ended abruptly in scandal and enforced relegation in both cases.
Paris Saint Germain (PSG), currently the wealthiest French club, was
formed only in the early 1970s and therefore has no historical tradition of
championship success.
Elsewhere, the most successful teams in Germany, Italy, Scotland and
Spain have enjoyed greater championship dominance than have
Liverpool and Manchester United in England. The clear lead over all
rivals achieved by Germanys Bayern Mnchen in table 2.6 is untypical;
in the other ve countries the scores of the rst and second most successful teams are much closer together (although in Scotland a signicant gap
between Rangers and Celtic has developed in recent seasons). Bayern
St Etienne 11.7
Nantes 11.4
Marseille 11.1
Monaco 10.2
Bordeaux 9.6
Stade Reims 8.6
Nice 5.6
Paris SG 5.2
Lille 4.6
Lens 3.4
Nmes 2.8
Sochaux 2.5
Auxerre 1.9
RC Paris 1.9
Roubaix 1.9
Lyon 1.5
Strasbourg 1.5
Metz 0.9
Toulouse 0.9
Angers 0.6
Sedan 0.6
Valenciennes 0.6
Bastia 0.3
Le Havre 0.3
Montpellier 0.3
Liverpool 19.5
Man Utd 17.9
Arsenal 8.2
Leeds 6.3
Tottenham 6.3
Wolverhampton 6.0
Everton 5.3
Nottm Forest 3.5
Ipswich 3.1
Burnley 2.5
Derby 2.5
Aston Villa 2.2
Portsmouth 2.2
Chelsea 1.9
Blackburn 1.6
Man City 1.6
Newcastle 1.6
Preston 1.6
Blackpool 0.9
She Weds 0.9
West Brom 0.9
QPR 0.6
Southampton 0.6
Watford 0.6
Crystal Palace 0.3
Hudderseld 0.3
Norwich 0.3
Sunderland 0.3
West Ham 0.3
B Mnchen 27.3
B Mncheng'bach 11.1
W Bremen 9.7
Hamburg 8.8
Kln 8.3
B Dortmund 6.0
Stuttgart 5.1
Kaiserslauten 4.2
B Leverkusen 2.8
Hertha Berlin 2.8
1860 Mnchen 2.3
Entracht Frankfurt 2.3
Schalke 04 2.3
E Braunschweig 1.9
Nrnberg 1.4
Alemannia 0.9
Duisberg 0.9
Dsseldorf 0.9
B dingen 0.5
Freiburg 0.5
GERMANY
SCOTLAND
Rangers 33.6
Celtic 24.8
Aberdeen 11.9
Hibernian 7.9
Hearts 7.5
Kilmarnock 4.1
Dundee Utd 3.1
Dundee 1.9
Motherwell 1.2
Partick 0.9
Dunfermline 0.6
East Fife 0.6
St Johnstone 0.6
Clyde 0.3
St Mirren 0.3
Third Lanark 0.3
ITALY
Juventus 26.2
AC Milan 22.4
Inter 16.1
Fiorentina 6.3
Torino 6.3
Napoli 6.0
Roma 4.1
Lazio 3.2
Bologna 1.6
Cagliari 1.6
Parma 1.3
Sampdoria 1.3
Udinese 0.9
Verona 0.9
Perugia 0.6
Vicenza 0.6
Modena 0.3
Padova 0.3
SPAIN
Notes:
Scores are calculated by awarding three points for a championship win, two for second place and one for third place. Totals are converted to percentages.
See also notes to Table 2.2.
FRANCE
ENGLAND
51
52
53
1 to Group 5), using three simple criteria to dene the clubs that make up
each group:
the clubs home-town population recorded in the 1961 Census of
Population
the date of the clubs initial entry into the league
the clubs geographical location (for which there are two broad categories: south and midlands/north).
Figure 2.3 shows the percentage share of the clubs in each of the ve
groups in an aggregate performance score, calculated by awarding 92
points to the club which nished rst in Division 1 (or the Premier
League), 91 points to the club which nished second, and so on, in each
season. Minor adjustments to the points system are made for seasons
when the number of teams in the league was greater or less than 92.
Teams nishing in equal positions in the old Division 3 (South) and
Division 3 (North) are awarded equal points: rst position in either division gains 48 points, second gains 46 points, and so on.
The relative constancy of Group 1s performance score emphasises the
fact that the dominance of the clubs from the largest cities has been a consistent feature throughout the leagues history, and is by no means a
uniquely recent phenomenon. In contrast, the Group 2 clubs from the
next city size-band, which include three championship winners from the
1920s, 1930s and 1950s (Newcastle United, Sunderland and
Wolverhampton Wanderers), experienced a signicant decline in their
performance as a group between the late 1930s and late 1950s. Since
then, the average performance of the Group 2 clubs has also been quite
stable.
Group 3 clubs enjoyed a steady and sustained improvement in their
average performance between the 1920s and the end of the 1980s. Their
progressive advance represents the most important long-term shift in the
geographical balance of football power. Since the Second World War, the
majority of clubs gaining admission to the league have been located in
the south: Colchester United (admitted in the 1951 season), Peterborough
United (1961), Oxford United (1963), Cambridge United (1971),
Hereford United (1973),Wimbledon (1978), Maidstone United (1990),
Barnet (1992) and Wycombe Wanderers (1994).3 More important,
however, has been a progressive rise in the status of many of the southern
clubs which entered the league during its early 1920s expansion or later,
and which have been engaged in a catching-up process during much of
13
Post-war entrants to the league from the north and midlands were Scunthorpe United
and Shrewsbury Town (both in the 1951 season), Wigan Athletic (1979), Scarborough
(1988) and Maccleseld Town (1998).These lists do not include several clubs that temporarily exited and re-entered.
55
56
57
Total
D1/
Prem
1922
1923
1924
1925
1926
1927
1928
1929
1930
1931
1932
1933
1934
1935
1936
1937
1938
1939
1947
1948
1949
1950
1951
1952
1953
1954
1955
1956
1957
1958
1959
1960
1961
1962
1963
1964
1965
1966
1967
1968
1969
1970
25.6
23.6
23.3
23.1
23.1
23.4
23.5
23.9
22.9
21.0
21.8
21.4
22.5
23.1
24.7
26.4
27.9
27.0
35.4
40.2
41.0
40.6
39.5
39.0
37.1
36.4
34.1
33.2
32.7
33.5
33.7
32.5
28.6
28.0
28.8
28.5
27.6
27.2
28.9
30.1
29.2
29.5
12.5
10.8
10.8
10.4
10.4
10.6
10.6
10.5
10.5
9.5
9.9
9.5
10.4
10.8
11.4
11.4
11.6
11.5
14.9
16.7
17.9
17.4
16.7
16.1
16.1
16.1
15.1
14.2
13.8
14.4
14.7
14.4
12.9
12.1
12.4
12.5
12.7
12.5
14.2
15.3
14.5
14.8
D2/
One
6.1
6.2
5.9
6.2
6.1
6.5
6.9
6.7
6.3
6.1
5.6
6.1
5.8
6.1
6.9
8.0
8.6
8.6
11.0
12.3
11.1
11.7
10.8
11.1
9.7
9.7
9.0
9.1
8.7
8.6
8.6
8.4
7.0
7.5
7.4
7.6
7.0
6.9
7.3
7.4
7.4
7.6
Percentage shares
D3/
Two
D4/
Three
D1/
Prem
D2/
One
D3/
Two
D4/
Three
4.4
4.2
4.1
4.1
3.9
3.8
3.6
4.0
3.9
3.2
3.9
3.5
3.8
3.8
4.0
4.4
4.6
4.0
5.6
6.7
7.0
7.1
7.3
6.9
6.7
6.3
6.0
5.7
5.6
6.1
6.1
5.7
4.8
5.2
5.7
5.4
4.4
4.8
4.4
4.0
4.3
4.2
2.5
2.4
2.5
2.3
2.6
2.5
2.5
2.7
2.3
2.3
2.2
2.3
2.5
2.4
2.4
2.6
3.1
2.9
3.9
4.6
5.0
4.4
4.8
4.9
4.7
4.2
4.0
4.3
4.6
4.3
4.2
4.0
3.9
3.3
3.2
3.0
3.5
3.0
3.0
3.4
3.0
2.9
48.8
45.6
46.1
45.2
45.3
45.2
44.9
43.9
45.6
45.0
45.7
44.6
46.5
46.8
46.1
43.0
41.7
42.4
42.1
41.5
43.7
42.8
42.2
41.3
43.2
44.4
44.2
42.7
42.2
43.1
43.6
44.4
45.3
43.1
43.3
43.8
46.0
45.8
49.2
50.8
49.8
50.2
23.9
26.4
25.5
27.0
26.6
27.9
29.3
28.0
27.5
28.9
25.9
28.4
25.8
26.4
28.0
30.3
30.9
31.9
31.1
30.5
27.1
28.8
27.3
28.4
26.0
26.7
26.4
27.3
26.6
25.8
25.6
25.8
24.5
26.6
25.8
26.6
25.3
25.4
25.1
24.7
25.2
25.7
17.4
17.8
17.7
17.7
16.8
16.1
15.2
16.6
17.1
15.3
18.0
16.3
16.8
16.4
16.1
16.8
16.5
14.8
15.9
16.5
17.0
17.5
18.5
17.8
18.0
17.4
17.6
17.1
17.1
18.2
18.2
17.6
16.7
18.6
19.7
19.0
16.0
17.6
15.3
13.4
14.8
14.2
9.9
10.1
10.6
10.1
11.4
10.8
10.6
11.5
9.8
10.8
10.3
10.7
11.0
10.5
9.8
9.9
11.0
10.8
10.9
11.4
12.2
10.9
12.0
12.5
12.7
11.5
11.9
12.9
14.0
12.9
12.6
12.3
13.5
11.7
11.3
10.6
12.7
11.1
10.3
11.1
10.2
9.9
58
Total
D1/
Prem
1971
1972
1973
1974
1975
1976
1977
1978
1979
1980
1981
1982
1983
1984
1985
1986
1987
1988
1989
1990
1991
1992
1993
1994
1995
1996
1997
1998
1999
28.2
28.7
25.4
25.0
25.6
24.9
26.0
25.4
24.5
24.6
21.9
20.0
18.8
18.3
17.8
16.5
17.4
18.0
18.5
19.5
19.5
20.4
20.6
21.7
21.9
21.9
22.8
24.8
25.4
14.0
14.5
14.0
13.1
12.0
13.1
13.6
13.3
12.7
12.2
11.4
10.4
9.3
8.7
9.7
9.0
9.1
8.1
7.8
7.9
8.6
10.0
9.7
10.7
10.7
10.5
10.8
11.1
11.6
Percentage shares
D2/
One
D3/
Two
D4/
Three
D1/
Prem
D2/
One
D3/
Two
D4/
Three
7.1
6.8
5.6
6.3
7.5
5.8
6.2
6.5
6.1
6.1
5.2
4.8
5.0
5.4
4.0
3.6
4.2
5.3
5.8
6.9
6.3
5.8
5.9
6.5
6.6
6.6
6.9
8.3
7.5
4.4
4.7
3.7
3.4
4.1
3.9
4.1
3.3
3.4
4.0
3.6
2.8
2.9
2.7
2.7
2.5
2.4
2.8
3.0
2.8
2.8
3.0
3.5
3.0
3.2
2.9
3.2
3.5
4.1
2.8
2.7
2.1
2.2
1.9
2.1
2.1
2.3
2.3
2.4
1.7
2.0
1.6
1.6
1.4
1.4
1.7
1.8
1.8
1.9
1.8
1.6
1.5
1.6
1.4
2.0
1.9
1.8
2.1
49.6
50.5
55.0
52.3
47.0
52.6
52.2
52.2
51.7
49.4
52.0
52.1
49.6
47.4
54.7
54.8
52.6
45.1
42.3
40.5
44.2
49.0
47.2
49.1
48.7
47.7
47.4
45.0
45.8
25.1
23.6
22.1
25.3
29.5
23.3
23.9
25.5
25.1
24.8
23.6
23.8
26.5
29.2
22.6
21.5
24.0
29.7
31.6
35.4
32.2
28.5
28.5
29.9
30.2
30.0
30.5
33.7
29.6
15.5
16.4
14.7
13.7
16.0
15.8
15.8
13.1
13.8
16.2
16.6
14.1
15.7
14.9
15.0
15.1
13.5
15.3
16.4
14.4
14.6
14.7
16.9
13.7
14.8
13.3
14.0
14.2
16.3
9.8
9.5
8.3
8.7
7.5
8.3
8.1
9.2
9.4
9.5
7.7
10.0
8.3
8.5
7.7
8.5
9.9
9.9
9.7
9.7
9.0
7.8
7.5
7.3
6.2
9.0
8.1
7.1
8.3
Note:
Up to and including the 1958 season, D3 is Division 3 (South) and D4 is Division 3
(North).
Sources: Tabner (1992); Rothmans (various years).
59
England
Figure 2.5 Trends in average home-league attendances, selected clubs, four European leagues
Sources: <http://rernes.free.fr/>; Rothmans.
Scotland
1976
1978
1980
1982
1984
1986
1988
1990
France
1992
1994
1996
1998
Hamburg
Germany
Werder
Bremen
Kln
Bayern
Mnchen
65
Aston Villa, Chelsea, Everton and Manchester United shown in panel (a).
So too is the 1990s upturn; although as indicated above, in many cases
the numbers of spectators lost in earlier decades have been only partially
recovered. In recent years, ground capacity restraints have restricted the
magnitude of the improvement, especially for Chelsea and Manchester
United.
For Celtic and Rangers, the 1970s and early 1980s Scottish data
shown in panel (b) tell a similar story of sustained decline in attendances.
During this period, Celtic and Rangers were more typical of Scottish
clubs in general than Aberdeen and Dundee United, which did not suer
from the same problem, thanks mainly to improvements in performance
on the eld of play. Since the mid-1980s, however, the divergence in fortunes between the two pairs of clubs could hardly be more pronounced.
In the 1970s Celtic and Rangers lled their stadia for matches against
each other, but regularly played home matches against lesser opposition
in grounds two-thirds or three-quarters empty. Nowadays both clubs normally ll their grounds eortlessly for league matches, regardless of the
opposition. Meanwhile, as the fortunes of Aberdeen and Dundee United
on the pitch receded (relative to the highs attained in the mid-1980s),
attendances at both clubs fell somewhat during the 1990s.
A similar story of spectacular recent attendance gains for the big-market
teams, accompanied by more modest performance by others, is also apparent in the French data shown in panel (c).The tabulations begin only after
the worst eects of French footballs declining popularity in the 1960s and
early 1970s had been experienced. Average attendances remained quite
steady during the 1980s and early 1990s, prior to the recent uplift. Clearly
the 1998World Cup did much to improve the popularity of domestic club
football in France during the 1999 and 2000 seasons. But in common with
trends apparent in other countries, the two big-market teams included in
panel (c), Marseille and Paris Saint Germain (PSG), have beneted far
more than the other two, Bordeaux and Nantes.
For Germany, the long-term pattern diers somewhat from the other
three countries. Perhaps because attendances started from a relatively low
base when a national league competition resumed in the 1964 season,
West German attendances remained reasonably buoyant throughout the
1970s. Victories for the national team in the 1972 European
Championships and 1974 World Cup were also helpful. As elsewhere,
attendances did fall during the early 1980s, while the 1990s were a period
of sustained improvement. As panel (d) shows, in recent seasons the
attendances of Bayern Mnchen, Germanys most successful and economically powerful club, have pulled away from those of others including
Hamburg, Kln and Werder Bremen.
66
67
seems to have declined further during the 1940s and early 1950s, before
returning with a vengeance in the mid-1950s.
Demobilisation and the relaxation of restrictions after the Second
World War released a widespread and unprecedented pent-up demand
for recreation and entertainment, which manifested itself not only in
booming football attendances, but also in record-breaking cinema audience gures and packed beaches at British seaside holiday resorts during
the summers. In explaining the subsequent sustained decline in footballs
popularity as a spectator sport, social historians have emphasised the
impact of broader socio-economic changes. The growth of the auent
society saw the emergence of more home-oriented and privatised patterns of leisure activity, with the range of options extended in particular
by television and the car. Many women became less willing to acquiesce
in a home-based or child-minding function while mens leisure was
centred on the pub, club or football terrace. The spread of home ownership led to an increase in the popularity of alternative weekend activities
such as DIY and gardening (Dunning, Murphy and Williams, 1988;
Walvin, 1994).
The increase in incidents of crowd misconduct both inside and outside
football stadia, of which the train-wrecking exploits of Everton and
Liverpool supporters in the late 1950s were an early manifestation, can
also be attributed at least in part to broader social changes. Dunning,
Murphy and Williams (1988) see football hooliganism as one unfortunate
consequence of a general upsurge in working class assertiveness and
decline in social deference, stimulated by a new self-condence boosted
by high and stable levels of employment and increasing auence in the
late 1950s and 1960s.This also manifested itself more positively through
the novels, lms, plays and popular music of the period. Whatever the
causes, there is no doubt that over time, hooliganism and the countermeasures taken by clubs and the police prompted many spectators to
switch from the terraces to the living room and TV set. Football grounds
became increasingly the preserve of the more fanatical supporters, as the
less committed came to the conclusion that the entertainment on oer
was simply inadequate to compensate for the inconvenience, discomforts
and occasional physical dangers caused by the hooligan phenomenon.
From the late 1950s onwards, rising auence combined with improvements in both public and private transport, including the spread of car
ownership and the construction of the motorway network, led to important changes in the geographical composition of demand. With longdistance travel now practical both nancially and logistically, the top clubs
began to drain support from their smaller counterparts. In the late 1960s
there emerged a distinctive and vociferous London contingent among the
68
69
70
Waddington and Malcolm (1998) also provide some comparative data, based on a
number of surveys carried out in Britain, France, Spain, Austria and Belgium.
71
Table 2.9. Aggregate revenues and costs, Premier League and Football
League clubs, 1999 season
Premier League
( million)
Item
Football League
( million)
All clubs
( million)
Revenue
11Matchday
11Television
11Other
247
196
226
136
135
110
383
231
336
Costs
11Wages and salaries
11Net transfer fees
11Other costs
392
127
194
229
113
106
621
140
300
144
167
111
72
Table 2.10. Financial and employment data, top six Premier League clubs
and divisional averages, 1999 season
Club
Total revenue
(excluding
transfers)
Wages and
salaries
Operating
prot (before
transfers)
Pre-tax prot
Number of
employees
30,533
8,929
7,307
2,585
3,930
8,917
22,411
23
2,068
8,061
1,373
1,293
498
415
180
251
225
185
19,545
3,441
684
216
15,343
1,317
1,547
127
12,866
913
1,217
188
11,317
626
410
183
Overall, Tables 2.9 and 2.10 show that most of the leading Premier
League clubs, as well as the average Premier League club, made a prot in
1999, while the average Football League club made a loss.
Until recently, gate revenues from league matches represented by far
the largest source of revenue for most football clubs. It was not until the
start of the 1990s that clubs revenue sources became suciently
diversied that the share of league gate revenues in aggregate revenue fell
below 50 per cent for the league as a whole. Of course, television, sponsorship, merchandising and other forms of commercial activity all make
much bigger contributions now than ever before. Over the longer term,
however, gate revenues data provides an accurate representation of trends
in footballs overall revenue-raising capability and performance. This
section draws on a data set comprising annual gate revenues from league
matches for all Premier League and Football League clubs, obtained
from Football League archives and (for recent seasons) from Football
Trust (various issues).The 1926 season is the rst for which comprehen-
73
sive records are available. An attraction of the gate revenues data set for
researchers is that it can be combined with the attendance data described
in section 2.3, in order to calculate an average admission price for each
club in each season. Average admission price is obtained simply by dividing total gate revenue by total attendance.
Table 2.11 shows the average admission price per spectator in nominal
and real terms and average nominal admission prices by division, for each
season between 1926 and 1999. For the period up to the mid-1970s,
admission prices for spectators paying at the gate were subject to a statutory minimum determined by the Football League, which is also shown
in table 2.11. At certain times, however, concessions were available to
groups such as juniors, pensioners and the unemployed. Clubs were also
permitted to charge more than the minimum. Many did so for seated
accommodation (especially) and, increasingly as time went by, for standing accommodation as well. The original objective of the Leagues
minimum admission price was to prevent clubs from attempting to attract
spectators at the expense of other clubs in the same geographical catchment area by cutting prices.Table 2.12 shows aggregate gate revenue data
in nominal and real terms, together with divisional gate revenue shares.
Figure 2.6 shows the percentage shares of the Group 1 to Group 5 clubs
(see table 2.7) in aggregate gate revenue.
Table 2.11 suggests that there was little dierence between clubs in
admission prices during the inter-war period, with most clubs charging
the league minimum of 1 shilling (0.05) to all but a relatively small
minority of more auent, seated spectators. The gentle increase in the
real admission price series up to the mid-1930s and its subsequent reversal reects the constancy of admission prices in nominal terms, against a
background of retail prices that were falling in the late 1920s and early
1930s, and rising thereafter. When football resumed after the Second
World War, the increase in admission prices was just sucient to compensate for price ination during the war. The post-war attendance boom
therefore generated revenues more than 50 per cent higher than before
the war in real terms for several seasons until the early 1950s.
The relative homogeneity of footballs pricing structure was preserved
throughout the 1950s. In the 1960 season, Division 1 clubs were charging
only 25 per cent more on average than their Division 4 counterparts, a
ratio not much greater than that which had prevailed during the inter-war
period. From the 1960s onwards, however, the story is one of steadily
increasing divergence between the prices of clubs at dierent levels within
the league. By the 1970 season, Division 1 clubs were charging 48 per
cent more per spectator on average than Division 4 clubs. This gap
increased to 64 per cent in 1980, 70 per cent in 1990 and 155 per cent in
74
Minimum
admission
price ()
0.06
0.06
0.06
0.06
0.06
0.06
0.06
0.06
0.06
0.06
0.06
0.06
0.06
0.06
0.09
0.09
0.10
0.10
0.10
0.11
0.13
0.13
0.13
0.14
0.14
0.14
0.15
0.15
0.18
0.20
0.20
0.21
0.24
0.26
0.27
0.29
0.33
0.34
0.41
0.43
0.52
0.05
0.05
0.05
0.05
0.05
0.05
0.05
0.05
0.05
0.05
0.05
0.05
0.05
0.05
0.0625
0.0625
0.0625
0.0625
0.0625
0.075
0.075
0.075
0.075
0.10
0.10
0.10
0.10
0.10
0.125
0.125
0.125
0.125
0.125
0.20
0.20
0.20
0.25
0.25
0.30
0.30
0.40
Nominal Real
()
(1926=100)
0.06
0.06
0.06
0.06
0.06
0.06
0.06
0.06
0.06
0.06
0.06
0.06
0.06
0.06
0.08
0.09
0.09
0.09
0.10
0.11
0.12
0.12
0.12
0.13
0.13
0.13
0.14
0.14
0.16
0.18
0.18
0.19
0.21
0.23
0.24
0.26
0.29
0.30
0.35
0.38
0.46
100
103
106
107
112
118
119
124
123
122
118
114
113
110
120
118
120
120
112
112
126
124
120
122
118
115
117
120
135
141
141
141
151
160
162
164
176
172
186
185
209
0.06
0.06
0.06
0.06
0.06
0.06
0.06
0.06
0.06
0.06
0.06
0.06
0.06
0.06
0.08
0.09
0.10
0.10
0.10
0.11
0.12
0.12
0.12
0.13
0.13
0.13
0.14
0.14
0.17
0.18
0.19
0.19
0.21
0.22
0.23
0.24
0.26
0.28
0.32
0.35
0.43
0.06
0.06
0.06
0.06
0.06
0.06
0.06
0.06
0.06
0.06
0.06
0.06
0.06
0.06
0.08
0.08
0.09
0.09
0.09
0.10
0.11
0.12
0.12
0.12
0.12
0.12
0.12
0.13
0.14
0.16
0.16
0.16
0.18
0.20
0.21
0.22
0.24
0.24
0.29
0.31
0.38
0.05
0.05
0.05
0.05
0.05
0.05
0.05
0.05
0.05
0.05
0.05
0.05
0.05
0.05
0.07
0.08
0.08
0.08
0.08
0.09
0.10
0.10
0.10
0.11
0.11
0.11
0.11
0.12
0.13
0.13
0.14
0.15
0.16
0.18
0.18
0.20
0.22
0.23
0.27
0.28
0.36
75
Season
Nominal Real
()
(1926=100)
Minimum
admission
price ()
1974
1975
1976
1977
1978
1979
1980
1981
1982
1983
1984
1985
1986
1987
1988
1989
1990
1991
1992
1993
1994
1995
1996
1997
1998
1999
0.53
0.59
0.76
0.85
1.05
1.18
1.50
1.84
2.03
2.24
2.44
2.77
2.97
3.21
3.56
3.95
4.48
5.32
6.25
7.09
7.55
8.75
9.64
10.74
11.37
12.30
0.59
0.66
0.83
0.96
1.19
1.29
1.67
2.05
2.32
2.55
2.76
3.21
3.42
3.70
4.26
4.73
5.39
6.46
7.55
8.77
9.17
11.58
12.74
14.59
14.99
16.27
0.40
0.40
0.65
n/a
n/a
n/a
n/a
n/a
n/a
n/a
n/a
n/a
n/a
n/a
n/a
n/a
n/a
n/a
n/a
n/a
n/a
n/a
n/a
n/a
n/a
n/a
204
185
202
197
224
222
239
261
265
281
291
312
323
336
354
365
378
424
481
537
557
625
672
722
749
799
0.49
0.60
0.75
0.83
0.99
1.16
1.53
1.85
1.93
2.17
2.42
2.46
2.63
2.92
3.22
3.71
4.21
4.82
5.47
6.21
6.64
6.46
7.62
7.96
9.30
9.81
0.43
0.46
0.61
0.68
0.82
0.99
1.21
1.43
1.59
1.83
1.91
2.11
2.33
2.48
2.83
3.13
3.42
3.94
4.45
5.11
5.32
5.46
5.97
6.66
7.51
8.64
0.39
0.41
0.59
0.58
0.75
0.91
1.02
1.20
1.35
1.45
1.63
1.79
2.05
2.35
2.51
2.69
3.17
3.70
4.28
4.31
4.49
5.60
5.37
5.72
5.85
6.45
Notes:
Up to and including the 1958 season, D3 is Division 3 (South) and D4 is Division 3
(North). 1999 season data are based on estimates for some clubs.
Sources: Football League; Football Trust; Bird (1982).
n/a not available.
76
Season
Nominal
( million)
Real
(1926=100)
D1/Prem
D2/One
D3/Two
D4/Three
1926
1927
1928
1929
1930
1931
1932
1933
1934
1935
1936
1937
1938
1939
1947
1948
1949
1950
1951
1952
1953
1954
1955
1956
1957
1958
1959
1960
1961
1962
1963
1964
1965
1966
1967
1968
1969
1970
1971
1972
1973
1.352
1.373
1.410
1.419
1.372
1.246
1.263
1.261
1.321
1.373
1.456
1.575
1.661
1.594
2.933
3.578
3.780
3.801
3.786
4.135
4.541
4.431
4.215
4.355
4.311
4.448
4.557
4.549
4.677
4.981
5.242
5.341
5.829
6.308
6.931
7.682
8.388
8.828
9.960
10.814
11.823
100
105
109
111
111
108
112
115
120
123
127
131
136
129
185
205
214
211
192
189
202
196
177
176
167
167
171
168
168
171
176
174
181
189
203
214
222
220
227
229
230
46.9
47.0
47.2
45.7
47.8
47.4
48.1
46.9
48.9
49.9
48.5
45.2
44.5
45.8
45.0
43.8
45.5
44.3
44.2
44.1
46.4
46.9
47.4
45.4
45.3
46.9
47.2
48.1
49.6
47.6
47.2
48.0
52.4
52.0
55.0
56.7
57.1
56.9
56.9
57.8
61.4
26.8
28.2
28.9
28.3
27.5
28.5
25.8
28.8
26.2
26.3
28.2
30.3
30.5
31.1
30.9
30.4
28.2
29.9
28.4
28.5
26.2
27.2
26.5
27.6
26.8
25.6
25.7
26.0
24.8
27.2
26.3
27.1
24.7
24.0
23.8
23.1
22.9
23.9
22.7
21.7
20.4
16.3
15.5
14.7
16.1
16.5
15.0
17.6
15.4
16.0
15.3
15.3
16.1
15.6
13.9
14.7
15.7
15.9
16.5
17.3
16.7
16.8
16.4
16.5
16.0
16.1
16.8
16.4
15.8
14.7
16.5
17.7
16.5
13.5
15.2
13.3
11.6
12.2
11.5
12.8
13.4
11.9
10.0
9.3
9.2
9.9
8.2
9.1
8.5
9.0
8.9
8.5
7.9
8.4
9.4
9.2
9.4
10.2
10.5
9.3
10.1
10.7
10.6
9.5
9.6
11.0
11.9
10.6
10.7
10.1
10.9
8.8
8.9
8.4
9.5
8.7
7.9
8.5
7.8
7.7
7.6
7.1
6.3
77
Season
1974
1975
1976
1977
1978
1979
1980
1981
1982
1983
1984
1985
1986
1987
1988
1989
1990
1991
1992
1993
1994
1995
1996
1997
1998
1999
Nominal
( million)
13.174
15.180
18.822
22.220
26.651
28.960
36.911
40.239
40.523
42.096
44.760
49.276
48.901
55.844
63.906
72.885
87.219
103.691
127.329
146.238
163.655
191.610
211.500
244.838
283.313
312.056
Real
(1926=100)
D1/Prem
D2/One
D3/Two
221
205
218
222
246
236
255
248
230
229
232
240
231
253
276
292
319
359
425
480
524
593
639
714
804
879
58.8
52.6
57.7
58.6
59.3
56.6
55.0
58.2
59.6
56.3
53.7
63.5
63.2
60.6
54.0
50.7
48.8
53.7
59.2
58.3
59.7
64.4
63.0
64.4
59.4
60.6
23.7
29.8
23.0
23.3
23.9
24.6
25.4
23.8
22.6
25.6
29.0
20.1
19.1
21.8
26.9
29.7
33.3
29.2
25.0
24.9
26.3
22.3
23.7
22.6
27.5
23.6
11.0
12.4
12.8
12.6
10.3
11.6
13.1
12.9
11.1
12.8
11.6
11.4
11.9
10.4
12.2
13.0
11.0
10.8
10.5
12.2
9.7
9.3
8.3
8.7
9.4
11.5
Notes:
Up to and including the 1958 season, D3 is Division 3 (South) and D4 is
Division 3 (North). 1999 season data are based on estimates for some clubs.
Sources: Football League; Football Trust.
D4/Three
6.4
5.2
6.4
5.5
6.5
7.3
6.5
5.0
6.6
5.3
5.7
5.0
5.9
7.2
7.0
6.6
6.9
6.3
5.3
4.5
4.3
4.0
5.0
4.3
3.7
4.3
79
80
81
network in 1968. From then until the late 1970s, as buyers of broadcasting rights the BBC and ITV appear to have operated an informal cartel,
the main eect of which was to limit the fees paid to football by the television companies. The proceeds were distributed evenly among all league
members, with each club receiving a meagre 1,300 from the fee of
120,000 paid during the 1968 season.The total payment had increased
modestly to 5,800 per club, or just under 534,000 in total, by the 1979
season (Goldberg and Wragg, 1991).
The rst attempted breach in the BBCITV stranglehold occurred in
1978 when London Weekend Television (LWT) broke ranks with the
other ITV companies, and attempted to negotiate exclusive rights to
cover and distribute televised football to the rest of the ITV network. An
Oce of Fair Trading (OFT) ruling prevented the deal from going ahead,
but in the subsequently renegotiated contract, the amount received by
each club rose to 23,900 from an aggregate annual fee of 2.2 million
from the 1980 season onwards. Regular live coverage of league xtures,
shared between BBC and ITV, took place for the rst time in the 1984
season, in a two-year deal which provided for coverage of 10 matches per
season at a cost of 2.6 million per season. Live coverage was suspended
temporarily during the 1986 season, following the breakdown of negotiations for a new contract. Footballs parlous nancial condition at the
time, combined with increasing tensions between the larger and smaller
clubs over the distribution of the proceeds, probably allowed the TV companies to retain the upper hand in the negotiations. A contract was eventually settled at the start of 1986, reinstating live coverage during the
remainder of the 1986 season and for the following two seasons. The
annual price was 3.1 million for 14 matches, which implied the price per
match had fallen relative to the previous contract.The 1986 contract was
also the rst to breach the principle of equal distribution of revenues
between all 92 league member clubs; instead, Division 1 clubs received
50 per cent of the proceeds, Division 2 clubs 25 per cent and Divisions 3
and 4 clubs 12.5 per cent each.
The arrival of the two satellite broadcasters BSB and Sky in the late
1980s signalled a signicant shift in the relative bargaining power of the
football industry as the seller of broadcasting rights and the television
companies as buyers. The BBCITV duopsony was gone for good, and
the television companies were competing against one another to acquire
broadcasting rights far more intensely than ever before.The implications
of the shift were already apparent in the earliest days of satellite broadcasting. In the 1988 round of negotiations, ITV secured exclusive rights to
show 18 matches per season over the next four seasons, from 1989 to
1992 inclusive.The annual fee was 11 million.The balance of bargaining
82
power within football was also changing rapidly.Threats from the leading
clubs to withdraw from the Football League and enter into separate negotiations with the television companies secured a settlement in which
roughly three-quarters of the ITV fee went to the Division 1 clubs. Of this,
more than 40 per cent was shared between the so-called big ve, which
were to become the subject of the majority of ITVs coverage: Arsenal,
Everton, Liverpool, Manchester United and Tottenham Hotspur. By way
of consolation for losing altogether the right to cover league football, BBC
formed a partnership with BSB to purchase broadcasting rights for FA
Cup and international matches. Collaborative arrangements of this kind
between the satellite and terrestrial broadcasters, intended to allay public
concerns about the progressive disappearance of sports coverage from
free-to-air television channels, were a key component of the satellite companies strategies to expand their portfolio of live sports coverage during
the 1990s.
As the ITV contracts 1992 expiry date started to approach, it was clear
that the next round of negotiations would be shaped both by the ambitions of the newly merged satellite operator BSkyB to secure a
signicantly greater share of the coverage, and by the manoeuvrings of the
leading clubs anxious to maximise their own share of the proceeds.
Strategically, BSkyB identied the acquisition of regular live coverage of
league football as the key step that would generate sucient take-up of
satellite services to enable the company to extend subscription charges
beyond lms, to sports and other channels. Meanwhile in early 1991, the
Football Associations Blueprint for the Future of Football (Football
Association, 1991) announced that at the end of the 1992 season, the 22
Division 1 clubs would withdraw completely from the Football League to
form the new Premier League, with its own governing body under the
auspices of the Football Association. Naturally, the Premier League
would negotiate its own television contract, which would be separate
from any arrangements made by the Football League on behalf of its 70
remaining members.
In negotiations documented extensively by Fynn and Guest (1994),
BSkyB succeeded (narrowly) in outbidding ITV for the contract providing exclusive live broadcasting rights during the Premier Leagues rst
ve seasons, from 1993 to 1997 inclusive. Subsequently the BSkyB contract was extended for a further four seasons from 1998 to 2001. The
partnership with BBC inherited by BSkyB following the BSBSky merger
was to continue, with BBC screening regular Saturday evening highlights
of league matches played earlier the same day, from the 1993 season
onwards. The headline cost of the combined BSkyBBBC Premier
League coverage was 304 million for the initial ve-year contract, and
83
then 743 million for the subsequent four-year extension. The total
amounts received by the Premier League clubs are smaller than these
headline gures, however, as some of the revenue ows to the Football
Association. In each of the 1993 to 1996 seasons, the Premier League
clubs received around 40 million from the BSkyBBBC contract. In
1997 the aggregate receipts increased to 83 million. In 1998, as the new
contract got underway, the Premier League clubs received 125 million.
These proceeds are shared according to a formula that depends upon
each teams nishing position in the league, and on the number of television appearances. The distribution of the television coverage is more
equitable than under the old ITV contract, a feature inuential in encouraging the majority of Premier League clubs to favour BSkyB over ITV,
when the rst BSkyB contract was concluded in 1992.5 In the 1998
season, Arsenal, Manchester United, Liverpool, Chelsea and Leeds
United were the top ve earners from the contract, accounting for around
35 per cent of the Premier League clubs total proceeds.
The arrangement whereby BBC and BSkyB shared live coverage of FA
Cup and international matches also continued until the end of the 1998
season, when ITV replaced BBC in this role. ITVs initial consolation for
failing to secure Premier League coverage for the 1993 season was a contract to show selected Football League matches live on a regional basis on
Sunday afternoons, for varying fees that amounted to more in total than
had been paid to the former Division 2, 3 and 4 clubs under the old ITV
contract. But from the 1997 season onwards the rights for live Football
League coverage were also acquired by BSkyB.The television earnings of
Football League clubs have increased dramatically as a result, from
around 16 million in 1996, to more than 30 million per season
between 1997 and 2000. In 1997 Division One clubs received around 57
per cent of Football League television revenues, Division Two around 26
per cent and Division Three around 17 per cent (Football Trust; Deloitte
and Touche, 1999).
The outcome of the third sale of televised broadcasting rights since the
formation of the Premier League was announced in June 2000. BSkyBs
bid to secure the continuation of its Premier League coverage from the
2002 season onwards was successful. Initially, a second live Premier
15
Four of the big ve are reported to have opposed the BSkyB contract. As the main
beneciaries of the old ITV contract, Arsenal, Everton, Liverpool and Manchester
United naturally favoured its continuation. It is reported that Aston Villa and Leeds
United also voted in favour of ITV. The decision of the fth big-ve member,
Tottenham Hotspur, to vote in favour of BSkyB, was crucial in ensuring the two-thirds
majority needed for the deal to go through.Tottenham chairman Alan Sugars interest in
the consumer electronics (and satellite dish) manufacturer Amstrad, was declared at the
meeting (Conn, 1997, pp. 1920).
84
League package was sold to the cable operator NTL, to permit matches
to be screened on a pay-per-view basis; though a few weeks later NTL
announced their withdrawal. ITV replaced BBC as provider of the
Saturday evening Premier League highlights package, while ONdigital
won the rights to live Football League coverage from BSkyB. All of the
contracts awarded in June 2000 were secured by bids far in excess of the
prices paid in the previous round. As yet, footballs gravy train does not
appear to have run out of steam, though the momentum may be slowing.
Chapter 9 considers the prospects for the future of footballs relationship
with the broadcasting media.
The BSkyB takeover bid for Manchester United, and the collective
sale of broadcasting rights
The issue of vertical integration between football clubs and television
companies rst attracted headline attention in England with the
announcement on 7 September 1998 that BSkyB were bidding 575
million to achieve a 100 per cent takeover of Manchester United. Two
days later, the United board recommended acceptance of an improved
oer valuing the company at 623 million. Trade Secretary Peter
Mandelson subsequently referred the BSkyB bid to the Monopolies and
Mergers Commission (MMC). On 12 March 1999 the MMC delivered
its ndings: the merger would damage competition between broadcasters, and would be detrimental to the wider interests of British football
(Monopolies and Mergers Commission, 1999). On 9 April 1999, the
Trade Secretary Stephen Byers announced his acceptance of the MMCs
recommendations, and blocked the BSkyB takeover.
The arguments put forward for and against the BSkyBManchester
United bid are worth rehearsing briey. From BSkyBs perspective, the
attractions were obvious: rst, to obtain control of the worlds most
protable and most widely supported football club at a price which many
analysts considered conservative; second, to obtain inuence in possible
future negotiations over the creation of a European Super League, if such
negotiations ever take place; and third, to obtain similar inuence in
negotiations over the next domestic television contract, due to begin in
2001.
Signicantly, it was the impact of the merger in the domestic (rather
than the European) arena that provided the main grounds for the MMCs
rejection of the bid. To have the same company involved simultaneously
on both sides of the negotiations for broadcasting rights, it was argued,
would damage competition between broadcasters. Assurances that the
companys broadcasting arm would refrain from inuencing the football
85
86
that this technicality may have been inuential in persuading the Court to
endorse the status quo.
The OFTs case also hinged upon the question as to whether access to
exclusive live Premier League coverage conferred monopoly power upon
BSkyB as sole suppliers to a distinct market of a specic programme type,
or whether it merely enabled BSkyB to compete more eectively with
other broadcasters for audiences, within a much wider market for television programmes in general (Cave, 2000). If Premier League coverage
constituted a market in its own right, the existing arrangements, by
restricting access and driving up price, could well be anti-competitive and
against the public interest. But if Premier League coverage was simply
one of many programme types competing for audience share, the existing
arrangements could be benecial in helping to promote competition
through improved programme quality. Overall, the RPC was sympathetic
to this latter argument (Cave, 2000). From footballs point of view the
collective sale of broadcasting rights enabled the Premier League to
market its championship as a single entity. It also ensured greater equality
between the television revenues earned by richer and poorer clubs than
would be the case if clubs sold their broadcasting rights individually.The
existing arrangements therefore helped to promote not only competition
between broadcasters, but also competitive balance between clubs within
football; a consideration recognised by the Court as benecial to the
public interest.
In the long term, the 1999 rulings in the BSkyBManchester United
takeover bid and the OFTPremier League case at the RPC are unlikely
to represent the nal word on the matters which the two cases have raised.
On the issue of vertical integration, it is perhaps signicant that a number
of media groups already hold controlling interests in leading clubs on the
continent.6 Since the BSkyBManchester United bid was turned down,
there has been a spate of acquisitions of minority shareholdings in English
clubs by media groups determined to press their inuence on football as
far as Football Association regulations on cross-ownership will allow. At
the time of writing, BSkyB holds shares in Manchester United, Leeds
United, Chelsea, Sunderland and Manchester City; NTL has stakes in
Newcastle United, Middlesbrough, Aston Villa and Leicester City; and
Granada Media has stakes in Arsenal and Liverpool.Technological developments such as the introduction of multi-channel digital television and
internet broadcasting, ensure that technology will soon permit virtually
16
For example, Finivest controls AC Milan; Canal Plus controls Paris Saint Germain and
Servette Geneva; and Path holds a large minority interest in Olympique Lyonnais
(Deloitte and Touche, 1999).
87
unlimited screenings of live matches, on either a subscription or a payper-view basis, assuming that the market will bear such a level of exposure
(Cameron, 1997). On the issue of broadcasting rights, it seems likely that
the leading English clubs may eventually decide to take matters into their
own hands, and follow their Italian and Spanish counterparts in negotiating their own individual deals.
2.6 Other sources of revenue and cross-subsidy
Section 2.4 has identied a number of historical trends in a comprehensive data set covering gate revenues from league xtures. Data on the total
revenues the clubs have generated from all sources is much more dicult
to obtain, because historically many clubs have not produced company
accounts on a regular or consistent basis. Szymanski and Kuypers (1999)
have compiled annual data on turnover, wages and salaries, net transfer
expenditure and prot for a number of clubs that have led accounts at
Companies House. Their wages and salaries data include all employees;
at present, Deloitte and Touche (1999) estimate that players salaries
account for around two-thirds of total wages and salaries at larger clubs,
and up to four-fths at smaller clubs.
The Szymanski and Kuypers data set shows that for most of the post-war
period, smaller clubs were more dependent than larger clubs on income
from sources other than league gate revenues. Recently, however, the position has reversed. During the 1990s gate revenues from league matches
were a smaller proportion of total revenues for Premier League and
Division One clubs than for clubs in DivisionsTwo andThree. For all clubs
reliance on gate revenues from league matches has fallen considerably. In
the 1950s, revenues from league matches constituted over 80 per cent of
total revenues. By the 1990s, the corresponding gure was below 40 per
cent. Growth in revenues from television, as described in section 2.5,
accounts for much of the shift. According to table 2.9 television money
accounted for 29.3 per cent of Premier League and 12.5 per cent of
Football League revenues in the 1999 season.The growth of sponsorship
and other types of commercial activity, especially merchandising, has also
been very important. According to table 2.9 these sources accounted for
33.8 per cent of Premier League revenues and 39.1 per cent of Football
League revenues in the 1999 season. Comprehensive data on the revenue
contribution made by dierent types of commercial activity is not available, but table 2.13 shows the estimated annual values of the shirt sponsorship deals of all Premier League clubs current during the 2000 season. At
the start of the 2001 season,Vodafones 30 million four-year shirt sponsorship deal with Manchester United set a new English record.
Sega
LDV
JCT600
Autoglass
SubaruIsuzu
ED5
One-2-One
Packard Bell
Walkers Crisps
Carlsberg
Arsenal
Aston Villa
Bradford
Chelsea
Coventry
Derby
Everton
Leeds
Leicester
Liverpool
3.3
0.8
0.3
1.1
0.7
0.7
0.7
2.0
0.5
1.0
Fee p.a.
(million)
Sponsor
Club
2000
1999
1999
1998
1998
2000
1998
1997
2000
1999
Start
season
Man Utd
Middlesbrough
Newcastle
She Weds
Southampton
Sunderland
Tottenham
Watford
West Ham
Wimbledon
Club
Table 2.13. Shirt sponsorship contracts, Premier League clubs, 2000 season
Sharp
BT Cellnet
Newcastle Breweries
Sanderson
Friends Provident
Reg Vardy
Holsten
Phones 4 U
Dr Martens
Tiny
Sponsor
1.3
0.3
1.0
0.5
0.5
0.5
1.3
0.5
1.0
0.3
Fee p.a.
(million)
1997
1996
2000
1995
2000
2000
2000
2000
1999
2000
Start
season
89
90
tournaments, the FA Cup and the League Cup, has come under strain in
recent seasons.The extension of the European calendar in particular has
brought genuine problems of xture congestion, which have been tackled
in a number of ways: by delaying the entry of teams involved in Europe to
the League Cup; by substituting penalty shoot-outs for replays as a means
of settling drawn cup ties; by the clubs themselves choosing to eld teams
comprising reserve team players for cup matches; and by the complete
withdrawal of Manchester United from the 2000 seasons FA Cup.
Manchester Uniteds decision sparked an unprecedented level of debate
and complaint in the media and even in Parliament. In the eyes of many
traditionalists, all of the developments mentioned above have tended to
devalue the two domestic cup competitions, to the detriment of the
leagues smaller member clubs in particular and, in the case of the FA
Cup, non-league clubs as well.
2.7 Footballs labour market: players wages and the
transfer system
The retain-and-transfer system and the maximum wage were both key
features of English footballs labour market from the very earliest days of
professionalism. Both were designed to prevent the clubs with the most
resources from acquiring all of the most talented players simply by outbidding other clubs for their services the outcome which, it was argued,
might occur if the football players labour market operated without any
restrictions. Both survived in their original form until the maximum wage
was abolished in 1961, and the retain-and-transfer system was substantially overhauled in 1963.
The retain-and-transfer system originated in an FA regulation introduced in 1885, requiring clubs to register their players annually with the
Football Association. Player registrations immediately became tradeable
commodities between clubs, since unregistered players were not permitted to appear (Morrow, 1999). All player contracts were renewable annually at the clubs discretion, and clubs were entitled to retain a players
registration even if his contract was not being renewed. In theory and
sometimes in practice, this enabled a club to prevent an out-of-contract
player from earning a living. A player could move only if his present club
was prepared to sell or release his registration. Under this system all clubs
had the discretion to retain individual players for as long as they wished,
as well as the guarantee of nancial compensation when players were
eventually permitted to depart.
The incentive for players to seek to move was also restricted by the existence of the maximum wage, originally xed at 4 per week when it was
91
92
Even in the 1990s several leading English clubs have been unwilling to break internal pay
structures in order to match the salaries paid to superstar players in Italy and Spain.
93
structure to the removal of the maximum wage appears to have been virtually complete by the mid-1970s. Between 1974 and the late 1980s, the
rates of increase in revenues and wages were very similar. Having crept
from below 40 per cent in the 1950s to above 50 per cent by the mid1970s, the average ratio of total wages and salaries to total revenues
remained stable thereafter. Naturally, however, the average data tend to
mask considerable intra-club and year-on-year variation.
A further important change to the transfer system was introduced in
1978, when players were awarded the right to decide themselves whether
to move on the expiry of their contracts. Though the new system was
termed freedom-of-contract, in practice the out-of-contract player was
still constrained to some extent. If his existing club was oering a new
contract at least as rewarding as the expired contract, it could still
demand a compensation fee, either to be agreed between the two clubs, or
(in cases where agreement could not be reached) to be decided by a FA
tribunal. The tribunal decision was binding on both parties. This meant
that in order to move, an out-of-contract player had to nd a new club
that would either agree a fee with his existing club, or be willing to take
the risk of going to arbitration at the Football Association.
Comprehensive data on transfer expenditure prior to the 1970s is not
available. Even so, it seems certain that the labour market reforms of the
early 1960s tended to encourage increased levels of expenditure in the
transfer market, by creating the opportunity for the richer clubs to attract
players by oering more generous packages of remuneration. On the
other hand, freedom-of-contract, by limiting the value to his present club
of the player whose contract was drawing towards its conclusion, might be
expected to have had the opposite eect from 1978 onwards.
Relative to gate revenues, however, there has been no clear trend in
aggregate expenditure on transfers between league clubs since the 1970s,
as table 2.14 shows.There is certainly considerable year-on-year variability: in part this reects the relatively small number of transactions which
make up each annual total. The data, however, do not seem to justify the
frequent complaint that the transfer market is out of control. In fact,
perhaps the most surprising feature of table 2.14 is the long-term stability
of the ratio of total transfer expenditure to gate revenue. The one apparently unsustainable burst of transfer activity, in which some of the characteristics of a boombust cycle are apparent, followed English footballs
rst 1 million transfer (Trevor Francis from Birmingham City to
Nottingham Forest) in February 1979 and peaked during the following
season. A spate of transfers for fees of 1 million or more, involving
players such as Steve Daley, Gary Birtles (twice), Andy Gray and Peter
Ward, culminated in Bryan Robsons 1.5 million move from West
11.8
12.0
11.8
10.5
11.5
12.6
14.8
19.3
16.3
12.6
12.3
12.7
13.3
12.8
13.9
19.5
15.9
19.4
13.5
24.4
19.1
20.4
19.0
36.4
36.8
46.5
24.3
1973
1974
1975
1976
1977
1978
1979
1980
1981
1982
1983
1984
1985
1986
1987
1988
1989
1990
1991
1992
1993
1994
1995
1996
1997
1998
1999
10.7
10.9
10.8
10.5
10.6
11.1
12.5
13.7
12.7
11.3
11.5
11.5
11.1
11.4
11.8
13.2
13.4
15.8
14.9
12.7
12.8
13.2
15.8
16.1
17.1
16.7
12.2
D3/Two
0.2
0.3
0.2
0.2
0.2
0.4
0.8
1.5
0.3
0.5
0.1
0.4
0.3
0.4
0.7
1.0
1.6
3.3
1.2
1.4
0.5
1.3
1.1
2.8
2.0
1.5
2.0
D4/Three
115.9
116.9
117.0
113.4
117.0
111.7
117.2
132.8
121.2
122.4
110.2
113.0
114.2
115.5
122.8
136.7
146.2
155.1
146.0
174.9
173.2
191.8
109.9
139.5
140.7
160.0
174.7
All
100
106
102
43
78
112
124
214
121
163
53
68
73
80
117
157
162
159
143
241
256
332
403
441
428
422
594
D1/Prem
100
104
102
118
146
168
117
196
115
142
136
140
146
137
150
117
183
208
131
228
173
182
164
306
299
370
190
D2/One
100
117
177
150
145
182
168
206
130
156
162
157
141
149
161
105
106
167
129
167
167
177
135
139
156
359
258
D3/Two
100
136
192
163
156
190
185
290
153
176
117
150
144
150
185
113
164
318
102
115
142
102
183
207
143
105
139
D4/Three
Sources: Football Trust, Deloitte and Touche (1999, 2000), Football League, Premier League.
113.2
113.7
114.2
112.2
114.7
117.6
119.1
118.3
111.9
118.0
116.3
118.4
119.5
110.9
116.4
123.0
125.3
126.6
126.4
146.4
150.8
166.9
184.0
194.2
194.8
195.3
136.2
D2/One
Season D1/Prem
Table 2.14. Gross expenditure on transfers between English league member clubs, 19731999
100
108
193
137
164
195
129
210
118
111
148
158
160
163
189
138
162
180
137
213
202
250
289
358
349
389
418
All
49.1
52.4
46.1
18.6
31.1
43.5
59.4
88.9
52.7
55.0
24.2
28.8
28.8
31.9
40.8
57.6
63.4
63.2
44.4
58.7
50.1
56.1
57.4
66.0
57.5
50.0
45.1
Total as per
cent of league
gate revenues
95
96
huge inux of foreign players now being employed by English clubs, especially in the Premier League (see also section 2.1 and chapter 4). Many
have been attracted by remuneration perhaps well above what they would
be likely to attain elsewhere. According to data reported by Deloitte and
Touche (1999), the net transfer expenditure outow of the English league
football clubs as a whole has increased dramatically, from around 11
million in the 1993 season to 15 million (1994), 36 million (1995),
reaching a peak of 93 million in the 1996 season. After falling to around
70 million in each of the 1997 and 1998 seasons, it increased sharply to
140 million in the 1999 season. While the post-Bosman increase was
spectacular, the trend was already underway before 1995. The spectacular success achieved by Eric Cantona, rst at Leeds United and then at
Manchester United, following his transfer from Nmes in 1991, seems to
have been particularly inuential in raising awareness within the traditionally insular British game of the kind of contribution gifted overseas
players might be capable of bringing.
Second, since the price of acquiring an out-of-contract player is now
paid entirely to the player (rather than shared between the player and his
former club) wage ination is a predictable consequence of Bosman.
Players seeking a new club are in a much stronger position to bargain for
high remuneration if their acquisition is not encumbered with the obligation to pay a transfer fee. As table 2.15 shows, wage ination in English
football has certainly been exceptional during the post-Bosman period,
but so too has growth in revenue. Bosman has clearly assisted some individuals, especially star players, to capture more of the large surpluses that
are being generated from sources such as television and merchandising. It
seems likely, however, that high wage ination would have occurred in
any case (as it did before 1995) as a consequence of the popular revival
football has enjoyed during the 1990s.
Third, many clubs have attempted to protect themselves against the
possibility of losing their best players without compensation by oering
longer-term contracts, especially to their younger stars. Morrow (1999)
suggests that this development may also have contributed to wage
ination in the short term, if a higher initial level of remuneration is
needed to persuade a player to commit for a longer period. Longer contracts may also be attractive to players, however, since they insure future
earnings against the consequences of serious injury or loss of form.They
may also help restrain wage ination over the longer term, if they are
enforced over their full duration. There are nevertheless incentives for
clubs to oer enhanced terms to their top stars far in advance of the expiry
of their present contracts, to avoid involvement in an eventual bidding
contest for the players signatures. The insurance argument provides
97
Table 2.15. Revenue, wages and salaries and net transfer expenditure,
19941999
Year
Premier
League
Div One
Div Two
Div Three
3,768
3,569
4,331
5,599
7,597
6,666
1,451
1,740
1,681
2,482
2,771
3,700
1,046
1,103
1,380
1,159
1,312
1,417
4,348
5,158
5,692
7,562
9,370
10,353
1,096
1,256
1,324
1,734
2,179
2,861
743
788
941
1,051
1,201
1,302
2,397
2,735
3,289
4,080
5,653
6,729
243
98
141
320
175
468
114
54
159
76
267
239
149
404
997
782
1,303
1,402
All
75.5
72.2
78.7
69.9
78.6
77.3
71.1
71.4
68.2
90.6
91.5
91.8
65.4
64.8
68.7
69.4
75.0
77.7
10.9
4.9
11.5
6.6
16.0
15.2
5.3
2.4
0.2
1.5
3.9
2.0
16
14
16
16
14
14
81
81
81
81
72
72
24
24
24
23
21
20
19
21
21
22
18
18
98
99
100
D1/Prem
1973
1974
1975
1976
1977
1978
1979
1980
1981
1982
1983
1984
1985
1986
1987
1988
1989
1990
1991
1992
1993
1994
1995
1996
1997
1998
1999
0.4
0.2
0.6
0.7
0.5
1.8
0.2
0.6
0.9
2.4
0.4
2.2
2.2
1.2
4.1
5.5
0.8
3.5
7.4
3.6
6.0
13.0
9.6
25.2
14.5
1.5
27.5
D2/One
0.1
0.3
0.2
0.7
0.0
1.1
0.6
0.6
0.5
1.9
0.4
1.3
1.2
0.7
2.4
2.6
1.2
0.8
5.1
3.7
2.0
6.0
8.9
18.3
3.1
0.6
25.3
D3/Two
D4/Three
0.3
0.2
0.3
0.0
0.3
0.5
0.3
0.7
0.6
0.3
0.5
0.5
0.9
0.3
1.4
1.5
1.3
1.5
0.3
5.3
3.2
5.9
0.3
4.0
9.3
1.8
0.7
0.1
0.3
0.1
0.0
0.2
0.2
0.2
0.5
0.2
0.3
0.5
0.4
0.1
0.2
0.4
1.3
0.7
1.1
2.0
2.0
0.8
1.1
1.0
2.9
2.1
3.9
2.9
the leading clubs and the rest. For several Division One clubs that have
achieved promotion to the Premier League in recent seasons, the chasm
has turned out to be unbridgeable. Deloitte and Touche (1999) have
likened the jump from Division One to the Premier League to that which
would be faced by a convenience store attempting to become a supermarket in the space of just 10 weeks. But the temptation for Football League
to gamble on elevation to Premier League status by increasing wage and
transfer expenditure is ever-present, even though at the aggregate level
such eort cancels itself out completely if all teams increase spending
101
simultaneously (see also chapter 9). One of the more alarming trends of
recent seasons has been the sharp rise in wages and salaries as a percentage of revenue in the Football League (see table 2.15). Nevertheless, predictions of the imminent nancial demise of the lower division clubs were
just as prevalent during the early 1980s as they are today, and were
perhaps even more credible given that the nances of the football industry as a whole were in seemingly terminal decline at the time.
Clearly, footballs nances are now passing through a period of structural change at least as far-reaching as that which followed the reforms of
the early 1960s. Even if the transfer system is eventually reformed or even
abolished as a result of the European Commissions interventions, the
present authors are more optimistic than many that the resilience that has
seen many clubs overcome the most outrageous nancial duress in the
past, will continue to favour the future survival of English professional
club football at all its levels, in a form not too dissimilar to that which has
proven so durable over the past 80 years.
2.8 The historical development of professional football:
international comparisons
Sections 2.2 to 2.7 have provided an account of the historical development of professional football in England and Wales, as both a sport and a
business.The fact that the codied game of football now played throughout the world originated on the playing elds of the public schools of
Victorian Britain naturally makes the story that unfolded in the previous
sections of chapter 2 a hugely important part of the history of the sports
international development. When one investigates the history of football
worldwide, however, one nds that for every country there is a unique
story to be told, which explains the evolution of the sport into its presentday form. Section 2.8 aims to provide just a avour of this international
diversity, by outlining the historical development of professional football
in two cases which contrast starkly, both with each other and with
England.
In France as in England, the historical development of football has
been mainly organic rather than consciously planned, but the sport has
always struggled to achieve the heights of popularity attained in a number
of other European countries. Despite spectacular recent French success
at European and world level, club football appears to be at a crossroads:
on the one hand commercial pressures are threatening to undermine a
distinctive national model of sports organisation, while on the other the
leading French clubs seem to be struggling to keep up with their Italian,
Spanish, German and British counterparts.
102
In Japan in contrast, professional football is an exclusively recent phenomenon. As in France the origins of Japanese amateur football can be
traced as far back as the 1870s, but the sport has always struggled for recognition under the shadow of baseball. The formation of the J.League in
1993 represents a conscious top-down attempt to launch a new sports
brand in the consumer marketplace, requiring radical innovation capable
of shifting entrenched cultural values, and deploying the full battalions of
modern marketing and advertising techniques toward this end. Although
initial public enthusiasm for the new brand has waned recently to some
extent, professional football does appear to have established a foothold
suciently secure to permit it to develop further in the future.
France
As in many countries, football was rst introduced into France by travelling British businessmen during the late nineteenth century.The rst club
was founded in the port city of Le Havre in 1872 by members of shipping
and transit companies who wished to play both football and rugby. A
number of other multi-sports clubs were formed in the 1870s and 1880s,
before clubs devoted exclusively to football began to appear during the
1890s. The rst national cup tournament was won by Standard AC in
1894, and an embryonic league competition was won by Club Franais in
1896. By the turn of the century, football was played extensively throughout northern France including Paris (Pickup, 1999).
At the start of the twentieth century, football spread rapidly to the east
and south of France. Frances rst international xture was played
against Belgium in 1904, but the administration of the sport during the
pre-war period was chaotic, owing to rivalries between a number of alternative governing bodies. This situation was resolved in 1919 by
unication of the competing bodies under the Federation Franaise de
Football (FFF). A French ocial, Jules Rimet, was the rst president of
the newly created international football federation FIFA when it was
formed in 1920, and became the prime mover behind the creation of the
rst international World Cup tournament, played in Uruguay in 1930.
In domestic football, a major advance came with the acceptance of professionalism in 1932, followed by the introduction of a national league
competition won by Ste in its inaugural 1934 season.The 1935 winners
were Sochaux, a club which had been formed in 1929 on a professional
basis with the backing of the Peugeot car manufacturer. This development had been instrumental in prompting the authorities to adopt professionalism. Racing Club (RC) Paris, formed by a Parisian estate agent in
1932, became the next champions in 1936.
103
104
Average
(000)
Season
Average
(000)
Season
Average
(000)
Season
Average
(000)
1981
1982
1983
1984
1985
11.4
10.8
11.5
11.2
10.9
1986
1987
1988
1989
1990
10.9
12.6
13.5
11.8
14.3
1991
1992
1993
1994
1995
12.2
12.4
15.1
13.1
14.6
1996
1997
1998
1999
2000
16.0
16.4
17.9
19.3
22.3
Source: <http://rernes.free.fr./>
105
(and several other club ocials) being jailed in 1995 for a variety of
bribery and corruption oences including match xing during the 1993
season.8
Perhaps more sustainable models for the commercial management of
French football at the highest level are provided by Olympic Marseilles
gradual rehabilitation under the presidency of Robert-Louis Dreyfus,
head of the Adidas sportswear manufacturer; and by the recent progress
of Paris Saint Germain, under the eective control of the subscription
television company Canal Plus (Eastham, 1999). Formed in 1984, Canal
Plus success in marketing subscriptions to view live football on terrestrial
television provided an early exemplar for BSkyBs subsequent success in
Britain. In another move that proved to be prescient Canal Plus acquired
a 40 per cent share in the ailing Paris Saint Germain in 1991. Although
one of the television companys aims was to inuence the future development of relations between football and television, their nancial and
organisational support led to a revival in Paris Saint Germains competitive and commercial fortunes: the club were French champions in 1994,
and European Cup Winners Cup winners in 1996. By the end of the
decade Paris Saint Germain had by far the largest turnover of any French
club, and in summer 2000 were (ironically) able to pay 22 million in
order to re-sign Anelka from Real Madrid.
The conict between the traditional national model and what
Bourdieu (1999) has described as the neoliberal commercial model for
the control of French football is still underway. Current symptoms
include the continuing exodus of leading players to foreign clubs; the
debate and legal actions concerning the proposed termination of municipal funding for football clubs on the grounds that it is anti-competitive;
and the expressed desire of a handful of major clubs, including Olympic
Marseille and Paris Saint Germain, to oat on the stock exchange.
Clearly these clubs believe that the commercial freedom that otation
would secure is essential if they are to keep pace with other leading
European clubs in the future. But France is not the only country in which
the high ambitions of the richest clubs have raised fears about the competitive and commercial implications for the rest.
Japan
The rst football match to be played in Japan is reported to have taken
place just one year after the sport rst appeared in France. In 1873
18
Eastham (1999) describes the Marseille case in detail. Eastham also describes the entrepreneur Jean-Luc Lagardres attempt to revive the fortunes of Racing Club Paris in the
late 1980s.This venture also ended in sporting and commercial failure, but there was no
illegal or criminal activity.
106
107
108
American or baseball inuences are apparent in several aspects of the J.Leagues competitive structure. In all seasons but one, the championship has been played in two stages
separated by a mid-season break, with the winners of each stage meeting in a play-o to
determine the overall champions. An experiment with a European-style single-stage
format was abandoned after one season in 1996. At rst, penalty shoot-outs were used to
settle draws. Now, matches level after 90 minutes go into 30 minutes extra time under
the golden goal rule.Three league points are awarded for a win in 90 minutes, two for a
win in extra time and one for a draw at the end of extra time.
109
2.9 Convergence and divergence in the league gate revenues of English football clubs
The previous sections of chapter 2 have described a number of fundamental changes in the competitive and business structure of professional
football that took place during the twentieth century, and especially in
recent times. In England as elsewhere, their cumulative eect has been to
widen the economic and nancial chasm that now separates clubs operating at dierent levels within the league. Inevitably, the magnitude of the
gulf which currently exists raises concerns about the future viability of
English footballs traditional competitive structure of 92 clubs organised
into four divisions, all employing a squad of full-time professional players.
These concerns have been prominent during footballs recent renaissance, but it is worth recalling that they are by no means solely a recent
phenomenon. Discussion of the merits and demerits of proposals such as
break-away competitions for the top clubs at national or European levels,
semi-professional status for lower-division players, or a return to regionalisation of the lower divisions was just as extensive during the early 1980s
when footballs revenue base was in decline, as it has been recently.
This section does not attempt to provide answers or solutions to all of
these issues. Instead, it pursues a less ambitious objective: to quantify the
rise in inequality that has taken place over the long term. To this end, a
number of standard economic measures of convergence, divergence and
inequality are applied to the league gate revenues data set that was
described in section 2.4. A direct comparison of the economic structure
of professional football between the 1920s and the 1990s will of course
show that economic inequality between clubs has risen considerably
during the period. But the analysis will also reinforce the impression, conveyed less formally in the earlier sections of this chapter, that much of the
increase in inequality within the 75-year period which the gate revenues
data set describes has taken place in relatively short bursts. Specically,
the period between the late 1950s and the mid-1970s, which includes the
abolition of the maximum wage and the reform of the retain-and-transfer
system, was one of fundamental change in the economic structure of
English football. So too was the period (still underway) since the start of
the 1990s, which takes in the formation of the Premier League, the development of pay-television, and the Bosman ruling.
Measuring convergence, divergence and inequality
Describing the pattern and modelling the process by which a group of
economic units increases or decreases in size (on some appropriate
110
measure) over time, both in absolute terms and relative to one another, is
a key endeavour in a number of elds of empirical economic enquiry, as
the following examples illustrate:
International economists may wish to know whether the average prosperity of a group of countries, measured by their per capita Gross
Domestic Product (GDP), can be expected to converge or diverge over
time. Do wealthier countries enjoy faster or slower growth in per capita
GDP than their poorer counterparts? If the rich are becoming richer at
a faster rate than the poor are becoming richer (or if the poor are actually becoming poorer) there may be a need for interventionist policies
by development agencies and the governments of the richer countries
to oset rising inequality. Alternatively, if the poor are becoming richer
at a faster rate than the rich, then inequalities between nations may
tend to diminish naturally over time without the need for any such
intervention (Solow, 1956; Baumol, 1986; Barro, 1991; Temple,
1999).
On a slightly smaller scale, industrial economists may wish to identify
and predict trends in industry concentration measures, which indicate the
extent to which an industry is dominated by a small number of large
rms. To model the evolution of an industrys structure, it is useful to
study the empirical relationship between the relative sizes of the member
rms, and their growth performance. Do the larger rms tend to grow
faster than the smaller ones? If so, we should expect to see market concentration increase over time. Alternatively, is there no statistical relationship between rm size and growth? Is the probability of growing at a
certain rate over any given period the same for a large rm and a small
rm? If so (perhaps less obviously) market concentration can also be
expected to increase over time. Or, nally, do the smaller rms tend to
grow faster on average than their larger counterparts? In this case,
market concentration might either increase or decrease in the short
term. But in the long term, concentration will converge towards an equilibrium, beyond which there is no further tendency for it either to
increase or decrease (Gibrat, 1931; Hart and Prais, 1956; Sutton, 1997).
Returning to football, the type of empirical model that has been used to
investigate the evolution of per capita GDP in international economics,
and the evolution of industry structure and concentration in industrial
economics, can be adapted to model the distribution of aggregate gate
revenues among the member clubs of the English football league. To do
so, some mathematical notation is needed, and some standard economists denitions of convergence and divergence must be introduced.The
notation is as follows:
111
(2.1)
where ui,tT is a random error term with a mean of zero and a standard
deviation of
u, and and are coecients to be estimated.The numerical value of the coecient in (2.1) determines the relationship between
revenue at time t, and growth in revenue over the period t to tT:
If
0, growth in revenue over any period tends to be related positively
to initial revenue at the start of the period, so on average the larger clubs
grow faster than the smaller clubs. As suggested above, the dispersion of
revenues between clubs will tend to widen over time, and there is no
sigma convergence.
If 0, there is no relationship between growth in revenue and initial
revenue. In this case as well, the distribution of revenues between clubs
tends to widen over time, and there is no sigma convergence. If growth is
112
In terms of the parameters of (2.1), the expression for the long-term average revenue is
/. The expression for the standard deviation of the distribution of revenues at any
given time around this average is
u/(2 2). In the short term, if the current standard deviation of revenues is smaller than
u/(2 2), the distribution of revenues
will tend to widen, so there is no sigma convergence. If the current standard deviation is
larger than
u/(2 2), the distribution will tend to narrow, and there is sigma convergence.
Cross-sectional
relationship between
initial revenue and growth
(i) > o
log
growth
113
log
revenue
log initial
revenue
time
(ii) = o
log
growth
log
revenue
log initial
revenue
time
(iii) < o
log
growth
log
revenue
log initial
revenue
time
114
time
Figure 2.8 Time-path of revenue under conditional beta convergence
Sigma convergence and unconditional beta convergence oer adequate convergence measures, as long as (2.1) provides an accurate representation of the processes that are generating the rates of growth of
revenues for all clubs in each period.11 It is possible, however, that this is
not the case.There may be factors other than the size of the clubs revenue
at the start of each period, which inuence the growth in revenue over the
same period. In the case where 0, this is equivalent to saying that, in
the long term, each clubs revenue may be mean-reverting towards an
average value which is specic to that club, and which depends on observable characteristics of the club. This is illustrated in gure 2.8, in which
the two clubs revenues tend to be mean-reverting towards dierent longterm average values.
The three additional variables dened above, loge( pi), loge(ai) and ni,
will be used in the empirical analysis to test whether club-specic characteristics are in fact relevant in determining the long-term average values of
revenues. In the long term, it might be expected that each clubs attendance would be positively related to its local population base, negatively
related to its age (if early entry into the league allowed the clubs concerned
to benet from rst-mover advantages in establishing a high initial base
11
115
level of support), and negatively related to the number of other clubs in its
near geographical vicinity. If so, then (2.1) should be reformulated as
follows:
T loge(ri,tT) loge(ri,t ) 1 loge( pi ) 2 loge(ai )
3 ni ui,tT
(2.2)
Fm,t
i,t
i1
for m 1 . . . n. Fm,t is the sum of the m smallest revenues in the data set,
for any value of m.
The Gini coecient compares the shape of Fm,t as m increases from 1
to n, with the shape Fm,t would take if there was perfect equality between
the revenues of the n clubs. This is illustrated in gure 2.9. The straight
line linking O and B is the shape of Fm,t if total league revenue was shared
equally between the n member teams. The curve OAB lying beneath the
straight line is the actual shape of Fm,t.The greater the degree of inequality, the lower down in gure 2.9 will be the mid-point A, and the greater
will be the curvature of OAB. The Gini coecient is the ratio of the
shaded area between the straight line OB and the curve OAB, to the triangle OBC. The greater the degree of inequality, the closer is the Gini
coecient to one.13
12
13
In the case 0, the average value to which the revenues of club i are mean-reverting in
the long term is ( 1 loge( pi ) 2 loge(ai ) 3 ni )/.
The formula for the Gini coecient is
nFn,t 2
Gt
m,t
m1
nFn,t
116
Fm,t =
cumulative
revenue
B
C
m=n
clubs ranked in
ascending order of
revenue (m = I ... n)
Gt
(2)
s(loge(ri,t))
(1)
0.690
0.756
0.777
0.740
0.754
0.744
0.726
0.731
0.731
0.778
0.781
0.752
0.728
0.734
0.757
0.686
0.694
0.695
0.681
0.680
0.690
0.708
0.701
Season
1926
1927
1928
1929
1930
1931
1932
1933
1934
1935
1936
1937
1938
1939
1947
1948
1949
1950
1951
1952
1953
1954
1955
0.349
0.365
0.378
0.352
0.368
0.376
0.366
0.379
0.373
0.397
0.396
0.369
0.360
0.369
0.368
0.344
0.346
0.343
0.341
0.341
0.349
0.362
0.361
Gini coecient
Sigma
convergence
(4)
0.178b
0.237a
0.241a
0.216a
0.143b
0.246a
0.232a
0.179a
0.172a
0.214a
0.215a
0.197a
0.212a
0.210a
0.231a
0.112c
0.077
0.076
0.116
0.095
0.078
0.023
0.072
(3)
0.067
0.136c
0.161b
0.129b
0.111c
0.203a
0.186a
0.151a
0.153a
0.204a
0.230a
0.180a
0.188a
0.215a
0.213a
0.088c
0.057
0.028
0.073
0.057
0.020
0.010
0.021
0.128a
0.128a
0.104a
0.101a
0.045
0.032
0.035
0.022
0.014
0.018
0.011
0.013
0.012
0.000
0.010
0.038
0.025
0.056
0.053
0.077c
0.099b
0.059
0.124b
1
(5)
Unconditional beta
convergence
Table 2.18. Convergence in gate revenues from English league matches, 19261997
3
(7)
0.218
0.227
0.098
0.161
0.245
0.297b
0.266
0.194
0.215c
0.146
0.146
0.124
0.243
0.254
0.101
0.034
0.016
0.024
0.033
0.169
0.151
0.224
0.255
2
(6)
0.013
0.014
0.012
0.010
0.014c
0.005
0.006
0.007
0.008
0.011c
0.009
0.004
0.005
0.014c
0.000
0.006
0.005
0.008
0.005
0.008
0.007
0.008
0.017
Gt
(2)
s(loge(ri,t))
(1)
0.692
0.659
0.682
0.736
0.767
0.744
0.755
0.795
0.801
0.837
0.798
0.851
0.878
0.880
0.928
0.928
0.963
0.985
1.013
1.052
1.009
1.043
1.030
Season
1956
1957
1958
1959
1960
1961
1962
1963
1964
1965
1966
1967
1968
1969
1970
1971
1972
1973
1974
1975
1976
1977
1978
0.354
0.347
0.355
0.370
0.382
0.387
0.388
0.403
0.415
0.435
0.422
0.446
0.451
0.463
0.468
0.471
0.477
0.494
0.489
0.498
0.488
0.495
0.500
Gini coecient
Sigma
convergence
(4)
0.088
0.006
0.035
0.051
0.001
0.052
0.037
0.018
0.018
0.016
0.009
0.009
0.027
0.080
0.012
0.074
0.098
0.204a
0.220a
0.215a
0.215a
0.196a
0.262a
(3)
0.045
0.060
0.095
0.016
0.065
0.045
0.083
0.074
0.107
0.078
0.072
0.081
0.059
0.026
0.006
0.030
0.089
0.143b
0.137b
0.137b
0.156b
0.174a
0.221a
0.100b
0.094c
0.069
0.105b
0.086c
0.123b
0.150a
0.127b
0.121b
0.071
0.100c
0.105b
0.119b
0.068
0.021
0.062
0.001
0.076
0.119b
0.124b
0.086
0.042
0.087
1
(5)
Unconditional beta
convergence
3
(7)
0.280
0.251
0.053
0.218
0.118
0.047
0.103
0.090
0.129
0.186
0.174
0.199
0.064
0.114
0.118
0.164
0.039
0.108
0.076
0.022
0.064
0.165
0.183
2
(6)
0.003
0.001
0.000
0.003
0.003
0.004
0.017
0.017
0.016
0.007
0.013
0.021c
0.012
0.008
0.005
0.014
0.007
0.001
0.004
0.000
0.004
0.006
0.001
0.979
1.002
1.029
0.994
1.015
1.003
1.050
0.998
0.963
0.963
0.929
0.934
0.979
1.038
1.033
1.074
1.124
1.124
1.146
0.478
0.483
0.504
0.502
0.497
0.505
0.539
0.525
0.511
0.486
0.476
0.478
0.500
0.518
0.513
0.522
0.552
0.565
0.576
0.213a
0.218a
0.224a
0.143b
0.184a
0.102c
0.101
0.052
0.016
Notes:
a
signicantly dierent from zero, two-tail test, 1 per cent level;
b
5 per cent level; c 10 per cent level.
1979
1980
1981
1982
1983
1984
1985
1986
1987
1988
1989
1990
1991
1992
1993
1994
1995
1996
1997
0.301a
0.306a
0.301a
0.211a
0.261a
0.159b
0.160b
0.115
0.040
0.146a
0.162a
0.154a
0.126b
0.129b
0.071
0.057
0.047
0.035
0.010
0.012
0.005
0.003
0.000
0.001
0.006
0.011
0.015
0.014
0.034
0.102
0.099
0.035
0.074
0.230
0.280
0.415a
120
under 0.70. Between the mid-1970s and the end of the 1980s, a new
plateau was established, with an average value just over 1.00. After falling
slightly at the start of the 1990s, s(loge(ri,t )) has since increased further,
achieving its highest-ever value in the 1997 season. For the period covered
by table 2.18 as a whole, there is clearly no evidence of sigma convergence
in the gate revenues data. Instead the dispersion of aggregate gate revenue
between clubs has widened considerably. As suggested above, however,
much of the increase in the standard deviation has taken place over relatively short sub-periods within the last 75 years, specically between the
late 1950s and mid-1970s, and again more recently during the 1990s.The
trend in the Gini coecient, shown in column (2) of table 2.18, tells a
very similar story to the sigma convergence measure. Overall, Gt increased
from around 0.34 in the late 1940s to just under 0.58 by the late 1990s.
For the 75 clubs that were league members in both the 1926 and 1997
seasons, tests for unconditional and conditional beta convergence of
revenue over the entire observation period involved estimation of (2.1) and
(2.2), using 1926 as the base season and the change in log revenue during
the 55 seasons completed between 1926 and 1997, denoted 55 loge(ri,97),
as the dependent variable.The estimation results are as follows:
55loge(ri,97)1.2630.368b loge(ri,26) ui,97
(1.321) (0.138)
n75
s u 0.731
R2 0.089
(2.3)
s u 0.762
R2 0.111
(2.4)
121
While there is evidence of neither beta convergence nor sigma convergence in the results reported so far, there is still the possibility of observing convergence in revenues over shorter sub-periods within the
observation period. Indeed, this might well be expected in view of the
pattern exhibited by s(loge(ri,t )), which as column (1) of table 2.18 shows,
is quite stable over relatively long sub-periods. Columns (3)(7) of table
2.18 report the results of a series of estimations of (2.1) and (2.2), using
each season from 1926 to 1987 in turn as the base season, and the change
in log revenue over the next T10 seasons as the dependent variable.
Therefore the rst row reports regressions of log growth between the
1926 and 1936 seasons on log revenues in the 1926 season. The gap
created by the Second World War is ignored in dening the 10 season
intervals, so (for example) the fth row reports regressions of log growth
between 1930 and 1947 on log revenue in 1930.
Column (3) of table 2.18 reports the estimated values of in (2.1)
(denoted ). In the unconditional convergence model, negative estimates
of are obtained in all estimations up to and including base season 1956.
At the 5 per cent level, the estimated coecients are signicantly smaller
than zero for all base seasons between 1931 and 1947. Therefore there
was unconditional beta convergence in revenues during this period.
During the early and mid-1950s, the negative estimated values of
become smaller in absolute terms, before eventually turning positive for
all but one base season between 1957 and 1970. During this period,
unconditional beta convergence broke down. From base season 1971
onwards, negative estimates of are obtained for all base seasons up to
and including 1986.These are signicantly dierent from zero for all base
seasons between 1973 and 1983. From base season 1981 onwards,
however, the negative estimates of start to become smaller in absolute
terms. In the nal estimation for base season 1987, a small positive estimated value of is obtained.
Columns (4)(7) report the estimated values of , 1, 2 and 3 in the
conditional convergence model (2.2). In general, the conditional beta
convergence results are similar to their unconditional counterparts. The
estimated values of are signicantly dierent from zero for all base
seasons up to and including 1947, and from 1973 to 1985 inclusive.
During the 1920s and 1930s, the negative conditional convergence estimates are generally larger in absolute terms than the corresponding
unconditional estimates. Similarly, between the late 1950s and early
1970s, the conditional estimator shows less tendency than the unconditional to return positive estimates. For the 14 base seasons between 1957
and 1970, the unconditional estimator returns 13 positive estimates,
while the conditional estimator does so on only ve occasions.
122
The estimates of 1, the coecient on log population, are positive in all
55 estimations of (2.2), and signicantly dierent from zero at the 5 per
cent level in 21 cases. This represents reasonably strong and consistent
evidence that when there is beta convergence, the equilibria to which
individual clubs revenues are mean-reverting are dependent on the local
population base. Despite a progressive increase over time in the geographical mobility of spectators (as described in section 2.3) there is no
indication that the relationship between population and revenue was any
weaker at the end of the observation period than at the start. Several of
the largest estimates of 1 are obtained for base seasons in the early
1980s. Presently the most fashionable clubs are mainly located in the
largest cities, but because they attract support from all parts of the
country these clubs no longer depend on their local population base for
their popularity.
The signs of the estimates of 2, the coecient on the number of other
local clubs, are also mostly as expected, with 41 negative and only 14 positive estimates obtained. The negative sign of the majority of these
coecients suggests that other things being equal, a larger amount of
local competition tends to depress a clubs equilibrium gate revenue.
However, none of the 41 negative estimates is signicantly dierent from
zero at the 5 per cent level, so the inuence does appear to be relatively
weak. In this case, there is some indication of a change over time: most of
the negative estimates of 2 (and most of those which come closest to
being statistically signicant) are found near the beginning of the observation period. In the 1920s, a large local population base was presumably
necessary to generate a large level of (local) support, because spectators
would go to watch the nearest team. The presence of other teams in the
same locality would tend to diminish a clubs revenue base. In the 1990s a
large local population base was still necessary to generate a large level of
(national) support, because the big-city teams are the most fashionable.
The presence of other teams in the same locality, however, would make
little or no dierence to these clubs revenue base.
Finally, the duration of the clubs league membership appears to exert
only a very weak inuence (if any) on equilibrium revenue, with
sequences of both positive and negative estimates of 3 obtained for
dierent sub-periods.The positive estimate obtained in the very nal estimation for base season 1987 is the only one in column (7) that is
signicantly dierent from zero, at the 1 per cent level. This result could
perhaps be meaningful, however, in view of the ndings in section 2.2
concerning the revival in the fortunes of several of the leagues oldest
member clubs during the 1990s.
123
2.10 Conclusion
Chapter 2 has presented an analysis of both the historical development
and the current state of English club footballs commercial structure, and
has provided some international comparisons. The empirical results
reported section 2.9, together with the description of the historical development of English professional football as a business in sections 2.22.8,
suggest that between the early 1920s and the end of the century, the economic structure of English football has passed through two distinct equilibrium phases, interrupted by two transitional periods.
The rst and highly durable equilibrium phase lasted from the early
1920s until the late 1950s. Considerable uniformity in the nancial structure of clubs operating at all levels within the league was preserved
through regulation of both admission prices (which determined the main
component of revenue) and players wages (the principal item of cost). By
the end of the 1950s, however, as the post-war recovery gathered momentum and general living standards rose dramatically, the pressure for relaxation of these regulations became irresistible. The abolition of the
maximum wage in 1961, and the reform of the retain-and-transfer system
in 1963, heralded a period of rising nancial inequality between clubs
operating at dierent levels within the league hierarchy, as football made
the transition towards a new (and considerably less durable) equilibrium
economic structure.
The period from the mid-1970s to the end of the 1980s is identied as
the second equilibrium phase. During this period footballs economic
fortunes were in decline on a number of fronts. Pressures emanating from
social and demographic changes, the hooligan phenomenon, and the
failure of football clubs themselves to maintain their own physical infrastructure, had taken their toll on attendances over many years, to the
extent that the future survival of large numbers of clubs (both large and
small) seemed to be in jeopardy. As the crisis deepened, the earlier trend
towards rising inequality between clubs appears to have been checked;
the nancial positions of all clubs were deteriorating at roughly the same
rate. Tragically, however, it was only after the disasters of Heysel,
Bradford and Hillsborough that the momentum for fundamental reform
became overwhelming.
Footballs rehabilitation during the 1990s as the most popular and
fashionable national sport in England and elsewhere has coincided with a
further sharp rise in nancial and competitive inequality between clubs.
This increase has taken place against a background of rising rather than
falling demand at all levels, which has done much to oset the most
124
Over the past 45 years, a consensus has evolved among most sports economists that in a professional sports league, the free operation of market
forces should maintain a reasonable degree of competitive balance among
member teams. The interdependencies between teams inherent in the
competitive structure of any sports league create disincentives for a
wealthy team to attempt to accumulate talent to the extent that the league
as a whole loses competitive viability. Contrary to widely held belief,
measures such as a salary cap or maximum wage, a reserve clause restricting player mobility between clubs, or gate revenue sharing, are not
expected to create closer competitive balance than would tend to emerge
naturally as a result of the free play of market forces. Such measures may
have implications, however, for the distribution of revenues between
strong clubs and weak clubs and between clubs (in general) and players.
In this chapter, section 3.1 describes the North American model for the
organisation and regulation of professional team sports, and draws some
comparisons with the British and European models.This is followed by a
review of the theoretical literature on competitive balance within sports
leagues, most of which has been developed with North American team
sports in mind. After the main ndings of this literature have been examined, its applicability to European football is considered. Two important
premises of the US literature seem questionable in respect of European
football: rst, the assumption of prot maximising behaviour on the part
of clubs; and, second, the assumption that the available stock of playing
talent is xed. The implications of relaxing or altering these assumptions
are explored. Section 3.1 also reviews the empirical evidence on the
eects of the introduction of free agency in a number of North American
professional team sports on competitive balance within the leagues
concerned.
Competitive balance matters, of course, because it ensures uncertainty
of outcome, both for individual matches and for the league championship
as a whole. Uncertainty of outcome is the lifeblood of any sporting event:
take away the element of uncertainty and competitive sport degenerates
125
126
127
Quirk and Fort (1992, 1999) and Scully (1995) provide fascinating descriptions of the
historical development of the institutional structures of the main North American professional team sports.This sub-section draws on these accounts.
128
holders. Inducements take the form of oers to construct lavish new stadiums at public expense: a signicant source of public subsidy for leagues
and teams that are already rich in cash.To demonstrate that threats on the
part of team owners to relocate are far from idle, Quirk and Fort (1999)
document 11 team moves that took place in MLB between 1950 and
1997; nine in the NFL; and 18 in the NBA.The leagues have been able to
maintain their monopoly positions largely because the US courts,
acknowledging the peculiarities of the economics of team sports (see
chapter 1), have tended to accept the legitimacy of restrictions on the production of sporting events that would be inadmissible elsewhere. In a
landmark ruling in 1922, the Supreme Court granted MLB exemption
from the Sherman antitrust laws. In 1961 the Sports Broadcasting Act
was passed by Congress, entitling the leagues to sell broadcasting rights
collectively on behalf of their member teams (Quirk and Fort, 1992;
Scully, 1995).
The main role of the leagues within this framework is to implement
rules aimed at furthering the collective interest of the teams in achieving
joint prot maximisation. The leagues therefore impose restrictions on
the behaviour of the teams in both the product and labour markets, which
seek to prevent any individual team from achieving a level of competitive
dominance that would be damaging to the interest that all teams share in
maintaining a reasonable degree of competitive balance. In the early
history of all three sports, the most important restriction of this kind was
the reserve clause, described in chapter 1. Having signed a contract as a
professional, the players team retained the option to renew his contract,
eectively binding the player to his present employer. The reserve clause
was justied on the grounds that it was necessary to prevent the richest
teams from outbidding the rest for the services of the top players; but its
tendency to depress players compensation was a highly convenient sideeect from the team owners perspective.
Baseballs reserve clause, eective since 1880, was challenged successfully in the Supreme Court in 1976, on the grounds that it should be
interpreted as only a one-year option clause. The Court ruled that by
playing for a year without signing a new contract, a player could satisfy his
obligations under the reserve clause, and subsequently become a free
agent. Under the 1976 settlement, all MLB players became free agents
after completing a minimum number of years of major league service
(currently six years). In basketball, the one-year interpretation had always
applied, but teams that signed a free agent had to pay compensation to his
former employer. This system was also abandoned in 1976, when moves
towards the full implementation of free agency were initiated. Players
became free agents after four years NBA service or upon expiry of their
129
130
become more rather than less concentrated since the cap was introduced
(Quirk and Fort, 1999).
In all three sports revenue sharing plays an important part in osetting
the basic inequalities in drawing power between teams. Practices
concerning the distribution of gate revenues vary between sports. NFL
home and away teams share gate revenues on a 60:40 split. In baseball,
the split is 95:5 in the NL and 80:20 in the AL. In the NBA there is no
sharing of gate revenues. Broadly speaking, national television revenues
are shared equally between all league member teams, while the home
team keeps local television revenues. Broadcasting rights in all three
sports are sold in separate packages to national and local broadcasters. In
contrast to their European counterparts, North American sports have
not pursued the pay-television option at national level. National television contracts are with the free-to-air networks ABC, CBS, NBC and
Fox, which seek to recoup their fees through advertising revenues. As in
Europe television revenues increased exponentially during the 1990s.
Quirk and Fort (1999) quote 1996 estimates for total television and
radio income of $706 million for MLB, $615 million for the NBA and
$1.29 billion for the NFL. The latter gure has since been eclipsed by a
new set of contracts yielding $2.2 billion per year between 1998 and
2005.
With the leagues maintaining a stranglehold over the supply of professional team sports, it is unsurprising that the histories of the three sports
are littered with attempts on the part of outsiders to set up rival competitions. Following the creation of the NL in 1876, there were several
attempts to establish rival leagues, including the AL which was set up in
1901. The current NLAL monopoly, achieved by agreement between
the two leagues in 1903, has survived subsequent challenges from the
Federal League (1914), the Mexican League (in the 1940s) and a threatened Pacic Coast League (which was averted in 1958, partly as a result
of the moves of the Brooklyn Dodgers and New York Giants to Los
Angeles and San Francisco, respectively). In basketball the NBA was
created in 1949 from the remains of two rival leagues. Its most powerful
challenger since has been the American Basketball Association (ABA), set
up in 1967. Mounting nancial diculties eventually led to the demise of
the ABA, with four of its surviving teams being incorporated into the
NBA in 1976. Since the creation of the American Professional Football
Association (the forerunner to the NFL) in 1920, there have been several
attempts to establish rival leagues in American football; the most successful being the fourth American Football League (AFL), set up after the
NFL had refused to award a number of expansion franchises in the late
1950s. The AFL operated alongside the NFL between 1960 and 1969,
131
before a merger incorporating the AFL teams into the NFL realised the
original aims expressed by the AFLs founders 10 years earlier.
Although there is a prolic and colourful history of attempts to create
rival leagues, only the AL has endured for more than a few years, and then
only by operating together with the NL as a de facto joint monopoly supplier. In some instances, however, rival league proposals (whether implemented or only threatened) have been successful in forcing the
established leagues to make policy changes. Clearly, there is a strong
imperative in any sport to operate a competitive structure which delivers a
single champion. But while the established leagues remain eectively
closed to new entrants, excluding a number of cities capable of supporting potentially valuable franchises, the incentives that have previously
brought so many rival league proposals to (temporary) fruition will
undoubtedly remain in place.
Of course, it is important not to exaggerate the dierences between the
markets for professional team sports in North America and Europe. As
shown, there are a number of signicant institutional dierences. But as
Fort (2000) argues, spectators wherever they live exhibit an overwhelming preference for seeing their own team win, while in both regions sports
organizations put the highest level of sports quality before the fans that
can aord to support it. And the joint ventures undertaken by teams
through their organizations make them nancially better o than they
would be acting alone (Fort, 2000, p. 451). Organisational and institutional structures have implications for the division of the spoils (both
pecuniary and non-pecuniary) among the main protagonists: owners,
players, broadcasters, spectators and governing bodies. But the similarities between North America and Europe in the underlying incentives
motivating these parties perhaps far outweigh the dierences.
Theoretical models of sports leagues: some preliminaries
As seen in chapter 1, Rottenberg (1956) rst articulated the idea that
market mechanisms can be relied upon to maintain a reasonable degree
of competitive balance among the member teams of a sports league,
without the need for extensive regulation of player compensation or
mobility by the sports governing body. In discussing the economic structure and characteristics of the North American baseball players labour
market, Rottenberg considers the case of two teams located in dierent
towns, one of which has a larger population (or potential market) than the
other. Other things being equal the marginal revenue product of a player
of a given level of ability is greater with the big-market team than with the
small-market one. But because the marginal revenue function declines as
132
the quantity of playing talent already held increases, it does not pay the
big-market team to accumulate the most talented players to the point
where complete competitive dominance is achieved.
In essence, this argument does not depend upon contractual arrangements or the structure of player compensation. It is valid under free
agency, in which case the player may be in a strong position to secure
most or all of his marginal revenue product in wage negotiations, since his
reservation wage is the wage he could command by signing for another
team. It is also valid under a reserve clause, in which case teams may have
the opportunity to drive player compensation down, towards the highest
wage the player could command in employment outside the sport.2
El-Hodiri and Quirk (1971) develop a mathematical model of an nteam professional sports league that captures these and other insights.
Specically, they demonstrate that perfect competitive balance, with all
teams having equal playing strengths, is consistent with an assumption of
prot maximising behaviour only if there is no buying and selling of
players contracts, or if the revenue functions of all teams are the same. In
the type of model presented below, the latter would require that all teams
are located in towns of equal (or similar) population size, so that each
teams potential market is the same. Since neither of these requirements is
likely to be met in practice, [i]t is not surprising, then, that casual empiricism indicates that in no professional team sport under current rules of
operation is there a tendency toward equal playing strengths (El-Hodiri
and Quirk, 1971, p. 1313).
Subsequently, Quirk and El-Hodiri (1974), Quirk and Fort (1992),
Fort and Quirk (1995), Vrooman (1995, 1996, 1997), Hausman and
Leonard (1997), Rascher (1997) and Ksenne (2000a, 2000b) have used
variants of the same model to investigate various policy issues. The main
ndings of this body of work are reviewed below, using a modied version
of the model described by Vrooman (1995). Several of these authors
present simplied and stylised versions of the original n-team model, in
which the league consists of just two teams.Without sacricing any of the
important insights available from the n-team version, the two-team model
can be derived using less algebra, and is also suitable for a diagrammatic
form of presentation.
It is assumed that team i (for i1, 2) has the following functions for
total revenue, Ri and total cost, Ci:
Prior to its gradual elimination from most US sports between the 1970s and 1990s, the
reserve clause operated in a manner similar to the original retain-and-transfer system in
English football, by tying a player for his entire career to the holder of his registration.
Ri rPiaWib
Ci cP iW id
133
(3.1)
where
Ri
Ti
PiaWi 1
MCi c
(3.2)
134
the extra unit, MCi. As shown below, however, depending on the assumptions made about the rules governing player compensation, MRi MCi
may or may not be an attainable condition in the closed model, owing to
the limited availability of playing talent. If the stock of talent is xed and
the cost of hiring talent is low, all teams may be forced to operate with
marginal revenue above marginal cost.
If both teams face the same marginal cost of talent ( 0 and 1 in
(3.1)), the marginal revenues of both teams should be the same in equilibrium (MR1 MR2). If MR1
MR2, a transfer fee can be found at which
it is attractive to both teams for some talent to be transferred from team 2
to team 1. If the transfer fee is pitched somewhere between the discounted values of the future revenue gains to team 1 and losses to team 2,
both teams are made better o by the transfer. It is protable for both
teams to continue to trade in this way, until the condition MR1 MR2 is
attained.
This equilibrium condition MR1 MR2 can be used to obtain an
expression for the equilibrium ratio of win ratios, a simple measure of
competitive balance, as follows:
W1/W2 (P1/P2)/(1)
(3.3)
135
MR,MC
MR2(P2)
D
B
c*
A
c
MR1(P2)
O
0.5
T1*
MR1(P1)
T1
Figure 3.1 Competitive balance under reserve clause and under free
agency
136
Although it would be perhaps easy for club owners to collude to hold down the wage in a
two-team league, the two-team model is only a stylised version of the more realistic nteam case, in which eective collusion would be harder, though perhaps not impossible
to achieve.
137
138
MR,MC
MC1
MR2 (P2)
D1
B
A
D2
c
MR1(P1)
MC2
T1** T1*
T1
that, over time, there has been a trend towards the erosion of both explicit
and implicit mechanisms for revenue sharing in English league football.
Intuition might suggest that by causing the gap between rich and poor to
widen, this is likely to create greater competitive imbalance. But as before,
intuition turns out to be misleading. Suppose each team keeps only a proportion, , of its own gate revenues, while the remaining (1) are
awarded to the other team (with 0.5 1).The net revenues of teams 1
and team 2, after sharing, are as follows:
NR1 R1 (1 )R2
(3.4)
NR1
T1
NR2
T2
(3.5)
139
MR,MC
MR 2 (P2)
MNR 2 (P2)
B
c*
E
c
MR1(P1)
MNR1(P1)
O
T 1*
T1
140
Fort and Quirk (1995) show that this conclusion is qualied if, as is actually the case in
North America, revenue comprises two components: gate revenue which is shared, and
local TV revenue which is captured wholly by the home team. In this case, variation in
the proportion of gate revenue which is shared does aect the distribution of talent
between the two teams.
141
the domestic league. For simplicity, it can be assumed that the wage per
unit of talent, c, is determined in the international players labour market,
and is therefore exogenous to the model.
This formulation seems more useful than the closed model for the
analysis of European football, especially since the liberalisation of various
institutional features of the players labour market in the mid-1990s. In
fact, it is of some interest to compare directly the equilibrium properties
of the closed and open models. Although the analogy is rather stylised, the
pre-1995 regulations in European football, under which teams could eld
no more than three overseas players, might be interpreted as comparable
to a (partially) closed model, in which teams had to rely primarily on the
domestic transfer market to improve their playing squads. Since 1995,
however, the labour market has become very much more open, with many
of the top teams now more likely to acquire new players from continental
Europe or further aeld,5 than from other clubs in their domestic leagues
(see also chapter 4).
In the open model, the marginal revenues of teams 1 and 2, obtained by
dierentiating (3.1) under the assumption that the total stock of playing
talent T1 T2 is variable, are as follows:
MR1
MR2
R1
T1
R2
T2
P a1 W 1 1 T2/(T1 T2)2
P2aW2 1 T1/(T1 T2)2
(3.6)
(3.7)
For any given set of values for P1, P2, and , (3.7) produces a smaller
equilibrium value of W1/W2 in the open model than does (3.3) in the
closed model. In the open model, the acquisition of an extra unit of talent
by one team does not deplete the talent held by the other. Its impact on
the acquiring teams win ratio and revenue is therefore smaller than in the
5
142
T1
1'
C
1''
T2
Figure 3.4 Derivation of teams iso-prot curves, open model
closed model. It becomes harder for the big-market team to achieve dominance; by buying players it strengthens its own squad, but it does not
weaken the other teams squad. The big-market team does not nd it
protable to dominate the league to the same extent as in the closed
model, so competition becomes more balanced.
At this point, the algebra for the open model becomes more complex
and will not be pursued. A diagrammatic treatment of this model,
however, is still possible. Because T1 and T2 can vary independently, a
new diagram showing iso-prot curves for the two teams is required. An
iso-prot curve for team 1 shows all combinations of values for T1 and T2
that produce identical prots for team 1. For each value of prot that
team 1 could earn, there is a separate iso-prot curve. Similarly there is
another complete set of iso-prot curves for team 2.
Figure 3.4 illustrates the construction of iso-prot curves for team 1.
Starting at A, where T1 is small relative to T2, team 1 earns a certain
prot, say 1, but could improve by employing more talent (especially as
team 1 has the higher home-town population). As T1 is increased while T2
is held constant, B is approached.Team 1s win ratio increases, and so too
does its revenue, by an amount that exceeds its increase in costs.Team 1s
143
144
C
T1
W
T1''''
T1'''
B
P
=0
2
1=0
team 1
iso-profit
X
T1'
L
N
T1''
45
team 2
iso-profit
F
T2'' T2'
T2'''' T2'''
T2
reaction by earning a higher prot at point M than at L. In the terminology of classical duopoly theory, M is a Stackelberg equilibrium.
It is worth noting that MR1
MR2 at M. Team 1 refrains from buying
players (either from team 2 or from outside the league), however,
because it anticipates that if it did so, team 2 would respond in kind, and
the equilibrium would quickly revert back towards L. If, as described
above, the big-market team acts strategically (by using its ability to anticipate the other teams reaction to enhance its prot), competition
becomes more balanced than at the CournotNash equilibrium.
145
An unrealistic feature of this formulation is that both teams could become more
protable than at point L by simultaneously reducing their stocks of talent in the same
proportions, moving south-west along the ray OB towards the origin. If revenue
depends only on the relative values of T1 and T2 (which determine the win ratio) and not
on their absolute values, both teams would have incentives to shed all of their talented
players and eld less expensive journeymen. To prevent this from happening in the
model, revenues could (realistically) be made partly dependent on the quality of talent
elded by the home team or by both teams.This kind of adjustment adds to the algebra,
but does not aect any of the main qualitative conclusions of this sub-section. Hausman
and Leonard (1997) also make each teams revenue a function of the quality of other
teams, in order to study the positive externality which arises if superstars help generate
gate or television revenues for teams other than their own (see below).
Ferguson et al. (1991) use a standard approach to test the assumption of prot maximising behaviour on the part of teams participating in the North American National Hockey
League.The results tend to support the prot maximising hypothesis.
146
this is the only point at which team 2 can break even if team 1 hires T1.
At Q team 1 has a higher win ratio and higher revenue than at P, and also
makes a positive prot. Competition, however, has become still more
unbalanced. Since team 1 is not bound by the prot constraint at Q,
would it be advisable for team 1 to invest even further in additional talent?
The answer is probably not. Anywhere to the north of Q, team 2 is
unable to break even at any value of T2, and is therefore insolvent.
Because of the joint nature of production, this outcome is presumably in
neither teams interest.
This sub-section concludes with a summary of the main ndings.
Other things being equal, a change from a closed to an open player labour
market should make competition within the league more balanced.
Under open conditions, it becomes harder for a rich team to use its
spending power to achieve league dominance by monopolising the best of
a limited supply of talent. In the open model, competition should be more
balanced if teams pursue prot maximising objectives than if they pursue
traditional sporting objectives of win ratio or revenue maximisation,
subject to a nancial constraint. Prot maximising teams tend not to
over-spend on players, and this makes for a more balanced league competition.With constrained revenue maximisation, the strength of each clubs
nancial muscle is the decisive factor in determining expenditure on
talent, so competition tends to be less balanced. In either case, strategic
behaviour by the big-market club is likely to strengthen these conclusions.
The two-team closed model does of course emphasise the degree of
interdependence between the teams, which would be lower in an n-team
league. On the other hand, the comments of Kevin Keegan, the buying
manager in the 15 million transfer of Alan Shearer from Blackburn
Rovers to Newcastle United in the summer of 1996, demonstrate that
team managers at the top level are highly conscious of the interdependence of their actions. During the 1996 season, Newcastle had been
Manchester Uniteds closest rival for the Premier League title.
[W]hatever happened, we knew that if we had not beaten Manchester
United to Shearers signature, it would have nished the league as a
contest . . . The combination of Shearer and United, the way they played
and the pace they possessed, would have been devastating, and we could
not let it take place (Keegan, 1999, p. 327).
Empirical investigations and extensions
The most popular empirical test of the propositions of the closed model
for North American sports is based on a simple measure of competitive
balance in any one season:
147
148
over the previous three, four or ve seasons, they nd that teams that draft
early tend to benet subsequently by winning signicantly more games.
This suggests the draft does tend to enhance competitive balance, contrary to the invariance result.
Taking a longer-term perspective, Eckard (1998) examines the eect
on competitive balance in college football of a range of restrictions on
player recruitment, eligibility and compensation that were introduced by
the NCAA in 1952. Pooled cross-sectional time series data is used to
decompose the overall variance in win ratios into a time component for
each team and a cumulative component across teams:
(W
i,t
0.5)2/NT
(W
i,t
Wi )2/NT
(W 0.5) /N
2
(3.8)
tion in the rst term and an increase in the second term on the right-hand
side of (3.8).This suggests that the regulations made competition less balanced: dierences between strong and weak teams became more
entrenched, resulting in a reduction in the variation of each teams performance over time and an increase in the variation in average performance
between teams.This nding is conrmed using a variety of other competitive balance indicators, including several championship dominance
measures.
Hausman and Leonard (1997) demonstrate empirically that between
1989 and 1992, the match attendances and television ratings achieved by
clubs generally in the NBA were enhanced signicantly in matches
involving three basketball superstars: Larry Bird, Michael Jordan and
Magic Johnson. As well as generating revenue for their own teams, superstars also have a positive eect on the revenues of opponents.This positive
externality may have implications for competitive balance, and for the
ecient distribution of talent between the leagues member clubs. Under
prot maximising assumptions, small-market teams tend to free ride and
hire less talent than they would in the absence of the externality.The distribution of talent is therefore weighted more heavily in favour of bigmarket teams, and competition is less balanced. The equilibrium is also
inecient, in the sense that it fails to maximise league revenues.
149
Hausman and Leonard argue against the salary cap as a mechanism for
correcting this imbalance. They suggest that the salary cap is likely to
cause a substantial departure from the ecient equilibrium in the other
direction, distributing talent too heavily in favour of small-market teams,
and over-compensating for the eect of the superstar externality.
In a recent application of the closed model to European football,
Hoehn and Szymanski (1999) investigate the implications for competitive balance in the domestic league of involvement of big-market teams in
European competition.While team 2s revenue function depends only on
its win ratio against team 1 in domestic competition, team 1s total revenues are obtained from both domestic and European competition. In the
latter, team 1 competes against a European team with a similar (large)
home-town population.Team 1 has an incentive to invest more heavily in
talent than it otherwise would, in order to achieve a reasonable win ratio
in Europe and maximise prots from both competitions. The eect is to
make competition more unbalanced in the domestic league. The higher
the proportion of team 1s total revenue derived from Europe, the greater
is the degree of domestic imbalance. The decline in team 2s win ratio
reduces team 2s revenue, perhaps to the point where prots become
unattainable and losses inevitable. At this point the domestic league may
cease to be nancially viable, and may even close down. To avoid this
outcome, Hoehn and Szymanski advocate the creation of a European
league for the leading teams, which would entail their complete withdrawal from their domestic leagues.The present arrangements, requiring
these teams to participate simultaneously in both domestic and European
competition, are unlikely to be sustainable in the long term for the
reasons outlined above.
Hoehn and Szymanski provide an analysis of a situation that manifestly
does create serious tensions for the leading teams when balancing their
domestic and European commitments. Not least among the concerns are
the sheer physical demands on players of regular involvement in two
matches per week.To some extent, however, these pressures may eventually help foster competitive balance in the domestic league: teams playing
in Europe may choose to rest key players for domestic matches, or may
lose matches they would normally expect to win as a result of tiredness or
de-motivation. As yet, experience of regular league competition at
European level may be insucient for clubs or policy makers to be completely condent that gate and television revenues would hold up if
Europe were to become the norm for the leading clubs, rather than a
glamorous distraction. The prospects for the eventual creation of a
European Super League are discussed further in chapter 9.
150
151
Table 3.1. Percentages of home wins, draws and away wins and average
number of goals scored by home and away teams per season, English league,
19731999
Season
Home wins
(per cent)
Draws
(per cent)
Away wins
(per cent)
Average goals
per game:
home team
Average goals
per game:
away team
1973
1974
1975
1976
1977
1978
1979
1980
1981
1982
1983
1984
1985
1986
1987
1988
1989
1990
1991
1992
1993
1994
1995
1996
1997
1998
1999
51.7
48.9
52.4
50.2
51.9
50.1
48.0
49.9
49.6
47.1
52.0
50.8
50.1
50.1
48.9
46.1
46.8
46.8
48.4
47.3
46.5
45.6
46.4
45.0
46.4
47.9
45.0
28.4
30.2
27.9
27.7
27.9
30.9
29.6
27.9
28.0
27.4
26.3
26.0
24.3
24.6
27.4
27.3
28.6
27.3
27.6
27.1
26.3
27.4
27.5
29.6
28.1
27.8
28.0
19.9
21.0
19.7
22.1
20.2
19.0
22.4
22.2
22.4
25.5
21.6
23.2
25.6
25.3
23.7
26.6
24.6
25.9
24.1
25.6
27.1
27.0
26.1
25.3
25.5
24.3
27.0
1.57
1.51
1.59
1.59
1.66
1.62
1.55
1.58
1.53
1.53
1.72
1.69
1.62
1.69
1.55
1.55
1.58
1.54
1.56
1.51
1.57
1.53
1.51
1.48
1.47
1.51
1.46
0.92
0.92
0.92
0.99
0.99
0.98
1.01
0.98
0.94
1.07
1.06
1.06
1.10
1.11
1.05
1.08
1.08
1.09
1.06
1.07
1.12
1.14
1.08
1.06
1.03
1.05
1.08
season periods (seasons 1973 to 1981 inclusive; 1982 to 1990; and 1991
to 1999). Evidently the post-1982 improvement in the performance of
away teams identied for the English league as a whole in table 3.1 also
applies to the Division 1 teams shown in table 3.2. For the other four
countries, the data are similarly aggregated, except for the nal nine-year
period which is split into two sub-periods at the point when three points
for a win was introduced: the 1995 season in France and Scotland, and
the 1996 season in Germany and Spain.
The results for France, Germany and Spain indicate that long-term
152
Table 3.2. Percentages of home wins, draws and away wins and average
number of goals scored by home and away teams per season, international
comparisons, 19731999
Seasons
Home wins
(per cent)
Draws
(per cent)
Away wins
(per cent)
Average goals
per game:
home team
Average goals
per game:
away team
England
197381
198290a
19919a
49.6
48.4
45.7
28.7
26.3
28.5
21.7
25.3
25.8
1.58
1.58
1.52
1.00
1.07
1.09
France
197381
198290
19914
19959a
57.0
55.1
51.7
49.2
25.1
27.4
31.7
31.0
18.0
17.4
16.6
19.8
1.92
1.67
1.45
1.47
1.01
0.87
0.76
0.87
Germany
197381
198290
19915
19969a
54.3
50.9
45.4
44.9
25.5
27.7
31.2
29.3
20.1
21.5
23.4
25.8
2.29
2.18
1.76
1.73
1.32
1.26
1.19
1.22
Scotland
197381
198290
19914
19959a
46.7
45.4
40.7
43.4
24.7
25.3
29.3
27.3
28.5
29.3
30.0
29.2
1.62
1.53
1.36
1.51
1.21
1.14
1.11
1.16
Spain
197381
198290
19915
19969a
61.6
53.6
50.4
47.9
25.1
26.9
28.4
27.1
13.3
19.5
21.2
25.0
1.74
1.57
1.51
1.60
0.79
0.88
0.94
1.09
Note:
a
Denotes three league points awarded for a win.
Source: Match results data set.
erosion of the importance of home advantage is very much an international phenomenon, which has in fact been taking place at a faster rate in
these three countries than in England. No doubt this partly reects the
fact that home advantage counted for more in these countries than it did
in England in the 1970s. Since then a process of international convergence has been underway; the dierences between countries were very
much greater at the start of the period covered by table 3.2 than at the
153
154
Goals
scored
by
the
home
team
0
1
2
3
4
5
6
Total
6
Total
19.3
12.0
19.3
15.1
12.0
10.7
10.2
38.6
16.7
13.2
19.4
14.7
11.7
10.7
10.2
36.6
13.2
15.1
15.2
12.4
10.9
10.4
10.2
17.3
1.1
2.0
1.3
0.9
0.3
0.2
0.0
5.9
0.3
0.5
0.3
0.1
0.1
0.0
0.0
1.3
0.0
0.1
0.0
0.0
0.0
0.0
0.0
0.3
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.1
120.7
132.9
125.6
113.2
115.1
111.9
110.6
100.0
Goals
scored
by
the
home
team
0
1
2
3
4
5
6
Total
6
Total
17.8
10.6
18.2
14.4
12.2
10.7
10.3
34.3
17.2
12.5
19.5
14.8
12.0
10.7
10.3
37.0
13.5
15.9
15.2
12.7
11.0
10.4
10.1
18.8
1.6
2.2
1.7
1.0
0.5
0.1
0.0
7.3
0.4
0.7
0.4
0.2
0.1
0.1
0.0
2.0
0.1
0.2
0.1
0.1
0.0
0.0
0.0
0.5
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.1
120.7
132.2
125.2
113.3
115.8
112.0
110.8
100.0
of all matches.The results in table 3.4 for P2 are very similar. A comparison of both of these sets of results with those for P1 reported in table 3.3
lends further credence to the suggestion that the change from two to three
points for a win had the desired eect and encouraged attacking play, particularly by away teams. In P1, the proportion of matches that nished 00 was 9.3 per cent, a higher percentage than in either of the two
subsequent periods. Similarly among other low-scoring results, the proportions of 1-0 home wins (12.0 per cent) and 1-1 draws (13.2 per cent)
were higher in P1 than subsequently. There was, however, a smaller
155
Goals
scored
by
the
home
team
0
1
2
3
4
5
6
Total
6
Total
18.6
11.2
18.1
14.3
11.4
10.6
10.2
34.4
17.5
12.9
19.4
14.5
11.6
10.6
10.2
36.7
14.0
16.4
14.9
12.5
10.9
10.3
10.1
19.2
1.4
2.3
1.7
1.1
0.4
0.1
0.0
7.1
0.4
0.6
0.5
0.3
0.1
0.0
0.0
2.0
0.1
0.1
0.2
0.0
0.0
0.0
0.0
0.4
0.0
0.1
0.0
0.0
0.0
0.0
0.0
0.2
122.2
133.6
124.9
112.7
114.5
111.6
110.6
100.0
proportion of 0-1 away wins before the introduction of three points for a
win (6.7 per cent) than after.
The nal columns and rows of tables 3.33.5 show the marginal probability distributions of goals per match scored by the home and away
teams, respectively, calculated by summing horizontally across the other
columns and vertically down the other rows of each table. For example,
table 3.5 indicates that the home team failed to score in 22.2 per cent of
matches played in P3, and the home team scored once in 33.6 per cent
of these matches. Similarly, the away team failed to score in 34.4 per cent
of these matches, and scored once in 36.7 per cent of these matches. For
the away teams in particular, there has been a marked change in the shape
of the marginal probability distribution of goals scored.The modal (most
frequent) number of away goals scored changed from zero in P1, to one in
P2 and P3. After increasing from 0.98 in P1 to 1.08 in P2, the average
number of away team goals per game remained unchanged in P3. For the
home teams, the marginal distribution of goals scored was similar in
periods 1 and 2, but dropped signicantly in P3.The average numbers of
goals scored by the home team were 1.58, 1.61 and 1.51, respectively in
the three periods. Overall, the improvement in the performance of away
teams over the last two decades was attributable initially (in the 1980s) to
an improvement in their oensive or goal scoring prowess, and has been
sustained subsequently (in the 1990s) by an improvement in their defensive capacity to avoid conceding goals.
Tables 3.6 and 3.7 show the conditional distributions for the numbers
of goals scored by the home and away team.To save space, only the results
156
Goals
scored
by
the
home
team
0
1
2
3
4
5
6
Average
6
25.1
32.5
23.6
12.4
14.2
11.6
10.6
11.45
20.4
35.2
25.7
12.1
14.4
11.6
10.6
11.52
21.0
33.2
25.7
13.1
14.7
11.6
10.7
11.55
20.0
32.5
24.2
15.5
16.1
11.1
10.5
11.61
22.1
29.6
24.6
15.6
15.9
11.7
10.6
11.61
20.0
32.5
36.3
15.0
15.0
11.3
10.0
11.46
20.6
41.2
17.6
15.9
14.7
10.0
10.0
11.53
for P3 covering the 1991 to 1999 seasons inclusive are included.The conditioning is on the number of goals scored by the opposing team. The
conditional distributions can be used to identify whether there are interdependencies between the numbers of goals scored by the home and
away teams in each match, or whether the numbers of goals scored by the
two teams can be considered as approximately independent.
For example, the rst column of table 3.6 shows, for matches in which
the away team failed to score, the proportions of occasions on which the
home team scored 0, 1, 2, 3, etc. goals. In 25.1 per cent of matches in
which the away team failed to score, therefore, the home team also failed
to score; in 32.5 per cent of matches in which the away team failed to
score, the home team scored once, and so on.The second column shows
the same for matches in which the away team scored once. In table 3.7,
the conditioning is reversed. The rst row shows, for matches in which
the home team failed to score, the proportions of occasions on which the
away team scored 0, 1, 2, 3 goals, and so on. The second row shows the
same for matches in which the home team scored once.
If the number of goals scored by the home team and the away team are
independent of each other, then the number of away goals should not
aect the probabilities for the number of home goals. This means the
probability distributions in each column of table 3.6 should look similar
to one another (and similar to the marginal distributions for the number
of home goals in table 3.5). Likewise, the number of home goals should
not aect the probabilities for the number of away goals, so the probability distributions in each row of table 3.7 should also look similar to one
157
Goals
scored
by
the
home
team
0
1
2
3
4
5
6
6
Average
39.0
33.3
32.7
33.7
31.7
35.2
33.6
33.8
38.4
37.9
35.1
35.3
37.2
37.4
18.2
19.0
19.8
19.8
19.7
20.0
21.5
6.4
6.9
6.9
8.7
9.6
5.2
5.6
1.9
1.7
1.9
2.4
2.5
2.1
1.9
0.4
0.4
0.6
0.2
0.5
0.3
0.0
0.2
0.2
0.1
0.1
0.6
0.0
0.0
1.00
1.08
1.10
1.12
1.20
1.03
1.05
another (and similar to the marginal distributions for the number of away
goals in table 3.5). In fact, however, the probabilities do not appear to
conform to this condition for independence. In table 3.7, for example, if
the home team scores 0, then 0 is also the most likely score for the away
team (with a probability of 0.390). The probability that the away team
scores 1 is lower, at 0.338. If the home team scores 1, however, then the
probability that the away team scores 0 is only 0.333, while the probability that the away team also scores 1 is higher, at 0.384. The conditional
mean scores reported in the nal column of table 3.7 indicate that the
(conditional) average number of away goals was an increasing function of
the number of home goals over the range 0 to 4 home goals. Similarly,
according to the bottom row of table 3.6, the (conditional) average
number of home goals was an increasing function of the number of away
goals over the range 0 to 4 away goals.
In common with the results presented by Dixon and Coles (1997),
these tabulations suggest that the greatest degree of interdependence
occurs in low-scoring games, since the largest discrepancies between the
probabilities in adjacent rows or columns are towards the top left-hand
corners of tables 3.6 and 3.7. Nevertheless, it is apparent that the shapes
of the conditional probability distributions do vary to some extent over
the full range of values for the conditioning variable. The numbers of
goals scored by the home and away teams are interdependent. This fact
complicates the task of modelling match results through scores. As will be
seen in section 3.3, many of the diculties can be avoided by modelling
match results directly (as home win, draw or away win) rather than indirectly (through the home and away team scores). Modelling results
158
159
42
32
31
Consecutive wins
Reading
Newcastle
Liverpool
1986
1993
1991
13
13
12
63
51
18
16
15
15
20
17
16
16
15
17
17
17
17
17
17
17
17
31
27
Consecutive losses
Walsall
Brighton
Stoke
West Bromwich
15
12
11
11
1989
1973
1986
1996
16
16
16
16
16
10
18
18
46
42
41
18
15
15
15
15
15
15
Note: Seasons quoted are the seasons in which each run ended.
Source: Match results data set.
0.485
0.500
0.510
0.526
0.540
0.558
0.581
0.609
0.635
0.672
0.485
0.474
0.459
0.445
0.441
0.419
0.401
0.398
0.350
0.485
0.511
0.528
0.545
0.537
0.603
Number of
Number of
previous matches previous
without a win
consecutive wins
Uncond.
N 1
N 2
N 3
N 4
N 6
N 8
N 10
N 15
N 20
Number of
previous
matches
unbeaten
N
0.485
0.464
0.440
0.414
0.408
0.349
0.278
Number of
previous
consecutive losses
0.485
0.496
0.505
0.516
0.521
0.534
0.544
0.559
0.578
0.586
Number of
previous home
matches unbeaten
0.485
0.463
0.450
0.428
0.415
0.412
0.390
0.419
Number of
previous home
matches without
a win
0.485
0.508
0.524
0.545
0.549
0.551
0.594
Number of
previous
consecutive
home wins
0.485
0.448
0.429
0.390
0.381
Number of
previous
consecutive
home losses
0.238
0.223
0.213
0.201
0.191
0.171
0.151
0.124
0.110
0.105
0.238
0.249
0.262
0.273
0.278
0.295
0.308
0.317
0.346
0.238
0.214
0.198
0.180
0.177
0.140
Number of
Number of
previous matches previous
without a win
consecutive wins
Uncond.
N 1
N 2
N 3
N 4
N 6
N 8
N 10
N 15
N 20
Number of
previous
matches
unbeaten
N
0.238
0.260
0.277
0.294
0.299
0.340
0.407
Number of
previous
consecutive losses
0.238
0.229
0.221
0.213
0.210
0.197
0.193
0.184
0.177
0.166
Number of
previous home
matches unbeaten
0.238
0.258
0.271
0.286
0.298
0.317
0.351
0.374
Number of
previous home
matches without
a win
0.238
0.218
0.202
0.187
0.177
0.158
0.144
Number of
previous
consecutive
home wins
0.238
0.269
0.294
0.311
0.344
Number of
previous
consecutive
home losses
0.238
0.246
0.257
0.264
0.279
0.304
0.314
0.349
0.385
0.385
0.238
0.229
0.223
0.212
0.209
0.195
0.191
0.187
0.191
0.238
0.251
0.270
0.290
0.308
0.374
Number of
Number of
previous matches previous
without a win
consecutive wins
Uncond.
N 1
N 2
N 3
N 4
N 6
N 8
N 10
N 15
N 20
Number of
previous
matches
unbeaten
0.238
0.220
0.211
0.209
0.220
0.142
Number of
previous
consecutive losses
N
0.238
0.252
0.269
0.283
0.298
0.332
0.349
0.338
Number of
previous away
matches unbeaten
0.238
0.231
0.224
0.220
0.215
0.205
0.196
0.183
0.170
0.159
Number of
previous away
matches without
a win
0.238
0.260
0.288
0.334
0.347
Number of
previous
consecutive
away wins
0.238
0.224
0.215
0.209
0.201
0.188
0.185
0.168
Number of
previous
consecutive
away losses
0.485
0.473
0.458
0.446
0.431
0.405
0.396
0.357
0.320
0.333
0.485
0.500
0.509
0.525
0.529
0.553
0.561
0.593
0.590
0.485
0.465
0.437
0.420
0.407
0.330
Number of
Number of
previous matches previous
without a win
consecutive wins
Uncond.
N 1
N 2
N 3
N 4
N 6
N 8
N 10
N 15
N 20
Number of
previous
matches
unbeaten
N
0.485
0.514
0.532
0.550
0.565
0.653
Number of
previous
consecutive losses
0.485
0.464
0.446
0.430
0.413
0.382
0.361
0.390
Number of
previous away
matches unbeaten
0.485
0.495
0.503
0.509
0.517
0.525
0.531
0.543
0.543
0.561
Number of
previous away
matches without
a win
0.485
0.455
0.421
0.361
0.359
Number of
previous
consecutive
away wins
0.485
0.508
0.521
0.534
0.543
0.562
0.584
0.575
Number of
previous
consecutive
away losses
164
165
if
if
if
2 i j i, j
1 i j i, j 2
i j i,j 1
(3.9)
166
10
See, for example, Greene (1999).The details of the ordered probit model are described
in more detail in section 3.3, where a forecasting model for match results is developed.
200 simulations of the 27-season period are used to generate the results reported below.
167
Column (4) shows the mean number of unbeaten spells of each duration
across the 200 replications. Columns (3) and (5) show the 5 per cent and
95 per cent percentiles for the number of unbeaten spells of each duration.
The range of values between columns (3) and (5) can be interpreted as a
90 per cent condence interval for the number of spells of each duration,
based on the assumption that there are no short-term persistence eects.
For example, 2,619 unbeaten spells of at least 10 matches should be
observed if this assumption is correct. Making reasonable allowance for
the random or unpredictable element in match results, any number of
such spells between 2,435 and 2,824 would be consistent with the hypothesis that there are no short-term persistence eects. The actual number
shown in column (2), however, falls outside this interval, at 2,366.
This example illustrates a consistent feature of the simulation results
for all four types of spell considered in table 3.13: the actual numbers of
spells of any duration are almost always smaller than the expected
numbers, and often below the lower limit of the 90 per cent condence
interval. For unbeaten spells, the null hypothesis of no short-term persistence eects is therefore rejected at the 10 per cent signicance level at
all durations between six and 13 matches. For win-less spells, the result is
even stronger: the same null hypothesis is rejected at any duration of two
or more matches. For spells of consecutive wins and consecutive losses,
the results are similar.The actual numbers fall below the lower limit of the
90 per cent condence interval at all durations of two or more matches for
spells of consecutive wins, and for all durations between three and nine
matches for spells of consecutive losses.
As discussed above, these ndings suggest that there are negative shortterm persistence eects, which tend to curtail the durations of both good
and poor spells. A team currently enjoying a run of good results has a
higher probability (relative to the probability conditioned only on the
teams underlying strength) of a poor result in its next match, and a team
on a run of poor results has a higher probability of a good result in its next
match. Negative short-term persistence eects of this kind tend to curtail
the durations of both good and poor spells, accounting for the patterns
evident in the comparison between the actual and the simulated results
shown in table 3.13.
3.3 Modelling and forecasting match results
Previous literature on modelling football results
In section 3.3 a regression model is developed to describe the patterns
that can be identied in the 27-year English league match results data set
Actual
(2)
69,762
44,696
29,255
19,527
13,265
9,158
4,565
2,366
580
163
Duration
(matches)
(1)
1
2
3
4
5
6
8
10
15
20
69,507
44,435
29,084
19,465
13,253
9,212
4,624
2,435
563
141
Lower
5 per cent
(3)
69,751
44,747
29,438
19,806
13,583
9,487
4,854
2,619
668
204
Mean
(4)
Expected
69,980
45,055
29,807
20,182
13,952
9,814
5,121
2,824
772
268
Upper
5 per cent
(5)
69,775
44,463
28,875
19,140
12,733
8,589
4,047
1,982
361
64
Actual
(6)
69,531
44,512
29,141
19,473
13,214
9,121
4,478
2,303
484
112
Lower
5 per cent
(7)
69,776
44,815
29,494
19,816
13,537
9,391
4,707
2,473
583
168
Mean
(8)
Expected
70,016
45,143
29,850
20,161
13,850
9,666
4,923
2,649
709
240
Upper
5 per cent
(9)
Table 3.13. Actual and expected numbers of unbeaten and win-less spells, and spells of consecutive wins and losses, of
various durations
39,592
14,284
5,422
2,174
888
382
88
21
1
2
3
4
5
6
8
10
39,322
14,381
5,657
2,394
1,059
487
110
24
Lower
5 per cent
(3)
Actual
(2)
Duration
(matches)
(1)
39,575
14,621
5,847
2,514
1,142
547
142
41
Mean
(4)
Expected
39,813
14,846
6,028
2,635
1,218
612
184
64
Upper
5 per cent
(5)
39,582
14,511
5,605
2,259
931
411
92
21
Actual
(6)
39,346
14,365
5,613
2,316
990
439
93
20
Lower
5 per cent
(7)
39,574
14,578
5,759
2,425
1,071
497
123
36
Mean
(8)
Expected
39,785
14,789
5,917
2,529
1,154
557
153
54
Upper
5 per cent
(9)
170
described above. Having been estimated from historical data, this model
can be used to assess probabilities and generate forecasts for future match
results.
In the academic literature, there are relatively few studies that seek to
model match results data for football. Early contributions by Moroney
(1956) and Reep, Polard and Benjamin (1971) use the poisson and negative binomial distributions to model the distributions of the numbers of
goals scored per game. The aggregated approach adopted, however, precludes the generation of specic forecasts for individual matches based on
information about the respective strengths of the two teams concerned.
By comparing nal league placings with experts pre-season forecasts,
Hill (1974) demonstrates that individual match results do nevertheless
have a predictable element, and are not determined solely by chance.
Maher (1982) develops a model in which the home and away team
scores follow independent poisson distributions, with means which are
the product of parameters reecting the attacking and defensive capabilities of the two teams. If H denotes the goals scored by home team i and A
the goals scored by away team j, the respective probability functions are:
P(Hh)exp(ij ) (ij )h / h!
and
P(Aa)exp(ij) (ij)a / a!
where i and j reect the attacking capability of team i at home and the
defensive capability of team j away, and i and j reect the defensive
capability of team i at home and the attacking capability of team j away.
Tests show that i and i can be regarded as proportional to i and i
respectively, so it is necessary to estimate only one set of attacking and
one set of defensive parameters for each team. This can be done ex post,
after the full set of match results has been observed, using maximum likelihood methods. The model does not predict scores or results ex ante.
Although goodness-of-t tests show that the model provides a reasonably
accurate approximation to the data, separate examination of the observed
and expected distributions of the dierence between the scores of the two
teams reveals a tendency to under-estimate the proportion of drawn
matches. This is attributed to interdependence between the scores of the
home and away teams, and is corrected by modelling scores using a bivariate poisson distribution. The marginal distributions are the same as
before, but allowance is also made for positive correlation between the
home and away team scores in each match.
Dixon and Coles (1997) employ Mahers (1982) modelling approach
for dierent ends: they seek to develop a forecasting model capable of
generating ex ante match outcome probabilities. Instead of using the
171
172
Of course, the incentive for a trailing team to play recklessly in the hope of
grabbing a goal increases as the end of the match draws near. So perhaps
the observed tendency for the scoring rate to increase simply mirrors an
increase (as time elapses) in the probability of the scores being unequal.
In fact Dixon and Robinson nd some evidence to support both hypotheses. The scoring rate does depend upon whether the scores are currently
level; but even after controlling for the current score, the scoring rate still
increases with the duration of the match elapsed.
Finally, Knings (2000) approach to the modelling of match results
(rather than individual team scores) using ordered probit estimation has
been described in section 3.2. In common with Maher (1982), the aim is
to obtain a set of parameters that describes a collection of match results ex
post, rather than to generate forecasts or estimated probabilities ex ante.
Kning advocates modelling results directly, rather than indirectly
through scores, partly on grounds of simplicity: fewer parameters are
required, the estimation procedures are more straightforward and the
specied ordered probit model lends itself quite easily to the inclusion of
dynamics or other explanatory variables. Furthermore, the thorny
problem of interdependence between the home and away team scores is
nessed completely using this approach.
Kuypers (2000) uses a variety of explanatory variables drawn from
recent match results to estimate an ex ante forecasting model, also using
ordered probit regression.These include: the average points per game and
the cumulative points attained by the home and away teams in the current
season; the league positions and goal dierences of the two teams; and the
points and goal dierences obtained by the two teams from the last three
games.The main focus of the paper, however, is on investigating whether
the forecasts provide a basis for protable xed-odds betting on football
match results. The full specication and estimations of the forecasting
model are not presented. In the remainder of this section an explicit
ordered probit regression model is developed to generate estimated ex ante
probabilities for match outcomes, based on historical match results data.
Specication of the model
For the purposes of describing the construction of the model, it is
assumed initially that the underlying process causing matches to nish
either as home wins, draws or away wins (or using statistical terminology,
the underlying data generating process) was unchanged throughout the
27-year observation period (covering the seasons from 1973 to 1999
inclusive). As seen in section 3.2, this is an over-simplication. Having
developed and estimated the model using data for the full 27-year period,
173
m 2 = 0.038
m 1 = 0.713
0.238
0.277
y *= 0
0.485
Normal random
variable
174
175
m 2 = 0.038
(a)
m 1 = 0.713
0.208
0.267
0.525
y * = 0.1
Normal random
variable
m 2 = 0.038
(b)
m 1 = 0.713
0.271
0.285
y * = 0.1
0.444
Normal random
variable
176
strengths and the recent form of the two teams be used to dene the most
appropriate value of y* for the match in question? The answer comes by
exploiting the wealth of statistical evidence in the 27-year data set, to construct a formula to convert suitably dened team strength, recent form
and match signicance variables into values of y*. The formula is chosen
so that the values of y* which it delivers will generate match result probabilities that correspond as closely as possible to the patterns of home
wins, draws and away wins that actually occurred in the data set.
The expression used to generate the value of y* for the match between
home team i and away team j, now denoted y*i,j , is as follows:
1
d1
1
d0,1 P di,0,1
d1
1
00,0 P 0j,0,0
d1
m1
1S H
2 S Ai,j
i,j
d2
1
d0,1 P dj,0,1
d1
HH
RH
m
i,m
2
d1,1 P di,1,1
n1
2
d1,1 P dj,1,1
d2
A
R i,n
HA
n
d
Pd
1,2 i,1,2
n1
d
Pd
1,2 j,1,2
R Hj,n
AH
n
m1
AA A
R j,m
m
(3.10)
177
S Ai,j 1 if match has signicance (as dened above) for away team j,
and does not have signicance for home team i; 0 otherwise.
The specication of the model is completed by specifying the link
between y*i,j and the result of each match, denoted yi,j , as follows:
Home win yi,j 1
Draw
if 2 y*i,j i, j
if
y*i,j i, j 1
(3.11)
where i,j is an error term which allows for the random element in the
result of the match between teams i and j, and which is assumed to follow
the standard normal distribution. Together, (3.10) and (3.11) constitute
an ordered probit model. The estimated version of (3.11) can be rearranged to obtain probabilities for each outcome, as follows:
Home win probability pi,jH p(i, j
m
2 y *i, j )1 ( m
2 y *i, j )
Draw probability p D
p( m
1 y *i, j i, j m
2 y *i, j )
i, j
(m
2 y *i, j ) ( m
1 y *i, j )
Away win probability p Ai,j p(i, j m
1 y *i, j ) ( m
1 y *i, j )(3.12)
where ( ) is the standard normal distribution function. In (3.10), the
variables P di,y,s (for team i, and their counterparts for team j ) are the main
team quality indicators. The higher the value of y *i, j , the higher the probability of a home win, so ady,s
0 and bdy,s 0 are expected. It is assumed
that team is underlying quality is captured by its win ratio over the 12
months prior to the match in question, represented by
1
P 0i,0,0
d1
d
, and
i,0,1
1
d1
2
P di,1,1
d2
d
.The
i,1,2
model
allows the individual components of these sums to make dierent contributions to the team quality measure. For example, if present season
results are a better indicator of current team quality than previous season
results in the same division within the same 12-month period, a00,0
a00,1
and b00,0 b00,1 would be expected. If previous-season results from a
higher division indicate higher quality than those from a lower division,
1
1
1
1 for y0,1 would be expected. An
a0y,1
y,1
and by,1
0y,1 y,1
ay,1
important issue is the number of previous years over which team quality
should be measured. Experimentation indicated that {P di,0,s, P dj,0,s} and
178
The model was also tested using a stricter denition, classifying a match as insignicant
only when it was mathematically impossible for the team in question to win the championship, promotion or relegation under any assumptions concerning the performance
179
denitions are SH
1 if the match is signicant for home team i but not
i,j
for away team j, and SH
0 elsewhere. Similarly, SAi,j 1 if the match is
i,j
signicant for away team j but not for home team i, and SAi,j 0 elsewhere.
Therefore 1
0 and 2 0 would be consistent with incentive eects of
the kind described above.
Estimation of the model
Table 3.14 reports the results of two sets of estimations of the coecients
of (3.10) and (3.11).The set reported in columns (1)(2) and (5)(6) use
all of the available data for the 1973 to 1999 seasons inclusive. Columns (1)
and (5) show the coecient estimates, and columns (2) and (6) show their
standard errors.There are 53,820 useable match result observations.13
Although using all of the data maximises the information set available
for the estimation, it relies on an assumption of parameter constancy
which, as has already been seen from the trends in the proportions of
home and away wins identied in section 3.2, is invalid. In columns
(3)(4) and (7)(8) of table 3.14, therefore, the estimation is repeated
using only the 19,746 useable match result observations for the 10
seasons 1989 to 1998 inclusive.Table 3.1 suggests that the proportions of
home and away wins did not vary systematically within this period, so the
assumption of parameter stability seems plausible. Slightly fewer
coecients are found to be signicantly dierent from zero in the tenseason model. This is a natural consequence of the fact that the estimation is based on fewer observations. The estimation for this particular
10-season period is reported here, because this model is used in chapter 8
to make ex ante predictions of match results for the 1999 season, in order
to investigate hypotheses about the eciency of prices in the xed-odds
betting market.
In general, the parameter estimates reported in table 3.14 are in accordance with prior expectations. For example, the estimated coecients on
P0i,0,0, P0i,0,1and P0i,1,1show that for the home team, the points ratio from
matches played during the present season is a stronger indicator of
current team quality than the points ratio from matches played in the
13
A
of other teams.The chosen algorithm, however, produced values of SH
i,j and Si,j which had
greater explanatory power in the ordered probit model. Although the algorithm is rather
crude, it does succeed in a simple manner in identifying those matches of signicance to
one or both teams during the last few weeks of the season, the results of which are most
likely to be aected by incentive eects.
The number of useable observations is slightly less than the 54,848 matches completed,
because full sets of lagged results are unavailable for all teams at the start of the observation period, and for teams entering or re-entering the league during their rst seasons of
membership.
P 0i,0,0
1
P i,0,1
0
P i,0,1
1
P i,0,1
1
P i,1,1
0
P i,1,1
1
P i,1,1
2
P i,1,2
1
P i,1,2
P 0i,1,2
1
P i,1,2
2
P i,1,2
P 0j,0,0
1
P j,0,1
P 0j,0,1
1
P j,0,1
1
P j,1,1
P 0j,1,1
1
P j,1,1
2
P j,1,2
1
P j,1,2
0
P j,1,2
1.555a
1.523a
1.128a
0.830a
0.592a
0.541a
0.460a
0.172
0.716a
0.403a
0.313a
0.111
1.292a
1.331a
0.912a
0.619a
0.845a
0.603a
0.466a
0.398
0.717a
0.469a
(1)
coe.
0.109
0.179
0.100
0.091
0.154
0.096
0.085
0.350
0.142
0.094
0.080
0.158
0.109
0.177
0.099
0.091
0.154
0.095
0.086
0.354
0.141
0.093
(2)
s.e.
197399
0.183
0.292
0.163
0.151
0.257
0.157
0.143
0.889
0.230
0.152
0.130
0.236
0.180
0.287
0.160
0.152
0.257
0.156
0.145
0.893
0.228
0.150
(4)
s.e.
198998
1.736a
1.535a
0.979a
0.829a
0.699a
0.610a
0.469a
0.783
0.626a
0.364b
0.411a
0.195
1.439a
1.151a
0.714a
0.467a
0.567b
0.491a
0.346b
0.867
0.732a
0.415a
(3)
coe.
Estimation period
RH
i,1
RH
i,2
RH
i,3
RH
i,4
RH
i,5
RH
i,6
RH
i,7
RH
i,8
RH
i,9
R Ai,1
R Ai,2
R Ai,3
R Ai,4
RH
j,1
RH
j,2
RH
j,3
RH
j,4
R Aj,1
R Aj,2
R Aj,3
R Aj,4
R Aj,5
0.066a
0.032b
0.075a
0.010
0.020
0.002
0.015
0.028b
0.042a
0.020
0.037a
0.047a
0.007
0.028b
0.031b
0.031b
0.037a
0.045a
0.045a
0.043a
0.046a
0.028b
(5)
coe.
0.013
0.013
0.013
0.013
0.013
0.013
0.013
0.013
0.013
0.013
0.013
0.013
0.013
0.013
0.013
0.013
0.013
0.013
0.013
0.013
0.013
0.013
(6)
s.e.
197399
198998
0.027
0.016
0.074a
0.009
0.010
0.001
0.013
0.014
0.019
0.017
0.043b
0.033
0.029
0.038c
0.025
0.027
0.038c
0.031
0.013
0.033
0.043b
0.006
(7)
coe.
Estimation period
Table 3.14. Ordered probit estimation of match results forecasting model, 19731999 and 19891998
0.021
0.021
0.021
0.021
0.021
0.021
0.021
0.021
0.021
0.021
0.021
0.021
0.021
0.021
0.021
0.021
0.021
0.021
0.021
0.021
0.021
0.021
(8)
s.e.
0.281a
0.413a
0.148a
0.114a
0.629
0.145
0.080
0.161
0.027
0.027
0.055
0.055
0.143
0.536b
0.121b
0.118b
0.574
0.189
0.130
0.242
0.048
0.048
0.092
0.092
obs
R Aj,6
R Aj,7
R Aj,8
R Aj,9
0.037a
0.031b
0.030b
0.021
53,820
0.013
0.013
0.013
0.013
0.027
0.022
0.021
0.055b
19,746
Notes:
= signicantly dierent from zero, two-tail test, 1% level; b = 5% level; c = 10% level.
Key:
P 0i,0,0= total points gained by home team: last 12 months, current season, current division; divided by matches played in last 12 months
1
P i,0,1
= as above: last 12 months, last season, one division above current division
0
P i,0,1= as above: last 12 months, last season, current division
1
P i,0,1
= as above: last 12 months, last season, one division below current division
1
P i,1,1
= as above: 1224 months ago, last season, one division above current division; divided by matches played 1224 months ago
P 0i,1,1= as above: 1224 months ago, last season, in current division
1
P i,1,1
= as above: 1224 months ago, last season, one division below current division
2
Pi,1,2
= as above: 1224 months ago, two seasons ago, two divisions above current division
1
Pi,1,2
= as above: 1224 months ago, two seasons ago, one division above current division
0
Pi,1,2
= as above: 1224 months ago, two seasons ago, current division
1
Pi,1,2
= as above: 1224 months ago, two seasons ago, one division below current division
2
Pi,1,2
= as above: 1224 months ago, two seasons ago, two divisions below current division
1
P 0j,0,0, P j,0,1
. . . , same as above for away team.
SH
=
1
if match has signicance for home team but not for away team
i, j
Si,Aj = 1 if match has signicance for away team but not for home team
RH
= result of most recent home match played by home team
i,1
RH
= result of second most recent home match played by home team, and so on
i,2
R Ai,1 = result of most recent away match played by home team
R Ai,2 = result of second most recent away match played by home team, and so on
RH
, R Aj,1 . . . , same as above for away team.
j,1
1
P j,1,2
2
P j,1,2
SH
i, j
Si,Aj
1
2
0.021
0.021
0.021
0.021
182
same division within the same 12-month period but during the previous
season. In turn, the latter is a stronger indicator than the points ratio from
matches played in the same division during the previous 12-month period
and during the previous season. The estimated coecients on P0j,0,0, P0j,0,1
and P0j,1,1 show the same for the away team. Only one of the four estimated
2
2
2
2
coecients on Pi,1,2
, Pi,1,2
, Pj,1,2
and Pj,1,2
is signicantly dierent from
zero, however. This is primarily owing to a shortage of data, since movement up or down by two divisions in consecutive seasons is a relatively
rare occurrence.
The estimated coecients on RH
, RAi,n, RH
and RAj,mare rather more
i,m
j,n
erratic, both in terms of numerical magnitude and in terms of the
signicance of individual coecients. Despite the nding of a negative
persistence eect in section 3.2, the estimated coecients on the home
teams recent results are predominantly positive (and those for the away
team predominantly negative).This is undoubtedly because recent results
contribute information about the underlying strengths of the two teams.
The estimated coecients on these variables therefore combine information about team heterogeneity with information about persistence eects;
and the former appears to dominate the latter. Finally, despite the crudity
of the formula used to identify matches with signicance for championship, promotion or relegation outcomes for either team, the coecients
on SH
and SAi,j are correctly signed and signicantly dierent from zero in
i,j
both estimations. This seems to conrm that the eorts contributed by
football players (like most employees in other industries) are strongly
inuenced by incentives.14
Table 3.15 compares the 19891998 models predictions for the 1999
season with the actual match results.The mechanics of the calculation of
predictions are described in the next sub-section. On the basis of historical data, the model slightly over-estimates the proportions of home wins
and draws, and under-estimates the proportion of away wins. 486 of the
877 home-win predictions (55.4 per cent) turned out to be correct; 173
out of the 545 draw predictions (31.7 per cent) were correct; and 178 out
of the 522 away-win predictions (34.1 per cent) were correct. The
number of matches that would be expected in each cell if the forecasting
model had no predictive content are shown in parentheses. If the 877
home win predictions had been selected randomly, therefore, the
expected number of correct predictions would be 409 (or 46.6 per cent).
Likewise, 545 randomly predicted draws would be expected to yield 157
14
183
Table 3.15. Match results forecasting models predicted results and actual
results
Predicted
Actual
Home win
Draw
Away win
Total
Home win
Draw
Away win
Total
486 (409)
238 (253)
153 (215)
877
227 (252)
173 (157)
145 (133)
545
194 (244)
150 (151)
178 (128)
522
1907
1561
1476
1944
Note:
Cells show the actual number of matches in each category, and (in italics) the expected
number in each category if the forecasting model had no predictive content.
correct predictions (28.8 per cent) and 522 randomly predicted away
wins should yield 128 correct predictions (24.5 per cent).The forecasting
model therefore outperforms random prediction signicantly in the case
of home wins and away wins, but only marginally in the case of draws,
which appear to be near-random events.15
An illustration: Middlesbrough v. Arsenal, Premier League, 24 April
1999
This sub-section illustrates the calculation of win-draw-loss probabilities
and a match result forecast, using the 1999 seasons Middlesbrough v.
Arsenal Premier League xture as an example.This xture was played on
24 April 1999.
The complete win-draw-loss records of the two teams over the 24
months prior to the xture are shown in table 3.16.The total numbers of
games played 012 months and 1224 months before 24 April 1999 were
ni,0 36 and ni,1 48 for Middlesbrough, and nj,0 38 and nj,1 35 for
Arsenal. Using this data, the calculations of Pdi,y,s and Pdj,y,s for home team
iMiddlesbrough and away team jArsenal are shown in column (1) of
table 3.17. Column (2) shows the relevant coecients from the estimated
model, obtained from table 3.14. Column (3) shows the product of each
variable and its coecient. The sum of these products is carried forward
to the nal calculation of y *i, j below.
15
A test of the hypothesis that the actual numbers of correct or incorrect predictions in
each cell correspond to the numbers expected under random prediction rejects the null
at all realistic signicance levels. The test statistic is distributed chi-square with four
degrees of freedom under the null, and the realised value is 52.4.
184
Matches played
Middlesbrough
1997
on or after 24/4/97
1998
before 23/4/98
on or after 24/4/98
1999
before 23/4/99
Arsenal
1997
1998
1999
on or after 24/4/97
before 23/4/98
on or after 24/4/98
before 23/4/99
Division
Played
Won
Drawn
Lost
Premier
1
1
Premier
14
44
12
34
11
26
11
12
13
19
11
14
0
9
0
8
Premier
Premier
Premier
Premier
12
33
15
33
11
20
13
18
10
19
10
12
1
4
2
3
Calculation
(1)
Coe.
(table 3.14)
(2)
Product
(3)
P 0i,0,0
1.736
0.917
1
P i,0,1
0.829
0.035
1
P i,1,1
0.469
0.298
P 0i,1,2
0.364
0.019
P 0j,0,0
1.439
0.909
P 0j,0,1
0.714
0.056
P 0j,1,1
0.491
0.344
P 0j,1,2
0.415
0.012
y *i, j
0.052
185
Variable
Value
(4)
Coe.
(table 3.14)
(5)
Variable
Value
(1)
Product
(6)
RH
i,1
0.5
0.027
0.014
RH
j,1
0.038
0.038
RH
i,2
RH
i,3
RH
i,4
RH
i,5
RH
i,6
RH
i,7
RH
i,8
RH
i,9
R Ai,1
R Ai,2
R Ai,3
R Ai,4
0.016
0.016
RH
j,2
0.025
0.025
0.074
0.074
RH
j,3
0.027
0.027
0.009
0.009
RH
j,4
0.038
0.038
0.5
0.031
0.016
0.013
0.013
0.017
R Aj,1
R Aj,2
R Aj,3
R Aj,4
R Aj,5
R Aj,6
R Aj,7
R Aj,8
R Aj,9
0.199
0.5
0.010
0.005
0.5
0.001
0.001
0.5
0.013
0.007
0.014
0.019
0.019
0.017
0.017
0.5
0.043
0.022
0.033
0.033
0.029
0.5
0.033
0.5
0.043
0.022
0.006
0.006
0.027
0.027
0.022
0.022
0.021
0.5
0.055
0
0.028
0.279
Key:
See table 3.14.
The results achieved by the two teams in the matches immediately preceding 24 April 1999, in chronological order, were as follows:
Middlesbrough
Last 9 home:
Last 4 away:
WLDDDWWWD
LWDW
Arsenal
Last 4 home:
Last 9 away:
WWWW
DLWWWDDWD
186
Team
Played
Won
Drawn
Lost
Points
11
12
17
18
19
20
Manchester United
Arsenal
Middlesbrough
Blackburn Rovers
Southampton
Nottingham Forest
32
33
34
33
34
34
19
18
12
17
18
14
10
12
14
11
17
19
13
13
18
15
19
21
67
66
50
32
31
21
187
during the previous 24 months, their excellent recent form and the fact
that the match was signicant for them but not for their opponents, the
model adjusts these average probabilities to those obtained above: 0.330,
0.296 and 0.374, respectively. Arsenal are therefore signicantly more
strongly fancied in this xture than an average away team in an average
xture.
To obtain a specic match result prediction, the tted values y *i, j for the
19,746 matches played between the 1989 and 1998 seasons and included
in the estimation reported in table 3.14 are ranked in descending order.
The two cut-o values of y *i, j separating the 46.6 per cent of matches with
the highest values of y *i, j (the most likely home wins), the 27.8 per cent of
matches with the next highest values (the most likely draws), and the
remaining 25.6 per cent of matches with the lowest values (the most likely
away wins) are 0.119 and 0.050.Therefore a suitable forecasting rule for
matches played during the 1999 season is:
forecast home win if y *i, j
0.119
forecast draw if 0.050 y *i, j 0.119
forecast away win if y *i, j 0.050
On these criteria, the forecast for the Middlesbrough v. Arsenal xture
with y *i, j 0.250 is an away win. Readers who have worked right the way
through this sub-section may recall or may be interested to learn that the
match in question actually resulted in a 61 victory for the away team!
3.4 Conclusion
Competitive balance and uncertainty of match result outcomes have been
the main themes linking the theoretical and empirical sections of chapter
3. The theoretical literature on competitive balance places considerable
faith in the capacity of market forces to maintain a reasonable degree of
competitive balance within professional team sports leagues. It tends to
be suspicious of regulatory intervention intended to promote competitive
balance, in the form of reserve clauses, salary caps, or revenue sharing
arrangements. At best, the eect on competitive balance of such mechanisms may be neutral; at worst they could inadvertently make competition more unbalanced. North American assumptions of closed player
labour markets and prot maximising behaviour do not necessarily transfer directly across the Atlantic. Nevertheless, the theoretical literature
does suggest strongly that caution needs to be exercised before advocating simplistic remedies for perceived trends towards excessive championship dominance by a handful of big-market teams, both in England and
elsewhere in Europe.
188
Since the early 1960s there has been a series of major institutional
reforms to the organisation of the players labour market in English football, starting with the abolition of the maximum wage in 1961, and culminating in the 1995 European Court of Justice (ECJ) ruling in the
Jean-Marc Bosman case (see chapter 2). Some of the broader consequences of these changes are obvious and widely recognised. Spiralling
wages, especially for superstar players, are a consequence of the progressive shift towards freedom-of-contract that has been underway throughout this period. Although naturally the chronology and detail of
institutional reform varies between countries, the same long-term trend
has been evident worldwide. The trend towards the globalisalisation of
the players labour market has accelerated signicantly in recent years,
especially among the major professional football playing nations located
in the EU, as a consequence of the Bosman ruling.
This chapter begins in section 4.1 by presenting a descriptive prole of
certain key personal and career characteristics of professional football
players in the English league. The analysis focuses on those players who
turned out most regularly, week-in-week-out, for league matches played
in each of two seasons located 20 years apart: 1979 and 1999.Tabulations
of the data are used to draw comparisons between the composition of the
two samples. Among the players characteristics investigated are age,
place of birth and various aspects of each individuals career playing
record. This analysis is followed in section 4.2 by a rather more general
investigation of patterns of international migration on the part of professional footballers, which draws mainly on the results of previous sociological research.
Section 4.3 examines explanations for the very high wages earned by
the leading stars in modern-day professional football. Above-ination
increases in players wages, especially at the highest level, have been a permanent feature of English football since the abolition of the maximum
wage in the early 1960s. In section 4.3 it is argued that scarcity in the
supply of the highest talent is only part of the explanation. Before the
189
190
191
192
1979
1999
Age
1979
1999
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
0.2
0.5
1.6
4.3
6.0
7.1
9.8
8.2
7.3
9.3
8.4
6.9
7.1
4.9
4.4
0.0
0.2
0.9
2.5
3.9
4.3
6.2
6.6
7.5
8.6
7.5
9.0
7.9
6.1
7.6
31
32
33
34
35
36
37
38
39
40
41
15.1
14.0
12.2
11.4
10.7
10.3
10.3
10.1
10.0
10.0
10.0
16.3
14.4
13.2
13.3
12.0
10.9
10.5
10.2
10.0
10.3
10.1
Mean
St. dev.
25.3
14.2
26.8
14.4
about 18 months, from 25.3 years in 1979 to 26.8 years in 1999.The age
distribution appears to shift to the right in a fairly uniform manner across
the full range of ages, so the shape of the distribution remains similar.
2028 is the most frequent age range in the 1979 sample, and 2231 is the
most frequent range in the 1999 sample. There were fewer players 19 or
under in 1999 (3.6 per cent) than in 1979 (6.6 per cent), and there were
more players aged 32 or above in 1999 (14.9 per cent) than in 1979 (9.0
per cent).
A possible explanation for the shift in the age distribution is the recent
inux of players born overseas, who were present in much greater
numbers in 1999 than in 1979. If overseas players are older on average
than their local-born counterparts (perhaps because they tend to move
abroad at a relatively late stage in their career) this might account for the
increase in the average age.Table 4.2 shows the average age of players on
31 May, disaggregated by birthplace and division. On the basis of the evidence in table 4.2, the hypothesis that overseas players were responsible
for the increase in average age can be dismissed categorically; in 1999 the
average age of the overseas players was almost identical to their local-born
counterparts.1 Table 4.2 also shows that there was relatively little variation
1
There is rather more variation in the average age of the overseas-born players in the 1979
sample; this, however, is mainly a reection of the small numbers involved.
193
Table 4.2. Average age of players, by birthplace and division, 31 May 1979
or 1999
Birthplace
Britain and Ireland
Rest of world
All players
Average
No.
Average
No.
Average
No.
1979
D1
D2
D3
D4
All
25.8
25.1
25.4
25.1
25.4
1254
1258
1283
1286
1081
26.9
23.8
21.6
27.0
25.0
110
116
115
112
123
25.9
25.1
25.3
25.1
25.3
1264
1264
1288
1288
1104
1999
Prem
Div One
Div Two
Div Three
All
27.5
27.2
26.5
26.2
26.8
1160
1243
1267
1277
1947
27.2
27.2
26.4
27.1
27.1
180
145
121
111
157
27.4
27.2
26.5
26.2
26.8
1240
1288
1288
1288
1104
in the average age of players across the leagues four divisions: in both
seasons, the average age of Division 1/Premier League players was around
six months higher than their counterparts in the other three divisions.
Further analysis of players dates of birth reveals a feature of the data that
is quite striking, and also consistent between the two samples. Table 4.3
shows the percentage distribution of all players by month of birth. This
reveals that a signicant majority of professional players (59.9 per cent in
1979 and 58.1 per cent in 1999) were born during months that fall in the
rst half of the school year, from September to February inclusive. The
most likely explanation is that children whose birthdays fall in these months
are physically bigger and stronger, and are therefore more likely to succeed
in school sports, than children in the same school year with birthdays
between March and August, who nevertheless play in the same school
sports teams as their older counterparts. Although the physical disadvantage to the latter is most pronounced at the youngest ages, it seems certain
that an aversion to sport acquired early on in life would eectively terminate
an individuals prospects of ever becoming a professional sports person.
Birthplaces and rst clubs
Table 4.4 shows the percentage distribution of players by birthplace,
classied by standard region (old denitions) for British-born players and
194
1979
1999
Jan
Feb
Mar
Apr
May
June
July
Aug
Sept
Oct
Nov
Dec
18.9
10.1
18.2
15.9
17.2
16.6
16.0
16.3
10.2
11.3
10.1
19.3
18.0
17.3
18.5
16.7
16.8
16.3
16.1
17.5
10.7
12.0
10.4
19.7
1979
1999
18.7
13.5
11.5
17.9
17.5
10.9
17.6
10.2
15.8
10.6
11.9
11.7
20.7
14.9
12.2
16.2
17.7
10.0
14.9
17.8
13.4
14.2
11.2
12.4
97.8
85.6
France
Germany/Austria/Switzerland
Benelux
Scandinavia
Spain/Portugal/Malta
Italy
Greece/Turkey
Eastern Europe/former USSR
Asia
Africa
Australia/New Zealand
North or South America
West Indies
All outside Britain and Ireland
1979
1999
0.2
0.2
0.3
0.1
0.2
0.4
0.3
0.5
2.2
11.3
10.9
11.5
13.3
10.3
10.5
10.3
10.6
10.4
11.6
11.5
11.7
10.5
14.4
195
1999 season
Region of birth
Footballers
(per cent)
Population
(1961, per cent)
Footballers
(per cent)
Population
(1981, per cent)
London/South East
South West
East Anglia
East Midlands
West Midlands
Yorks and Humberside
North West
North
Wales
22.4
14.2
11.8
19.4
19.0
13.0
21.1
12.2
16.9
34.8
18.0
13.2
17.2
10.3
10.1
13.9
16.7
15.7
26.6
16.3
12.8
18.0
19.9
12.9
19.2
10.0
14.4
34.3
18.8
13.8
17.8
10.5
19.9
13.0
16.3
15.7
by country or broader geographical area for overseas players.The proportion of players born outside Britain and Ireland increased from 2.2 per
cent in 1979 to 14.4 per cent in 1999. France, Italy and the Scandinavian
countries are among those that contributed signicant numbers of
players in 1999, but were not represented at all in 1979.
Although the overall proportion of British and Irish-born players fell
from 97.8 per cent in 1979 to 85.6 per cent in 1979, the proportions of
players from some English regions increased: specically, London/South
East, the South West, East Anglia and the West Midlands. But relative to
population, all of these regions were under-represented in both seasons.
This is shown in table 4.5, which compares the 1979 and 1999 percentage
distributions of English- and Welsh-born players by region of birth with
regional populations from the 1961 and 1981 Censuses, respectively.The
North West, the North and Wales all produced signicantly fewer players
in 1999 than in 1979. Relative to population, table 4.5 shows that (with
the exception of Wales in 1999) these regions were over-represented in
both seasons. In table 4.4, the proportion of players born in Scotland
declined sharply between 1979 and 1999, while the proportion born in
the Irish Republic increased. The marked decline in the number of
Scottish players is dicult to explain, but as table 4.4 shows, it is consistent with the similar experience of regions in the north of England.
The proportions of clubs located in each region of England and Wales
changed very little over the period, because there were only minor adjustments to the leagues 92-club membership. This suggests two possible
196
1979
1999
London/South East
South West
East Anglia
East Midlands
West Midlands
Yorks and Humberside
North West
North
Wales
Other
23.3
15.5
12.6
19.6
12.7
12.9
19.8
16.0
13.3
14.4
24.9
15.3
14.9
17.4
17.9
12.0
17.7
16.0
12.5
11.6
197
Table 4.7. Most frequent rst clubs: top 20, 1979 or 1999
1979
6
8
10
12
16
20
Arsenal
Manchester City
Chelsea
Leicester City
Manchester United
Burnley
Crystal Palace
Leeds United
Wolverhampton Wdrs
Middlesbrough
Nottingham Forest
Tottenham Hotspur
Hudderseld Town
Blackpool
Derby County
Birmingham City
Coventry
Liverpool
West Bromwich Albion
Newcastle United
Sunderland
1999
No. of
players
27
26
24
24
24
22
22
20
20
19
19
18
18
18
18
17
17
17
17
16
16
Position
1
2
3
4
7
8
10
12
18
First club
No. of
players
Manchester United
Crystal Palace
Norwich City
Luton Town
Manchester City
Southampton
Leeds United
Millwall
Tottenham Hotspur
Arsenal
Hudderseld Town
Aston Villa
Blackpool
Crewe Alexandra
Ipswich Town
Leicester City
Nottingham Forest
Charlton Athletic
Liverpool
Middlesbrough
Sheeld United
Watford
29
24
21
20
20
20
19
18
18
17
17
16
16
16
16
16
16
15
15
15
15
15
Among the players whose rst club was in each region, columns (1)
and (2) of table 4.8 show the proportions who were born in the same
region. In both samples, clubs located in the North had the highest propensity to recruit new players locally, and clubs in East Anglia had the
lowest. For all regions except the West Midlands, the proportion of new
players recruited locally was smaller in the 1999 sample than in the 1979
sample. This suggests that in general, recruitment has become less concentrated geographically over time. Even for the biggest clubs, however,
local recruitment is still high, as table 4.9 illustrates. In the 1999 sample,
12 of the 29 players whose careers began at Manchester United were born
in the North West, and 10 of the 17 players whose rst club was Arsenal
were born in London/South East. Among the smaller clubs included in
table 4.9, the proportions are even higher: 12 of the 16 players who began
198
Table 4.8. Percentages of players whose rst club and whose current club is in
their region of birth, 1979 or 1999
Per cent of players whose
rst club is in stated region,
who were born in the same
region
Region
(1)
1979
(2)
1999
(3)
1979
(4)
1999
London/South East
South West
East Anglia
East Midlands
West Midlands
Yorks and Humberside
North West
North
Wales
56.4
45.9
41.4
53.8
42.9
64.8
68.0
81.8
75.0
43.1
44.8
27.8
43.9
58.6
58.3
61.5
74.2
46.4
40.2
21.4
8.3
28.7
28.3
39.6
47.5
38.9
43.8
20.8
19.0
20.8
13.0
22.2
27.1
36.6
34.7
16.7
at Blackpool were from the North West, and 21 of the 24 players who
began at Crystal Palace were from London/South East.
Among the players whose current club (in 1979 or 1999) was in each
region, columns (3) and (4) of table 4.8 show the proportions who were
born in the same region. Again, in most cases these proportions declined
signicantly between 1979 and 1999. Clubs from the North West and the
North had signicantly higher concentrations of locally born players than
clubs from all other regions in 1999.
Professional career histories
Table 4.10 shows the percentage distribution of players by duration of
spell at present club, and table 4.11 shows the percentage distribution by
number of clubs (past and present) in England and Wales. Loan spells are
not included in table 4.11. Both tables indicate that there has been a
marked increase in player mobility between 1979 and 1999. In 1979, for
example, 22.3 per cent of all players had been with their present club for
more than 5 years; in 1999, the corresponding gure was 12.8 per cent. In
1979, 39.2 per cent of all players were still with their rst professional
club in England and Wales; in 1999, the corresponding gure was 30.2
per cent. In both seasons, there was more mobility in the leagues lower
divisions than in the higher divisions, but the dierence does appear to
9
10
1979
1999
1979
1999
1979
1999
1979
1999
1979
1999
1979
1999
Arsenal
Blackpool
Crystal Palace
Leeds
Manchester Utd
Middlesbrough
17
21
1
2
L/SE
Club
SW
EA
1
2
1
1
EM
WM
2
1
9
5
3
1
YH
8
12
12
12
NW
13
10
2
1
1
5
1
1
4
1
Table 4.9. Distribution of players, by birthplace, selected rst clubs, 1979 or 1999
2
3
3
2
Wal
2
1
2
2
Scot
2
3
2
1
NI
5
2
2
1
Ire
RoW
19
15
24
29
20
19
22
24
18
16
27
17
Total
4.0
21.6
21.6
15.5
10.6
6.8
7.2
2.7
3.8
2.3
1.5
6.4
D2
5.2
12.9
18.6
14.8
6.1
11.7
7.2
6.1
4.9
3.0
4.2
10.5
01
12
23
34
45
56
67
78
89
910
More than 10
Mean
D1
Duration at
present club
(years)
3.7
22.2
20.8
17.4
11.1
6.9
5.9
5.6
2.8
3.8
1.4
2.1
D3
1979
2.6
31.9
30.2
19.1
5.6
4.2
5.2
1.0
0.7
1.4
0.7
0.0
D4
3.8
22.4
22.9
16.8
8.3
7.3
6.3
3.8
3.0
2.6
1.9
4.7
All
3.8
24.6
22.1
15.8
10.8
4.6
3.3
5.0
3.8
2.9
1.7
5.4
Prem
3.3
23.6
29.9
14.6
9.4
6.3
4.2
4.2
2.4
2.4
1.7
1.3
Div One
2.9
25.7
26.7
18.4
13.5
5.6
3.8
2.8
0.7
1.0
0.4
1.4
Div Two
1999
Table 4.10. Percentage distribution of players, by duration of present spell at present club, 1979 or 1999
2.4
36.8
30.6
16.7
7.3
4.5
1.7
1.0
0.4
0.4
0.4
0.2
Div Three
3.1
27.8
27.5
16.4
10.2
5.3
3.3
3.2
1.7
1.6
1.0
2.0
All
201
6 or more
Mean
1979
D1
D2
D3
D4
All
All British/Irish born
50.4
45.1
35.4
27.4
39.2
38.6
29.9
30.3
33.3
29.2
30.7
31.0
14.0
16.7
19.1
23.3
18.4
18.5
4.2
4.5
9.7
10.8
7.4
7.6
1.5
1.5
2.1
6.3
2.9
3.0
0.0
1.9
0.3
3.1
1.4
1.4
1.8
1.9
2.1
2.5
2.1
2.1
1999
Prem
Div One
Div Two
Div Three
All
All British/Irish born
45.4
33.0
22.6
22.2
30.2
24.2
22.5
24.3
34.4
27.1
27.3
28.3
17.5
19.8
21.9
20.8
20.1
22.2
7.1
12.8
9.0
12.2
10.4
11.7
5.8
4.2
7.3
7.3
6.2
7.0
1.7
5.9
4.9
10.4
5.9
6.7
2.1
2.5
2.6
2.9
2.6
2.7
202
D1
D2
D3
D4
All
D1
D2
D3
D4
All
67
68
50
26
211
39
31
49
47
166
14
26
47
50
137
9
13
23
65
110
129
138
169
188
624
1999
Previous division
Current
division
Prem/
D1
Div One/
D2
Div Two/
D3
Div Three/
D4
All
Prem
Div One
Div Two
Div Three
All
71
58
36
14
179
41
61
64
38
204
9
41
69
81
200
2
24
32
69
127
123
184
201
202
710
Table 4.12 tabulates their current divisional status (in May 1979 or 1999)
against the divisional status of their previous club at the time of departure. The results suggest that over time, large jumps between divisions
have become rarer. In the 1979 season, there were nine individuals
playing regularly in Division 1 who had transferred to their present club
from Division 4; in 1999, the corresponding gure was only two.
Similarly, in 1979, 26 individuals were playing regularly in Division 4
having transferred from Division 1; in 1999, the corresponding gure was
14. It seems reasonable to infer that reductions in the numbers of large
jumps between divisions probably reect a widening in the gap in playing
standards between divisions during the intervening period.
The main ndings of section 4.1 can be summarised as follows:
The average age of regular professional players in England and Wales
has increased by about 18 months over the past 20 years. The age
203
204
205
206
Cup nals places received by each of six continental regions are determined by quota, with parallel regional qualifying competitions used to
determine which nations obtain places. Europe received 15 of the 32
places (47 per cent) in the 1998 tournament, while 436 of the 704 squad
players (62 per cent) were employed by European clubs. Among the 704
squad players, there were 122 non-Europeans playing for European
clubs. In stark contrast, there were no non-South Americans playing in
South America, and no non-Africans playing in Africa. Of the 110 South
American squad players, 37 per cent played outside South America and
of these 83 per cent played in Europe. Italy and Spain were the most
popular destinations for South American players. Of the 110 African
squad players, 67 per cent played outside Africa and of these 96 per cent
played in Europe. France and Spain were the most popular destinations
for African players.
Within Europe, Italy, Spain and England were the biggest importers of
World Cup players, ahead of Belgium, Germany and France, whose clubs
employed the highest proportions of foreign nationals in Maguire and
Steads (1998) earlier survey (based on 1995 data on all rst-division
players). Maguire and Pearton attribute the discrepancy partly to the
higher preponderance in the World Cup survey of elite players, who naturally tend to gravitate towards the leagues with the highest spending
power. Any changes in patterns of intra-European migration between
1995 and 1998 are also likely to reect the impact of the Bosman ruling
(see chapter 2). By 1998 France appears to have replaced Holland as the
most popular recruiting ground within the EU for the leading Italian
clubs.The tendency for Spain to recruit principally from Eastern Europe
and South America rather than from the EU remained evident, however,
as did the imbalance between the magnitudes of northsouth and
southnorth intra-European migration ows described above.
Section 4.2 concludes by reporting an analysis of the home and destination countries of the members of the 16 national squads which took
part in the nals of the 2000 European National Championships (Euro
2000) staged in Belgium and Holland.The analysis, along the same lines
as that of the 1998 World Cup reported by Maguire and Stead (2000),
provides a more recent snapshot of patterns of intra-European player
migration at the highest level. As in the World Cup, 22 players are permitted in each squad, or 352 players in total. A notable feature of Euro 2000
was that just over half of those selected, 177 players or 50.3 per cent of the
total, were foreign-based players.
The full data are shown in table 4.13. The columns of table 4.13 show
the 16 participating nations, arranged from left to right in ascending
order of the numbers of foreign-based players included in their squads,
22
Ita
22
Spa
England
Spain
Italy
Germany
Holland
Belgium
France
Portugal
Turkey
Denmark
Norway
Yugoslavia
Czech
Romania
Sweden
Slovenia
Scotland
Austria
Greece
USA
Croatia
Switzerland
Japan
All
Destination country
21
11
Eng
11
19
11
11
Tur
11
11
18
12
Ger
11
11
11
11
11
14
14
Por
12
11
14
13
10
11
11
12
Bel
14
11
18
11
11
13
11
12
15
Cze
15
17
13
11
16
11
11
12
11
Rom
Home country
15
11
11
11
17
11
11
19
11
Nor
15
17
16
12
15
12
Fra
16
11
11
15
11
16
11
13
11
13
Slov
Table 4.13. Analysis of home countries and countries of employment, Euro 2000 squad players
17
12
15
12
11
14
13
13
12
Swe
17
12
15
15
17
13
Hol
11
18
14
11
11
17
15
11
12
Yug
19
11
11
13
11
14
14
13
15
Den
139
133
129
122
110
117
115
114
113
113
111
111
110
110
110
110
116
116
113
112
111
111
111
177
All
208
shown in the nal row. At one extreme, Italy and Spain selected squads
consisting solely of home-based players. At the other extreme, Denmark
and Yugoslavia selected 19 and 18 foreign-based players, respectively.The
rows of table 4.13 show the countries of employment of the squad
members of each country.The rst 16 rows show players employed in the
participating nations; the remaining seven rows show players employed in
non-participating nations. Each group of nations is arranged in descending order of the total numbers of imported players employed, shown in
the nal column. The numbers of home-based players in each squad,
shown in italics in the relevant cells of table 4.13, are not included in the
nal row and column totals.
Table 4.13 identies England as by some distance the leading importer
of players at this level, followed by Spain, Italy and Germany. Englands
position, together with the fact that there was only one foreign-based
England squad member, undoubtedly reects the present-day attractiveness of the English league to leading players as a result of its commercial
vibrancy, despite the fact that doubts remain (fuelled partly by the
England teams own performances in Euro 2000!) as to whether playing
standards in England are truly comparable with the very best that can be
found elsewhere in Europe. The long-standing tendency for the leading
Spanish and Italian players to remain home-based has continued in 2000,
though the absence of foreign-based players from the Italian squad
appears to reect a conscious policy decision rather than a complete lack
of suitably skilled foreign-based talent.
Englands apparent ascendancy within Europe as an importer of
foreign-based players is clearly a recent phenomenon, but several longerestablished features of intra-European migration patterns are also apparent in table 4.13.These include the popularity of England as a destination
for Scandinavians (especially Norwegians); the ability of Scandinavians
to nd employment across a wide range of northern and southern
European countries; the popularity of Spain (especially) and Italy as a
destination for Eastern Europeans (represented here principally by
Yugoslavia and Romania); and the importance of geographical proximity
or historical links in determining migration ows, especially for players
from nations with less economically powerful domestic leagues. In Euro
2000 these included the Czech Republic with ve squad members based
in Germany, and Slovenia with ve Austrian-based players.
Finally it is interesting to note that the number of squad members
based in either Italy or Spain provides a rather accurate predictor of eventual success in Euro 2000. Apart from these two countries themselves, the
ve nations with the next highest totals on this criterion (Yugoslavia,
Holland, Portugal, France and Romania) all progressed beyond the rst
209
210
211
212
employs only just over 3,000 professionals in its core activity of playing
football matches. Rosens account of these phenomena rests on two key
aspects of consumer tastes and production technology:
Imperfect substitution between sellers of certain services. The very best
talent is in scarce supply. For consumers lesser talent is not perfectly
substitutable for greater talent, even if there is compensation in terms of
quantity. For example, Manchester United might be three times more
talented than Birmingham City, but watching Birmingham City on
three occasions is not equivalent to watching Manchester United once.
Scale economies in joint consumption of certain services. The technology
of production enables very large audiences to be serviced at the same
time. The eort and therefore the production costs incurred by
Manchester Uniteds players are much the same if they are playing in a
half-empty stadium, or in front of a capacity crowd, or additionally
before a large television audience. The technology, in other words, has
characteristics similar to that employed to supply certain public goods
like street lighting and national defence: production costs depend only
slightly, or not at all, on the number of people consuming the service.
But in contrast to these examples, in football there are no free rider
problems preventing the seller from appropriating a charge for providing the service. Would-be audience members who are unwilling to pay
(by purchasing a match ticket or a pay-television subscription) can
easily be excluded from consumption.
The implications of these attributes of tastes and technology for the distribution of earnings and employment opportunities depend on the
precise nature of the relationship between production costs and audience
size. Rosen suggests two reasons why marginal costs of production may
eventually start to increase as the audience size increases. First, internal
diseconomies include all the usual reasons why marginal costs rise as
output increases. In professional football, marginal costs start to rise
when players become tired or injured as the number of matches played
increases, or perhaps as a result of increased stress or psychological pressure in matches played before a large audience. Second, external diseconomies are factors that reduce the quality of the product as the audience size
increases. A stadium with a capacity of 50,000, for example, might
provide a better viewing experience than one twice that size, owing to
excessive distance from the pitch in the latter case. Beyond the stadium
capacity, the television audience consumes a product that is presumably
inferior to that obtained by live spectators, though one whose characteristics do not vary further as the television audience size increases.
Rosen demonstrates rst that imperfect substitution between sellers,
combined with internal diseconomies, induces a skewed distribution of
213
earnings, even if there are no joint consumption economies.The more talented sellers charge each consumer a higher price and capture a larger
share of the market than their less talented counterparts. But in the
absence of joint consumption economies, talent dierentials do not lead
to extremely high seller concentration: all sellers are able to capture some
market share. Under certain simplifying assumptions, both the price paid
by each consumer and each sellers audience share can be shown to be
linear functions of the sellers talent. This implies that sellers total earnings, the product of price and audience, are quadratic in talent.The distribution of earnings is therefore more skewed than the distribution of
talent.
To account for high seller concentration as well as a skewed earnings
distribution, joint consumption economies are also required in the
model. Suppose initially that there are no internal or external diseconomies at all, so joint consumption economies are not limited by the size of
the audience. In other words, the audience size can be expanded
indenitely, without imposing additional costs upon the seller, and
without causing any deterioration in the quality of the product obtained
by the consumer. In this extreme case, the market has the characteristics
of a natural monopoly. The most talented seller is able to drive all other
sellers out of business, and in the long run services the entire market.The
equilibrium price, however, is constrained by the threat of entry.The rent
element in the equilibrium price, which is proportional to the talent
dierential between the most and second most talented sellers, is modest.
But crucially, the total rent obtained by the most talented seller is very
large indeed, because the latter captures the entire audience. As Rosen
and Sanderson (2000) observe, superstars achieve high earnings by operating at high volume (by servicing very large audiences) but at low
margins (by charging each audience member only a modest mark-up relative to less talented suppliers).
The zero diseconomies case described above is somewhat unrealistic,
but useful because it illustrates the main insight of the Rosen model.The
most talented seller captures the entire market because the production
technology permits limitless expansion of the audience at zero cost to the
seller.This is in stark contrast to the experience of doctors, nurses, teachers and police ocers, who nd that the number of patients they can
treat, schoolchildren they can teach and criminals they can catch on any
one day is strictly nite. The key to the high earnings of superstars, in
other words, lies in the vast extent of the audience they can reach at any
one time, aorded by the existence of scale economies in joint consumption (Rosen and Sanderson, 2000).
In practice, of course, one football team does not dominate the entire
214
market.The Rosen model suggests that this is not solely due to considerations of jointness in production or competitive balance (see chapters 1
and 3). It is also to be expected if joint scale economies in consumption
are eventually limited by internal or external diseconomies of the kind
described above. In this case several sellers obtain market shares in longrun equilibrium.The price each seller charges is again constrained by the
(actual not potential) presence in the market of other sellers with less
talent, and the rent element in the price is dependent on the talent
dierential. If the rate at which internal or external diseconomies increase
varies inversely with talent, audience share and price both increase with
talent, but audience share increases at a faster rate than does price. As
before the high earnings of the most talented sellers derive principally
from the vast size of their audiences, rather than from the modest price
dierential paid by each audience member.
An extension of Rosens superstars model is discussed by MacDonald
(1988), who considers the career progression of performers who eventually become superstars in their elds, despite the fact that in the early stages
of their careers they may have imperfect information about their future
performance and future earnings. Aspiring stars enter the profession at a
young age and perform cheaply.The audience feedback they receive provides useful but incomplete information about their probable future performance.Those who receive negative feedback are persuaded to withdraw
from the profession.Those who receive positive feedback are encouraged
to continue, and as the favourable notices accumulate their audience
rapidly expands. So, too, do their earnings, and as in the Rosen model at a
faster rate than the talent gradient would suggest. Although MacDonald
seems to have in mind primarily groups such as actors or musicians, the
notion of a professional sports apprenticeship as an information-gathering
process, enabling individuals to take rational decisions at various stages as
to whether to continue in pursuit of a career as a professional, seems an
appropriate description of the path followed by many aspiring youngsters
in professional sports.
In summary, the contrast between the congurations of high earningslow utility achieved by sports stars, and low earningshigh utility
achieved by groups such as nurses and teachers is explained mainly by the
technologies under which the former operate, which enable them to
provide services, albeit of modest value, to very large audiences (Rosen,
1981; Rosen and Sanderson, 2000).The latter groups provide services of
much higher value, but to a very nite number of users.The same technological conditions explain the high levels of seller concentration found in
professional sports. Internal and external diseconomies of scale, however,
impose limits on the audience share of the most talented sellers, enabling
215
the less talented to capture some share of the market. The road from
apprenticeship to stardom can be viewed as a type of sequential information-gathering process, in which cumulative feedback on performance
enables the individual to take rational decisions on whether to continue or
withdraw from the profession (MacDonald, 1988).
Rosens essentially technological explanation for the high earnings of
superstars is clearly relevant in explaining wage ination in professional
football during the 1990s. Before the introduction of pay-television,
English football reached large television audiences via free-to-air services
provided by the BBC and ITV. But constrained by this type of public
goods production technology, football was unable to appropriate the full
economic price for providing the service to individual audience members.
Since the late 1980s, however, technological innovation in broadcasting
has transformed this situation. Television has unquestionably made the
single most important contribution to recent growth in football revenues.
Meanwhile, with further moves in the direction of free agency taking
place in the footballers labour market at the same time, the leading
players bargaining position in wage negotiations has strengthened appreciably. This has enabled them to capture a large proportion of the rent
element in the additional revenues that are being generated, resulting in
rampant wage ination.
Rank-order tournaments and intra-team earnings distributions
As noted previously, Rosens insights on the economics of superstars do
not address the complexities of intra-team earnings distributions. Yet at
rst sight, these can seem as puzzling as the disparities between the earnings of sports stars in general and those of other groups. In the English
Premier League, for example, the highest paid players can earn many
times as much as the average rst-team player, even though it seems
doubtful whether this dierential could be explained by a corresponding
dierential in MRP.
In a competitive labour market, the intra-team wage structure is equivalent to a piece-rate system, since each player receives a wage equivalent
to his MRP. Dierences in wages between players in the same club should
reect the fact that those contributing most to club revenues get paid the
most. Casual observation, however, suggests that some players receive
large wage increases even when there is no noticeable improvement in
performance. Often, players who might be judged only a small fraction
better than others appear to earn signicantly more. In these cases, the
intra-team wage structure seems to be hierarchical. The models outlined
above are unable to explain why top stars can earn n times more than
216
other players, if the talent gradient is shallower than n, and if their contribution to revenue is not n times as high.
In a highly inuential paper on the determination of executive salaries
in corporate business, Lazear and Rosen (1981) argue by analogy that top
executives high salaries are equivalent to the outcome of a rank-order
tournament. In situations where it is costly to monitor or measure exactly
the productivity of individual workers, it may be ecient for rms to rank
workers according to their productivity, and then reward them according
to the rankings that they achieve. In the corporate sector, the salaries of
vice-presidents are typically much lower than those of presidents. Yet
presidents are often promoted from the ranks of vice-presidents. On the
particular day an individual is promoted, his salary might triple, but it is
dicult to argue that his productivity has also suddenly increased by an
equivalent amount on the same day.
According to the Lazear and Rosen model, the individual who achieves
promotion to president is like the winner of a contest for which there is a
very large rst prize. The presidents salary does not reect his or her
current productivity, but it does induce all junior employees (from the
rank of vice-president downwards) to perform appropriately, in the hope
that one day they too will win the prize and achieve promotion to president. In other words, the president earns a high salary not because he is
highly productive as president, but because this type of salary structure
provides incentives for all employees to be more productive over their
entire working lives.
Scully (1995) adapts this type of analysis in order to examine player
wage determination in a number of North American sports. It is worth
noting that in contrast to the situation in the corporate sector, the productivity of each player or worker is relatively easy to measure (baseball
being perhaps the most obvious case in point). Both the employer and the
customer observe the joint output of the team, and have ample opportunity to form their own subjective assessments of the individual contributions of each team member. With reference to English football, this
relationship is aptly described as follows. The rst team player faces over
forty public examinations a year before an attendance of thousands of
spectators. He will be talked about, shouted at, criticised or praised, given
extensive coverage in the press and radio . . . he is in the limelight and
must be seen to behave as bets his profession (PEP, 1966, p. 134).
Even so, professional team sports players do not seem to be paid on a
piece-rate basis. Scully (1995) argues that the rank-order tournament
model may provide a more convincing explanation of compensation
scales than the traditional supply-and-demand model. Clubs that are
motivated either by playing success or by prot need players to develop
217
their skills to their maximum potential, and to exert the maximum eort
on behalf of the team. A compensation structure in which wages depend
upon each players ranking relative to others within the same club provides suitable incentives for all players to compete (eectively against one
another) in pursuit of these aims. Even in sports such as football, where
there is high input complementarity and individual outputs are less easily
identied, the rank-order tournament model may still be applicable.
In English football, there are no restrictions on the number of players a
club can register. A typical league club may have a rst team squad of
2530 full-time professional players, while some Premier League clubs
have squads twice as large as this. Each clubs professionals compete
against one another for a place in the rst team. At the start of each game
the manager selects one goalkeeper, 10 outeld players and ve substitutes, three of whom can be used to replace any of the 11 players on the
eld of play at any time during the game, either for tactical reasons or
because of injury. The numbers of defenders, mideld players and forwards among the outeld players depends upon the tactical choices of the
manager. Typical formations include 4-3-3 (four defenders, three
midelders and three forwards), 4-4-2, 5-3-2 and 3-5-2. For each playing
position, the squad will include several players who can provide cover.
Especially useful in this respect are utility players, who are able to play in
more than one outeld position. Performance is related to athletic
endowment such as speed, strength, agility, balance and co-ordination;
intangible attributes of talent such as determination and concentration;
and player and team investments in enhancing skill through coaching,
tactical awareness and practice at passing, shooting, heading and tackling.
Theoretical and empirical models of rank-order tournaments have
been applied extensively to individual sports such as tennis and golf.
Competitors face a xed prize structure and achieve the next-higher
reward by moving up by one place in the ranking of competitors. Scully
(1995) argues, however, that competition among team sports players for
a rst-team place is not entirely dissimilar. Those in the rst team are
always threatened explicitly by the performance of lower-ranked reserve
players, and implicitly by the possibility that the club might sign a
replacement player for the same position from outside. If a rst-team
player experiences a slump in form, a reserve is substituted into the team
or the club acquires a new player. Injury or suspension for disciplinary
transgressions can also lead to the replacement of one player by another.
A replacement that performs well can often capture a rst-team place,
until he in turn eventually drops out owing to loss of form, injury or
suspension.
The rest of this sub-section contains an exposition of the rank-order
218
(4.1)
where Qi ti i for i1, 2; E(2 1)0; and F( ) is the cumulative distribution function of 2 1.
The intra-team wage structure pays SF to the rst-choice player and SR
to the reserve player, with SF
SR.The expected wages of the two players,
S1 and S2, are determined by the probability that player 1 beats player 2
for the rst team place ( p) and the wages for the two positions (SF, SR):
S1 pSF (1p)SR;
S2 (1p)SF pSR
(4.2)
C2
t2
(for i1, 2)
(4.3)
219
MC2
MC1
t 2** t 2*
t 1**
t 1*
t 1,t 2
(4.4)
220
Second, assuming that the club has a nancial constraint which prohibits
it from making a loss, there is an upper bound to the total wage bill, equal
to the clubs total revenue.
Scully simplies the derivation of the equilibrium intra-team wage
structure by introducing a parameter c (0c 1), which reects the
degree of competitiveness in the players labour market. c is the total wage
bill expressed as a proportion of total revenue, as follows:
SF SR cTR
(4.5)
At the maximum value of c1, the players wages capture all of the clubs
revenue.This would apply with free agency in the players labour market,
so the players bargaining power is maximised.2 The distance by which c
falls below one is a measure of the degree of monopsony power held by
the club as a buyer of players.
Equations (4.4) and (4.5) can be used to derive the following expressions for the wage levels, SF and SR:
SF cTR[ p/(2p 1)] [(C2(t2)
C )/(2p 1)]
SR [(C2(t2)
C )/(2p 1)] cTR[(1 p)/(2p 1)]
(4.6)
Implicitly, for c1 there is also an assumption that the club seeks to maximise revenue or
playing success, subject to a zero prot constraint.
221
The model also predicts that moves towards free agency should lead to
a widening of the gap in compensation between top players and the rest,
and should lead to an increase in investment in playing skills all round.
There is no doubt that the former has occurred in English football since
movement towards free agency rst started in the early 1960s. The latter
is more dicult to verify conclusively, but there is overwhelming anecdotal evidence: video footage suggests, and most commentators agree, that
modern-day players are much tter, faster and stronger than their
counterparts from earlier eras.3 More tentatively, it is possible to speculate that increased investment in human capital by players could account
for the increase over the past 20 years in the average age of the players
appearing most regularly in the league (see section 4.1). If as seems likely,
the full benets of such investment take time to materialise, players performance may tend to peak later, and careers may tend to last longer,
under a high-investment than under a low-investment regime.This would
be consistent with the shift in the age distribution identied in table 4.1.
4.4 Determinants of players compensation: empirical
evidence
Empirical evidence on the links between competitive conditions in the
players labour market, wage structures and performance in football is
virtually non-existent at the microeconomic level, because in most countries individual players wage details are not disclosed. For a number of
North American professional team sports, however, there is sucient
individual disclosure to permit empirical investigation of these relationships at the micro level. An inuential early contribution to this literature
was made by Scully (1974), who developed a framework for making comparisons between estimated MRPs for dierent categories of player in
Major League Baseball (MLB), and the average wages of players in the
same categories. Scullys study dates from the pre-free-agency period
when the reserve clause was still eective, so systematic divergence
between wages and MRPs is to be expected.
Scully obtains estimated marginal productivities for baseball hitters
and pitchers using the slope coecients in a regression of team win ratios
on team slugging averages (for hitters) and team strikeout-to-walk ratios
(for pitchers). Dummy variables are also included in the regression to
3
It is not at all clear, however, that modern-day players are more skilful on average than
players of earlier periods. Natural skill is, of course, mainly an endowed rather than an
acquired attribute, and is therefore not sensitive to changes in the structure of players
compensation.
222
control for dierences between the National League (NL) and the
American League (AL), and to dierentiate teams according to the
extent to which they were in contention to be divisional winners in
the season concerned. Gross MRPs for hitters and for pitchers are
imputed by multiplying the marginal productivity estimates by the slope
coecient obtained from a second regression of team revenues on win
ratios.Various additional controls used in the second regression include a
NLAL dummy, home-town population and a variable reecting underlying dierences between teams in intensity of support. Net MRPs are
calculated by deducting allowances for player training and other factor
costs from the gross MRPs.
Scully then estimates wage equations for hitters and pitchers, using
individual data. Among the explanatory variables are personal career
slugging averages and strikeout-to-walk ratios, career durations and variables (such as population and intensity of support) that are also used in
the revenue regression to control for demand factors contributing to
monopoly rent.The latter are mainly insignicant in the wages equations,
suggesting that players failed to capture any share of monopoly rents
under the reserve clause.The comparisons between MRPs and wages are
drawn by substituting sets of explanatory variables describing representative hitters and pitchers of mediocre, average and star quality into the
various estimated equations. The MRPtalent gradient is found to be
much steeper than the wage-talent gradient. Consequently, while mediocre players earned more than their net MRPs, the wages of average players
and star players were around only 20 per cent and 15 per cent of their net
MRPs, respectively. This constitutes evidence that before free agency
there was substantial monopsony exploitation in the players labour
market.
Using an empirical framework similar to that employed by Scully 20
years earlier, MacDonald and Reynolds (1994) use data on player performance and earnings in MLB in the 1987 season to show that after the
introduction of free agency, players wages matched their estimated
career MRP more closely than in the pre-free-agency period. But there
were still some anomalies: young players were paid less than their MRP,
perhaps because of institutional restrictions on their mobility within the
labour market, and senior pitchers (but not senior hitters) were over-paid.
MacDonald and Reynolds also present evidence that dierences between
the wages of top and middle-rank players are consistent with Rosens
model of superstars.
Scully (1995) uses data on the individual performance and wages of
baseball and basketball players before and after the introduction of free
agency in both sports to examine the impact of this change. There is of
223
course no doubt that the wagetalent gradient has become much steeper
during the free-agency period, and for both sports the sensitivity of wages
to performance in the current season and to measures of past experience
(included in the regressions to control for talent, investment and eort) is
higher than it was before free agency. For baseball, an average individual
performance index exceeds an average individual wage index for up to
547 career games; thereafter, the wage index exceeds the performance
index. This nding seems consistent with the time-prole of wages predicted by the rank-order tournament model. For the pre-free-agency
period, in contrast, the wage index is higher than the performance index
both for inexperienced players (less than 405 games) and for players
nearing the end of their careers (1,550 games).
On the other hand, some empirical research has suggested that there
are limits to the applicability of the rank-order tournament model to team
sports. An intra-team wage distribution that is too unequal may be divisive, perhaps damaging team cohesion and spirit to an extent that outweighs the incentive benet described above. It is certainly common
enough for English football clubs to explain failure to reach agreement
with a player they are seeking to sign in terms of a reluctance or refusal to
overturn existing internal wage structures. Richards and Guell (1998)
argue on similar lines that a high intra-team variance in wages may have a
negative eect on performance, since it implies a dispersed level of talent
and an inability to function eectively as a team. One or two star players
do not necessarily carry a team. In their empirical results for MLB over
the period 19925, the mean wage has a positive eect on the win ratio
and the propensity to win championships, but an increase in the variance
of wages reduces the win ratio (with a coecient that is signicant at the
10 per cent level). A greater dispersion of wages does not seem to aect
the probability of winning the championship, however.
The investigation of discrimination in professional team sports has
motivated many empirical studies of the determinants of player compensation, especially in the USA.This issue has been of particular interest in
North American sports because before the Second World War there was
complete racial segregation at professional level in a number of sports: the
teams in the top leagues employed only white players.The colour bar was
lifted for the rst time in both baseball and basketball during the late
1940s, but there is no doubt that discriminatory practices still continued
for many more years afterwards.
Speaking generally, the forms that discrimination can take include
inequality in the level of compensation; inequality in hiring standards;
inequality in the allocation of playing positions; and inequality in the
availability of opportunities for endorsements or sponsorship. Possible
224
225
discrimination in baseball is relatively weak.There is, however, some evidence of discrimination in terms of the length of career durations and in
the allocation of team positions. In basketball, in contrast, there is consistent evidence of both wage and customer discrimination. Apart from
some evidence of positional discrimination, less evidence on other forms
of discrimination is available for American football, mainly because of
diculties in dening suitable individual performance indicators.
It is also apparent that evidence of discrimination in North American
team sports is becoming weaker over time (Kahn, 2000; Rosen and
Sanderson, 2000). Given that most of the evidence of discrimination
during the 1980s was stronger for basketball than for other sports, it is
unsurprising that basketball was the main focus for several discrimination
studies during the 1990s. Using NBA data for the 1995 season, Hamilton
(1997) nds evidence of a 1 per cent across-the-board wage dierential
between white and black players. At the lower end of the wage range,
whites earned less than blacks; though when the number of matches is
included in the regressions as a control, the eect reverses. At the upper
end of the wage range, whites earned more than equally qualied blacks.
The premium was the same regardless of the race of the teams manager,
and so is attributed by Hamilton to customer rather than employer discrimination.
Bodvarsson and Brastow (1999) use wage regressions to test for
employer discrimination in the NBA in the 1986 and 1991 seasons. Again
the race of the teams manager is assumed to convey information about
team owners attitudes: owners that employ black managers are assumed
to be non-discriminators. The test therefore examines whether black
players working for white managers experienced discrimination. The
wage regressions include the percentages of team members and of the
local population that were black, as controls for co-worker and customer
discrimination, respectively. Wage discrimination, present in 1986, is
found to have disappeared by 1991. This nding is consistent with the
hypothesis that two important structural changes in the interim increased
competition between teams as employers, and therefore reduced the discretion for team owners to exercise discrimination as monopsony purchasers in the labour market. These were the 1988 NBA Collective
Bargaining Agreement which introduced free agency, and the addition of
four new teams which stimulated competition between teams as employers of players.
Szymanski (2000b) has presented the rst test for employer discrimination against black players in English football. Although football has
never operated a formal colour bar, instances of black players appearing
in the Football League before the 1970s were isolated.There is abundant
226
227
the evidence that player recruitment is, in practice, geographically segmented (see section 4.1). Szymanski suggests that the fact that the market
for corporate control of football teams has been relatively weak and
inecient may explain why the more ecient, non-discriminating teams
did not drive the less ecient, discriminating ones out of business
through economic competition between clubs as employers.
4.5 The football transfer market
In contrast to major team sports in North America, the trading of players
for cash is common in professional football. In any football season the
large majority of player transfers go almost unnoticed by the public, yet
there are a few deals that make headlines, on the front as well as the back
pages of newspapers, simply because of the enormous sums of money
involved.These big-money transfers help to raise the public prole of the
football transfer market across Europe; but recently they have also
attracted the critical scrutiny of the European Commission who, at the
time of writing, are proposing to enforce major changes to the transfer
system. Nevertheless, the production and circulation of rumours of
impending big-money moves has clearly developed into a highly lucrative
cottage industry in its own right. Such rumours enable players (or their
agents) to send signals of availability to prospective buyers and, as a sideeect, help to fuel wage ination. During the summer months in particular, the same rumours generate welcome publicity for clubs attempting to
sell season tickets for the coming season. And of course they also help to
ll the sports sections of newspapers starved of match action to report.
Not so many years ago a world record transfer fee would be set every
few years, and typically the players involved would be great names like
Johan Cruy, Diego Maradona, Ruud Gullit and Roberto Baggio.Today
it is not unusual for the world record fee to be broken several times in the
same year. As of summer 2000, the world record transfer fee stands at
37 million, paid by Real Madrid to Barcelona for Luis Figo (see section
2.1).Yet having also acquired exclusive rights to sell merchandise bearing
Figos name and image, Real believe that they will easily be able to recoup
the fee paid. To put recent fees into perspective, in 1990 Baggio moved
from Fiorentina to Juventus for a fee of just 8 million. And by todays
standards, the fee of 5 million paid in 1984 by Napoli to Barcelona for
the signature of Maradona looks very small indeed.
Although transfer ination has become a matter of signicant public
debate and concern, until recently relatively little academic analysis of the
football transfer market has been undertaken by economists. Section 4.5
begins by reviewing a small number of empirical studies of the English
228
transfer market. In all of these studies, the attributes of the player and the
characteristics of the buying and selling clubs determine the transfer fee.
Section 4.5 then presents an empirical analysis of the determinants of
transfer fees in English football.
In the rst empirical study of the transfer market in English football,
Carmichael and Thomas (1993) assume that the transfer fee is determined by characteristics of the player concerned and by the relative bargaining power of the buying and selling clubs. Using a similar
methodology, Reilly and Witt (1995) investigate the degree to which
there is racial discrimination in the transfer market. No evidence of discrimination is found. Speight and Thomas (1997) analyse the determination of transfer fees settled through the arbitration process for
disputed fees for out-of-contract players, under the system that operated
from 1978 until the late 1990s (see section 2.7). Speight and Thomas
nd that fees are determined mainly by player and club characteristics,
rather than by compromise between the nal oers of the buying and
selling clubs.
An innovative feature of the paper by Carmichael, Forrest and
Simmons (1999) is that the propensity of players to be transferred, and
the transfer fees of players that are transferred, is modelled separately in a
two-stage estimation process. The rst stage involves the estimation of a
discrete choice model of the propensity for a player to be transferred.The
second stage involves the estimation of a model of the transfer fee, which
incorporates a sample selection bias correction.The latter is derived from
the rst stage, and allows for the fact that the characteristics of players
who are transferred may dier systematically from the characteristics of
players in general. Dobson and Gerrard (2000) point out a diculty that
arises because the propensity to be transferred has to be modelled excluding information about actual or potential buying clubs.The buying club is
observed for players who are transferred, but the potential buyers of nontransferred players are not observed. Excluding information on the characteristics of the actual and potential buyers may create a misspecication
problem at the rst stage, which is carried over into the transfer fee equation via the sample selection bias correction.
In contrast to some of the earlier studies, the empirical analysis of
Dobson and Gerrard (1999, 2000) is based explicitly on a theoretical
model of the transfer market. The starting point is Sloanes (1971)
assumption that the football club is a utility maximiser subject to a
nancial solvency constraint (see also chapter 1). The clubs utility function is assumed to have two arguments: team performance and club
prots. The relative weightings of these arguments in the utility function
depend on, among other things, the ownership and control structure of
229
the club. Each clubs valuation of a player depends upon his expected
impact on team performance and club prots, and on the relative importance attached to the two arguments. The decision to be involved in the
transfer of a player (either as seller or buyer) is interpreted as an attempt
on the part of the club to move towards its optimal position on the performanceprot frontier. Clubs that are net sellers of players need to raise
revenue and reduce costs in order to increase prots or (in many cases)
reduce losses. In contrast, clubs that are net buyers of players are motivated by the need to improve playing success, having already satised
their prot requirements.
The determination of transfer fees: a theoretical model
The model of the player transfer market described in this sub-section is
based on that of Dobson and Gerrard (1999). Consider the trade of
player k from selling club m to buying club n.The purchase of the player is
expected to improve team performance and the buying clubs revenue. In
essence, the buying clubs valuation function represents a money measure
of the increase in utility from the expected change in team performance
(keQn ) and club revenue (keRn ). Formally, the buying-club valuation
function (V Bn,k ) over the period of the contract can be written:
V Bn,k V Bn,k(keQn, keRn )
(4.7)
(4.8)
In equation (4.8) Wn,k is the wage costs of the player during the period of
the contract, Sn,k is the signing-on fee paid to the player when the contract
is agreed and eTn,k is the expected proceeds (if any) from selling the player
before his contract legally expires. Each component of the maximum bid
price is a present value at the time of the transfer. Assuming the club to be
rational, the maximum bid price sets the upper limit for the observed
transfer fee.
Selling-club ms valuation of player k depends on the expected
dierence between the teams current level of performance (keQm ) and
230
revenue (keRm ) and the post-transfer levels over the remaining period of
the players current contract.The selling clubs valuation function (V Sm,k )
is as follows:
V Sm,k V Sm,k(keQm, keRm)
(4.9)
If the selling club is rational, its minimum acceptable transfer fee is the
current net present value of the player to the club. The minimum price
that selling-club m is prepared to accept for player k or player ks reservation price (T Sm,k) is as follows:
T Sm,k V Sm,k Wm,k eTm,k
(4.10)
In (4.10) Wm,k is the present value of the wage costs for the remainder of
the players contract with club m, and eTm,k is the present value of the
expected transfer fee if player k were to be sold at a later stage in his
contract.
The observed transfer fee for player k, (TFk) should lie somewhere
between the upper and lower limits, T Bn,k and T Sm,k. If the observed fee lies
above the lower limit, there is a monopoly rent: the selling club is able to
extract some of the excess of the buying clubs net valuation over the reservation price. The monopoly rent is (TFk T Sm,k)/(T Bn,k T Sm,k ). For
simplicity, it is assumed that is determined exogenously, and remains
constant for all transfers. The observed transfer fee for player k can be
written:
TFk T Bn,k (1 )T Sm,k ,
T Bn,k T Sm,k ,
0 1
(4.11)
(4.12)
231
(4.13)
232
The number of registered players with the 92 league clubs has remained relatively stable
since the 1970s at around 2,000. The number declined temporarily to around 1,600
during the period of nancial retrenchment in the sport in the early 1980s.
The entertainment and other leisure goods retail price index is used as the best available
proxy for the football industrys own price index. Constraints such as the partial availability of gate price data and the non-disclosure of individual player wage data prevent
the construction of a more specic football price index.
261
190
363
267
331
427
343
410
2,592
19731975
19761978
19791981
19821984
19851987
19881990
19911993
19941996
Totals
503
476
471
367
495
655
492
673
4,132
In season
Close season
Seasons
Permanent
764
666
834
634
826
1,082
835
1,083
6,724
Total
37
29
25
23
54
50
71
124
413
Close season
401
351
160
479
622
672
840
1,375
4,900
In season
Temporary
438
380
185
502
676
722
911
1,499
5,313
Total
1,202
1,046
1,019
1,136
1,502
1,804
1,746
2,582
12,037
Total
64
64
82
56
55
60
48
42
56
Permanent as per
cent of total
234
league appearances, career league scoring record, international recognition, current form (measured by league appearances), goals scored in the
previous season and positional dummy variables. Buying- and sellingclub characteristics include league position, goal dierence and average
gate attendance in league matches. The denitions of the independent
variables and the expected signs of their coecients are detailed in table
4.15. A number of statistically signicant quadratic terms are included in
the estimations to allow for non-linear eects.
Table 4.16 shows summary statistics for selected variables. The mean
transfer fee in nominal terms is 418,334 with fees ranging from a
minimum of 5,000 to a maximum of 15 million. The sample includes
players aged from 17 years 2 months to 36 years 2 months; players with
no league experience to players with nearly 600 career league appearances; and players with no full or Under-21 international appearances to
86 full international appearances and 18 Under-21 appearances. The
sample also includes buying and selling clubs covering the full spectrum
of performance and attendance levels.
The estimation results are presented in table 4.17.The estimated model
explains almost 77 per cent of the variation in real transfer fees, suggesting
that there is a considerable degree of systematic variation in transfer fees.
The estimated model also provides evidence of the existence of a monopoly
rent element in transfer fees.The buying-clubs characteristics are jointly
signicant as determinants of transfer fees. Observed transfer fees therefore appear to exceed the reservation price of the selling club.The selling
club is able to extract a share of the non-negative dierential between its
reservation price and the buying-clubs maximum bid price. These
ndings lend support to the claim of Carmichael andThomas (1993) that
observed transfer fees are the outcome of a bargaining process.7
Dobson and Gerrard (1999) report that the average nominal transfer
fee increased at an annual rate of around 25 per cent between the 1991
and 1996 seasons. This gure may tend to over-state the underlying rate
of ination in transfer fees, however, for two reasons. First, owing to skewness in the distribution of transfer fees, the average fee may be misleading
as a representation of the sample as a whole. Average logarithmic growth
in transfer fees can be used as a measure of transfer ination that avoids
this problem. This transformation yields an average annual rate of transfer ination of 19.4 per cent.
The second diculty in measuring the underlying rate of transfer
ination is that average ination is aected by any change in the mix of
players being transferred. The transfer fees model can be used to control
7
In contrast, Szymanski and Smith (1997) assume that the transfer fee is established at
the reservation price of the selling club.
235
Variable
Denition
LRLFEE
AGE
PRCLUB
LGTOT
GLRATE
CAPS
U21CAP
LEAG
GOALS
DEF1
FOR1
SELPOS
SELAST
SELGD
SELGAT
DSELGD
FSELGD
BUYPOS
BUYLST
BUYGD
BUYGAT
DBUYGD
FBUYGD
Expected
sign
Quadratic term
(expected sign
in parentheses)
n/a
n/a
?
AGESQ ()
n/a
LTOTSQ ()
n/a
CAPSSQ ()
U21SQ ()
n/a
n/a
DEF1SQ (?)
FOR1SQ (?)
n/a
SLSTSQ (?)
n/a
n/a
n/a
n/a
n/a
BUYLSTSQ (?)
n/a
n/a
n/a
n/a
236
Mean
St. dev.
Minimum
Maximum
FEE
RLFEE
AGE
LGTOT
LEAG
GOALS
GLRATE
CAPS
U21CAP
SELGD
SELGAT
BUYGD
BUYGAT
418,334
306,435
25.97
172.93
24.82
3.74
0.134
1.41
0.75
1.20
12,546.92
1.75
12,088.21
819,845
560,699
3.57
126.67
14.88
5.45
0.129
6.53
2.24
18.90
9,079.93
18.83
8,361.08
5,000
3,212
17.17
0
0
0
0.000
0
0
57
1,450
53
1,532
15,000,000
9,461,756
36.16
589
46
42
1.392
86
18
56
44,984
56
44,984
237
6.1887a
(1.0975)
LEAG
0.1736a
(0.0189)
DSELGD
0.0776a
(0.0261)
AGE
4.8381a
(0.8518)
GOALS
0.2109a
(0.0585)
FSELGD
0.0864a
(0.0235)
4.0062a
(0.9206)
BUYPOS
0.1637a
(0.0220)
10.6374a
(3.2181)
BUYLST
0.0997a
(0.0387)
AGESQ
1.0591a
(0.1630)
DEF1
PRCLUB
0.06449a
(0.0165)
DEF1SQ
LGTOT
0.0301a
(0.0067)
LTOTSQ
0.000416a
(0.0001)
FOR1SQ
GLRATE
1.0365a
(0.3939)
CAPS
0.1677
(0.0726)b
FOR1
0.6854c
(0.3557)
BUYLSTSQ
0.0035
(0.0035)
1.2445a
(0.4848)
BUYGAT
0.3615a
(0.0421)
SELPOS
0.0656a
(0.0216)
BUYGD
0.0523a
(0.0189)
SELAST
0.1375a
(0.0399)
DBUYGD
0.0420
(0.0260)
CAPSSQ
0.0242c
(0.0127)
SLSTSQ
0.0061c
(0.0035)
FBUYGD
0.0788a
(0.0243)
U21CAP
0.625a
(0.2373)
SELGD
0.0816a
(0.0177)
obs.
1350
s.e. of
regression
0.6934
0.4922b
(0.2166)
SELGAT
0.0659
(0.0404)
R2= 0.77
F(Bn,k)
170.97a
U21SQ
Notes:
a
signicantly dierent from zero, two-tail test, 1% level;
b
5% level;
c
10% level.
Standard errors of estimated coecients are shown in parentheses.
Dependent variable is LRLFEE.
F(Bn,k) is an F-statistic for the joint signicance of the buying club characteristics.
238
239
dierentials provide all players with the incentive to invest in the development of their skills, in the hope of securing a rst-team place and a higher
wage. Individual players wages therefore tend to dier systematically
from their marginal revenue products.
Empirical research has quantied the relationships between players
characteristics, performance and compensation, especially in the case of
North American team sports. In many studies there is signicant evidence of discrimination against blacks during the post-war period,
though the situation has improved markedly during the last decade.
During the 1970s and 1980s English football does not appear to have
been completely innocent of similar tendencies. In the nal empirical
section of chapter 4, the English transfer market has been modelled by
assuming that transfer fees depend on attributes of the player concerned,
as well as characteristics of both the buying and selling clubs. The model
has been used to test for the existence of a monopoly rent element in
transfer fees.The evidence is consistent with transfer fees settled through
a bargaining process, in which the selling club captures some of the nonnegative dierential between its own reservation price and the buying
clubs maximum bid price.
241
242
243
formal training, only limited training opportunities exist for the football
manager.The Football Association organises coaching courses, but most
managers still rely primarily on their playing experience as preparation
for a career in football management.
The role of the modern-day football manager as the person responsible
for team aairs rst started to evolve during the inter-war period.
Previously, in the nineteenth and the early part of the twentieth centuries,
responsibility for team aairs rested primarily with the clubs directors
and chairman. Throughout this period, it was quite common for a
manager to see his players on match days only. During the rest of the
week, the manager would undertake scouting missions, watch other
teams play and carry out administrative duties. Walvin (1994) identies
the growth of a stronger ethos of professionalism within football during
the 1920s and 1930s as instrumental in encouraging directors to begin
shifting responsibility for performance on the pitch towards the professional manager.
Two individuals from the inter-war period, Herbert Chapman
(Hudderseld Town 19215, Arsenal 192534) and Major Frank
Buckley (Wolverhampton Wanderers 192744) typify the new breed of
professional, hands-on team manager. Before Chapmans arrival,
Hudderseld and Arsenal were both under-achieving clubs. Having
enjoyed only a modest career as a player himself, Chapman steered
Hudderseld to the rst two of three consecutive championships in 1924
and 1925, and then won further titles with Arsenal in 1931 and 1933.The
latter was also the rst of three consecutive championships; tragically,
however, Chapman died during the course of the 1934 season, aged 55.
Chapmans success at Arsenal (in particular) was based on a regime of
physical tness, strength and skill previously unseen in the English game.
Chapmans tactical acumen was also impressive: Arsenal are widely credited as the rst team to adapt tactically to the introduction of the presentday oside law in 1925:1
Arsenal came in with the idea of a stopper centre-half. Previously centre-halfs
were allowed to wander, and it now became a pendulum. The stopper centrehalves stayed there and then the two full-backs were like pendulums on a clock. If
play was on one wing, one full-back would take it and the other one came back to
cover behind the centre-half.You had three forwards up front, two in the V point
of the W with a lot of alternatives. One of those men had two men in front of him
and one at the side. Hed got three alternatives with the ball. (Jack Curtis, quoted
in Taylor and Ward, 1995, pp. 278)
1
A player is oside if, when the ball is played forward to him, fewer than two opposing
players stand between him and the goal line. Before 1925 a player was oside if there
were fewer than three opposing players.
244
245
246
X2/Y
247
A
C
X1/Y
248
(5.1)
where yi,t is the output of the ith team in the tth season (for i1 . . . N;
t1 . . . T ); xi,t is a vector of inputs used by the ith team (including an
intercept term); is a vector of unknown parameters to be estimated; and
ui,t is a one-sided (non-negative) error term that reects managerial
ineciency. yi,t is therefore bounded from above.
In the stochastic production frontier model, the error term consists of
two parts: a one-sided component (to capture technical ineciency, as
before) and a two-sided component (to capture random inuences on
performance).The stochastic frontier takes the following form:
yi,t xi,t (vi,t ui,t)
(5.2)
In (5.2) the variables are dened as before, with the addition of a symmetrical component (vi,t) in the error term that captures the eects of all
random inuences on the performance of the ith team.
Eectively, the inclusion of the random component (vi,t) in the error
term produces a dierent ecient frontier for each team, incorporating
the eects of luck or other random inuences on the teams performance.
Managerial eciency is then measured by comparing actual performance
with the ecient performance implied by the team-specic frontier.
Intuitively, this approach is appealing since deviations from the average
frontier might be partly the result of events beyond the managers control.
In gure 5.2, QQ represents a deterministic frontier. Good fortune results
in the stochastic frontier for team A being placed below the deterministic
frontier (QQ, for which the random error is positive). Bad fortune
results in the stochastic frontier for team C being placed above the deterministic frontier (QQ, for which the random error is negative). Although
A is closer to the deterministic frontier than C in gure 5.2, C turns out to
be more ecient than A when eciency is measured relative to each
teams stochastic frontier, because OD/OC
OB/OA.
In order to make this approach operational, certain assumptions have
to be introduced. With cross-sectional data, and sometimes with panel
data, it is necessary to make a specic distributional assumption about
the ineciency term ui,t. The theoretical literature suggests half-normal,
exponential, gamma or truncated normal distributions. A weakness of the
production frontier approach is that dierent assumptions about the distribution of ui,t can sometimes produce widely diering results. Recently,
the increasing availability of panel data has increased the scope of frontier
249
Q ''
Q
X2/Y
Q'
A
C
B
Q ''
Q
Q'
X1/Y
Figure 5.2 Stochastic production frontier
estimation procedures. Using panel data there is no need to specify a particular distribution for the one-sided team-specic error term. With
several observations on each team, it is possible to obtain estimates of this
term directly, using either xed-eects (FE) or random-eects (RE)
specications. Alternatively, if assumptions about the distribution of the
ineciency term are retained, maximum likelihood estimation can be
used to locate the ecient frontier and obtain the managerial eciency
scores.
Estimation of team production functions
Scullys (1974) study of baseball, already reviewed in some detail in
chapter 4, was the rst to estimate a team production function with team
250
performance as an output measure. Using an average production function rather than a production frontier, Scully modelled team output (percentage of matches won) as a function of playing and non-playing
(management, capital and team spirit) inputs. This method of modelling
the sporting production process soon became the norm in production
function studies. In Zechs (1981) study of baseball, output is a function
of the following playing skills: hitting, running, defence and pitching.
Similar approaches have been used to measure output in basketball
(Scott, Long and Somppi, 1985) and American football (Atkinson,
Stanley and Tschirhart, 1988). In Schoelds (1988) study of English
county cricket and Carmichael and Thomass (1995) study of English
rugby league, a single-equation model is estimated, treating the player
inputs as exogenous in the same way as the US studies. The production
function is determined through a system of recursive equations, thus
endogenising playing performance.
None of the above studies uses the production frontier approach, so all
of them assume implicitly that team production takes place at maximum
eciency. Very little consideration is given to the contribution of the
manager to team output. In some cases, one or two managerial variables
are included on the right-hand side of the estimated equation. Zech
(1981), for example, includes the number of years spent managing
in major league baseball and the managers lifetime win-loss ratio, but
neither variable is found to be a statistically signicant determinant
of current performance. Carmichael and Thomas (1995) include
the number of years coaching experience to capture the managerial
contribution.
Surprisingly, production frontier analysis has been used in relatively
few professional team sports studies. Zak, Huang and Siegfried (1979)
estimate a deterministic frontier using match data for the 19767
National Basketball Association (NBA) season. The eciency of each
team is found to be close to 100 per cent (the lowest eciency rating is
99.7 per cent). This is somewhat surprising since in a deterministic
model, all of the deviation from the optimum is the result of ineciency.
The highly competitive nature of the sport and the extensive data coverage might account for the ndings. Porter and Scully (1982) also use a
deterministic specication to estimate a baseball production frontier.
There are two measures of team quality: team slugging percentage (a
measure of hitting skill) and team strikeout-to-walk ratio (a measure of
pitching quality). Managerial eciency and job tenure are found to be
positively correlated. The contribution made by the manager to team
output is found to compare favourably with that of the superstar players.
In subsequent research, Scully (1994, 1995) compares eciency in
251
252
253
254
(5.3)
The managers eciency within each spell is assumed not to vary over
time (i.e. ui,t ui). Each spell is treated separately, however, so a manager
can obtain several eciency scores from separate spells.4
In order to exploit both the cross-sectional and time-series dimensions
of his panel data set, Dawson estimates eciency scores using both xedeects (FE) and random-eects (RE) specications. For purposes of
comparison, Dawson also estimates another set of eciency scores using
maximum likelihood.The FE model is estimated using the within estimation procedure (Hallam and Machado, 1996); the RE model is estimated
as a generalised least squares (GLS) regression. The team production
function, which converts playing and managerial inputs into output
measured by win ratio data, is assumed to be the same for all teams,
because each team plays to a given set of rules and uses the same production technology. It is also assumed to satisfy the standard properties of any
neoclassical production function.
Associated with each player is a certain level of playing skill, reecting
the players ability to pass, shoot and head the ball, speed, agility, strength,
awareness, concentration, determination, and so on.These attributes may
be intrinsic or acquired through investment in training and development
(by either the player or the club). Although it is dicult to evaluate the
quality of a player independently of his team-mates and manager, for simplicity Dawson assumes that team quality is the linear sum of the skills of
the individual players. In other words, the inputs are treated as separable.
This approach is standard in the US literature. It may cause the managers
performance to be over- or under-stated, however, depending on whether
team synergy eects are positive or negative.
In contrast to North American sports like baseball, in football little
data is recorded systematically on dierent aspects of the contributions
made by individual players to team performance.5 This clearly poses a
4
If the performance of the manager is believed to vary over time, perhaps because the
value of the managers human capital increases as he gains experience, it may seem
rather restrictive to measure eciency assuming time invariance. For this reason
Dawson also estimates managerial eciency using a number of time-varying models
based on (5.2), but each with its own temporal structure. The interested reader is
referred to Dawson, Dobson and Gerrard (2000a, 2000b).
The Carling Opta Index of playing performance in the Premier League has been compiled in recent seasons (Carling Opta, 1999, 2000).This is constructed from data measuring a variety of contributions made by players during each match: passes, tackles, shots
on goal, and so on. Carmichael, Thomas and Ward (2000) have recently presented an
analysis of the relationship between team performance and a variety of player contribution measures of this kind. It seems likely that the Carling Opta Index will provide a valuable source of data for future studies of the type described in this chapter.
255
As a way of independently checking the validity of the transfer value approach, Dawson
also measures team quality using a raw characteristics approach. This denes team
quality as a vector of the means of the player variables (described above) for the team. In
this approach the coecients are estimated as part of the team production function. In
contrast, the transfer value approach denes the mean of player quality as the weighted
sum of the player variables for each individual player, with the weights being derived
from a parsimonious version of the model of transfer fees in Dobson and Gerrard (1999,
2000) based on 1,350 transfers. In the results reported below team quality is measured
using the transfer value approach only.
256
257
Table 5.1. Fixed eects, random eects and maximum likelihood estimates of
production frontier, Premier League, 19931998
Estimation method
FE
Constant
ln(qi,t)
t
obs.
R2
RE
1.005
(0.211)
1.362a
(0.108)
0.048a
(0.015)
147
0.623
a
ML
1.125
(0.159)
1.352a
(0.082)
0.009
(0.100)
147
0.672
a
0.801a
(0.129)
1.345a
(0.081)
0.011
(0.021)
147
n/a
Notes:
signicantly dierent from zero, two-tail test, 1% level.
Standard errors of estimated coecients are shown in parentheses.
Dependent variable is ln(Wi,t) where Wi,t win ratio.
n/a not available.
Source: Dawson (2000).
a
(5.5)
where Wi,t is team is win ratio in season t; qi,t is the mean of the estimated
transfer values of all players who featured in a minimum number of rstteam matches; t is a linear time trend, included to allow for possible
changes in the position of the ecient production frontier over time; and
ui and vi,t are the ineciency and random components of the error term,
as before.
Dawson generates manager eciency scores using English football
league data covering the 1993 to 1998 seasons inclusive. For purposes of
illustration, some results for the Premier League only are presented,
assuming eciency to be time invariant. Table 5.1 shows the results
obtained from the FE and RE estimations obtained by pooling the data
over the six seasons; and from a maximum likelihood (ML) estimation in
which ui are distributed half-normal, obtained from separate cross-sectional estimations for each of the six seasons individually.The coecient
for the team quality measure (qi,t) in each of the estimated models is positive and statistically signicant. The time trend is insignicant in all
models except FE, indicating that the position of the ecient frontier
does not change over time. The overall explanatory power of the models
258
RE
ML
0.549
0.163
1.000
0.071
0.633
0.103
1.000
0.391
0.825
0.059
0.915
0.561
1
0.983
0.926
0.709
1
0.878
0.733
1
0.614
1
2
8
15
24
12
4
3
1
2
1
2
11
33
18
5
2
0
0
0
4
50
15
2
1
0
0
0
0
0
259
Ferguson does perform better according to specications that use the raw player characteristics as frontier inputs: he is assigned rankings of 7th and 1st respectively in the RE
and ML versions of this model (see Dawson, Dobson and Gerrard, 2000b).
260
J. Gregory
(Aston Villa)
A. Wenger
(Arsenal)
G. Vialli
(Chelsea)
P. Reid
(Sunderland)
R. Gullitt
(Chelsea)
W. Bonds
(West Ham)
D. Webb
(Chelsea)
M. ONeill
(Leicester)
R. Wilkins
(QPR)
M. Walker
(Norwich)
K. Keegan
(Newcastle)
D. Jones
(Southampton)
P. Neal
(Coventry)
C. Todd
(Bolton)
R. Gould
(Coventry)
O. Ardiles
(Tottenham)
K. Dalglish
(Blackburn)
G. Graham
(Leeds)
D. Wilson
(Barnsley)
G. Souness
(Liverpool)
J. Smith
(Derby)
G. Francis
(QPR)
C. Gross
(Tottenham)
B. Rioch
(Arsenal)
D. Livermore
(Tottenham)
Eciency
(RE)
1.000
1.000
0.852
0.862
0.812
0.829
3=
0.779
0.799
3=
0.777
0.796
0.773
0.771
0.729
0.761
0.729
0.742
0.721
0.739
0.720
10
0.732
10
0.706
11
0.730
11
0.693
12
0.729
12=
0.674
13
0.709
12=
0.663
14
0.701
14
0.661
15
0.695
15=
0.645
16
0.694
15=
0.643
17
0.686
17=
0.635
18
0.685
17=
0.633
19
0.684
19=
0.629
20
0.681
19=
0.623
21
0.679
21
0.618
22
0.672
22
0.618
23
0.668
23
0.610
24
0.666
24
0.609
25
0.664
25
J. Gregory
(Aston Villa)
A. Wenger
(Arsenal)
G. Vialli
(Chelsea)
R. Gullitt
(Chelsea)
P. Reid
(Sunderland)
W. Bonds
(West Ham)
M. ONeill
(Leicester)
R. Wilkins
(QPR)
D. Jones
(Southampton)
K. Keegan
(Newcastle)
D. Webb
(Chelsea)
M. Walker
(Norwich)
C. Todd
(Bolton)
P. Neal
(Coventry)
D. Wilson
(Barnsley)
G. Graham
(Leeds)
J. Smith
(Derby)
R. Gould
(Coventry)
C. Gross
(Tottenham)
O. Ardiles
(Tottenham)
K. Dalglish
(Blackburn)
R. Atkinson
(She Wed)
B. Rioch
(Arsenal)
G. Souness
(Liverpool)
R. Hodgson
(Blackburn)
Rank Manager
(club)
A. Wenger
(Arsenal)
K. Keegan
(Newcastle)
J. Gregory
(Aston Villa)
R. Gullitt
(Chelsea)
M. ONeill
(Leicester)
R. Wilkins
(QPR)
M. Walker
(Norwich)
A. Ferguson
(Man Utd)
G. Vialli
(Chelsea)
K. Dalglish
(Blackburn)
P. Reid
(Sunderland)
P. Neal
(Coventry)
C. Todd
(Bolton)
W. Bonds
(West Ham)
R. Gould
(Coventry)
G. Graham
(Leeds)
O. Ardiles
(Tottenham)
G. Francis
(QPR)
J. Kinnear
(Wimbledon)
J. Smith
(Derby)
D. Webb
(Chelsea)
D. Jones
(Southampton)
G. Souness
(Liverpool)
H. Redknapp
(West Ham)
H. Kendall
(Everton)
Eciency
(ML)
0.915
0.911
0.902
0.902
0.891
0.887
0.885
0.881
0.879
0.876
0.874
0.873
0.873
0.871
0.867
0.867
0.864
0.864
0.863
0.863
0.862
0.861
0.859
0.855
0.854
261
Included among the manager human capital measures are the amount
of experience the manager gained as a player (number of league games
played and number of clubs played for); the quality of that experience
(number of international caps); and a number of measures reecting
managerial experience. These include the number of previous clubs
managed; the number of months spent managing before the current job;
the total number of months spent managing in the current division; the
total number of months spent managing the current club; and dummy
variables reecting whether or not the manager had a prior aliation with
the club, either as player, assistant manager or manager.The club-specic
factors include the total wage bill of the club and the average managerial
tenure at the club over the past 10 and 25 seasons.
Dawson nds that most of the manager human capital measures are
signicant in explaining variations in managerial eciency. The number
of clubs the manager played for and the number of international caps
both improve eciency. The latter nding reects an important
dierence between football (worldwide), and North American team
sports in which there is little or no competition at international level.
Football players apparently do gain valuable experience while on international duty, both by playing under the guidance of the national team
manager, and by gaining experience of dierent playing styles and tactics
adopted by opposition teams. This experience provides tangible benets
subsequently, when a former international player pursues a career in
management.
Elsewhere, Dawsons empirical results suggest that older managers are
more ecient than their younger counterparts. Managerial experience is
important, and eciency increases if there was a previous association with
the managers current club. Contrary to expectation, however, the number
of months spent managing in the current division and the total number of
months spent managing before the present job are both inversely related to
eciency. All of the club-specic factors are signicant. Managerial
eciency increases with the absolute level of wages paid relative to the
average.Assuming that the managers relative pay is correlated with the relative value of the clubs overall wages and salaries bill, this is consistent with
the notion that higher managerial pay induces higher performance.
Managers at clubs with high average tenure levels are found to be less
ecient than those employed by clubs where the threat of dismissal is
higher.
Dawsons study is the rst systematic empirical investigation of the
eciency of managers in English professional football, using a novel way
of measuring the quality of the playing input based on predicted transfer
values.The results appear to be robust over a variety of estimation procedures.The most ecient managers are not necessarily those that manage
262
the most successful teams. Managers that gain honours or keep a team
aoat with relatively limited playing resources feature prominently
towards the top of the estimated eciency rankings, as of course they
should.
5.4 Conclusion
The contribution of the football manager to team performance can be
broken down into a direct and an indirect component.Taking the collection of players at his disposal as given, the managers direct contribution is
to maximise performance through astute team selection, superior tactics
and powers of motivation. Over the longer term, the managers indirect
contribution is to enhance his existing squad, by coaching players so as to
improve their skills, and by eective dealings in the transfer market.
Despite the fact that the output of the professional sports team is easily
measured by win ratios or league standings, and much data on the characteristics of the major inputs (players and manager) is easily available and
in the public domain, the indivisibility of the team eort which ultimately
determines performance poses a major challenge for any researcher
seeking to isolate and measure precisely the contribution made by the
manager to the eectiveness of the teams performance.
A realistic assessment of the quality of the managerial contribution can
be made only if the quality of the playing resources that were at the
managers disposal is measured and taken into account. In the case of
North American professional team sports, considerable academic eort
has been invested in separating the managerial from the playing input to
team production, using sophisticated econometric techniques such as
stochastic frontier analysis. In this literature, it is clear that empirical
results can depend heavily on the way in which the inputs are measured;
on the specication of the production function and the distributional
assumptions concerning the ineciency and random components of the
error term; and on the estimation method employed. Clearly it is important to be able to demonstrate that any set of empirical results is reasonably consistent over a range of denitions, model specications and
estimation procedures.
The recent work of Peter Dawson, which has been reviewed in some
detail in this chapter, represents a promising rst attempt at bringing
similar methods to bear on English professional football.The lack of adequate historical data on most aspects of player performance apart from
goals scored and conceded, and the fact that the playing and managerial
inputs are in any case less easily separable than in North American sports
such as baseball, make this task a formidable challenge. The valuation of
263
players using a model created from player transfer data oers a novel way
round some of the problems, though not without presenting some further
diculties of its own.The inclusion of nancial data in the measurement
of inputs also acknowledges the critical fact that the quality of the
resources at the English football managers disposal depends, to a large
and ever-increasing extent, on the strength of his clubs nancial muscle.
265
extent to which the removal of a manager inuences the teams performance subsequently. According to one view, it is simple common sense
for a manager to be replaced when his team is performing badly.
Assuming that a more eective replacement can be appointed, team performance should improve post-succession. An alternative view is that a
change of leadership is usually disruptive, especially if it occurs while the
season is in progress, and tends to make matters worse rather than better.
A third view suggests that on average, the appointment of a new manager
has no eect whatever on team performance, because performance
depends primarily on the playing talent available to the team, and not on
the managerial input. Managerial dismissal is a form of scapegoating:
during a poor run of results team owners can appease disgruntled spectators by removing the manager, even if doing so is not expected to have any
real eect on team performance. The US evidence on these matters is
considered in section 6.6. Some new results for English football are then
presented in sections 6.76.8. These seek to identify managerial succession eects, using season-level data (in section 6.7) and match-level data
(in section 6.8).
6.1 Patterns of managerial change in English football
The analysis of managerial change and team performance draws on data
for 1,092 managerial spells with English clubs that were recorded
between the 1973 and 1999 seasons (inclusive). Exactly 1,000 of these
spells began after the start of the 1973 season, while 1,010 terminated
before the start of the 2000 season. In total, the data set includes 918
complete spells that both began after the start of the 1973 season and terminated before the start of the 2000 season.
For the 1,000 spells that began after the start of the 1973 season, table
6.1 shows the top 50 ranked in descending order of duration, measured in
matches completed. Of course, in view of the notorious insecurity of the
English football managers job tenure, the longevity of a spell is itself often
a meaningful measure of its success. In some cases it may also reect
attributes or attitudes of the clubs owners, which inuence their willingness either to stick with or dismiss a manager during a lean spell. Among
the longest-serving managers shown in table 6.1, Brian Clough
(Nottingham Forest), Dario Gradi (Crewe Alexandra), John Rudge (Port
Vale), Joe Royle (Oldham Athletic) and Graham Taylor (Watford) all
steered their clubs through one or more promotion campaigns, and then
consolidated and achieved further success at a higher level. In a slightly
dierent category, Alex Ferguson (Manchester United), Bob Paisley
(Liverpool) and George Graham (Arsenal) all achieved lasting success
266
Team
Dates
Matches
Win ratio
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27=
27=
29
30
31
32
33
34=
34=
36
37
38
39
40=
40=
42=
42=
44=
44=
46
47
48
49
50
Nottingham Forest
Crewe Alexandra
Port Vale
West Ham United
Oldham Athletic
Manchester United
Southampton
Wrexham
Watford
Tranmere Rovers
Bournemouth
Derby County
Liverpool
Crystal Palace
Hudderseld Town
Leyton Orient
Arsenal
Charlton Athletic
Charlton Athletic
Wolverhampton Wandrs
Luton Town
Sheeld United
Tottenham Hotspur
Leeds United
Chester City
Southend United
Blackpool
Colchester United
Arsenal
Aston Villa
Doncaster Rovers
Brighton
Lincoln City
Bolton Wanderers
Hartlepool United
Reading
Norwich City
Gillingham
Hull City
Rotherham United
Wimbledon
Grimsby Town
Norwich City
Chestereld
Wimbledon
York City
Manseld Town
Stockport County
Bristol City
Aldershot
197593
1983
198499
197489
198294
1986
197385
1989
197787
198796
198392
198493
197483
198493
197886
198291
198695
1991
198291
198694
197886
198895
197684
198896
198592
197683
19829
197582
197683
197482
197885
198793
197885
198592
197683
197784
19807
19817
19917
19739
19929
198894
197380
1993
19817
19939
19839
198995
19828
198591
758
726
679
621
534
503
488
439
434
415
401
397
378
376
374
368
363
360
358
355
352
339
336
335
333
322
320
320
318
317
316
312
311
302
302
301
299
297
296
295
295
288
288
284
284
277
276
274
270
267
0.574
0.518
0.497
0.507
0.504
0.669
0.550
0.502
0.568
0.564
0.510
0.524
0.692
0.535
0.537
0.507
0.606
0.517
0.443
0.569
0.523
0.506
0.531
0.584
0.482
0.550
0.517
0.527
0.563
0.576
0.494
0.466
0.532
0.527
0.440
0.542
0.517
0.571
0.458
0.517
0.483
0.531
0.476
0.560
0.600
0.487
0.495
0.582
0.574
0.412
Brian Clough
Dario Gradi
John Rudge
John Lyall
Joe Royle
Alex Ferguson
Lawrie McMenemy
Brian Flynn
Graham Taylor
John King
Harry Redknapp
Arthur Cox
Bob Paisley
Steve Coppell
Mike Buxton
Frank Clark
George Graham
Alan Curbishley
Lennie Lawrence
Graham Turner
David Pleat
Dave Bassett
Keith Burkinshaw
Howard Wilkinson
Harry McNally
Dave Smith
Sam Ellis
Bobby Roberts
Terry Neill
Ron Saunders
Billy Bremner
Barry Lloyd
Colin Murphy
Phil Neal
Billy Horner
Maurice Evans
Ken Brown
Keith Peacock
Terry Dolan
Jimmy McGuigan
Joe Kinnear
Alan Buckley
John Bond
John Duncan
Dave Bassett
Alan Little
Ian Greaves
Danny Bergera
Terry Cooper
Len Walker
267
268
Team
Dates
Matches
Win ratio
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
Fulham
Liverpool
Arsenal
Liverpool
Liverpool
Hull City
Manchester United
Walsall
Blackburn Rovers
Oxford United
Chelsea
Newcastle United
Watford
Southend United
Sunderland
Leeds United
Bradford City
Southend United
Brentford
West Bromwich
Barnet
Everton
Wolverhampton Wandrs
Sheeld United
Manchester United
Southend United
Wigan Athletic
Shrewsbury Town
Peterborough United
Blackburn Rovers
Swindon Town
Crystal Palace
Wycombe Wandrs
Fulham
Queens Park Rangers
Bristol Rovers
Cambridge United
Liverpool
Manchester City
Lincoln City
Cambridge United
Hartlepool United
Arsenal
Sheeld United
Colchester United
Chelsea
Bournemouth
Arsenal
Rotherham United
Bristol City
19989
198591
1996
197483
19835
19824
1986
1998
19915
19825
1998
19927
1997
19867
1995
1998
19812
1995
1998
19923
19913
19817
1998
19978
19816
1993
19856
1978
19912
197981
19849
19967
19935
19967
19804
19835
19902
19948
1998
19727
19748
198991
198695
19957
1994
198891
19802
19846
197981
198890
46
224
114
378
84
111
503
46
169
160
51
205
92
33
183
31
70
14
46
46
82
252
30
34
225
28
60
33
87
88
228
52
88
71
157
90
128
184
60
209
149
75
363
71
19
129
87
97
73
108
0.761
0.732
0.702
0.692
0.673
0.671
0.669
0.663
0.663
0.663
0.657
0.656
0.652
0.652
0.650
0.645
0.643
0.643
0.641
0.641
0.640
0.637
0.633
0.632
0.629
0.625
0.625
0.621
0.621
0.619
0.618
0.615
0.614
0.613
0.612
0.611
0.609
0.609
0.608
0.608
0.607
0.607
0.606
0.606
0.605
0.605
0.603
0.603
0.603
0.602
Kevin Keegan
Kenny Dalglish
Arsne Wenger
Bob Paisley
Joe Fagan
Colin Appleton
Alex Ferguson
Ray Graydon
Kenny Dalglish
Jim Smith
Gianluca Vialli
Kevin Keegan
Graham Taylor
David Webb
Peter Reid
David OLeary
Roy McFarland
Steve Thompson
Ron Noades
Osvaldo Ardiles
Barry Fry
Howard Kendall
Colin Lee
Nigel Spackman
Ron Atkinson
Barry Fry
Bryan Hamilton
Richie Barker
Chris Turner
Howard Kendall
Lou Macari
Dave Bassett
Martin ONeill
Micky Adams
Terry Venables
David Williams
John Beck
Roy Evans
Joe Royle
Graham Taylor
Ron Atkinson
Cyril Knowles
George Graham
Howard Kendall
George Burley
Bobby Campbell
David Webb
Don Howe
Ian Portereld
Joe Jordan
269
ended at the close of the 1985 season following two consecutive promotions, is another.
Table 6.3 shows an alternative set of rankings, which make adjustments
to the points awarded for wins, draws and losses during the two seasons
following promotion or relegation. In calculating the adjusted win ratios,
0.2 extra points are awarded per match played during the season following promotion, and 0.1 extra points per match are awarded during the
second season following promotion. Similarly, negative adjustments of
0.2 and 0.1 points per match are made during the rst and second
seasons following relegation.The precise numerical adjustments are arbitrary, but they do address the problems discussed in the previous paragraph in a straightforward manner. Naturally, managers whose spells
included one or more seasons following promotion are beneciaries of
the adjustment: notably, John Beck (Cambridge United), George Kerr
(Grimsby Town) and Chris Turner (Peterborough United), who occupy
the top three positions, each having achieved two promotions. In the
process of taking Wimbledon from Division 4 to Division 1, Dave Bassett
achieved no fewer than four promotions and one relegation. Accordingly,
the ranking of Bassetts spell with Wimbledon jumps from just outside the
top 50 (52nd) in table 6.2 to 11th in table 6.3.
Audas, Dobson and Goddard (1999) discuss a number of factors contributing to the chronic insecurity of the English football managers position. On the one hand, match results (the managers most important
performance measure) are completely transparent, easily interpreted and
instantly in the public domain.1 On the other hand, the managers ability
to inuence results is constrained by uncertainty concerning the tness,
performance and motivation of players.Whether or not a group of players
will combine to form a successful team has always depended on a
mixture of luck and judgement, in proportions which ultimately are
impossible to fathom. In any event, each match and each seasons league
programme are by their very nature zero-sum aairs, in which failure for
some participants is inevitable. As characteristics of the managers position, the combination of direct accountability for outcomes that are
transparent and public, and imperfect control over the processes which
determine the same outcomes, could not be better designed to minimise
job security.
Whereas inadequate performance in business tends to lead to the loss
of customers followed by events such as internal restructuring, acquisition by new owners or liquidation, football clubs are highly resilient, even
1
In contrast, the types of performance measure on which business managers are judged,
such as sales, growth or prots, are relatively opaque and subject to interpretation.
270
Team
Dates
Matches
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
Cambridge United
Grimsby Town
Peterborough United
Fulham
Hull City
Watford
Bury
Blackburn Rovers
Swindon Town
Liverpool
Wimbledon
Wycombe Wandrs
Newcastle United
Oxford United
Arsenal
Walsall
Exeter City
Liverpool
Bolton Wandrs
Brighton and Hove Albion
Bradford City
West Bromwich
Manseld Town
Port Vale
Liverpool
Manchester United
Hartlepool United
Crystal Palace
Gillingham
Barnsley
Queens Park Rangers
Newcastle United
Walsall
Wigan Athletic
Chelsea
Watford
Rotherham United
Charlton Athletic
Preston North End
Middlesbrough
Lincoln City
Southend United
Wigan Athletic
Bournemouth
Leeds United
Barnsley
Bristol City
Sunderland
Southend United
Cambridge United
19902
197982
19912
19989
19824
1997
19958
19915
19849
198591
19817
19935
19927
19825
1996
19947
19779
197483
19925
197681
19812
19757
19746
19803
19835
1986
19913
197680
19959
197880
19804
19967
1998
19813
1998
197787
19813
19812
19948
19737
19727
19867
19958
19802
1998
19804
19978
197981
1995
19748
128
105
87
46
111
92
133
169
228
224
284
88
205
160
114
130
95
378
138
214
70
86
103
195
84
503
74
182
184
108
157
42
46
96
51
434
74
42
143
178
209
33
127
87
31
142
71
112
14
149
0.798
0.795
0.789
0.761
0.754
0.752
0.752
0.737
0.737
0.732
0.722
0.718
0.718
0.715
0.702
0.695
0.693
0.692
0.691
0.690
0.689
0.680
0.674
0.673
0.673
0.669
0.669
0.668
0.667
0.667
0.665
0.664
0.663
0.663
0.657
0.657
0.657
0.652
0.652
0.652
0.652
0.652
0.647
0.647
0.645
0.645
0.644
0.644
0.643
0.642
John Beck
George Kerr
Chris Turner
Kevin Keegan
Colin Appleton
Graham Taylor
Stan Ternent
Kenny Dalglish
Lou Macari
Kenny Dalglish
Dave Bassett
Martin ONeill
Kevin Keegan
Jim Smith
Arsne Wenger
Chris Nicholl
Bobby Saxton
Bob Paisley
Bruce Rioch
Alan Mullery
Roy McFarland
Johnny Giles
Dave Smith
John McGrath
Joe Fagan
Alex Ferguson
Alan Murray
Terry Venables
Tony Pulis
Allan Clarke
Terry Venables
Jack Charlton
Ray Graydon
Larry Lloyd
Gianluca Vialli
Graham Taylor
Emlyn Hughes
Alan Mullery
Gary Peters
Jack Charlton
Graham Taylor
David Webb
John Deehan
David Webb
David OLeary
Norman Hunter
John Ward
Ken Knighton
Steve Thompson
Ron Atkinson
271
In the past 30 years, many clubs (including big names such as Bristol City, Crystal
Palace, Middlesbrough and Wolverhampton Wanderers) have teetered on the edge of
bankruptcy but only two (Aldershot and Maidstone United) have gone out of business
while members of the league.
272
explanation of the reason for the change as either a dismissal or a resignation. To do so would probably lead to signicant over-counting of the
number of voluntary departures, since many resignations actually occur
in response to behind-the-scenes pressure from the chairman or directors. In extremis this can take the form of an ultimatum to resign or be
sacked (Audas, Dobson and Goddard, 1999).
In classifying managerial departures as either involuntary or voluntary,
any information available on the circumstances at the time of the departure was assessed. Particularly relevant criteria were the teams performance prior to termination; the nature of any new appointment accepted
by the manager immediately after his departure; and the reasons given by
the club or the manager for the latters departure. Involuntary departures
(the majority) include those that were at the clubs instigation, as well as
resignations for which there was no obvious reason other than disappointing results or other pressures emanating from the managers current
appointment.Voluntary departures include those for which there is clear
and tangible evidence that the manager instigated the move willingly.
Such evidence is usually in the form of an immediate or imminent move
to another job at a club (at home or overseas) of comparable or higher
status, or a move to a national team managers position.
A couple of recent examples will illustrate the method of classication.
First, of the 10 close season departures during the summer of 1999, only
Kevin Keegans move from Fulham to the England managers post is
classied unambiguously as voluntary. Although Steve Bruce (Sheeld
United) and Brian Little (Stoke City) both subsequently obtained managerial appointments at Hudderseld Town and West Bromwich Albion,
respectively, these appointments do not appear to have motivated their
departures. Bruces move from Sheeld United was attributed to
instability and disputes at boardroom level, while Little was dismissed
after a disappointing campaign and an especially poor run of results
towards the end of the 1999 season.3
Second, after a successful spell at Reading, Mark McGhee made two
voluntary moves in quick succession, rst to Leicester City (in December
1994) and then Wolverhampton Wanderers (December 1995). Three
campaigns in which promotion to the Premier League failed to materialise resulted in McGhees involuntary removal from Wolverhampton in
November 1998. McGhees move from Leicester triggered Martin
3
At the time of writing it remains to be seen whether Ron Atkinson (Nottingham Forest)
will eventually accept a new appointment or retire permanently from management.
Other departures during the summer of 1999 included Nigel Pearson (Carlisle United),
Peter Jackson (Hudderseld Town), Dennis Smith (West Bromwich Albion), Ray
Mathias (Wigan Athletic) and Joe Kinnear (Wimbledon).
273
ONeills switch from Norwich City as his successor. This move is also
classied as voluntary; but it is a borderline case, since at the time ONeill
was facing signicant spectator discontent only six months into his
appointment at Norwich. Ironically, while Wolverhampton and Norwich
have both remained rmly rooted in Division One, Leicester gained promotion at the end of ONeills rst season in 1996, and have since become
an established Premier League team. Recently ONeill has once again
moved voluntarily, from Leicester to Celtic; though this move took place
in the summer of 2000, after the end of the sample period for the data set
used in this chapter.
Table 6.4 shows the numbers of involuntary and voluntary managerial
departures per season.The data suggest that the level of job insecurity of
football managers has increased over time. Although there is considerable year-on-year variation in the total number of departures, the longterm trend is clearly upward.The nine-year averages for the total number
of departures increase from 33.8 per year (for seasons 19731981 inclusive) through 35.6 (19821990) to 42.9 (19911999). Signicantly the
increase in the total is driven by a rise in the number of involuntary
departures, while the number of voluntary departures remains roughly
constant. In recent seasons an increasing tendency for senior clubs to
appoint current or recently retired players directly into managerial posts,
or to recruit managers from overseas, may account for the pattern.4 With
no fewer than 61 managerial departures, the 1995 season established
what is presumably an all-time record for managerial turnover. Although
this gure looks like an outlier, it is perhaps signicant that at the end of
this season, the league structure was changed from 22-24-24-22 to 2024-24-24 teams in the four divisions. The change meant there was one
less promotion berth and one more relegation berth than usual in each
division. This seems to have created sucient panic at boardroom level
to have cost many managers their jobs during the course of the 1995
season.
Table 6.5 summarises some other features of the managers data set.
There is a high level of monthly variation in the incidence of managerial
departure. 705 of the 1,010 departures (69.8 per cent) took place during
the course of the season, and only 305 (30.2 per cent) during the close
4
There has been no tendency, however, for the average age of football managers to decline
over time. For the league as a whole, the average age of managers in post at the start of
the 1973 season was 43.4 years. The average age fell slightly to 42.2 at the start of the
1986 season, but then increased to 44.8 by the start of the 1999 season. For Division
1/Premier League managers, the corresponding gures were slightly higher: 46.5
(1973), 44.1 (1986) and 47.7 (1999). These averages are based on all managers for
whom date-of-birth information was available; a very small number of managers (at
most two out of 92) are excluded from these calculations for certain seasons.
274
Involuntary
Voluntary
Total
1973
1974
1975
1976
1977
1978
1979
1980
1981
27
20
25
23
30
37
23
20
34
4
13
7
4
6
8
9
1
13
31
33
32
27
36
45
32
21
47
Average
(19731981)
26.6
7.2
33.8
1982
1983
1984
1985
1986
1987
1988
1989
1990
33
32
32
32
34
30
25
23
33
4
4
6
3
6
8
3
8
4
37
36
38
35
40
38
28
31
37
Average
(19821990)
30.4
5.1
35.6
1991
1992
1993
1994
1995
1996
1997
1998
1999
35
31
34
30
50
34
38
41
28
Average
(19911999)
35.7
12
2
6
6
11
5
10
7
6
7.2
47
33
40
36
61
39
48
48
34
42.9
275
Voluntary
Total
Month
August
September
October
November
December
January
February
March
April/May
Close season
18
51
82
86
70
77
69
79
82
220
2
9
16
12
18
14
6
8
6
85
1,020
1,060
1,098
1,098
1,088
1,091
1,075
1,087
1,088
1,305
Division
D1/Prem
D2/One
D3/Two
D4/Three
156
198
219
261
43
48
54
31
1,199
1,246
1,273
1,292
Total
834
176
1,010
season. Of the 705 within-season departures, 91 (12.9 per cent) were voluntary, whereas of the 305 close season departures, 85 (27.9 per cent)
were voluntary. Departure within the season is therefore relatively more
likely to be at the clubs behest, while departure during the close season
(when most managerial contracts tend to expire) is more likely to be at
the managers behest. October and November, when failure to realise
over-optimistic pre-season aspirations rst becomes inevitable for many
clubs, are the peak months for within-season departures. January,
perhaps the latest time at which a new appointment might reasonably be
expected to bring about a change of fortune within the current season, is
another peak period; while MarchMay, when many clubs begin planning
for the next season, are also periods of high turnover. There appear to be
(slight) lulls in the rate of turnover during the months of December and
February.
Table 6.5 also shows that managerial insecurity is greater in the lower
than in the higher divisions. Both the number of involuntary departures
and the total number of departures vary inversely with divisional status. If
276
D1/Prem
D2/One
D3/Two
D4/Three
918
99.4
42
75
132
176
119.7
43.5
86.5
161.5
222
105.8
44
82.5
138
247
102.2
46
78
143
273
78.6
38
64
107
Involuntary departures
No.
765
Mean duration
94.9
1st quartile
41
Median
72
3rd quartile
127
136
110.4
42
82.5
141
183
99.1
41
75
133
199
99.3
45
75
137
247
79.9
39
63
112
Voluntary departures
No.
153
Mean duration
121.6
1st quartile
58
Median
92
3rd quartile
165
40
151.5
61
136
198
39
137.5
58
114
177
48
114.0
61.5
92
147.5
26
65.9
33
66
92
All departures
No.
Mean duration
1st quartile
Median
3rd quartile
Figure 6.1 Distribution of complete managerial spells, by duration, 19731999, all departures
Source: Managers data set.
Figure 6.2 Distribution of complete managerial spells, by duration, 19731999, involuntary departures
Source: Managers data set.
Figure 6.3 Distribution of complete managerial spells, by duration, 19731999, voluntary departures
Source: Managers data set.
280
divisions was higher than the mean duration of spells that ended in the
lower divisions. Finally, the shape of the distribution by duration diers
between spells that ended involuntarily and those that ended voluntarily.
For the latter the inverse relationship between mean duration and divisional status holds uniformly across all four divisions. For the former
there is little or no dierence between the mean and median duration of
spells that ended in Divisions 2/One and 3/Two. Although spells that
ended involuntarily in Division 1/Premier League were longer on average,
and those that ended in Division 4/Three were shorter, the dierence in
mean duration (about 30 matches) is much smaller than the corresponding dierence for spells that ended voluntarily (about 85 matches).
6.2 Determinants of managerial change: evidence from
North American team sports
Gruskys (1963) investigation of managerial succession in North
American baseball appears to have been the rst academic study to consider the possibility of a bidirectional link between managerial change and
team performance. For 16 professional baseball teams, Grusky obtains a
negative correlation between the number of managerial changes and the
average standing of each team, for both of the periods 192141 and
19518. In other words, teams that made the most managerial changes
tended to perform badly. Grusky also obtains a negative correlation
between the change in the average duration of managerial tenure between
these two periods, and the change in average standing. The teams that
increased their rate of managerial turnover the fastest therefore tended to
experience a decline in average standing. Grusky (1963) and Gamson
and Scotch (1964), in a comment on the Grusky paper, go on to identify
three possible theories of managerial turnover:
The common sense theory. When a team is under-performing, the
manager is held accountable and is likely to be replaced. If a more
eective replacement is hired, or if the incoming manager can learn from
the mistakes of his predecessor, team performance should improve postsuccession.This would be sucient to explain Gruskys nding of negative correlation between managerial turnover and team performance.
The vicious circle theory. Again, poor performance tends to trigger managerial change. But the disruptive eect of a change of leadership, particularly if it occurs within-season, tends to make matters worse rather
than better. Faltering teams can become trapped in a vicious circle of
high managerial turnover and declining performance. Gruskys empirical results therefore reect a bi-directional relationship between turnover and performance.
281
282
managers perform the rst of these tasks with equal eciency, so all
teams achieve the maximum attainable points scored and the minimum
attainable points conceded given their playing resources. But there are
dierences between managers in terms of the eciency with which points
scored and conceded are converted into win ratios. Each manager
achieves an eciency score based on the residuals, , from a regression of
the form:
ln(W ) 0 1ln(S/OS )
(6.1)
where W is the win ratio and S and OS are points scored and conceded,
respectively.
Scully (1995) obtains from OLS estimation of (6.1). It is shown that
career eciency is positively related to career length, demonstrating
(indirectly) that eciency does inuence the decision to terminate the
managers appointment. Evidence is also found that in baseball and basketball, player-managers are less ecient than their non-playing counterparts, and that average eciency increased over time in all three sports.
Scully (1995) models the decision to terminate the managers appointment directly. In principle the decision to terminate or retain should be
based on a comparison between the actual and the maximum attainable
win ratio, given the quality of playing resources at the managers disposal.
In practice, however, many other factors also come into play: owners subjective judgements about the quality of their own and their opponents
players; the managers relationships with both the owner and the players;
the managers past experience; the possibility that the owner may wish to
sacrice the manager primarily in order to send signals to players, supporters or the media; and last but not least, the inuence of luck on team
performance.
In Scullys empirical estimations, the dependent variable is a 0-1
dummy that indicates either continuation or termination of the
managers position within or at the end of each season.The independent
variable is the teams league standing at the end of the season (or at the
time of departure for within-season departures). Although this measure is
crude because it takes no account of either playing strengths or managerial eciency (as dened above), it is justied on the grounds that [f]or
both owner and manager, [the clubs standing] is the bottom line (Scully,
1995, p.161). No distinction is made between involuntary and voluntary
managerial departures.The linear probability model and logit and probit
models are used for estimation.The main nding, that in both sports the
decision to terminate is highly sensitive to league standing, is robust
across all specications.
Scullys (1994) study provides a number of (mainly technical)
283
284
285
286
(6.2)
287
(b
m
m
t) R
vi,t ui,t
i,t3 ... t1
(6.3)
m1
vi,t
ui,t
eciency score.6
In (6.3) the eciency scores are time-varying: there is a separate score for
each team in each season. For the purpose of dening variables for the
hazard function estimations, the indices i and t refer to teams and
seasons, respectively, from here on. Figure 6.4 illustrates the shape of the
estimated ecient frontier for the 1973 and 1999 seasons, over a realistic
range of values for
Ri,t3...t1 of up to 4 per cent. The frontier becomes
6
The distributional assumptions for the random error terms and the eciency scores are
vi,t ~ N(0,
2v) and ui,t | u i,t|where u i,t ~ N(0,
u2).
0.0
20.0
40.0
60.0
80.0
0.002
0.01
0.018
0.026
Figure 6.4 Estimated stochastic production frontiers, 1973 and 1999 seasons
Source: Managers data set.
Efficient
performance
(maximum
league
position
attainable:
92 = top of
first division,
91 = second,
etc.)
0.042
Share of league gate revenues
0.034
1973
1999
289
The full expression for the estimated frontier, with standard errors of estimated
coecients shown in parentheses, is as follows:
Pi,t [0.00790.4415at][11.35a0.10at]
R i,t3...t1 [0.6205a0.0092at]
R 2i,t3...t1
(2.283) (0.109)
(0.68) (0.03)
(0.0602) (0.0029)
R 3i,t3...t1 [0.000133a0.000003at]
R 4i,t3...t1 vi,t ui,t
[0.0152a0.0003at]
(0.0019) (0.0001)
(0.000019) (0.000001)
s 2v 121.8
s 2u 70.2
obs2,450
signicantly dierent from zero, two-tail test, 1 per cent level.
The inclusion of the interactions between each coecient and the time trend allow the
position of the ecient frontier to shift progressively over time. This is necessary,
because while the distribution of the output variable, Pi,t does not change over time, the
distribution of the inputs,
R i,t3...t1 does change with variation over time in the distribution of revenues between clubs. In 1999 it was possible to attain a relatively high position
in the lower divisions with a smaller share of total revenue than would have been possible
in 1973. Conversely, the teams occupying the highest positions in the top division in
1999 tended to have a higher revenue share than their counterparts in 1973.
290
Manager
Team
Season
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
Dave Bassett
Dario Gradi
Dave Bassett
John Docherty
Gary Megson
John Beck
Dave Bassett
Stan Ternent
Mark McGhee
Bobby Gould
John Docherty
John Docherty
Kenny Dalglish
Bob Stokoe
Dave Bassett
Ron Atkinson
Dario Gradi
Graham Turner
Joe Kinnear
Jimmy Sirrell
Stan Ternent
Emlyn Hughes
Bob Stokoe
Alan Ashman
Brian Clough
Jimmy Adamson
Tom Johnston
Joe Royle
Gary Megson
Alan Buckley
Neil Warnock
Alan Nelson
Chic Bates
Joe Kinnear
George Kerr
Jim Smith
Graham Turner
Kenny Dalglish
Chris Turner
John Docherty
Jimmy McGuigan
Dennis Smith
Richie Barker
Neil Warnock
Jimmy Sirrell
Danny Wilson
Neil Warnock
Danny Wilson
Alan Durban
Alan Curbishley
Wimbledon
Crewe Alexandra
Wimbledon
Cambridge United
Stockport County
Cambridge United
Wimbledon
Bury
Reading
Wimbledon
Millwall
Cambridge United
Blackburn Rovers
Carlisle United
Wimbledon
Cambridge United
Crewe Alexandra
Shrewsbury Town
Wimbledon
Notts County
Bury
Rotherham United
Carlisle United
Carlisle United
Nottingham Forest
Burnley
York City
Oldham Athletic
Stockport County
Grimsby Town
Bury
Charlton Athletic
Shrewsbury Town
Wimbledon
Grimsby Town
Oxford United
Shrewsbury Town
Blackburn Rovers
Cambridge United
Cambridge United
Rotherham United
Oxford United
Shrewsbury Town
Notts County
Notts County
Barnsley
Notts County
Barnsley
Shrewsbury Town
Charlton Athletic
1987
1998
1986
1979
1998
1992
1985
1998
1995
1988
1989
1980
1993
1984
1984
1978
1999
1980
1994
1982
1997
1982
1983
1975
1978
1974
1975
1987
1999
1999
1999
1976
1985
1995
1981
1985
1984
1994
1991
1983
1977
1998
1979
1992
1981
1998
1991
1997
1976
1999
16, D1
11, Div One
13, D2
12, D2
18, Div One
15, D2
12, D2
17, Div One
12, Div One
17, D1
10, D1
18, D2
14, Prem
17, D2
12, D3
12, D3
18, Div One
13, D2
16, Prem
15, D1
11, Div Two
17, D2
14, D2
22, D1
11, D1
16, D1
15, D2
13, D2
16, Div One
11, Div One
22, Div One
12, D2
18, D2
19, Prem
17, D2
11, D2
18, D2
12, Prem
11, D3
12, D2
14, D3
12, Div One
11, D3
21, D1
12, D2
12, Div One
14, D2
19, Prem
19, D3
18, Prem
12.224
12.667
12.679
12.706
12.729
12.748
12.766
12.813
12.862
12.901
12.936
12.971
13.023
13.078
13.089
13.092
13.103
13.106
13.107
13.107
13.112
13.121
13.153
13.170
13.176
13.177
13.181
13.239
13.266
13.323
13.336
13.339
13.345
13.362
13.364
13.365
13.385
13.408
13.433
13.451
13.458
13.462
13.464
13.465
13.470
13.473
13.478
13.479
13.482
13.484
291
Manager
Team
Season
2441
2442
2443
2444
2445
2446
2447
2448
2449
2450
Jimmy Dickinson
Steve Gritt
Steve Burkinshaw
Mike Everitt
Harry Haslam
Bill McGarry
Colin Appleton
Ian Portereld
Bobby Houghton
Terry Cooper
Portsmouth
Brighton and Hove Albion
Sheeld Wednesday
Brentford
Sheeld United
Wolverhampton
Swansea City
Sheeld United
Bristol City
Bristol City
1979
1997
1976
1974
1981
1986
1985
1982
1982
1983
17, D4
23, Div Three
20, D3
19, D4
21, D3
23, D3
20, D3
11, D4
21, D2
23, D3
15.609
15.648
15.726
16.251
16.682
16.699
17.615
17.843
18.331
19.670
league position achieved by team i in the season preceding the current managers initial appointment
These denitions imply that appointments terminated very early in a season (during
August or September) are attributed by the model to performance indicators (or other
covariates) from the previous season, while appointments terminated subsequently are
attributed to current-season performance. A new manager appointed after October does
not take the blame (or the credit) for performance during the season of appointment,
and starts to appear in the estimations only in the season after appointment (if he is still
in post in October of that season). Managers appointed after October and terminated
before the following October do not appear in the estimations at all.
292
Oi
Ii
843
550
320
192
193
160
139
127
116
118
117
Duration in
seasons, t
0
1
2
3
4
5
6
7
8
9
10
Spells of
duration t
live on 1/10
in any season
270
216
119
191
130
120
112
110
117
111
Departures
before 1/10 of
following season
23
14
19
18
13
11
10
11
11
10
Right-censored
spells, live
on 1/10/99
at duration
t, and still live
on 1/10/2000
1.000
0.680
0.413
0.259
0.136
0.092
0.062
0.043
0.027
0.015
0.013
Survivor
function, St
0.000
0.320
0.587
0.741
0.864
0.908
0.938
0.957
0.973
0.985
0.987
Distribution
function, Ft
0.320
0.267
0.154
0.123
0.044
0.030
0.019
0.016
0.012
0.002
Probability
function, ft
0.320
0.393
0.372
0.474
0.323
0.333
0.308
0.370
0.438
0.125
Hazard
function, ht
Table 6.8. Distribution of complete and right-censored managerial spells, by duration (in seasons) and crude estimates of the
survivor, distribution, probability and hazard functions
294
Covariates
i,t
(Pi,tPi )
Ci,t
Ai,t
A2i,t
Mi
Si
Oi
Ii
Yi,t
obs.
departures
Inv.
(1)
Inv.
(2)
Vol.
(3)
Vol.
(4)
0.0604a
(0.0189)
0.0277a
(0.0036)
0.1106a
(0.0337)
0.0867
(0.0640)
0.0012c
(0.0007)
0.0012
(0.0009)
0.0126
(0.0079)
0.1999
(0.3705)
0.0298
(0.0896)
0.0125b
(0.0057)
2,157
649
0.0600a
(0.0189)
0.0278a
(0.0036)
0.1112a
(0.0337)
0.0887
(0.0639)
0.0012c
(0.0007)
0.0012
(0.0009)
0.0143b
(0.0073)
0.1497a
(0.0585)
0.0121c
(0.0071)
0.0006
(0.0640)
0.2592
(0.1910)
0.0033
(0.0022)
0.0045b
(0.0021)
0.0011
(0.0176)
0.7741
(0.5685)
0.2909
(0.1944)
0.0061
(0.0130)
2,157
132
0.1500a
(0.0583)
0.0122c
(0.0071)
0.0123b
(0.0057)
2,157
649
0.2660
(0.1897)
0.0033
(0.0022)
0.0044b
(0.0021)
0.8138
(0.5260)
0.2928
(0.1831)
2,157
132
Notes:
a
signicantly dierent from zero, two-tail test, 1% level;
b
5% level;
c
10% level.
Standard errors of estimated coecients are shown in parentheses.
Key:
i,t managerial eciency score, obtained from (6.3)
Pi,t league position
Pi league position achieved in the season preceding the managers appointment
Ci,t victories achieved in most recent FA Cup campaign
Ai,t managers age (years)
Mi managers previous managerial experience in English league football (months)
Si managers playing experience in English league football (seasons)
Oi Scottish/overseas managerial experience dummy
Ii international playing experience dummy
Yi,t time trend (1 1973 season, 2 1974, and so on).
295
The hazard is minimised at about age 32; most managerial careers start later than this.
The FA Cup variable, although signicant in the involuntary departure hazard, does not
seem to aect the voluntary departure hazard. If cup success is largely a matter of chance
(often depending heavily on luck in the draw) it is unsurprising that cup success typically
does not tend to attract job oers from other clubs. A cup run may nevertheless be useful
to an incumbent manager in mollifying supporters and helping to stave o the threat of
involuntary departure.
296
297
converge towards the divisional average. The model also does not allow
for cross-sectional interdependence between the probabilities of simultaneous departure for managers of dierent clubs. In practice, however, one
can think of instances when, having just lost the services of its manager to
Club A, Club B has oered the post to the manager of Club C, which in
turn has attempted to secure Club Ds manager as his successor.
Nevertheless, the incidence of chain-reactions of this kind is probably
lower today than it has been in the past, thanks partly to an increased tendency for clubs to appoint foreign managers, as well as serious attempts
that have been made by the football authorities (including the League
Managers Association) to discourage unauthorised poaching of managers under contract with other clubs.
Most of the covariates in the match-level hazard function model are
similar to those in the season-level model, with the exception of the managerial eciency scores, for which no obvious match-level equivalent is
available, and which are therefore omitted. In their place, to capture the
short-term eect of match results on the job-departure hazard, the
match-level model includes the individual results of the last few league
matches as additional covariates.The full set of covariate denitions is as
follows:
Ri,t-k
Pi,t
Pi
C1i,t
C2i,t
Ai,t
298
Table 6.10 shows the crude estimates of the functions St , Ft , ft and ht for
the 981 managerial spells included in the match-level model. The
denitions of the survivor function St and the distribution function Ft are
the same as before (see section 6.3), but with t now denoting durations
in matches. The probability and hazard functions show total jobdeparture probabilities within duration bands 30 matches in width.With
durations measured in matches, the peaks in the hazard are located
around durations 180 and 360 matches.
Table 6.11 reports the estimates of the hazard function coecients in
the match-level model. Columns (1) and (2) show estimates of the involuntary departure hazard function. Column (1) includes all of the covariates; column (2) omits those with estimated coecients in column (1)
that are close to zero.The importance of current and recent match results
is reected in coecients on Ri,tk that are negative and signicant being
obtained for up to 16 matches prior to the current match. The absolute
magnitude of the eect diminishes quite regularly as the number of
lagged matches increases. The most recent matches therefore have a
greater eect on the hazard than the more distant ones, as expected. In
general, the coecients on Ri,tk conrm that the decision to terminate a
managers appointment involuntarily tends to be heavily inuenced by
very short-term considerations.
Elsewhere the results are mostly consistent with those for the seasonlevel involuntary departure hazard function.The change in league position
since the start of the managers spell is a highly signicant determinant of
the involuntary hazard. Dummy variables indicating the teams position in
the current seasons FA Cup competition did not produce signicant estimated coecients; nor did FA Cup dummies created using various alternative denitions.The failure to show that a managers job is more secure
before elimination from the cup and less secure after elimination may
reect diculties in disentangling the cup eect from other monthwithin-season variation in the hazard, captured by the dummies Xj,i,t.The
981
844
596
423
295
201
134
180
160
148
135
124
118
112
119
Duration in
matches, t
0
30
60
90
120
150
180
210
240
270
300
330
360
390
420
Spells that
survived
to duration t
124
228
166
117
184
164
146
119
112
112
111
116
115
113
Departures
between
durations
t and t30
13
20
17
11
10
13
18
11
10
11
10
10
11
10
Right-censored
spells, live at
duration t, and still
live at duration
t30
1.000
0.872
0.629
0.450
0.317
0.218
0.146
0.088
0.066
0.053
0.039
0.026
0.020
0.013
0.009
Survivor
function, St
0.000
0.128
0.371
0.550
0.683
0.782
0.854
0.912
0.934
0.947
0.961
0.974
0.980
0.987
0.991
Distribution
function, Ft
0.128
0.243
0.179
0.132
0.099
0.072
0.058
0.022
0.013
0.014
0.012
0.007
0.007
0.004
Hazard
function for
departure
between
durations
t and t30
0.128
0.279
0.285
0.294
0.311
0.331
0.398
0.249
0.200
0.270
0.314
0.250
0.333
0.307
Probability
function for
departure
between
durations
t and t30
Table 6.10. Distribution of complete and right-censored managerial spells, by duration (in matches) and crude estimates of
the survivor, distribution, probability and hazard functions
Ri,t13
Ri,t12
Ri,t11
Ri,t10
Ri,t9
Ri,t8
Ri,t7
Ri,t6
Ri,t5
Ri,t4
Ri,t3
Ri,t2
Ri,t1
Ri,t
Covariates
Inv.
(2)
1.05a
(0.10)
1.04a
(0.10)
0.493a
(0.091)
0.508a
(0.092)
0.674a
(0.093)
0.441a
(0.090)
0.604a
(0.092)
0.407a
(0.091)
0.350a
(0.090)
0.395a
(0.091)
0.202b
(0.090)
0.233a
(0.090)
0.217b
(0.090)
0.222b
(0.091)
Inv.
(1)
1.05a
(0.10)
1.03a
(0.10)
0.492a
(0.091)
0.503a
(0.092)
0.670a
(0.093)
0.440a
(0.090)
0.604a
(0.092)
0.405a
(0.091)
0.350a
(0.090)
0.396a
(0.091)
0.203b
(0.090)
0.235a
(0.090)
0.217b
(0.090)
0.217b
(0.091)
Mode of departure
0.166
(0.202)
0.024
(0.199)
0.396b
(0.200)
0.530a
(0.202)
0.147
(0.199)
0.160
(0.202)
0.028
(0.203)
0.075
(0.203)
0.054
(0.201)
0.172
(0.206)
0.022
(0.204)
0.396
(0.207)
0.321
(0.208)
0.211
(0.206)
Vol.
(3)
A2i,t
Mi,t
Si,t
Oi,t
Ii,t
Yi,t
X8,i,t
X9,i,t
X10,i,t
X11,i,t
Ai,t
C2i,t
C1i,t
(Pi,tPi ,t )
Covariates
0.137
(0.200)
0.013
(0.197)
0.376c
(0.198)
0.507b
(0.199)
Vol.
(4)
0.027a
(0.003)
0.169
(0.148)
0.158
(0.209)
0.222a
(0.061)
0.003a
(0.001)
0.001
(0.001)
0.006
(0.007)
0.251
(0.322)
0.064
(0.086)
0.022a
(0.005)
3.37a
(0.30)
2.96a
(0.22)
2.38a
(0.20)
2.12a
(0.18)
Inv.
(1)
Mode of departure
0.023a
(0.005)
3.51a
(0.27)
3.10a
(0.18)
2.53a
(0.14)
2.26a
(0.14)
0.221a
(0.060)
0.003a
(0.001)
0.028a
(0.003)
Inv.
(2)
Table 6.11. Involuntary and voluntary managerial job-departure hazard functions: duration in matches
0.018a
(0.006)
0.292
(0.353)
0.826
(0.517)
0.371b
(0.182)
0.004b
(0.002)
0.004c
(0.002)
0.033b
(0.016)
0.563
(0.570)
0.451b
(0.190)
0.014
(0.012)
5.04a
(0.79)
4.50a
(0.54)
3.78a
(0.45)
3.83a
(0.47)
Vol.
(3)
4.75a
(0.72)
4.26a
(0.43)
3.52a
(0.32)
3.63a
(0.36)
0.479b
(0.189)
0.387b
(0.178)
0.004b
(0.002)
0.003c
(0.002)
0.038b
(0.016)
0.021a
(0.006)
Vol.
(4)
765
0.207b
(0.091)
0.162c
(0.091)
0.231b
(0.091)
0.127
(0.091)
98,799
765
98,799
0.212b
(0.091)
0.168c
(0.091)
0.233b
(0.091)
153
0.369c
(0.210)
0.172
(0.209)
0.183
(0.204)
0.549a
(0.213)
98,799
X3,i,t
153
Key:
Ri,tk
Pi,t
Pi ,t
C1i,t
C2i,t
Ai,t
Mi
Si
Oi
Ii
Yi,t
Xj,i,t
2.45a
(0.17)
2.06a
(0.15)
2.32a
(0.15)
2.66a
(0.14)
2.98a
(0.15)
2.67a
(0.27)
result of the kth match before match t, dened for k 0 . . . 17 (1 win, 0.5 draw, 0 defeat)
league position going into match t
league position going into rst match of current managers spell
1 if teams FA Cup campaign is underway and incomplete, 0 otherwise
1 if teams FA Cup campaign has nished, 0 otherwise
age of manager (years)
managers previous managerial experience in English league football (months)
managers playing experience in English league football (seasons)
Scottish/overseas managerial experience dummy
international playing experience dummy
time trend (1 1973 season, 2 1974, and so on)
calendar month dummies.
X5,i,t
X4,i,t
X2,i,t
98,799
X1,i,t
X12,i,t
Notes:
a
signicantly dierent from zero, two-tail test, 1% level;
b
5% level;
c
10% level.
standard errors of estimated coecients are shown in parentheses.
departures
obs.
Ri,t17
Ri,t16
Ri,t15
Ri,t14
2.57a
(0.15)
2.12a
(0.14)
2.33a
(0.15)
2.66a
(0.14)
2.97a
(0.15)
2.65a
(0.27)
3.33a
(0.37)
3.25a
(0.35)
3.88a
(0.43)
4.24a
(0.40)
5.28a
(0.60)
3.89a
(0.72)
3.24a
(0.29)
3.29a
(0.34)
3.97a
(0.43)
4.28a
(0.40)
5.29a
(0.60)
3.93a
(0.72)
302
The highly signicant coecient on Ri,t17 in column (3) appears to be a statistical quirk,
and this covariate is not retained in column (4).
303
6.6 The eect of managerial change on team performance: evidence from North American team sports
Sections 6.66.8 investigate whether, on average, a change of manager
tends to lead to an improvement in team performance. In a comment on
Gruskys (1963) article (see section 6.2), Gamson and Scotch (1964)
identify a key issue that must be addressed by any empirical investigation
of this question, by means of an analogy.
If we compared average rainfall in the month preceding and the month
following the performance of the Hopi rain dance, we would nd more
rain in the period after. The dance is not performed unless there is a
drought, so such a comparison would be misleading. Nevertheless, this
slump-ending eect may help to account for the tenacity of belief in the
eectiveness of the ritual (Gamson and Scotch, 1964, p. 71).
Team performance, like rainfall and many other natural and man-made
phenomena, exhibits a tendency for mean-reversion or regression towards
the mean over time. Successful teams do not remain successful forever;
and teams that are unsuccessful eventually nd the wherewithal to
improve. This means that on average, the performance of the most successful teams at time t will tend to deteriorate at time t1, and the performance of the least successful teams will tend to improve. If teams that
change their manager are predominantly teams that are unsuccessful at
the time they make the change, some improvement in their average performance in the season following the change is to expected as a result of
the mean-reversion eect. This is the case even if the manager is just a
gurehead, who makes no contribution whatever to his teams performance. Any investigation of the relationship between managerial change
and subsequent team performance must therefore control for this meanreversion eect.
Gamson and Scotch attempt to control for mean-reversion in a relatively crude but straightforward manner. In their comparison of the win
ratios before and after 22 mid-season managerial changes in Major
League Baseball (MLB) which took place between 1954 and 1961,
match results during the two weeks prior to each change are excluded
from the calculations. In 13 of the 22 cases, the team performed better
under the new manager than it did up to two weeks before the removal of
the old manager.This nding could be consistent with either the common
sense or the ritual scapegoating theories of managerial succession (see
section 6.2), but it does not support the vicious circle theory, which
would anticipate deterioration in post-succession performance. Of
course, the assumption that the omission of two weeks results is sucient
to exclude the mean-reversion eect is rather arbitrary.
304
305
306
307
data. The managerial departure dummies distinguish between withinseason and close season departures and between involuntary and voluntary departures (as before). Measuring performance in the season
following succession oers the advantage that the dependent variable is
not contaminated by any match results which accrued during the latter
stages of the outgoing managers spell in charge. A price is paid, however,
for this benet: performance immediately following a within-season
departure (until the end of the same season) is excluded from the
denition of the dependent variable.
Both ordinary least squares (OLS) and xed-eects (FE) panel estimates of the season-level succession eects model are reported in table
6.12.The FE estimator allows for the presence of club-specic inuences
on performance, referred to more generally as individual eects. These
are modelled using a full set of club-specic dummy variables; their estimated coecients are not reported in table 6.12. Columns (1) and (3)
show the OLS and FE estimates of a model that uses just one dummy variable for all four types of managerial change, obtained by summing the
four dummies dened above
2
(w
j
j
ci,t1
)
i,t1
j1
Columns (2) and (4) show the results for the model that includes the four
dummies as separate covariates.12
The coecients on the lagged managerial change dummy in columns
(1) and (3) of table 6.12 are both positive and signicant at the 1 per cent
level.When this dummy variable is disaggregated into its four constituent
components in columns (2) and (4), the coecients on the involuntary
departure dummies are positive (involuntary departure leads to an
improvement in the following season), but only the coecient on the
within-season involuntary departure dummy is statistically signicant.
Conversely the coecients on the voluntary departure dummies are negative (voluntary departure leads to a decline in the following season). In
this case, the coecient on the close season voluntary departure dummy
is signicant at the 5 per cent level in the OLS estimation, but not in the
(preferred) FE estimation. The fact that the coecients on the involuntary departure dummies are positive and those on the voluntary dummies
negative, does at least tend to support the common sense theory of
12
Although the coecients on the promotion and relegation dummies are quite similar in
both sets of estimations, the estimated coecients on lagged league position are considerably lower in the FE models than in the OLS versions. This reects the fact that this
coecient is upward biased if OLS is used on a panel data set when individual eects are
in fact present.
308
Covariates:
Constant
Pi,t1
pri,t1
rei,t1
2
(w
j1
w1i,t1
j
j
c i,t1
)
i,t1
OLS
(1)
1.465a
(0.336)
0.0434a
(0.0055)
12.87a
(0.47)
12.26a
(0.47)
1.111a
(0.307)
c1i,t1
w2i,t1
c2i,t1
obs
R2
2,238
0.43
OLS
(2)
1.250a
(0.337)
0.0392a
(0.0055)
13.09a
(0.47)
12.62a
(0.47)
2.029a
(0.364)
0.770
(0.567)
0.617
(0.749)
1.748b
(0.821)
2,238
0.44
FE
(3)
FE
(4)
0.1853a
(0.0103)
12.38a
(0.46)
11.97a
(0.46)
0.861a
(0.302)
2,238
0.49
0.1789a
(0.0104)
12.53a
(0.47)
12.22a
(0.46)
1.553a
(0.361)
0.299
(0.559)
0.178
(0.737)
0.868
(0.806)
2,238
0.50
Note:
a
signicantly dierent from zero, two-tail test, 1% level;
b
5% level.
Standard errors of estimated coecients are shown in parentheses.
Dependent variable is Pi,t Pi,tPi,t1.
Key
Pi,t1
pri,t1
rei,t1
w1i,t1
c1i,t1
w2i,t1
c2i,t1
309
310
For consistency with the notation used in chapter 3, the t-subscript for match t is suppressed.
311
20
f Z
k
k i,j
(6.4)
k1
Table 6.13 reports the estimated version of (6.4), obtained using all
match observations for which complete data are available, between the
1973 and 1999 seasons inclusive. Apart from minor numerical variations,
the estimated coecients and standard errors on all covariates that were
included in the previous version of this model are virtually unchanged
(see table 3.14, columns (1), (2), (5) and (6)).The estimates of the additional coecients on the manager change dummy variables, Zki,j , and their
standard errors, are reported in columns (5) and (6) of table 6.13. The
estimated coecient for the match immediately following a change of
manager is negative and signicant at the 1 per cent level. This suggests
that a within-season managerial change is disruptive in the very short
term: a team that changes its manager tends to under-perform, to an
extent that is discernible on conventional statistical criteria, in the match
after the change takes place.
Among the other results in column (5), only the negative coecient on
is signicant at the 5 per cent level.While there is a mix of positive and
Z 11
i,j
negative estimated coecients, overall the preponderance is negative. It
can be argued, however, the impact of a managerial change on any individual result is of less interest than the cumulative impact on results over
1
P j,0,1
0
P j,0,1
1
P j,0,1
1
P j,1,1
0
P j,1,1
1
P j,1,1
2
P j,1,2
1
P j,1,2
1
P i,1,2
2
P i,1,2
0
P j,0,0
2
P i,1,2
1
P i,1,2
0
P i,1,2
P 0i,0,0
1
P i,0,1
0
P i,0,1
1
P i,0,1
1
P i,1,1
0
P i,1,1
1
P i,1,1
0.912
0.618a
0.857a
0.612a
0.472a
0.405
0.718a
0.115
1.278a
1.329a
0.712
0.401a
0.313a
1.545a
1.525a
1.129a
0.832a
0.601a
0.546a
0.461a
0.165
(1)
Coe.
0.177
0.099
0.091
0.154
0.096
0.086
0.354
0.142
0.080
0.158
0.110
0.111
0.179
0.100
0.091
0.154
0.096
0.085
0.350
0.142
0.094
(2)
s.e.
RH
j,1
RH
j,2
RH
j,3
RH
j,4
R Aj,1
R Aj,2
R Aj,3
R Aj,4
R Ai,2
R Ai,3
R Ai,4
RH
i,8
RH
i,9
R Ai,1
RH
i,1
RH
i,2
RH
i,3
RH
i,4
RH
i,5
RH
i,6
RH
i,7
0.013
0.013
0.013
0.013
0.013
0.013
0.013
0.013
0.013
0.013
0.013
0.031b
0.031b
0.031b
0.036a
0.044a
0.045a
0.043a
0.046a
0.013
0.013
0.013
0.013
0.013
0.013
0.013
0.013
0.013
0.013
(4)
s.e.
0.037a
0.047a
0.006
0.065a
0.031b
0.074a
0.008
0.018
0.002
0.015
0.027b
0.042a
0.019
(3)
Coe.
Z 14
i,j
Z 15
i,j
Z 16
i,j
Z 17
i,j
Z 18
i,j
Z 19
i,j
Z 20
i,j
Z 11
i,j
Z 12
i,j
Z 13
i,j
Z 8i,j
Z 9i,j
Z 10
i,j
Z 1i,j
Z 2i,j
Z 3i,j
Z 4i,j
Z 5i,j
Z 6i,j
Z 7i,j
0.048
0.048
0.049
0.049
0.050
0.051
0.039
0.016
0.010
0.105b
0.002
0.031
0.052
0.053
0.054
0.055
0.056
0.057
0.059
0.045
0.045
0.045
0.046
0.046
0.047
0.047
0.122a
0.051
0.023
0.015
0.032
0.072
0.038
0.070
0.049
0.008
0.041
0.021
0.015
0.048
(6)
s.e.
(5)
Coe.
Table 6.13. Relationship between managerial succession and performance, measured by match
0.316
0.365c
0.373c
0.332
0.352
0.338
0.290
0.353b
0.355b
0.386b
0.253c
0.237
0.248
0.122a
0.173a
0.150c
0.135
0.104
0.176
0.214c
(7)
Cum. coe.
0.193
0.202
0.212
0.221
0.230
0.240
0.250
0.165
0.175
0.184
0.136
0.146
0.156
0.045
0.064
0.079
0.092
0.105
0.116
0.126
(8)
s.e.
0.468a
0.282a
0.412a
0.148a
0.114a
0.631
0.143
53,820
0.161
0.027
0.027
0.055
0.055
0.080
0.093
R Aj,7
R Aj,8
R Aj,9
R Aj,6
R Aj,5
0.013
0.013
0.013
0.013
0.013
0.028b
0.037a
0.030b
0.030b
0.020
Key:
Z ki,j D ki D kj
D ki 1 if home team i changed manager between the kth and k1th matches before the current match (within the current season)
for k 1 . . . 20, 0 otherwise
Djk is the same for away team j
Key to all other variables is the same as in table 3.14.
Notes:
a
signicantly dierent from zero, two-tail test, 1% level;
b
5% level;
c
10% level.
Estimation period is 197399
2
P j,1,2
SH
i, j
Si,Aj
1
2
obs.
1
P j,1,2
P 0j,1,2
314
315
6.26.5 have investigated the causes of managerial departure, and sections 6.66.8 have examined succession eects, of managerial change on
subsequent team performance. The extent to which poor team performance tends to trigger managerial change has been examined using both
season-level and match-level data. In the season-level estimations, the
likelihood of involuntary departure is higher for inecient managers; is
lower if there has been an improvement in the teams league position since
the manager was appointed; and is lower if the team made progress in the
FA Cup. The involuntary departure hazard has increased over time. The
voluntary departure hazard tends to be higher for ecient managers and
managers who have improved their teams league position. Experienced
managers are also found to be more likely to depart voluntarily.
The virtue of estimating the job-departure hazard with match-level
data is that it permits identication of short-term inuences on the departure decision, which tend to be obscured when durations are measured by
season. A statistically signicant link between individual match results
and the involuntary departure hazard is found for up to 16 matches prior
to the current match.The involuntary departure hazard decreases with an
improvement in league position; decreases with the age of the manager;
but does not seem to depend on the managers other human capital
attributes. The few most recent match results are also signicant in the
voluntary departure hazard function, but the eect is weaker. Perhaps
surprisingly, a run of poor results is more likely to trigger a voluntary
departure than a run of good results; this may be because managers are
more inclined to accept alternative employment oers when their current
team is struggling than when it is doing well. Managers that have
improved their teams position are more likely to leave voluntarily, and
previous managerial experience and international recognition as a player
both increase the likelihood of voluntary departure.
The eect of a change of manager on team performance after the change
takes place has also been considered in chapter 6. In order to measure
managerial succession eects correctly, it is necessary to control for a
natural tendency for mean-reversion in team performance. No team
carries on losing forever, so a team that has recently experienced a poor
spell should be expected to improve, whether it changes its manager or not.
To measure the average eect of a change of manager on post-succession
performance, this natural tendency to improve should not be attributed to
the decision to terminate the previous managers appointment. Only if
teams tend to improve by more than the expected amount is it correct to
infer that removing the manager has a benecial average eect.
In estimations of the managerial succession eect using match-level
data, it is found that a within-season managerial departure tends to be
316
disruptive in the short term. On average, and after controlling for the
mean-reversion eect, a team that changes its manager tends to underperform in its next few matches. In the longer term, however, estimations
using season-level data suggest that involuntary within-season managerial departure tends to lead to an improvement in performance in the following season.The improvement is greater than would be expected given
the mean-reversion eect. Ironically these results suggest that club
owners, who are often accused of short-termism and myopia when taking
the decision to terminate an under-performing managers appointment,
may actually be incurring short-term pain in order to realise long-term
gains by so doing.
A number of broad trends in league match attendances in English football at the aggregate level have been identied in chapter 2. During the
post-Second World War boom, league attendances surged, reaching an
all-time high of 41.0 million in the 1949 season.The boom, however, was
relatively short-lived. It was followed by a period of sustained decline that
continued, almost uninterrupted, until the 1986 season, when attendances fell to 16.5 million. The period since the late 1980s has seen a
gradual but signicant improvement. By the 1999 season, total attendances had risen to 25.4 million. The recent growth in attendance actually under-states the growth in demand because of ground-capacity
constraints, with many leading clubs currently experiencing demand for
tickets far in excess of existing capacities.
Chapter 2 has also reviewed the academic debate about the causes of
the long post-war decline in football attendances, and its recent reversal.
Social and demographic change, increasing material auence, the option
to watch football on television rather than in person, crowd misbehaviour, the deteriorating physical state of many of footballs stadia and the
dubious quality of the some of the entertainment on oer on the eld of
play are among the factors considered to have contributed to the decline
in the popularity of attending live football. More recently, improved facilities in all-seated stadia, together with the near-eradication of the hooligan problem, have helped strengthen footballs appeal as a middle-class
spectator sport. In addition, a sport so heavily steeped in history and tradition could hardly have failed to benet from the 1990s heritage fad.
Improved standards of entertainment on the eld of play, boosted by a
large inux of talented overseas players, have also contributed much to
footballs recent popular resurgence.
Social history and sociology provide many useful insights into the
causes of uctuating football attendances. In section 7.1, however, an
alternative approach to the analysis of attendances is pursued.The statistical or econometric analysis of attendances at the micro level forms the
subject matter. Statistical modelling of variations in the attendances of
317
318
319
from academics since the 1970s. Hart, Hutton and Sharot (1975) published the rst econometric analysis of patterns of attendance at English
football matches. At around the same time, Demmert (1973) published
an investigation of US baseball attendances, while Noll (1974) compared
the determinants of attendances at four US professional team sports:
baseball, basketball, football and hockey. Since the appearance of these
early papers, the topic has continued to receive the attentions of econometricians on a regular basis. The purpose of this section is to review critically this literature, focusing primarily on studies that have analysed
attendances at English and Scottish football, but also drawing insights
from others where applicable. Cairns (1990) has written a comprehensive
review of the literature from the 1970s and 1980s, and parts of this
section draw on his ndings.
Before embarking on the review, it is worthwhile to identify a useful
way of classifying attendance studies. All previous empirical analyses of
attendance fall into one of two categories: rst, those which model individual match attendances; and second, those which model entire season
or annual attendances (either for the league as a whole or for individual
clubs). As will be seen, both types of study provide useful insights, though
the types of inuence on attendance that they are best equipped to identify dier in each case. The rst sub-section of section 7.1 discusses the
use of attendance data as the dependent variable in models of spectator
demand for attendance. The following four sub-sections then describe
the explanatory or independent variables that have been used most commonly in match and season attendance models.
Measuring the demand for football attendance
Attendance data, usually announced by the home club while the match is
in progress, published in the following days newspapers and subsequently collected in annual statistical publications such as Rothmans
FootballYearbook, provide the main source of data for the dependent variable in econometric models of the demand for football attendance. In
general, academic researchers have not hesitated to exploit such voluminous and easily accessible data sets, while acknowledging that some
diculties can arise when interpreting attendance data as a measure of
spectator demand.
Football attendance data is usually aggregated over two types of spectator: season ticket holders, who pay for an entire seasons admission in
advance; and purchasers of tickets for individual matches, who may have
committed to attend up to one month in advance, or as late as the day of
the match itself. Clearly the factors that inuence the decisions of
members of either group to attend will tend to dier. Once a season ticket
320
has been purchased, the only costs arising from the decision to attend an
individual match relate to time and transport. The disincentives for
season ticket holders to attend are therefore small, unless the home
teams performance is so bad that it becomes preferable to stay at home!
Match ticket purchasers on the other hand, are likely to be more selective
in their choice of xtures and more sensitive to temporary uctuations in
recent home-team performance or the quality of the visiting team. It is
regrettable that most match attendance data does not distinguish
between the two groups. Becker and Suls (1983) estimate separate
regressions for match attendances and season ticket sales for Major
League Baseball (MLB) matches played during the 1970s.Their ndings
for the relationship between team performance and demand are quite
similar in both cases. For football the annual attendance data set used by
Simmons (1996) also distinguishes between season ticket and match
ticket sales.
Published match attendance data also includes groups of ticket holders
who may have paid dierent prices for seats of varying quality. Before the
advent of all-seated stadia, match attendance data was aggregated over
both seated and standing spectators. Again, the nature of the aggregation
tends to restrict or preclude the investigation of a range of interesting
issues, including measurement of the price and other elasticities of
demand for tickets in dierent price categories, or in dierent parts of the
ground. Using a survey data set which disaggregates attendances into a
standing and seated component, Dobson and Goddard (1992) nd some
evidence that the attendance of standing spectators was more variable
than that of their seated counterparts. Standing attendances were particularly sensitive to the recent performance of the home team and the
extent to which the match had signicance for championship issues. The
data set used in this study was quite small, however, and with the subsequent elimination of terracing from the leagues higher echelons, the
results are now more of historical than practical interest.
Most studies analyse attendance by estimating single-equation regression models, and interpreting the resulting equations as demand functions. This is acceptable if it can be assumed that supply is perfectly
elastic. If not, estimation of the coecients of the demand equation may
be problematic. Suppose, for example, attendance depends on admission
price, but clubs adjust admission price in the light of the previous
seasons attendance. Current admission price is therefore correlated with
whatever random factors inuenced last seasons attendance. If the
eect of such random factors tends to persist over more than one
season, current admission price is also correlated with the current
random element in attendance.This violates one of the key assumptions
321
Namely, the assumption that the error term is uncorrelated with any of the independent
variables.
322
1979
1984
1989
1994
1999
Average attendances
D1/Prem
28,294
D2/One
13,693
D3/Two
16,198
D4/Three
13,925
G1
29,325
G2
18,554
G3
10,372
G4
18,736
G5
13,713
All
12,681
27,428
13,303
16,113
14,182
30,049
16,297
19,636
18,298
14,693
12,426
18,834
11,601
14,939
12,817
22,855
13,020
17,097
15,777
13,258
19,301
20,553
10,558
15,499
13,245
22,804
13,055
17,290
16,314
13,504
19,503
23,055
11,738
15,392
13,423
26,550
15,425
17,750
17,520
13,559
10,800
30,580
13,605
17,510
13,794
31,901
19,763
18,652
12,160
13,679
13,146
44,483
37,559
28,629
24,862
50,594
40,698
29,092
33,269
24,052
33,573
37,491
32,089
25,583
20,520
47,745
31,684
24,991
26,005
20,767
28,665
35,388
27,930
19,664
13,910
45,343
28,390
20,642
19,494
13,044
23,738
28,157
20,948
13,716
19,047
32,808
23,049
14,837
16,448
19,609
17,939
33,573
21,123
14,202
10,479
36,762
26,166
15,072
18,761
19,580
19,247
61.7
35.4
21.4
16.8
59.4
40.0
33.1
24.9
19.5
37.0
50.2
36.2
19.3
13.7
47.9
41.1
28.4
22.2
15.7
32.4
58.1
37.8
28.0
23.3
50.3
46.0
35.3
32.4
26.9
40.0
81.9
56.0
39.3
37.8
80.9
66.9
52.2
45.7
37.0
60.2
91.1
64.4
52.9
36.2
86.8
75.5
57.4
64.8
38.4
68.3
323
324
management philosophy which was ultimately to lead to the most catastrophic consequences.
Today, as table 7.1 suggests, many of the top clubs sell out regularly
for all but the most insignicant matches, and attendance data can no
longer be interpreted as a meaningful measure of demand. Although estimation techniques such as tobit allow valid estimation in cases where the
distribution of the dependent variable is constrained or truncated, a data
set containing a reasonable mix of constrained and unconstrained attendance gures is still required. At present, the information content in a
recent seasons array of home-match attendances for clubs such as
Arsenal, Manchester United and Newcastle United is precisely zero
(unless one merely wishes to discover what the ground-capacity constraint is). While this situation continues, it is clearly not possible to estimate attendance models of the type reviewed in this section using current
data for capacity-constrained clubs.
Demographic and geographic determinants of attendance
A link between the size of the market from which each club draws its
support and the level of attendance at the clubs matches seems to be both
theoretically and intuitively obvious. Empirically, however, it is less
straightforward either to dene each clubs catchment area, or to come up
with a market-size measure that is not arbitrary to some extent. Dening
the size of the market is made dicult by the progressive erosion over
time of the geographical segmentation of supporters, especially of the
leading clubs. Manchester United, for example, are now famous (or notorious) for drawing their support from all regions of the UK, and even
from beyond.
Even if markets could be segmented geographically, measurement is
made dicult by the arbitrary nature of the municipal boundaries used to
compile most local population statistics, and by the need to make adjustments in cases where two or more clubs are situated in close proximity.
Despite these diculties, in principle it should be possible to identify
empirically a link between local population and attendance by comparing
a cross-section of clubs. It may be more dicult to do so for a single club
over time, since local populations change perhaps far more slowly than
other factors which inuence a clubs attendance. In fact, many (but not
all) match-attendance studies have included cross-sectional market-size
measures for either or both of the home and away teams among their
explanatory variables. Where such measures are used, almost invariably
they produce regression coecients which are of the expected sign and
statistically signicant.
325
326
327
increase with average earnings whenever the latter are above about 400
per week. The signs of the estimated coecients on price, however,
suggest a positive relationship between price and attendance over a realistic range of prices, so the model fails to identify an economically meaningful price eect.
Annual attendance models, using data with a longer time dimension,
have been more successful in identifying relationships between price,
income and unemployment variables and football attendances. Birds
(1982) study, which reports equations for attendance aggregated across
the entire league and by division, was the rst of its kind for English football, and is perhaps still the most widely quoted empirical attendance
study. In an attempt to capture the full cost of attending, Birds price
measure combines the leaguess minimum admission price with an
implicit price index for motoring and other transport costs, taken from
the UKs national income accounts. Total consumer expenditure is used
as an income measure.The results suggest a price elasticity of demand for
football attendance of 0.22, and an income elasticity of 0.62.
Birds negative income elasticity implies that attending football is an
inferior good. As incomes increase, consumers tend to give up attending
football, and follow other, perhaps more expensive or up-market leisure
activities instead. As a description of what actually occurred between the
1950s and 1970s, this seems plausible and consistent with the historical
account given in chapter 2. However, developments in time-series econometrics since Birds paper was published raise doubts about the statistical
validity of some of the relationships which he identies. Specically,
although it is correct to say that incomes rose and football attendances fell
between the 1950s and the 1970s, nothing more than this can be inferred
from a negative and signicant coecient obtained in a crude regression
of attendance on income. It is not possible to infer that the rise in income
caused or even inuenced the decline in attendance in any sense. All
that can really be said is that income and attendance were trended in
opposite directions during the period in question.
Szymanski and Smith (1997) estimate attendance and revenue equations as part of a larger model, which attempts to capture both demandand supply-side inuences on the trade-o between team performance
and prot faced by club owners. In an attendance estimation pooled over
48 clubs using data for the period 197489, an estimated price elasticity
of 0.76 is obtained. When a set of time dummy variables is included,
this estimate drops to 0.34. No income or unemployment measure is
included among the explanatory variables. In this paper, the estimation
period is too short for the full battery of modern time-series econometric
techniques to be deployed, and in any case the attendance equation is not
328
A time series is stationary if its mean and variance are constant with respect to time, and
non-stationary if its mean and variance are changing with respect to time. An untrended
series can be stationary or non-stationary: a non-stationary untrended series is one that
tends to wander over time. A trended series is, by denition, non-stationary because the
trend causes its mean to change with respect to time. Some trended series, however, can
be made stationary by de-trending. Other trended series remain non-stationary even
after any deterministic trend component is removed.
329
330
Hart, Hutton and Sharot (1975) use the current league positions of the
home and away teams as explanatory variables, and nd that while highly
placed away teams attract higher attendances, the home-position
coecients are generally insignicant. The latter nding may reect the
fact that after the rst few weeks of the season (during which spectators
may in any case tend to disregard the embryonic league table), there is
often little further variation in the home-position variable in a clubspecic equation until (perhaps) the start of the next season.
By estimating club-specic equations over a longer period, thereby
obtaining greater variation into the home-position variable, Cairns
(1987) obtains signicant coecients on both the home- and awayposition variables, and by pooling their data across clubs Walker (1986),
Peel and Thomas (1988) and Dobson and Goddard (1992) nd the
same. Baimbridge, Cameron and Dawson (1996) nd linear and quadratic terms in the absolute dierence between the positions of the two
teams to be insignicant, possibly because various other team-quality
measures (such as the previous seasons nishing position, and the
number of international and overseas players in each team) are also
included. In addition to league position variables, Walker (1986), Cairns
(1987) and Dobson and Goddard (1992) include the number of points
gained from the three, four or ve most recent matches played by either or
both teams. A recent run of good results is usually found to have a
signicant and positive eect on attendance.
Incorporating team-quality information into a match-attendance
model naturally raises questions about the link between uncertainty of
outcome and attendance (see also chapter 3). Hart, Hutton and Sharot
(1975) and Peel and Thomas (1988, 1992) focus on uncertainty of individual match outcomes as a determinant of attendance, while Jennett
(1984) and Cairns (1987) are concerned with uncertainty of season outcomes: the impact on attendance of the importance (if any) of the match
for championship, promotion or relegation issues.
Hart, Hutton and Sharot try including the absolute value of the
dierence between home and away positions as an explanatory variable in
their attendance equation, but fail to obtain the negative coecient that
would suggest that the more evenly balanced the teams and uncertain the
match outcome, the higher the attendance. This procedure ignores the
inuence of home advantage on outcome uncertainty. Peel and Thomas
(1988) use home-win probabilities calculated from bookmakers posted
betting odds, and nd that attendance is positively related to the probability of a home win. However, by incorporating this variable in linear
(rather than quadratic) form, they eectively measure relative team
quality and not uncertainty of outcome directly. Matches at opposite ends
331
of the home-win probability scale (at 0.05 and 0.95, for example) should
rate as roughly equal (i.e. very low) in terms of uncertainty of outcome.
Nevertheless, from their results Peel and Thomas argue that measures to
increase uncertainty of outcome might have a detrimental eect on attendance, if they reduce home-win probabilities.They therefore question the
empirical validity of the hypothesis that uncertainty of outcome has a positive eect on attendance.
Jennett (1984) constructs dummy variables to indicate the degree of
championship signicance of each match for the home and away teams.
While it remains mathematically possible for a team to achieve the points
total observed (after the event) to be required to win the championship,
the dummy is set equal to the reciprocal of the number of games that
remain to be played. The dummy therefore increases progressively for
teams that remain in contention as the prospect of winning the championship draws near, reecting the heightened signicance of the nal few
matches for such teams. For teams that drop out of contention after
reaching a point from which the points required are no longer attainable,
the dummy is re-set to zero for the seasons remaining matches. In the
estimations, the championship signicance dummies are found to be
highly signicant for both the home and away teams, though similarly
designed relegation signicance dummies are insignicant.
Jennetts championship signicance variable has been widely, and
perhaps excessively, criticised (Cairns, Jennett and Sloane, 1986; Cairns,
1987; Peel and Thomas, 1988) on the grounds that its computation
requires knowledge of the ultimate championship-winning points total.
Of course, such knowledge is unavailable to spectators at the time they
take their decisions whether or not to attend. The Jennett measure does
nevertheless succeed in distinguishing quite accurately between matches
that (on any reasonable criteria) would be regarded as having championship signicance, and those that would not. If the purpose is to model
attendances ex post, rather than forecast them ex ante, Jennetts measure
achieves its objective. Presumably, data on changes throughout the season
in bookmakers odds on the championship outcome would provide a
more widely acceptable measure, although to our knowledge no researchers have exploited this possibility as yet. Instead, Cairns (1987) denes a
team to be in contention if it could still win the championship by gaining
80 per cent of the points available from its remaining matches, assuming
that the current leader gains only 50 per cent. While the percentages are
arbitrary, Cairns also demonstrates that matches with championship
signicance on this measure attract signicantly higher attendances.
On the whole, the issues involved in incorporating team-quality measures into a season attendance model are relatively straightforward. The
332
333
variations in the weather on attendance, but generally do not nd evidence of any eect. Other miscellaneous issues that have been investigated in attendance studies include the eect of league restructuring on
attendance; the eect (if any) of team racial composition on attendance;
and the impact on attendance of simultaneous live broadcasts on satellite television.
Cairns (1987) investigates the eects of the change from a two- to a
three-division league structure in Scotland from the 1976 season
onwards, by including dummy variables to test for changes in the
coecients of the attendance model after the restructuring took place.
Recent match results seem to have become a more signicant determinant of attendance than they were before restructuring, and in a smaller
Premier Division, a higher proportion of matches appear to have been
signicant in inuencing championship outcomes. The coecients on
dummies for matches against Scotlands two big-market teams (Celtic
and Rangers) became smaller after the change, perhaps because in the
new 10-club Premier Division teams met four times rather than twice per
season, reducing the importance of these matches for supporters of other
teams. Overall the downward trend in attendance became less pronounced after the change. Cairns judges the introduction of the Premier
Division to have been a success in terms of its impact on attendance.3
Burkitt and Cameron (1992) estimate a season-attendance model for
English rugby league clubs over the period 196690.The period includes
a major restructuring of the league from a single- to a two-division format
in 1973. A dummy variable is used to identify the eect of the split on
each clubs attendance. Attendances increased overall, but while teams
that moved into the higher division beneted, teams that moved into the
lower division suered as a result of the split.
The English Rugby Football League underwent several changes of
league structure during the 1990s. Rugby league therefore oers promising territory in which to investigate the relationship between league structure and match attendance. Dobson, Goddard and Wilson (2001)
estimate an empirical match-attendance model using data for the period
19905. Subsequently Monte Carlo methods are used to generate simulated match results data under alternative league structure assumptions.
The simulated data are used to generate simulated values for the independent variables of the attendance model, and simulated attendances under
3
This seems to be borne out by the fact that during 25 years of further, sometimes quite
intense argument (when membership of the Premier Division has been adjusted up and
down between 10 and 12 clubs on several occasions) on no occasion have proponents of
a return to anything similar to the pre-1975 two-division structure been able to carry the
debate.
334
335
336
From the start FC Barcelona had strong links with local middle-class
high schools, and was an active participant in demonstrations of
Catalonian identity organised by local cultural organisations. In contrast,
its ercest local rival RSC Espaol reacted against the middle-class and
separatist tendencies of Barcelona, and cultivated a consciously Castilian
identity. Espaols support included a strong working-class element,
while originally the club relied mainly on Catalan and Spanish players
rather than foreigners. The clubs allegiance to the unied Spanish state
was formalised when King Alfonso XIII conferred upon it the Royal
Sports Club (RSC) appellation in 1912 (Colom, 1997).
Catalonia prospered economically from Spains neutrality in the First
World War, and several of FC Barcelonas manufacturing patrons were
prominent exporters of manufactures to several of the combatants. By the
1920s, footballs popularity had spread to the extent that the class composition of the clubs support was becoming more broadly based. Following
the voluntary abdication of Alphonso XIII, however, the club became a
focal point for local opposition to the military government of General
Primo de Rivera (192330). In public demonstrations against the de
Rivera dictatorship, the blue-and-scarlet Barcelona club ag was substituted for the banned red-and-gold Catalan national ag. In an incident
during a friendly match against local rivals Jupiter in 1924, the crowd at
the clubs Les Corts stadium jeered while a British naval band played the
Spanish national anthem, provoking the government to temporarily close
Les Corts and ban the club from operating (Duke and Crolley, 1996;
Burns, 1999).
During the Spanish Civil War (19369) FC Barcelona inevitably
became identied with the forces of republicanism and separatism,
while RSC Espaol symbolised sympathy with Francos Falangist
forces. In 1936 Barcelona club president Josep Sunyol was detained by
soldiers at a Falangist checkpoint near Guadarrama and summarily executed. The episode received little attention in Spain during the Franco
years, but is described in detail by Burns (1999). In 1937 Barcelona
embarked on a tour of central and North America, partly in an attempt
to solicit political support. But with the Falangists slowly gaining ground
in the Civil War, few players elected to return home when the tour concluded.
Under the dictatorship of General Francisco Franco (1939-75),
symbols of Catalan nationalism including the language and national ag
were again suppressed. Meanwhile the football clubs name FC Barcelona
was changed to the Castilian equivalent Club de Futbol Barcelona. A new
board of management comprising pro-Franco ocials was installed, and
no one suspected of harbouring pro-Catalan sympathies was permitted to
337
338
independence (Duke and Crolley, 1996). Catalonia now boasts its own
Parliament and president, and wields considerable autonomy over a wide
range of matters such as policing, welfare and taxation.
The strength of the shift towards Catalan regional autonomy was
reected symbolically in football in 1995, when Barcelonas local rivals
also changed their name, from the Castilian Espaol to the Catalan
Espaynol. But the club has never been able to escape from the
shadows of its most illustrious neighbour: average league attendances
in the 2000 season of 65,500 (Barcelona) and 18,900 (Espaynol)
reect accurately the gap in popularity between the two clubs. While
most of todays Espaynol supporters were born and raised in
Catalonia, locals and migrants alike who are unattracted or repelled by
the symbolism of the Barcelona juggernaut are just as likely to ally
themselves with Real Madrid as with the local poor relations. In 1997
Espaynols plight was symbolised by the sale of their own stadium to
repay debts; since then the club have been tenants in the city of
Barcelonas Olympic Stadium.
Today, any local animosity between Barcelona and Espaynol pales by
comparison with the strength of the BarcelonaReal Madrid rivalry, certainly one of the most intense anywhere in world football. But associations with the political schism of the Franco era are now mainly stylised
and symbolic. Twenty-ve years after Francos death, Real can no longer
be described as the regimes team, while most Catalonian residents take
pride in both a Catalan and a Spanish identity. Nevertheless, the symbolism surrounding the BarcelonaReal Madrid rivalry still lives on in
some quarters: for example, in the antics of Reals Ultras Sur who continue to display the imperial ag of Francos Spain, and of Barcelonas
Boixos Nois who enjoy burning the ag of democratic Spain (King,
2000). Few regular supporters of either club make the journey to watch
away xtures involving the two clubs, and even club ocials often tend
not to travel.
On the eld of play, Barcelona has always operated a policy of employing foreign players, but this tendency has never been more evident than
during the post-Bosman era. And the same now applies equally to Real
Madrid. In playing terms the CatalanCastilian dichotomy has therefore
become mainly irrelevant in determining the fortunes of the two clubs.
Nevertheless, Barcelonas pro-Catalan associations remain a powerful
part of its mystique and mythology, as do Athletic Club de Bilbaos
Basque ailations, and Real Madrids and Espaynols traditional centrist
leanings. Clearly political, nationalist and cultural sentiment has played a
fundamental part in shaping the present-day support that all of these
clubs attract.
339
340
In 1887 the decision was taken to form a new catholic football club as a
vehicle for raising money for the poor of Glasgows east-end. Brother
Walfrid, a member of the catholic Marist Order, is usually credited as the
founder of Celtic. The clubs stadium was opened in May 1888, and its
rst match took place three weeks later, against Rangers. Celtics subsequent meteoric rise undoubtedly helped transform Scottish football, both
on and o the pitch.The club was a prominent advocate of the new league
competition, launched in the 1891 season, as well as the subsequent drive
for the recognition of professionalism, eventually achieved in 1893. Celtic
were cup runners-up in 1889 and winners for the rst time in 1892.
Further cup success proved elusive until 1899, but as four times league
winners between 1893 and 1898, Celtic was the dominant force in
Scottish football throughout the mid-1890s.
By insisting upon their retention of amateur status, Queens Park
appear to have sealed their own decline as a major competitive force. It
therefore fell to Rangers to assume the leading role as Scottish challengers to what was perceived by many as alien Irish competitive dominance.
Rangers rst cup victory came in 1894, but the club had to wait until
1899 for the league championship. Realising that erce partisanship was
also good box oce, the management of both Celtic and Rangers seem to
have been responsible for fanning the ames of hostility to some extent
during the 1890s. Footballs developing commercial potential was
reected in Celtics decision in 1897 to abandon charitable status and
form a limited company. Rangers followed suit two years later. In 1904 a
cartoonist satirising the two clubs competitive and commercial dominance may have been the rst to coin the enduring Old Firm epithet
(Murray, 1984).
As the intensity of rivalry between the clubs increased during the
1890s, Rangers positive protestant identity inevitably coalesced with
more negative anti-catholic and anti-Irish sentiment in the protestant
community. Class and social prejudices also tended to reinforce religious
and ethnic divisions, as a direct consequence of the inferior economic
status of the migrant community. While most of Rangers early directors
and shareholders were managers and professionals, many of their Celtic
counterparts were publicans. Many ordinary Rangers supporters were
foremen and skilled workers, while Celtic attracted large numbers of
unskilled manual labourers.
By the turn of the century all the ingredients were in place for a century
of often bitter, sectarian strife between the two clubs and their supporters.
Specic actions of the clubs which became enduring bones of contention
included Rangers undeclared but rigorously enforced ban on employing
catholic players, management and sta; Celtic, by contrast, have always
341
342
With more data now available for the post-Second World War period, data for the 1926 to
1939 seasons inclusive (used in the earlier study) has been dropped from the estimations
reported here.
343
344
Census data for local authority districts are available only at ten-year intervals.The geographical boundaries of local authority districts change signicantly from time to time.
345
(7.1)
(ln(Ai,t ) At ) / At
pi,t
gi,t
ln(Ai,t)
li,t
Pi,t
346
Gi,t
D1,i,t
D2,i,t, D3,i,t are dened similarly for 2nd and 3rd place nishes
pri,t1
rei,t1
At, At
Pk,t,
Pk,t mean and standard deviation of Pi,t across all teams
in division k in season t
Gk,t,
Gk,t mean and standard deviation of Gi,t across all teams
in division k in season t
ui,t
error term.
The inclusion in (7.1) of the lagged terms in ai,t1 and li,t1, although
required in order to capture the dynamics of the short-term interactions
between league position and attendance, makes (7.1) dicult to interpret
if the main interest is in the long-term equilibrium or steady-state relationship between variables such as league position, price and goals scored
and attendance.When estimating a model with lagged variables, however,
it is possible to derive an equivalent steady-state model, by setting current
and lagged values of each variable equal to their assumed steady-state
values, and then rearranging the equation. In this case, the relevant
steady-state variables are:
ai,t ai,t1 . . . asi ;
pi,t pi,t1 . . . psi ;
li,t li,t1 . . . l si ;
git git1 . . . g si
Setting all dummy variables to zero for simplicity, the implied steadystate equation is:
asi i 1 l si 2 psi 3 g si
where i i /(1 1); 1 (2 3)/(1 1); 2 4/(1 1); and
3 5 / (1 1).
(7.2)
347
(7.3)
R2 0.95
(7.4)
In preliminary estimations of (7.1), diagnostic tests revealed evidence of serial correlation and heteroscedasticity in the residuals. To correct for serial correlation, the nal
version is estimated using Hatanakas (1974) two-step procedure, assuming that the
error term ui,t follows a rst-order autoregressive process.To correct for heteroscedasticity, all standard errors and signicance tests are based on Whites (1980) adjustment.
44,775
36,090
36,082
36,493
32,995
40,289
30,811
30,271
27,751
29,361
27,226
24,842
23,972
25,777
21,479
22,112
21,796
22,464
20,491
19,034
18,539
16,337
20,092
19,640
18,220
Rank Team
11
12
13
14
15
16
17
18
19
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
Manchester United
Tottenham Hotspur
Everton
Arsenal
Newcastle United
Liverpool
Manchester City
Chelsea
Sunderland
Aston Villa
Leeds United
Sheeld Wednesday
West Ham United
Wolverhampton Wandrs
Leicester City
Birmingham City
West Bromwich Albion
Nottingham Forest
Sheeld United
Coventry City
Norwich City
Crystal Palace
Middlesbrough
Derby County
Southampton
Average
attendance
(1)
6.0
10.1
10.9
7.0
16.8
7.8
15.1
15.2
23.1
15.7
17.1
22.5
19.3
19.2
21.6
25.6
19.3
17.9
25.9
28.5
32.9
43.4
25.3
24.3
24.2
Average
position
(2)
0.113
0.075
0.075
0.075
0.057
0.043
0.019
0.007
0.004
0.001
0.059
0.062
0.066
0.103
0.124
0.127
0.136
0.136
0.144
0.149
0.151
0.151
0.157
0.172
0.179
i
(3)
2i
(6)
0.910
0.166c
0.044
0.051
0.056
0.080
0.212a
0.068a
0.117
0.064
0.155
0.081
0.071
0.059
0.000
0.096
0.030
0.041
0.018
0.090
0.300a
0.007
0.058
0.091
0.027
1i
(5)
0.042
0.013
0.009
0.016
0.005
0.003
0.012
0.015a
0.016c
0.007
0.009
0.011
0.036a
0.007
0.009
0.020b
0.010a
0.027a
0.015a
0.019a
0.008a
0.015a
0.024a
0.001
0.017c
1i
(4)
0.926a
0.646a
0.697a
0.623a
0.683a
0.646a
0.620a
0.354b
0.659a
0.768a
0.834a
0.692a
0.846a
0.557a
0.603a
0.626a
0.528a
0.589a
0.555a
0.496a
0.641a
0.739a
0.575a
0.876a
0.614a
0.211
0.088
0.138
0.107
0.232c
0.042
0.004
0.048
0.145
0.204
0.207
0.152
0.027
0.150c
0.031
0.238b
0.068
0.043
0.134c
0.020
0.018
0.231
0.157
0.443
0.118
3i
(7)
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
Stoke City
Portsmouth
Bristol City
Brighton and Hove Albion
Ipswich Town
Burnley
Bolton Wandrs
Preston North End
Plymouth Argyle
Fulham
Blackburn Rovers
Queens Park Rangers
Swindon Town
Watford
Hudderseld Town
Millwall
Cardi City
Charlton Athletic
Hull City
Bristol Rovers
Brentford
Blackpool
Notts County
Luton Town
Oldham Athletic
Swansea City
Bradford City
Peterborough United
Bournemouth
Reading
Barnsley
19,230
17,499
14,689
13,367
17,028
16,667
17,063
14,686
12,500
14,895
16,068
14,240
10,656
11,015
13,108
11,495
14,454
14,689
12,549
11,044
19,897
13,613
11,507
12,382
19,574
10,212
17,475
16,215
17,848
18,200
19,560
23.5
33.0
44.5
49.0
26.7
30.6
30.1
40.1
47.4
37.5
27.2
32.8
53.8
49.7
38.7
47.8
41.6
30.3
46.5
48.0
57.4
34.5
47.6
32.0
50.8
52.2
64.1
66.1
59.9
57.6
49.2
0.182
0.186
0.188
0.209
0.250
0.259
0.259
0.272
0.275
0.278
0.281
0.298
0.307
0.312
0.326
0.327
0.329
0.331
0.344
0.352
0.353
0.356
0.364
0.365
0.391
0.400
0.406
0.413
0.421
0.421
0.424
0.412a
0.533a
0.571a
0.533a
0.654a
0.723a
0.567a
0.627a
0.605a
0.581a
0.698a
0.600a
0.501a
0.738a
0.017
0.565a
0.604a
0.543a
0.697a
0.859a
0.503a
0.413a
0.741a
0.458a
0.512a
0.676a
0.257
0.315c
0.528a
0.437a
0.585a
0.017a
0.013b
0.013a
0.019a
0.028a
0.006
0.025a
0.013b
0.021b
0.019a
0.022a
0.027a
0.020a
0.018a
0.025a
0.009b
0.028a
0.028a
0.010
0.016
0.015a
0.022a
0.007
0.016a
0.011b
0.025a
0.029a
0.018a
0.012b
0.020a
0.010
0.047
0.091
0.085
0.056
0.295a
0.205
0.107
0.176
0.086
0.211b
0.275c
0.079
0.120c
0.060
0.077
0.174a
0.613b
0.156a
0.060
0.449
0.058
0.032
0.242
0.071b
0.218c
0.188
0.068
0.054
0.053
0.058
0.272
0.102
0.189
0.041
0.024
0.072
0.043
0.085
0.280b
0.064
0.130c
0.187
0.046
0.134c
0.152
0.217a
0.016
0.043
0.030
0.023
0.744b
0.035
0.108c
0.385c
0.013
0.188c
0.199
0.060
0.054
0.311a
0.012
0.366b
18,881
16,969
17,521
17,187
17,926
18,400
16,298
16,975
18,851
18,414
16,770
16,572
16,658
17,211
15,391
17,165
16,194
16,188
15,580
16,608
13,832
17,276
15,340
14,638
14,917
Rank Team
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
Grimsby Town
Northampton Town
Oxford United
Southend United
Port Vale
Leyton Orient
Gillingham
Walsall
Bradford Park Avenue
Rotherham United
Lincoln City
Wrexham
Chestereld
Carlisle United
Exeter City
Doncaster Rovers
Manseld Town
Tranmere Rovers
Stockport County
Wimbledon
Hereford United
Bury
York City
Aldershot
Colchester United
Average
attendance
(1)
50.6
69.8
45.3
61.1
59.7
53.5
67.1
63.3
69.9
49.6
65.5
65.5
63.1
58.5
75.1
68.2
66.7
64.4
72.4
31.7
77.5
54.9
70.2
77.8
70.6
Average
position
(2)
1i
(4)
0.569a
0.455a
0.612a
0.498a
0.517a
0.533a
0.568a
0.497a
0.206
0.407c
0.447a
0.658a
0.400a
0.463a
0.285c
0.464a
0.449a
0.783a
0.367a
0.830a
0.328a
0.582a
0.308b
0.711a
0.463a
i
(3)
0.434
0.437
0.444
0.449
0.453
0.460
0.472
0.473
0.476
0.495
0.501
0.505
0.507
0.507
0.517
0.525
0.531
0.535
0.540
0.553
0.553
0.555
0.566
0.580
0.588
0.011a
0.015b
0.015a
0.006
0.022a
0.013a
0.011c
0.011c
0.019a
0.023a
0.013a
0.015
0.013a
0.018a
0.007c
0.024a
0.013b
0.009
0.012b
0.075
0.022a
0.021a
0.014a
0.008
0.006
1i
(5)
0.017
0.094
0.017
0.181a
0.113
0.121
0.024
0.169b
0.048
0.162
0.072
0.210c
0.105
0.086
0.083
0.198
0.061
0.115
0.188a
0.994
0.196a
0.145b
0.025
0.061
0.084
2i
(6)
0.175b
0.011
0.103
0.145c
0.070
0.086
0.109
0.055
0.000c
0.047
0.134c
0.144
0.179c
0.029
0.071
0.233
0.124
0.277
0.122
0.918
0.046
0.027
0.154b
0.259c
0.131
3i
(7)
Scunthorpe United
Cambridge United
Shrewsbury Town
Newport County
Wigan Athletic
Torquay United
Chester City
Crewe Alexandra
Hartlepool United
Darlington
Halifax Town
Barrow
Rochdale
Southport
Workington Town
15,024
14,190
15,599
15,435
13,713
14,691
14,481
14,292
14,283
14,022
14,010
15,099
13,665
14,169
13,970
68.9
61.2
57.4
73.0
65.5
73.2
74.3
76.5
79.5
80.0
76.9
76.1
77.5
76.3
77.7
Notes:
a
signicantly dierent from zero, two-tail test, 1% level;
b
5% level;
c
10% level.
i s are not indicated.
Statistically signicant
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
0.589
0.590
0.599
0.607
0.614
0.617
0.621
0.628
0.638
0.655
0.714
0.729
0.742
0.801
0.845
0.328b
0.228
0.610a
0.638a
0.753a
0.604a
0.651a
0.611a
0.379b
0.169
0.346b
0.184
0.444a
0.164
0.564b
0.015a
0.014a
0.014a
0.004
0.028
0.022a
0.002
0.006
0.019c
0.019a
0.011a
0.007
0.010
0.001
0.004
0.114c
0.047
0.050
0.173
0.849
0.093
0.173
0.090
0.027
0.059
0.004
0.114b
0.026
0.115
0.829a
0.013
0.041
0.009
0.238
0.530
0.100
0.331b
0.132
0.111
0.060
0.030
0.099
0.095
0.212a
0.264
352
353
expected signs. All are signicantly dierent from zero at the 5 per cent
level except the estimate of 5 (the coecient on gi,t, goals scored), which is
insignicant at any conventional signicance level.The estimates of 1 (the
short-term loyalty coecient on ai,t1, lagged attendance) and 2 and 3
(the coecients on li,t and li,t1, currrent and lagged league position) have
particularly high t-statistics, indicating that these variables are the strongest determinants of attendance according to the estimated model.The estimate of 4 (the coecient on pi,t, admission price), while also signicant at
the 1 per cent level, nevertheless has a much smaller t-statistic.
In the steady-state equation (7.4), the estimates of 1 (the steady-state
league position coecient) and 2 (the steady-state price coecient) are
signicantly dierent from zero at the 1 per cent level, and the estimate of
3 (the steady-state goals scored coecient) is signicantly dierent from
zero at the 5 per cent level. Using data for the 1997 season for Premier
League clubs, (7.4) implies that an improvement of one place in a clubs
nal league position would add 1.68 per cent to its average attendance. A
1 increase in admission price would lead to a loss of attendance of 0.81
per cent, implying a price elasticity of 0.114 for a club charging the divisional average admission price. These estimates are similar to those for
the period 192692 reported in Dobson and Goddard (1995).
It is also possible to estimate a set of club-specic attendance equations, in which the slope coecients (1 10) are allowed to vary
between clubs in the same way as the intercept coecient (i ) in (7.1)
and (7.3).7 The model specication, shown below as equation (7.5), is
essentially the same as (7.1), except that i-subscripts are attached to each
slope coecient to indicate that these coecients are now specic to each
club. Following the same procedures as before, it is also possible to obtain
the steady-state equivalent, (7.6).
ai,t i 1i ai,t1 2i li,t 3i li,t1 4i pi,t 5i gi,t 6i D1,i,t
7i D2,i,t 8i D3,i,t 9i pri,t1 10i rei,t1 ui,t
(7.5)
(7.6)
Downward and Dawson (1999) have pointed out that the estimates of the coecient on
the lagged dependent variable ai,t1 in the pooled model ((7.3)) are downward biased
and inconsistent, owing to the presence in the model of the individual club xed eects
or alpha coecients. This problem disappears in the club-specic estimations. The fact
that the estimated 1 (equal to 0.5125) in (7.3) is smaller than 61of the 96 estimated 1i s
in table 7.2 reects the bias in the pooled model.
354
2i (the steady-state price coecient) and 3i (the steady-state goals
scored coecient).8 There is considerable variation between clubs in the
estimates of these coecients, as would be expected given the relatively
small number of time-series observations for each club. Nevertheless, the
general pattern is the same as before. The estimates of 1i are signed as
expected in all 96 cases, and are signicantly dierent from zero at the 5
per cent level on a two-tail test in 86 cases.The estimates of 1i are signed
as expected in 89 out of 96 cases, and signicant in 58 of these cases.
Price and goals scored are found to exert a less powerful eect on attendance, as before.The estimates of 2i and 3i are signed as expected in 59
and 80 cases, respectively. The numbers of these coecients that are
signicant at the 5 per cent level are 13 and 10, respectively.
The results indicate, unsurprisingly, that variations in league position
are by far the strongest determinant of variations in attendance. Figure
7.1 shows plots of (unadjusted) average league attendances and league
position (measured, as above, on a scale of 92 to 1) for six selected clubs:
Arsenal, Blackpool, Bolton Wanderers, Derby County, Ipswich Town and
Sheeld United. In most of these cases, there is obvious similarity
between the shapes of the two plots. Arsenal, however, whose d 1i in table
7.2 is not statistically signicant, are an exception. Relative to other clubs,
there is little variation in Arsenals performance, as they remained in the
top division throughout the period.Variations in Arsenals attendance do
not appear to be closely correlated with movements in their position
within the top division. Blackpool and Ipswich Town are included in
gure 7.1 as representatives of the downwardly mobile northern and
upwardly mobile southern contingents, respectively (see chapter 2). For
Derby County, one of six clubs whose d 1i in table 7.2 is negative (though
not signicant), the attendance and performance plots appear to correspond in the early 1970s, when the team surged from relative obscurity to
win the Division 1 title twice in four seasons.There seems to be little similarity elsewhere, however.
BoltonWanderers (d 1i 0.25) and Sheeld United (d 1i 0.15) form an
interesting contrast. Changes in Boltons performance are tracked closely
throughout by changes in their attendance. Several seasons spent near the
foot of Division 3 followed by one season in Division 4 in 1988 saw
average attendances fall to around 20 per cent of their immediate post-war
8
With a maximum of 50 time series observations per club, diagnostic tests show little evidence of serial correlation in the residuals in the club-level estimations.The estimations
of (7.5) are therefore carried out using ordinary least squares (OLS), without incorporating an autoregressive error structure. There is, however, some evidence of heteroscedasticity, so the standard errors and signicance tests are again based on Whites (1980)
adjustment.
355
levels. For Sheeld United, the link between performance and attendance is also apparent, but less pronounced.When the club dropped into
Division 4 in the 1982 season, average attendances still held up to around
35 per cent of their immediate post-war levels.
Explaining base attendances, and the loyalty, league position, price
and goals scored coecients
The second stage of the empirical analysis investigates whether there are
observable characteristics of the clubs or their home towns that are
important in determining the estimated values of i (the alpha
coecients), 1i (the short-term loyalty coecients), 1i (the steady-state
league position coecients), 2i (the steady-state price coecients) and
3i (the steady-state goals scored coecients) reported in table 7.2. At the
second stage, estimations are carried out of ve cross-sectional regressions, in which the estimated coecients listed above are used as dependent variables, to be explained by a set of variables which measure club or
home town characteristics. The full list of explanatory variables used in
the second-stage regressions is as follows:
Hi natural logarithm of the population of the local authority
district in which club i is located, recorded in the 1981
Census of Population
Xi the natural logarithm of the number of years since club i
rst entered the league, measured at the end of the observation period in 1997
Ci the number of other clubs that were members of the league
for at least 18 seasons during the post-Second World War
period, located within a 30 mile radius of club is stadium
Fi the percentage of employed residents working in agriculture in club is local authority district, 1981 Census of
Population
Ii the percentage of employed residents working in energy
and water, manufacturing and construction in club is local
authority district, 1981 Census of Population
Ui Males aged 1664 out of employment as a percentage of
those classed as economically active in club is local
authority district, 1981 Census of Population.
Where several clubs are located in the same town, all have the same Hi, Fi,
Ii and Ui based on the towns census data. Inevitably, the use of data from
Figure 7.1 Trends in league attendance and performance, selected English clubs
Sources: Butler (1987), Smailes (1992),Tabner (1992), Rothmans.
368
369
i
Base
attendance
(1)
1i
Short-term
loyalty
(2)
1i
League
position
(3)
Price
(4)
3i
Goals
scored
(5)
2.4978a
(0.3750)
0.1791
(0.3796)
0.01358
(0.01547)
0.3460
(0.2428)
0.3341
(0.2144)
Hi
Home-town
population
0.0748a
(0.0183)
0.0168
(0.0201)
0.00019
(0.00100)
0.0150
(0.0147)
0.0268c
(0.0137)
Xi
Duration of
league membership
0.2841a
(0.0653)
0.1429b
(0.0681)
0.00012
(0.00217)
0.0773c
(0.0445)
0.0393
(0.0268)
Ci
Local competition
0.0095b
(0.0042)
0.0008
(0.0051)
0.00027
(0.00024)
0.0037
(0.0039)
0.0002
(0.0031)
Fi
Agricultural
employment
0.016
(0.0198)
0.0035
(0.0117)
0.0001
(0.00063)
0.0055
(0.0098)
0.0021
(0.0075)
Ii
Industrial
employment
0.003
(0.0021)
0.003
(0.0026)
0.00004
(0.00013)
0.0009
(0.0013)
0.0018
(0.0016)
0.0132a
(0.0044)
0.0042
(0.0057)
0.00008
(0.00017)
0.0063b
(0.0028)
0.0058b
(0.0027)
Independent
variables and
denition
Constant
Ui
Unemployment
obs.
96
96
2i
96
96
96
R2
0.83
0.94
0.86
0.27
0.19
s.e. of regression
4.94
1.31
1.50
1.26
1.26
Notes:
a
signicantly dierent from zero, two-tail test, 1% level;
b
5% level;
c
10% level.
standard errors of estimated coecients are shown in parentheses.
370
governing entry and exit to and from the league should be relatively cautious in terms of the criteria laid down by which new clubs are able to
replace established ones (Dobson and Goddard, 1995). Of course, this
does not preclude the possibility that other considerations in favour of
admitting a new club may outweigh the tradition factor. Other things
being equal, however, the results suggest that historical tradition does
have a major bearing on spectator demand, and should be an important
consideration when framing rules governing promotion and relegation to
and from the league.
As expected from elementary economic theory, the number of clubs
located within close geographical proximity (a crude measure of the level
of competition facing each club), Ci, has a negative and signicant association with base attendance. On the other hand, the rate of unemployment, Ui , is found to have a positive association. This nding perhaps
runs counter to theoretical expectations, but is in line with the results of
the cross-sectional attendance model of Baimbridge, Cameron and
Dawson (1996). As discussed earlier in this chapter, it may simply reect
the fact that for various sociological and historical reasons, many of the
most keenly supported clubs are located in cities or regions which tend to
suer from high structural unemployment.
In contrast to the results for the alpha coecients, few of the explanatory variables are signicant in the regressions which use the short-term
loyalty, league position, price and goals scored coecients (in turn) as the
dependent variable, reported in columns (2)(5) of table 7.3. To some
extent this is to be expected. While there are strong reasons to expect
factors such as population, age or the strength of local competition to
inuence the average or base level of a clubs attendance, there seems less
reason to expect such factors to inuence the sensitivity of attendances to
variations in league position, price, and so on.
The only explanatory variable which is signicant in the regression
with the estimated short-term loyalty coecient, b 1i, as dependent variable, is the number of years since the club entered the league, Xi. Clubs
which entered early are found to have a higher short-term loyalty
coecient, a nding which seems to emphasise the importance of this
factor in shaping attendance patterns. None of the explanatory variables
is signicant in the regression for the estimated steady-state league position coecient, d 1i. In the regression for the estimated steady-state price
coecient, d 2i, the coecient on the rate of unemployment, Ui, is positive
and signicant at the 5 per cent level, and the coecient on the duration
of league membership, Xi, is negative and signicant at the 10 per cent
level.The former suggests that attendances in towns with high unemployment are more sensitive to variations in price than attendances elsewhere:
371
372
and composition of the crowds that watch certain clubs, as the histories of
FC Barcelona in Spain and Celtic and Rangers in Scotland make abundantly clear.There are signs, however, that as the relevant historical divisions in society tend to fade (and as the playing stas of the respective
clubs tend to become increasingly cosmopolitan) xtures such as
CelticRangers are increasingly becoming spectacles played out in stadia
full of relatively orderly seated spectators, whose displays of the paraphernalia of sectarianism (ags, banners and songs) are mainly symbolic or
nostalgic. The time may be not too far distant when Old Firm derbies,
while still supercharged in terms of competitive intensity, are not fundamentally dierent from other xtures involving leading teams in other
countries, that lack the same historical associations.
In the empirical section of chapter 7, a two-stage econometric model of
variation in English annual average club-specic league attendances
during the post-Second World War period is presented. At the rst stage,
an attendance model is estimated, which ranks all clubs according to their
estimated base attendance, and which examines the eect on attendance
of short-term loyalty, team performance, admission price and goals
scored. At the second stage, the cross-sectional variation between clubs
base levels of attendance, and their short-term loyalty, performance, price
and entertainment coecients, is explained in terms of socio-economic,
demographic and football-related characteristics of each club and its
home town. Home-town population, the amount of competition from
other local football clubs and the duration of each clubs league membership are all highly signicant determinants of base attendances. Duration
of membership is also a signicant determinant of the short-term loyalty
coecient.These results suggest that clubs which joined the league early
continue to enjoy signicant rst-mover advantages, in many cases more
than a century after they rst gained membership! It is evident that sporting history and tradition remain important factors in shaping patterns of
demand for attendance at English football, even in the modern era.
One of the most inuential and durable propositions of neoclassical economics is the idea that the interactions between large numbers of buyers
and sellers operating in free, unregulated markets, each taking spontaneous and uncoordinated decisions in pursuit of their own self-interest, can
normally be relied upon to deliver an ecient allocation of scarce
resources.The originator of this notion was Adam Smith, whose invisible
hand metaphor, rst articulated The Wealth of Nations published in 1776,
represents one of the most spectacular intellectual achievements of the
late-eighteenth-century Scottish enlightenment. A key assumption of the
many theoretical models that have since been developed to formalise
Smiths brilliant insight is that all relevant information about the prices
and characteristics of tradeable commodities is transmitted rapidly and
accurately among buyers and sellers. Otherwise, trade that takes place on
the basis of incomplete or inaccurate information may well lead to a misallocation of resources.
Unsurprisingly, therefore, investigating the speed and eciency with
which relevant information is impounded into the prices at which trade
takes place in markets is of considerable interest to economists and
others. Its importance is not purely theoretical, for the reasons outlined
above, but also highly practical. If relevant information is available to
some traders but not to others, the well informed can exploit their informational advantage to trade with the uninformed at prices which the
latter would not accept if they had access to the same information. In
other words, access to privileged information confers the opportunity to
prot at the expense of the informationally disadvantaged.
Chapter 8 investigates two cases in which information about the performance of football teams on the eld of play is relevant to the formation of
prices at which trade takes place in particular markets. In both cases, the
focus is on the eciency with which useful information is impounded into
the relevant prices.The rst case, considered in sections 8.1 and 8.2, is the
secondary market for trade in the shares of football clubs that are quoted
on the London Stock Exchange (LSE). Rapid changes in footballs
373
374
375
376
evolve. Walvin (1994) identies the growth of a stronger ethos of professionalism within football during the 1920s and 1930s as instrumental in
encouraging directors to begin shifting responsibility for performance on
the pitch towards the team manager (see chapters 5 and 6).
Morrow (1999) identies a number of characteristics of the nancial
structure of the typical English football club prior to recent changes in the
nancing of football. Traditionally, clubs were small, privately owned
companies, which tended to be under-capitalised, and to raise little or no
nance from retained prot. Under-capitalisation (in other words, a relatively low ratio of equity to total nancing) seems to have been a consequence of the reluctance of many club owners to dilute or jeopardise their
personal control by using new equity issues as a means of raising nance.
In many cases, however, club owners or major shareholders provided
signicant additional long-term nance in the form of personal loans.
Although these are technically distinct from equity nance, the practical
implications of the distinction may often be minimal (Deloitte and
Touche, 1999). Borrowing from banks, often on a short-term basis
through overdraft facilities, was (and in many cases still is) another major
source of nance for clubs whose outgoings are in excess of their revenues, either temporarily or in the longer term.
With hindsight, the otation of Tottenham Hotspur on the LSE Ocial
List in October 1983 might appear to have been a groundbreaking development. The issue, which was four times over-subscribed, raised 3.3
million, a sum considerably more substantial at the time than it seems by
present-day standards. But for a number of years Tottenham remained
the only club with a Stock Exchange quotation, perhaps because the
companys subsequent performance did not oer strong encouragement
to others who might have been considering making the same move. By the
end of 1990, a series of unsuccessful diversication ventures, implemented through the creation of subsidiary companies, left Tottenham
shares trading slightly below their opening price seven years earlier.1
Nevertheless, other clubs did eventually follow suit. Millwall raised
4.8 million in a otation in October 1989, and Manchester United 6.7
million in June 1991. Again, the early experiences of the quoted companies were not encouraging.Within two years, Millwall shares were trading
at around one-tenth of their original value. The teams relegation from
Division 1 in May 1990, followed by a further decline in its fortunes
within Division 2, led investors to scale back drastically their assessment
1
In December 1990, trading was suspended for a year, until the clubs principal shareholder, Irving Scholar, was eventually bought out by consumer electronics entrepreneur
Alan Sugar and team manager Terry Venables.
377
of the companys value, before trading in the shares was eventually suspended. Manchester United shares were more than 50 per cent undersubscribed at the time of otation in June 1991, and were still trading
below their original oer price more than 18 months later.
At the start of the 1990s, other clubs were experimenting with alternative methods of raising external nance. Highly controversial at the time
was Arsenals Highbury Bond issue, targeted at the clubs own supporters. Bond purchasers were guaranteed the right (eectively in perpetuity)
to purchase a season ticket for a specic seat in return for an investment
of 1,500 or 1,100 (Campbell and Shields, 1993). Despite a slow initial
take-up, the bond eventually raised about 14 million towards the construction of a 22.5 million new stand. A handful of other clubs launched
similar bond schemes in the early 1990s, with varying degrees of success.
West Ham Uniteds Hammers Bond, launched in 1991, succeeded only
in triggering a large-scale supporters revolt, and by mid-1992 the club
were admitting defeat. By the end of the 1990s, Newcastle United were
nding that long-term guarantees of the right to a specic seat made to
bond purchasers several years earlier were an inconvenient and embarrassing (temporary) impediment to the further redevelopment of their
stadium.
As described in chapter 2, the rst BSkyBPremier League television
contract, covering the ve seasons from 1993 to 1997, injected television
revenues on a scale previously unimagined into football. But almost as
soon as this contract was underway, it was realised that potential still
remained for even higher television revenues to be earned in the future.
By 1995 speculation concerning the possible value of the next television
contract was commonplace and often hyperbolic. A surge of optimistic
sentiment concerning the sectors prospects for continued revenue
growth led to massive appreciation in the prices of Tottenham Hotspur
and Manchester United shares; by the end of December 1996, the
increases over the previous 18 months were 368 per cent and 336 per
cent, respectively.The circumstances were clearly favourable for a spate of
15 further otations, all of which took place between September 1995
and October 1997. The costs of stadium redevelopment, and spiralling
player salaries and transfer fees, were both creating signicant pressure
for football clubs to nd new sources of nance. Needless to say, in the
eyes of institutional investors the former was a more acceptable
justication for otation than the latter.The desire of current club owners
to realise gains from past investments may also have been a factor motivating the decision to oat in certain cases.
It is worth describing briey some of the mechanics of otation. Table
8.1 provides some of the details, and charts the six-monthly percentage
FTSE 100
PO
RT
RT
PO
T/PO
PO
PO
T/PO
PO
PO
Method
13
15.2
50.4
15.5
17.5
20.5
11.4
10.7
12
17.1
16.7
16.7
13.3
Proceeds
(m)
11
28
68
10
19
12.95
6.95
13
35
121
119
6.96
11
28
18
23
69
54
28
46
14
10
10
45
7
15
12
21
4
32
51
35
41
24
31
17
38
40
26
35
12
14
32
14
12
48
30
40
7
24
42
5
30
20
18
21
1
6
40
8
16
42
11
11
14
21
41
10
12.98
31
29
43
35
3
7
18
8
12
4
6.98
45
25
9
12.97
20
12.96 6.97
09/10/97
24/10/97
06/05/97
Nottingham Forest
22/04/97
Southampton Leisure
Aston Villa
Leicester City
20/03/97
01/04/97
06/03/97
Charlton Athletic
Newcastle United
14/01/97
02/01/97
Birmingham City
23/10/96
Loftus Road
(Queens Park Rangers)
23/12/96
01/08/96
Leeds Sporting
01/01/97
29/03/96
Chelsea Village
Sheeld United
29/09/95
Sunderland
12/10/83
07/06/91
Manchester United
Company
Tottenham Hotspur
Date of otation,
and list
9
22
36
16
24
8
1
9
22
30
50
7
18
16
11
12
6.99
Table 8.1. Date and method of otation and share price returns from otation to December 1999, oated English clubs
10
9
36
15
15
16
20
8
42
33
28
11
6
12.99
379
Companies seeking acceptance onto the Ocial List must be able to produce audited
accounts for the previous three years, must satisfy a minimum capitalisation requirement
and must make some shares available to the general public. None of these criteria applies
to companies joining AIM.The absence of the need to demonstrate a historical record of
protability is in contrast to the rules of several European exchanges. This fact has
encouraged some leading European clubs to consider the possibility of acquiring a
listing in London (Morrow, 1999).
The timing was also fortuitous with regard to Newcastle Uniteds on-eld performance.
Newcastle nished 2nd to Manchester United in the Premier League in both the 1996
and 1997 seasons, and Newcastle was seen widely as one of the two or three leading
English teams. Finishes in 13th position in both the 1998 and 1999 seasons, however,
rather belied this impression. By December 1999, Newcastle shares were trading at just
over 50 per cent of their April 1997 opening price.
380
Morrow (1999) suggests that, in some cases, the presence of a dominant chairman may
act as a deterrent to the acquisition of shareholdings by institutional investors.
381
382
383
384
Sunderlands eventual promotion from Division One was conrmed several matches
before the end of the following (1999) season, and had been recognised as a forgone conclusion for weeks or months in advance. Little or no share price reaction is discernible on
the trading days immediately after promotion was conrmed.
385
Winning a cup tie merely guarantees that the team will play another tie in the same tournament, with the prospect of further ties (and more revenue), but only for as long as the
team avoids defeat. Winning either the FA Cup or a European tournament outright
guarantees lucrative admission to one of the next seasons European tournaments.
Losing a cup tie, on the other hand, guarantees immediately that no further revenue will
be earned from that seasons cup competition, and closes o a possible route into
Europe for the following season.
386
387
Table 8.2. Estimation results for determinants of share price returns, pooled
model
Market returns and
league match results
FA Cup/Europe
1.62a
(0.50)
0.91c
(0.50)
0.21
(0.50)
B1,1,t
Fi,4,t
0.40
(0.51)
Fi,5,t
0.18
(0.51)
1.97b
(0.95)
3.19a
(0.95)
1.43
(0.95)
0.31
(0.95)
0.17
(0.95)
0.11a
(0.03)
0.20a
(0.03)
1.96a
(0.20)
Fi,1,t
Promotion/relegation
22.48a
(1.45)
2.20
Ui,2,t
(1.45)
1.27
Ui,3,t
(1.45)
1.18
Ui,4,t
(1.45)
2.91
Ui,5,t
(1.45)
15.59a
Di,1,t
(0.95)
2.59a
Di,2,t
(0.95)
0.11
Di,3,t
(0.95)
2.49a
Di,4,t
(0.95)
0.62c
Di,5,t
(0.95)
obs. 6,604
s.e. of regression 2.51
R2 0.15
Constant
rm,t
Yi,t
Ui,1,t
Fi,2,t
Fi,3,t
Ei,1,t
Ei,2,t
Ei,3,t
Ei,4,t
Ei,5,t
25.57a
(2.51)
3.08
(2.51)
7.13a
(2.51)
B1,1,t
B1,4,t
3.75
(2.51)
B1,4,t
3.72a
(0.73)
B1,5,t
0.04
(2.51)
8.85a
(2.51)
0.65
(2.51)
1.31
(2.51)
0.13
(2.51)
0.41
(2.51)
16.10a
(2.51)
0.06
(2.51)
0.43
(2.51)
2.82
(2.51)
6.03b
(2.51)
B1,5,t
7.19a
(0.73)
2.45a
(0.73)
0.90
(0.73)
0.11
(0.73)
0.70
(0.73)
0.05
(0.73)
3.87a
(0.73)
0.19
(0.73)
0.16
(0.73)
1.41c
(0.73)
0.50
(0.73)
B1,2,t
B1,3,t
B2,1,t
B2,2,t
B2,3,t
B2,4,t
B2,5,t
B3,1,t
B3,2,t
B3,3,t
B3,4,t
B3,5,t
Notes:
a
signicantly dierent from zero, two-tail test, 1% level;
b
5% level;
c
10% level.
Standard errors of estimated coecients are shown in parentheses.
Dependent variable is ri,t (daily return on club share prices).
B1,2,t
B1,3,t
B2,1,t
B2,2,t
B2,3,t
B2,4,t
B2,5,t
B3,1,t
B3,2,t
B3,3,t
B3,4,t
B3,5,t
5.33a
(0.73)
1.28c
(0.73)
2.46a
(0.73)
388
SUR takes advantage of similarities in the structure of the error term across equations to
provide more ecient estimates of the models coecients than would be available if
each equation were estimated separately using OLS. The use of SUR to estimate a
system of equations is advantageous if (as seems likely in the present case) there are
regular or occasional shocks, other than those controlled for by the event dummies,
which aect the dependent variable in all equations in a similar manner on the same day.
389
gained in value over the two-year period. In terms of its size, and the
diversied nature and international scope of its business, it seems reasonable to consider Manchester United, whose share price does not behave
in the same manner as those of other quoted clubs, as a special case.
Sunderlands gains appear to reect a period of sustained improvement
on the eld of play, as the club gradually constructed a team which, by the
end of the 1999 season, had achieved and appeared capable of retaining
Premier League status.
The estimated coecient on rm,t of 0.20 (the textbook beta) in table 8.2
indicates a relatively weak link between general market movements and
movements in the football sector. Other things being equal, a 1 per cent
rise in the market index translates into an average rise of 0.2 per cent in
football club share prices on the same day.Tables 8.3 and 8.4 suggest that
there is considerable variation in this coecient between clubs, though
perhaps without very much discernible pattern.
In table 8.2, the pooled coecient on Yi,t of 1.96, which is signicant at
the 1 per cent level, quanties the link between the unanticipated component of league match results and the share price movement on the next
trading day, as discussed above. If a quoted team wins a match it was
expected to lose (with win/draw/lose probabilities of 0.1/0.3/0.6, say, pro 0.25) the models expected share
ducing an expected win ratio of W
i,j
price movement is 1.96 (1 0.25), or just under 1.5 per cent. If the
team wins a match it was expected to win (with probabilities of 0.6/0.3/0.1,
0.75) the expected price movement is 1.96 (1 0.75),
producingW
i,j
or around 0.5 per cent. It is interesting to note that in tables 8.3 and 8.4,
Manchester United is the only club for which the coecient on Yi,t is not
statistically signicant at the 5 per cent level. As before, this seems to indicate that Manchester Uniteds fortunes as a business are no longer dependent on short-term uctuations in the teams fortunes on the eld of play,
especially in regular domestic league competition.
Promotion or avoidance of relegation (captured by Ui,k,t) and relegation
or failure to achieve promotion (captured by Di,k,t), are both found to have
a dramatic and immediate eect on the share prices of the clubs concerned in table 8.2. On average, the former produced an upward adjustment of around 22.5 per cent on the next trading day, and the latter
caused a downward adjustment of around 15.6 per cent (with further
smaller but statistically signicant uctuations over the next couple of
trading days). Immediately before the nal outcome is decided, the share
price of a club involved in promotion or relegation issues should be a
weighted average of the prices that would apply given either outcome (as
discussed above). A next-day reaction of 22.5 per cent when promotion
is conrmed therefore starts from a share price which already reects
7.93a
0.17
0.89
13.05a
10.96a
6.27a
2.76
BSkyB bid
B1,1,t
B1,2,t
B1,3,t
B1,4,t
B1,5,t
B2,1,t
B2,2,t
8.33a
0.41
18.39a
17.92a
22.92a
9.36a
2.68
1.39
8.41a
0.05
9.49a
5.90b
3.28
7.53a
0.30
0.16
3.51b
0.85
1.34
1.35
0.47
8.81a
0.09
2.56
4.53a
3.09c
3.05c
2.85c
0.23c
0.36a
2.01a
3.30c
0.16
2.16
Leeds Utd
0.15c
0.27a
1.31b
3.23a
1.07
0.73
Chelsea
Aston Villa
25.76a
3.20b
7.48a
3.32a
0.19
9.06a
0.49
2.14c
1.16
2.37b
0.39
0.06
0.04
2.42b
0.01
0.11b
0.15
0.32
0.23
1.07
Man Utd
Table 8.3. Estimation results for determinants of share price returns, individual models (1)
7.23a
5.85b
2.08
2.18
23.44a
5.16b
1.80
0.41
0.31
0.06
5.21a
3.86b
0.98
0.17
0.08
0.16c
1.41b
4.48a
0.43
2.47
Newcastle Utd
5.25a
0.08
0.34
3.59b
8.82a
3.35b
0.73
8.10a
1.96b
0.07
0.13b
1.13a
0.39
4.09a
0.99
Tottenham
0.56
2.64
1.11
0.14
0.26
0.40
508
1.89
0.23
0.78
9.17a
1.51
0.18
3.43
0.91
508
2.67
0.35
1.86
0.48
1.51
0.17
9.38a
0.76
0.02
0.10
0.63
508
1.63
0.20
1.21
0.06
0.93
Notes:
signicantly dierent from zero, two-tail test, 1% level;
b
5% level;
c
10% level.
Dependent variable is ri,t (daily return on club share prices).
B2,5,t
B3,1,t
B3,2,t
B3,3,t
B3,4,t
B3,5,t
obs.
s.e. of regression
R2
B2,3,t
B2,4,t
1.26
0.15
16.17a
0.23
0.41
2.67b
5.87a
508
1.25
0.61
0.01
2.47c
3.91a
3.95a
7.15a
5.44a
0.39
9.41a
508
1.46
0.25
0.23
11.86a
7.81a
2.58
0.50
0.78
508
2.59
0.26
0.03
0.03
0.59
Fi,2,t
Fi,3,t
Fi,4,t
Fi,5,t
FA Cup
Fi,1,t
Di,5,t
Ui,4,t
Ui,5,t
Di,1,t
Di,2,t
Di,3,t
Di,4,t
1.54
0.65
0.53
0.49
0.35
5.29c
0.72
1.45
0.26
0.16
0.08
1.36
0.25
0.05
26.46a
6.69b
1.96
9.82a
2.89a
0.15
0.09
0.62a
6.54a
Promotion/relegation
Ui,1,t
Ui,2,t
Ui,3,t
Bolton
Birmingham
0.47
0.20
0.45
0.25
0.15
0.04
1.75
8.47a
1.20
0.03
0.10
3.24b
6.90a
20.80a
0.25
0.10
0.09c
0.98a
Charlton
0.24
3.23
0.41
0.19
4.01c
3.46
5.80a
0.14
0.32
0.27
5.35c
1.16
7.53b
0.97
1.54
3.44
0.11
4.02
0.19
0.45a
2.56a
She Utd
31.66a
0.10
0.20
0.15
2.49a
QPR
Table 8.4. Estimation results for determinants of share price returns, individual models (2)
0.22
0.43
0.05
0.06
0.57
0.07
0.25
1.68
14.47a
6.87a
0.23a
0.15b
2.03a
Southampton
1.02
0.71
0.21
4.08a
0.19
0.00
0.45
20.96a
0.02
2.22
0.09
0.08
2.13a
Sunderland
2.02
1.03
0.25
2.56
2.29
5.61
6.24
508
4.17
0.04
0.11
7.46a
0.48
0.02
0.07
0.07
508
1.69
0.19
B3,3,t
B3,4,t
B3,5,t
obs.
s.e. of regression
R2
Notes:
a
= signicantly dierent from zero, two-tail test, 1% level;
b
= 5% level;
c
= 10% level.
Dependent variable is ri,t (daily return on club share prices).
B2,5,t
B3,1,t
B3,2,t
0.24
2.16
2.97
0.70
0.35
0.40
0.17
0.07
0.09
0.07
0.03
B1,5,t
B2,1,t
B2,2,t
B2,3,t
B2,4,t
B1,4,t
6.42
3.28
0.47
0.56
0.39
1.65
BSkyB bid
B1,1,t
B1,2,t
B1,3,t
0.44
0.14
508
1.35
0.45
0.01
0.13
0.11
0.24
1.18
1.23
1.73
0.94
0.14
0.02
4.30a
5.28a
3.75a
0.19
0.26
508
3.41
0.05
0.38
508
3.51
1.52
0.31
0.33
0.40
0.88
0.06
0.73
0.43
0.11
0.77
0.94
0.25
1.71
0.35
2.09
0.38
0.17
2.28
0.60
0.47
0.72
0.25
0.35
0.27
0.08
0.33
0.31
0.29
1.85
0.97
508
0.28
3.79b
508
0.27
1.90
0.42
0.88
7.93a
3.16c
0.45
1.80
1.66
0.43
0.44
0.05
0.15
2.72
2.06
0.54
0.99
0.06
0.27
0.59
0.46
1.18
0.74
11.40a
0.38
0.99
0.11
11.49a
3.89b
0.32
394
1999
Figure 8.1 Trends in index value of logarithmic share price: Bolton Wanderers
Source: Datastream.
97
Au g -
70
75
80
85
90
98
Au g -
95
97
N ov -
Bolton Wanderers
98
N ov -
100
Feb9
9
Feb9
105
-98
May
Play-off final
defeat May
1999
-99
May
Figure 8.2 Trends in index value of logarithmic share price: Charlton Athletic
Source: Datastream.
397
This can be contrasted with the conditions for semi-strong-form eciency, which requires
that all publicly available information (not just historical information) is reected in the
price; and strong-form eciency, which requires that all publicly and privately held information is reected in the price.
Figure 8.3 Trends in index value of logarithmic share price: Manchester United
Source: Datastream.
97
Au g -
Figure 8.4 Trends in index value of logarithmic share price: Newcastle United
Source: Datastream.
97
N ov -
80
85
90
95
98
Au g -
100
98
N ov -
105
Feb9
9
Feb9
Newcastle United
-98
May
MMC's decision to
turn down bid
leaked March
1999
-99
May
400
it should be possible to use the forecasting model to devise betting strategies that either make a prot, or (at least) reduce the losses expected in
view of the inclusion of the bookmakers margin in the odds, and the obligation to pay betting duty. If the prices are weak-form ecient, then it is
not possible to devise such strategies using historical data that is publicly
available (Fama, 1970).
In the economics literature, there have been numerous studies of the
eciency of betting markets. Much of this literature focuses on racetrack
betting, and only a representative selection of these papers is considered
here. Several early contributions focused on the relationship between the
forecast prices (FP), published in the sporting press on the race-day
morning, and the starting prices (SP), at which all bets are settled, and
which are compiled and published minutes before the race takes place.
Dowie (1976) interpreted a nding of little or no dierence between the
correlations between FP and race outcomes on the one hand, and SP and
race outcomes on the other, as evidence against the hypothesis that bets
were being placed on the basis of inside information between the publication of FP and the start of the race. If inside information were being used,
the SP correlation should exceed the FP correlation. Crafts (1985),
however, demonstrated that this interpretation was not necessarily
justied. Similar FP and SP correlations across all outcomes could be
consistent with substantial discrepancies between FP and SP on individual outcomes, and with SP containing signicantly superior information.
By betting at SP on outcomes whose SP were substantially shorter than
the longest prices that had previously been quoted, signicantly smaller
losses than average could have been realised.
More recent literature has focused on a consistent empirical regularity
in racetrack betting odds known as longshot bias, implying that the
expected return to a bet placed at longer odds is lower than at shorter
odds. Several kinds of explanation have been put forward. One possibility
is that bettors may be risk lovers, and are therefore willing to accept bets
on longshots oering low expected returns but high variance (Weitzman,
1965; Ali, 1977). Another possibility is that bettors subjective probabilities of losing are systematically downward biased (Henery, 1985). If so, it
is straightforward to show that SP are subject to longshot bias. Golec and
Tamarkin (1998) have suggested incorporating skewness, as well as the
mean and variance of the return, into the bettors utility function. They
demonstrate that it may be rational for risk-averse bettors who love skewness (bets oering a high probability of losing and a low probability of a
very large return) to accept bets on longshots despite the low expected
returns.
An alternative kind of explanation regards longshot bias as a rational
401
402
403
Testing the eciency of the bookmakers prices for betting on the results
of English football matches raises a slightly dierent set of issues. In contrast to the situation with racetrack betting, prices are xed by the bookmakers several days before the match takes place, so both bookmaker and
bettor know the price and the payos when the bet is placed.And in contrast
to North American team sports betting, the bookmaker does not adjust the
prices as bets are placed, even if new information is received as the time of
the match draws near.The pricing system requires the bookmaker to quote
odds in the form: a-to-b home win; c-to-d draw; and e-to-f away win.
This means, for example, that if a bettor stakes b on a home win, the overall
net payos are a (the bookmaker pays the winnings and returns the stake)
if the outcome is a home win, and b (the bookmaker keeps the stake) if the
outcome is anything else. Except for televised matches, bettors must place
combination bets on at least ve matches (if home wins are selected) or
three matches (if draws or away wins are selected).
A bookmakers quoted prices can be converted to match outcome
probabilities as follows: uH
b/(ab); uD
d/(cd ); u Ai,j f/(ef ). The
i,j
i,j
sum of these expressions invariably exceeds one, because the odds
contain a margin to cover the bookmakers costs and prots. But implicit
home-win, draw and away-win probabilities which sum to one can be
obtained easily, by applying the adjustment: fH
uH
/ (uH
uD
u Ai,j ), and
i,j
i,j
i,j
i,j
by applying similar adjustments for fD
and fAi,j . The bookmakers total
i,j
margin is i, j uH
uD
u Ai,j 1.
i,j
i,j
For English football, Pope and Peel (1989) investigate the eciency of
the prices set by four national high-street bookmakers for matches played
during the 1982 season. A simple, though as before not necessarily a very
powerful test of the weak-form eciency hypothesis is based on regressions of match outcomes against implicit bookmakers probabilities. For
the purposes of describing these tests, the following variable denitions
are used:
Hi,j 1 if the match between team i and team j results in a home
win, and 0 otherwise
Di,j 1 if the match between team i and team j results in a draw,
and 0 otherwise
Ai,j 1 if the match between team i and team j results in an
away win, and 0 otherwise.
In the linear probability model: Hi,j 1 2fH
vi,j , a necessary condii,j
tion for weak-form eciency is {1 0, 2 1}. Equivalent conditions
404
apply in the corresponding regressions for Di,j and Ai,j . Any other combination of values for {1, 2} suggests that there may be ineciencies.The
combination {1
0, 2 1}, for example, means that bookmakers overestimate the home-win probabilities for the matches least likely to
produce home wins (with long odds), and under-estimate the probabilities for those most likely to produce home wins (with short odds). Noting
that ordinary least squares (OLS) estimation of the linear probability
model gives rise to a heteroscedastic error structure, Pope and Peel
(1 H
)
suggest using weighted least squares (WLS) estimation, using H
i,j
i,j
as weights, where Hi,j are the tted values of the dependent variable
obtained from (preliminary) estimation of the model using OLS.
Alternatively, the model can be estimated as a logit regression, in which
case dierent numerical estimates of the coecients 1 and 2 are
expected.
In the linear probability model regressions, Pope and Peel nd no
signicant departures from the condition {1 0, 2 1} for the odds on
home- and away-win outcomes. On the other hand, the draw odds have
no signicant predictive content for draw outcomes, suggesting a departure from eciency conditions. The returns achievable from two betting
strategies are investigated.The rst, using only information derived from
bookmakers odds, involves betting on outcomes for which the predicted
probability derived from the logit regression of match outcome on bookmakers implicit probabilities exceeds the geometric mean of the four
bookmakers implicit probabilities. The second, also exploiting extraneous information, involves betting on a consensus of the recommendations
of six national newspaper tipsters. Under both strategies, there is some
evidence that positive pre-tax returns could have been earned, though
these would not have been sucient to generate a prot after allowing for
payment of betting duty.
Dixon and Coles (1997) use their match scores forecasting model (see
chapter 3) to test the protability of a strategy of betting on English football matches with the highest expected return according to the match
outcome probabilities derived from the model. Betting on outcomes for
which the ratio of the models probability to the bookmakers implicit
probability exceeds about 20 per cent is found to produce a pre-tax return
signicantly greater than zero, after allowing for the bookmakers margin.
Cain, Law and Peel (2000) report evidence of longshot bias in the
xed-odds betting market for match results and scores in English football.The odds available for bets on specic scores are dependent only on
the odds posted for the three possible match results (home win, draw or
away win), and it is possible to bet on the score of a single match. The
poisson and the negative binomial distributions are used to model the
405
home and away team scores independently. Estimates of the fair odds for
specic scores are computed, conditional on the bookmakers posted
home-win odds. In comparisons of the estimated fair odds with the bookmakers actual odds for specic scores, the former are generally found to
be signicantly longer than the latter for longshot bets, indicating longshot bias.The fair odds are sometimes shorter than the bookmakers odds
for bets on strong favourites. Direct calculations of average protability
for dierent categories of bet suggests that bets on strong favourites may
oer limited protable betting opportunities.
In the rest of section 8.3, the match results forecasting model developed in section 3.3 and estimated using data for the 1989 to 1998 seasons
inclusive (see table 3.14) is used to test for the eciency of the prices
quoted by one of the leading high-street bookmakers for xed-odds
betting on match results during the 1999 season. To test the weak-form
eciency hypothesis, a full set of xed-odds betting coupons for matches
played over each weekend was collected throughout the season. From a
total of 2,036 league matches played during the 1999 season, the forecasting model is capable of generating predictions for 1,944 matches.10
The bookmakers weekend coupons provided odds for 1,568 of these
matches, which constitute the data set for the analysis in this section.The
average value of the bookmakers margin across the 1,568 matches
included in the data set was i,j 0.115, or 11.5 per cent.
Probabilities and forecasts of the results of the 1,568 matches are
obtained by substituting the full set of covariate values for each match
into the estimated version of (3.10), as reported in columns (3)(4) and
(7)(8) of table 3.14.This procedure is described in detail in section 3.3.
For each match, a set of estimated home-win, draw and away-win probA
abilities is obtained. As before, these are denoted p H
, pD
and p i,j
(see
i,j
i,j
(3.12)). Following the methods described in section 3.3, a specic forecast for the result of each match is also obtained.
Tables 8.5 and 8.6 show some of the key features of the implicit bookmakers probabilities and the evaluated match result probabilities. When
the bookmakers and the models probabilities are disaggregated by
month and by division, there is little systematic variation in the mean
probabilities for each outcome, but some signicant systematic variation
in the standard deviations. First, the standard deviations of both the
bookmakers probabilities and the models probabilities increase systematically during the course of the season.Towards the end of the season, it
10
Ninety-two matches involving two teams (Maccleseld Town and Halifax Town) which
entered the league at the start of the 1998 and 1999 seasons respectively, had to be discarded as the forecasting model requires a full set of match results for a two-year period
prior to the match in question.
406
August
September
October
November
December
January
February
March
April/May
Premier
D1
D2
D3
All
Actual proportions
Di,j : Draw
A
: Away win
i,j
Mean
St. dev.
Mean
St. dev.
Mean
St. dev.
0.446
0.446
0.447
0.452
0.450
0.447
0.449
0.467
0.459
0.444
0.453
0.456
0.452
0.452
0.445
0.080
0.088
0.089
0.095
0.091
0.094
0.110
0.112
0.131
0.121
0.101
0.098
0.088
0.102
0.274
0.273
0.273
0.272
0.272
0.273
0.272
0.267
0.258
0.271
0.270
0.269
0.270
0.270
0.285
0.009
0.008
0.009
0.011
0.012
0.009
0.014
0.017
0.022
0.019
0.014
0.015
0.012
0.015
0.280
0.281
0.280
0.275
0.279
0.280
0.279
0.266
0.283
0.286
0.277
0.275
0.277
0.279
0.269
0.074
0.082
0.082
0.089
0.085
0.089
0.103
0.101
0.118
0.110
0.093
0.090
0.082
0.093
August
September
October
November
December
January
February
March
April/May
Premier
D1
D2
D3
All
Actual proportions
pD
: Draw
i,j
A
p i,j
- Away win
Mean
St. dev.
Mean
St. dev.
Mean
St. dev.
0.457
0.453
0.461
0.468
0.464
0.463
0.469
0.483
0.477
0.463
0.471
0.473
0.459
0.467
0.445
0.068
0.073
0.083
0.095
0.089
0.091
0.108
0.108
0.127
0.110
0.106
0.090
0.081
0.097
0.283
0.283
0.280
0.277
0.279
0.279
0.275
0.272
0.270
0.276
0.275
0.277
0.281
0.277
0.285
0.013
0.013
0.017
0.021
0.020
0.020
0.025
0.026
0.030
0.025
0.025
0.020
0.017
0.022
0.260
0.264
0.258
0.255
0.257
0.258
0.256
0.244
0.253
0.261
0.254
0.250
0.260
0.256
0.269
0.057
0.062
0.068
0.078
0.072
0.074
0.088
0.085
0.103
0.090
0.086
0.072
0.066
0.079
407
is possible to make a more specic assessment of the win/draw/lose probabilities based on the current seasons results, than is possible at the start,
when the previous seasons results provide the only available guidance.
Second, there is also some systematic variation in the standard deviations
by division: these are highest in the Premier League and lowest in
Division Three.This seems likely to reect a greater degree of competitive
balance in the lower than in the upper divisions of the league. The forecasting model appears to be reasonably accurate in replicating the main
features of the bookmakers implicit probabilities.
Testing for weak-form eciency without using the forecasting model
WLS estimation results, obtained using the regression-based test procedure employed by Pope and Peel (1989) and Golec and Tamarkin (1991),
are as follows:
Hi,j 0.03231.0572 fH
vi,j
i,j
(0.0524) (0.1145)
F(2,1566)0.28
Di,j 0.19421.7779 fD
vi,j
i,j
(0.1878) (0.6974)
F(2,1566)1.48
A
Ai,j 0.04550.8055c f i,j
vi,j
(0.0319) (0.1154)
F(2,1566)1.52
(8.1)
408
Table 8.7. Average returns from 1 bets, long, medium and short
odds
Betting regime
R1
R2
R3
Home win
Long odds
Medium odds
Short odds
All
0.032
0.001
0.027
0.018
0.132
0.103
0.126
0.119
0.232
0.203
0.226
0.219
Draw
Long odds
Medium odds
Short odds
All
0.015
0.027
0.130
0.056
0.090
0.080
0.015
0.053
0.190
0.180
0.085
0.153
Away win
Long odds
Medium odds
Short odds
All
0.204
0.122
0.090
0.008
0.080
0.212
0.183
0.110
0.020
0.312
0.283
0.210
Draw, AugDec
Long odds
Medium odds
Short odds
All
0.041
0.008
0.091
0.045
0.066
0.097
0.020
0.063
0.166
0.197
0.120
0.163
Draw, JanMay
Long odds
Medium odds
Short odds
All
0.006
0.055
0.183
0.070
0.109
0.054
0.063
0.040
0.209
0.154
0.038
0.140
0.050
0.148
0.113
0.079
0.058
0.237
0.204
0.174
0.158
0.337
0.304
0.274
0.360
0.081
0.064
0.079
0.220
0.175
0.161
0.032
0.120
0.275
0.261
0.132
Key:
R1: bets at odds consistent with bookmakers implicit probabilities (no prot
margin)
R2: bets at quoted odds inclusive of margin, but not allowing for betting duty
R3: bets at quoted odds, with betting duty of 10 pence on each 1 bet.
409
highest values of fH
are classied as having short odds; and the remaining
i,j
568 matches are classied as having medium odds.The same classication
scheme is used for draws and away wins.Three betting regimes are considered. Under R1, betting takes place at odds consistent with the bookmakers implicit probabilities with no prot margin, so the bookmakers
expected return across all bets is zero. Under R2, betting takes place at the
quoted odds inclusive of the bookmakers margin, but not allowing for
betting duty. Several high-street bookmakers oered tax-free betting via
the Internet during the course of the 2000 season, (though this service was
not available during the 1999 season, to which the data used in the present
analysis refers). Finally, under R3, betting takes place at the bookmakers
quoted odds, with betting duty of 10 pence levied on each 1 bet.
The average returns across all bets under R1 reect the tendency of the
bookmakers implicit probabilities to under-estimate the proportion of
draws and to over-estimate the proportions of home wins and away wins
(see also table 8.5). Although a strategy of betting on every match to nish
drawn would produce a prot of just below 6 per cent under R1, this converts to losses of 5 per cent under R2, and 15 per cent under R3.
Strategies of betting selectively on draws with short odds and away wins
with long odds, on the other hand, are capable of producing prots under
both R1 (13 per cent and 20 per cent, respectively) and R2 (2 per cent
and 8 per cent, respectively). But even in these cases, payment of
betting duty converts these prots into losses under R3 of 8 per cent
and 2 per cent, respectively.
If betting duty is ignored, do these results constitute evidence of weakform ineciency and protable trading opportunities? The answer
depends upon whether it would have been possible at any stage to have
anticipated that betting on draws with short odds and away wins with
long odds was the appropriate strategy.This question is investigated in the
lower rows of table 8.7. The season is split into two periods
(AugustDecember and JanuaryMay), and the average returns for the
matches in each period classied previously as having long, medium and
short odds for draws and away wins are reported. There appears to be
some consistency over time in the results. A strategy of betting on draws
with short odds could have been observed to produce a relatively high
return over the period AugustDecember. If continued subsequently, the
same strategy would have produced an average return of 18 per cent
under R1 over the period JanuaryMay. The same applies to the strategy
of betting on away wins with long odds which, if continued over the
period JanuaryMay, would have produced an average return under R1 of
36 per cent. In this case, a positive return of 12 per cent for JanuaryMay
would have been realised even under R3. At the start of January, however,
there would of course have been no guarantee that betting strategies
410
(8.2)
Notes:
b
c
411
August
September
October
November
December
January
February
March
April/May
All
R1
R2
R3
0.121
0.155
0.067
0.081
0.053
0.127
0.056
0.049
0.070
0.031
0.005
0.242
0.043
0.031
0.055
0.011
0.052
0.059
0.167
0.075
0.095
0.342
0.143
0.131
0.155
0.089
0.152
0.159
0.267
0.175
Key:
R1: bets at odds consistent with bookmakers implicit probabilities (no
prot margin)
R2: bets at quoted odds inclusive of margin, but not allowing for betting
duty
R3: bets at quoted odds, with betting duty of 10 pence on each 1 bet.
The F-statistics test the joint hypothesis H0:{1 0, 2 1, 3 0}. This
hypothesis is rejected at the 1 per cent level in the regression for draws
and at the 10 per cent level in the regression for home wins, but is
accepted in the regression for away wins. t-tests for H0:3 0 reject the
null at the 5 per cent level for home wins; at the 1 per cent level for
draws; and at the 10 per cent level for away wins.This seems to constitute
reasonably strong and consistent evidence that the forecasting model
does contain additional information not already impounded into the
bookmakers odds. The latter are therefore found to be weak-form
inecient.
Tables 8.88.10 investigate the extent to which the additional information available from the forecasting model might be capable of generating a
trading prot, under the three (hypothetical and actual) betting regimes,
R1R3.Table 8.8 shows the average returns obtained from betting on the
forecasting models predicted result for every match. The predicted
results are obtained using the procedure described in section 3.3. Under
R1, the average return from all possible bets, placed indiscriminately, is
around 1 per cent. According to table 8.8 however, the average return
from bets placed only on the models predicted outcome for each match is
412
August
September
October
November
December
January
February
March
April/May
All
R1
R2
R3
0.165
0.142
0.082
0.067
0.028
0.081
0.109
0.200
0.218
0.124
0.045
0.024
0.029
0.043
0.077
0.030
0.005
0.077
0.092
0.008
0.055
0.076
0.129
0.143
0.177
0.130
0.105
0.023
0.008
0.092
Key:
R1: bets at odds consistent with bookmakers implicit probabilities (no
prot margin)
R2: bets at quoted odds inclusive of margin, but not allowing for betting
duty
R3: bets at quoted odds, with betting duty of 10 pence on each 1 bet.
Best 900
Second-best 900
Middle 1,104
Second-worst 900
Worst 900
All
R1
R2
R3
0.071
0.151
0.048
0.122
0.106
0.010
0.038
0.032
0.060
0.213
0.198
0.094
0.138
0.068
0.160
0.313
0.298
0.194
Key:
R1: bets at odds consistent with bookmakers implicit probabilities (no
prot margin)
R2: bets at quoted odds inclusive of margin, but not allowing for betting
duty
R3: bets at quoted odds, with betting duty of 10 pence on each 1 bet.
413
just over 3 per cent.The model seems to perform best during the middle
of the season, and less well near the beginning (September) and at the end
(April/May), though bets placed at the very beginning of the season
(August) do generate a positive return. Overall, however, the positive
average return under R1 is insucient to prevent losses from being realised under both R2 and R3.
Betting on the forecasting models predicted outcomes does not represent the most eective use of the information the model generates,
because it does not take advantage of perceived ineciencies in the bookmakers prices.Table 8.9 shows the average returns obtained from betting
on the outcome of each match for which, according to the forecasting
model, the bookmakers odds generate the highest expected return.
Overall, this strategy generates a much higher average return than betting
on the models predicted result. Under R1, there is a positive average
return of more than 12 per cent. The average return under R2 is close to
zero, while R3 still produces an average loss of 9 per cent.This strategy
appears to perform better at the start and at the end of the season than it
does in the middle. It seems likely that it is more dicult to formulate reliable assessments of the match outcome probabilities at the start of
the season (when there is little or no present-season evidence on team
qualities) and at the end (when results are heavily inuenced by incentive
eects) than in the middle of the season.
Finally, table 8.10 shows the average returns from all 4,704 possible
bets (3 bets on each of 1,568 matches), ranked in descending order of
their expected return according to the forecasting model. All of the bets
included in table 8.9 are assessed by the model to represent relatively
good value for money (since only the bet with the highest expected return
for each match is included). In contrast, by including all available bets
table 8.10 provides a better impression of the models capacity to distinguish between bets which represent good or poor value. The 900 bets
assessed by the model as oering the highest expected return produce an
actual average return of 7 per cent under R1, whereas the next 900 generate an actual return of 15 per cent. The average return then decreases
across the remaining three groups, with average losses of 11 per cent
realised on the 1,800 bets assessed by the model to oer the poorest
expected returns, under R1. Of course, the fact that the actual returns
correlate with the rankings by expected return produced by the model
constitutes further evidence of weak-form ineciency in the bookmakers
prices. Unfortunately, however, the ineciencies do not appear to be
suciently large to create protable betting strategies after allowing for
the bookmakers margin and for betting duty.
414
2/5
4/5
Evens
8/11
11/10
10/3
Evens
5/4
Fixture list
H
(1)
3/1
12/5
5/2
5/2
12/5
2/1
9/4
9/4
D
(2)
11/2
11/4
2/1
3/1
15/8
4/5
9/4
7/4
A
(3)
Bookmakers odds
0.639
0.498
0.447
0.519
0.426
0.206
0.448
0.398
H
(4)
0.224
0.263
0.255
0.256
0.263
0.298
0.276
0.276
D
(5)
Bookmakers
probabilities
0.138
0.239
0.298
0.224
0.311
0.496
0.276
0.326
A
(6)
Table 8.11. Fixed-odds betting on Premier League xtures, 24th April 1999
0.633
0.467
0.577
0.438
0.549
0.330
0.493
0.425
H
(7)
0.232
0.285
0.254
0.290
0.263
0.297
0.279
0.292
D
(8)
0.135
0.249
0.170
0.272
0.188
0.374
0.229
0.284
A
(9)
Forecasting models
probabilities
H
D
H
D
H
A
H
D
2-0
1-3
0-0
4-1
1-0
1-6
1-2
1-1
Models
Match
prediction result
(10)
(11)
416
unfortunately this translates into a net loss of 25p after payment of 10p
duty on each bet. In comparison with table 8.8, 24 April 1999 appears to
have been a slightly better-than-average day on which to rely on the
models predictions (for Premier League matches at least). In contrast,
the strategy of betting on the outcome of each match with the highest
expected return according to the models probabilities would have produced only one winning bet out of eight, on Leicester to beat Coventry.
In comparison with the average performance of this strategy for the 1999
season as a whole shown in table 8.9, a gross loss of 5.90 represents a
worse-than-average return.
It is perhaps relevant to note that all four of the matches for which there
was relatively large disagreement between the bookmakers and the forecasting models probabilities carried signicance for championship or
relegation outcomes, for either or both of the teams involved.The disagreements between the probabilities may reect the fact that end-of-season
matches in which the two teams have dierent levels of incentive to win
are hard to assess. As suggested by table 8.11, the bookmaker appears
to have predicted three of these four matches (Derby County v.
Southampton, Everton v. Charlton Athletic and Middlesbrough v.
Arsenal) more successfully than the model (the exception being Leicester
City v. Coventry City).Without wishing to over-generalise on the basis of
a very small number of observations, it does seem possible that the
models championship, promotion or relegation signicance dummy variable is insuciently sensitive to detect all of the incentive eects that
various congurations of placings in the late-season league table might
throw up. It must also be remembered that the bookmakers are able to
incorporate into their prices all kinds of qualitative data (e.g. about suspensions or injuries, or about the quality of recent team performances,
match results notwithstanding) which are not used in the forecasting
model in its present form. But in principle, there is no reason why the
model could not be rened to incorporate any additional relevant information that can be collected and quantied systematically.
8.4 Conclusion
The two empirical investigations reported in chapter 8 are linked by a
common theme: both are concerned with the transmission of information
to secondary markets, in which trade takes place at prices ultimately contingent on occurrences on footballs elds of play. Investigation of the
speed and eciency with which relevant information is impounded into
prices is an important and widely researched topic in both economics and
nance. Interest in this topic is partly theoretical: rapid and accurate ows
417
419
420
421
If rapacious club owners are not the main driving force behind footballs growth in competitive and economic imbalance, then perhaps it is
the players who should shoulder most of the responsibility. Chapter 2 has
charted a number of institutional changes to the structure of players
employment contracts and the organisation of the transfer market, dating
back to the abolition of the maximum wage in 1961, and the reforms to
the retain-and-transfer system that followed the Eastham v. Newcastle
United case at the High Court in 1963. Clearly much has changed
between the early 1960s when Fulhams Johnny Haynes became the rst
English player to earn 100 per week, and the dawn of the new millennium when Roy Keanes new 52,000-a-week contract with Manchester
United made him at that point the English leagues highest-paid player,
earning more in a week than most UK economics professors make in a
year! So is it correct to say that the astronomical wage demands of the top
players have created a situation in which only a handful of wealthy clubs
can aord to employ them, while at the same time conveniently guaranteeing that the same clubs will dominate the league championship yearin-year-out?
It is certainly true that the tendency for the best players to gravitate to
the wealthiest clubs is far more pronounced today than it was, say, 30 or
40 years ago. Englands 1966 World Cup-winning team, for example,
included players whose careers were spent mostly or entirely with relatively unfashionable clubs: Bobby Moore and Geo Hurst both played
exclusively for West Ham United until they were in their thirties and
nearing the end of their careers; George Cohen spent his entire career at
Fulham; and Gordon Banks top-level career was split between Leicester
City and Stoke City. But in order to explain why 35 years later such
players (or their successors) would have been snapped up by Manchester
United, Arsenal, Liverpool or one of a handful of overseas clubs, it is
important to separate cause from eect. The players enhanced bargaining power as free agents does not necessarily provide the ultimate explanation.
It is clear that leading players have been able to increase their remuneration spectacularly because changes to the contractual and transfer
systems have eroded the monopsony power of individual clubs as buyers
of players services. Under the pre-1963 retain-and-transfer system a
player had no choice other than to abide by the wishes of the club currently holding his registration. Quite literally the clubs held all the cards.
In contrast in the post-1995, post-Bosman world of free agency, an outof-contract player is completely free to sell his services to the highest
bidder. Small wonder that players earnings have rocketed as a result.
And, as argued in chapter 4, small wonder that the earnings of the most
422
talented players have rocketed the most, since these are the players with
the greatest monopoly power as sellers of services for which few or no
direct substitutes exist.
Some caution is needed, however, before attributing the rise in competitive imbalance within the league to the same process.As argued in chapter
3, the economically optimal distribution of a resource such as playing
talent should be independent of the ownership of the property rights to
that resource. In other words, shifting the property rights or ownership of
playing talent from the clubs to the players should not aect the optimal
distribution of playing talent between the clubs. Essentially, the question
as to whether a players marginal revenue product (MRP) could be
increased by moving from Club B to Club A, and therefore the question as
to whether a trade can be arranged that leaves both clubs and the player
better o (or at least no worse o) is independent of the players contractual position. If the player is currently under contract and the trade does
take place, Club A pays some portion of the capitalised value of the players
MRP in compensation to Club B. Since Club B (not the player) owns the
property rights, the players reward for the move may be small. If the player
is out of contract, he can bargain with Club A to increase his remuneration
until he captures the bulk of his own MRP with Club A. Since the player
(not Club B) owns the property rights, Club B receives nothing when the
player departs. But the question as to whether the gains from trade actually
exist is dierent from the question as to who captures the gains, assuming
that they do exist and the move does take place. While the rising level of
player earnings can be explained, long-term changes in the equilibrium
distribution of playing talent between clubs remain unexplained.
Sceptics may argue that the invariance result is a property of a highly
stylised abstract theoretical model, but does not necessarily bear much
relation to what is happening in the real world. A fair comment, perhaps,
but empirical studies in the USA that have investigated the eect on competitive balance of the change from the reserve clause to free agency in a
number of sports do not suggest that the change had any discernible
eect on the distribution of performance between teams. Nor does the
evidence suggest that other regulatory measures intended to inuence
competitive balance directly, such as the rookie draft or salary capping,
are eective in achieving this aim. Admittedly the top US sports operate
within an institutional and competitive framework somewhat dierent to
that which applies in European football. But the US experience does
suggest that the invariance result is a powerful one, and that considerable
caution should be exercised before attempting to infer any direct causal
link between the growth in players earnings and the rise in competitive
imbalance.
423
This is not to say that spiralling wages for the top players are a good
thing, or that they do not have adverse consequences for the nancial
health of individual football clubs or the sport as a whole. Indeed, to the
extent that a steeper earningstalent gradient tends to raise the stakes
which clubs just below the highest echelon must be prepared to place if
they wish to gamble on improving their status, it may well have harmful
consequences. Division One clubs like Crystal Palace andWolverhampton
Wanderers, for whom high-spending gambles on achieving Premier
League status have so far failed to pay o, spring readily to mind.An ambitious attempt by Palaces former owners to spend their way back to
Premier League status foundered spectacularly in 1999, bringing the club
close to the point of bankruptcy and closure. Wolverhampton have spent
heavily on players consistently over a number of years, but have regularly
nished tantalisingly short of achieving a promotion berth. Only the willingness of the clubs owners to subsidise heavy losses has allowed this strategy to continue.
If owners and players are absolved from ultimate responsibility for the
rise in competitive imbalance in football, then where does the blame really
lie? The theoretical models discussed in chapters 3 and 4 again provide a
strong clue to the answer. Economic models of competitive balance and
the distribution of playing talent in sporting leagues suggest that the relative size of each teams actual or potential audience is the nal arbiter of
the degree of competitive balance or imbalance that exists within the
league. Economic models also suggest that the ability of the top performers in certain elds to service vast audiences, with little or no additional
costs accruing as the audience size increases, explains not only the high
earnings of superstars, but also their tendency to capture very large shares
of the markets they serve. One should therefore look to changes in the
pattern and composition of demand for the ultimate explanation of
changes in the distribution of playing talent between clubs. And what has
been the biggest single factor inuencing the size and composition of the
audience for professional football over the past 40 years? A one-word
answer to this question will suce, and clearly that word is television.
As seen in chapter 2 the regular televised screening of league match
highlights dates back to 1964.The direct nancial impact of television on
football was minimal during the 1960s and 1970s because the fees paid
for broadcasting rights were puny. Before the days of pay-television,
broadcasters could not charge the audience directly for providing the
service. Televised football had many of the hallmarks of a public good:
audience members unwilling to pay could not be excluded from obtaining the service, so audience members who would have been willing to pay
could not actually be made to do so.
424
425
426
would play in 17. The Cup Winners and UEFA Cups would be merged
into an expanded UEFA Cup competition for clubs failing to qualify for
the Champions League, as well as eight of the clubs eliminated from the
Champions League in the rst group stage.
As the expanded Champions League ran its course for the rst time
during the 2000 season, many commentators felt that the format had
become cumbersome and bloated. Although the knock-out stages (in
particular) produced some breathtaking football, many group matches
were either sterile or defensive aairs that lacked the excitement of
sudden-death competition, or uninteresting because the teams were
unequally balanced, or meaningless because teams had already qualied
or could not qualify for the next stage. Some were played in front of halfempty stadia which contributed to the lack of atmosphere. On the one
hand this may have provided a rather sobering view of a future consisting
of regular league competition at European level. But on the other hand
television viewing gures throughout Europe remained buoyant, with
334 million raised from the sale of broadcasting rights and sponsorship
packages for the 2000 season alone (Marketing Week, 25 May 2000). The
point at which further expansion of the scale of the tournament starts
causing viewers to switch o is unknown, but appears not to have been
breached so far.
Although the same format has been retained for the 2001 season,
UEFA have decided to look again at the structure of the Champions
League, with a reduction in the number of matches one possibility under
consideration. Evidently there are conicting pressures on organisers and
clubs, which apply equally to the Champions League and to any hypothetical Super League. A smaller tournament reduces wear-and-tear on
players and is more exciting, while a larger tournament enables more
clubs to compete and generates more revenue.
It seems likely that advocates of a complete withdrawal from domestic
league competition on the part of the leading clubs have under-estimated
the importance of domestic history, tradition and continuity as dening
characteristics of these clubs identities. Though Manchester United v.
Liverpool and Manchester United v. Real Madrid are more prestigious
than Manchester United v. Southampton, the latter is still a bigger event
than many xtures, involving teams like Sturm Graz, Rosenborg and
Brondby, that could well be staple fare if an inclusive Super League
became a reality; especially if (as is always possible) Manchester United
themselves turned out to be no more than a mid-table team in a European
setting. Of course the same point applies equally to Real Madrid,
Barcelona, Juventus, AC Milan and others.
Even so, wholesale withdrawal by the top clubs from domestic league
427
428
negotiations. Despite continuing resistance from UEFA, it still looks possible, though by no means certain, that the Atlantic League proposals will
eventually materialise in some form.
Wage ination and footballs arms race
Returning to the domestic scene, declining competitive balance raises
important policy issues at all levels within the English league hierarchy.
The widening nancial and competitive gulf between the Premier League
and Football League has already been touched upon in this chapter, and
has been discussed extensively elsewhere (Deloitte and Touche, 1999,
2000). So is it true to say that the escalating wage settlements of the
Premier League stars are creating ripple-like eects throughout the
Football League as aspiring lower-division clubs strive to keep pace, and
that the resulting wage ination is threatening the viability and survival of
many of the smaller clubs?
It certainly seems to be the case that the degree of intra-league mobility
in English football is less than it used to be. And nowhere is this more
apparent than at the boundary between the Premier and Football
Leagues. Only nine of the 20 teams promoted to the Premier League at
the end of its rst seven seasons (1993 to 1999 inclusive) avoided relegation in the rst season after promotion. In contrast, 17 of the 21 teams
promoted to Division 1 in the seven seasons following the introduction of
three-up three-down in 1974 avoided immediate relegation. In the
reverse direction, however, the gures are much closer. Only seven of the
22 teams relegated to Division One between 1993 and 1999 were promoted the following season, while just ve of the 21 relegated to Division
2 between 1974 and 1980 achieved the same feat.
In the 1970s one had the impression that any mid-table Division 2
team could realistically aspire to become a mid-table Division 1 team
within a year or two, either by making a couple of good signings or
(perhaps more by luck than judgement) hiring a particularly eective
manager. Not all such teams did so of course, but the possibility was
always there. But by the turn of the century, a serious assault on achieving
a (relatively) secure foothold in the Premier League appears to have
become an operation requiring millions of pounds of up-front investment
with no guaranteed return, and several years of patient and consistent
development to build up an adequate playing squad, as well as all the
accompanying infrastructure that is now commonplace at the highest
level. True, Division One teams can still achieve promotion without
making all this investment, but a speedy return via the relegation trapdoor
can usually be anticipated in such cases. For clubs such as Leicester City,
429
430
431
system for in-contract players. As suggested in chapter 2, the crosssubsidy the transfer system currently delivers could be achieved by
means of direct transfer of revenue (known euphemistically as revenue
sharing) from the Premier League to the Football League, on a scale that
would be modest relative to the income the Premier League currently
generates. But of course it can also be argued that the incentive for smallmarket teams to discover and develop talented players would be less if
the cross-subsidy were automatic, rather than payable only when a player
is sold.
Details of the arrangements that will replace the transfer system are
unclear at the time of writing. Early speculation suggested that something
similar to a spot market in playing talent could emerge. Under the latest
proposals, players will only be allowed to change clubs during two transfer windows during the summer and winter, each of which will last for a
few weeks. No player will be allowed to change clubs more than once in
any twelve-month period. In-contract players will be permitted to break
their contracts unilaterally, subject to a period of notice of a few months
duration. Such players would be banned from playing for other clubs
during the period of notice. Compensation arrangements will still apply
for moves involving in-contract players.This system will give players who
are in form and in demand massively enhanced bargaining power in wage
negotiations. It will also make the managers tasks of team building and
long-term planning more dicult than under existing arrangements.
Predictions made in 1995 that the Bosman ruling would spell the end
for the transfer market, resulting in wholesale chaos, were clearly wide of
the mark. One of the main reasons is that clubs and players both saw clear
benets in the security that long-term contracts could provide. Securing
the services of a star player for several years gives time for a manager to
build a successful team, or for a club to exploit fully the players merchandising potential. A long-term contract provides a player with insurance
against loss of form or injury. Consequently clubs were keen to tie their
best players in long-term arrangements, and many players were keen to
extend contracts before they expired, without reaching the stage of
becoming free agents. At present, one can only wonder how these mutual
interests in stability and security would be served if European law does
eventually dictate that something similar to that a spot market in playing
talent must operate in football. Experience suggests, however, that if a
need is suciently strong, new forms of contractual arrangement will
develop capable of satisfying that need within whatever parameters the
law imposes.
Meanwhile lower-division clubs (in particular) will worry about
the implications of the possible loss of transfer income, especially if the
432
433
lems are sending signals mainly about the shortcomings of their own
management (in the broad sense) rather than signals that the sector as a
whole is facing imminent ruin.
What about spiralling player wages? Should nothing at all be done to
bring these under control? For reasons that are not entirely clear the
inated earnings of footballs top performers do seem to rankle with the
public to a greater extent than those of comparable groups such as pop
musicians and lm stars.The economic analysis, however, regards spiralling wages very much as a consequence rather than a cause of the football
sectors changing economic structure. This means that at the aggregate
level there is a good deal of economic determinism in what is happening.
Ultimately the top players are the monopoly suppliers of the performances the public most wants to see and is willing to pay to see. If restrictions on the players ownership of the property rights to their own
performances are removed, and if intermediate links in the supply chain
(such as clubs and broadcasters) are subject to the full rigours of competition, then it is inevitable that most of the rent generated by the sale of the
rights to view these performances will eventually accrue to the players.
Could a salary cap be used to impose some restraint on wage ination?
Such a measure would presumably have to operate along the lines currently employed in basketball and football in the USA, limiting clubs
total wages expenditure to a set percentage of designated revenues, and
not on the lines of English footballs original maximum wage that capped
the wage payable to any individual player. As already seen, in the USA
such restrictions are designed primarily to inuence the distribution of
playing talent so as to enhance competitive balance, rather than to
restrain growth in wages directly (though this may also be a consequence). In the USA there are serious doubts about the eectiveness of
salary capping. But in Europe it appears even less likely that similar measures could work in practice. Quite apart from the obvious incentives for
clubs and players to cheat, and the probable extreme scepticism of the
European courts, the openness of the players labour market means a cap
would need to apply throughout Europe to be acceptable or eective.
Clearly it would be dicult or impossible to achieve agreement, let alone
monitor compliance, given the sheer number of parties and disparate
range of interests involved.
In any case, economic reasoning suggests that achieving restraint over
player wage ination is not necessary in order to secure the nancial survival of the football industry. If the stratospheric earnings of icons such as
Keane, Beckham, Figo and del Piero are really so distasteful or socially
divisive that they demand some form of intervention, that is a dierent
matter, and one that should ultimately be addressed through the tax
434
system. But since there is no sign of any public or political appetite for the
reinstatement of scal regimes signicantly more egalitarian than those in
force at present, it seems a little disingenuous to castigate players and
clubs for simply acting rationally in pursuit of their private interests
within the connes of existing, democratically sanctioned structures.
In view of the previous analysis, the emergence in recent years of a consistent pattern of protability among a handful of leading English clubs
requires some further comment. In theoretical terms, it can be explained
by qualifying slightly the conclusion that market power on the supply side
of the football industry rests predominantly with the players as monopoly
sellers in the labour market. Manchester United in particular, as well as a
handful of other clubs, obviously have highly marketable brand identities themselves, that exist irrespective of the identities of the players currently appearing in their rst teams. As monopoly owners of the
merchandising rights to these identities embodied in club insignia and
logos, the clubs can appropriate monopoly rents from the public through
the sale of these property rights. Lower down the leagues hierarchy,
however, although club brand loyalties among their own supporters are
just as strong (if not stronger), they are too localised to be of much value
to merchandisers reliant on economies of scale. Manchester United has
obviously been more successful than any other club worldwide in exploiting merchandising as a source of revenue. In view of the adulation the
club apparently attracts in countries such as China and Malaysia, the
future commercial potential of global merchandising may be very great
indeed, if still largely unquantied and untapped.
The top clubs brand identities may also help to counter-balance
some of the monopoly power enjoyed by the top players as sellers in the
labour market. Apart from the pecuniary rewards, the oer of a contract
to play for Manchester United (or Juventus, Real Madrid or Barcelona)
has intangible value, in terms of the prestige or the sense of having
reached the pinnacle of ones career, that it confers upon even the most
highly paid superstars. At the very top of the profession, just as the
number of individuals that compete directly with the likes of Zidane, Figo
and Beckham as suppliers of playing talent is strictly nite, so too is the
number of clubs that compete directly with Juventus, Real Madrid and
Manchester United as buyers. As there are limits to competition on both
sides of the labour market, both players and clubs can expect to earn
monopoly rents. Moving down the league hierarchy, the numbers of clubs
of similar status and players of similar ability both increase rapidly, but
the intensity of arms race competition between clubs as buyers in the
labour market appears to increase faster than the intensity of economic
competition between players as sellers. Consequently lower-level players
435
still tend to enjoy monopoly rents (earnings above their reservation levels)
while the clubs that employ them consistently turn in losses.
Ownership, control and media interests in football
As noted previously, the high protability of Manchester United is a
matter of some controversy among its own supporters, and one that also
raises more general questions about the suitability of alternative models
of ownership, nancial management and corporate governance for football clubs. Manchester Uniteds 1999 season turnover of 111 million
was larger than that of any other club in the world.Yet partly because of
the clubs duty to control its expenditure so as to full its responsibilities
to its shareholders, United experiences diculties in competing with the
leading Italian and Spanish clubs to attract the worlds top players.
At the close of the 2000 season the contrasting styles of nancial management of Manchester United and Real Madrid attracted attention in
several articles in the UK nancial press (Financial Times, 22 July 2000,
18 August 2000; Scotland on Sunday, 30 July 2000; The Independent, 5
August 2000). In contrast to Uniteds status as a publicly limited
company, Real is mutually owned by around 50,000 members, mostly
ordinary supporters, who elect a president to run the club. A promise to
sign Portuguese star Luis Figo from Barcelona, fullled for a world record
transfer fee of 37 million in the summer of 2000, was one of the key
pledges of recently installed Real president Florentino Lopezs election
campaign. Despite having a 1999 season budget of around 72 million,
Real earns little or no prot, and has reported debts anywhere between
75 million and 180 million. The club maintains close links with its
bankers, however, who are unlikely to withhold access to further lending
owing to Reals status as a national institution.
From the supporters perspective, the most telling comparisons
concern expenditures on players. In a survey of the wages of the worlds
highest paid players (World Soccer, November 1999) Reals three highest
earners were McManaman (65,000 per week), Anelka (56,000) and
Roberto Carlos (40,000). Uniteds three highest were Beckham, Keane
and Giggs (all on 17,500 per week).The survey was carried out shortly
before Keanes new contract was concluded, but the contrast is plain
enough. At the start of the 2000 season, only ve of Reals 27 rst-team
squad members were home-grown; the rest had been signed from other
clubs. United had 16 home-grown players in a squad of 34 (Macas,
2000). In the summer of 2000, Reals record transfer signings were Figo
(for 37 million from Barcelona) in 2000 and Anelka (22.5 million from
Arsenal) in 1999. Uniteds record signings were Yorke (for a relatively
436
modest 12.6 million from AstonVilla) and Stam (10 million from PSV
Eindhoven) both in 1998.
Nevertheless, both models of ownership and nancial management are
subject to pressures for change within a rapidly changing commercial
environment. Under their new presidency, Real Madrid is seeking to
reduce losses and to develop previously under-exploited sources of
revenue such as merchandising. The idea of stock exchange otation has
already been mooted by rivals Barcelona. In England, it is signicant that
BSkyB have now replaced the Edwards family as the largest single shareholders in Manchester United. Media group interests in English football
clubs are constrained by an FA regulation restricting any party with interests in more than one club to shareholdings below 10 per cent in each
case. Nevertheless BSkyB has acquired shareholdings up to the 10 per
cent limit in Manchester United, Leeds United, Chelsea, Sunderland and
Manchester City. Granada Media has acquired a similar stake in
Liverpool, while cable operator NTL has investments in Newcastle
United, Middlesbrough, Aston Villa and Leicester City. It is easy to
predict that further vertical integration and joint ventures between broadcasters and football clubs will take place in the future, and that the interface between football and television will continue to be a major
battleground in which many of the issues inuencing footballs future
development will be fought out and decided.
As already seen, the role played by technological innovation, diusion
and competition in broadcasting in determining the size and structure of
professional footballs eective audience cannot be emphasised too
strongly. Highly signicant for the future is the fact that while BSkyB
enjoyed considerable monopoly power as the dominant supplier of paytelevision services through most of the 1990s, at the time of writing this
phase seems to be drawing rapidly to a close. Soon after the 19927 and
19972001 television contracts were signed, each providing huge windfall gains to the football industry, analysts were suggesting that relative to
the amounts that could theoretically have been raised, both deals might
actually have been under-priced. In large measure this was because
BSkyB was the only bidder with the necessary pay-television technology
and infrastructure in position on both occasions. The outcome of the
third sale of televised broadcasting rights since the formation of the
Premier League covering the 2002, 2003 and 2004 seasons, which was
announced in June 2000, demonstrated that BSkyBs monopoly position
has already been considerably eroded.
For a total fee of 1.1 billion BSkyB did succeed in ensuring the continuation of its contract to screen 66 live Premier League matches per season.
It was initially intended, however, that the exclusivity of BSkyBs live
437
coverage would not continue, however. Originally the cable operator NTL
secured a contract to screen another 40 matches on a pay-per-view basis
for 328 million. Under the agreement, the two companies would not
screen the same match simultaneously, and BSkyB would have priority in
selecting its xtures. A feeling among commentators that the demand for
pay-per-view football in Britain might not be sucient to sustain the NTL
contract appeared to be conrmed several weeks after the initial
announcement, when NTL gave notice of its withdrawal. An alternative
buyer for pay-per-view had not emerged at the time of writing, and it
remains to be seen whether any further components of the June 2000
package will also unravel. For 183 million ITV secured the rights to
screen the Saturday evening highlights package. BBCs consolation was a
contract with the Football Association for joint coverage of the FA Cup
and England internationals shared with BSkyB, for which the joint fee was
340 million. Finally for 315 million the terrestrial digital broadcaster
ONdigital secured the rights to screen live Football League matches.
The 20014 contracts undoubtedly point towards a future in which
heightened competition between broadcasters will reduce the returns to
the television companies, and raise the football industrys proceeds from
the sale of broadcasting rights, perhaps far beyond even the levels attained
during the 1990s. From the television companies perspective, the
winners curse in sealed-bid auctions of the kind that operated in 2000
has been widely debated. By denition, the successful bidder is the
company that made the highest assessment of the value of the package
under auction. Given that such assessments are uncertain and prone to
error, the successful bidder is the one most likely to have over-estimated
the true value, and consequently faces a high risk of realising a loss on the
transaction. Certainly the Citys reaction to BSkyBs success, reected in
the companys share price, was less than euphoric. Just before the June
2000 announcement, with speculation mounting that BSkyB might lose
the Premier League contract altogether, BSkyB shares were trading more
than 20 per cent below than their mid-April value. Most of this loss was
recovered immediately after the announcement, but by mid-August the
price had drifted down again, to around 15 per cent below the mid-April
value. Anguish at the BBC over the demise of its Match of the Day highlights programme may have been tempered slightly by the knowledge that
ITVs expenditure was far too high ever to be recouped directly via
increased advertising revenues on Saturday evenings.
Beyond 2004 Internet broadcasting seems likely to play a signicant
role in the relationship between football clubs and broadcasting companies. Although the technology was not in place to permit transmission of
pictures of suciently high denition in time for Internet broadcasting
438
439
9.3 Conclusion
The main implications of the arguments developed in this concluding
chapter are that in the future, broadcasting companies will tend to act
increasingly as a conduit through which revenues are channelled from the
public as nal consumers to the football clubs. In turn, erce competition
between the clubs as employers in a world of player free agency will
ensure, as is already happening today, that most of the proceeds are
passed on directly to the players, as the ultimate monopoly suppliers of
the performing services the public wishes to view. As monopoly owners
themselves of the property rights in marketable brand identities that
exist independently of the players they employ, the top clubs will be able
to generate sucient prots at least to satisfy the requirements of their
shareholders and other investors. Lower down the league hierarchy,
where the sizes of the markets in club brand identities are too small for
merchandisers and others to realise economies of scale, any rents accruing from the sale of broadcasting rights will be tend to be dissipated and
passed on to players as a result of the impact of arms race competition.
At this level, most clubs will continue to realise whatever losses their
owners are prepared to tolerate, as has typically been the case in the past.
From an economists perspective all of this is entirely consistent with
the predictions of theoretical models of optimising behaviour on the part
of individuals, participating socially as buyers and sellers in markets.
Whether the kinds of changes unleashed by market forces in the highlycharged world of modern-day professional football are good or bad is a
normative question, of a kind that economists are usually ill-equipped or
unwilling to address; indeed, many economists would not even accept
that the question is meaningful. Once the genie of market forces has
escaped from the bottle, economists may be as powerless as anyone else to
do much to inuence the consequences, in sports as in other walks of life.
But if economists can help clarify the distinction between the underlying
causes of human actions and the intended or unintended consequences
that follow, they will at least have made some contribution to our understanding.
References
Aigner, D.J., Lovell, C.A.K. and Schmidt, P.J. (1977) Formulation and estimation
of stochastic frontier models, Journal of Econometrics, 6, 2137
Alchian, A.A. and Demsetz H. (1972) Production, information costs and economic organisation, American Economic Review, 62, 77795
Ali, M.M. (1977) Probability and utility estimates for racetrack bettors, Journal of
Political Economy, 85, 80315
Allen, M.P., Panian, S.K. and Lotz, R.E. (1979) Managerial succession and
organizational performance: a recalcitrant problem revisited, Administrative
Science Quarterly, 24, 16780
Atkinson, S.E., Stanley, L.R. and Tschirhart, J. (1988) Revenue sharing as an
incentive in an agency problem: an example from the National Football
League, RAND Journal of Economics, 19, 2743
Audas, R., Dobson, S.M. and Goddard, J.A. (1997) Team performance and managerial change in the English Football League, Economic Aairs, 17, 306
(1999) Organizational performance and managerial turnover, Managerial and
Decision Economics, 20, 30518
Baade, R. (1996) Professional sports as catalysts for metropolitan economic
development, Journal of Urban Aairs, 18, 117
Baade, R. and Dye, R. (1990) The impact of stadiums and professional sports on
metropolitan area development, Growth and Change, 21, 117
Badarinathi, R. and Kochman, L. (1996) Football betting and the ecient market
hypothesis, American Economist, 40, 525
Baimbridge, M., Cameron, S. and Dawson, P.M. (1996) Satellite television and
the demand for football: a whole new ball game?, Scottish Journal of Political
Economy, 43, 31733
Bale, J. (1993) Sport, Space and the City, London: Routledge
Ball, R. and Brown, P. (1968) An empirical evaluation of accounting income
numbers, Journal of Accounting Research, 15978
Barnett,V. and Hilditch, S. (1993) The eect of an articial pitch surface on home
team perfromance in football (soccer), Journal of the Royal Statistical Society
Series A, 156, 3950
Barro, R.J. (1991) Economic growth in a cross section of countries, Quarterly
Journal of Economics, 104, 40744
Baumol,W.J. (1959) Business Behaviour,Value and Growth, New York: Macmillan
(1986) Productivity growth, convergence and welfare: what the long-run data
show, American Economic Review, 76, 107285
440
List of references
441
442
List of references
List of references
443
444
List of references
Fama, E., Fisher, L., Jensen, M. and Roll, R. (1969) The adjustment of stock
prices to new information, International Economic Review, 10, 121
Ferguson, D.G., Stewart, K.G., Jones, J.C.H. and le Drassay, A. (1991) The
pricing of sports events: do teams maximize prot?, Journal of Industrial
Economics, 39, 297310
Ferrall, A. and Smith, A.A. Jr. (1999) A sequential game model of sports championship series: theory and estimation, Review of Economics and Statistics, 81,
70419
Finn, G.P.T. (2000) Scottish myopia and global prejudices, in G.P.T. Finn and R.
Giulianotti (eds.), Football Culture: Local Contests, Global Visions, London:
Frank Cass
Fishwick, N. (1989) English Football and Society, 19101950, Manchester:
Manchester University Press
Fizel, J.L. and DItri, M. (1996) Estimating managerial eciency: the case of
college basketball coaches, Journal of Sport Management, 10, 43545
(1997) Managerial eciency, managerial succession and organizational performance, Managerial and Decision Economics, 18, 295308
Football Association (1991) The Blueprint for the Future of Football, London:
Football Association
Football Trust (various issues) Digest of Football Statistics, Leicester: University of
Leicester
Fort, R. (2000) European and North American sports dierences(?), Scottish
Journal of Political Economy, 47, 43155
Fort, R. and Quirk, J. (1995) Cross-subsidization, incentives and outcomes
in professional team sports leagues, Journal of Economic Literature, 33,
126599
Foster, K. (2000) European law and football: whos in charge? in J. Garland, D.
Malcolm and M. Rowe (eds.), The Future of Football: Challenges for the Twentyrst Century, London: Frank Cass
Fynn, A. and Guest, L. (1994) Out of Time:Why Football isnt Working, London:
Simon & Schuster
Gamson, W.A. and Scotch, N.A. (1964) Scapegoating in baseball, American
Journal of Sociology, 70, 6972
Gandar, J.M., Dare,W.H., Brown, C.R. and Zuber, R.A. (1998) Informed traders
and price variations in the betting market for professional basketball games,
Journal of Finance, 53, 385401
Gandar, J.M., Zuber, R.A., OBrien,T. and Russo, B. (1988) Testing rationality in
the point spread betting market, Journal of Finance, 43, 9951008
Gibrat, R. (1931) Les ingalits conomiques, Paris: Librairie du Recueil Sirey
Goldberg, A. and Wragg, S. (1991) Its not a knock-out: English football and globalisation, in J. Williams and S. Wragg (eds.), British Football and Social
Change: Getting Into Europe, Leicester: Leicester University Press
Golec, J. and Tamarkin, M. (1991) The degree of ineciency in the football
betting market: statistical tests, Journal of Financial Economics, 30, 31123
(1998) Bettors love skewness, not risk, at the horse track, Journal of Political
Economy, 106, 20525
(1999) Econometric Analysis, 4th edn., Englewood Clis, New Jersey: PrenticeHall
List of references
445
Grier, K.B. and Tollison, R.D. (1994) The rookie draft and competitive balance:
the case of professional football, Journal of Economic Behaviour and
Organization, 25, 2938
Grusky, O. (1963) Managerial succession and organizational eectiveness,
American Journal of Sociology, 69, 2131
(1964) Reply to scapegoating in baseball, American Journal of Sociology, 70,
726
Guilianotti, R. (1999) Football:A Sociology of the Global Game, Cambridge: Polity
Gujarati, D.N. (1995) Basic Econometrics, 3rd edn., New York: McGraw-Hill
Hallam, D. and Machado, F. (1996) Eciency analysis with panel data: a study of
Portuguese dairy farms, European Review of Agricultural Economics, 23, 7993
Hamilton, B.H. (1997) Racial discrimination and professional basketball salaries
in the 1990s, Applied Economics, 29, 28796
Hart, P.E. and Prais, S.J. (1956) The analysis of business concentration: a statistical approach, Journal of the Royal Statistical Society Series A, 119, 15091
Hart, R.A., Hutton, J. and Sharot, T. (1975) A statistical analysis of association
football attendances, Applied Statistics, 24, 1727
Hatanaka, M. (1974) An ecient estimator for the dynamic adjustment model
with autocorrelated errors, Journal of Econometrics, 2, 199220
Hausman, J.A. and Leonard, G.K. (1997) Superstars in the National Basketball
Association: economic value and policy, Journal of Labour Economics, 15,
586624
Henerey, R.J. (1985) On the average probability of losing bets on horses with given
starting price odds, Journal of the Royal Statistical Society Series A, 148, 3429
Hill, I.D. (1974). Association football and statistical inference, Applied Statistics,
23, 2038
Hill, J.R., Madura, J. and Zuber, R.A. (1982) The short run demand for major
league baseball, Atlantic Economic Journal, 10, 315
Hoehn, T. and Szymanski S. (1999) The Americanization of European football,
Economic Policy, 28, 20540
Hoer, R.A. and Payne, J.E. (1996) How close to their oensive potential do
National Football League teams play?, Applied Economics Letters, 3, 7437
(1997) Measuring eciency in the National Basketball Association, Economics
Letters, 55, 2939
Horne, J. (1996) Sakka in Japan, Media, Culture and Society, 18, 52747
(2000) Soccer in Japan: is wa all you need? in G.P.T. Finn and R. Giulianotti
(eds.), Football Culture: Local Contests, GlobalVisions, London: Frank Cass
Horowitz, I. (1994) On the manager as principal clerk, Managerial and Decision
Economics, 15, 41319
Horton, E. (1997) Moving the Goalposts: Footballs Exploitation, Edinburgh:
Mainstream
Hugman, B.J. (1998) The PFA Premier and Football League Players Records
19461998, Harpenden: Queen Anne Press
Hunt, J. and Lewis, K. (1976) Dominance, recontracting and the reserve clause:
major league baseball, American Economic Review, 66, 93643
Jacobs, D. and Singell, L. (1993) Leadership and organizational performance: isolating links between managers and collective success, Social Science Research,
22 16589
446
List of references
List of references
447
MacDonald, D.N. and Reynolds, M.O. (1994) Are baseball players paid their
marginal products? Managerial and Decision Economics, 15, 44357
MacDonald, G. (1988) The economics of rising stars, American Economic Review,
78, 15566
MacKinlay, A.C. (1997) Event studies in economics and nance, Journal of
Economic Literature, 35, 1339
Macas, P. (2000) AZ of European Football, Cceres: Copegraf, S.L.
Maguire, J. (1994) Preliminary observations on globalisation and the migration of
sport labour, Sociological Review, 42, 45280
Maguire, J. and Pearton, R. (2000) Global sport and the migration patterns of
France 98 World Cup nals players: some preliminary observations, in J.
Garland, D. Malcolm and M. Rowe (eds.), The Future of Football: Challenges
for the Twenty-rst Century, London: Frank Cass
Maguire, J. and Stead, D. (1998) Border crossings: soccer labour migration and
the European Union, International Review for the Sociology of Sport, 33,
5973
Maher, M.J. (1982) Modelling association football scores, Statistica Neerlandica,
36, 10918
Malcolm, D., Jones, I. and Waddington, I. (2000) The peoples game? Football
spectatorship and demographic change, in J. Garland, D. Malcolm and M.
Rowe (eds.), The Future of Football: Challenges for the Twenty-rst Century,
London: Frank Cass
Marris, R.L. (1964) The Economic Theory of Managerial Capitalism, London:
Macmillan
Michie, J. and Oughton, C. (1999) Football and broadcasting and the MMC
case, in S. Hamil, J. Michie and C. Oughton (eds.), The Business of Football:A
Game of Two Halves, Edinburgh: Mainstream
Mignon, P. (2000) French football after the 1998 World Cup: the state and the
modernity of football, in G.P.T. Finn and R. Giulianotti (eds.), Football
Culture: Local Contests, GlobalVisions, London: Frank Cass
Mintzberg, H. (1973) The Nature of ManagerialWork, New York: Prentice-Hall
Mises, L. von (1966) Human Action, 3rd edn., Chicago: Contemporary Books
Monopolies and Mergers Commission (1999) British Sky Broadcasting Group
PLC and Manchester United PLC: A Report on the Proposed Merger, London:
TSO, Cm 4305
Moorhouse, H.F. (1984) Professional football and working class culture: English
theories and Scottish evidence, The Sociological Review, 32, 285315
Moran, R. (2000) Racism in football: a victims perspective, in J. Garland, D.
Malcolm and M. Rowe (eds.), The Future of Football: Challenges for the Twentyrst Century, London: Frank Cass
Moroney, M.J. (1956) Facts from Figures, 3rd edn., London: Penguin
Morrow, S. (1999) The New Business of Football: Accountability and Finance in
Football, London: Macmillan
Murray, B. (1984) The Old Firm: Sectarianism, Sport and Society in Scotland,
Edinburgh, John Donald
(1998) The Old Firm in the New Age: Celtic and Rangers since the Souness
Revolution, Edinburgh, Mainstream
Neale, W.C. (1964) The peculiar economics of professional sports, Quarterly
Journal of Economics, 78, 114
448
List of references
Nogawa, H. and Maeda, H. (1999) The Japanese dream: soccer culture towards
the new millennium, in G. Armstrong and R. Giulianotti (eds.), Football
Cultures and Identities, Basingstoke: Macmillan
Noll, R.G. (1974) Attendance and price setting, in R.G. Noll (ed.), Government
and the Sports Business, Washington, DC: Brookings Institution
Noll, R.G. and Zimbalist, A. (1997) Sport, Jobs and Taxes, Washington, DC:
Brookings Institution
Panko, L.D. (1968) Market eciency and football betting, Journal of Business,
41, 20314
Peel, D.A. and Thomas, D. (1988) Outcome uncertainty and the demand for football, Scottish Journal of Political Economy, 35, 2429
(1992) The demand for football: some evidence on outcome uncertainty,
Empirical Economics, 17, 32331
PEP (1966) English professional football, Planning, 32 (496)
Pickup, I. (1999) French football from its origins to 1984, in H. Dauncey and G.
Hare (eds.), France and the 1998World Cup, London: Frank Cass
Pindyck, R.S. and Rubinfeld, D.L. (1998) Econometric Models and Economic
Forecasts, 4th edn., Boston: McGraw-Hill
Pope, P.F. and Peel, D.A. (1989) Information, prices and eciency in a xed odds
betting market, Economica, 56, 32341
Porter, R. and Scully, G.W. (1982) Measuring managerial eciency: the case of
baseball, Southern Economic Journal, 48, 64250
Preston, I. and Szymanski, S. (2000) Racial discrimination in English football,
Scottish Journal of Political Economy, 47, 34163
Quirk, J. and El-Hodiri, M. (1974) The economic theory of a professional league,
in R.G. Noll (ed.), Government and the Sports Business, Washington, DC:
Brookings Institution
Quirk, J. and Fort, R. (1992) Pay Dirt: The Business of Professional Team Sports,
Princeton: Princeton University Press
(1999) Hard Ball:The Abuse of Power in Pro Team Sports, Princeton: Princeton
University Press
Rascher, D.A. (1997) A model of a professional sports league, in W. Hendricks
(ed.), Advances in the Economics of Sport, 2, Greenwich: JAI Press
Reep, C., Pollard, R. and Benjamin, B. (1971) Skill and chance in ball games,
Journal of the Royal Statistical Society Series A, 131, 5815
Reilly, B. and Witt, R. (1995) English league transfer prices: is there a racial
dimension?, Applied Economics Letters, 2, 2202
Richards, D.G. and Guell, R.C. (1998) Baseball success and the structure of salaries, Applied Economics Letters, 5, 2916
Ridder, G., Cramer, J.S. and Hopstaken, P. (1994) Estimating the eect of a red
card in soccer, Journal of the American Statistical Association, 89, 11247
Robbins, S.P. (1994) Management, 4th edn., Engleeld Clis, New Jersey:
Prentice-Hall
Rosen, S. (1981) The economics of superstars, American Economic Review, 71,
84558
Rosen, S. and Sanderson, A. (2000) Labour markets in professional sports,
National Bureau of Economic Research, Working Paper, 7573, Cambridge,
Mass.: National Bureau of Economic Research
List of references
449
450
List of references
Singell, L.D. (1993) Managers, specic human capital and productivity in Major
League Baseball, Atlantic Economic Journal, 21, 4759
Sloane, P. (1969) The labour market in professional football, British Journal of
Industrial Relations, 7, 18199
(1971) The economics of professional football: the football club as utility maximiser, Scottish Journal of Political Economy, 17, 12146
Smailes, G. (1992) The Breedon Book of Football League Records, Derby: Breedon
Solow, R.M. (1956) A contribution to the theory of economic growth, Quarterly
Journal of Economics, 70, 6594
Speight, A. and Thomas, D. (1997) Arbitrator decision-making in the transfer
market: an empirical analysis, Scottish Journal of Political Economy, 44,
198215
Sugden, J. and Tomlinson, A. (1998) FIFA and the Contest for World Football,
Cambridge: Polity
Sutton, J. (1997) Gibrats legacy, Journal of Economic Literature, 35, 4059
Szymanski, S. (2000a) Hearts, minds and the Restrictive Practices Court case, in
S. Hamil, J. Michie, C. Oughton and S. Warby (eds.), Football in the Digital
Age:Whose Game is it Anyway?, Edinburgh: Mainstream
(2000b) A market test for discrimination in the English professional soccer
leagues, Journal of Political Economy, 108, 590603
Szymanski, S. and Kuypers,T. (1999) Winners and Losers:The Business Strategy of
Football, London:Viking
Szymanski, S. and Smith, R. (1997) The English football industry: prot, performance and industrial structure, International Review of Applied Economics, 11,
13553
Tabner, B. (1992) Through the Turnstiles, Hareeld:Yore
Taylor, A.J.P. (1965) English History: 19141945, Oxford: Oxford University Press
Taylor, Lord Justice (1990) The Hillsborough stadium disaster (15 April 1989),
inquiry by the Rt. Hon. Lord Justice Taylor, Final Report, Cm. 962, London:
HMSO
Taylor, R. and Ward, A. (1995) Kicking and Screaming: An Oral History of Football
in England, London: Robson
Temple, J. (1999) The new growth evidence, Journal of Economic Literature, 37,
11256
Terrell, D. and Farmer, A. (1996) Optimal betting and eciency in parimutul
betting markets with information costs, Economic Journal, 106, 84668
Turner, D. and White, A. (1993) The Breedon Book of Football Managers, Derby:
Breedon Books
Vaughan Williams, L. and Paton, D. (1997) Why is there a favourite longshot bias
in British racetrack betting markets?, Economic Journal, 107, 1508
Vrooman, J. (1995) A general theory of professional sports leagues, Southern
Economic Journal, 61, 97190
(1996) The baseball players marker reconsidered, Southern Economic Journal,
63, 33960
(1997) A unied theory of capital and labour markets in major league baseball,
Southern Economic Journal 63, 594619
(2000)The economics of American Sports leagues, Scottish Journal of Political
Economy, 47, 36498
List of references
451
Index
Aberdeen 51, 65
AC Milan 24, 30, 51, 86, 426
admission price 12, 15, 22, 30, 7180,
91, 123, 245, 318, 320, 3269,
34272
Ajax Amsterdam 427
Aldershot 271
Al-Fayed, Mohammed 380
all-seated stadia 56, 6971, 124, 317, 320,
342
American Basketball Association 130
American Broadcasting Company (ABC)
130
American football 2, 14, 12631, 1478,
225, 2502, 253, 2812, 3045, 318,
4012, 433
American Football League 130
American League 6, 12631, 147, 222
American Professional Football Association
130
Amsterdam Treaty 430
Amstrad 83
Anelka, Nicolas 104, 105, 435
Ardiles, Osvaldo 108
Arnold,Thomas 38
Arsenal 39, 42, 69, 80, 82, 83, 86, 104,
1837, 197, 209, 243, 259, 265, 323,
324, 347, 354, 377, 41416, 420, 421,
435
Aston Villa 40, 59, 65, 66, 83, 86, 259, 352,
380, 382, 385, 414, 420, 436
Athletic Club de Bilbao 3378
Atkinson, Ron 272
Atlantic League 51, 4278
Atletico Madrid 30
attendence 3, 8, 9, 13, 145, 20, 22, 24, 30,
5671, 73, 79, 80, 91, 104, 108, 114,
123, 148, 178, 211, 216, 224, 226,
234, 245, 289, 31772, 424
Australian Rules Football 321, 332
Austria 70, 205, 208
Austrian economics 240
452
Index
Bosman ruling 12, 23, 31, 9599, 104, 109,
189, 190, 206, 231, 430, 431
Bournemouth 267
Bradford City 69, 123
Bradford Park Avenue 352
Brazil 108
Brighton and Hove Albion 432
Bristol City 68, 271
British Broadcasting Corporation (BBC)
804, 215, 334, 424, 437
British Satelite Broadcasting (BSB) 80, 81,
82
British Sky Broadcasting (BSkyB) 30, 80,
8287, 105, 334, 377, 381, 386, 394,
397, 417, 424, 4368
Brndby 426
Brooklyn Dodgers 130
Brother Walfrid 340
Bruce, Steve 272
Buckley, Major Frank 2434
Burnley 43
Byers, Stephen 84, 386
Cambridge United 53, 269, 289
Cambridge University 39
Canal Plus 86, 105
Cantona, Eric 96
Carbone, Benito 205
Carlisle United 272, 289
Carlos, Roberto 435
cartel 7, 81, 85, 424
causality tests 321
Celtic 49, 51, 65, 273, 318, 333, 335,
33942, 372, 427
Central League 106
Champions League 41, 4256, 427
Channel 4, 80
Channel 5, 80
Chapman, Herbert 42, 243
Charlton Athletic 55, 196, 289, 382, 384,
394, 397, 4146, 429, 432
Chelsea 30, 59, 65, 69, 83, 86, 141, 196,
209, 259, 380, 382, 385, 386, 436, 438
China 434
classical economics 240
Clough, Brian 265
Club Franais 102
CobbDouglas production function 251
Cohen, George 421
cointegration 3289
Colchester United 53
collusion 4, 136
Columbia Broadcasting System (CBS) 130
competitive balance 3, 9, 11, 12, 16, 42, 86,
91, 12549, 150, 153, 187, 214, 407,
41924, 428, 430, 433
453
constant returns to scale 246
convergence 12, 20, 23, 10922, 152
CournotNash equilibrium 1434, 218
Coventry City 352, 4146
Crewe Alexandra 89, 265
cricket 6, 250
Croatia 31, 205
Cruy, Johan 227, 335
Crystal Palace 49, 55, 197, 271, 352, 423,
432
Czech Republic 208
Daley, Steve 93
Dalglish, Kenny 42, 267
data envelopment analysis 283
del Piero, Alessandro 35, 433
Denmark 208, 427
Deportivo La Corua
Derby County 43, 354, 4146
de Rivera, General Primo 336
deterministic production frontier see production frontier
di Canio, Paulo 205
di Matteo, Roberto 205
di Stefano, Alfredo 337
discrete choice model 20, 228, 283
discrimination 10, 11, 190, 2237, 228,
239, 334
Douglas, Lieutenant Commander
Archibald 106
draft 129, 136, 1478, 253, 422
Dreyfus, RobertLouis 105
Dundee United 65
Dunga 108
Dunkerque 31
duration analysis 284302
hazard function 20, 264, 284302, 315
survivor function 283, 285, 298
Eastham, George 92, 421
economies of scale 2125, 224, 434, 439
Edwards family 436
ecient markets hypothesis 156,
373417
elasticity of demand
income 327
price 320, 327, 329, 353, 371
Ellis, Doug 380
Espaol see Espanyol
Espanyol 3368
European Championship 65, 104, 203,
2069
European Commission 98, 99, 101, 190,
227, 430
European Court of Justice (ECJ) 23, 31,
95, 98, 189, 430
454
Index
Index
invariance result 136, 147, 148, 422
Ipswich Town 43, 55, 196, 354
Ireland 31, 195, 2023, 237, 238, 33941
iso-prot curve 1426
isoquant 2467
Italy 22, 2438, 4351, 68, 87, 92, 101,
104, 127, 195, 201, 205, 206, 208,
209, 237, 238, 420, 427, 435, 438
J.League 102, 1078
Jackson, Peter 272
Japan 11, 23, 102, 1058
Japanese Football Association 106
Japan Soccer League 1067
JEF United Ichihara 108
Johnson, Magic 148
Johnston, Maurice 341
Jordan, Michael 148
Jupiter 336
Juventus 30, 35, 51, 68, 227, 426, 434
Kaiserslauten 51
Kamper, Hans see Gamper, Joan
Kashima, Antlers 108
Keane, Roy 421, 433, 435
Keegan, Kevin 146, 259, 267, 272
Kendall, Howard 267
Kerr, George 269
King Alphonso XIII 336
Kinnear, Joe 272
Kln 65
Lagardre, Jean-Luc 105
Lazio 30, 51
League Cup 40, 42, 90
League Managers Association 297
Leeds United 37, 59, 83, 86, 96, 380, 382,
385, 436
Leicester City 86, 259, 272, 273, 380, 382,
41416, 421, 428, 436, 438
Lens 30
Lige 31
Lincoln City 267
linear probability model 2823, 4034, 410
Lineker, Gary 108
Litbarski, Pierre 108
Little, Brian 272
Liverpool 30, 39, 42, 43, 49, 67, 68, 69, 82,
83, 86, 92, 95, 209, 265, 267, 326,
352, 414, 421, 426, 436, 438
logit 282, 404
London Stock Exchange (LSE) 15, 373,
37697, 419
Alternative Investment Market (AIM)
379, 382
Ocial List 376, 379, 382
455
London Weekend Television 81
Lopez, Florentino 435
Louis Schmelling paradox 5
Luton Town 196
Lyon 30, 86
Major League Baseball see baseball
Malaysia 334, 434
managers see football managers
Manchester City 59, 86, 352, 436
Manchester United 24, 30, 38, 39, 41, 42,
43, 49, 65, 68, 69, 82, 83, 846, 87, 90,
92, 95, 96, 146, 196, 197, 209, 212,
259, 265, 267, 324, 347, 3767, 379,
380, 381, 382, 385, 386, 388, 389,
394, 397, 417, 420, 422, 426, 427,
434, 435, 436, 438, 439
Mandelson, Peter 84
Maradona, Diego 227, 335
Marinos 108
Marseille 30, 49, 65, 104, 105
match results see football match results
Mathias, Ray 272
maximum bid price 22931, 234, 239
maximum likelihood (ML) 18, 170, 249,
254, 2579, 286
maximum wage 11, 13, 79, 903, 109, 123,
135, 189, 418, 421, 433
McCann, Fergus 342
McGhee, Mark 272
McManaman, Steve 35, 435
McNeil family 339
Media Partners 425
merchandising 15, 22, 30, 38, 71, 72, 87,
96, 107, 108, 227, 418, 424, 431, 434,
436, 439
Merchant Taylors public school 335
Mexican League 130
Middlesbrough 55, 68, 86, 1837, 271,
41416, 429, 436
migration 13, 189, 191, 2039, 238, 339
Millwall 196, 376, 432
minimum admission price see admission
price
Minor League Baseball see baseball
Mitsubishi Motors 108
Monaco 30
Monopolies and Mergers Commision
(MMC) 845, 381, 386, 397
monopoly 3, 5, 7, 86, 127, 128, 131, 222,
2301, 234, 239, 422, 433, 434, 436,
439
monopsony 3, 4, 6, 220, 222, 225, 421
Moore, Bobby 421
moral hazard 241
Murray, David 342
456
Index
Nantes 65
Napoli 36, 51, 227
National Basketball Association (NBA)
126131, 147, 1489, 225, 250, 251,
334, 402
National Broadcasting Company (NBC)
130
National Collegiate Athletic Association
(NCAA) 147148
National Football Institute 103
National Football League (NFL) 12631,
147, 251, 4012
National Hockey League (NHL) 145
National League (NL) 6, 12631, 147, 222
National Transcommunications Limited
(NTL) 84, 86, 436, 437, 438
natural monopoly 6, 7, 213
negative binomial distribution 170, 404
neoclassical economics 15, 240, 281, 373
New York Yankees 5
Newcastle United 36, 37, 53, 59, 86, 92,
95, 146, 259, 267, 324, 326, 352, 377,
379, 382, 385, 394, 397, 414, 421, 436
Newton Heath see Manchester United
Nmes 96
NipponShinpan 107
non-stationarity 3289
normal distribution 18, 165, 1737, 186,
248
North American Model 12, 12631
Northern Ireland 31, 203, 231, 238, 341
Norwich City 49, 55, 196, 259, 273, 352
Nottingham Forest, 43, 69, 93, 265, 272,
382, 414
Notts County 289
ONeill, Martin 259, 273
Oce of Fair Trading (OFT) 81, 856,
381
Oldham Athletic 265
ONDigital 80, 84, 437
ordered probit 20, 1656, 172, 17688
ordinary least squares (OLS) 18, 282, 283,
3078, 254, 386, 388, 404
Oxford United 53, 267, 352
Oxford University 39
Pacic League 106
Pacic Coast League 130
Paisley, Bob 42, 265, 267
Paris Saint Germain 30, 49, 65, 86, 103,
105
Parma 30
Pearson, Nigel 272
Peterborough United 53, 269, 352
Peugeot 102
player transfers
transfer expenditure 23, 3681, 87,
90101, 420
transfer fees 9, 11, 13, 20, 223, 367,
134, 146, 191, 22736, 239, 255, 256,
377, 435
transfer market 13, 16, 22, 23, 35, 51,
90101, 129, 141, 146, 1901, 201,
202, 203, 204, 22736, 239, 244, 246,
253, 255, 262, 421, 4302
transfer values 2557
players wages and salaries 4, 10, 13, 16, 23,
358, 71, 87, 90101, 123, 132,
13340, 20931, 2389, 4213,
42835
poisson distribution 1701, 404
Poland 31
Port Vale 265
Portsmouth 43, 55
Portugal 51, 205, 208, 427
Premier League 1, 2438, 40, 41101,
10924, 141, 146, 15089, 189203,
209, 217, 2257, 2316, 2378,
25263, 26580, 286302, 30616,
31734, 34272, 377, 379, 38197,
40316, 41839
Preston North End 39, 80, 382
probit 282, 284
production frontier 20, 241, 2456, 264,
283, 28791
Professional Footballers Association (PFA)
92
prot maximisation 8, 9, 12, 125, 127, 128,
133, 135, 140, 143, 145, 146, 148,
187, 210, 211, 240, 241, 245, 375, 420
PSV Eindhoven 436
Queens Park 33940
Queens Park Rangers 55, 259, 380, 382,
384
Racing Club Paris 102, 103, 105
random eects (RE) 18, 249, 254, 25760
Rangers 49, 51, 65, 318, 333, 335, 33942,
372, 438
rank-order tournament model 13, 190,
21521, 223, 2389
Ravinelli, Fabrizio 205
reaction function 1436
Reading 66, 272
Real Madrid 24, 30, 35, 49, 51, 103, 105,
227, 3378, 426, 427, 434, 435, 436
Reith, John 80
Republic of Ireland see Ireland
Index
reserve clause 34, 9, 10, 125, 1289, 132,
1356, 140, 187, 210, 211, 2212, 422
reservation price 22931, 234, 239
reservation wage 4, 132, 135, 140, 210
Restrictive Practices Court (RPC) 856,
381, 438
retain-and-transfer system 11, 902, 109,
123, 132, 421
revenue maximisation 137, 140, 145, 146,
241
revenu sharing 5, 9, 12, 89, 125, 12930,
1379, 187, 431
Riijkard, Frank 205
Rimet, Jules 102
Robson, Bryan 93
Roma 24, 30, 51
Romania 208
Rosenborg 426
Rotherham United 40
Royal Engineers 39
Royle, Joe 265
Rudge, John 265
Rugby public school 38
rugby league 6, 250, 321, 3334
rugby union 6
salary cap 5, 9, 125, 12930, 1367, 140,
147, 149, 187, 422, 433
Sampdoria 51
Savoldi, Guiseppe 36
Scarborough 53
Scholar, Irving 376
Scotland 6, 15, 31, 4351, 5965, 70, 80,
1503, 195, 203, 231, 238, 292, 318,
319, 326, 333, 335, 33942, 372, 427
Scunthorpe United 53, 89
seemingly unrelated regressions (SUR) 18,
388
Servette Geneva 86
Ste 102
Shankly, Bill 42
share prices 1516, 20, 3734, 379,
38097, 417
Shearer, Alan 35, 37, 146
Sheeld United 80, 267, 272, 289, 3545,
380, 382, 384, 394
Sheeld Wednesday 69, 289
Shrewsbury Town 53
sigma convergence see convergence
simulations 1657, 3334
Sky 80, 81
Slovenia 208
Small Heath see Birmingham City
Smith, Dennis 272
Smith, Jim 267
457
Sochaux 102, 103
Sony Creative Products 107
Souness, Graeme 341
South Korea 108
Southampton 49, 55, 141, 196, 323, 380,
382, 385, 397, 41416, 426
Southend United 267
Southern League 39
Southport 352
Soain 15, 22, 2438, 4351, 70, 87, 101,
104, 127, 1503, 205, 206, 208, 209,
318, 3358, 372, 420, 427, 435, 438
sponsorship 15, 16, 22, 30, 38, 71, 87, 89,
107, 108, 223, 245, 374, 424, 425,
426, 438
St Etienne 49, 51, 103, 104
Stackelberg equilibrium 1445
Stade Reims 49, 103
Stam, Jaap 436
Standard AC 102
stochastic production frontier see production frontier
Stoke City 272, 421
Sturm Graz 426
Stuttgart 51
Sugar, Alan 83, 376
Sumitomo Motors see Kashima Antlers
Sunderland 53, 80, 86, 326, 352, 382, 384,
389, 394, 429, 436
Suntory 107
Sunyol, Josep 336
superstars, economics of 1011, 13,
21115, 238, 423
Sweden 427
Swindon Town 55, 66
Tapie, Bernard 104
Taylor, Graham 265, 267
Taylor, Lord Justice 69
television 7, 12, 13, 15, 22, 30, 41, 67, 71,
79, 807, 96, 99, 105, 106, 107, 108,
109, 124, 130, 147, 148, 149, 190,
212, 215, 238, 317, 333, 334, 374,
377, 380, 381, 418, 4238, 4359
Thatcher, Margaret 68
Tokyo Giants 106
tobit 324
Torino 51
Tottenham Hotspur 82, 83, 347, 376, 377,
380, 382, 414, 420
Tranmere Rovers 89
transaction costs 240, 271
transfers see player transfers
Turkey 209
Turner, Chris 269
458
Index
Werder Bremen 65
West Bromwich Albion 66, 93, 272, 352,
380, 382
West Ham United 37, 49, 352, 377, 414,
421, 432
Wigan Athletic 53, 272
Wimbledon 53, 269, 289, 352, 414
Witty family 335
Wolverhampton Wanderers 43, 53, 68,
2434, 271, 272, 273, 289, 352, 380,
423
Woolwich Arsenal see Arsenal
World Club Championship 41
World Cup 56, 65, 102, 103, 104, 107, 108,
2056, 421
World Series 5, 6
World Youth Championship 107
Workington 352
Wrexham 352
Wycombe Wanderers 53
Yokohama Flugels 108
Yominuro FC seeVerdy Kawasaki
Yorke, Dwight 435
Yugoslavia 205, 208
Zico 108
Zidane, Zinedene 434
Zola, Gianfranco 205