MANAGERIAL AND DECISION ECONOMICS

Manage. Decis. Econ. 28: 5770 (2007)

Published online in Wiley InterScience
(www.interscience.wiley.com) DOI: 10.1002/mde.1313

Hidden Monopsony Rents in Winner-take-all

Markets}Sport and Economic Contribution
of Spanish Soccer Playersy
Pedro Garcia-del-Barrio* and Francesc Pujol
Department of Economics, University of Navarra, Edicio Bibliotecas (entrada Este), Pamplona, Spain

In labor markets where few companies compete for many workers, economic theory predicts
monopsony rents. Surprisingly, soccer clubs do not prot from the expected rents. The purpose
of this study is to explain such contradictory evidence.
Our model and empirical test, using data obtained from the Spanish professional soccer league
for the season 2001/2002, suggests that monopsony rents that the clubs were to obtain from
most of the soccer players would eventually revert to the superstars. The study also illustrates
that the market value of players stems both from their sporting performance and their
economic contribution. Copyright # 2007 John Wiley & Sons, Ltd.

INTRODUCTION for US Major League Baseball and concludes that

it is of considerable magnitude. Hunt and Lewis
Much of the work on monopsony rents has been (2003) analyze, empirically, the same league and
tested in the context of professional sports. also claim that it exhibits monopsony rents for the
Economic theory points to the exploitation of teams owners.
professional sportsmen under the reserve clause, Monopsony power was initially associated in
through the introduction of monopsony power. literature to the reserve clause. Scully (1974) insists
Rottenberg (1956), for instance, describes the that canceling the reserve clause would result in
clubs monopsony power in the Baseball industry. benets for the players with no damage to the
From a basic theoretical framework, Scully (1974) game.1 Recent papers on the issue broadly
calculates the rate of monopsonistic exploitation acknowledge that professional team sports had
monopsonistic labor markets in the past, whereas
*Correspondence to: Department of Economics, University of they now have free agency. Some other studies
Navarra, Edicio Bibliotecas (entrada Este), 31080, Pamplona, argue that certain monopsonistic exploitation will
Spain. E-mail: [email protected] still remain once the clause has been removed, due
y
We would like to thank Stefan Szymanski, Francisco Galera,
to specic contractual arrangements. When con-
Miguel Cardenal, Javier Gomez-Biscarri, J. Alberto Ruiz and
Javier Hualde for their useful comments and suggestions. sidering the soccer industry, we argue that mono-
Thanks also are due to the participants at the 57th IAES psony power persists even once the reserve clause
Conference, March 2004, Lisbon; and at the XXIX Simposio de has been eliminated. Since 1995, any out-of-
Analisis Economico, December 2004, Pamplona. We have contract player can freely negotiate with a team
greatly beneted from the comments of one anonymous referee.
and does not have to pay any compensation fee
The usual disclaimer applies. Pedro Garcia-del-Barrio grate-
fully acknowledges nancial support from the Ministerio de to his former club. But the clubs anticipate
Ciencia y Tecnologia (SEJ2004-04649, Spain). this feature and encourage the players to sign

Copyright # 2007 John Wiley & Sons, Ltd.

58 P. GARCIA-DEL-BARRIO AND F. PUJOL

long-term contracts.2 In fact, although the player The purpose of this study is to provide an
is allowed to play out his option and become a free answer for a contradictory evidence, which we
agent, the player could only switch to another argue to be just an outwardly contradiction. In the
team provided the two teams agree on a transfer soccer leagues a few teams compete for many
price. Hence, the transfer fee works here as the workers, including those young players coming
Roselle Rule did in NFL American football, from junior categories. In these situations the
implying monopsony elements in soccer.3 economic theory predicts the existence of mono-
At this stage, a second stream of arguments psony rents. However, soccer clubs do not prot
enters into the discussion. It is not uncommon for from the expected rents.6 Moreover, in the case of
some markets to support high concentrations of the Spanish LFP, the vast majority of clubs
rewards among small numbers of participants. systematically exhibit a negative economic out-
Following the term settled by Frank and Cook come, as summarized in Table 1.
(1995), we will refer to them as the winner-take-all Yet, the negative results of soccer teams do not
markets. This hypothesis states that those workers derive from a loss of potential revenues in the
who are slightly better than other workers become industry as a whole. Szymanski (2001) analyzes the
the winners of the market, earning much larger English premier League and concludes that the
earnings than the losers (the wages of the former increasing income inequality between clubs (and
group exhibit a more than proportional magnitude therefore the unbalance of skill endowments)
with respect to its productivity). Note that Noll produces a negligible, if any, decline in competitive
(1974) and Rosen (1981) had already referred to the balance. Signicantly, there is no evidence at all
phenomenon of superstars. More recently, Dobson that this feature has reduced the professional
and Goddard (2001) stress again how skewed soccer league capability of producing revenues.
earnings distributions may stem from scarcity of Table 1 shows as well the collective revenues of the
supply of outstanding talent, together with the large clubs competing in the ve big European leagues
audiences that they attract. Therefore, a few in the last decade, providing evidence on the large
individuals earn enormous rewards, thereby dom- share that stems from broadcast rights contracts.
inating the activities in which they engage. Gate revenues have progressively become a
Frank and Cook (1995) remark the fact that secondary part of clubs earnings, although they
quite a large number of markets (like professional still represent 30% of the total revenues in
sports, pop culture, and arts) experience similar Spain and the UK. Note that unlike most of the
reward structures among them, in which many other soccer leagues, since the late 90s, some
individuals compete for a handful of big prizes at Spanish teams negotiate individually the televi-
the top. Perhaps the most novel and provocative sions rights, which presumably leads to increasing
statement made in their book is that socially economic dispersion to the advantage of the
wasteful patterns of competitive investment result strongest teams.
from winner-take-all contests. In other words, they This paper hinges on a simple and original idea,
argue that too many resources are devoted to the which is built upon the simultaneous analysis of
competition itself and not enough resources to real monopsony rents and the winner-take-all issue. In
production, thereby encouraging some arrange- order to jointly examine these two aspects, we
ments to restrict investments in superstars.4 consider it accurate to assume that the Spanish
Typically, this sort of wasteful activities occurs soccer industry is a dual labor market. One
whenever there is an arms race between countries. segment of the market could be characterized by
Rosen and Sanderson (2001) stress how sports the conventional analysis of monopsony, in which
markets experience a curious combination of co- a small number of clubs wish to attract many
operation and competition, which would be potential candidates. The majority of those are
another variant of the arms race phenomenon. average candidates, whose instruction has often
They also say that substantial ineciencies result taken place in second-category teams. But the
from market failures in contests. Furthermore, market is also comprises of a few outstanding
they consider that the winner-take-all phenomen- workers (soccer superstars), who accumulate
on characterizes a broad range of work activities in market power. A relatively large number of
the US economy, implying part of the increased entities (not only Spanish clubs) would ercely
income inequality.5 compete for hiring those superstars, so that the

Copyright # 2007 John Wiley & Sons, Ltd. Manage. Decis. Econ. 28: 5770 (2007)
DOI: 10.1002/mde
 HIDDEN MONOPSONY RENTS IN WINNER-TAKE-ALL MARKETS 59

Table 1. Prots and Revenues in the Big Five Soccer Leagues

Prots (million h) 1996 1997 1998 1999 2000 2001 2002 2003 2004
English Premiership 77 129 143 104 80 121 125 185 223
Spanish 1a Liga 23 19 124 170
Spanish 1a Liga 21 17 101 150 159 369 a
682 a
402 a

Italian Serie A 3 8 36 114 46 216 404 381 341

German Bundesliga 37 27 47 35 87 100 138 52
French Ligue 1 5 7 46 70 36 41 98 61 102

Revenues (million h)
English Premiership 534 716 898 1034 1192 1447 1748 1832 1949
Spanish 1a Liga 366 524 569 612 683 713a 776 847 953
Italian Serie A 452 551 650 714 1059 1151 1127 1137 1126
German Bundesliga 373 444 513 577 681 880 1043 1114 1058
French Ligue 1 277 293 323 393 607 644 643 689 655

Revenues 2004 (million h) Gate Broad- Sponsor- Com- Total % GateRev % % %

receipts cast ship mercial Broad Spons. Com.
English Premiership 585 877 487b 1949 30 45 25b
Spanish 1a Liga 276 391 200 86 953 29 41 21 9
Italian Serie A 180 619 158 169 1126 16 55 14 15
German Bundesliga 212 296 328 222 1058 20 28 31 21
French Ligue 1 118 308 144 85 655 18 47 22 13
Source: Deloitte&Touche Annual Review of Football Finance (2003, 2005). Authors calculations from clubs accounts.
a
Records extrapolated from a sample of 16 clubs accounts. (The total number of clubs in the league is 20).
b
Sponsorship and other commercial income combined.

winner-take-all hypothesis undoubtedly applies. market. On the other hand, Scully (2004b) notices
Again, the arms race phenomenon among teams that a dual market has to be considered by dealing
can help to explain the prevalence of superstar dierently with veteran and non-veteran players,
monopoly power in bargaining. whereas our view advises a separate treatment of
In summary, this paper suggests that the normal and outstanding players (or rather skills,
monopsony rents that the clubs were to obtain which the players possess in dierent combina-
from the soccer players will revert to the few tions). This dierent point of view is not irrelevant,
superstar players who enjoy strong bargaining in as much as it can explain that hidden mono-
power. This approach can explain why prot psony rents persist in the industry and are
making clubs have been very much the exception obtained from all the players, since all of them
and not the rule, while monopsony rents have been posses certain amount of normal skills. In addition
present throughout most of the history of the to that, we explicitly mention that the increasing
industry. compensation dispersion between the two seg-
This issue seems to have implicitly been recog- ments of the market can be justied by taking into
nized by Scully, but we propose a slightly dierent account the winner-take-all argument.7
explanation of the facts, using an alternative Other essential feature of this study is that, when
methodology and analyzing the Spanish soccer evaluating the contribution that a soccer player
market. On one hand, Scully (2004a) refers to the brings forth to his team, we pay especial attention
non-linearities that characterize this type of to the players capability of generating economic
markets (where individuals who are only a revenues (achieved through broadcast rights,
small fraction better that others, in terms of merchandizing contracts, etc.). As stated by
productivity, may receive several times as much), Horowitz and Zappe (1998), it is broadly acknowl-
which is a key ingredient in our analysis. He also edged that, in professional-team-sport labor mar-
points out that superstars receive a multiple of kets, players compensation is based on sporting
pay-per-unit on performance (compared to other performance. They also stress that the retribution
players), and that it happens since long before the varies depending on the experience, bargaining
change from a restricted to a free agency labor power and demarcation of the player in the pitch,

Copyright # 2007 John Wiley & Sons, Ltd. Manage. Decis. Econ. 28: 5770 (2007)
DOI: 10.1002/mde
60 P. GARCIA-DEL-BARRIO AND F. PUJOL

as well as on some team characteristics. Never- presented by Scully (1974), the formal process of
theless, we venture that one essential variable has MRP and salary determination should incorporate
been often neglected in this regard, since the player a major feature: monopsonistic factors that reduce
economic contribution should be considered no skill rewards below players MRP.8
less important than his sporting performance. Specically, we assume that soccer players
Precisely the empirical study will also test this possess a vector of talents, which can be grouped
idea by including an appropriate variable into the into two broad types: ordinary or average skills
regressions. (Ai) and superstar abilities (Sj). To determine the
prevalent salary for the former type of inputs, a
monopsonistic framework is suitable. The case of
the latter is dierent: since it is an exclusive
A BASIC THEORETICAL FRAMEWORK production factor, no monopoly rents will be
extracted from the superstar abilities. In order to
After describing the motives of the paper, we attract the services of outstanding talents, a erce
will}in the following pages}provide a basic competition among teams takes place. This kind of
theoretical framework of our study. In some arms race environment, in which teams are
aspects, the starting point of our model parallels involved, results in monopolistic power on the
the one presented by Scully (1974): a model of part of the supplier of superstar abilities.
marginal revenue product (MRP) and salary Let us start by characterizing the revenues. The
determination, in which the demand of some team obtains earnings from two main sources:
inputs is established in a monopsony framework. (i) the direct revenues (coming from performance
Remember that if the labor market were perfectly and gate receipts, merchandizing, etc.) and (ii)
competitive, player salaries would be equated with the share of the total industry prots, correspond-
player MRP; whereas it is not the case if the ing to each particular team (which will depend
reserve clause (or other contract arrangements) mainly on popularity, broadcast rights, etc.).
restricts player bargaining to one owner. In the Besides that, our analysis divides the earnings
latter case, players and owners share the players into two groups: those linked to sporting perfor-
MRP. mance, and those which stem from the merchan-
In order to set the wages, and following the view dizing power associated to each particular soccer
of Neale (1964), we initially assume that the league skill.
as a whole (rather than the individual club) is the To analyze the decision-making challenge faced
decision-making unit. This approach has been by each team, it is necessary to previously dene
challenged in the literature on the grounds that the expression of prots for the industry as a
most economic decisions (such as how much whole. These prots are associated with the
money to spend on stadium development and monopsony rents achieved by the unique agent
how many players to hire), are made by the clubs. placed in the demand side of the market. We may
Accordingly, at a later stage, we will assume that think of all the teams behaving like a cartel that
the club is the relevant economic decision maker, agrees to pay, in a monopsonistic framework, the
as in Sloane (1971). He considered playing success wage that maximizes prots. The revenues for the
as the most important objective to which chair- industry could then be dened by:
person, directors, managers, players and specta-
tors can all subscribe. Therefore, the model is
going to be eventually arranged presuming that the Rind Rf Ai ; Sj ; c; mAi ; Sj ; c;
objective of the soccer club is one of utility 8i 1; . . . ; I and 8j 1; . . . ; J: 1
maximization, subject to a nancial solvency
constraint.
Consider the players MRP in the soccer where functions f and m represent, respectively, the
industry and assume that the inputs (abilities) productivity related to the sporting performance
generate revenues from two sources: through and to the economic protability associated with
individual performance (and its eect on team each particular skill. In addition, c accounts for
standing) as well as through merchandizing specic characteristics: league competitiveness,
activities. In addition, following the framework number of fans, etc. On the other hand, the cost

Copyright # 2007 John Wiley & Sons, Ltd. Manage. Decis. Econ. 28: 5770 (2007)
DOI: 10.1002/mde
 HIDDEN MONOPSONY RENTS IN WINNER-TAKE-ALL MARKETS 61

of hiring soccer skills is given by Sj or . . . PSind 0; and the opposite occurs with
X X the average skills, since there are extraordinary
C wa fi Ai
Ai wa m
i Ai
Ai
i i
(monopsony) rents associated with them:
X X !!
f
ws
Sj ws m
j
Sj : 2 A   @wa fi @wam
i
j Pind Ai
Ai
> 0: 6
j j @Ai @Ai
The wages have been split out into two elements: Once the prots of the whole industry have been
rewards of player performance contribution (w f ), calculated, the next step is to determine the
and payments associated to merchandizing activ- amount of Sj that each club is willing to hire. At
ities (wm). Obviously, there are as many wages as this stage, in accordance with several previous
the number of dierent skills of labor input. Note studies, we consider that the clubs are not pure
also that the two rst terms of (2) correspond with prot maximizers.9 Henceforth, we assume that
labor inputs constrained by monopsony situations. the objective function to be maximized by each
Accordingly, total prots for the whole soccer team (n) is driven essentially by the sportive
industry can be dened by success, rather than the prot. We consider the
Pind Rf Ai ; Sj ; c; mAi ; Sj ; c utility function for a team (Un) to depend on its
X X potential playing success, which is directly related
 wa fi Ai
Ai wam
i Ai
Ai to the amount of talent hired (Sn) as well as to the
i i
X f
X specic characteristics of each particular club jn :
ws
Sj
j ws m
j
Sj : 3
P  
j j
Ai
Un U ; Sn ; jn 8n 1; . . . ; N: 7
N
Obviously, the problem consists of maximizing the
previous expression with respect to Ai. There are where jn accounts for teams characteristics and
i j rst-order conditions for this problem, which peculiarities. By assuming that all the teams
are dened by require the presence of the same number of
players, the amount of ordinary inputs}the
@P
0 ) wa fi Ai wa m i Ai proportion of all kinds of ordinary skills that
@Ai corresponds to each team SAi =N}is taken to be
@R @f @R @m the same for all the teams.



@f @Ai @m @Ai Besides, given that there is a positive relation-
!
@wa fi @wam i
ship between superstar inputs in each team (Sn)
 Ai
; 4 and the teams utility function, the way to
@Ai @Ai
maximize the latter is to maximize the amount of
superstar input. In other words, the objective of
@P @R @f @R @m maximizing expression (7) will be achieved by
0 ) ws fj ws m
j

: 5 maximizing Sn. And yet, teams do not want to
@Sj @f @Sj @m @Sj
incur any decit (at least in the long run) and,
These two equations reveal that the league as a consequently, the teams optimization problem
whole maximizes prots by selecting a level of can be characterized using the following expres-
ordinary skills (Ai) such that each skill is rewarded sion, where the constraint drives the major role:
with a salary equal to its MRP, minus the
P  
Ai
monopsony rents (represented by the last term of Max Un U ; Sn ; jn 8
Sn N
Equation (4)). The outstanding abilities (Sj) do not
suer from monopsonistic distortions, so that they


are remunerated at a level equal to their MRP. Sn
s:t: : Pn an ; j
PA
ind En f Sn ; mSn ; jn
These equations allow us to obtain the optimal S n
amount of each type of ordinary skills to be hired
 wsf
Sn wsm
Sn 50:
in the industry, A i , as well as the corresponding
wages wafi(A m 
i ) and wai (Ai ). Similarly, one could We treat jn as exogenous and assume that each
determine the total amount of the superstar team is willing to spend all its extraordinary prots
abilities and its wage: S f m
j , wsj and wsj . No prots to hire superstar talents. The rst term of the
are extracted from the wages associated to constraint accounts for the portion of the prots

Copyright # 2007 John Wiley & Sons, Ltd. Manage. Decis. Econ. 28: 5770 (2007)
DOI: 10.1002/mde
62 P. GARCIA-DEL-BARRIO AND F. PUJOL

achieved by the industry as a whole that corre- receive extraordinary rewards (rents). Specically,
sponds to each particular team. This amount is we expect that rich teams will get big superstars
dened by the product of two elements: the and poor clubs get small ones. This behavior
proportion of earnings that correspond to each hinges on the erce competition that clubs main-
team, an
; and the total revenues generated by the tain in order to get the best players. In any case,
whole industry, PA ind : As stated in the constraint of this kind of arms race between clubs, results in a
expression (8), the former factor depends basically strong bargaining position on the part of the
on the special talents hired by the team with suppliers of superstar skills.
respect to the total special abilities in the league The last step consists of analyzing the point of
Sn =S ; as well as on jn : Note also that the latter view of the superstar players. The challenge
element relates to revenues as it can potentially consists simply of matching clubs (which are also
account for match attendance and the level of lined up according to economic criteria) to super-
support that each team draws from its fans, which star players, so that their specic interests meet
ultimately depends on team performance and each other. Of course, the players enjoy a stronger
the size of the metropolitan area. The amount bargaining position, so that they will choose the
S has already been calculated and is considered team which, in accordance with this model, oers
as given. There is another source of revenues, them greater rewards. The foreseeable outcome of
En(
), that enters into the prot function of this matching problem is simple: the best players
each team, and that also depends primarily on will enroll the richest clubs. This is the reason why
Sn and jn : those players can be considered the winners of the
One would expect both an
and En
to present market (to whom the winner-take-all hypothesis
a positive rst derivative with respect to Sn. In is applied).
addition, as long as the second derivatives have a Note that in our approach each soccer player
negative sign, the optimal amount of Sn, for each possesses a set of skills and, if hired, he should be
particular team, can be identied. The result is rewarded for each and every of them. Hence, the
straightforward: the team rewards the outstanding total rewards received by a soccer player will
skills by simply paying all the available resources comprise the previously determined market sal-
that it possesses. Given that the wage paid for each aries. The superstar players accumulate both
unit of Sn is xed, the clubs hire a dierent amount outstanding abilities as well as average skills. On
of this factor depending on their nancial status. one hand, since they possess a certain amount of
The total wage bill the team devotes to this type of normal skills constrained to monopsony power,
skills can be dened by they will be underpaid for them. On the other

 hand, they will be compensated as they will charge
Sn huge amounts of money for their outstanding
WBn ws ws
Sn an  ; jn
PA
f m
ind
S abilities, which eventually will overcome the
En f Sn ; mSn ; jn : 9 monopsony rents that they were deprived of. This
situation would not aect the majority of the
Expression (9) states that each club (n) establishes soccer players, who do not endow such out-
a wage bill equal to the revenue it directly achieves standing skills. The monopsony rents extracted
from the superstar skills plus the proportion of from these normal players will eventually enrich
league prots that corresponds to that particular the pockets of a few superstars.
team. The key issue here is that such proportion
will be strongly inuenced by the higher or lower
presence of special skills. Alternatively, this result
could be read in average terms: the club will hire DESCRIPTION OF THE VARIABLES AND
the amount Sn such that the wage per unit equals DATA SOURCE
the average revenue. No team would hire more
than that, unless it does not care about decit. In order to empirically test this issue, the paper
In the context of factual player contracting, each analyzes the data of the Spanish professional
and every team will desire to enroll at least one soccer league, for the season 2001/2002.10 The
outstanding player, who endows more or less data source for this study is the sports journal
superstar talents. Such a superstar player will MARCA: Gua MARCA (Liga 2003) and Gua

Copyright # 2007 John Wiley & Sons, Ltd. Manage. Decis. Econ. 28: 5770 (2007)
DOI: 10.1002/mde
 HIDDEN MONOPSONY RENTS IN WINNER-TAKE-ALL MARKETS 63

MARCA (Liga 2002). The latter has been used to thousands.14 The foreseeable multicollinearity
complement those pieces of information that were between (Google) and the indexes of sporting
not recorded in the former. The observations for performance requires that the former variable be
which all the information was available are 369 ltered before entering the regression, as we
(the total number of players registered in the explain afterwards.
league was 518). We will pay specic attention to There is still a third essential element to be
the contribution of the economic impact asso- brought into the analysis. We aim to prove
ciated with each particular player, as far as it may whether or not the winner-take-all hypothesis
help to explain the large earnings dierentials that holds in the labor market of Spanish professional
could not be explained due to dierences in soccer players. A way of testing this feature is
sporting performance. dening a series of dummy variables, taking the
The core of the analysis is composed by the value 1 for the case of the few winners of the
dependent variable plus three main explanatory market (soccer superstars), and 0 for the other
variables. It also incorporates a number of individuals. We dene as superstars those players
variables to control some characteristics of each with the highest Internet exposure as measured by
soccer player. The dependent variable is desig- the Google variable. The names reserved for these
nated as (MarketValue) and represents the esti- qualitative variables are (Winner-all-5), (Winner-
mated market value of a soccer player. After all-10) and (Winner-all-20), for the greatest 5
searching for a reliable proxy variable to evaluate superstars, the group of the 5 second-best players
market value, we have chosen (MarketValue) as or those rated between the 11th and 20th position,
the more accurate one. This variable was collected respectively. As pointed out in the theoretical
from MARCA, and measures the amount of description, it may also be signicant to include
Euros (in millions) that the clubs are willing to the variable (Team-winner) to account for the best
pay for the services of each soccer player.11 Note player in each of the 20 teams in the league, those
that, apart from the diculty in obtaining public who are the winners of the market at the team
information on players salaries, transfer fees, level, instead of the industry level, again in terms
commercial contracts, etc., these earnings are of highest Internet exposure at the team level.
unsuited to our purpose, since the gures are Finally, we consider a number of variables to
distorted depending on the date on which the control dierent factors. First, (Europe) accounts
contracts took place and often the contractual for the number of games that have taken place in
transfer fee does not reect the players real the Champions League or in the UEFA Cup;
market value. On the contrary, (MarketValue) second, (International) collects the number of
provides a homogenous evaluation, which is games in which a player has defended his own
computed for all the players at the very same date national team; and nally (Age) examines the
in the beginning of the season. potential inuence of the age in this context. In the
Then, two alternative variables of worker case of these three variables, the corresponding
productivity are used in the study: an index of signicant quadratic terms are included in the
performance measured by puntos Marca (Pun- estimations to allow for non-linear eects. In
tosMarca);12 and another index of productivity addition to that, the regressions also incorporate
evaluated through liga Fantastica (LigaFantasti- one dummy for foreign non-European players
ca).13 Since both variables capture the same reality (No-EUplayer), and another one for non-Spanish
and are undoubtedly correlated, they will not be players who belong to the European Union
introduced together in the same regression. (EUplayer). Finally, we use three dummies asso-
Another principal issue analyzed here is the ciated to the eld position of soccer players:
economic protability that a soccer player reports defenders (Defender), midelders (Midelder)
to his club besides his direct contribution to and attackers (Attackers). Hence, the goalkeepers
sporting performance. In order to evaluate the remain as the reference group. These variables will
revenue-generating capability that a player pos- fulll the role of controlling the heterogeneity
sesses (through broadcast rights, merchandizing associated with the players eld position, which is
contracts, etc.), we propose to line up the players a key issue since the indexes of performance used
according to the total number of links reported by the same criteria for all the soccer players, no
the Internet search engine Google, expressed in matter what their demarcation in the pitch was.

Copyright # 2007 John Wiley & Sons, Ltd. Manage. Decis. Econ. 28: 5770 (2007)
DOI: 10.1002/mde
64 P. GARCIA-DEL-BARRIO AND F. PUJOL

Table 2. Summary Statistics for the Main Variables

Variable Mean St. dev. Minimum Maximum
a
MarketValue 6.4173 8.9503 2 80
Winner-all-5 0.0135 0.1157 0 1
Winner-all-10 0.0135 0.1157 0 1
Winner-all-20 0.0243 0.1544 0 1
Team-winner 0.0487 0.2157 0 1

PuntosMarca 1.1607 0.2598 0.20 2.00

LigaFantastica 2.0891 0.6669 0.01 4.50
Googleb 4.8467 6.5577 0.27 63.40

Age 28.6504 3.8405 20 41

International 9.4200 18.9793 0 94
Europe 4.0162 6.4748 0 18
EUplayer 0.1436 0.3511 0 1
No-EUplayer 0.3387 0.4739 0 1
Defender 0.3577 0.4799 0 1
Midelder 0.3604 0.4807 0 1
Attacker 0.1924 0.3947 0 1
a
In millions of Euros.
b
In thousands of links in the Internet.

Table 2 reports some descriptive statistics of the tastica), inducing clear risks of multicolliniarity. It
variables employed in this study. is then necessary to lter the initial measure of
soccer players Internet presence (Google) against
all the factors related to individual sporting
performance and other measurable individual
EMPIRICAL RESULTS characteristics. We will use the residuals of this
regression as a ltered measure of strict non-sport
The empirical analysis is carried out based on data value (FilteredGoogle) generated by the soccer
from the Spanish soccer league for the Season player as a result of his public exposure. This
2001/2002. It aims to explain the determinants of marketability of players is a contribution that
the economic market value of soccer players, who should be taken into account in the reward scheme
participate in a labor market of a special nature. if the sport labor market is not far from eciency.
We expect that the monetary valuation of these The results of the mentioned ltering procedure
agents is linked to the sources of revenues that they are not going to be displayed here, since they are a
generate for their teams: their sporting perfor- mere device to produce (FilteredGoogle). How-
mance, and also their contribution to sources of ever, it is interesting to note that, for the baseline
revenues coming from the support from fans, model, the adjusted R2 of the corresponding
which is translated into soccer team sales. (Google) regression is around 0.58. It means that
Sporting performance of each single player is an important share of the soccer player presence
captured through two alternative composite values on the Internet cannot be explained exclusively by
that have already been described (PuntosMarca) sporting performance considerations, which rein-
and (LigaFantastica). The contribution of soccer forces the necessity of incorporating a measure of
players to their team revenues through merchan- economic impact of the players.15
dizing-related sales is measured by their individual We have estimated various specications, from
presence on the Internet, with the number of sites which the preferred models are displayed in
in which they appear, according to the web search Table 3. Initially, we computed the estimations
engine Google. One cannot directly introduce the introducing the variables without transformations;
variable (Google) into the model, as one can but then, due to the lack of normality in the
suspect a problem of interdependence between the residuals, we run the regressions after having taken
players popularity (Google) and the measure of logarithms of the dependent variable.16 This
sporting performance (PuntosMarca or LigaFan- conventional procedure solves the problems of

Copyright # 2007 John Wiley & Sons, Ltd. Manage. Decis. Econ. 28: 5770 (2007)
DOI: 10.1002/mde
 HIDDEN MONOPSONY RENTS IN WINNER-TAKE-ALL MARKETS 65

Table 3. Explaining Soccer Players Economic Valuation

Dep.var. MarketValue Monopsony model Winner-take-all model
Variable (1) (2) (3) (4)
Winner-all-5 1.5360 1.5850
(9.21) (9.12)
Winner-all-10 0.4818 0.5444
(3.08) (2.89)
Winner-all-20 0.2206 0.0940
(1.40) (0.58)
Team-winner 0.3473 0.3997
(2.65) (2.78)

PuntosMarca 0.9207 0.8351

(8.01) (7.45)
LigaFantastica 0.2358 0.2062
(4.93) (4.61)
FilteredGoogle 0.3837 0.4106 0.3205 0.3444
(7.21) (7.64) (6.26) (6.65)

International 0.0155 0.0171 0.0158 0.0173

(4.14) (4.41) (4.65) (5.02)
InternationSq 0.0001 0.0001 0.0001 0.0001
(1.50) (1.20) (3.52) (3.27)
Europe 0.0965 0.1001 0.1099 0.1147
(3.84) (3.81) (4.44) (4.41)
EuropeSq 0.0031 0.0038 0.0040 0.0046
(2.35) (2.70) (3.13) (3.40)
Age 0.0552 0.0119 0.0656 0.0123
(0.81) (0.16) (0.99) (0.17)
AgeSq 0.0014 0.0003 0.0015 0.0006
(1.22) (0.26) (1.35) (0.55)
EUplayer 0.1701 0.1502 0.1497 0.1365
(2.28) (1.94) (2.18) (1.93)
No-EUplayer 0.1075 0.1430 0.0960 0.1345
(1.86) (2.22) (1.72) (2.18)
Defender 0.0188 0.3223 0.0062 0.2764
(0.20) (3.22) (0.07) (2.80)
Midelder 0.2169 0.0771 0.1935 0.0676
(2.33) (0.76) (2.09) (0.67)
Attacker 0.5498 0.2131 0.448 0.1535
(5.57) (2.04) (4.44) (1.36)
Espanyol 0.3109 0.2626 0.3219 0.2799
(2.21) (1.74) (2.31) (1.88)
Osasuna 0.2199 0.2238 0.2131 0.2157
(1.83) (1.88) (1.79) (1.81)
RealMadrid 0.2687 0.3387 0.1165 0.1742
(1.958) (2.39) (1.00) (1.44)
Cons 0.4686 1.7827 0.2872 1.3579
(0.47) (1.58) (0.29) (1.25)

R2 0.703 0.669 0.727 0.695

Root MSE 0.412 0.435 0.397 0.421
N. Obs 369 369 369 369

Skewness/kurtosis [0.984] [0.995] [0.789] [0.641]

ShapiroWilk [0.763] [0.822] [0.714] [0.738]
Estimations computed with robust standard errors. MarketValue, PuntosMarca, LigaFantastica and Google in logarithms.
(t-statistics) in parenthesis and [P-values] in brackets. Indicates signicant at 1%;  at 5%; and  at 90%.

diagnostics and is in accord with the fact that On one hand, in order to deal with the problem
earnings are often not linear with relation to of heteroskedasticity, all the regressions were
performance (Scully, 2004a). computed using robust (white-corrected) standard

Copyright # 2007 John Wiley & Sons, Ltd. Manage. Decis. Econ. 28: 5770 (2007)
DOI: 10.1002/mde
66 P. GARCIA-DEL-BARRIO AND F. PUJOL

errors. On the other hand, we have veried group), the valuation of attackers receives much
through two alternative tests that the residuals higher attention and reward. The same could be
are normal: the P-value for both the Skewness/ said of the midelders, but only for the regression
Curtosis and the ShapiroWilk tests, indicate estimated using (PuntosMarca). The opposite
enormous probability of mistake if rejecting the occurs for the case of the defenders, who seem to
null of normality. be undervalued when estimating the regression
We start analyzing the Monopsony model, in with (LigaFantastica). Remember that these re-
which we suppose that all soccer players share gressions have already tried to capture the sporting
substantially similar labor assets and, conse- contribution of each player through the corre-
quently, are economically valued according to sponding performance index.
their individual contribution to the team income Finally, attending to the estimations of the
(i.e. we do not introduce dummies for superstars). Monopsony model, it seems that (RealMadrid)
The estimations are shown in column 1 for the case and, to some extent, (Espanyol) reward their
of (PuntosMarca) and in column 2 when (Liga- players more than it would be expected, as regards
Fantastica) is used. First, the explanatory power of to the usual practices of the other clubs. The
the model is substantially high, as the R2 is about opposite can be stated of Osasuna. Note that we
0.70. Second, the results clearly display the fact have included in the regressions as many dummies
that the commercial attractiveness of players is an as teams in the league, but we only report the
important factor in explaining their market value. statistically signicant ones in Table 3.
As expected, the sporting performance}either At this stage, we are ready to tackle the other
measured by (PuntosMarca) or by (LigaFantasti- central issue of this study: checking the presence of
ca)}is a main explanatory factor of players the winner-take-all hypothesis in the Spanish
economic valuation. And yet, the public exposure soccer labor market. In order to do that, we
of players (FilteredGoogle) appears as a major propose to retrace the model incorporating a
factor in determining the soccer players market number of dummy variables. For the soccer
value. We consider this result as one of the most superstars, it is expected that the initial relation-
relevant ndings of the paper. ship assumed between market value and perfor-
Another interesting result is that players who mance, does not hold true anymore. The marginal
have played matches with the national team higher level of sport or economic performance that
(International) experience a higher economic those players represent should then be associated
reward. Playing matches in international competi- with a signicant positive jump in their economic
tions, like the Champions League and the UEFA valuation. As explained, our proposal is to
Cup (Europe), also generates a positive signal, indirectly identify those superstars by matching
from which the player takes economic advantage. the higher extreme values of (Google).17
According to our data set, European non-Spanish On one hand, the majority of the players, with
players seem to be systematically overrated less than 800010 000 web appearances in Google,
whereas the opposite occur to non-European seem to follow a similar path. On the other hand,
foreign players. The t-statistics are just on the there are 5 clear outliers (there is a substantial
border of the signicance levels, though. This positive jump) and another 1020 players subject
result suggests the presence of some geographical to a special pattern.18 In accordance with sugges-
discrimination or weaker bargaining power of tions in the theoretical framework, we have
non-European players. In any case, there is no created two dierent kinds of dummy variables.
evidence for claiming discrimination on the part of One type gathers the league-superstars: those
national players, in contrast with commonplace players who have the higher Internet coverage
comments in Spanish sports circles on this issue. within the Spanish League (because of the digital
The inuence of age, which also relates experience, press, the fans sites, etc.). The exact number of
is introduced into the regression in the conven- superstars identied is evidently a matter of choice.
tional quadratic form. However, this does not We have run the regressions using three dummies
have any statistical relevance. of this type. The rst dummy is (Winner-all-5) and
Concerning the inuence of players demarca- collects the 5 league top players; the second one,
tion on economic valuation, we can observe that, called (Winner-all-10), gathers the ve second-best
with respect to goalkeepers (who are the reference players. Finally, (Winner-all-20) is created to

Copyright # 2007 John Wiley & Sons, Ltd. Manage. Decis. Econ. 28: 5770 (2007)
DOI: 10.1002/mde
 HIDDEN MONOPSONY RENTS IN WINNER-TAKE-ALL MARKETS 67

account for the players whose Google records rank enrolled by Espanyol. The overrating due to
between the 11 and the 19th position. Similarly, participating in European (Europe) or Interna-
another sort of dummy (Team-winner), is con- tional games (International) is now more signi-
structed to identify 18 team-superstars}one from cant and its quadratic term works better. More
each team}who have been collected by selecting importantly, the inuence of public exposure
the players who present the highest (Google) (FilteredGoogle) on market value is maintained
record of each particular team.19 The results of and is important to determine the payment scheme
adopting this procedure are shown in columns 3 in the soccer market.
and 4 of Table 3, and are the preferred ones.
Comparing these results to those in columns 1
and 2, one can conclude the crucial role of the
winner-take-all hypothesis in this particular labor
market. The examination of columns 3 and 4, CONCLUSIONS
teaches us that regardless of the measure chosen to
capture sporting performance (PuntosMarca or This study has examined new arguments to explain
LigaFantastica), the scission of the market of the contradictory evidence that Spanish soccer
soccer players into two groups is meaningful. This clubs do not prot from the expected monopsony
conclusion emerges from the analysis of the rents. The soccer industry has been described as a
estimated coecient for the dierent dummies dual market, in which one segment of the market is
that have been devised to capture the winner-take- characterized within a monopsonistic framework,
all element. The size of the coecient of (winner- while the other is composed by a small number of
take-5) is considerable, and it is strongly statisti- outstanding labor inputs that accumulate market
cally signicant (t-statistic above 9.1 in the two power. The empirical analysis is carried out using
regressions). It means that the selected group of data from the Spanish professional soccer league
the best 5 league superstars appears to receive a (LFP), for the season 2001/2002. Throughout the
positive additional extra valuation (compared with study, especial attention has been paid to the
the remaining 364 soccer players of the data set). economic impact associated with each particular
This fact also holds, even if not in such a big player, as far as it helps to explain the wage
signicant manner, for the case of the ve players dierentials that could not be explained due to
represented in (Winner-all-10) and the 18 of sporting performance dierentials.
(Team-winner). Interestingly, note that (Winner- The empirical signicance of the corresponding
all-20) would be statistically signicant if (Team- variables allows us to conclude that there are, in
winner) were left aside of the estimations. How- eect, two segments in the labor market of soccer
ever, the eect of the latter seems to be stronger, players (We consider that one can condently
supporting our thesis that winner-take-all elements make an extrapolation of our results from the
also operates within the teams sphere and not just Spanish LFP to other major sport leagues).
at the league level. Renowned players receive an increase in reward,
In addition, the last analysis permits us to which is out of proportion to the marginal increase
conclude that the number of top league-superstars of sport and economic contribution that they
in the season 2001/2002 was presumably 10, since provide to their team, suggesting the presence of
the coecient for (Winner-all-20) is neither statis- the winner-take-all elements. The existence of this
tically signicant when using (PuntosMarca), feature in the Spanish soccer league can thus help
column 3; nor it is signicant when using us to understand why most of the teams are unable
(LigaFantastica), column 4. Due to the big to generate extraordinary prots even if the global
disparity between the coecient for (Winner-all- structure of the market would allow them, in
5) and for (Winner-all-10), it can also be claimed theory, to benet from monopsony rents. Our
that anyway there seem to be dierent categories empirical results are in line with the theoretical
of league-superstar. belief that an important share of the monopsony
Note also that when taking into account the rents captured from ordinary soccer players, are
peculiar rewarding behavior of top superstars, absorbed in a second moment by those few
Real Madrid players seem no longer to be over- superstars who enjoy strong bargaining power.
valued; whereas it is still the case of the players The addition of the winner-take-all issue into the

Copyright # 2007 John Wiley & Sons, Ltd. Manage. Decis. Econ. 28: 5770 (2007)
DOI: 10.1002/mde
68 P. GARCIA-DEL-BARRIO AND F. PUJOL

econometric specication improves the results playing skills would be the same with or without the
reached with the Monopsony model. reserve clause. Hunt and Lewis (2003) argued that,
We have been able to reach these conclusions as long as property rights are well dened and
reconstructing through sales of players contracts is
largely due to the incorporation of an inventive and permissible and costless, only the distribution of
reliable proxy measure of football players popu- wealth between players and owners would be
larity. This measure, based in the number of aected. According to this view, protable invest-
Internet sites referring to each individual player, ments in teams are still possible in the absence of
is straightforward and has been used in this monopsony rents, since monopoly prots remain.
They justify this by considering the whole league
research in an innovative way, at least in the eld (instead of the team) as a monopoly, rather than as
of economics and as far as we know. The ltering an inecient cartel. Quirk and Fort (1992), in
method that we have used increases the robustness Chapters 57, explore the division of monopoly
of this variable as a pertinent measure of a football rents between players and owners.
players popularity, since we have purged all 4. They also stress that excess competition is strongest
when the activity is zero sum and when competition
Internet impact factors directly produced by itself is worthless. If only rank order (not perfor-
sporting performance. This measure of popularity mance level) matters, the game is pure lotto and
constitutes one of the key elements, which has resources used to improve performance to gain
enabled us to test the hypothesis of the labor competitive edge are wasted.
market structure in the soccer industry. Right now, 5. The winner-take-all hypothesis, as stated by Frank
and Cook, received some critical remarks in the
almost no study related with the sport and review made by Rosen (1996), who believes that the
entertainment industry incorporates popularity apparent abundance of superstars today is unrelated
indexes. And yet, we fear that such a variable can to the decrease in wages of unskilled labor that
eventually play a signicant role in helping to dominate inequality statistics.
explain the economic behavior in the sport and 6. The fact that the soccer clubs seldom make money is
a common feature of all the European soccer
entertainment industry, at least as a control leagues. For the case of the English leagues, the
variable. The popularity measure that we propose fact is documented by Szymanski and Kuypers
in our study, even if it is somehow time demanding, (2000). In other study, Szymanski and Smith (1997,
can thus enhance the robustness or, in the opposite, p. 2426) state that only 7 out of 40 English team
lead to a revision of many empirical results. averaged prots over the period 19781997, and
these prots were small.
7. As concerns the empirical analysis both approaches
are quite similar in practice. Firstly, the regressions
NOTES have to be estimated upon data on individuals, even
if the model recognized duality in the skills.
1. The reward given to soccer players is so large that Secondly, Scully (2004b) seems to conceive the
there is no reason why lower earnings would not veteran free agents as a limited pool of high-quality
elude a comparable eort and performance from players, which greatly resembles our understanding
them. Hence, the eciency wage hypothesis does of superstar players.
not seem a crucial issue here, since there are inherent 8. The players MRP in baseball is the ability or
incentives, other than monetary ones, that persuade performance that he contributes to the team and the
soccer players to make their best. eect of that performance on gate receipts. The
2. See, for instance, Ascari and Gagnepain (2003). The eect of player performance on revenue may be
paper by Scully (2004b) initially considers the direct or indirect. Ability contributes to team
markets of professional team sports as free labor performance and victories raise gate receipts and
markets, at least for eligible veteran players. Never- broadcast revenues; this is the substantial eect of
theless, he acknowledges that many non-veteran the individuals performance. Additionally, it is
players are paid below their net marginal revenue possible that some players may attract fans over
product. On the other hand, we will mirror Scully and above their individual contribution through the
(2004b) in a number of ways. First, he presumable team. (Scully, 1974: 916). Since Scully, other studies
thinks of these labor markets as dual markets, in have tried to measure the players MRP in dierent
which the inequality between veteran and non- ways (see, for instance, Berri, 1999). For a descrip-
veteran player compensation is continuously in- tion of the market for players in the US, describing
creasing. Second, he notices that a remarkable part the structure of their wages and the distribution of
of player compensation must be conceived as a their earnings, see Scully (1995).
share of league revenue. 9. Since the seminal paper by Sloane (1971), many
3. Another main policy issue is the allocational eects researchers have rejected the idea that soccer clubs
that result from monopsony practices. El Hodiri are pure prot miximizers. Instead, they argue that
and Quirk (1971) proved that the distribution of the clubs maximize a composite objective function

Copyright # 2007 John Wiley & Sons, Ltd. Manage. Decis. Econ. 28: 5770 (2007)
DOI: 10.1002/mde
 HIDDEN MONOPSONY RENTS IN WINNER-TAKE-ALL MARKETS 69

based on attendances, recognition, playing perfor- 16. We want to acknowledge the comments of one
mances, and prots. referees of the Journal, which helped us to improve
10. For a general view of the type of industry that is the statistical properties of the estimated model.
going to be examined here, it may be helpful to Table 3 shows the results of using log(MarketVa-
consult Ascari and Gagnepain (2003), as well as lue), as well as log(Google) and log(PuntosMarca)
Hoehn and Szymanski (1999). or log(LigaFantastica). Had we applied the loga-
11. Diario Marca is the main sports daily journal in rithmic transformation solely to the dependent
Spain. The records for (MarketValue) are reported variable, we would have appreciated no meaningful
in the issue printed in the 22nd September 2002. dierences in the estimations.
They are established by a board of experts, that 17. The records of the Google search for each individual
meets under ocial supervision and accord the player are available on the Internet: Cf.: Pay and
accurate value of the individual, both relying on the performance in Spanish soccer league: who gets the
quality of the player and depending on the economic expected monopsony rents? Universidad de Navarra:
contribution that the player is supposed to draw for WP 05/04 (http://www.unav.es/econom/investiga-
the team. This index is provided to the readers by cion/working/wp0504.pdf).
MARCA in order for them to elaborate a hypothe- 18. This feature can be seen by plotting the data. Even if
tical team, which will compete in the game: Liga we use a logarithmic presentation of the Google
Fantastica. For the 53 cases in which the rank was values, the uncommon path of the top 5 soccer
not complete for this year, we have taken the value players remains. The 5 league superstars in that
reported for the previous season (Cf. diario MAR- season, as far as (Google) records concern, are:
CA, 27.IX.2001), after checking the consistency of Raul, Zidane, Figo, Roberto Carlos and Rivaldo.
the information. 19. We did also test the winner-take-all hypothesis by
12. It consists of a cumulative rank, elaborated upon means of creating dummy variables related to the
the records (between 0 and 3) that journalists assign indexes of sporting performance. The corresponding
to each soccer player after the game. At the end of results yielded basically the same conclusions. In
the season, the ranking is obtained adding up the particular, the estimations were more satisfactory
points granted to each individual along the league. for the case of (LigaFantastica) than for the case of
This index is a natural measure of productivity, (PuntosMarca). Anyway, as far as the variable used
including aspects that could hardly be quantied to construct the dummies was Google (instead of
otherwise. We use the average of the ranking values FilteredGoogle), the chosen procedure to identify
for the seasons: 2000/2001 and 2001/2002. winners takes simultaneously into account the
13. This alternative index is computed by adding sporting performance as well as the economic
quantiable indicators on play performance in impact of the players.
addition to the score given by the journalists. It
means that the ranking provided by (LigaFantasti-
ca) enriches and encompasses that of (PuntosMar-
ca). We also use the average of the ranking values
for the last two seasons. REFERENCES
14. The specic values associated to each footballer
were those that resulted of including in the search: Ascari G, Gagnepain P. 2003. How inecient are
name of the player AND name of the team AND football clubs? An evaluation of the Spanish arms
football. (The last element was written in the most race. http://economia.unipv.it/docs/dipeco/quad/ps/
common languages in which soccer has popular q152.pdf [February 2003].
support.) This rank was computed in May 2003. Berri J. 1999. Who is Most Valuable? Measuring the
15. The estimations of the Google regressions are not players production of wins in the National Basketball
shown, since they escape from the goal of the paper, Association. Managerial and Decision Economics 20:
but they can be found in Pay and performance in 411427.
Spanish soccer league: who gets the expected Dobson S, Goddard J. 2001. The Economics of Football.
monopsony rents? Universidad de Navarra: WP Cambridge University Press: Cambridge, UK and
05/04 (http://www.unav.es/econom/investigacion/ New York, USA.
working/wp0504.pdf). It is worth succinctly men- El Hodiri M, Quirk J. 1971. An Econometric Model of a
tioning that Internet presence depends, in a statis- Professional Sports League. Journal of Political
tically signicant way, on the soccer player sporting Economy 79: 13021319.
performance (either measured by PuntosMarca or Frank RH, Cook PJ. 1995. The Winner-take-all Society:
by LigaFantastica). It also depends on the number How More and More Americans Compete for Ever
of matches played in international competitions, at Fewer and Bigger Prizes, Encouraging Economic
teams level (Europe), as well as at country level Waste, Income Inequality, and an Impoverished Cultur-
(International). The age seems not to have a decisive al Life, Simon and Schuster, Free Press, Martin
impact on the relative presence of the players in the Kessler Books: New York, London and Toronto.
Internet. With respect to the position of soccer Hoehn T, Szymanski S. 1999. European football. The
players in the pitch, only the overexposure of structure of leagues and revenue sharing. Economic
attackers is statistically signicant. Policy: A European Forum 14(28): 203240.

Copyright # 2007 John Wiley & Sons, Ltd. Manage. Decis. Econ. 28: 5770 (2007)
DOI: 10.1002/mde
70 P. GARCIA-DEL-BARRIO AND F. PUJOL

Horowitz I, Zappe C. 1998. Thanks for the memories: Scully GW. 1974. Pay and performance in major league
baseball veterans end-of-career salaries. Managerial baseball. American Economic Review 64: 915930.
and Decision Economics 19: 377382. Scully GW. 1995. The Market Structure of Sports.
Hunt JW, Lewis KA. 2003. Dominance, recontracting, University of Chicago Press: Chicago and London.
and the reserve clause: major league baseball. Scully GW. 2004a. Introduction to the special issue on
American Economic Review 66(5): 936943. sports economics. Managerial and Decision Economics
Neale P. 1964. The peculiar economics of professional 25: 6162.
sports. Quarterly Journal of Economics 78: 114. Scully GW. 2004b. Player salary share and the distribu-
Noll RG. 1974. Attendance, prices, and prots in tion of player earnings. Managerial and Decision
professional sports business. In Government and the Economics 25: 7786.
Sports Business, Noll R (ed.). The Brookings Institu- Sloane P. 1971. The economics of professional football:
tion: Washington, D.C. the football club as utility maximiser. Scottish Journal
Quirk J, Fort RD. 1992. Pay Dirt: The Business of of Political Economy 17: 121146.
Professional Team Sports. Princeton University Press: Szymanski S. 2001. Income inequality, competitive
Princeton, NJ. balance and the attractiveness of team sports: some
Rottenberg S. 1956. The baseball players labor market. evidence and a natural experiment from English
Journal of Political Economy 64: 242258. soccer. The Economic Journal 111: F69F84.
Rosen S. 1981. The economics of superstars. Journal of Szymanski S, Kuypers T. 2000. Winners and losers. The
Political Economy 79: 13021319. Business Strategy of Football. Penguin Books: Lon-
Rosen S. 1996. Review of: the winner-take-all society. don, England.
Journal of Economic Literature 34(1): 133135. Szymanski S, Smith R. 1997. The English football
Rosen S, Sanderson A. 2001. Labor markets in industry: prot performance and industrial structure.
professional sports. The Economic Journal 111 International Review of Applied Economics 11:
(February): F47F69. 135153.

Copyright # 2007 John Wiley & Sons, Ltd. Manage. Decis. Econ. 28: 5770 (2007)
DOI: 10.1002/mde