Salary differences under the salary cap in Major League Soccer

Abstract

Professional soccer is the world’s most popular sport; a number of National Leagues are under the control of National Associations. The economic theory behind soccer is the continuing competition to earn much more than other sports do in the sports market. Since the supply of talent is limited, teams’ demand for certain professionals is so strong that it leads to salary differences between players. Therefore, in this study, attention is given to the determinants of the differences in workers’ salaries in the Major League Soccer labor market using Generalized Least Squares (GLS) estimation on panel data from 2007 to 2016. Birth place is the most influential determinant of a player’s salary, along with a player’s position, a player’s age, whether the player has a national team duty, and the number of games in which the player started in the first eleven. Conversely, moving from one Major League Soccer team to another and the number of games played as a substitute have a negative effect on players’ salaries.

Keywords

Association football labor market sport economics talent

Introduction

Sporting events are a worldwide economic phenomenon. Around the world, soccer, baseball, basketball, and hockey are more than just games. In particular, soccer is a major spectator sport and has a long history. Among “ball sports,” modern soccer can be traced back to the 19th century, when it was developed in the elite public schools of England. Prior to the Industrial Revolution, sports and leisure activities were luxuries for the population, consisting mainly of poor farmers who had to devote their hours to growing enough food to survive. After the Industrial Revolution in England, the living standards for most of the population improved and the working class could spend time on sports and other leisure activities. The growing popularity of soccer among the working class led to the professionalization of soccer. In the meantime, upper-class sportsmen’s roles changed from amateur players to owners of professional teams.¹

The economic approach to understanding professional soccer is somewhat different from many academic disciplines. Sociologists, psychologists, and historians focus on the context, while economists primarily adopt a very different perspective on the issues.

Therefore, the economics of sports is based on the use of theories and reproduced models to generate estimations and look for alternative methods of development within and across sports teams. It is clear that economists can produce useful explanations about the sports market by making predictions about sports teams and players.²

Professional sports labor markets have their own unique properties and provide great advantages for academic research. As Kahn³ states,

Professional sport offers a unique opportunity for labor market research. There is no research setting other than sports where we know the name, face, and life history of every production worker and supervisor in the industry. Total compensation packages and performance statistics for each individual are widely available, and we have a complete data set of worker-employer matches over the career of each production worker and supervisor in the industry.

Moreover, the rise of professional sports teams in soccer, the effectiveness of salary caps and wage determination in the players’ labor market, the determination of transfer fees, the professionalism in sports, and the superficial demand for the commercialization of sports seem like plausible solutions with respect to exploring economic studies of the players’ labor market. The aim of this paper is to provide an economic modeling of player remuneration under the salary cap in Major League Soccer (MLS) and to consider the key factors affecting players’ salary differences.

Literature review

In 2013, professional sports, including players’ salaries, television contracts, gate receipts, and merchandise sales, generated over $305 billion in revenue in the USA.^a With the industrialization of professional sports, sports clubs started enjoying much higher levels of revenues from sports-related activities and products, which led to higher salaries for professional players. In many cases, popular players are paid much more than the CEOs of well-known companies. Consequently, high salaries in professional sports have raised a clear question in labor economics: are players paid what they are worth, or are they paid too much? In this research, determinants of workers’ salaries in the sports labor market—namely, MLS, the top-tier professional soccer league in the USA and Canada—are analyzed using panel data from 2007 to 2016. It is hoped that the insights gained from a specific focus area could be a useful inference for future research.

Many researchers suggest that the relationship between team salaries and players’ performance in the world depends heavily on a loose corporation.^4–6 Those statements are not supported by empirical evidence in general and influenced by an analysis of soccer. Scully⁷ states that increasing expenditures on coaching, transferring experienced players and playing performance are simply necessary conditions for improving a team’s win percent ratio in the league; indeed, they are not sufficient to improve the situation. Szymanski⁸ analyzes the performance of English soccer and finds a positive relationship between players’ performances and the salaries they earn. He interprets this to mean that players with the most talent gravitate to those teams best suited to their experience, performance, and requirements. To the best of our knowledge, teams are trying to maximize their wins while putting more effort into having the most experienced players.

There has been an increasing amount of research on sports analysis and player salary in recent years. However, most of this research concerns the relationship between players’ salaries (including the “superstar effect”) and team success. Trustworthy research with sound results regarding the empirical side of the professional soccer leagues is very rare in the current literature. The main obstacle is that in many countries, players’ salaries are not disclosed to the public; hence, not much reliable data exist for academic research. For this reason, most studies in the existing literature use estimates or proxies for players’ salaries.

Since the birth place of soccer is Western Europe, soccer is the dominant and most popular sport in and around Europe. Today, European leagues are considered the highest level and highest quality soccer competition arena for fans and players. This popularity has also attracted researchers’ interest. Much work in this field studies European leagues, mostly the German Bundesliga, the English Premier League, the Spanish Primera Division La Liga, and Italian Serie A and Serie B. Feess et al.,⁹ Frick,¹⁰ Lehman and Schulze¹¹,and Battre et al.¹² are just a few of many studies that explore the determinants of differences in players’ salaries in the German Bundesliga using a cross-section or panel data covering various short-time spans. Garcia-del-Barrio and Pujol^13,14 study Spanish La Liga’s 2001–2002 season, while Lucifora and Simmons¹⁵ work on Italian Serie A and Serie B during the 1995–1996 season. Employing different econometric approaches (mostly Ordinary Least Squares (OLS)), the most significant and positive factors that affected player salaries in almost all the studies were age, an international cap, status as a midfielder or forward, region of origin, number of games played, number of goals scored, and number of assists made.

Szymanski⁸ analyzes race discrimination in salaries in the English soccer league for the period 1978–1993, using a balanced panel data of 39 clubs. This study provides a specific example of unequal pay for equal work by looking at players’ market efficiency and how race can relate to a team’s willingness to pay higher salaries for a certain type of player. Following the literature on the salary cap in soccer,^16–18 Szymanski⁸ uses panel data set on salary to measure team salary inequality. Gaining prominence in a team correlates with differences in salaries according to playing positions and should be positively related to the productivity of the players. As a further measure of salary differences, the number of players who have played in the first eleven is also included. Therefore, it is expected that MLS players on the team with experience should be correlated with a higher salary.

MLS players’ official salary information is shared with the public by the Major League Soccer Players Union. Although MLS^b is one of very few soccer leagues that can be studied with a reliable salary data set, little research has focused on MLS soccer players’ labor market. Moreover, those existing studies on MLS use smaller samples or short periods—mostly just one season.

Lee and Harris¹⁹ look at the relationship between player performance and salary in MLS for the seasons 2007 to 2009. They employ various econometric techniques, including fixed effects regression, robust regression, Tobit regression, and two-stage least squares regression. When they use fixed effects, robust and Tobit regression, goals, assists, and minutes of play are significant and positive while the number of games played is significant and negative. On the other hand, in their TSLS model, the number of games played exhibits a positive and significant effect on player remuneration. Therefore, they conclude that “games played” is inconclusive.

There are several problems with their econometric approach. Minutes of play, number of games played, goals, and assists should not appear as explanatory variables in the same estimation equation since they are all correlated. Playing more games means more minutes of play and, thus, more chances to score and assist. The authors attempt to address this issue by employing two-stage least squares estimation in their fourth model, but they use goals and assists in the current year. In sports, contracts are signed at the beginning of the season and players’ past performance and career statistics determine salaries. In each season, salary is first determined and the contract is signed; then a player’s performance is observed. Hence, in such models, lagged values of the proxies for player performance should be used. In addition, if the data set contains information about goalkeepers and defenders along with midfielders and forwards, position-specific performance measures such as goals and assists will be misleading measures for the players in these positions. For instance, a goalkeeper has almost no chance of making an assist or scoring a goal during the game. Our model in this study addresses these variable problems by paying attention to proxies that will be used to measure player performance; our study employs random effects GLS estimation.

Zimbalist²⁰ tries to analyze the relationship between payrolls and team performance. He finds that if teams are doing well in the mid-season, they are more likely to spend on better players to increase the win percent ratio. Therefore, with regard to his analysis, the increment of players’ salaries results from the anticipation of achievement of the plays as well as players’ performance. Kuethe and Motamed²¹ focus on the superstar effect on player earnings using 2007 MLS salary data consisting of 193 players. Their results show that experience, ethnicity, and the international cap are significant and positive factors. Suspiciously, age exhibits a negative effect on salary, and players’ positions are not significantly different from zero. According to their contradictory findings, experience has a positive effect, but age has a negative effect on players’ salaries. Reilly and Witt²² examine pay determination in the MLS labor market using OLS and median regressions. The data set is from only one league season, namely the 2007 MLS league season. Although their model does not include any position-specific performance measures, goalkeepers are excluded from their data set.

Conclusions derived from limited data sets and contradictory results in the existing literature on the MLS labor market forced us to perform a more reliable, accurate, and consistent analysis with a much larger data set. In this current study, the panel data set for 10 years (2007–2016), consisting of 4905 observations of 1734 players under contract by any MLS team for at least one season, is used for the investigation. Results from this study provide more sound and trustworthy information about the determinants of MLS players’ salaries. As noted earlier, multiple variables must be analyzed when seeking the determinants of higher salaries for certain players on a team. This comparative study, therefore, has an advantage over previous studies because it provides clearer estimates of the relationship between salary differences and the number of games played, the country of origin and the experience of the player.

Data

In MLS, there is a limit, called a salary cap, on the amount of money that teams can spend on player salaries and fees. For the 2017 season, each team is allowed to spend only $3.845 million on player salaries in total and the maximum amount a player can be paid is $480,625. One of the critical points in MLS history was the introduction of the Designated Player Rule in 2007. This rule is the most important characteristic of MLS compared to European soccer leagues. There is no salary cap; hence, there is no designated player rule in any European soccer league. The British star player David Beckham was the first player who signed a contract under this rule, also known as the Beckham Rule. Under this rule, some of the player’s salary is charged to the salary cap and paid by MLS, while the rest of the salary is paid by the team owner. There is no limit to the portion paid by the team owner. This rule allowed clubs to pay higher salaries for high-quality players. Thus, MLS teams became more competitive for star players in the international soccer market. Since 2007, there has been no such major change in the rules in MLS that may affect players’ salaries. For this reason, 2007 is a good point at which to start an analysis of players’ salaries. In this current research, unbalanced panel data on players’ annual base salaries from 2007 to 2016 in MLS is studied. Players’ salary information is gathered from the official Major League Soccer Players Union website, mlsplayers.org. The data set includes base salary information for all MLS players under contract during these years; the numbers exclude any signing, marketing, or guaranteed bonuses. Players’ characteristics—such as age, birth place, playing position, club information, career statistics such as experience in MLS, national team duties, number of games started in the first eleven and number of games pitched as a substitute, number of goals and assists, and number of clean sheets and goals allowed (for goalkeepers)—are from the official MLS website, mlssoccer.com.

The data set contains 4905 observations of 1734 professional soccer players under contract for at least one year in MLS between 2007 and 2016. When changes in labor wages over time are studied, the effect of inflation should be considered. Inflation-adjusted salary data provides more accurate results than do nominal salary values. In this study, players’ salaries are corrected for inflation, with 2007 as the base year.^c Since designated players receive higher salaries, to control for this effect a dummy variable is introduced for designated players. MLS experience is the total number of years a player was on an MLS team’s roster prior to the beginning of each season. Youth dummy is equal to 1 if the player ever had his name in his national youth team’s roster but had not yet made it to the highest level national team in his career, while the international cap is the dummy variable for the players who appeared in a senior-level national team roster at least once. Instead of an exact birthdate, birth year is used to calculate the age of each player. Club dummies are introduced for 20 clubs competing in MLS. Finally, region/continent dummies are used to categorize players based on their birth place. For this purpose, this study refers to the United Nations Statistics Division’s country groupings for the continents and geographical sub-regions. A list of variables in the data set and their descriptions are given in Table 1.

Table 1.

Descriptions of variables.

Variable	Description
Ln (Real Base Salary)	Natural logarithm of the annual base salary corrected for inflation
Designated	1 if the player is a designated player for that year
Age	Player’s age (birth year is used)
Youth	1 if the player has ever had his name on his national youth team’s roster at least once but could not make it to the senior-level national team
International cap	1 if the player has ever had his name on a senior-level national team’s roster at least once
Transfer	1 if the player is transferred from an MLS team to another MLS team
MLS experience	Total number of years the player was on an MLS team’s roster prior to the beginning of the season
Birth place dummies
North America (ref. group)	1 if the player was born in a North American country
Central America	1 if the player was born in a Central American country
South America	1 if the player was born in a South American country
Western Europe	1 if the player was born in a Western European country
Northern Europe	1 if the player was born in a Northern European country
Southern Europe	1 if the player was born in a Southern European country
Eastern Europe	1 if the player was born in an Eastern European country
The Caribbean	1 if the player was born in a Caribbean country
Africa	1 if the player was born in an African country
Asia	1 if the player was born in an Asian country
Oceania	1 if the player was born in an Oceanian country
Position Dummies
Goalkeeper (ref. group)	1 if the player is a goalkeeper
Defender	1 if the player is a defender
Midfielder	1 if the player is a midfielder
Forward	1 if the player is a forward player
Lag (games started)	Number of games the player was in the first eleven in the previous season
Lag (games substituted)	Number of games the player went on the pitch as a substitute in the previous season
Lag (goals per game)	Number of goals per game the player scored when he played in the previous season
Lag (assists per game)	Number of assists per game the player made when he played in the previous season
Lag (shutout per game)	Number of shutouts/clean sheets per game the goalkeeper had in the previous season
Lag (goals allowed per game)	Number of goals per game the goalkeeper allowed in the previous season
Club dummies	Dummy variables for 20 MLS clubs
Year dummies	Year dummies (2007–2016)

Gini coefficient is the most widely used statistic to measure income inequality among individuals. Its value ranges from 0 to 1, where 0 represents perfect equality and 1 indicates perfect inequality. A Gini coefficient of 0.6 or above means extremely high income inequality, while any value between 0.5 and 0.59 means relatively high levels of income inequality. Table 2 and Figure 1 provide Gini coefficients for salary dispersion for each year from 2007 to 2016. The league-wide Gini coefficient reaches its highest value of 0.647 in 2015 and its minimum value of 0.544 in 2013. These numbers clearly show that the MLS labor market is shuttling between relatively high and extremely high levels of income inequality. Table 2 also reveals that the highest levels of income distribution inequality are among midfield players and forward players.

Table 2.

Gini coefficients of salary distribution in each year from 2007 to 2016.

	2007	2008	2009	2010	2011	2012	2013	2014	2015	2016	Average
Goalkeeper	0.490	0.470	0.427	0.343	0.398	0.350	0.354	0.351	0.331	0.471	0.399
Defender	0.431	0.438	0.348	0.533	0.496	0.486	0.372	0.400	0.416	0.409	0.433
Midfielder	0.724	0.714	0.665	0.609	0.591	0.565	0.511	0.604	0.697	0.687	0.637
Forward	0.582	0.604	0.513	0.631	0.659	0.652	0.687	0.728	0.691	0.672	0.642
League-wide	0.632	0.627	0.567	0.588	0.579	0.567	0.544	0.606	0.647	0.644	0.600

Figure 1.

Season-by-season change in Gini coefficient in MLS.

For further analysis, as seen in Table 3, forward players have the highest median real base salary, while goalkeepers have the lowest. Forward players earn almost 43% more than do goalkeepers. To determine how the dependent variable—the natural logarithm of inflation-adjusted base salary—varies over time and over individuals, “between variation” and “within variation” are also compared in Table 3. When the whole data set is considered, standard deviations are 0.943 for “between variation” and 0.374 for “within variation.” It is noteworthy that the data set exhibits higher variation in the base salary from one individual player to the next while the variation in the salary of a player over time is smaller. Therefore, differences among players are much more deterministic than differences over time. For a visual illustration and comparison Figure 2 also provides median nominal base salary season-by-season in MLS by playing position. Since there are extreme values in the data set, median values are used instead of mean values in Figure 2.

Table 3.

Statistics about the real salary distribution of the players in MLS (2007–2016).

	Number of players	Number of observations	Number of designated players	Average real salary	Median real salary	Within variation of Ln (real base salary)	Between variation Ln (real base salary)
Goalkeeper	167	550	2	80,602	54,094	0.353	0.768
Defender	520	1526	6	100,670	66,451	0.360	0.751
Midfielder	671	1835	56	184,872	71,300	0.371	1.021
Forward	376	994	59	230,648	77,183	0.412	1.063
All players	1734	4905	123	156,260	68,348	0.374	0.943

Figure 2.

Season-by-season median salary.

Model

This study intends to explore the factors that cause dispersion in player salaries in MLS using panel data from 2007 to 2016. For this purpose, a natural logarithm of the inflation-adjusted annual base salary is used as the dependent variable. To get the most out of the data set, in the first model, position-specific performance measures (such as number of goals scored, number of assists made, number of shutouts, and number of goals allowed) are excluded. This allows us to use the salary information of all players in the regression. A set of variables for non-position-specific player performance, individual player characteristics, control variables for clubs, and the year dummies are introduced in the first model.

As discussed previously, designated players receive higher salaries. To find the salary difference between designated players and players under salary cap, a dummy variable, designated, is introduced to the model. In the professional sports labor market, older and more experienced players are more valuable than the rookies. To control for the effects of these variables, age and MLS experience are used as regressors. Age is measured in years, while MLS experience is the total number of years spent in MLS prior to the beginning of each season. Since no reliable racial information exists with respect to individual MLS players, readers should be careful about any research investigating the effect of race on pay differences. In this study, to investigate salary differences, a dummy variable is used for the birth place of the player rather than his race. Birth places of MLS players are split into geographic regions such as North America (reference group), Central America, South America, the Caribbean, Northern Europe, Southern Europe, Eastern Europe, Western Europe, Africa, Asia, and Oceania. To further investigate and compare the salary differences relative to the top-tier soccer world, Europe is split into four sub-regions.

The playing position of the player is another determinant of the salary under investigation, since scoring a goal or saving a shot requires much more skill than simply clearing a ball in defense. For this reason, position is the set of dummy variables for playing positions in soccer—goalkeeper, defender, midfielder, and forward—where the goalkeeper is taken as the reference group. We would like to call attention to position dummies. These dummies control for the goal-scoring ability and goal productivity of the player as well.

Since it is a sign of quality and skill, players who serve in their senior-level national teams are expected to receive higher salaries. In addition to the international cap dummy variable, the youth dummy variable is used for players who are currently or were formerly in youth-level national teams but who have never been an international cap.

In soccer, payments are set by long-term contracts. Frick¹⁰ calculates that the average contract duration in German Bundesliga is about three years. Transferring to another club means a new contract; hence, it may affect players’ salaries. The transfer variable is a dummy variable equal to one if a player transfers from an MLS club to another MLS club.

Statistics such as the number of goals, assists, shots, offsides, fouls committed, yellow cards, and red cards are position-specific measures of player performance. Forward players and attacking midfielders are more likely to have higher numbers of goals scored, assists, shots, and offsides, whereas defenders and defensive midfielders are more likely to commit fouls and receive yellow and red cards. A goalkeeper has almost no chance to score, shoot or be caught in offside. Therefore, these position-specific measures cannot be used as performance indicators for all MLS players. The number of games the player played is the most common variable used in the models in the literature but it may also be misleading. A player who played 90 min and a player who went on the field as a substitute in the 89th min will add one more game to their number of games played. An econometric estimation that uses the number of games played as an explanatory variable will treat these two players’ performances the same. However, the performances of these two players are not the same. For this reason, unlike the existing literature, this study uses two different performance measures, namely, the number of games in the first eleven and the number of games as a substitute. These variables are not position-specific and they are common performance indicators for all players. There are other advantages of using these variables as a performance measure. When a player is the best in his position on the team, he will start in the first eleven. For factors (such as injuries) that affect individual player performance, red cards will be reflected in this statistic. Players who perform poorly, are frequently injured, or are suspended due to cards will have a smaller number for the games they started in the first eleven. The same arguments are also valid for the number of games as a substitute when the player is not performing well enough for the first eleven but is preferred as a player of hard times and strategic decisions during a game.

As mentioned previously, the number of games played and the position-specific variables should not be used as independent variables simultaneously in the same estimation equation. Otherwise, it will cause a multicollinearity problem, as players who play more games will have higher statistics for the number of goals scored, assists, offsides, fouls committed, and yellow and red cards received. Battre et al.¹² found that in German Bundesliga the most recent performance, i.e. in the last season, has a greater impact on players’ salaries than a previous career performance. Hence, in the first model here, the number of games started in the first eleven in the previous season and the number of games as a substitute in the last season are used as performance indicators rather than the current year. Finally, year is the year dummy to control for year effects, club is the dummy to capture fixed club-specific characteristics and ɛ_it is the error term

\begin{matrix} Ln (RealBaseSalary)_{i, t} \\ = β_{0} + β_{1} * {Designated}_{i} + β_{2} * {Age}_{i, t} + β_{3} * {Youth}_{i} \\ + β_{4} * {InternationalCap}_{i} + β_{5} * {Transfer}_{i, t} + β_{6} \\ * {MLSExperience}_{i, t} + β_{7} * {BirthPlace}_{i} + β_{8} \\ * {Position}_{i} + β_{9} * {GamesStarted}_{i, t - 1} + β_{10} \\ * {GamesSubstituted}_{i, t - 1} + β_{11} * {Club}_{i, t} + β_{12} \\ * {Year}_{t} + ɛ_{i, t} \end{matrix}

To use position-specific performance statistics, goalkeepers are removed from the data set in Model II. As a solution to the aforementioned multicollinearity problem, instead of raw numbers, averages per game are used. The number of goals per game in the previous season and the number of assists per game in the previous season are included in Model II. Different from Model I, defender is the reference group for the position dummy in Model II below

\begin{matrix} Ln (RealBaseSalary)_{i, t} \\ = β_{0} + β_{1} * {Designated}_{i} + β_{2} * {Age}_{i, t} \\ + β_{3} * {Youth}_{i} + β_{4} * {InternationalCap}_{i} \\ + β_{5} * {Transfer}_{i, t} + β_{6} * {MLSExperience}_{i, t} \\ + β_{7} * {BirthPlace}_{i} + β_{8} * {Position}_{i} + β_{9} \\ * {GamesStarted}_{i, t - 1} + β_{10} * {GamesSubstituted}_{i, t - 1} \\ + β_{11} * {GoalsperGame}_{i, t - 1} + β_{12} \\ * {AssistsperGame}_{i, t - 1} + β_{13} * {Club}_{i, t} + β_{14} \\ * {Year}_{t} + ɛ_{i, t} \end{matrix}

Finally, salary differences among goalkeepers are investigated in Model III. Instead of goal and assist statistics in Model II, shutout per game in the previous season and goals allowed per game in the previous season are introduced in Model III as performance statistics for goalkeepers

\begin{matrix} Ln (RealBaseSalary)_{i, t} \\ = β_{0} + β_{1} * {Designated}_{i} + β_{2} * {Age}_{i, t} + β_{3} \\ * {Youth}_{i} + β_{4} * {InternationalCap}_{i} \\ + β_{5} * {Transfer}_{i, t} + β_{6} * {MLSExperience}_{i, t} + β_{7} \\ * {BirthPlace}_{i} + β_{8} * {Position}_{i} + β_{9} \\ * {GamesStarted}_{i, t - 1} + β_{10} * {GamesSubstituted}_{i, t - 1} \\ + β_{11} * {ShutoutperGame}_{i, t - 1} + β_{12} \\ * GoalsAllowed {perGame}_{i, t - 1} + β_{13} * {Club}_{i, t} \\ + β_{14} * {Year}_{t} + ɛ_{i, t} \end{matrix}

Empirical results

Estimation results from panel GLS regression for all three models are shown in Table 4. For all three estimations, club and year dummies are included to control for fixed club-specific characteristics and year effects. Note that since they are log-linear estimations, coefficient estimates from these regressions should be exponentiated for accurate interpretation of the effects of independent variables on the real base salary of MLS players. Coefficients are transformed using the formulation (e^βi− 1), and the values are given in Table 5.

Table 4.

Results from GLS estimation.

Dependent variable: Ln (Real Base Salary)	Model I	Model II (no GK)	Model III (goalkeeper)
Intercept	9.454*** (0.126)	9.571*** (0.129)	10.160*** (0.236)
Designated	1.176*** (0.046)	1.103*** (0.048)	–
Age	0.055*** (0.004)	0.054*** (0.005)	0.029*** (0.009)
Youth	0.160*** (0.038)	0.166*** (0.041)	0.220*** (0.086)
International cap	0.415*** (0.037)	0.412*** (0.040)	0.284*** (0.068)
Transfer	−0.049*** (0.018)	−0.033* (0.019)	−0.120** (0.048)
MLS experience	0.010* (0.006)	0.012* (0.006)	0.020 (0.012)
Birth place
Western Europe	0.503*** (0.087)	0.498*** (0.092)	0.356** (0.172)
Northern Europe	0.369*** (0.066)	0.360*** (0.069)	0.412** (0.180)
Southern Europe	0.637*** (0.105)	0.742*** (0.112)	−0.334* (0.183)
Eastern Europe	0.103 (0.186)	0.118 (0.187)	–
Central America	0.088 (0.067)	0.087 (0.069)	0.145 (0.253)
South America	0.220*** (0.051)	0.212*** (0.052)	0.516** (0.227)
The Caribbean	−0.222*** (0.07)	−0.235*** (0.072)	0.262 (0.189)
Africa	0.033 (0.059)	0.029 (0.060)	0.260 (0.241)
Asia	−0.128 (0.151)	−0.130 (0.162)	−0.016 (0.254)
Oceania	0.050 (0.175)	0.094 (0.207)	0.011 (0.205)
Position Dummies
Defender	0.088* (0.051)	–	–
Midfielder	0.199*** (0.052)	0.070* (0.036)	–
Forward	0.314*** (0.057)	0.127*** (0.046)	–
Lag (games started)	0.015*** (0.001)	0.013*** (0.001)	0.152*** (0.002)
Lag (games substituted)	−0.014*** (0.002)	−0.012*** (0.002)	−0.035 (0.046)
Lag (goals per game)	–	0.527*** (0.087)	–
Lag (assists per game)	–	0.288*** (0.096)	–
Lag (shutout per game)	–	–	0.003*** (0.001)
Lag (goals allowed per game)	–	–	−0.047 (0.039)
Club dummies	Yes	Yes	Yes
Year dummies	Yes	Yes	Yes
Total observations^a	3117	2754	362

Goal and assist statistics for one player are not available; hence, the sum of the total observations for Model II and Model III is one less than that of Model I.

Significant at 10%.

Significant at 5%.

***

Significant at 1%.

Table 5.

Exponentiated coefficients from GLS estimation (e^βi− 1).

Dependent variable: Ln (Real Base Salary)	Model I	Model II (no GK)	Model III (goalkeeper)
Intercept	12,758.099***	14,341.752***	25,847.297***
Designated	2.241***	2.013***	–
Age	0.057***	0.055***	0.029***
Youth	0.174***	0.181***	0.246***
International cap	0.514***	0.510***	0.328***
Transfer	−0.048***	−0.032*	−0.113**
MLS experience	0.010*	0.012*	0.020
Birth place
Western Europe	0.654***	0.645***	0.428**
Northern Europe	0.446***	0.433***	0.510**
Southern Europe	0.891***	1.100***	−0.284*
Eastern Europe	0.108	0.125	–
Central America	0.092	0.091	0.156
South America	0.246***	0.236***	0.675**
The Caribbean	−0.199***	−0.209***	0.300
Africa	0.034	0.029	0.297
Asia	−0.120	−0.122	−0.016
Oceania	0.051	0.099	0.011
Position dummies
Defender	0.092*	–	–
Midfielder	0.220***	0.073*	–
Forward	0.369***	0.135***	–
Lag (games started)	0.015***	0.013***	0.164***
Lag (games substituted)	−0.014***	−0.012***	−0.034
Lag (goals per game)	–	0.694***	–
Lag (assists per game)	–	0.334***	–
Lag (shutout per game)	–	–	0.003***
Lag (goals allowed per game)	–	–	−0.046

Significant at 10%.

Significant at 5%.

***

Significant at 1%.

In the regression results for the first model, in which the whole data set is used without position-specific performance statistics, designated players’ salaries are more than double those of players under the salary cap. Regression results also reveal that older players receive significantly higher payments in MLS. Each year of age contributes 5.7% more to players’ salaries. In addition, each year of experience in MLS increases players’ salaries by 1%.

Having an international team duty is also an important factor in players’ salaries. Obviously, senior-level national team players are more valuable than prospective young players on youth-level national teams. Senior team players’ real base salaries are 51.4% higher, whereas youth team players’ real base salaries are 17.4% higher than those of a player without any national team duty.

An interesting finding of this study is the negative and significant coefficient for transfer dummy. For players who move from an MLS club to another MLS club, a 4.8% decline in real annual base salary is observed. The transferred player signs a new contract with a new club for lower pay. This is because MLS clubs do not release or sell well-performing players to another MLS club; only poor-performing players are given away. Hence, clubs offer less to these players.

When the birth place of the player is considered, the real base salary of players born in Africa, Asia, Oceania, Central America, or Eastern Europe is not significantly different from that of their North American counterparts. South American, Western European, Southern European, and Northern European players are paid significantly higher salaries. Southern Europeans’ base salaries in real terms are 89.1% higher than those of North America-born players. Western European players have 65.4%, Northern European players have 44.6%, and South American players have 24.6% higher base salary income than do North American players. Another interesting result in this model is the significantly lower real base salary for Caribbean players—almost 20% less than that of North Americans.

When it comes to the effect of players’ positions on the field, defenders, midfielders, and forward players earn significantly more than do goalkeepers. Forward players earn 36.9% more, midfielders earn 22% more, and defenders earn 9.2% more than do goalkeepers in MLS. This difference reflects the importance placed on players who contribute to team success by scoring a goal or making an assist.

When the effect of players’ performance statistics on salary is considered, one more game in the first eleven increases the numbers on the paycheck by 1.5% the next year while each game as a substitute causes a decrease in salary by 1.4% the next year.

Finally, the authors are aware of the expansion draft in MLS. Expansion draft allows the new teams that enter the league to pick players from the existing teams. Expansion draft is thought to have an effect on player salaries and authors have attempted to test for its impact through a binary variable. But, regression results showed that the impact is insignificant so the variable is not included in the model reported here.

In Model II, removing goalkeepers from the data set causes a loss in information; however, it also allows us to use goals per game and assists per game in the previous season as performance statistics. The third column of Table 5 shows that despite small changes in the values of coefficient estimates, there is no change in the significance of the explanatory variables in the second model. Goals per game in the previous season increases player salary by 69.4%, while assists per game in the previous season increases salary by 33.4%. This result confirms that scoring a goal is more valuable than making an assist. It also explains why forward players are paid higher than any other position. Midfielders receive 7.3% more and forward players receive 13.5% more than do defenders.

However, in Model III, the significance and magnitudes of the coefficients change for goalkeepers. Since the lag of performance statistics requires two years of existence in the league and there is no designated goalkeeper and Eastern European goalkeeper who has played for two consecutive years in MLS, there is no observation for these variables in the third model. The contribution of age to player remuneration decreases to 2.9% for goalkeepers. However, the effect of youth-level national team experience increases to 24.6%, while the effect of being a senior-level international cap decreases to 32.8% compared to Model II.

Moving from one MLS team to another affects goalkeepers’ salaries much more than those of defenders, midfielders, and forward players. A new contract causes a decrease in goalkeeper salary by 11.3%. Different from Model II, MLS experience is insignificant for goalkeepers, and South European goalkeepers earn 28.4% less than their North American counterparts do. In addition, South American goalkeepers are the most valuable in MLS. They receive 67.5% more than do North American goalkeepers.

Substituting a goalkeeper in a game is very rare in soccer. For this reason, if a player is considered a substitute, he is the second goalkeeper of the team. This fact is reflected in the coefficients of the performance statistics as well. Being a substitute goalkeeper has no effect on salary, but being the team’s first goalkeeper has much more of an effect on salary. In Model II, the number of games started in the first eleven in the previous season affects the salaries by only 1.3%, but for goalkeepers the effect is 16.4%.

Another striking result for goalkeepers is that although the effect on salary is only 0.3%, the number of shutouts in the previous season has a significant and positive effect. However, although conceding goals has a negative effect, it is not significant. This can be rephrased as indicating that good goalkeepers are rewarded in monetary terms but poor goalkeepers are not penalized. An explanation for this can be that not only the goalkeeper but also the poor defense may be the reason for goals allowed.

Conclusion

Determinants of players’ salaries in MLS are investigated in this study. As a starting point, the average of Gini coefficients for player salaries in MLS is found to be 0.6 for the whole data set from 2007 to 2016, which indicates high levels of income inequality among players in MLS. Moreover, the data set tells us that between variation is higher than within variation. That is, differences in the salaries among players are much higher than the differences over time.

This study stands aside from the existing literature on salary determination in several ways. First, instead of performance indicators in the current season, statistics in the previous season are used because player contracts depend on the performance of the player during the previous season; the current-season performance is not observed at the time of the signing of the contract. Second, to determine the difference between a player who played 90 min and a player who played just 1 min, instead of the number of games played, the number of games started in the first eleven and the number of games substituted in the previous season are used as explanatory variables. Finally, a transfer dummy helps us understand the effect of signing a new contract on salaries in MLS.

Birth place is the most influential determinant of players’ salaries. This is especially true for defenders, midfielders, and forward players; players born in Southern Europe double their salaries. South American, Northern European, and Southern European players also receive significantly higher salaries than do US-born soccer players, while Caribbean players earn significantly less.

Results from this study confirm the differences in salaries according to playing positions, namely that midfield and forward players have higher salaries than do goalkeepers and defenders. Being a youth-level or senior-level international team member also adds dollars to the paycheck of a player.

Another interesting finding involves the MLS clubs’ approach to player transfers. Players move from one MLS club to another due to poor performance in their current clubs; they sign contracts with their new clubs for lower salaries. MLS clubs are not willing to release or sell players who show high performance. In addition, clubs are not eager to transfer well-performing players from other MLS clubs and offer them higher-paying contracts.

Among performance-related factors, the number of goals and the number of assists are among the most influential factors. On the other hand, while the number of games started in the first eleven has a positive effect, the number of games as a substitute has a negative effect on players’ salaries.

When goalkeepers are considered, the most influential factor is being born in South America. The final words about goalkeepers would be that good goalkeepers are rewarded but poor ones are not penalized.

Footnotes

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

Notes

References

Leeds

von Allmen

. The economics of sports, 5th ed. New Jersey: Pearson Education Inc., 2014, pp. 39–41.

Bougheas

Downward

. The economics of professional sports leagues: a bargaining approach Discussion Papers in Economics DP 00/6, University of Nottingham, UK, 2000.

Kahn

. The sports business as a labor market laboratory. J Econ Perspect 2000; 14: 75–94.

Buchanan

Slottje

. Pay and performance in the NBA, Greenwich, CT: JAI Press, 1996.

Quirk

Fort

. Hardball: the abuse of power in pro team sports, New Jersey: Princeton University Press, 1999.

Sanderson

Siegfried

. The implications of athlete freedom to contract: lessons from North America. Econ Aff 1997; 17: 7–12.

Scully

. The Market Structure of Sports, Chicago: University of Chicago Press, 1995.

Szymanski

. A market test for discrimination in the English professional soccer leagues. J Polit Econ 2000; 108: 590–603.

Feess

Frick

Muehlheusser

. Legal restrictions on outside trade clauses: theory and evidence from German soccer Discussion Paper No. 1140, Institut Zukunft der Arbeit, Bonn, 2004.

10.

Frick

Salary determination and the pay-performance relationship in professional soccer: evidence from Germany. In: Rodríguez

Késenne

García

(eds). Sports economics after fifty years: essays in honour of Simon Rottenberg, Oviedo: Universidad de Oviedo, 2006, pp. 125–146.

11.

Lehmann

Schulze

. What does it take to be a star? The role of performance and the media for German soccer players. Appl Econ Q 2008; 54: 59–70.

12.

Battre

Deutscher

Frick

. Salary determination in the German Bundesliga: a panel study IASE Conference Paper No. 0811, International Association of Sports Economists, 2008.

13.

Garcia-del-Barrio

Pujol

. Pay and performance in Spanish soccer league: who gets the expected monopsony rents? Faculty Working Papers 05/04, School of Economics and Business Administration, University of Navarra, 2004.

14.

Garcia-del-Barrio

Pujol

. Hidden monopsony rents in winner-take-all markets-sport and economic contribution of Spanish soccer players. Manage Decis Econ 2007; 28: 57–70.

15.

Lucifora

Simmons

. Superstar effects in sport: evidence from Italian soccer. J Sports Econ 2003; 4: 35–55.

16.

Frick

Lee

. Temporal variations in technical efficiency: evidence from German soccer. J Productiv Anal 2011; 35: 15–24.

17.

Franck

Nuesch

. The effect of wage dispersion on team outcome and the way the outcome is produced. Appl Econ 2011; 43: 3037–3049.

18.

Coates

Frick

Jewell

. Superstar salaries and soccer success: the impact of designated players in Major League Soccer. J Sports Econ 2016; 17: 716–735.

19.

Lee

Harris

. Managing excellence in USA Major League Soccer: an analysis of the relationship between player performance and salary. Manag Leisure 2012; 17: 106–123.

20.

Zimbalist

. Competitive balance in sports leagues: an introduction. J Sports Econ 2002; 3: 111–121.

21.

Kuethe

Motamed

. Returns to stardom: evidence from US Major League Soccer. J Sports Econ 2010; 11: 567–579.

22.

Reilly

Witt

. The determinants of base pay and the role of race in Major League Soccer: evidence from the 2007 League Season School of Economics Discussion Papers 1907, School of Economics, University of Surrey, 2007.