Performance, salaries, and contract length: empirical evidence from German soccer.

Frick, Bernd

The recent transfers of Christiano Ronaldo from Manchester United, and of Ricardo Izecson dos Santos Leite (Kaka) from Associazione Calcio Milan to Real Madrid--as well as the increasing financial problems of many of the top teams in the big five European leagues--have again increased the public's attention for the global football players' labor market. Therefore, the paper addresses two important, and highly contested, issues: player remuneration and contract duration (players are usually considered as overpaid and poorly motivated. Using two different unbalanced panels from the German Bundesliga league that cover six and 13 consecutive seasons respectively (1997-98 to 2002-03 and 1995-96 to 2007-08), I show, first, that the variance in player salaries can be explained, to a large extent, by the variance in individual performance. That is, salaries can be explained by career games played and games played last season, previous and recent international appearances, and goals scored. Moreover, player position, leadership skills, and region of birth clearly matter as well. The impact of these characteristics varies across the salary distribution. Second, I find robust evidence that player performance--measured primarily, but not exclusively, by a subjective overall player rating from Kicker, a highly respected soccer magazine-significantly increases in the last year of the contract. In addition, the variance in player performance is significantly lower in the last year of the contract. These findings suggest that moral hazard is a widespread phenomenon, even in professional soccer.

Keywords: salaries, contract duration, pay determination, moral hazard, soccer Introduction

The development of player salaries in professional football in Germany The escalating and/or skyrocketing salaries of professional football players have only recently become a highly controversial issue in Germany. Perhaps surprisingly, this has not always been the case: When, in the summer of 1954, the members of the German national team returned home after their glorious victory in the World Cup final against Hungary, each player received a gratification of 2,000 DM--about six months' pay of a male full-time employee (Muller-Jentsch, 1989). By that time, the enormous amount was considered by most people to be a well-deserved recognition for an outstanding performance.

The public opinion, however, changed gradually. On July 28, 1962, when the representatives of the 21 different regional football associations in Germany agreed to introduce a single first division, they also introduced a minimum and a maximum salary--the former being 250 DM per month and the latter 1,200 DM per month. (1) Moreover, the maximum transfer fee was set at 50,000 DM, of which a maximum of 5,000 DM could be paid to the player; all of these caps were finally abandoned in 1972. The salaries of the top players soon started to rise: In 1966, Uwe Seeler--at the time, he was captain of the national team--earned 50,000 DM, while midfielder Gunther Netzer was paid 100,000 DM already in 1972. Five years later, top-scorer Gerd Muller earned 500,000 DM per season. In 1987, Rudi Voller was paid 1.1 Mio. DM, and in 1992, Andreas Moller made 1.7 Mio. DM. Upon his return from the Italian Serie A to the Bundesliga in 1995, Lothar Matthaus was paid 2.5 Mio. DM: An amount that he more than tripled until 1998. (2) In 2001, Stefan Effenberg, as well as Oliver Kahn, were paid 9.5 Mio. DM (Sonnenberg, 2002).

This development, which can mainly be attributed to the development of the television revenue generated by the clubs, has, for most of the time, been accompanied by public discussion about the adequacy of player salaries. In addition, it has recently even attracted the attention of a number of politicians. Since the mid-1960s, the increasing liabilities of some first division clubs were considered as early signals of the forthcoming "collapse" of professional football due to "excessive" player and head coach salaries (Die Zeit, May 17, 1968; Der Spiegel, January 22, 1968). This discussion went on for decades and culminated shortly before Christmas 2007 when Norbert Lammert, the president of the German Bundestag--the nation's parliament--released the following statement:

"I am particularly annoyed by the salary explosion that we have recently experienced in professional sport in general and in soccer in particular ... This is something I cannot understand at all" (Onabrucker Zeitung, December 23, 2007).

This caused the president of the German Football Association, Theo Zwanziger, to respond with the following statement:

"From a 'moral' point of view, the salaries of many professional soccer players are too high-as are the incomes of most actors and some top managers" (Suddeutsche Zeitung, January 9, 2008).

Given the steadily increasing ticket sales and merchandising revenues, it is hardly surprising that football fans seem to be quite relaxed with regard to the level and the development of player salaries. In an online opinion poll that was started shortly after the interviews were published, the daily newspaper Die Welt asked its readers whether politicians should be concerned about the development of player salaries in professional football. The results are shown in Table 1.

The fans' position is nicely summarized in the following quote by sports journalist Oskar Beck:

"We football fans are a rather strange species. We complain when our heroes earn enormous amounts of money, but at the same time we readily accept higher ticket prices if this enables our favorite club to sign yet another top-scorer. Moreover, we are prepared to pay 19.90 [euro] for the memoirs of Stefan Effenberg and the diaries of Lothar Matthaus as if it were the most recent works of Nobel laureates Heinrich Boll and Gunter Grass" (Die Welt, December 30, 2007).

In summary, it appears that fans have fewer problems with the escalating and skyrocketing salaries than politicians and journalists seem to expect. From an economic point of view, however, the question is not whether the salaries are adequate or excessive, but whether the observable variation in player remuneration can be explained by differences in individual performance and the clubs' ability to pay, which, in turn, is a function of past and recent sporting success, market size, and tradition. These and related questions will be answered in Section 2 of the following paper.

The development of contract duration in professional football in Germany

The issue of contract duration is as contested as the remuneration of players. Dragoslav Stepanovic, former head coach of Eintracht Frankfurt, argued the following in an interview in the summer of 1992 when his team finished third in the Bundesliga--a position the club has not since accomplished again:

"In principle, player contracts should not exceed three months. In case of excellent performance such contracts can always be extended for another three months."

Norbert Pflippen--a well-known player-agent who represented former star player Christian Ziege by the time the young man was 21 years old and had just signed a five-year contract with Bayern Munich--reasoned similarly:

"An ambitious young player should never sign a long-term contract. He must always be convinced that within one or two years he will again be underpaid. Having the opportunity to renegotiate is crucial."

Thus, not only sports fans, but head coaches and player agents also seem to believe that players can strategically vary their performance--an impression that is consistent with principal agent theory. One of the major insights of this theory is that properly designed incentive contracts will align the interests of a rational and opportunistic agent with those of a principal not completely informed about the talent and the abilities of the agent. (3) Explicit incentives, such as performance-related pay, are not the only motivating factor. Workers with fixed-term contracts, for example, have incentives to vary effort at different points of their contract cycle--that is, to increase effort just before a new contract is signed and to reduce it after a lucrative multiyear contract is secured. The duration-related incentives create a considerable moral hazard problem, a topic which has been empirically examined only on occasion. (4)

Although the issue that will be dealt with in the second part of my paper (see Section 3 below) is of critical importance for the managers of professional soccer teams as well as the managers of "normal" firms, most of the available studies rely on data from the sports industry. This is not surprising because individual performance can be measured easily and the data is readily available to the researcher:

"Professional sport offers a unique opportunity for labor market research. There is no other research setting 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, professional sports leagues have experienced major changes in labor market rules and structure ... creating interesting natural experiments that offer opportunities for analysis" (Kahn, 2000, p. 75).

Thus, Section 3 of my paper will empirically analyze the behavioral consequences of short- versus long-term contracts. The main questions to be addressed are as follows: Is "shirking"--as alleged by fans and sports journalists--really an issue in professional football? Second, does this kind of undesired behavior have an impact on the sporting performance of the clubs? Most of the available studies answer these questions by comparing a player's performance in the first season after he has signed a long-term contract with that same player's performance in the last year of his old contract. Given the obvious problems of this approach, I take a different route: I look at changes in players' performance as they approach renegotiation. That is, I compare their performance in the last year of a particular contract to the performance in the season(s) before that contract expires. The assumption here is that, if performance improves in the last year of the old contract, it is most likely the result of a deliberate change in behavior. If, on the other hand, performance deteriorates in the first year of the new, long-term contract, this could be due to a number of different factors, such as stochastic variations in performance and/or random shocks that are beyond the player's control.

The remuneration of professional football players

Theory. In the absence of labor market restrictions--such as salary caps, reserve clauses, and/or draft rules--players will be paid according to their marginal product; that is, the wage an individual player receives is a function of his contribution to the team's revenues which is, in turn, influenced by his talent and experience on the one hand and his "fan appeal" on the other hand (Rottenberg, 1956). However, since the clubs differ with respect to their drawing potential--there are small market and large market teams--they also differ with respect to their ability to pay. For example, the marginal product of Arjen Robben or Franck Ribery is certainly higher in Munich than it is in Freiburg or in Mainz. However, since it rests on a number of critical assumptions--such as player mobility, complete information, and risk neutrality--the neoclassical model of wage determination has often been rejected, not only by sports fans, but by some highly respected economists as well: "... the elementary classical model presents a very poor description of employment relations in advanced economies" (Milgrom & Roberts, 1992, p. 329).

However, the problems that are characteristic for most-if not all-"real life" labor contracts (e.g., information asymmetries, incompleteness, or importance of implicit elements) are clearly less important in professional team sports. Here, an individual player's performance can easily be measured, "shirking" can be detected at low cost, effort and talent can be evaluated, not only by a player's current club but by other teams as well. It is, therefore, plausible to assume that in the German Bundesliga-as in other professional team sports leagues with an unregulated labor market-players are paid mainly according to their past and recent performance and their ability to attract fans (5). Thus, the term marginal product is used here to describe the value of an individual player's contribution to the spectacle (i.e., the matches in which he appears). (6)

Moreover, professions in which talent is highly valued by consumers are usually characterized by a highly skewed distribution of earnings: Small differences in talent translate into large differences in pay (Rosen, 1981). Player reputation attracts additional spectators; however, advances in technology also facilitate the reproduction of matches at low cost. Together, these two effects lead to a considerable expansion of the market. In general, players are neither completely homogenous nor completely specialized. This, in turn, creates a situation of bilateral monopoly in which players and teams share a surplus or economic rent. Only a few players who are sufficiently differentiated can shift surpluses (i.e., rents) completely into salaries; these players will tend to be the superstars of their sports.

Previous Evidence. To the best of my knowledge, only four studies have been published in English so far that seek to identify the determinants of player salaries in professional football. Lucifora and Simmons (2003) use information on 533 outfield players from the Italian Serie A and Serie B at the beginning of the 1995-1996 season (i.e., a cross section). They find that individual performance--measured primarily by the number of games played and goals scored--has a statistically significant and economically relevant influence on salaries. Moreover, earnings are highly convex in the individual's career goal-scoring rate and the assist rate, suggesting the existence of a considerable superstar effect. Lehmann and Schulze (2008) use 651 player-year-observations from the German Bundesliga in the seasons 1998-1999 and 1999-2000. Their performance measures also have the expected, and statistically significant, influence on salaries. Surprisingly, however, media presence has a positive, but declining, influence. This suggests decreasing returns to popularity--a finding that is difficult to reconcile with the concept of superstardom (7). Feess, Frick, and Muhlheusser (2004) use a sample of players appearing in the German Bundesliga in the period 1994-1995 to 1999-2000 (n = 604 observations). They found that, above and beyond the traditional performance measures--such as games played, goals scored, and international appearances--contract length also has a positive and statistically significant impact on a player's annual wage. They also found that this effect has become much stronger in the "Post Bosman Era"--that is, after the transfer of property rights from the clubs to the players, a respective decision induced by the European Court of Justice in December 1995. The finding that contract length and annual salary are complements, rather than substitutes, again suggests that superstar effects are of particular importance in the pay determination process. Finally, Kuethe and Motamed (2010) use data on 193 athletes who played in the MLS in 2007 and were under league contracts at the start of the 2008 season in order to identify the impact of the designated player rule and all-star game participation on individual salaries. Controlling for player age, experience, goals, assists, and region of origin, both variables are found to have a statistically significant impact on remuneration. This suggests that superstardom is an important determinant of player salaries.

In summary, these papers show that salaries of professional athletes are more than just a random occurrence, that systematic factors determine these salaries to a large extent, and that these systematic factors--such as age, experience, and performance--are very similar to those found in other occupations. Where sports teams differ in structure of earnings is in the distribution of salaries; that is, they are even more highly skewed than in standard occupations. In addition, sports teams apply more stringent selection procedures into occupations. For example, poor performance by a player results in being dropped from team squad and very quickly being discarded; there are high levels of mobility within the industry (e.g., between teams) and into and out of the industry, with shorter careers than in most occupations. (8)

Testable Hypotheses. The observable variance in player salaries is primarily due to the variance in talent and performance:

1. Player salaries will increase with performance (e.g., league appearances or goals), experience (i.e., age), and popularity (e.g., appearances in the national team). (9)

2. The most recent performance (i.e., in the last season) will have a greater impact on player salaries than previous career performance.

Moreover, the club's differing ability to pay--which, in turn, is a function of the size of the respective market, the club's history, and its sporting performance--will also affect player salaries significantly.

The structure and development of player salaries in the German Bundesliga

Available Data. My primary source of information is Kicker, a highly respected soccer magazine that offers market valuations of players assessed at the beginning of a season for 13 consecutive years (1995-1996 to 2007-2008) as a proxy for undisclosed salary; the salary remains private and confidential not only in Germany but in the rest of in Europe too. I am confident of the reliability of these proxies for several reasons. First, the correlation between Kicker salary figures and those from another reliable source (see http://www.transfermarkt.de) is high at .75 (Torgler, Schmidt, & Frey, 2006). Second, the player valuations in Kicker magazine have been compiled by a stable team of experts who have established consistent practice over a long period. I, therefore, interpret the players' market values as published by Kicker as particularly reliable. Aggregating the individual market values across teams, and dividing these by a constant factor of 1.5, results in the aggregated wage bills of the 18 teams in the Bundesliga as published in the annual reports of the German Football Association for the period between 1996-2007. Furthermore, the correlation between Kicker player valuations and a subset of actual salary data obtained from the Bundesliga has been found to be high at .80 (Frick, 2003).

The size of my sample is quite large with 6,147 player-year-observations for 1,993 different players who were assigned player characteristics, including number of career games played, number of games played last season, number of career goals scored, number of goals scored last season, number of career international appearances, number of international appearances last season, team captain (dummy), position (a set of three dummies), region of birth (six dummies), and previous league. In addition, they

[FIGURE 1 OMITTED]

[FIGURE 2 OMITTED]

Descriptive Evidence

It appears from Figure 1 that average player salaries have increased from 550,000 [euro] in 1995-1996 to about 1.3 Mio. [euro] in the 2007-2008 season. Interestingly, the standard deviation constantly oscillates around the mean, suggesting that the dispersion of player salaries has remained more or less constant over time. (10) The decline in player salaries in the 2003-2004 and 2004-2005 seasons has to be attributed to the insolvency of the Kirch group--the company that had bought the television rights for a record amount of 695 Mio. DM per year, starting with the 2000-2001 season. Moreover, player salaries differ considerably by position: In the 2007-2008 season, goalkeepers earned on average about 900,000 [euro], while forwards were paid an average of 1.45 Mio. [euro] (see Figure 2). The salaries of defenders and midfielders are higher than those of goalkeepers but lower than those of forwards.

Although statistically significant (2007-2008: F = 3.08, p < .05), these averages hide considerable variation within the different groups of players. Particularly in the case of goalkeepers, the standard deviation of individual salaries--and, therefore, the corresponding coefficient of variation--is rather high (see Table A1). Perhaps also surprising is the fact that the wage premium of forwards seems to decline over the years. Whether this is due to changes in the supply of forwards (relative to other positions) or to changes in the quality of all players under contract (relative to other positions), remains to be seen.

Econometric Findings

I start with the estimation of an ordinary least squares (OLS) model with robust standard errors, a random-effects (RE) model, and a median regression (MR) model. (11) I then present the findings of various quantile regressions (i.e., .10, .25, .75, and .90) with bootstrapped standard errors (200 repetitions). The results are comparable to those obtained from OLS as well as RE- and MR-estimation. However, few of the coefficients remain constant over the percentiles.

The model to be estimated is of the following general form:

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII]

where AGE: Player age

GPL: Number of appearances in Bundesliga in last season

CGP: Number of career appearances in Bundesliga

IAL: International appearances last season

IAP: International appearances in career

GSL: Goals scored last season in Bundesliga

CGS: Career goals scored in Bundesliga

CAP: Captain of team (0 = no; 1 = yes)

FDA: Previous team in first division abroad (0 = no; 1 = yes)

PD: Vector of position dummies (ref.: goalkeeper)

RD: Vector of region of birth dummies (ref.: Germany)

TD: Vector of team dummies (ref.: Borussia Moenchengladbach)

YD: Vector of year dummies (ref.: 2001/2002)

Thus, my models distinguish between a player's career performance and his most recent (i.e., last season) performance. The most recent performance--measured by inter alia, the number of games played, the number of international appearances, and the number of goals scored--is, of course not included in the career performance (the results of OLS, RE, and MR estimations are displayed in Table 2 below). (12)

Most studies of pay determination in football rely on the standard conditional expectations model. However, the focus on the conditional mean is likely to misrepresent the relationship between pay and performance, if there are differences in the returns to performance along the conditional distribution. Several studies of salary determination in other professional (e.g., North American) team sports use quantile regression estimation since log salary measures tend to have even greater kurtosis values than in standard occupations (Berri & Simmons, 2009; Hamilton, 1997; Leeds & Kowalewski, 2001; Reilly & Witt, 2007; Simmons & Berri, 2009; Vincent & Eastman, 2009). OLS salary regressions are sensitive to the presence of outliers and can be inefficient if the log salary measure has a highly non-normal distribution, as is often the case in professional team sports. In contrast, quantile regression estimates are more robust. Presence of non-normality is indicated by a large kurtosis value and D'Agostino and colleagues' (1990) test is performed by the sktest command in Stata 10.1. In my panel, the p-value for the test statistic of the null hypothesis is .000, kurtosis does not depart from the value associated with a normal distribution; hence, my log salary data depart from normality, a result that is similar to those found in some studies of North American sports (e.g., Berri & Simmons, 2009 study on the National Football League).

One further advantage of quantile regression is that it facilitates examination of salary returns to characteristics at different points in the salary distribution (Buchinsky, 1998; Koenker & Bassett, 1978). That is, I can investigate the impacts of the available performance measures at any quantile of the salary distribution, not just the conditional mean. Moreover, the quantile regression approach is semi-parametric in that it avoids assumptions about the parametric distribution of the regression error term, an especially suitable feature where the data are heteroskedastic as in my case.

To ensure robustness of standard errors, I bootstrap with 200 replications. I report quantile regression estimates in Table 3. My main findings can be summarized as follows (see Tables 2 and 3):

* First, age, career games played, international appearances over the entire career, and international appearances in the last season all have a statistically significant non-linear influence on salaries. The statistically significant coefficient of the cubic term suggests existence of superstar effects (Lucifora & Simmons, 2003).

* A strange result is obtained for career goals scored: The coefficient of the linear and the cubic term are significant and negative, while the coefficient of the squared term is positive and significant. (13)

* Second, goals scored last season, as well as games played last season, have a significantly positive and strictly linear influence on annual income; that is, there seem to be no decreasing returns to either goals scored or games played.

* Comparing the returns to career performance and to performance in the last season, it appears that historical merits do not count very much; that is, recent performance is--as expected--far more important than past performance.

* Third, defenders, midfielders, and forwards earn significantly higher salaries than goalkeepers. The premiums for these positions, however, differ considerably across estimations: The effect is most pronounced in the RE estimation and weakest in the MR model.

* Fourth, region of birth is also important: Players from South America and Western Europe receive a considerable pay premium while players from the rest of the world are neither favored nor discriminated against. (14) The pay premium for South Americans and West Europeans is not surprising: Other things equal, players from these regions attract larger crowds (Wilson & Ying, 2003) and contribute more to merchandising revenues (Kalter, 1999).

* The longer a player has been active in his current club, the lower the c.p. of his annual salary. Whether this is the result of an adverse selection process (e.g., better players are traded while less talented players remain with their old club), or whether some players are willing to forfeit money to stay at home, is not yet clear. (15)

* Finally, team captains and players who moved abroad from a first division club to Germany are paid a significant premium too. As for team captains, this is obviously due to leadership skills that are required for the job and that are, therefore, particularly rewarded in the market (Kuhn & Weinberger, 2005).

Few of the coefficients retain their magnitude across the different quantiles of the salary distribution:16

* Generally, the maximum income is reached at an age of about 27 or 28 years. The age-earnings profile, however, is much flatter for the players with the highest incomes.

* The impact of games played last season, as well as career games played on annual salaries, is much stronger for players at the bottom of the income distribution.

* International appearances--past as well as current--seem to have a much stronger influence on the salaries of the players at the top of the income distribution.

* Goals scored--past as well as current--tenure with the current club, and being a team captain seem to have a more or less constant impact on player salaries (i.e., the coefficients are quite similar for the different quantiles).

* The coefficients of the position dummies change considerably across the income distribution, indicating that goalkeepers are the real superstars in the business. (17)

* The pay premium enjoyed by players from South America increases across the pay distribution, while the premium for players from Western Europe decreases. (18,19)

Contract Duration and Player Performance

What can we learn from the available literature?

The common perception among sports fans is that players become lazy and expend less effort once they have signed a long-term contract. The available evidence (summarized in Table 4 below) is less clear: While some of the studies find robust evidence supporting the shirking hypothesis, others do not find any sign of such behavior. However, even if no shirking can be detected, opportunistic behavior may well be an issue. First, reputation considerations may keep players from reducing their effort levels. In this case, only a player who knows that he has recently signed his last contract will have an incentive to withhold effort. Anticipating such behavior, managers will refrain from giving long-term contracts to older workers. Second, many player contracts will include incentive clauses tying individual and/or team performance to compensation. This, in turn, is likely to result in a higher wage bill because risk-averse players may expect a premium in exchange for their readiness to accept contingent pay. (20)

Summarizing, agency theory identifies two different options for teams to control moral hazard. First, monitoring can reduce information asymmetries, and second, incentive contracts may be used to mitigate the underlying motivation deficit. Since monitoring is often rather costly and difficult to implement--especially with regard to the player's behavior outside the game--teams tend to at least partly reward their players depending on the output produced, assuming that measuring inputs is more or less impossible. However, outcomes are not fully under the control of the agent and, at the same time, risk aversion on behalf of the player limits the team's ability to use output-related pay only. Thus, an efficient contract balances the costs of risk bearing against the benefits of improved incentives. (21)

Data, estimation and empirical findings

The sample used in this study includes all regular players (22) who were under contract with any of the teams in the first German soccer division (n = 760) obtained from various annual editions of Kicker, the leading soccer magazine in the country. Altogether, the sample includes 1,866 player-year-observations from the 1998-1999 through the 2002-2003 seasons. The subjective performance measures used here are school grades ranging from 1 (exceptional) to 6 (very poor) and summarize a player's effort and contribution to his team's performance. The number of graded appearances varies between 7,113 and 7,239 per season. This means that, on average, between 23 and 24 graded players appear during each match--a grade is awarded only if the player spends at least 30 minutes on the pitch. Due to the definition chosen in this paper, approximately 72% of the players are considered regular--recall that these are players with nine and more appearances per season--and 28% are considered back-up players with less than nine appearances. The number of regular players varies less than the number of back-up players (n = 301-321; n = 110-154). This is mainly due to the fact that in 2000-2001, two of the three relegated teams increased their roster sizes considerably during the season. Since regular players by definition appear significantly more often than the back-up players, more than 93% of the graded appearances (i.e., those of a minimum duration of 30 min) were by regular players. Two hundred and eighty-six (37.6%) of the players appear in my dataset for only one season and disappear again thereafter due to a transfer to a lower division club or a transfer to a club abroad, or because the player's club had been relegated at the end of the season. On the other hand, 106 players (14.0%) managed to survive in the Bundesliga for at least five seasons. (23)

[FIGURE 3 OMITTED]

Turning to the contract variable, it appears that 24% of the observations are in their last contract year (see Table 5 and Figure 3). About 34% have one year remaining on their contracts and 27% have two seasons remaining. Since it is plausible to assume that, in the case of a multiyear contract, a player's incentives to perform well will increase linearly, the estimations presented below use the number of remaining contract years as an exogenous variable; a censored contract variable with any duration of more than two years was recoded as two. The implicit assumption is that, with two years remaining on the contract, players gradually start to deliver better performances in order to reach their optimal bargaining position in the last season before the contract expires.

The estimated models are of the following general form:

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII]

where PP: player performance in season t [(rag = relative average grade; see Figure A2), (apg = average player grade; see Figure A3), (vpp = variance of player performance; see Figure A4)] (24)

CS: contract status in season t [remaining contract duration (estimates 1.1, 2.1, and 3.1); censored remaining contract duration (estimates 1.2, 2.2, and 3.2); last contract year (dummy; 0 = no; 1 = yes; estimates 1.3, 2.3, and 3.3)]

GS: goals scored in season t

RC: number of red cards in season t

YC: number of yellow cards in season t

SP: semiprofessional (dummy; 0 = no; 1 = yes)

CGP: career games played in Bundesliga

CGP2: career games squared

AGE: player age

[AGE.sup.2]: age squared

INT: appearance in national team (dummy; 0 = no; 1 = yes)

TCCB: team change during Christmas break (dummy; 0 = no; 1 = yes)

DEF: defender (dummy; 0 = no; 1 = yes)

MID: midfielder (dummy; 0 = no; 1 = yes)

FOR: forward (dummy; 0 = no; 1 = yes)

ATG: average team grade

In models 1.1-1.3, the dependent variable is the individual player's average grade; in models 2.1-2.3, it is the relative individual average corrected by the average grade of the player's team. Given the grade system used, higher values of the dependent variable denote a weak or even a poor performance. (25) The expected sign of the contract status variable is, therefore, positive in estimates 1.1, 2.1, and 3.1 as well as in 1.2, 2.2, and 3.2, and negative in estimates 1.3, 2.3, and 3.3.

1. The higher the remaining duration of a player's contract, the poorer his performance will be and the higher the variance in his performance.

2. The performance will significantly improve and the variance will be significantly lower in the last year of the contract.

Looking at the control variables (see Tables 6-8), it appears that the number of goals scored, and the number of yellow cards per season, have a significantly positive influence on player performance--recall the worse the performance, the higher the average grade. (26) Player age and experience--measured by the number of career games played--have the expected nonlinear impact on performance while being a member of the national team, the number of red cards per season, and the position dummies are--by and large--statistically insignificant. Perhaps surprisingly, semiprofessionals, and players who have been traded over the Christmas break, perform significantly better than otherwise comparable players without these characteristics--perhaps expectations are lower in these cases and players are, therefore, graded more generously.

With regard to the variance of player performance, the picture is slightly different (see Table 8). First, the number of goals scored and the number of red cards increase the variation, as does membership in the national team. Second, none of the coefficients of the other control variables--apart from one of the position dummies--comes close to statistical significance. Third, the higher the number of appearances in the last season, the lower the variation in a player's performance.

Turning to the coefficients of the contract status variable, it appears that, irrespective of its concrete specification, convincing evidence in favor of the shirking hypothesis can be found. That is, the shorter the remaining duration of a player's contract, the better his performance. Moreover, the consistency in a player's performance increases as he approaches renegotiation. Depending on the specification of the model, a player's performance increases by 2%-3% per year as his contract elapses. This is by no means trivial: Players are often monitored day-by-day, not only by their coaches, but by millions of sports fans as well. Such an increase in performance is certainly surprising as it mirrors a player's possibilities to increase his effort as he expects to benefit from being more devoted to his job.

Equally interesting in the context of the paper is yet another question: Do these contract-related changes in individual performance affect team performance? If the performance measures used in the paper are valuable to the teams, we should observe team outcomes to follow the individual player's performance (i.e., looking to a rise when many players are in the last year of their contracts and to a fall when many have signed new multiyear contracts). The relevant literature has identified several wedges that might exist between individual and team performance. First, if only some valuable tasks are measurable, incentive effects can lead players to misallocate resources toward the measurable tasks and away from other, equally important ones (see Holmstrom & Milgrom, 1991). Second, the readiness to cooperate may suffer under some incentive structures; that is, if players are paid according to the number of goals scored or the number of appearances (see Baker, 1992). Finally, rational individuals might behave opportunistically when individuals who reduce their effort levels cannot be identified (see Holmstrom, 1982). While the latter problem is unlikely to occur in professional team sports, the former two are certainly worth investigation.

To examine whether changes in individual player performance actually affect team performance, I estimate a fixed effects model with the average team grade as the endogenous variable and the number of points at the end of the season as the dependent variable. (27) Taking into account that an individual player's performance improves considerably in the last contract year, the potential improvement in the team's performance can be easily calculated. On average, four players are up for contract negotiations each season. If that figure increases by two, the team will secure slightly more than one additional point (i.e., a draw instead of a loss). If half of the roster--instead of one quarter--is in the last contract year, the team will win two additional points. (28) Given the usually close competition--in some of the seasons under consideration, already one point more would have resulted in either avoiding relegation (e.g., Karlsruhe in 1997-1998 and Nuremberg in 1998-1999) or in qualification for a European cup competition (e.g., Berlin in 1998-1999, Leverkusen in 2000-2001, Munich in 2001-2002, and Dortmund in 2002-2003)--these marginal changes in individual performance can have massive economic consequences for the clubs affected.

Summary and Implications

Using two large longitudinal datasets from German professional football, the paper demonstrates that first players are remunerated by the market according to their innate talent and their performance, with the most recent performance being far more important than the performance delivered years ago. The OLS and the RE models explain more than 60% of the observable variance in player salaries. This is quite high and indicates that the available performance measures--although far from ideal--are indeed well suited for the empirical analysis. The quantile regressions, in turn, demonstrate that restricting the analysis to the standard models is problematic insofar as the focus on the conditional mean is likely to misrepresent the relationship between pay and performance; this is because there are considerable differences in the returns to performance along the conditional distribution. Second, the paper finds clear evidence of increasing player effort over the duration of individual contracts. Other things equal, a player's performance increases by 2%-3% in the last year of the contract, indicating that players can--and indeed do--vary their effort levels strategically.

The analyses can, and will be, extended in different directions. (29) First, the wage equations will be estimated separately for goalkeepers, defenders, midfielders, and forwards as the determinants of player wages likely differ across positions. (30) Re-estimating the models for regular and substitute players can also reveal interesting insights in the wage determination process. Moreover, the number of international appearances, previous as well as recent, should be weighted by the quality of the respective national team; that is, its position in the annual ranking of International Federation of Association Football (FIFA). Finally, estimating the models for different sub-periods will possibly yield information about changes in the wage determination process over time.

Second, the contract models will be extended too. First, annual salaries can be included in the estimations to control for unobserved heterogeneity among players. Second, young and old players clearly have different incentives; therefore, it is necessary to include a variable in the estimation that interacts the dummy for last year of contract with player age. Perhaps even more important is the fact that player contracts are of a rolling nature; that is, they are very often renewed before the old one is about to expire. Thus, the timing of renewal of a contract should also be included in the refined estimations as an additional explanatory variable. Finally, the question whether contract length and annual salaries are complements or substitutes needs to be addressed too (Link & Yosifov, 2011).

Clearly, current as well as proposed policy interventions in the now globalized football players' labor market would benefit from better contextual empirical evidence on the economic mechanisms that influence current practices in professional sports. Hence, future analyses should provide empirical evidence on how sports labor markets function economically, using this evidence to predict the likely consequences of proposed reforms. So far, an economic welfare analysis of effects of sports policies has been largely absent from recent debate, which tends to be dominated by specialists in law, sociology, and sports management. In particular, the need for some of the proposed interventions, such as quotas on team composition, is best assessed by asking whether the labor market for players is aalocatively efficiently, and if not, why not? So far, an analysis of labor market efficiency in professional sports has not been forthcoming due to data limitations. Using the available, and assembling new data from football across several countries, rigorous investigations of labor market structure, conduct, and performance in professional sports are possible and rewarding.

Appendix

[FIGURE A1 OMITTED]

[FIGURE A2 OMITTED]

[FIGURE A3 OMITTED]

[FIGURE A4 OMITTED]

Table A1: Means and standard deviations

Variable Mean Std. Dev. Min. Max.

Pay 909,014 889,577 17,043 10,000,000
Log(Pay) 13.31 .96 9.74 16.12
Games Played 13.27 12.62 0 34
Goals Scored 1.63 3.14 0 28
Intern. Appearances 1.43 3.08 0 25
Career Games 55.81 80.61 0 540
Career Goals 6.34 14.93 0 171
Career Intern. Appear. 7.54 16.56 0 130
Tenure 2.67 3.12 0 21
Captain .04 - 0 1
First Division Abroad .04 - 0 1
Goalkeeper .11 - 0 1
Defender .28 - 0 1
Midfielder .39 - 0 1
Forward .22 - 0 1
Germany .58 - 0 1
South America .05 - 0 1
North America .01 - 0 1
West Europe .13 - 0 1
East Europe .16 - 0 1
Africa .05 - 0 1
Asia/Australia .02 - 0 1

References

Baker, G. P. (1992). Incentive contracts and performance measurement. Journal of Political Economy, 100, 598-614.

Berri, D. J., &, Simmons, R. (2009). Race and the evaluation of signal callers in the National Football League. Journal of Sports Economics, 10, 23-43.

Bryson, A., Frick, B., & Simmons, R. (2009). The returns to scarce talent: Footedness and player remuneration in European football. Unpublished manuscript.

Buchinsky, M. (1998). Recent advances in quantile regression models: A practical guideline for empirical research. Journal of Human Resources, 33, 88-126.

Cantor, A. (1990). Firm-specific training and contract length. Economica, 57, 1-14.

Cantor, R. (1988). Work effort and contract length. Economica, 55, 343-353.

Conlin, M. (1999). Empirical tests of a separating equilibrium in National Football League contract negotiations. Rand Journal of Economics, 30, 289-304.

Conlin, M., & Emerson, P. M. (1999). Double moral hazard and a multi-dimensional separating equilibrium in contract negotiations: A study of the National Football League. Department of Economics, Cornell University, mimeo.

D'Agostino, R. B., Balanger, A., & D'Agostino, R. B., Jr. (1990). A suggestion for using powerful and informative tests of normality. American Statistician, 44, 316-327.

Dye, R. A. (1985). Optimal length of labor contracts. International Economic Review, 26, 251-270.

Engellandt, A., & Riphahn, R. (2005). Temporary contracts and employee effort. Labour Economics, 12, 281-299.

Feess, E., Frick, B., & Muhlheusser, G. (2004). Legal restrictions on buyout fees: Theory and evidence from German soccer. IZA Discussion Paper No. 1180, Bonn.

Fernie, S., & Metcalf, D. (1999). It's not what you pay it's the way that you pay it and that's what gets results: Jockeys' pay and performance. Labour, 13, 385-411.

Fort, R., & Maxcy, J. (1998). Last-period problems in sports. Department of Economics, Washington State University, mimeo.

Frick, B. (2003). Contest theory and sport. Oxford Review of Economic Policy, 19, 512-529.

Frick, B. (2007). The football players' labor market: Empirical evidence from the major European leagues. Scottish Journal of Political Economy, 54, 422-446.

Frick, B., & Deutscher, C. (2009). Die Entlohnung von Fuhrungsfahigkeiten: Evidenz aus dem professionellen Fufiball [The returns to leadership skills: Evidence from professional football]. Department of Management, University of Paderborn, mimeo.

Frick, B., Dilger, A., & Prinz, J. (2002). Arbeitsmarktregulierung und nachvertraglicher Oppor tunismus: Die Verhaltensfolgen garantierter Handgeldzahlungen in der National Football League [Labor market regulation and post-contractural opportunism: The effects of signing bonuses on player performance in the National Football League]. Zeitschrift fur Betriebswirtschaft, 72, 163-180.

Frick, B., Pietzner, G., & Prinz, J. (2007). Career duration in a competitive environment: The labor market for soccer players in Germany. Eastern Economic Journal, 33, 429-442.

Frick, B., Pietzner, G., & Prinz, J. (2009). Team performance and individual career duration: Evidence from the German "Bundesliga" In P. Andersson, P. Ayton, & C. Schmidt (Eds.), Myths and facts about football: The economics and psychology of the world's greatest sport (pp. 327-348). Cambridge, UK: Cambridge Scholars Publishing.

Gramm, C. L., & Schnell, J. F. (1994). Difficult choices: Crossing the picket line during the 1987 National Football League strike. Journal of Labor Economics, 12, 41-73.

Gramm, C. L., & Schnell, J. F. (1997). Following the leader: Race and player behavior in the 1987 NFL strike. Advances in the Economics of Sports, 2, 115-143.

Guadalupe, M. (2003). The hidden costs of fixed-term contracts: The impact on work accidents. Labour Economics, 10, 339-357.

Hamilton, B. H. (1997). Racial discrimination and professional basketball salaries in the 1990s. Applied Economics, 29, 287-296.

Holmes, P. (in press). New evidence on salary discrimination in Major League Baseball. Labour Economics, 18.

Holmstrom, B. (1982). Moral hazard in teams. Bell Journal of Economics, 13, 324-340.

Holmstrom, B., & Milgrom, P. (1991). Multitask principal-agent analyses: Incentive contracts, asset ownership and job design. Journal of Law, Economics, and Organization, 7, 24-52.

Horowitz, I., & Zappe, C. (1998). Thanks for the memories: Baseball veterans' end-of-career salaries. Managerial) and Decision Economics, 19, 377-382.

Huebl, L., & Swieter, D. (2002). Der Spielermarkt in der FuBball-Bundesliga [The player market in German professional football]. Zeitschrift fur Betriebswirtschaft, Ergdnzungsheft 4, Sportokonomie, 72, 105-125.

Ichino, A., & Riphahn, R. (2005). The effect of employment protection on worker effort. A comparison of absenteeism during and after probation. Journal of the European Economic Association, 3, 120-143.

Jimeno, J. F., & Toharia, L. (1996). Effort, absenteeism, and fixed term employment contracts. Revista Espanola de Economia, 13, 105-119.

Kahn, L. M. (2000). The sports business as a labor market laboratory. Journal of Economic Perspectives, 14, 75-94.

Kalter, F. (1999). Ethische Kundenpraferenzen im professionellen sport? Der Fall der FussballBundesliga [Ethical consumer preferences in professional sport? The case of the German "Bundesliga"]. Zeitschrift fur Soziologie, 28, 219-234.

Kicker. (1995-2008). Special Issues "Bundesliga". Nurnberg.

Koenker, R., & Bassett, G. (1978). Regression quantiles. Econometrica, 46, 33-50.

Krautman, A. C. (1990). Shirking or stochastic productivity in Major League Baseball? Southern Economic Journal, 56, 961-968.

Krueger, A. B. (2005). The economics of real superstars: The market for rock concerts in the material world. Journal of Labor Economics, 23, 1-30.

Kuethe, T. H., & Motamed, M. (2010). Returns to stardom: Evidence from U.S. Major League Soccer. Journal of Sports Economics, 11, 567-579.

Kuhn, P., & Weinberger, C. (2005). Leadership skills and wages. Journal of Labor Economics, 23, 395-436.

Leeds, M., & Kowalewski, S. (2001). Winner take all in the NFL. The effect of the salary cap and free agency on the compensation of skill position players. Journal of Sports Economics, 2, 244-256.

Lehmann, E. (2000). Verdienen Fussballspieler was sie verdienen? [Do football players deserve what they earn?] In H.-M. Schellhaass (Ed.), Sportveranstaltungen zwischen Liga--und Medieninteressen (pp. 97-121). Schorndorf: Karl Hofmann.

Lehmann, E. E., & Schulze, G. G. (2008). What does it take to be a star? The role of performance and the media for German soccer players. Applied Economics Quarterly, 54, 59-70.

Lehmann, E., & Weigand, J. (1999). Determinanten der Entlohnung von Profifussballspielern Eine empirische analyse fur die deutsche Bundesliga [The determinants of player salaries--An empirical analysis with data from German soccer]. Betriebswirts chaftliche Forschung und Praxis, 51, 124-135.

Lehn, K. (1982). Property rights, risk sharing, and player disability in Major League Baseball. Journal of Law and Economics, 25, 347-366.

Lehn, K. (1984). Information asymmetries in baseball's free agent market. Economic Inquiry, 22, 37-44.

Link, C. R., & Yosifov, M. (in press). Contract length and salaries: Compensating wage differentials in Major League Baseball. Journal of Sports Economics, 12.

Lucifora, C., & Simmons, R. (2003). Superstar effects in sports: Evidence from Italian soccer. Journal of Sports Economics, 4,-55.

Marburger, D. R. (2003). Does the assignment of property rights encourage or discourage shirking? Evidence from Major League Baseball. Journal of Sports Economics, 4, 19-34.

Maxcy, J. G. (1997). Do long-term contracts influence performance in Major League Baseball? Advances in the Economics of Sports, 2, 157-176.

Maxcy, J. G., Fort, R. D., & Krautman, A. C. (2002). The effectiveness of incentive mechanisms in Major League Baseball. Journal of Sports Economics, 3, 246-255.

Milgrom, P., & Roberts, J. (1992). Economics, management, and organization. Englewood Cliffs, NJ: Prentice-Hall.

Muller-Jentsch, W. (1989). Basisdaten der industriellen Beziehunge [Long-term data on German industrial relations]. Frankfurt, Germany: Campus

van Ommeren, J., & Hazans, M. (2008). Workers' valuation of the remaining contract duration. Economica, 75, 116-139.

Pedace, R. (2008). Earnings, performance, and nationality discrimination in a highly competitive labor market as an analysis of the English professional soccer league. Journal of Sports Economics, 9, 115-140.

Reilly, B., & Witt, R. (2007). Determinants of base pay and the role of race in Major League Soccer: Evidence from the 2007 league season. Department of Economics, University of Surrey, Discussion Paper D19/07.

Riphahn, R., & Thalmeier, A. (2001). Behavioral effects of probation periods: An analysis of worker absenteeism. Jahrbucher fur Nationalokonomie und Statistik, 221, 179-201.

Rosen, S. (1981). The economics of superstars. American Economic Review, 71, 845-858.

Rottenberg, S. (1956). The baseball player's labor market. Journal of Political Economy, 64, 242-258.

Simmons, R., & Berri, D. J. (2009). Gains from specialization and free agency: The story from the gridiron. Review of Industrial Organization, 34, 81-98.

Sonnenberg, M. (2002). Als die Gehalter laufen lernten [When salaries started skyrocketing]. SportBild, 2, 24-29.

Stiroh, K. J. (2007). Playing for keeps: Pay and performance in the NBA. Economic Inquiry, 45, 145-161.

Torgler, B., Schmidt, S., & Frey, B. (2006). Relative income position and performance: An empirical panel analysis. FEEM Discussion Paper No. 39, Milano.

Vincent, C., & Eastman, B. (2009). Determinants of pay in the NFL. A quantile regression approach. Journal of Sports Economics, 10, 256-277.

Wilson, D., & Ying, Y.-H. (2003). Nationality preferences for labour in the international football industry. Applied Economics, 35, 1551-1559.

Yang, C.-H., & Lin, H.-Y. (in press). Is there salary discrimination by nationality in the NBA? Foreign talent or foreign market. Journal of Sports Economics, 12.

Endnotes

(1) In the same year, the average salary of a full-time blue collar worker amounted to 7,775 DM. (i.e., about 60% of a football player's annual income). Today, the average player salary is about 45 times the average salary in Germany.

(2) In the latter year, Oliver Bierhoff--later Matthaus' teammate in Munich--earned more than 12 Mio. DM in Italy.

(3) With respect to contract length as a "discipline device," see A. Cantor (1990), R. Cantor (1988), Dye (1985), and van Ommeren and Hazans (2008).

(4) The few studies that have been conducted with representative samples of employees (see Engellandt & Riphahn, 2005; Riphahn & Thalmeier, 2001; Guadalupe, 2003; Ichino & Riphahn, 2005; Jimeno & Toharia, 1996) use information on work accidents, unpaid overtime, and absence days as dependent variables. Although the result reported in these papers is surprisingly consistent--fixed-term contracts induced higher levels of performance--the independent variables are far from perfect.

(5) Contrary to the findings reported by Horowitz and Zappe (1998) for baseball veterans, this suggests that nostalgia effects will be of minor importance only.

(6) I am grateful to an anonymous referee for making this point because it has obvious implications for the empirical analysis presented below. Contrary to the situation in most American team sports, few individual performance measures are recorded in football. Apart from the number of goals scored, assists made, tackles won, and yellow and red cards received, as well as number of substitute appearances, nothing is available at acceptable cost. It is, therefore, difficult to distinguish talent from popularity and/or fan appeal. Fortunately, it turns out that the set of measures that I use below to describe a player's talent, popularity, and fan appeal are not very highly correlated (i.e., multicollinearity is not a problem).

(7) Publications in German include Frick and Deutscher (2009), Huebl and Swieter (2002), Lehmann (2000), and Lehmann and Weigand (1999). With the exception of the latter, all these papers use much smaller samples from short sub-periods since the early 1990s.

(8) On the determinants of individual career length, see inter alia, Frick (2007) as well as Frick, Pietzner, and Prinz (2007, 2009).

(9) Thus, the variables I use to explain the observable variance in player salaries are indirect at best and measure an individual player's contribution to his team's economic performance only imperfectly.

(10) This is interesting, insofar as Theo Zwanziger, in the interview quoted above, also argued that many politicians, by supporting the developments that have been induced by the Bosnian-ruling of the European Court of Justice in December 1995, "have made few particularly gifted players richer and richer and the clubs poorer and poorer." He then went on to argue that "UEFA and the national associations will do their very best to introduce an individual salary cap and to reach a more egalitarian wage structure in professional football." However, the distribution of player salaries (as measured by the Gini coefficient) has remained more or less constant between 1995-1996 and 2007-2008 providing little reason for such an intervention.

(11) Although the Hausman-Test suggests using the results from the fixed effects estimation, I report the findings of the random effects estimation. The problem is that region of birth is a constant for each player and cannot be used in a fixed effects estimation. However, the differences between the remaining coefficients in the RE and the FE estimations are negligible.

(12) Contrary to the situation in most American team sports leagues with their abundance of performance figures, measurement of individual player performance in European football can be problematic, especially for defenders whose task it is to prevent the opposing teams' forwards to score goals. While counting the number of goals scored, shots on goal, and assists is straightforward, it is far more difficult to assess the performance of defensive players. In future work, I will therefore estimate the models separately for the different groups of players.

(13) This unexpected result survives a number of different specifications: Interacting the number of career goals with the position dummies leaves the finding virtually unaffected. Moreover, estimating the model separately by position yields the same result for forwards and midfielders but not for defenders. Estimating the model only for position players (i.e., without the goalkeepers) again yields the strange coefficients.

(14) This is in line with Pedace (2008) who finds that players from South America also receive preferential labor market treatment in the English Premier League. He argues that this positive discrimination is a rational response from owners who observe increased attendance with a larger presence of South American players.

(15) Anecdotal evidence seems to support the argument that some players suffer from home sickness once they are traded to another club.

(16) Estimating the models with the lagged annual salary to control for unobserved heterogeneity reduces the sample size considerably from 6,100 player-year-observations to 4,700. Although most of the coefficients retain their statistical significance, their magnitudes are somewhat reduced. The complete results are available from the author upon request.

(17) This term has first been used by Alan Krueger (2005), analyzing the revenues generated by particularly successful rock bands and musicians.

(18) However, recent evidence from the National Basketball Association (Yang & Lin, 2010) as well as from Major League Baseball (Holmes, 2010) suggests that, particularly at the lower end of the salary distribution, discrimination by race and/or nationality seems to persist.

(19) In further research, subjective evaluations of a player's performance (i.e., school grades) will also be used to estimate the hedonic wage equations (for a first application see Section 3 below).

(20) Using data from ten consecutive seasons (1990-1991 to 1999-2000) from the first German soccer division, Frick (2003) shows that the c.p. of the percentage of variable pay positively affects the performance of the teams. This finding, however, raises a further question: If the teams that pay their players, to a large extent via bonuses, are more successful than those who prefer fixed payments, why do not all teams turn to performance-related pay? The negative correlation between the log of total pay and the percentage of variable pay suggests that poor teams motivate their employees via bonuses while rich teams achieve this goal by paying high fixed salaries.

(21) In this view, contractually secured income may entice the player to shirk if the utility sacrificed with effort is not offset with income. However, it is also possible that long-term contracts are used as tournament devices. The reward of a secured multiyear contract may be part of a lucrative compensation package designed to increase competition among workers. Thus, in a tournament setting, such contracts may serve as incentives for which workers compete by increasing their individual effort levels. Moreover, long-term contracts may also be offered to players by risk-averse managers for risk management purposes.

(22) These are players appearing in at least 25% of all regular season matches. Since the league is formed by 18 teams, each team has 17 home matches and 17 away matches.

(23) Since the dataset is an unbalanced panel, the number of years the individual players have been active in the Bundesliga differs considerably. Note, however, that the presence in the dataset is not identical with the duration of individual careers. First, if a player cannot retain his status as a regular player (e.g., due to lack of fitness or injury), he disappears from the dataset even though he is still active as a back-up player for one of the first division teams. Second, many of the regular players started their career as substitutes who later managed to become established players. This means that they have been playing in the Bundesliga already before they appear in the dataset.

(24) The school grades that are used here to express a player's performance--ranging from very good to very poor--are clearly, but not exclusively, affected by whether a player receives a yellow or a red card and whether he scores a goal or produces an assist. Thus, the school grades express, in a simple one digit figure, a player's contribution to his team's performance on the pitch in a particular match.

(25) Kernel density estimates of the dependent variables are displayed in Figures A2-A4 in the Appendix.

(26) Estimating models 1.1-1.3 without the average team grade as an exogenous variable leaves the coefficient of the contract status variable unaffected. The results are, of course, available from the author upon request.

(27) The results are, of course, available from the author upon request.

(28) A larger share of players negotiating a new contract can, of course, also be problematic for the team's managers, because they may find themselves in a hold-up situation where, particularly the stars, can credibly threaten to sign with another team.

(29) A first example for an extension that is currently being performed is Bryson, Frick, and Simmons (2009) who use an unbalanced panel from the Bundesliga, as well as a cross-section from the Big Five European leagues to analyze the impact of both-footedness and left-footedness on player remuneration. Controlling for player age, height, position, and national league, they find that both-feet players enjoy a pay premium of more than 20%, while left-footed players receive a statistically significant premium of about 10%.

(30) Another possibility is to interact the position dummies with the number of goals scored, the number of career appearances, the number of international appearances, etc. to see whether the returns to experience and popularity differ by, or are equal across, position.

Author's Note

I would like to thank Marcel Battre, Christian Deutscher, Thomas Fritz, Julia Nagelschneider, and Wiebke Held for their assistance in compiling the datasets used in this study. Errors and omissions are, of course, my own responsibility.

Bernd Frick [1]

[1] University of Paderborn

Bernd Frick is a professor in the Department of Management. His research interests include personnel economics and sport economics.

Table 1: Are fans envious?

Possible Responses %

Yes, because politicians are 19
 obliged to intervene if
 certain developments in
 society are causing
 discontent.
Yes, because the salaries in 0
 football are simply too
 high.
No, because politicians
 should in principle 36
 abstain from intervening
 in private businesses.
No, because the salaries are 45
 the result of market
 forces.

Source: welt.de

Table 2: Estimation results I: Various methods

Variable Random Effects Robust OLS

 B T B T

Age .5121 22.43 *** .4559 18.99 ***
[Age.sup.2] -.0092 -21.48 *** -.0083 -18.69 ***
Games Played .0191 25.66 *** .0240 31.95 ***
Career Games .0042 7.48 *** .0056 11.27 ***
Career -.0021 -5.97 *** -.0028 -9.18 ***
 [Games.sup.2] * 100
Career [Games.sup.]3 .0033 5.46 *** .0043 8.07 ***
 * 10000
International Caps .0848 6.86 *** .0903 6.04 ***
International -.0071 -3.56 *** -.0081 -2.79 ***
 [Caps.sup.2]
International .0002 2.19 ** .0002 1.74 *
 [Caps.sup.3]
Career Caps .0118 4.19 *** .0125 5.36 ***
Career [Caps.sup.2] -.0003 -3.40 *** -.0003 4.17 ***
Career [Caps.sup.3] .0017 2.99 *** .0016 3.67 ***
 * 1000
Goals Scored .0444 14.24 *** .0465 16.28 ***
Career Goals -.0129 -4.71 *** -.0114 -4.69 ***
Career [Goals.sup.2] .0002 4.13 *** .0002 4.38 ***
Career [Goals.sup.3] -.0011 -3.68 *** -.0011 -4.15 ***
 * 1000
Tenure -.0142 -4.43 *** -.0187 -6.46 ***
Captain (1 = yes) .2692 6.60 *** .3406 10.17 ***
First Division Abroad .5910 12.46 *** .6159 11.41 ***
Defender .2113 5.17 *** .0990 3.20 ***
Midfielder .2677 6.65 *** .1667 5.34 ***
Forward .3157 7.14 *** .2167 5.97 ***
South America .4494 8.23 *** .3778 9.87 ***
North America -.0822 -.73 + -.1785 -1.92 *
West Europe .2442 6.62 *** .1848 7.00 ***
East Europe .0774 2.23 ** .0329 1.36 *
Africa .0654 1.24 + -.0117 -.30 +
Asia/Australia .0928 1.28 + .0099 .20 +
Constant 5.8725 19.30 *** 6.8245 21.14 ***
Team Dummies included
Season Dummies included
N of Observations 6,147 6,147
Obs. per Player 1-13 -
Number of Players 1,993 -
[R.sup.2] * 100 61,7 62,7
F-Value - 164.5 ***
Wald Chi2 6,672.0 *** -
LM-Test 392.0 *** -
Raw Sum of Dev. - -
Min Sum of Dev. - -

Variable Median Regression

 B T

Age .4361 23.71 ***
[Age.sup.2] -.0079 -23.12 ***
Games Played .0226 33.12 ***
Career Games .0057 12.46 ***
Career -.0030 -10.26 ***
 [Games.sup.2] * 100
Career [Games.sup.]3 .0046 9.06 ***
 * 10000
International Caps .0909 8.02 ***
International -.0094 -5.01 ***
 [Caps.sup.2]
International .0003 4.09 ***
 [Caps.sup.3]
Career Caps .0131 5.94 ***
Career [Caps.sup.2] -.0003 -4.48 ***
Career [Caps.sup.3] .0016 3.67 ***
 * 1000
Goals Scored .0513 18.26 ***
Career Goals -.0077 -3.56 ***
Career [Goals.sup.2] .0001 3.31 ***
Career [Goals.sup.3] -.0007 -3.11 ***
 * 1000
Tenure -.0153 -6.53 ***
Captain (1 = yes) .3718 10.50 ***
First Division Abroad .6346 15.11 ***
Defender .0539 2.24 **
Midfielder .0965 4.04 **
Forward .1020 3.68 **
South America .3824 11.91 ***
North America -.1510 -2.10 **
West Europe .1969 8.53 ***
East Europe .0200 .95 +
Africa -.0166 -.52 +
Asia/Australia .0185 .42 +
Constant 7.1631 29.21 ***
Team Dummies
Season Dummies
N of Observations 6,147
Obs. per Player -
Number of Players -
[R.sup.2] * 100 40,5
F-Value -
Wald Chi2 -
LM-Test -
Raw Sum of Dev. 4,656.6
Min Sum of Dev. 2,772.6

Notes: + not significant, * p < .10, ** p < .05, *** p < .01

Table 3: Estimation results II: Quantile regressions

Variable .1 Quantile .25 Quantile

Age .5415 *** .5485 ***
[Age.sup.2] -.0097 *** -.0099 ***
Games Played .0347 *** .0271 ***
Career Games .0050 *** .0058 ***
Career [Games.sup.2] -.0027 *** -.0034 ***
 * 100
Career [Games.sup.3] .0042 *** .0057 ***
 * 10000
International Caps .0340 ** .0568 ***
International -.0003 + -.0034 *
 [Caps.sup.2]
International .0000 + .0000 +
 [Caps.sup.3]
Career Caps .0108 *** .0119 ***
Career [Caps.sup.2] -.0002 ** -.0003 ***
Career [Caps.sup.3] .0014 ** .0019 ***
 * 1000
Goals Scored .0453 *** .0511 ***
Career Goals -.0094 ** -.0038 +
Career [Goals.sup.2] .0002 *** .0000 +
Career [Goals.sup.3] -.0014 *** .0000 +
 * 1000
Tenure -.0134 *** -.0181 ***
Captain (1 = yes) .3662 *** .3742 ***
First Division .7485 *** .6895 ***
 Abroad
Defender .2154 *** .1049 ***
Midfielder .2414 *** .1458 ***
Forward .2832 *** .1634 ***
South America .3010 *** .3086 ***
North America -.1989 + -.0509 +
West Europe .1999 *** .1992 ***
East Europe .0635 * .0690 ***
Africa -.0153 + .0538 +
Asia/Australia .1296 + .1042 **
Constant 4.6571 *** 5.1341 ***
Team Dummies included
Season Dummies included
N of Cases 6,147 6,147
Pseudo [R.sup.2] 43.6 42.4
 * 100
Raw Sum of Dev. 2,196.5 3,891.5
Min Sum of Dev. 1,239.1 2,240.8

Variable .75 Quantile .9 Quantile

Age .3660 *** .2829 ***
[Age.sup.2] -.0068 *** -.0055 ***
Games Played .0173 *** .0124 ***
Career Games .0047 *** .0030 ***
Career [Games.sup.2] -.0021 *** -.0001 **
 * 100
Career [Games.sup.3] .0030 *** .0013 +
 * 10000
International Caps .1241 *** .1129 ***
International -.0149 *** -.0114 ***
 [Caps.sup.2]
International .0006 *** .0004 ***
 [Caps.sup.3]
Career Caps .0126 *** .0122 ***
Career [Caps.sup.2] -.0002 *** -.0002 *
Career [Caps.sup.3] .0013 *** .0009 +
 * 1000
Goals Scored .0486 *** .0425 ***
Career Goals -.0132 *** -.0077 *
Career [Goals.sup.2] .0003 *** .0002 **
Career [Goals.sup.3] -.0001 *** -.0009 **
 * 1000
Tenure -.0201 *** -.0177 ***
Captain (1 = yes) .3114 *** .3296 ***
First Division .5848 *** .4772 ***
 Abroad
Defender -.0002 + -.1560 ***
Midfielder .0756 *** -.0537 +
Forward .1111 *** -.0170 +
South America .3863 *** .4230 ***
North America -.2002 *** -.2519 *
West Europe .1637 *** .1627 ***
East Europe -.0344 + .0085 +
Africa -.0389 + -.0320 +
Asia/Australia -.2022 *** -.1494 *
Constant 8.6862 *** 10.4911 ***
Team Dummies
Season Dummies
N of Cases 6,147 6,147
Pseudo [R.sup.2] 39.2 39.2
 * 100
Raw Sum of Dev. 3,577.2 1,934.0
Min Sum of Dev. 2,139.6 1,175.5

Notes: + not significant, * p < .10, ** p < .05, *** p < .01

Table 4: Player opportunism in professional team sports: A selective
review of the literature

Author(s) and League and Data Basic Findings
Year of Used
Publication

Lehn (1982) 650 MLB players Long-term contracts increase
 in 1980 the amount of time spent on
 the disabled list: Each
 additional year remaining on
 the contract is associated
 with a 25% increase in the
 average number of days spent
 on the disabled list. This
 is due to the fact that
 guaranteed multiyear
 contracts reduce the
 incentives for players to
 invest in proper physical
 conditioning. However, the
 disincentive effect of
 long-term contracts can be
 mitigated by inclusion of
 incentive bonuses in player
 contracts.
Lehn (1984) 155 MLB players Players who re-sign for at
 in 1980 least three years with their
 old team experience a
 significantly smaller
 increase in days spent on
 the disabled list than
 players who signed for three
 years or even longer with
 another team.
Krautman (1990) 110 MLB players There is no evidence of a
 (only hitters) significant departure from
 signing contracts the means of players'
 of more than 5 productivity distributions
 years duration, due to proximity to contract
 1976-1983 negotiations. Thus, the
 observable variation in
 performance is the result of
 a stochastic process rather
 than shirking.
Scoggins (1993) 110 MLB players A convincing answer to the
 (only hitters) question whether shirking
 signing contracts occurs or not depends on the
 of more than 5 choice of the performance
 years duration, measure (i.e., if total
 1976-1983 bases instead of slugging
 average is used as a
 performance measure,
 shirking can be detected).
Gramm and 1,106 out of Players with long-term
 Schnell 1,260 players contracts were less likely
 (1994, 1997) under contract to participate in the 1987
 with one of the strike. Since the main
 28 NFL teams on reason for the strike was
 September 21, the NFLPA's demand for free
 1987 agency and since average
 career duration in the NFL
 is rather short (i.e., about
 three seasons), players with
 long-term contracts were
 less likely to reap the
 benefits of free agency and
 were, therefore, more
 interested in maximizing
 their current incomes.
Maxcy (1997) MLB 1986-1993; For players with long-term
 1,343-2,284 contracts, status in regard
 player-year to re-contracting at the end
 observations for of the current season does
 hitters and not influence performance.
 882-1,552 The reason is that long-term
 player-year contracts are given to
 observations for players who have already
 pitchers proven themselves as
 reliable and consistent
 performers and are,
 therefore, not likely to
 shirk.
Fort and Maxcy MLB 1986-1993; Performance does not
 (1998) 2,238 player-year increase as players approach
 observations for renegotiation (i.e., when
 hitters and 1,625 the individual contract is
 player-year about to expire and when the
 observations for individual, therefore,
 pitchers should be most likely to
 expend more effort).
 Particularly when players
 with long-term contracts are
 analyzed from the shirking
 perspective, there is no
 evidence of shirking.
Conlin (1999); 1,873 of the Rookies signing their first
 Conlin and 2,016 players contract after training camp
 Emerson (1999) selected in the has started reveal positive
 1986-1991 NFL private information about
 drafts their abilities. Thus,
 players who sign after
 longer contract negotiations
 are of higher ability levels
 (i.e., higher percentage of
 active contracts and higher
 number of games started in
 first three years).
 Moreover, a player's effort
 level is influenced by the
 remaining duration of his
 contract; that is, the
 number of games started is
 significantly higher in the
 last year of the contract.
Fernie and 50 British When performance-related pay
 Metcalf (1999) jockeys, is replaced by guaranteed
 1983-1995 annual salaries via
 so-called "retainers," the
 individual athlete's
 performance deteriorates
 dramatically. Thus,
 non-contingent payments
 introduce moral hazard into
 a payment system which had
 previously proved to be
 rather successful in
 overcoming such behavior.
 Not surprisingly, therefore,
 these non-contingent
 payments have been largely
 abandoned recently.
Frick, Dilger, 349 The only legal way to
 and Prinz team-year circumvent or partly avoid
 (2002) observations, the hard salary cap in the
 NFL, 1988-1999 NFL is by paying signing
 bonuses to free agents
 (these bonuses are
 prorated). However, these
 guaranteed up-front payments
 that are unrelated to actual
 performance induce players
 to behave opportunistically:
 The higher the percentage of
 the signing bonuses, the
 poorer the performance of
 the team.
Maxcy, Fort, 1,160 For both pitchers and
 and Krautman player-year hitters, time spent on the
 (2002) observations on disabled list decreases in
 213 hitters and the season immediately
 812 player-year preceding contract
 observations on negotiations. Moreover,
 140 pitchers in playing time is above
 MLB average in that season.
 However, there is no
 evidence of ex post
 opportunism because
 long-term contracts do not
 cause a subsequent decline
 in performance.
Marburger (2003) 279 free agent The redistribution of
 nonpitchers property rights that was
 signing contracts caused by the conversion
 between 1990 and from the reserve clause to
 1993 with any of free agency should have
 the MLB-teams and increased player effort.
 133 nonpitchers However, free agency also
 in 1970 who had saw an increase in multiyear
 at least six contracts which, in turn,
 years of major creates shirking incentives.
 league experience The net impact of property
 rights assignment on
 shirking in MLB is obvious:
 Free agents with one- and
 two-year contracts
 outperform comparable
 reserve clause players over
 the same time frame. This is
 not the case for free agents
 with contracts exceeding two
 years.
Berri and 515 player-year Depending on the specific
 Krautman (2006) observations in measure of performance used
 the NBA, 2000/ in the estimates, the
 2001-2002/2003 evidence appears to be
 mixed: Although in the first
 estimation the effect of
 signing a long-term contract
 on performance is
 significantly negative, the
 economic impact is small. In
 the second estimation,
 however, even this small
 impact disappears.
Stiroh (2007) 349 NBA-player Individual performance
 contracts signed improves in the year before
 1993-2001 (2,077 signing a multiyear contract
 player-year and declines after the
 observations) contract is signed. This is
 consistent with an observed
 salary structure that
 rewards both historical
 performance and recent
 improvement and thus
 provides strong incentives
 to increase effort and
 improve performance before
 signing a new multiyear
 contract.

Table 5: Alternative measures of remaining
contract duration *

Variable Mean Std Dev

Remaining Contract 1, 35 1, 7
 Years
Recoded Number of 1, 17 0, 79
 Remaining Years ##
Last Year of 0, 24 -
 Contract (0 = no;
 1 = yes)

Notes: # Number of cases = 1,866
individual-year-observations for regular
players ## Number of years > 2 recoded
as 2

Table 6: Contract status and player performance (Dependent
variable: Average player grade)

Variable Model 1.1 Model 1.2
 Contract Variable: Contract Variable:
 Remaining Years Remaining Years
 [less than or
 equal to] 2

Average Team .8052 19.72 *** .8054 19.76 ***
Grade .0408 .0408
Contract Status .0187 2.70 *** .0283 3.32 ***
 .0069 .0085
Defender -.0632 -.33 + -.0752 -.40 +
 .1890 .1887
Midfielder -.0884 -.47 + -.0999 -.53 +
 .1894 .1892
Forward -.0277 -.14 + -.0407 -.21 +
 .1940 .1937
Goals Scored -.0385 13.23 *** -.0385 -13.25 ***
 .0029 .0029
Red Cards .0079 .36 + .0074 .34 +
 .0220 .0219
Yellow Cards -.0105 -3.52 *** -.0103 -3.49 ***
 .0029 .0029
Semi-Profession. -.1570 -2.04 ** -.1556 -2.03 **
 .0768 .0767
Career Games .0012 1.82 * .0012 1.87 *
 .0006 .0006
Career [Games.
 sup.2] (#) -.0001 -1.97 ** -.0001 -2.03 **
 .0005 .0000
International .0001 .17 + .0001 .14 +
 .0011 .0011
Age -.0782 -2.05 ** -.0849 -2.23 **
 .0380 .0381
[Age.sup.2] .0011 1.95 * .0012 2.13 **
 .0006 .0006
Team Change -.1207 -3.16 *** -.1145 -3.03 ***
 .0382 .0377
Constant 2.0694 3.24 *** 2.1671 3.40 ***
 .6380 .6365

N of Obs. 1,863 1,863
N of Players 760 760
F-Value 54.7 55.1
LM-Test 138.8 *** 143.2 ***
Hausman 111.6 *** 111.6 ***

Variable Model 1.3
 Contract Variable:
 Last Year-Dummy

Average Team .8100 19.80 ***
Grade .0409
Contract Status -.0413 -2.85 ***
 .0145
Defender -.0717 -.38 +
 .1889
Midfielder -.0994 -.53 +
 .1894
Forward -.0395 -.20 +
 .1939
Goals Scored -.0385 -13.23 ***
 .0029
Red Cards .0069 .32 +
 .0219
Yellow Cards -.0104 -3.49 ***
 .0029
Semi-Profession. -.1616 -2.10 **
 .0768
Career Games .0012 1.84 *
 .0006
Career [Games.
 sup.2] (#) -.0001 -1.97 **
 .0000
International .0001 .09 +
 .0011
Age -.0846 -2.22 **
 .0381
[Age.sup.2] .0012 2.07 **
 .0006
Team Change -.1059 -2.81 ***
 .0376
Constant 2.2136 3.47 ***
 .6379

N of Obs. 1,863
N of Players 760
F-Value 54.8
LM-Test 143.5 ***
Hausman 108.6 ***

Notes: (#) coefficient multiplied by 1,000 for ease of
presentation, + not significant, * p < .10, ** p < .05, p < .01

Table 7: Contract status and player performance (Dependent
variable: Average player grade relative to average grade of team)

Variable Model 2.1 Model 2.2
 Contract Variable: Contract Variable:
 Remaining Years Remaining Years
 [less than or
 equal to] 2

Contract Status .0199 2.85 *** .0296 3.44 ***
 .0069 .0086
Defender -.0355 -.19 + -.0479 -.25 +
 .1907 .1905
Midfielder -.0665 -.35 + -.0784 -.41 +
 .1912 .1910
Forward -.0059 -.03 + -.0195 -.10 +
 .1958 .1956
Goals Scored -.0369 12.65 *** -.0369 -12.66 ***
 .0029 .0029
Red Cards -.0017 -.08 + -.0022 -.10 +
 .0221 .0220
Yellow Cards -.0115 -3.85 *** -.0114 -3.81 ***
 .0030 .0030
Semi-Profession. -.1509 -1.94 ** -.1495 -1.93 **
 .0776 .0775
Career Games .0010 1.45 * .0010 1.51 +
 .0006 .0006
Career [Games.
 sup.2] (#) -.0009 -1.59 + -.0009 -1.66 **
 .0005 .0005
International .0005 .51 + .0005 .48 +
 .0011 .0011
Age -.0753 -1.96 ** -.0823 -2.14 **
 .0384 .0384
[Age.sup.2] .0010 1.78 * .0012 1.97 **
 .0006 .0006
Team Change -.1397 -3.64 *** -.1328 -3.50 ***
 .0384 .0379
Constant 1.3492 2.16 ** 1.4533 2.33 ***
 .6260 .6249

Nof Obs. 1,863 1,863
N of Players 760 760
F-Value 17.7 18.0
LM-Test 130.8 *** 134.6 ***
Hausman 95.3 *** 95.7 ***

Variable Model 2.3
 Contract
 Variable: Last
 Year-Dummy

Contract Status -.0466 -3.19 ***
 .0146
Defender -.0462 -.24 +
 .1906
Midfielder -.0798 -.42 +
 .1911
Forward -.0199 -.10 +
 .1957
Goals Scored -.0369 -12.66 ***
 .0029
Red Cards -.0025 -.11 +
 .0221
Yellow Cards -.0114 -3.81 ***
 .0030
Semi-Profession. -.1560 -2.01 **
 .0775
Career Games .0010 1.50 +
 .0006
Career [Games.
 sup.2] (#) -.0009 -1.61 +
 .0005
International .0004 .42 +
 .0011
Age -.0827 -2.15 **
 .0385
[Age.sup.2] .0011 1.93 *
 .0006
Team Change -.1236 -3.27 ***
 .0378
Constant 1.5288 2.44 ***
 .6264

Nof Obs. 1,863
N of Players 760
F-Value 17.9
LM-Test 134.5 ***
Hausman 94.5 ***

Notes: (#) coefficient multiplied by 1,000 for ease of
presentation, + not significant, * p < .10, ** p < .05, p < .01

Table 8: Contract status and player performance (Dep.
variable: Variance of player grade)

Variable Model 3.1 Model 3.2 Contract
 Contract Variable: Variable: Remaining
 Remaining Years Years [less than or
 equal to] 2

Contract Status .0090 1.64 * .0150 2.08 **
 .0055 .0072
Number of Graded -.0034 -3.67 *** -.0034 -3.69 ***
 Appearances .0009 .0009
Average Team Grade .1105 3.57 *** .1110 3.59 ***
 .0310 .0309
Defender -.0561 -1.99 ** -.0569 -2.01 **
 .0282 .0282
Midfielder -.0264 -.94 + -.0271 -.96 +
 .0281 .0281
Forward .0178 .56 + .0169 .53 +
 .0319 .0320
Goals Scored .0377 16.57 *** .0377 16.58 ***
 .0022 .0022
Red Cards .0546 2.86 *** .0545 2.86 +
 .0190 .0190
Yellow Cards -.0024 -.99 + -.0024 -.97 +
 .0024 .0024
Career Games -.0001 -.83 + * -.0001 -.79 +
 .0001 .0001
Career [Games.
 sup.2] (#) .0001 .77 + .0001 .73 +
 .0002 .0002
International .0569 4.52 *** .0570 4.53 ***
 .0126 .0125
Age .0146 .78 + .0130 .69 +
 .0187 .0187
[Age.sup.2] -.0003 -1.02 + -.0003 -.93 +
 .0003 .0003
Semi- Professional -.0045 -.17 + -.0037 -.14 +
 .0262 .0262
Team Change -.0201 -.65 + -.0188 -.61 +
 .0310 .0308
Constant .1602 .60 + .1747 .65 +
 .2691 .2688

N of Obs. 1,863 1,863 1,863
N of Players 760 760 760
F-Value 29.4 29.5 29.3
Wald [X.sup.2] 611.2 *** 611.4 ***
LM-Test 26.5 *** 26.3 ***
Hausman 14.5 + 13.3 +

Variable Model 3.3 Contract
 Variable: Last
 Year-Dummy

Contract Status -.0142 -1.11 +
 .0128
Number of Graded -.0034 -3.65 ***
 Appearances .0009
Average Team Grade .1073 3.47 ***
 .0309
Defender -.0579 -2.05 **
 .0282
Midfielder -.0279 -.99 +
 .0281
Forward .0158 .49 +
 .0320
Goals Scored .0378 16.61 ***
 .0022
Red Cards .0547 2.86 ***
 .0191
Yellow Cards -.0024 -.98 +
 .0024
Career Games -.0001 -.86 +
 .0001
Career [Games.
 sup.2] (#) .0001 .80 +
 .0002
International .0584 4.64 ***
 .0125
Age .0142 .76 +
 .0187
[Age.sup.2] -.0003 -1.02 +
 .0003
Semi- Professional -.0046 -.18 +
 .0263
Team Change -.0138 -.45 +
 .0307
Constant .1985 .74 +
 .2694

N of Obs.
N of Players
F-Value
Wald [X.sup.2] 606.6 ***
LM-Test 26.9 ***
Hausman 12.4 +

Notes: (#) coefficient multiplied by 1,000 for ease of
presentation, + not significant, * p < .10, ** p < .05, p < .01