Financial incentives and strategic behavior in European professional football: A match day analysis of starting squads in the German Bundesliga and UEFA competitions.

Rohde, Marc ; Breuer, Christoph

Abstract

Europe's professional football clubs engage in intraseasonal races to win the league, qualify for Union of European Football Associations (UEFA) competitions, and avoid relegation. In these competitions, playing talent is a scarce good as players invest effort in the form of fitness and the risk of injuries. Thus, managers face incentives to adjust effort levels by means of changing the value of starting squads. This paper analyzes whether managers save efforts in the absence of financial incentives or ahead of more important games. The theoretical model extends previous papers from the contest theory and shirking literature to a match-day panel regression analysis of strategic behavior. We build on two unbalanced panels covering 10 consecutive seasons in the German Bundesliga (n=6,120) and UEFA Champions League (n=1,920). The insights generated in this article are important for national league administrators as well as the UEFA in aligning their competitions towards each other and increasing incentives towards participating teams.

Keywords: sport economics, soccer, league design, contest theory, managerial effort, financial incentives

Introduction

This article addresses team effort, financial incentives, and strategic behavior by club managers in European professional football. League organizers are interested in the effort levels and strategic behavior of clubs to maintain the integrity of the competition and increase the quality of the league and the excitement for visitors.

League organizers and clubs interact with each other in the complex environment of European professional football. League organizers pursue several objectives, including the participation and the attraction of effort by participants, as well as competitive balance within the competition (Rottenberg, 1956; Szymanski, 2003). In order to achieve these objectives, they can design the contest format and adjust financial incentives for participants. In European professional football, there are two layers of competitions, including national competitions (leagues and cups) and Union of European Football Associations (UEFA) competitions (Champions League, Europa League). Clubs often face a high workload from playing in several competitions simultaneously. Typically, clubs in the "big five" European leagues play 34 to 38 league games per season. In addition, they may play a number of games in national cups. Participants in UEFA competitions may play up to 15 further games per season. Since players use up fitness and risk injuries while playing, teams need to manage playing talent as a scarce resource. League organizers, on the other hand, compete with each other to attract team effort (Azmat & Moller, 2009).

Contest theory would suggest that teams engage in strategic behavior by varying effort levels in response to financial incentives in national and UEFA competitions. In a North American context, teams and athletes have been shown to vary effort levels based on contest or contractual incentives. For example, there is quite strong empirical evidence in favor of tanking (Balsdon, Fong, & Thayer, 2007; Price et al., 2010; Taylor & Trogdon, 2002) and shirking or strategic behavior (Maxcy, Fort, & Krautman, 2002). In the context of European professional football, there is only limited research on effort and contest incentives. In a notable exception, Feddersen et al. (2012) analyze the impact of contest incentives on match performance in 10 European leagues. In this first approach to analyze the impact of financial incentives on team behavior, the authors study the relationship between contest incentives and match results. Through ordered logit regressions they show that intraseasonal races impact match results. No such effect is found for UEFA Champions League games within the next five days of national league matches. Green et al. (2015) show that football club managers engage in strategic behavior by adjusting team investments dependent on the number of UEFA Champions League qualifying places per league.

The primary objective of this paper is to assist national league regulators and the UEFA in designing their league schedules. Therefore we will analyze the impact of financial incentives on managerial effort in European professional football by extending the model of Feddersen et al. (2012) in several ways. First, this study intends to explicitly consider managerial effort by analyzing the market values of the starting squads. Our hypothesis is that financial incentives drive effort. Financial incentives are analyzed both in national leagues and UEFA competitions. Furthermore, we test for an interaction between national and UEFA competitions by including dummies for UEFA matches in the next five days after a Bundesliga match. Also we add further control variables compared to Feddersen et al. (2012) and differentiate between successful qualification for the UEFA Champions League (UCL) and UEFA Europa League (UEL). We base our analyses on two samples. The first sample covers an unbalanced panel of match-day observations from German contenders in the Bundesliga from 2006 to 2015. The second sample includes matches in the UCL group stage during the same period. The results suggest that football clubs reduce effort if intra-seasonal objectives have been clinched (Bundesliga), the UCL knock-out round has been clinched (UCL), or if UEFA games follow within the next five days (Bundesliga). National league organizers may intervene by adjusting contest formats and incentives.

Literature Review

European Football and the Rise of the UEFA Champions League

According to UEFA's European Club Licensing Benchmarking Report for the Financial Year 2014, European club football is an estimated [euro]15.9b market with an average growth rate of 9.5% over the last 19 years (Perry, 2015). In the last decade, the UCL has emerged as a key European contest with prize money and audiences exceeding those of national competitions (Peeters, 2011). In the 2014-15 season, 77 clubs from 53 country associations competed in the UCL. Each national league association can send up to four clubs to the competition, depending on the success of the association in the UEFA five-year ranking. Lhe estimated participation, TV pool, and performance-related revenues allotted to the 32 teams having entered at least the group stage amounted to [euro]988 million (m) in the 2014-15 season, with a minimum club premium of [euro]13.0m and a maximum club premium of [euro]89.1m generated by Juventus Football Club (UEFA, 2015a). The second-most important European club competition is the UEL, with 195 clubs from 54 country associations competing in the 2014-15 season. In total, [euro]239.8m have been distributed by the UEFA to the participants, which compares to about 24% of the UCL bonuses. Sevilla FC, the winner of that season's competition, generated around [euro]19.1m, which relates to 21.4% of the UCL winner (UEFA, 2015b).

Contest Theory, Effort, and Incentives

The theory of sporting contests has been developed first by Rottenberg (1956), who analyzes the impact of player reservation and free agency systems on the distribution of baseball players to teams. Fort and Quirk (1995) review the incentive effects of sports leagues in a model that assumes profit maximization by teams, income maximization by players, and market equilibrium outcomes. They argue that teams choose the talent level in order to maximize profits. Lazear and Rosen (1981) formalize the theory of contests in which teams are rewarded for their performance through external prices. They suggest that the competition for rewards leads contestants to display effort, which, in turn, has a positive effect on the contestant's output. More generally, contest success functions translate efforts into chances of winning (Muehlheusser, 2006). A key influencer of contest success functions is the incentive structure. Specifically, effort has been hypothesized to depend positively upon the price differential, negatively upon the marginal cost of effort, and finally on the effect of increasing effort supplied on the probability of winning (Feddersen, Humphreys, & Soebbing, 2012). Also, the probability of winning and gaining the larger price depends positively on the team's own actions and negatively on the competitor's actions (Knoeber & Thurman, 1994).

Szymanski (2003) explicitly raises the question of what should be the optimal structure of prizes in a specific tournament, and relates the investment or effort of a team to the design of the sporting contest. Given the contest technology and considering that the objective of the contest organizer is to design contests in such a way that participants contribute investments or effort, the paper concludes that there is an optimal reward structure contingent on the distribution of the abilities and willingness to pay of contestants. Similarly, Peeters (2011) argues that contest formats such as leagues, playoffs, or knock-out competitions have an impact on the revenue distribution among teams. For example, heavy revenue sharing mechanisms may reduce team efforts to win, while high performance-based price structures may increase motivation to win. Deutscher and Frick (2015) analyze incentive effects in heterogeneous tournaments. By pursuing a match-day level analysis of effort exertion in the NBA, they show that heterogeneity among competitors reduces effort. They conclude that league organizers should adjust the prize distribution by increasing the spread or by skewing the price structure towards the top teams. Azmat and Moller (2009) argue that not only contestants themselves, but also different contests may compete with each other to attract the participation and effort levels by contestants.

Empirical applications of contest theory cover a wide range of sporting competitions, including baseball, basketball, football, road racing, stock car racing, bowling, horse racing, tennis, and golf (see Frick, 2003; Frick & Simmons, 2008). From the perspective of the contestants, most studies focus on one of three phenomena: tanking, shirking/strategic behavior, and the impact of pay dispersion on performance.

"Tanking" is a phenomenon in which teams or athletes face incentives to lose. Apart from a few exceptions, there is quite strong empirical evidence in favor of tanking in several league systems, including the NBA (Price et al., 2010; Soebbing & Humphreys, 2013; Soebbing et al., 2013; Taylor & Trogdon, 2002), NCAA college basketball (Balsdon, Fong, & Thayer, 2007), American football, ice hockey, badminton, (European) football, and cricket (Balsdon, Fong, & Thayer, 2007; Preston & Szymanski, 2003), while no evidence is found in the Australian football league (Borland, Chieu, & Macdonald, 2009).

A closely related concept to tanking that is often used interchangeably is "shirking" (e.g., Price et al., 2010). In a more narrow sense, however, this concept refers to behavior of individual athletes that do not invest their full effort in the first year after signing a new long-term labor contract. Maxcy et al. (2002) differentiate between ex ante "strategic behavior" as defined by increasing performance prior to contract negotiations and ex post "shirking" as defined by slacking off after contract signing. Especially long-term labor contracts have been argued to create incentives to shirk, because the individual benefits less if he is protected under a contract (Marburger, 2003). Empirical evidence for shirking is quite mixed, and the quality of analyses critically depends on the model chosen. While some studies find supporting evidence for shirking in Major League Baseball (Krautmann & Solow, 2009; Lehn, 1982; Scoggins, 1993; Stiroh, 2007), others do not (Krautmann, 1990; Maxcy, 1997). Maxcy et al. (2002) find indications of ex ante strategic behavior of players, but not of ex post shirking after signing new contracts. Krautmann and Donley (2009) test the shirking hypothesis using different models, and find no evidence of shirking when using player performance, but find evidence of shirking when repeating the analysis based on the marginal revenue product produced by the player for the team. Similarly, Berri and Krautmann (2006) analyze potential shirking in the National Basketball Association, and find weak evidence when using the standard player productivity measure, while a marginal productivity measure more consistent with economists' definition of marginal product generates no clear results. Marburger (2003) analyzes the impact of the change from reserve era to free agency system, but finds no clear evidence of a systematic effect on shirking.

Finally, another research stream analyzing the influence of incentives refers to the impact of pay dispersion and/or superstars on performance of professional teams. There are two competing hypotheses on the impact of pay dispersion (Franck & Nuesch, 2011; Mondello & Maxcy, 2009). According to the efficiency wage theory (Debrock et al., 2004), a higher salary dispersion creates a meritocratic work environment where superior performance is rewarded with additional pay. On the other hand, a lower salary dispersion may improve performance due to better teamwork (Lazear, 1989). Empirical evidence, thus far, is mixed. For example, Franck and Nuesch (2011) analyze non-linear effects of pay dispersion in professional soccer and find empirical evidence that teams perform best when there is very high or very low wage dispersion. Frick et al. (2003) find that a higher degree of pay inequality increases the performance of hockey and basketball teams but decreases the performance of American football and baseball teams. Focusing on the impact of superstars on individual effort, Brown (2011) analyzes effort of professional golfers dependent on the participation of Tiger Woods. She finds that large differences in skill as present during Woods's strong periods decrease competitors' effort.

Effort and Incentives in European Football

To the best knowledge of the authors, there is only limited research on effort and incentives in European professional football. While Europe's premium football club competitions typically do not offer incentives to lose as some North American leagues do, European football teams may reduce effort in order to generate fitness savings. Wicker et al. (2013) analyze the effort of individual players in German Bundesliga, and find weak evidence that the interaction between intensive runs and tackling rate has a positive effect on player market values. Weimar and Wicker (2017) suggest that there may be a Moneyball phenomenon in soccer in a way that the soccer labor market may undervalue running distance. Garicano and Palacios-Huerta (2006) evaluate the effort level of teams based on the number of strikers, midfielders, and defenders, number of goals scored, number of goals received, as well as number of fouls and yellow and red cards received. While the first measures are confirmed to be positive effort measures, the latter is shown to have a negative influence on success. However, effort measures of individual players remain controversial, since both runs and yellow or red cards may be a sign of inferior team quality rather than superior effort levels. Also, the number of strikers, midfielders, and defenders used may be difficult to evaluate as, for example, a differentiation between offensive midfielders and strikers may often be dubious. Van Ours and Tujil (2011) analyze the effect of team effort on goal scoring in the final minutes of national team matches, and find that some teams are more likely to score than others when it is important to do so. Page and Page (2009) study situations in the group phases of UEFA competitions from 1992 to 2009 in which clubs will definitely finish either first or last before the last game. They show that such situations regularly decrease performance of the top team and increase performance of the bottom team.

In the first analysis of financial incentives in European professional football, Feddersen et al. (2012) study the impact of financial incentives on match performance in 10 European domestic football leagues over 11 seasons. Based on the original model by Taylor and Trogdon (2002), the authors use match level data and control for key factors that could influence match outcomes other than financial incentives. Particularly, the study controls for home field advantage and winning percentage of both teams prior to the game. By performing ordered logit regressions with a rich data set of 31,746 matches, the authors find that intraseasonal races impact match results. In an extended model, dummy variables are added to control for UCL games by both teams taking place in the next five days. No significant results are found for UCL games taking place in the next five days after the national league game.

In a notable study analyzing strategic behavior of football club managers, Green et al. (2015) study the impact of UCL qualifying slots in domestic leagues on talent investment. They find that particularly those clubs that have been close to qualifying for the UCL the previous season invest more into the team, but also clubs that did participate in the UCL the previous season do invest more.

Theoretical Model

This research tests the prevalence of strategic behavior in European professional football. For the matter of our study, strategic behavior is defined as an increase (or decrease) in managerial effort supplied due to financial incentives (disincentives). Financial incentives vary by competition and will be explained in the section on interaction of Bundesliga and UEFA competitions. Our theoretical model will explore the sensitivity of managerial effort to financial incentives. Two different samples will be used to analyze the impact of financial incentives in the Bundesliga, UCL group stage, and the interaction of Bundesliga and UEFA games.

Market Value of Squads as Measure of Managerial Effort

Our definition of managerial effort is based on the assumption that club managers may engage in strategic behavior to save efforts of key players. (1) Two recent examples may illustrate how club managers can intervene:

Intraseasonal races within a competition

On the April 25, 2015, Bayern Munich completed its 25th German Bundesliga title with a win against Hertha BSC Berlin. After this match-day, four further match-days were to follow. However, Pep Guardiola stated that the Bundesliga has finished for Bayern Munich. In fact, the coach gave a chance to a number of youth players in the following games. The next three games against Leverkusen (0:2), Augsburg (0:1), and Freiburg (1:2) were lost. Media and competitors complained about a distortion of competition (Der Tagesspiegel, 2015).

Interaction between competitions

In 2014, Chelsea London played with its "second 11" in the title-deciding match against Liverpool ahead of the Champions League semi-final return game against Atletico Madrid. At that time, Manchester City was competing with Liverpool for the Premier League title. Chelsea London had no ambitions within the Premier League anymore, and focused its effort on the Champions League. Jose Mourinho openly admitted that he intended to rest his star players, and that he had to follow the decision by club owner Roman Abramovich. (Fox Sports, 2014).

In the sport economics literature, the quality of teams is typically measured by means of salaries or transfer fees paid (e.g., Frick, 2007). On an individual player basis, these measures are not published for German teams. However, evidence exists that player market values are an accurate and up-to-date proxy (e.g., Wicker et al., 2013). In order to measure managerial effort, we will refer to the market value of the starting squad.

The Sensitivity of Managerial Effort to Financial Incentives

Bundesliga--Intraseasonal races

The German Bundesliga is the first German professional football division consisting of 18 teams. Each team plays each of the other teams twice. The team in first place at the end of the season wins the championship. As of the 2014-15 season, the teams finishing second and third qualify directly for the UCL, while the fourth team plays in a qualification game for the UCL. The fifth and sixth teams qualify directly for the UEL, while the seventh team enters the third UEL qualifying round for the UEL. The last two teams are directly relegated, while the third-to-last team plays a relegation game against the third team of the second division. This contest format relates to several financial incentives, as the championship and UEFA competitions imply incremental revenues, and relegation implies detrimental revenues the next season. When teams manage to clinch these intraseasonal races several match-days before the end of the season, financial incentives are reduced for the remaining matches. On the other hand, if teams have clinched UEL or UCL qualification, but are not yet eliminated from the UCL qualification or the championship race, respectively, teams face incentives to show additional effort. It is noteworthy that determinants of expected and actual match outcome differ. While expected match outcome of Bundesliga games has been shown to be significantly influenced by the difference in standing between team and opponent, the difference in competitions played, the difference in points earned over the last five matches, the differences in standing last season, and home advantage, actual match outcome seems to be only influenced by the last two factors (von Hanau, Wicker, & Soebbing, 2015).

Hypothesis 1: Bundesliga managers adjust team effort in the Bundesliga as a response to contest incentives in intra-seasonal races.

UEFA Champions League

The UCL group phase allows for early qualification for knock-out games. As of 2014-15, the UCL is played in eight groups of four teams. Each team plays each of the other teams twice. The first two teams qualify for the next UCL round, while the third place teams continue to play in the UEL. The last 16 teams continue to play in knock-out games. Teams that have clinched qualification for the UCL knock-out round face reduced incentives for the remaining group stage games. Teams that are eliminated from both the UCL and UEL (i.e., will finish last in their group with certainty) also face reduced incentives in the remaining matches. On the other hand, Page and Page (2009) argue that these teams may increase performance due to social pressure or pride.

Hypothesis 2: Bundesliga managers adjust team effort as a response to contest incentives in the UCL group phase.

Bundesliga--Interaction with UEFA Competitions

German football clubs participating in UEFA competitions generate direct revenues both from the Bundesliga and from the UEFA competition. In the Bundesliga, direct revenues include domestic and international broadcasting revenues distributed to the teams. Domestic TV revenues are based on the final ranks in the first two German divisions during the last four seasons. The last season is weighted the highest. Also international broadcasting revenues have been allocated based on the final Bundesliga ranks. (2) In the UEFA competitions, direct revenues include participation bonuses, performance bonuses, and the national TV market pool. Figure 1 shows the direct revenues of UCL contestants per club in Germany from 2005-06 to 2014-15. Since 2006, German UCL contenders generate at least half of their direct revenues from the UCL or international broadcasting revenues. In contrast, German UEL contestants generate up to 30% through the UEL and international broadcasting revenues. To sum up, German UEFA competition contestants are significantly incentivized by UEFA competition prize money as well as international broadcasting revenues.

[FIGURE 1 OMITTED]

Furthermore, the UEFA competitions also offer superior spreads of contest rewards between the first and last team within the competition. In 2015, the proportion of maximum to minimum direct revenues generated per club and competition amounts to 6.9 in the UCL and 11.1 in the UEL. The proportions compare to a ratio of 2.5 in the German Bundesliga. Also the first divisions in England, Italy, and France discriminate less between the top and bottom teams than the UEFA competitions. Only the Spanish Primera Division with its decentralized broadcasting scheme discriminates stronger.

Typically, Bundesliga matches take place on Fridays, Saturdays, and Sundays, while UEFA matches are played on Tuesdays, Wednesdays (UCL), and Thursdays (UEL). Thus, Bundesliga teams regularly play UEFA games within five days after the Bundesliga game.

Hypothesis 3: Bundesliga managers adjust team effort in the Bundesliga ahead of important UEFA games in the following five days (i.e., excluding group phase games with a lack of contest incentives).

Methodology

Data Source and Collection

This paper is based on two samples. The first sample for this study includes an unbalanced sample of 31 German Bundesliga clubs playing in the German first division "Bundesliga" from 2006 to 2015. Our database comprises all 3,060 matches played during the 10 seasons. Since every match is included twice to account for the home-field advantage, the full sample includes 6,120 club-match observations. To test for an interaction of Bundesliga games and UEFA games, this sample is amended by information on 576 UEFA matches (263 UCL matches and 313 UEL matches) following within five days from the Bundesliga match. To establish whether our results are generalizable to other data sources, we will compare our results from the Bundesliga sample with previous findings in the basic model by Feddersen et al. (2012), and analyze a second sample of 1,920 club-match observations from the UCL group stage.

Market values of the overall team, starting squads, reserve players, and substitutes are available from Transfermarkt from 2006 onwards (http://www.transfermarkt.de). The German-based website is owned by Axel Springer (51%) and Matthias Seidel (49%) and offers football data such as market values and transfer fees. Although the player values are based on expert estimates, the data have been found to be reliable proxies (Bryson, Frick, & Simmons, 2013). Further information related to the match (match day, time, results, home-field advantage) and team (league points, win percentage) have been collected from Kicker, another reliable database covering comprehensive football statistics (http://www.kicker.de).

Measures

Table 1 provides an overview of the key variables used in this study. In their standard model of shirking, Maxcy et al. (2002) derive effort levels from a comparison with past performance. Similarly, Wicker et al. (2013) add a reference point that controls for players' initial starting value. For the purpose of our study, we tested three different definitions of managerial effort. First, we performed our analysis with an absolute measure of the market value of the starting squad. Second, we used a relative measure that considered the market value of the starting squad relative to the overall team market value. Third, we determined the deviation of the market value of the starting squad from the average market value of the starting squad of that team year-to-date. No significant differences have been determined. Thus, managerial effort will be measured by the market value of the starting squad (t1_mv_start_sum) as most simple measure. Due to the right-skewed distribution, we will use the natural logarithm of the market value (Mincer, 1974).

Our definition of contest incentives extends previous research. Instead of the four variables controlling for contest incentives by Taylor and Trogdon (2002) and the 10 variables in the final model by Feddersen et al. (2012), we control for 14 variables due to the special characteristics of European football and an important distinction between UCL and UEL qualification. In the Bundesliga, we account for intraseasonal races for the championship, UCL qualification, UEL qualification, and relegation. If the team has been eliminated from all races (t1_elim), we expect a negative impact on effort. The same applies if a team has clinched the championship (t1_clinch_champ) or will be relegated with certainty (t1_clinch_rel). Further reductions of financial incentives include certain UCL qualification with simultaneous elimination from the championship race (t1_clinch_uclqual_elim_champ) and certain UEL qualification with simultaneous elimination from the UCL qualification race (t1_clinch_uelqual_elim_uclqual). On the other hand, certain UCL qualification without simultaneous elimination from the championship race (t1_clinch_uclqual_nelim_champ) and certain UEL qualification without simultaneous elimination from the UCL qualification race (t1_clinch_uel-qual_nelim_uclqual) provide financial incentives to put forward additional effort in the remaining games. We follow established sport economics literature studying the impact of financial incentives (e.g., Feddersen et al., 2012; Taylor & Trogdon, 2002) and mirror the financial incentive variables for team 2. However, results for the opponent variables should be interpreted carefully given the technical foundations that require them to take the exact values with opposite signs of the team 1 variables.

Our novel definition of contest incentives for UEFA competitions is based on the unique features of UCL and UEL group stages. If a team has clinched the UEFA knockout phase (t1_uefa_clinch_ko_phase), financial incentives are reduced for the remaining games. This is also the case if a team has been eliminated from the UEFA competition (t1_uefa_elim). Again, the financial incentives of the opponent are expected to have the reverse impact on team effort.

Since the market value of the starting squad in Bundesliga and UEFA matches might be influenced by several factors other than contest incentives, we need to add certain control variables. Similarly to the Maxcy et al. (2002) standard model of shirking, we control for team quality, home-field advantage, and a random component. While Maxcy et al. (2002) analyzed player effort dependent on the time of contract negotiations, our focus is on managerial effort. This measure is subject to public reference points by the competitor (Kahneman & Tversky, 1979). The winning percentage of both teams before the game (t1_winpct, t2_winpct) controls for the short- to medium-term quality. Previous research has shown that the opponent's winning percentage has a negative effect on winning (Taylor & Trogdon, 2002; Feddersen, Humphreys, & Soebbing, 2012). We expect the team's winning percentage to have a positive effect, and the opponent's winning percentage to have a negative effect on the market value of the starting squad. However, winning percentages may not always reflect the overall quality and long-term potential of a team. The presence of team market values offers additional public reference points that may support managerial decisions and influence match outcomes. Thus, we account for the overall market value of both teams (t1_mv, t2_mv). Due to its right-skewness, we again use the natural logarithm of the variable. Additionally, previous research has shown a positive impact of home-field advantage on match performance (Carmichael & Thomas, 2005; Forrest et al., 2005; Goddard & Thomas, 2006; von Hanau, Wicker, & Soebbing, 2015). Next to the psychological benefit of playing at home, we also expect managers to be influenced in their decision on the starting squad. A dummy variable takes the value of 1 if a team enjoys a home-field advantage and 0 otherwise (t1_home).

Our measures accounting for the interaction of Bundesliga and UEFA competitions extend the work by Feddersen et al. (2012). In addition to dummies controlling for UCL games in the next five days (t1_ucl_5d), we also account for UEL games in the five days after the Bundesliga game (t1_uel_5d). As a proxy for the match importance of the UEFA game, separate variables will account for UCL (t1_ucl_ko_5d) and UEL knockout games within five days (t1_uel_ko_5d). A further important addition to Feddersen et al. (2012) are control variables for active contenders in the UCL (t1_ucl_cont) and UEL (t1_uel_cont). Due to the exposure to higher financial incentives and because of the higher workload of these contenders, both variables are expected to have a positive influence on the market value of the starting squad.

Data Analysis

Regression analyses are performed to test the impact of financial incentives on managerial effort. The first and second hypotheses that analyze the sensitivity of managerial effort to financial incentives in the German Bundesliga and UCL group stage are tested using fixed-effects panel regressions controlling for team quality and home-field advantage. In the Bundesliga, we control for short- and long-term team quality by adding win percentages and market values of both teams. Due to the mirrored sample and match schedule and because of using cluster-robust standard errors on match level, we expect exactly contrary coefficients for all variables for the two teams except for the home-field advantage (Feddersen, Humphreys, & Soebbing, 2012). We also control for match-day and club-year fixed effects to account for the heterogeneity of teams and potential match-day effects toward the end of the season.

t1_mv_start_[sum.sub.ijk] = f (t1_[elim.sub.ijk], t2_[elim.sub.ijk], t1_clinch_[champ.sub.ijk], t2_clinch_[champ.sub.ijk], t1_clinch_uclqual_elim_[champ.sub.ijk], t2_clinch_uclqual_elim_[champ.sub.ijk], t1_clinch_ uclqual_nelim_[champ.sub.ijk], t2_clinch_uclqual_nelim_[champ.sub.ijk], t1_clinch_uelqual_ elim_[uclqual.sub.ijk], t2_clinch_uelqual_elim_[uclqual.sub.ijk], t1_clinch_uelqual_nelim_[uclqual.sub.ijk], t2_clinch_uelqual_nelim_[uclqual.sub.ijk], t1 clinch_[rel.sub.ijk], t2_clinch_[rel.sub.ijk], t1_[mv.sub.ijk], t2_[mv.sub.ijk], t1_[winpct.sub.ijk], t2_[winpct.sub.ijk] ,t1 [home.sub.ijk]), (1)

whereby i = team, j = match, k = season.

In the UCL group stage sample, we also control for the market value of both teams and the home-field advantage. Due to the short-term nature of only six games, the win percentage would not be meaningful.

t1_mv_start_[sum.sub.ijk] = f (t1_uefa_clinch_ko_[phase.sub.ijk], t2_uefa_clinch_ko_[phase.sub.ijk], t1_uefa_[elim.sub.ijk], t2_uefa_[elim.sub.ijk], t1_[mv.sub.ijk], t2_[mv.sub.ijk], t1_[home.sub.ijk]) (2)

Finally, the third hypothesis analyzes if managerial effort in Bundesliga games is lower if the games are followed by important UEFA competition games in the next five days. Therefore we extend the analysis of Hypothesis 1 by adding dummy variables for UCL games and UEL games in the next five days. We exclude those UEFA group stage games in which the team has already clinched the knock-out round. While these games are still subject to certain match-day bonuses by the UEFA, teams are not incentivized anymore to put forward additional effort to qualify for the next round.

t1_mv_start_[sum.sub.ijk] = f (t1_[elim.sub.ijk], t2_[elim.sub.ijk], t1_clinch_[champ.sub.ijk], t2_clinch_[champ.sub.ijk], t1_clinch_uclqual_elim_[champ.sub.ijk], t2_clinch_uclqual_elim_[champ.sub.ijk], t1_clinch_ uclqual_nelim_[champ.sub.ijk], t2_clinch_uclqual_nelim_[champ.sub.ijk], t1_clinch_uelqual_ elim_[uclqual.sub.ijk], t2_clinch_uelqual_elim_[uclqual.sub.ijk], t1_clinch_uelqual_nelim_[uclqual.sub.ijk], t2_clinch_uelqual_nelim_[uclqual.sub.ijk], t1_clinch_[rel.sub.ijk],_t2 clinch_[rel.sub.ijk],_t1_ucl_[5d.sub.ijk], t2_ucl_[5d.sub.ijk], t1_uel_[5d.sub.ijk], t2_uel_[5d.sub.ijk], t1_[mv.sub.ijk], t2_[mv.sub.ijk], t1_[winpct.sub.ijk], t2_[winpct.sub.ijk], t1_ucl_[cont.sub.ijk], t2_ucl_[cont.sub.ijk], t1_uel_[cont.sub.ijk], t2_uel_[cont.sub.ijk], t1_[home.sub.ijk]) (3)

Moreover, team quality and financial incentives increase toward the end of the UEFA competitions. Thus, we also run separate regressions with dummy variables for UCL (t1_ucl_ko_phase_5d) and UEL knock-out games (t1_uel_ko_phase_5d) only.

t1_mv_start_[sum.sub.ijk] = f (t1_[elim.sub.ijk], t2_[elim.sub.ijk], t1_clinch_[champ.sub.ijk], t2_clinch_[champ.sub.ijk], t1_clinch_uclqual_elim_[champ.sub.ijk], t2_clinch_uclqual_elim_[champ.sub.ijk], t1_clinch_ uclqual_nelim_[champ.sub.ijk], t2_clinch_uclqual_nelim_[champ.sub.ijk], t1_clinch_uelqual_ elim_[uclqual.sub.ijk], t2_clinch_uelqual_elim_[uclqual.sub.ijk], t1_clinch_uelqual_nelim_[uclqual.sub.ijk], t2_clinch_uelqual_nelim_[uclqual.sub.ijk], t1_clinch_[rel.sub.ijk], t2_clinch_[rel.sub.ijk], t1_ucl_ko_[5d.sub.ijk], t2_ucl_ko_[5d.sub.ijk], t1_uel_ko_phase_[5d.sub.ijk], t2_uef_ko_phase_[5d.sub.ijk], t1_[mv.sub.ijk], t2_[mv.sub.ijk], t1_winpctijk,_t2_ [winpct.sub.ijk], t1_ucl_[cont.sub.ijk], t2_ucl_[cont.sub.ijk], t1_uel_[cont.sub.ijk], t2_uel_[cont.sub.ijk], t1_[home.sub.ijk],) (4)

Results and Discussion

Descriptive Statistics

Table 2 contains the summary statistics of the Bundesliga sample. The sample includes 6,120 club-match observations. Since contest incentives and win percentages are determined prior to the match day, these variables have no observations for the first match day. This results in 5,940 observations that can be used for the regression analysis. The average market value of a starting squad amounts to [euro]55.3m. However, there is a large spread between the minimum market value of a starting squad of [euro]7.8m and the maximum market value of a starting squad of [euro]387.5m. The share of observations with teams that have been eliminated from or have clinched intra-seasonal races is quite small, since these races are typically decided in the last few match days only. About 2.14% of teams have been eliminated from all races at the time of observation. Only 0.42% of teams have clinched the championship, and 0.37% will be certainly relegated. Teams that have clinched the UCL qualification are more often not yet eliminated from the championship race (0.29%) than they are (0.08%). Similarly, teams that clinched UEL qualification are about four times as often not yet eliminated from the UCL race (2.16%) than they are (0.59%). This supports the importance of our extension of the basic model by Teddersen et al. (2012) by differentiating between the UCL and UEL qualification. The overall team market value including reserve players and missing players averages about [euro]93.1m. That is, the average starting squad is worth about 60% of the overall team.

Table 3 shows the summary statistics of the UCL group stage sample. The average market value of starting squads amounts to [euro]104.2m. In about 5.5% of observations, teams have clinched the knock-out round already before the game. This compares to 1.8% of observations of teams that have been eliminated before the game. On average, overall squads are worth around [euro]182m.

The summary statistics on the interaction of the Bundesliga and UEFA competitions are also summarized in Table 2. In about a quarter of Bundesliga match observations, teams are still active either in the UCL (11.1%) or the UEL (13.6%). In 4.3% of the observations teams have a UCL game in the next five days, and in 1.3% of the observations this is a UCL knock-out game. This compares to 5.1% of the observations in which teams have a UEL game, and 1.4% of observations in which teams have a UEL knock-out game in the next five days.

Panel Regression

Table 4 contains the regression results for the estimated impact of financial incentives on managerial effort in the German Bundesliga. Due to the symmetrical nature of the sample and estimation method as well as the use of cluster-robust standard errors on match level, the variables for the two teams have the expected opposite signs. (3) The control variables show the expected results. Win percentages and market values of the team in question have a positive impact and those of the opponent a negative and significant impact on the market value of the starting squad. We also find that team market values of starting squads are higher when playing at home. However, this result is only robust at the 10% level. On average, the market value of the starting squad is about half a million Euros higher when playing at home. This result might indicate that managers are subject to additional pressure to win when playing in their home stadium.

As expected, elimination from all intraseasonal races has a negative and significant impact, while elimination of the opponent from all races has a positive and significant impact (p<0.01). If a team is eliminated from all competitions, managers give a rest to players with a market value of about [euro]6.9m compared to the average game. We also find a strongly negative and significant impact if the team has already clinched the championship (p<0.05). In this case, managers send a starting squad on the field that is on average [euro]30.7m less worth. If the team has clinched the UCL qualification, but cannot win the championship anymore, we find a negative but insignificant result. However, if the team has clinched UCL qualification and is not yet eliminated from the championship race, we find the expected positive though insignificant impact on the market value of the starting squad. The dummy variables representing teams that have clinched UEL qualification and are either eliminated or not yet eliminated from UCL qualification have the expected signs, but are not yet significant. Having clinched relegation has a negative and significant impact on managerial effort. On average, market values of starting squads are [euro]10.1m lower.

First, these results confirm the findings by Feddersen et al. (2012) in their European sample. External prices rewarding the performance of teams influence the investment or effort by teams (Tazear & Rosen, 1981; Szymanski, 2003). Second, our results show that managers reduce effort in situations when it is most clear that financial incentives are reduced. This is the case when a team has clinched the championship or relegation, or when it is eliminated from all races. Third, we find that managers also engage in strategic behavior by investing additional effort in the most important Bundesliga games. Specifically, this has been shown for the final title-deciding matches. Thus, the non-linear prize structure adopted by the German Bundesliga is effective in terms of eliciting additional effort by heterogeneous contestants. However, it is inefficient in terms of maintaining effort levels when intraseasonal races have been clinched (Deutscher & Frick, 2015).

Table 5 displays the key regression results for the UCL group stage sample. The controls for team quality and home-field advantage have the expected signs. Also, the contest incentive variables show the expected signs. Robust results are found if a team has already clinched the knock-out round (p<0.001). Then, managers rest players worth about [euro]44.9m. These results confirm our contest-theoretical predictions, and support the hypothesis that strategic behavior does also exist in UEFA competitions. Further, the results also strengthen the findings by Page and Page (2009), who found reduced performance of these top clubs that have already qualified for the knock-out round. In contrast, no significant results are found when the team has already been eliminated from the competition. This might be explained by the balancing effect of reduced financial incentives on the one hand, and social pressure by fans or pride of the teams to end their "UCL adventure" with a good game on the other hand (Page & Page, 2009).

Table 6 contains the regression results for the sensitivity of managerial effort in the Bundesliga to UEFA games in the next five days. Based on the findings from the UCL group stage sample, Bundesliga games followed by UEFA games of minor importance (i.e., UEFA group stage games when the team has already clinched the knock-out phase) are excluded from the analysis. Teams being active in the UCL or the UEL have significantly more valuable starting squads. If the Bundesliga game is followed by a UCL game in the next five days, starting squads have a significantly lower market value of about [euro]4.2m. If the Bundesliga game is followed by a UCL knock-out game in the next five days, this effect doubles to about [euro]8.5m. No significant results are found for the dummy variable controlling for UEL games in the next five days. This might show the lower perceived value of the UEL compared to the UCL; that is, national leagues and UEFA competitions interact with each other and need to compete for effort by teams (Azmat & Moller, 2009). Due to the higher contest rewards and price differentials in the UCL compared to the Bundesliga, effort supplied in the Bundesliga has been found to be lower ahead of UCL games. While previous papers hypothesized this relationship (Feddersen, Humphreys, & Soebbing, 2012), this paper provides novel empirical evidence supporting the phenomenon's existence.

Robustness Checks

To proof-check our results, several robustness checks have been performed. Our hypotheses analyzing the sensitivity of managerial effort to financial incentives have been tested with fixed-effects panel regressions controlling for team quality and home-field advantage. One could argue that injured players or players penalized due to red cards in previous games might impact market values. To test this idea, we performed fixed effects regressions with two alternative dependent variables. First, we used the "market value of the starting squad relative to the overall team market value." The sign of the contest incentive variables remains unchanged and the impact significant. Second, the results also remain unchanged when using the "deviation of the market value of the starting squad from the average market value of the starting squad year-to-date" as a dependent variable. Furthermore, we would expect relevant player injuries to lead to players fully missing the match rather than sitting on the bench. We tested if the addition of the market value of the bench players as control variable has an impact, and results are robust to this addition. Moreover one might argue that top clubs like Bayern Munich might use its second string when playing against much weaker teams rather than due to intraseasonal races. However, this effect is controlled for by using club-year fixed effects.

Conclusion

The purpose of this study was to analyze managerial effort, financial incentives, and strategic behavior in European professional football. The study has extended previous research analyzing the impact of intra-seasonal Bundesliga races on match results (Feddersen, Humphreys, & Soebbing, 2012). Specifically, our paper directly addresses the issue of managerial effort. We have empirically tested the sensitivity of managerial effort to intra-seasonal races in the Bundesliga, the UCL group phase, and the interaction of Bundesliga and UEFA competitions.

Summary of Findings

Our results show that contest incentives within and between national and European competitions influence managerial effort. Previous research has shown that various forms of strategic behavior exist in North American professional sports. Professional and college basketball teams have been shown to intentionally lose in conference tournaments motivated by either fitness savings or corruption (Balsdon, Fong, & Thayer, 2007; Taylor & Trogdon, 2002). Also, individual athletes have been argued to increase effort before and decrease effort after contract negotiations as a form of moral hazard typically referred to as shirking or strategic behavior (Maxcy, Fort, & Krautman, 2002). Football club managers have been shown to engage in strategic behavior by adjusting team investment based on the number of UCL qualifying spots in national leagues (Green, Lozano, & Simmons, 2015). Our paper analyzes a different form of strategic behavior by managers motivated by financial incentives on the one side and fitness savings for the team on the other side. Feddersen et al. (2012) developed an initial model that has shown an impact of contest incentives on match results without explicitly modelling team effort. Our research closes this gap and extends this work in several points. First, we are able to shed light on the "black box" between contest incentives and match outcomes. We show that contest incentives drive managerial effort. Second, we show that financial incentives are not only important within the Bundesliga, but also in UEFA competitions. In the UCL group stage, teams that have clinched the knock-out round seem to reduce managerial effort by resting its most valuable players. Third, our extended model controlling for UEFA games in the next five days has allowed us to confirm an interaction between the two competitions. Managers tend to rest players in Bundesliga games ahead of important UCL games. Thus, national leagues and the UEFA compete to attract full effort by clubs (Azmat & Moller, 2009). Also, Europe's national football regulators can (dis)incentivize effort levels by clubs by choosing the contest format and reward systems.

Managerial Implications

Football clubs reduce effort if intraseasonal objectives have been clinched (Bundesliga), the UCL knock-out round has been clinched (UCL), or important UCL games follow within the next five days (Bundesliga). This has important managerial implications both for national regulators and the UEFA. Both national leagues and UEFA competitions may be subject to adverse consequences, such as a loss in the integrity of the competition, and reduced interest by fans. To prevent such effects, league organizers may introduce incentive mechanisms that reward performance in the final matches. Such mechanisms may include, for example, a stronger differentiation between the value of individual ranks in the league table. In a different approach, Formula 1 racing introduced the "double points rule" for the last championship race in the 2014 season to increase excitement at the season's end. Furthermore, national regulators, such as the German football league (DFL), may intervene in order to increase the effort supplied by teams in national leagues compared to UEFA competitions. First, they may change contest rewards. While league organizers may be assumed to maximize overall contest rewards, they can typically decide in favor of a more or less discriminatory allocation of broadcasting revenues. At the moment, the Bundesliga features a much lower spread between direct revenues than the UEFA competitions. In fact, the Premier League is the only one of the Big Five leagues to discriminate less between the top and bottom team of the first division. Nevertheless, while the competition between national and UEFA contests may favor a more discriminatory approach, national league organizers may follow further objectives, such as competitive balance, that favor a less discriminatory approach (Szymanski, 2003). Second, league organizers may change the contest format. For example, the timing of match days might be adjusted to increase the time span towards the next UEFA game. Third, league organizers may adjust incentives or disincentives related to effort savings. For example, strategic behavior in the last match days of the Bundesliga season might be reduced if teams are incentivized by additional match-day bonuses.

Limitations

The application of two different samples--one national league and one UEFA competition sample--provide confidence that our results with respect to the impact of financial incentives are generalizable to European professional football. Nevertheless, we are subject to limited data availability since market value data provided by Transfermarkt are only available from 2005-06 onwards.

Also, managerial effort supplied in the last match days may be argued to be subject to corruption. Elaad et al. (2015) analyzed the impact of the corruption perceptions index on the outcome of competitions in 75 countries from 2001 to 2013. The results indicated that probabilities of a team that was at risk of relegation to achieve the required result are generally significantly higher when the country is more corrupt. However, according to Transparency International, Germany is considered one of the least corrupt countries. Also, the study confirmed that the chance of achieving the desired result is much lower in Germany (~58%) compared to other big leagues such as Spain (~83%) and Italy (~78%).

Recommendations for Future Research

Future research may extend the current model in several ways. First, this paper has focused on strategic behavior on a match-day level. Future studies may analyze the effects of intra-match strategies such as substitutions. Market values of substituted players are also available from Transfermarkt. Second, we have compared managerial effort in two different competitions, namely the Bundesliga and UEFA competitions. League designers would benefit from future research analyzing the effects of changes in payout systems on effort supplied by teams. Two examples include the reform of the UCL payout system in 1999-2000 and the introduction of the centralized broadcasting scheme in Italy from the 2010-11 season onward. Third, it may be tested whether betting markets are efficient in terms of anticipating effort supplied by teams. Fourth, our data set and methodology maybe extended to analyze the impact of pay dispersion on team performance. Finally, it would be interesting to analyze whether there is an adverse incentive effect similar to the superstar effect in golf (Brown, 2011), if a team plays against superstar teams such as Bayern Munich.

References

Azmat, G., & Moller, M. (2009). Competition among contests. The RAND Journal of Economics, 40, 743-768.

Balsdon, E., Fong, L., & Thayer, M. A. (2007). Corruption in college basketball? Evidence of tanking in postseason conference tournaments. Journal of Sports Economics, 8, 19-38.

Berri, D. J., & Krautmann, A. C. (2006). Shirking on the court: Testing for the incentive effects of guaranteed pay. Economic Inquiry, 44, 536-546.

Borland, J., Chicu, M., & Macdonald, R. D. (2009). Do teams always lose to win? Performance incentives and the player draft in the Australian Football League. Journal of Sports Economics, 10, 451-484.

Brown, J. (2011). Quitters never win: The (adverse) incentive effects of competing with superstars. Journal of Political Economy, 119, 982-1013.

Bryson, A., Frick, B., & Simmons, R. (2013). The returns to scarce talent: Footedness and player remuneration in European soccer. Journal of Sports Economics, 14, 606-628.

Carmichael, F., & Thomas, D. (2005). Home-field effect and team performance: Evidence from English Premiership football. Journal of Sports Economics, 6, 264-281.

DeBrock, L., Hendricks, W., & Koenker, R. (2004). Pay and performance. The impact of salary distribution on firm-level outcomes in baseball. Journal of Sports Economics, 5, 243-261.

Der Tagesspiegel. (2015, May 17). Kritik am FC Bayern Munchen: Wettbewerbsverzerrung - schwere Vorwurfe gegen den FC Bayern [Criticism on FC Bayern Munich: Distortion of competition--Severe claims against FC Bayern]. Der Tagesspiegel. Retrieved from http://www.tagesspiegel.de/sport/kritik-am-fc-bayern-muenchen-wettbewerbsverzerrung-schwere-vorwuerfe-gegen-den-fc-bayern/11787012.html

Deutscher, C, & Frick, B. (2015). The incentive effects of prize spreads in heterogeneous tournaments: Empirical findings from the NBA. Working paper, Department of Management, University of Paderborn.

Elaad, G., Kantor, J., & Krumer, A. (2015). Does the corruption perceptions index predict the outcome of competitions? Evidence from a natural experiment around the world. Retrieved from http://econ.biu.ac.il/files/economics/seminars/ssrn-id2479747.pdf

Feddersen, A., Humphreys, B. R., & Soebbing, B. P. (2012). Contest incentives in European football. Working paper No. 2012-13. Retrieved from http://www.ualberta.ca/~econwps/2012/wp2012-13.pdf

Forrest, D., Beaumont, J., Goddard, J., & Simmons, R. (2005). Home advantage and the debate about competitive balance in professional sports leagues. Journal of Sports Sciences, 23, 439-445.

Fort, R., & Quirk, J. (1995). Cross-subsidization, incentives, and outcomes in professional team sports leagues. Journal of Economic Literature, 33, 1265-1299.

Fox Sports. (2014, April 23). Paper talk: Jose Mourinho in amazing tanking admission. Retrieved from http://www.foxsports.com.au/football/paper-talk-jose-mourinho-in-am azing-tanking-admission-david-moyes-furious-about-sacking/news-story/1af7e02dee3d3a071c-6b5627e20e556f?sv=6884a0ac66d03cf2a6e4052d57f05df6

Franck, E., & Nuesch, S. (2003). The effect of wage dispersion on team outcome and the way team outcome is produced. Applied Economics, 43, 3037-3049.

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., Prinz, J., & Winkelmann, K. (2003). Pay inequalities and team performance. Empirical evidence from the North American major leagues. International Journal of Manpower, 24, 472-488.

Frick, B., & Simmons, R. (2008). The allocation of rewards in athletic contests. In B. R. Humphreys & D. R. Howard (Eds.), The business of sports: Economic perspectives on sport (Vol. 2, pp. 1-24). Westport, CT: Praeger Publishers.

Garicano, L., & Palacios-Huerta, I. (2006). Sabotage in tournaments: Making the beautiful game a bit less beautiful. CEPR discussion paper No. 5231. Retrieved from http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.619.666&rep=repl&type=pdf

Goddard, J., & Thomas, S. (2006). The efficiency of the UK fixed-odds betting market for Euro 2004. International Journal of Sport Finance, 1, 21-32.

Green, C., Lozano, F., & Simmons, R. (2015). Rank-order tournaments, probability of winning and investing in talent: Evidence from champions' league qualifying rules. National Institute Economic Review, 232, R30-R40.

Hanau, T. von, Wicker, P., & Soebbing, B. P. (2015). Determinants of expected vs. actual match outcome: An examination of the German Bundesliga. Soccer & Society, 16, 63-75.

Kahneman, D., & Tversky, A. (1979). Prospect theory: An analysis of decision under risk. Econometrica, 47, 263-292.

Knoeber, C. R., & Thurman, W N. (1994). Testing the theory of tournaments: An empirical analysis of broiler production. Journal of Labor Economics, 12, 155-179.

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

Krautmann, A. C., & Donley, T. D. (2009). Shirking in Major League Baseball revisited. Journal of Sports Economics, 10, 292-304.

Krautmann, A. C., & Solow, J. L. (2009). The dynamics of performance over the duration of Major League Baseball long-term contracts. Journal of Sports Economics, 10, 6-22.

Lazear, E. P., & Rosen, S. (1981). Rank-order tournaments as optimum labor contracts. Journal of Political Economy, 89, 841-864.

Lazear, E. P. (1989). Pay equality and industrial politics. Journal of Political Economy, 97, 561-580.

Lehn, K. (1982). Property rights, risk sharing, and player disability in Major League Baseball. Journal of Law and Economics, 25, 343-366.

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? In W. E. Hendricks (Ed.), Advances in the economics of sport (pp. 157-176). Greenwich, CT/London, UK: JAI Press.

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

Mincer, J. (1974). Schooling, experience, and earnings. New York, NY: Columbia University Press.

Mondello, M., & Maxcy, J. (2009). The impact of salary dispersion and performance bonuses in NFL organizations. Management Decision, 47, 110-123.

Muehlheusser, G. (2006). Implications from the theory of contests for modelling and designing sports competitions. In W. Andreff & S. Szymanski (Eds.), Elgar original reference. Handbook on the economics of sport (pp. 342-348). Cheltenham, UK: Edward Elgar.

Page, L., & Page, K. (2009). Stakes and motivation in tournaments: Playing when there is nothing to play for but pride. Economic Analysis and Policy, 39, 455-464.

Peeters, T. (2011). Broadcast rights and competitive balance in European soccer. International Journal of Sport Finance, 6, 23-39.

Perry, S. (2015). The European club licensing benchmarking report: Financial year 2014. Nyon, Switzerland: UEFA.

Preston, I., & Szymanski, S. (2003). Cheating in contests. Oxford Review of Economic Policy, 19, 612-624.

Price, J., Soebbing, B. P., Berri, D., & Humphreys, B. R. (2010). Tournament incentives, league policy, and NBA team performance revisited. Journal of Sports Economics, 11, 117-135.

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

Scoggins, J. F. (1993). Shirking or stochastic productivity in Major League Baseball: Comment. Southern Economic Journal, 60, 239-240.

Soebbing, B. P., & Humphreys, B. R. (2013). Do gamblers think that teams tank? Evidence from the NBA. Contemporary Economic Policy, 31, 301-313.

Soebbing, B. P., Humphreys, B. R., & Mason, D. S. (2013). Exploring incentives to lose in professional team sports: Do conference games matter? International Journal of Sport Finance, 8, 192-207.

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

Szymanski, S. (2003). The economic design of sporting contests. Journal of Economic Literature, 41, 1137-1187.

Taylor, B. A., & Trogdon, J. G (2002). Losing to win: Tournament incentives in the National Basketball Association. Journal of Labor Economics, 20, 23-41.

UEFA. (2015). UEFA Champions League - Distribution to clubs 2014/15. Retrieved from http://www.uefa.com/MultimediaFiles/Download/competitions/General/02/29/45/25/2294525_DOWNLOAD.pdf

UEFA. (2015). UEFA Europa League - Distribution to clubs 2014/15. Retrieved from http://www.uefa.com/MultimediaFiles/Download/competitions/General/02/29/45/29/2294529_DOWNLOAD.pdf

UEFA. (2014). European Club footballing landscape 2014: Benchmarking report: Licensed to Thrill. Nyon, Switzerland: UEFA.

Van Ours, J. C, & Tuijl, M. A. (2011). Country-specific goal-scoring in the 'dying seconds' of international football matches. International Journal of Sport Finance, 6, 138-154.

Weimar, D., & Wicker, P. (2017). Moneyball revisited: Effort and team performance in professional soccer. Journal of Sports Economics, 18, 140-161.

Wicker, P., Prinz, J., Weimar, D., Deutscher, G, & Upmann, T. (2013). No pain, no gain? Effort and productivity in professional soccer. International Journal of Sport Finance, 8, 124-139.

Marc Rohde (1) and Christoph Breuer (1)

(1) German Sport University Cologne

Marc Rohde is a PhD student in the Department of Sport Economics and Sport Management. His research interests include sport economics, sport finance, and the managerial influence on financial and sporting success of professional clubs.

Christoph Breuer, PhD, is a professor in the Department of Sport Economics and Sport Management. His research interests include sport organizational economics and economics of sponsoring.

Endnotes

(1) In Germany, industry experts tend to differentiate between coaches and managers based on a typical separation of job roles. In an international context, however, the name "manager" is more prevalent (e.g., Frick & Simmons, 2008).

(2) In the 2015-16 and 2016-17 seasons, the basic payment to all Bundesliga teams will be increased from [euro]27m to [euro]45m. Then, each Bundesliga team will receive a basic payment of [euro]2.5m. The incremental revenues will be allocated based on the UEFA club coefficients that evaluates teams based on their performance in UEFA competitions during the recent five seasons.

(3) The authors of this paper are convinced of the choice of cluster-robust standard errors on match level. Regression results with cluster-robust standard errors on club-year level confirm the key results and are available upon request.

Authors' Note

The authors would like to thank Bernd Frick and the participants of the X Gijon Conference on Sport Economics for their valuable feedback on a previous version of this paper. Table 1. Overview of Variables Variable Variable Description type t1_mv_start_sum Team 1 - Market value of starting Managerial squad ([euro]m) effort Contest t1_elim Team 1--Eliminated from both races incentives (1=yes; 0=no) t1_clinch_champ Team 1--Clinched championship race (1=yes; 0=no) t1_clinch_uclqual_ Team 1--Clinched UCL qualification, elim_champ eliminated from championship race (1=yes; 0=no) t1_clinch_uclqual_ Team 1--Clinched UCL qualification, nelim_champ not eliminated from championship race (1=yes; 0=no) t1_clinch_uelqual_ Team 1--Clinched UEL qualification, elim_ucl eliminated from UCL qualification race (1=yes; 0=no) t1_clinch_uelqual_ Team 1--Clinched UEL qualification, nelim_uclqual not eliminated from UCL qualification race (1=yes; 0=no) t1_clinch_rel Team 1--Clinched relegation (1=yes; 0=no) t1_uefa_clinch_ Team 1--Clinched UEFA knock-out ko_phase phase (1=yes; 0=no) t1_uefa_elim Team 1--Eliminated from UEFA competition (1=yes; 0=no) t1_ucl_5d Team 1--UEFA Champions League game within the next five days t1_uel_5d Team 1--UEFA Europa League game within the next five days t1_ucl_ko_ Team 1--UEFA Champions League phase_5d knock-out game within next five days t1_uel_ko_ Team 1--UEFA Europa League knockout phase 5d game within next five days t1_mv (1) Team 1--Market value ([euro]m) t1_winpct (1) Team 1--Win percentage in season ytd (%) t1_ucl_cont Team 1--Contender in UEFA Champions League in respective season t1_uel_cont Team 1--Contender in EUFA Europa League in respective season t1_home Team 1--Home team (1=yes; 0=no) Variable Scale type Metric Managerial effort Contest Dummy incentives Dummy Dummy Dummy Dummy Dummy Dummy Dummy Dummy Dummy Dummy Dummy Dummy Metric Metric Dummy Dummy Dummy Note: (1) Variables for team 2 built in the same way as those for team 1 Table 2. Summary Statistics--Bundesliga Sample Variable type Variable N Mean Managerial t1_mv_start_sum 6,120 55.34 effort Contest t1 elim 5,940 2.14% incentives t1_clinch_champ 5,940 0.42% t1_clinch_uclqual_elim_champ 5,940 0.08% t1_clinch_uclqual_nelim_champ 5,940 0.29% t1_clinch_uelqual_elim_uclqual 5,940 0.59% t1_clinch_uelqual_nelim_uclqual 5,940 2.16% t1 clinch rel 5,940 0.37% t1 ucl 5d 6,120 4.30% t1 uel 5d 6,120 5.11% t1 ucl ko 6,120 1.31% t1_uel_ko 6,120 1.44% Control t1 mv 6,120 93.09 variables t1_winpct 5,940 0.50 t1 ucl cont 6,120 11.0% t1 uel cont 6,120 13.6% t1_home 6,120 0.50 N 6,120 Variable type Std. Dev. Min Max Managerial 49.49 7.80 387.50 effort Contest - 0 1 incentives - 0 1 - 0 1 - 0 1 - 0 1 - 0 1 - 0 1 - 0 1 - 0 1 - 0 1 - 0 1 Control 77.88 15.20 564.18 variables - 0 1 - 0 1 - 0 1 - 0 1 Note: (1) Variables for team 2 take the same values are those for team 1; (2) Contest incentives and win percentages are based on pre-game statistics and are thus not available for the first match day. Table 3. Summary Statistics - UCL Group Stage Sample Variable type Variable N Mean Std. Dev. Min Max Managerial t1_uefa_clinch _ko_phase 1,920 5.52% - 0 1 t1_uefa_elim 1,920 1.77% - 0 1 Control t1_mv 1,920 182.01 146.81 2.55 645.1 t1_home 1,920 0.50 - 0 1 N 1,920 Note: (1) Variables for team 2 take the same values as those for team 1. Table 4. Key Regression Results: The Sensitivity of Managerial Effort to Financial Incentives in the German Bundesliga (DV = t1_mv_start_sum) Fixed Effects VARIABLES Coef. Robust SE t1_elim -6.8685 (***) 2.0834 t2_elim 6.8685 (***) 2.0834 t1_clinch_champ - 12. t2_clinch_champ 30.7151 (*) 12.1057 t1_clinch_uclqual_elim_champ - 22.0205 t2_clinch_uclqual_elim_champ 8.2204 22.0205 t1_clinch_uclqual_nelim_champ 4.5403 4.8 t2_clinch_uclqual_nelim_champ -4.5403 4.8149 t1_clinch_uelqual_elim_uclqual -5.6831 4.2504 t2_clinch_uelqual_elim_uclqual 5.6831 4.2504 t1_clinch_uelqual_nelim_uclqual 4.0244 3.2598 t2_clinch_uelqual_nelim_uclqual -4.0244 3.2598 t1_clinch_rel - 3. t2_clinch_rel 10.0595 (**) 3.2981 t1_winpct 3.6270 (*) 1.5138 t2_winpct -3.6270 (*) 1.5138 log_t1_mv 72.2929 2.1411 log_t2_mv -72.2929 (***) 2.1411 t1_home 0.5 0. constant - 9.8408 Observations 5,940 R-squared 0.9 Note: (***) p<0.001, (**) p<0.01, (*) p<0.05, (+) p<0.1; Match-day and club-year dummies not displayed; cluster-robust standard errors on match level. Table 5. Key Regression Results: The Sensitivity of Managerial Effort to Financial Incentives in the UCL Group Phase (DV = t1_mv_start_sum) Fixed Effects VARIABLES Coef. Robust SE t1_ucl_clinch_ko_phase -44.8632 (***) 6.5323 t2_ucl_clinch_ko_phase 44.8632 (***) 6.5323 t1_ucl_elim 5.9541 5.3799 t2_ucl_elim -5.9541 5.3799 log_t1_mv 64.0069 (***) 4.0635 log_t2_mv -64.0069 (***) 4.0635 t1_home 1.5581 1.0457 constant -197.2427 (***) 18.8732 Observations 1,920 R-squared 0,9464 Note: (***) p<0.001, (**) p<0.01, (*) p<0.05, (+) p<0.1; Match-day and club-year dummies not displayed; cluster-robust standard errors on match level. Table 6. Key Regression Results: The Sensitivity of Managerial Effort in the German Bundesliga to UEFA Games in the Next Five Days (DV = t1_mv_start_sum) Fixed Effects VARIABLES Coef. Robust SE t1_elim -6.99 2.04 t2_elim 6.99 2.04 t1_clinch_champ -29. 11.9 t2_clinch_champ 29. 11.9 t1_clinch_uclqual_elim_champ - 22. t2_clinch_uclqual_elim_champ 9.1746 22. t1_clinch_uclqual_nelim_champ 4.0188 4.6 t2_clinch_uclqual_nelim_champ - 4.6 t1_clinch_uelqual_elim_uclqual -4.7 4.2 t2_clinch_uelqual_elim_uclqual 4.7 4.2 t1_clinch_uelqual_nelim_uclqual 4.8 3.2 t2_clinch_uelqual_nelim_uclqual -4.8 3.2 t1_clinch_rel -10.0 3.2 t2_clinch_rel 10.0 3.2 t1_ucl_5d -4. 2.0 t2_ucl_5d 4. 2.0 t1_uel_5d 0. 0.96 t2_uel_5d -0. 0.96 t1_ucl_ko_phase_5d t2_ucl_ko_phase_5d t1_uel_ko_phase_5d t2_uel_ko_phase_5d t1_winpct 3. 1.52 t2_winpct -3. 1.52 log_t1_mv 71.0 2.2 log_t2_mv -71.0 2.2 t1_ucl_cont 3.8 1.7391 t2_ucl_cont -3.8 1.7391 t1_uel_cont 1.7 0.95 t2_uel_cont -1.7 0.95 t1_home 0. 0.27 constant -269.3018 (***) 10.2620 Observations 5,919 R-squared 0.9535 Fixed Effects VARIABLES Coef. Robust SE t1_elim -6.7092 (***) 2.0014 t2_elim 6.7092 (***) 2.0014 t1_clinch_champ -28.6675 (*) 11.8027 t2_clinch_champ 28.6675 (*) 11.8027 t1_clinch_uclqual_elim_champ -9.5633 22.2012 t2_clinch_uclqual_elim_champ 9.5633 22.2012 t1_clinch_uclqual_nelim_champ 4.1041 4.7407 t2_clinch_uclqual_nelim_champ -4.1041 4.7407 t1_clinch_uelqual_elim_uclqual -4.5755 4.2321 t2_clinch_uelqual_elim_uclqual 4.5755 4.2321 t1_clinch_uelqual_nelim_uclqual 5.3846 3.1186 t2_clinch_uelqual_nelim_uclqual -5.3846 3.1186 t1_clinch_rel -9.8656 (**) 3.1902 t2_clinch_rel 9.8656 (**) 3.1902 t1_ucl_5d t2_ucl_5d t1_uel_5d t2_uel_5d t1_ucl_ko_phase_5d -8.4847 (*) 4.1395 t2_ucl_ko_phase_5d 8.4847 4.1395 t1_uel_ko_phase_5d 1.2080 1.5989 t2_uel_ko_phase_5d -1.2080 1.5989 t1_winpct 3.6759 (*) 1.5181 t2_winpct -3.6759 (*) 1.5181 log_t1_mv 70.5759 (***) 2.2414 log_t2_mv -70.5759 (***) 2.2414 t1_ucl_cont 3.0933 (*) 1.5449 t2_ucl_cont -3.0933 (*) 1.5449 t1_uel_cont 1.9297 (*) 0.8597 t2_uel_cont -1.9297 (*) 0.8597 t1_home 0.5274 0.2727 constant -267.5145 (***) 10.2031 Observations 5,940 R-squared 0.9545 Note: (***) p<0.001, (**) p<0.01, (*) p<0.05, (+) p<0.1; Match-day and club-year dummies not displayed; cluster-robust standard errors on match level; (1) Excluding Bundesliga matches followed by UCL games of minor importance (i.e., UCL group stage games when team has already clinched the knockout phase).