Analysis of the football transfer market of the Russian Premier League

Kiefer (2014) in his research showed the importance of red and yellow cards. There is no doubt that red cards and yellow cards give negative to the utility for the players and club because it leads to the disqualification of players and to the game in unequal squads. However, the following tendency can be seen: if a player spends more minutes on the football pitch, then the probability of getting a red or yellow card is higher. We will see the overall effect of cards on the market values.

Another two common variables are YandexRequests and MVP or in other words most valuable player. After each match, the organizers decide who will be the MVP of the game. So we can predict that most skillful players who usually have higher market values have higher probability to become the best player of the match. YandexRequests is a proxy to measure the popularity of players. It seems to be a better proxy comparing to the proxies like mentions on the TV ads because popularity will be measured most objectively. And if the proxy mentions on the TV ads will be used then there will be 2 problems. The first one is with the sample: it is impossible for all football players to have contract with different brands. The second problem is that the character of football players influence on this proxy: some players love to appear on the TV screens while others not, so the endogeneity will be higher with this proxy. We can predict that players who are more popular usually have higher market value. Their popularity is usually based on their great performance on the football pitch so market value adjusts to their high skill and, as a result, such players are better known all over the world.

The last one is MarketValueDecember2017. I decided to include this variable in order to decrease the endogeneity in the model. The logic is following: it is obvious that market values in one lag are closely related to each other, so by including previous market value we will decrease the level of endogeneity.

And now I would to discuss and comment the individual explanatory variables for each of the model.

OwnGoals, DuelsWonPerMatch are two explanatory variables that were included in the model for defensive players. Own goals usually appear in football when players make awful technical mistakes, for example try to kick ball from the penalty area and cut the ball into the own net. So usually own goals can be seen as consequence of low technical skill of a player. DuelsWonPerMatch is included into the model for defensive players because duels between players have become an essential part of the game itself. By winning these duels, players increase the probability of taking the ball into their team's possession or increase the probability of scoring a goal/do not miss a goal. So, we can predict that higher amount of winning duels for the defender will increase their market value.

The specific explanatory variables for the model of midfielders are AccuracyOfPass, Goals, NumberOfAssists and DuelsWonPerMatch that was already discussed. These explanatory variable shows the skill and technique of football players because players with lower mistakes in passes give higher utility for the team. Goals and assists indicate player's scoring ability and player's contribution to help others to score goals, so these variables are also should be included. These variables were described in details in the works by Bryson et al., (2012) and Gerrard & Dobson (2000).

Now about individual variables for forward's model. There will be again Goals, NumberOfAssists. Moreover, NumberOfShotsPerGame, DispossedPerGame, NumberOfOffsidesPerGame and SuccessfulDribblingPerGame will be added. Number of shots per game shows the ability of forwards to find a position, space and time in order to score a goal with this shot. The same story with dribbling: players who are more confident in themselves are more likely to use dribbling in order to improve the situation on the field. So, we expect that higher number of shots and dribbling tends to increase the market value of forwards. (Medcalfe, (2008)) However, larger attempts to go into dribbling increase the probability for the player to lose the ball. In order to take into the account this situation, the explanatory variable DispossedPerGame need to be included into the model. And we expect that the higher value of this variable will lead to larger decrease of the market value. It is known that some teams use offside trap in their defensive strategies, so forwards of the opposite team become the victims of this strategy. When the offside is defined by the referee, then the ball and dispossession goes to the team which organize this offside. In other words, we can compare offside with the dispossession of the ball. Again, the expectations with the variable NumberOfOffsidesPerGame will be similar as with the variable DispossedPerGame.

3.3 Results and interpretation

One of the purposes of this work is to understand if the popularity affects the market value. And if it so, then what is the effect of popularity. Another hypothesis that I would like to test with this data touches the age of football players. Moreover, to see the overall picture: which variables will be significant for each model and what is the effect of each variable.

We can notice that our dependent variable is not a simple variable: it has a natural logarithm. This will lead to unusual interpretation of coefficients of the explanatory variables. If any explanatory variable rises for one unit, then the coefficient on this explanatory variable shows the change of market value in %.

The table with the results can be found in Table 10:

Table 10. The coefficients and standard errors of explanatory variables

*** means that P-value approximately equals to 0.01

** means that P-value approximately equals to 0.05

* means that P-value approximately equals to 0.1

The R2 for defender's model is 0.8519, for midfielder's model is 0.8469 and for forwards's model is 0.7973.

Robust standard errors were used in order to deal with the problem of heteroscedasticity. Speaking about multicollinearity, it should be said that it occurs in variables Age and Age2. But it is a specific assumption to include both variable Age and Age2.

Firstly, we look at the model for defenders. There is an evidence that YandexRequests is a significant explanatory variable at 90% significance level. So the result can be interpreted as following: every additional request in Yandex search engine increases the market value of a football player by 0.00777%. Variables Age and Age2 are significant even at 99% significance level. There is a quadratic form, so the interpretation will be more complicated. The peak of the market value for the defenders will be achieved at the 24.33 years old. It is calculated from the vertex of a parabola= -0.779067/(-0.03201224). We can show different marginal effects: 0.779067-0.03201224*Age. For example if the player is 24 years old then the marginal effect can be calculated in the following way: 0.779067-0.03201224*Age and Age=24 =>marginal effect=0.01077.

It should also be said that such variables are significant at least at the 90% significance level for the model of defenders: MarketValueDecember2017, MinutesPlayedThisSeason, Russian, Weight, NumberOfOwnGoals, PlayerOfANationalTeam, MinutesPlayedLastSeason. The corresponding coefficients can be seen from the table with the results. Every additional minute for a defender increases his market value by 0.03% on average. If the defender is Russian, then his market value is likely to decrease by 27.9%. This thing happens because of the limit on foreign players. This rule gives incentive for the club to buy only high skilled foreign players. High skilled players usually have higher market value. As a result, Russian defenders have lower market value.

Now about midfielders. Here we can say that there is an evidence that YandexRequests is a significance explanatory variable at 99% significance level. Interpretation is the following: every additional request on the Yandex search engine increases the market value of the midfielder on the 0.00134%. Age and Age2 are significant at 99% significance level. Using the same method of interpretation of quadratic form for age, we can state that usually age increases the market value up to 23.89 years old and after this age, market value decreases for midfielders. The marginal effect can be found in the following way: 0.6168743-0.0258212*Age. In order to find particular marginal effect we need to put the needed age into this equation.

There are other variables that appear to be significant at least at 90% significance level for the midfielder model: MarketValueDecember2017, MinutesPlayedThisSeason, PlayerOfANationalTeam, Weight, AccuracyOfPasses. All these significant variables have a positive effect on market value (see coefficients in the table with the results). Every additional minute for a midfielder played this season increases his market value by 0.039% on average. The interpretation of these variables is simple. For example, every additional successful per cent of passes increase the market value by 1.71% on average.

Finally, we are going to discuss the results for forwards. The population in the Russian Premier League is not very representative, only 90 observations. YandexRequests is not significant at all, P-value is 0.8. Possible explanation of this result is small number of observations. Age and Age2 are significant variables at 95% significance level. According to our sample of forwards, the highest market value is achieved at the age of 22.85 years old (calculated with the same method as for defenders and midfielders). The marginal effect can be found with the equation 0.4407201-0.0192878*Age.

Other significant variables for forwards at least at 90% significance level are MarketValueDecember2017, MinutesPlayedLastSeason, The interpretation of these variables is obvious. For example, every additional minute played last season increases the market value of forwards by 0.02281% on average. P-values for goals equals to 0.265, for DispossedPerGame equals to 0.113, for MinutesPlayedThisSeason equals to 0.18, for PlayerOfANationalTeam equals to 0.1368. Again, I suppose that these variables are not significant just because of small population for forwards in Russian Premier League. Because they are usually considered as the key statistics for forwards: goals show your ability to score the goals, invitation to the national team usually can be characterized by the fact that the skill of the player is relatively high.

Overall, it can be stated that popularity influences positively on the market value of player and tests confirm this for defenders and midfielders. Russian nationality influences negatively on the market value of players in the Russian Premier League, while age has a quadratic relationship and, according to this relationship, age increases the market value up to the age 24 years old and then decreases the market value. Finally, we can state that time on a football field increases player's market value and it was proved statistically for defenders and midfielders.

Conclusion

All the tasks that were written in the aim of the work were fulfilled. We showed the difference between market values and transfer fees paid for the player. Moreover, we understand that the market value is always changing over some period and discuss the most probable reasons of these changes: age, personal statistics during the season, physical abilities and popularity of players. In was done an overview of the transfer market of the Russian Premier League and concluded that this league is considered as a good opportunity to get some experience before going to play for the best European club. It was provided the evidence that Russian clubs are not ignored by the scouts of the best European clubs and there is a demand on the best players in this league. Examples of the biggest improvements of the market value for the players of the Russian Premier League confirm our verdict that players can develop in this league.

We saw that depreciation of ruble led to the lower purchasing power of Russian clubs. The analysis based on the purchase value showed which clubs were the most effective in the season 2017/2018 in such terms like costs per point and costs per goal.

The most important part of this work was to understand which variables influence on the market value of football players in Russian League. For this purpose three models were built: for defenders, midfielders and forwards. The most important variable YandexRequests for the analysis in this work stood for the popularity. Our predictions that popular players tend to have higher market value were confirmed statistically for defenders and midfielders. It was found that YandexRequests is significant variable at 90% significance level for defenders and significant at 99% significance level for midfielders. For forwards this variable is not significant but this may happen because of a small sample. Moreover, we found other significant variables for each of the model which were discussed in details in this paper. In particular, such variables that are responsible for previous market value of football player, amount of minutes played during this and last season, being a player of a national team were statistically significant at least at 90% significance level and positively influenced on the market value for a football player.

References

1. Victor A. Matheson, (2004). “European football: A survey of the literature”.

2. Bernd Frick, (2007). “The football player's labor market: Empirical evidence from the major European leagues”.

3. Stephen Dobson & Bill Gerrard, (1999 ). “The determination of player transfer fees in English Professional soccer”.

4. F. Carmichael & D. Thomas, (1993). “Bargaining in the transfer market: theory and evidence”.

5. Pedro Garcia-del-Barrio & Francesc Puyol, (2016). “Broadcasting Revenues and Media Value in European Football”.

6. Ruijg & van Ophem, (2014). “Determinants of football transfers”.

7. Barrio & Puyol, (2007). “Hidden Monopsony Rents in Winner-take-all Markets - Sport and Economic Contribution of Spanish Soccer Players”.

8. Bryson et al. (2012). “The returns to scarce talent: footedness and player remuneration in European soccer”; Journal of Sports Economics, 14 (6) (2012), pp. 606-628.

9. Fry, Galanos & Posso (2014). “Let's get Messi? Top-scorer productivity in the European Champions League”; Scottish Journal of Political Economy, 61 (3) (2014), pp. 261-279.

10. Herm et al. (2014). “When the crowd evaluates soccer player's market values: accuracy and evaluation attributes of an online community”; Sport Management Review, 17 (4) (2014), pp. 484-492.

11. Carmichael et al., (1999). “The labour market in association football: who gets transferred and for how much?”; Bulletin of Economic Research, 51 (2) (1999), pp. 125-150.

12. Franck & Nuesch, (2012). “Talent and/or popularity: what does it take to be a superstar?”; Economic inquiry, 50 (1) (2012), pp. 202-216.

13. Medcalfe, (2008). “English League transfer prices: is there a racial dimension? A re-examination with new data”; Applied Econometrics Letters, 15 (11) (2008), pp. 865-867.

14. Lehmann & Schulze, (2008). “What does it take to be a star? The role of performance and the media for German soccer players”; Applied Economics Quarterly, 54 (1) (2008), pp. 59-70.

15. Carmichael & Thomas, (2000). “Team performance: the case of English Premiership football”.

16. Antonioni, Cubben & Dilger, (2000). “The Bosman ruling and the emergence of a single market in soccer talent”; European journal of Law and Economics, 9, pp. 157-173.

17. Szymanski, (1999). “The Americanization of European football”; Economic policy, vol. 14, issue 28, pp. 203-240.

18. Reilly & Witt, (1995). “English league transfer prices: is there a racial discrimination?”; Applied Economic Letters, vol. 2, issue 7, pp. 220-222.

19. Lucifora & Simmons, (2003). “Superstar effects in Sport: Evidence from Italian Soccer”; Journal of Sports Economics, vol. 4, issue 1, pp. 35-55.

20. Feess, (2003). “The impact of Transfer Fees on Professional Sports: An Analysis of the New Transfer System for European Football”.

21. Kiefer, (2014). “The impact of the Euro 2012 on popularity and market values of football players”; International journal of Sport Finance, 9(2), pp. 95-110.

Размещено на Allbest.ru