Bosman and 2004: How the EU Expansion of Central and Eastern European Countries Changed Football?
Analysis of Miklós Kállay
In 1995, the Bosman ruling had a dramatic effect on West-European football. The judgement forbade discrimination between domestic footballers and players who are citizens of the European Union. This institutional change prevented national and European international football associations from imposing quotas and setting maximal numbers for foreign athletes. However, it did not lift limitations on non-EU players.
On 1 May 2004, ten new countries joined the EU, which saw its largest enlargement to date on that day. Of those ten new members, eight states are nations with a post-communist heritage. The expansion at once allowed more than 72 million new citizens into the common market. There were around 6700 first league players in the pre-2004 market, which increased by around 2000 after the expansion, and by more than 3000 if Bulgaria and Romania are included. Furthermore, clubs with the highest average age of players are typically Western, while younger athletes come from the East, increasing the potential for change made available by the enlargement. Naturally, this expansion had an effect on all segments of the market – football market included.
In my dissertation I analyse the mutual consequences of the two events and I try to come to a conclusion, which is based on the data I collected throughout my research. These data range from the 1999-2000 to the 2008-2009 season and cover the English (Premier League), the German (Bundesliga) and the Italian (Serie A) first divisional teams. Since this is a shortened version of the original paper, it does not go details as deeply as the dissertation itself. If you are interested in the whole work, you can download it here!
Before moving into the most interesting parts, let us briefly mention the root cause – the Bosman ruling!
Jean-Marc Bosman was a Belgian professional footballer who played in RC Liége, a Belgian first division club from 1988. His contract was due to expire on 30 June 1990, but the team offered a new contract for one season in April, which would have reduced his monthly salary from BFR 120 000 to BFR 30 000. Bosman refused this and was put on the transfer list, with a compensation fee of BFR 11 743 000 (around €300 000 today). However, no club was prepared to pay that fee for him, and Liége would not let Bosman go without payment.
Bosman decided to take his case to the Court of First Instance in Liége on 8 August 1990. The case eventually proved to deal with an international problem in Europe, and the FIFA and the UEFA also joined the hearings. The incident revealed not just the issue of limiting the freedom of movement of workers by asking compensation fees for footballers who are out of contract, but also the problem of the ‘3+2 rule’ adopted by the UEFA in 1990 for all first league clubs playing in an UEFA member federation, which permitted national associations to limit the number of foreign players on the pitch to three by each club, plus two foreign players who have been playing in that country for at least five consecutive years. The case gained prominence and had advanced from the court of Liége to the European Court of Justice by 1995.
Although the Bosman ruling gave some limited directions to European leagues, it did not mean that player restrictions became universal across them. In some cases, national football associations decided that there was no need for any restriction, and let everyone in, not only the players from the EU/EEA. Nonetheless, some leagues remained conservative and continued to impose quotas on non-EU/EEA footballers. The three leagues examined in this study are the English Premier League, the German Bundesliga and the Italian Serie A.
Therefore, to give a full perspective, it is necessary to sketch the regulations applying to these three leagues.
The Premier League was rigid. A team could field at most three players who were not citizens of the EU or the EEA. The Bundesliga were less harsh: they allowed an unlimited number of ‘UEFA players’. The Serie A is a somewhat confusing case, because the ruling was changed frequently: at first, there were a maximum of five non-EU players in a team, but this rule was abolished in the 2000/2001 season. From the 2003/2004 season, a club may sign only one non-EU/EEA citizen, and only to replace another non-EU/EEA player already at the club
My predictions are: one will see a jump in the number of East-European players after 2004 in the Premier League clubs, and nothing will change in the Bundesliga. I have three tips for the Serie A: more domestic players, more East-European players, and less non-EU players.
Our initial hypotheses formulated, the study now proceeds with the empirical analysis of the topic based on data collected from numerous Internet resources, with the following sites serving as the backbone for the data generation process: http://www.worldfootball.net/, and http://www.transfermarket.co.uk/en. The information gathered includes squad size, transfer fees, position of domestic league and international tournament (Champions League and UEFA Cup). The research includes English, German and Italian teams which spent at least five seasons in the first division of their domestic league between the seasons of 1999/2000 and 2008/2009. Data is thus gathered for 20 English, 21 German and 21 Italian football clubs, 62 in total, making 620 observations in total.
Four categories of player types have been established. The first one represents the domestic players in the squad. The slight exception is the variable of the Premier League teams, in which players from the United Kingdom, rather than from England only have been included.
The second category, includes those footballers, who have come from a country which joined the EU/EEA before 2004. Swiss players were also included here, because Switzerland has been part of the European Single Market since 1999.
The third category is the one we are most interested in: it shows the number of professionals coming from a post-communist country which joined the EU in 2004. These states are: the Czech Republic, Estonia, Hungary, Latvia, Lithuania, Poland, Slovakia and Slovenia.
Lastly, the fourth category contains all the athletes whose nationalities have not been included anywhere else before. The four categories can be labelled as ‘Home’, ‘West’, ‘East’ and ‘Other’ for short. I specifically highlighted four other countries to see if there is a relation between a good international campaign and the desire of those states’ players. These four countries are: Austria (unsuccessful West), the Czech Republic (successful East), Greece (successful West) and Hungary (unsuccessful East).
Naturally one player can only be included in one of the four (domestic, West, East, other) groups, which raises the issue of dual citizenship.
In this case, there are certain ways to decide where to place the person. First, the ‘stronger’ passport always wins principle. Tim Howard (American-Hungarian) is considered as Hungarian, thus East-European, even though he represents the American national team, and Drogba (French-Ivory Coast) is included in the West-European category. When there is no difference of ‘strength’ between the passports (say, a player with an English and a Spanish citizenship in the Premier League), then the primary nationality is the birthplace and the one in which national team the footballer has played. Furthermore, since it is possible to get a citizenship after a certain time period, I will consider a player domestic automatically after he spent five years in the same country.
The Domestic League variable shows the position of the team in its domestic league in that year. Since many clubs have not always been in the first division in these ten seasons, for the purposes of determining a second division team’s league position the first and second divisions have been merged, and counting proceeds from the first team of the second division as if the two leagues were one. For instance, if a team became fourth in the second division, and the first division includes twenty teams, then the variable will be 24. This way there is no confusion about how successful a team was in any year examined.
The last two variables show whether a team qualified for either of the Champions League or the UEFA Cup, and if so, how successfully they competed. Due to the structural changes in both cups, it is hard to compare results. Since these are qualitative variables, a categorical variable scaling from 1 (‘Winner’) to 6 (‘n.a’) had to be generated, where the lower the number is, the better the club’s performance was in that year.
And now let’s see the results!
First, let’s see how an average squad is built up in each league. The table below shows the average numbers of player types before and after 2004. The t-value shows how significant the difference is between the two averages. The bigger the t-value is (in absolute terms), the more significant the difference is. The p-value shows the probability that the second average (the post-2004) can be shown from the first average (the pre-2004). Here, the smaller the p-value is, the more likely that the average has been changed and this is not just a coincidence. Thus, the bigger the t-value is, the smaller the p-value will be. Moreover, the p-value shows how significant the change is: if it is below 0.01, it is strongly significant, if it is between 0.01-0.05, then it is significant, if it is between 0.05-0.1, it is weakly significant, and if it is above 0.1, then it is non-significant.
It is possible to see some changes. Overall, 2004 made an impact on both domestic and Eastern player numbers – the first group’s numbers are declining, the second one’s rising. The same can be said of the Premier League on its own. While the number of UK players are falling, Eastern footballers are becoming more common in the country. In the Bundesliga, this is not the case. While German players are in a slight downturn, it is statistically insignificant.
The number of Eastern players is barely up either. The surprising case is the rise of Western players, which is statistically highly significant, but there is no explanation for this trend. In the Serie A, Eastern players are more welcome now, while overseas players find it harder to stay. This is probably due to the restriction of non-EU players in the Italian first division in 2003, which was discussed above.
The next table shows the same results, but now with percentages rather than averages:
The previously mentioned values stagnated (or even became more significant), while in some cases, the p-value even went down a bit: in the Premier League, the Western players, are close to 0.1, and the non-EU footballers are even less than 0.05.
The t-tests for the four countries were also calculated. However, there are only a few results which are statistically significant and only for Czech and Hungarian players. The underlying reason for this is probably the small number of these nationals in the leagues, which has not changed dramatically, therefore, the t-values mostly indicate no rise and fall in the numbers and shares. Moreover, since the number of footballers per individual country is small, their shares in the squads are also miniature.
And now, the more complicated calculations. First here is the table, and I will discuss the results below:
Table 4. Estimated coefficients of the regressors’ impact on the number of players
The coefficients of the regressors show what kind of and how big influence they have on the numbers of the player types.
We start with the constant: if nothing is included, there will be this number of players in a team from that category. For example, the constant of the domestic players is 16.527, so there would be 16-17 home players in one team. However, the coefficient for the year is -0.449, which means that a team usually loses one player around every two years. The z-value can be considered as the t-value, while the p-value is the same.
Looking at the table, it is clear that many regressors are usually not, or just weakly significant statistically. However, the only independent variable which never really matters is the ranking in the UEFA Cup. The constant is always strongly significant, and the year and ranking in the domestic league do matter for most player types. Every year, the number of domestic players in a team tends to fall, while the number of Western footballers rise.
The most notable panel is Panel B, which examines Western footballers: every correlation except the one between UEFA Cup performance and Western player numbers, is highly significant, therefore that part of the table tells us the most. As mentioned earlier, as time progresses and as better domestic league results are achieved the number of Western players in the squad rises, while better runs in the Champions League have a negative impact.
This result is rather surprising given that usually those same teams which perform better in one tournament do likewise in the other one too. The R-squared it the biggest here, too: 0.8044. The R-squared shows how well these regressors explain the numbers and the changes. In this case, the calculations explain around 80% of the changes regarding Western players.
Lastly, the most complicated analysis. Here I gathered all the data together and made one table out of it:
Table 5. Estimated coefficients of the regressors’ impact on the number of relevant players
In this table, the constant shows the number of non-EU players if we ignore everything else. This means that usually four overseas players can be found in one team. The ‘Home’, West’ and ‘East’ shows their difference between the constant. This suggests that there are around 19 domestic, 5-6 Western and 1 Eastern players in every team. The ‘EU Access’ shows how the 2004 enlargement affected the non-EU footballers, and the ‘Access home/West/East states how this number is different regarding the specific groups.
Every second club lost a domestic player this year, there is almost no change in the number of Western players, a slight increase in Eastern players, and a decrease in overseas players (domestic: -0.227 + -0.444 = -0.671; West: -0.227 + 0.293 = 0.066; East: 0.33). The last numbers show how the categories’ player numbers changed yearly after 2004.
Therefore, while there are 0.231 less home athletes in each team, there are 0.046 more Western and 0.104 more Eastern players (and 0.031 more non-EU – that coefficient remains the same). In order to fully understand, the graphs below show how these numbers have changed between 1999-2009. The blue line shows the actual trend, which took place, and the yellow shows the trend which would have happened without the influence of 2004. Until this year, the two lines are naturally identical, since there was no effect in place.
It is clear that changes have been experienced in all four categories. Nonetheless, I must highlight that these changes might just be a coincidence, since the p-value is too high in many cases, thus making it hardly possible to prove anything.
Although 2004 definitely had an impact on European football, it is not as strong as previously assumed. By reviewing the leagues one-by-one, we can conclude that the Premier League underwent the most pronounced change in the period, since both domestic (sharp fall) and Eastern (sharp rise) player numbers were strongly influenced.
In the Serie A overseas footballers saw their numbers dwindle. Nonetheless, this could be ascribed to the regulations implemented in 2003 to encourage Italian clubs to hire Italian athletes. It seems the FIGC (Italitan Football Federation) did not accomplish this, and only Eastern players benefited from the restriction.
As to the result from the Bundesliga; we hypothesised that the number of Eastern players will not rise which is true, but Western player numbers increased – the reason for this is unclear.
The results of the two regression analyses are not strong enough to pronounce 2004 a watershed in football (at least it is not statistically significant). However, from those we can draw the conclusion that the number of domestic players remarkably fell.
Although country-specific experiments have not been much used in the study, the results, found in the appendix, are worth mentioning.
These analyses do not tell us much, since only four states’ footballers have been analysed, and none of them have been dominant football nations in the 21st century.
However, even this small sample tells us a bit. The Austrians and the Greeks are not better off because of 2004, but the Czechs and the Hungarians are – a result not unexpected.
Nonetheless, inclusion of further variables would be required to refine this study, geographical location or the strength of the domestic league for example. Conducting analysis on more nations is also desirable.
The analyses of this study can be expanded upon. There are numerous ways in which more certain results could be achieved, containing more significant statistical inferences. The most straightforward way, without altering any other parameters of the study would be working with a longer time-frame. The study could take data from as early as 1995, the introduction of the Bosman-rule; and as late as 2017/2018 as well. It is possible to lower the barrier I set of the minimum of five seasons spent in the first division required for inclusion in this study, thus entering more teams, yielding more data to work with. Moreover, the range of the study’s interest can be broadened to encompass not only the elite level of football; second or even third divisions of each country could be included in the analysis (although not lower than third division – the fourth divisions are often amateur leagues, except in England). The analysis could study more states’ football environment: the first divisions of other West-European countries. Highly recommended are: Spain, France and Portugal, or perhaps the Dutch league.
Copyright 2018. @ Calcioanalyst