PREMIER LEAGUE PLAYER EXPENSES DURING COVID-19 - DIVA PORTAL

 
CONTINUE READING
Bachelor Thesis

Premier League player
expenses during Covid-19
How spending on the transfer market of football
has shifted since the initial shock of the pandemic.

                                Author: Daniel Axelsson
                                Supervisor: Lars Behrenz
                                Examiner: Mats Hammarstedt
                                Term: VT21
                                Subject: Economics
                                Level: Bachelor
                                Course code: 2NA11E
Abstract
This thesis examines the impact of the initial shock of the Covid-19
pandemic on the transfer market within the English Premier League. To
determine whether teams still overspend the same amount of money above
the market values of players as they did prior to the pandemic. Using a
differences-in-differences approach and comparing actual transfer values to
established market values of player, both before and after the initial
pandemic shock, with data and player statistics collected from the popular
German website Transfermarkt.de. This resulted in an average 11.8 percent
decrease in spending over the market value per player, or about 950.000
Euros less per player above their market value. However, both results show
no statistical significance of being true. Concluding that teams may have
spent less after the initial shock of the Covid-19 pandemic compared to
player market values. To support this explorative data result, more data must
be collected and analysed in the future.
Key words
MV
Market values
DID
Difference-in-differences
EPL (English top flight)
English Premier League
The Top Five
The top five leagues of Europe; English Premier League, French Ligue 1,
German Bundesliga, Spanish La Liga, and Italian Serie A.
The Big Six
The biggest clubs of England; Arsenal FC, Chelsea FC, Liverpool FC,
Manchester City, Manchester United and Tottenham Hotspur.
Acknowledgments
I would like to thank Lars Behrenz for his supervision on this thesis, Mats
Hammarstedt for his helpful comments and Joakim Jansson for his guidance
with the statistical tools.

I would also like to give a special thanks to Amanda Abrahamsson and Thea
Andersson. Without your help over the past years this thesis, or this degree,
wouldn’t have been possible.
Table of contents
1     Introduction                 1
2     Background                   2
3      Literature review           5
    3.1     The transfer market    5
    3.2     External shocks        8
4     Hypothesis                  10
5     Data                        11
6     Empirical strategy          15
7     Results                     19
8     Conclusion                  25
9     References                  27
10      Appendix                   1
1 Introduction
The covid-19 pandemic has had an impact on almost every economic sector, one of them
being professional sports, with most sports leagues either cancelling or postponing play after
the first wave started in March 2020.
The European football championship (Euros 2020) and the Tokyo Olympic Games were both
postponed until 2021, while the French and Dutch top-flights in football “Ligue 1” and
“Eredivisie” were some of the leagues who were cancelled all together (Drewes et al 2020).
Since then, most leagues have returned, only under very different circumstances: no fans at
the stadiums and increased security measures to limit the spread of the virus. These changes
have certainly impacted the game of football as well as the economics of football. With the
loss of all match day revenue, many experts and scholars feared that clubs would struggle to
keep afloat and be forced to cease operation or fall into bankruptcy, mostly in the lower tiers
of the system. (Drewes et al 2020)
To understand how the economics of football have changed since the first wave of the
pandemic and the initial shock, this thesis aims to determine if the transfer sums of players
who were transferred after the initial shock, have changed compared to those before the
pandemics first lockdown in Europe in relation to their market values. The usage of this
period is to establish a natural experiment since the absence of fans and other factor has
created an exogenous shock which, in a unique way, can change the way the transfer market
within English football operates.
The English Premier League was one of the few leagues across Europe where teams did not
receive any loans or packages from their governments. This helps with seeing the real effects
of the pandemic’s shock on the player transfer market, since no new funds from new sources
were given to the clubs to aid them.
To look at differences in propensity to spend after the exogenous shock that was and is the
Covid-19 pandemic, this thesis looks at player transfer sums of teams in the English top-
flight, The Premier League, and compare them to their respective market values, to see if
clubs still spend the same amounts on players after the exogenous shock of a global
pandemic. Has the loss of match day tickets and the uncertain future made teams more
hesitant to spend their money on players or has it perhaps made them spend more?
The transfer market is an important part of modern football. Not only for those at the top, to
help them secure more talent and win more trophies, also the lower tier teams benefit as well

                                                                                             1(29)
and even depend on selling players to bigger clubs to make ends meet. (Deloitte, Annual
Review of Football Finance 2020)

The goal of this thesis is to evaluate whether the clubs in the English top-flight division “The
English Premier League” still overspend the same amount of money on transfers of players
after the exogenous shock of the Covid-19 pandemic, as they were before the shock. By using
a difference-in-differences approach with the help of market values of players and their actual
transfer sums from Transfermarkt.de. This thesis will use the unique shutdown of play
between March and May of 2020 as a natural experiment to determine if it had an impact on
how Premier League teams act on the transfer market. Will it still be the case that English
teams overpay, in terms of market values, for players to the same extent as before, or will
there be a shift in how the Premier League clubs act on the market?

This thesis adds to previous literature about the transfer market, and demonstrates how an
exogenous shock of this magnitude, impacts global markets not only on common goods and
services, but also on the exclusive transfer market of football players.

Section 2 will give a short history about the transfer market followed by some important
aspects of the market and how they are used. Section 3 will contain insight previous literature
about footballs transfers market as well as other research about natural disasters and usage of
the method Difference-in-differences. In section 4 the hypothesis and some relevant theories
will be presented. Section 5 handles the data and information about the gathering and usage
of the data. Section 6 contains the empirical strategy and the methods used in the thesis.
Section 7 provides the results and in section 8 the thesis is concluded.

2 Background
In this section the history of the transfer market within football will be discussed. It will
contain a more in-depth explanation to why the transfer market looks like it does today, why,
and how it is important for many clubs worldwide, and why some clubs might have to pay
more than others when buying players. Also briefly mentioned in this section are some
different aspects of the football transfer market, how it has changed over the years and what
impacts these aspects have had on the market today.

The earliest traces of a sort of transfer market emerged in the home of professional football:
England. It can be dated back to the very first years of the sport being played and the
founding of the “Football League” in 1888. In this system often referred to as the “retain and

                                                                                                2(29)
transfer” system. Players were not allowed to leave a club for which they had signed for
unless that club released them of their services. The club then, still had exclusive rights on the
player and could bargain for a fee of transfer since they would be missing out on future
services of said player. (F. Carmichael and D. Thomas 1993) Along with the restrictions on
moving, players also had a maximum salary which was put in place to stop some teams to
attain all the superior players. (Simmons 1997) This system was operated uninterrupted until
the early years of the 1960s when following a dispute with the FA, the Football Association,
and the players’ union. The maximum salary was abolished. (F. Carmichael and D. Thomas
1993) Later, in 1963, the system was brought in front of the high court. The FA defended the
system proposing that without it the biggest clubs would be taking all the best players,
disrupting competition of the smaller, less powerful clubs. Another argument was that if
clubs would not come to receive payment for players leaving, they would not have the
incentives to spend money on the training and development of its players. (Simmons 1997)
The system was then modified as it was seen to be an “unreasonable restraint of trade”.

At the start of the 1977-1978 season further regulations and rules were changed as the
“freedom of contract” was introduced. Players where now allowed to negotiate with other
clubs once their contract was about to expire, and if their old clubs did not offer an equivalent
or better offer than the last year of the players contract, said player would be entitled to a
“free transfer”. If the old club did offer this sort of renewal, the player could still reject it, but
the new club then had to pay a compensation fee of transfer to the player’s old club. Any
complaints of disputes would go to the FLAC, the Football League Appeals Committee, a
tribunal who works independently for all clubs.

However, if a player was under contract, things were much different. Clubs then would have
to negotiate alone with each other without any representation whatsoever from the player.
Not until after two clubs agree on a transfer sum for a specific player, could the new club
discuss terms of a new contract with him. (F. Carmichael and D. Thomas 1993)

The next and perhaps most monumental change on the football transfers system came after a
ruling in the European Court of Justice. The famous “Bosman case” where the Belgian
national Jean-Marc Bosman challenged and sued his club RFC Liége on grounds of restraint
of trade and freedom of movement for workers under Article 48 of the Rome treaty. After his
contract had expired Bosman received an extension from Liége which where inferior to the
contract which he had received from a French club, US Dunkerque. Liége however refused to

                                                                                                  3(29)
let Bosman join the French club. The ruling from the European Court of Justice came in
December 1995 and stated that the agreement of a transfer sum after a player’s contract had
expired was incompatible with Article 48 as it restricted freedom of movement between two
member states of the European Union. (European Court Reports I-04921 1995) The court
also stated that the restriction of the number of foreign players allowed on a team was a
breach of Article 48. Players where now allowed to go wherever they were offered a contract
after their old contracts expired, something that caused outrage among many clubs who stated
that it would put players jobs at risk and “kill” the transfer market. This was not the case, as
money spent on transfers kept on rising, especially within the English Leagues. (Simmons
1997)

The transfer market continued to develop in the 21st century when in 2002 the European
commission and FIFA, football’s worldwide governing body, further regulated the way
players could be transferred across teams. Under the new rules a player under contract could
buy themselves out of the contract by paying a so called “release clause” a clause with a
value usual much higher than the value of the remaining contract. Most commonly this clause
would not be paid by the player himself but rather by a different club who would wish to take
on the player. The old club of the player could also negotiate with the new club to a lower fee
than the original release clause making it once again easier for players to move. (Hoey et al
2020)

FIFA, abbreviate of The Federation of International Football Associations, the governing
body of football internationally dictates the way clubs make transfers. The different members
are mandated to have two different periods of registrations of transfers. This is where players
can move from one club to another. According to FIFA, the first registration period begins
after the season is finished and should normally end before the new season begins and not
exceed 12 weeks. The second period should be in the middle of the season and not exceed 4
weeks. Most top leagues across Europe use the same periods to facilitate international play
and the movement of players. In popular literature the periods are commonly called “transfer
windows” and are differentiated by the “Summer window” (period 1, between the seasons)
and the “Winter window” (period 2, during the season). (FIFA Regulations on the Status and
Transfer of Player 2020)

In addition to a transfer fee, selling clubs could bargain for a Sell-On fee, a fee which
compensates the selling club on future transfers of a certain player. (Gürtler 2012) This

                                                                                              4(29)
would simply mean that if Club A decides to sell a player to Club B with a 20% sell-on fee
and Club B later sells the player to Club C for 10 million euro. Club A would be entitled to 2
million euros from the Sell-on fee. This is a common way for clubs to secure income in the
future, for instance with a young player from a small club that might be worth more in the
future. It can also help the buying club as it can be possible for it to buy a player for a lower
initial fee if there are monetary constraints. (Gürtler 2012).

3 Literature review
In the first part of this section, previous literature on the football transfer market, how transfer
sums are determined, how teams act differently, which teams can spend the most and the
effects Covid-19 have had on it so far, will be discussed. The second and last part of this
section relates to literature about other external shocks and how they have been researched in
the past.

3.1 The transfer market
When Rottenberg (1956) argued for baseball to open its system to a free market, he argued
that allowing players to move freely from team to team, would enable them to not be
discriminated against while at the same time spurring teams to be competitive as they “must
be nearly equal if each is to prosper”. His work on the valuation of players came down to
more than just player talent, but also the size of the club, its success, and its fans mattered. In
the market of football as seen in the previous chapter, this came into fruition step by step (F.
Carmichael and D. Thomas 1993; Simmons 1997). Where the former do find evidence of a
positive relationship between a wealthy clubs’ ability to buy players and therefore establish
themselves on the transfer market. In today’s modern football, as in other industries, firms or
clubs pay more for better workers, meaning that monetary compensation for players and for
clubs who sell players are determined by what kind of value he can add to the team. The
success “on the field” translates to financial success when crowds get larger, advertisement
more lucrative and price money is awarded. According to Ruijg and Ophem (2015) hiring a
new player is unsecure business and comes with a threefold payment: First the player is paid
a salary, secondly the player might need to be released from a previous contract and require a
transfer fee and lastly the former club might require compensation for its investment in the
players development (Ruijg and Ophem 2015). Building on the work from Carmichael et al
(1999) who found that some players, within English football, are more likely to be transferred
than others, using the Heckman two-step procedure. Namely those players who can score

                                                                                                5(29)
more goals and those who have been out on a loan spell but with a small transfer history.
They also confirm that players who have a higher fee of transfer are more likely to be sold
than others. Keeping in mind that this article examined data prior to the revolutionary
Bosman ruling, as the percentage of transfers that involved cash drastically declined in the
mid-1990s after the ruling (Frick 2007). Frick’s (2007) article about the football players
labour market, establishes that the variations in transfer fees boils down to player attributes
such as age, career games and international appearances. He also confirms that more
successful clubs pay more money in transfers. Since the Bosman ruling Frick (2007) points to
that the numbers of years left on a player’s contract, is very likely to have a large impact on
the price tag of the player.

When determining transfer sums after the Bosman case, for all players, Ruijg and Ophem
(2015) found that only a handful of characteristics are important and bring a positive effect
on the team’s success: age, average number of minutes played and not being a goalkeeper
where some of the most important. Suggesting that teams in the market for financial and on
the field success, should focus on these characteristics. Other scholars argue that teams in
Europe focus more on the “on the field” success rather than financial success compared to the
equivalent in North America (Lagos et al 2006; Pérez-Gonzaléz et al 2020). This can be both
voluntary and by forced regulations. Keeping this in mind, the determination that not all clubs
buy and sell players for profit maximization but rather to victory maximize can be made.
Even though one might lead to the other.

In the article Football transfer fee premiums and Europe’s big five Depken II and Globan
(2020) determines that bigger teams across Europe, especially English Premier League teams,
pay a premium when buying players. On average English teams pay roughly 1.8 million
pounds (£) more per player. This having much to do with lucrative broadcasting rights both
internationally and domestically since they find no significant difference in match attendance.
Depken II and Globan used a difference-in-differences approach to examine whether the
various broadcasting deals in 2012 had made an impact on these premiums. Resulting in an
approximate £ 2 million difference-in-differences post 2012 for English teams compared to
other top five leagues. (Depken II and Globan 2020)

Pérez-Gonzaléz et al (2015) analysed the 13 most valuable teams according to the Deloitte
football money league to show the average value of each player in the 13 squads across
Europe. Within these 13 teams, the six teams from the English Premier League are

                                                                                               6(29)
Manchester United, Manchester City, Liverpool FC, Chelsea FC, Arsenal FC, and Tottenham
Hotspur. These teams form the, commonly used in popular media, “Big Six” clubs.

In the nine years Pérez-Gonzaléz et al (2015) studied the 13 teams they saw an increase in
both club revenue and average market value of players, which had increased by 100.3% and
97.1% respectively, showing a similar growth in both categories. These revenues come from
three main streams of income. Theses streams are commercial, broadcasting and match day.
During the 2017/2018 season Match Day revenue accounted for about 17 % of the revenue
although it had accounted for around 30 % just 10 years prior. Since the 2017/2018 Season
the percentage of match day revenue has continued to decrease slightly, however it is still a
significant part of top clubs’ income. As teams were forced to close its gates to fans in the
spring of 2020, the money usually generated on match day vanished from the income sheet.
(Drewes et al 2020). The Deloitte Football Money League (2021), who for 24 years have
profiled the financial performance of the highest generating teams in the world of football,
stated that the top 20 clubs (seven of them English Premier League clubs, the “Big six” and
Everton FC) have failed to collect over 2 billion euros from the missing match day revenue in
hand with different rebates from broadcasters.

However, English Premier League clubs manged to limit its rebate with broadcasters better
than most other top 5 leagues. Meaning they had a slightly smaller revenue loss compared to
the French Ligue 1 who cancelled the 2019/2020 season and lost a significant part of their
revenue from broadcasting but instead received government funding. (DFML 2021)

Drewers et al (2020) mentions a second impact that the Covid-19 Pandemic could have on
the future of football. With games played in front of empty seats without spectators, the
atmosphere of thousands of fans screaming, encouraging, singing and even booing has
vanished. Wunderlich et al (2021) conducted a study to measure how home advantages have
changed since the pandemic. They found a significant change to how referees judge game
situations, where their bias towards distributing more yellow cards and fouls for the away
team, had completely disappeared without fans in the stands. Although the outcome of games
regarding home field advantage where insignificant, they still saw a minor effect. Drewers et
al (2020) argues that without the extra element of atmosphere, fans could be less likely to pay
for an expensive TV-package when the product is inferior to what it once was. “Clubs are
responsible for the production of the match and the fans for the production of stadium
atmosphere” (Drewers et al 2020) If teams want fans to pay the same amount for watching

                                                                                                7(29)
their games on TV it may be of importance to have fans back as soon as possible, and with an
uncertain future of when that could be, clubs may have to expect an even larger decrease of
revenue, if fans in the broadcasting section of revenue decrease as well. An expected loss of
broadcasting revenue could influence how much a club is willing to spend on the transfer
market.

3.2 External shocks
Moving on to literature about other “natural disasters” or external shocks, and more specific
natural experiments on these disasters or shocks. Beginning with Impact of the Great East
Japan Earthquake on the oyster market: a difference-in-differences estimation, where Sakai
et al (2018) looks at the oyster industry before and after the Great East Japan earthquake in
2011. They compare the regions who were affected by the disaster to those who were not.
Establishing a difference-in-differences estimate which they validate by looking at the
common trends of production and price in different geographic locations before the disaster
struck. They find that because of the earthquake, production decreased by 65% and increased
the price of oysters by over 26%.

Similarly, Leiter et al (2009) investigates the effects floods have on European firms’ capital
accumulation, productivity, and employment growth. Separating affected regions and non-
affected region with before and after a flood, they analyse the changes in capital,
productivity, and employment. In their research they discover that while using the difference-
in-differences approach, physical capital accumulation is higher in those affected by floods as
well as the short run employment. However, productivity decreased in all cases for regions
affected by floods.

In a comparable way this sort of difference-in-differences methodology could be used in this
study. To be able to see how the difference in transfers sums after the pandemic are related to
those before, with the control group being the market values of the players at the time of the
transfer.

Connecting this to previous research about sports and more specifically to the transfer market
of European football, the work of Depken II and Globan (2020) return once more. They find
that the difference-in-differences approach gave significant result, when looking at the higher
transfer sum premiums for English teams after the Premier League signed new broadcasting
deals in 2012. They use the broadcasting deal as an external shock and compare the changes

                                                                                            8(29)
with other European “top five” leagues who did not receive a new broadcasting deal at the
time.

Many scholars expects to see a negative effect on income for all clubs after the Covid-19
pandemic (Drewes et al 2020; Depken II and Globan 2020; DFML 2021). Considering that
the teams in the top flights across Europe did not seek or did not receive financial help, but
rather gave support and funding to clubs in the lower tiers of their domestic system. For
example, in Germany where the top four teams who had participated in the UEFA
Champions League, provided 20 million euros to the remaining teams in the division
(Drewers et al 2020; PremierLeague.com 2020). Meaning that some “top five” leagues did
not take money from their respective governments to survive the pandemic. While discussing
market values Drewers et al (2020) states that the impact of the Covid-19 pandemic using
market values is an interesting research question as the shutdown in playing may act as a
natural experiment. A central part of this thesis.

In today’s modern football, transfers are an important part of the game. Whether you are in
the market for selling or buying, teams depend on the transfer market for either financial
reasons or for “on the field” success. In the 2019/2020 season Premier League clubs alone
spent over 1.8 billion Euro on the transfer market (Transfermarkt 2021). 17 out of 20 teams
were net buyers, meaning they bought players for more money than for what they sold
players. Spending money in the transfer windows has become a necessity to be competitive
on an elite level. During the years transfer sums have continued to increase on pace with the
revenue of teams (Pérez-Gonzaléz et al 2015). With sell on-fees, release-clauses and longer
contracts, teams have worked around ways to let players leave on their own terms making it
more imminent for teams to spend money on transfers. A more central market with teams
more focused on the winning criteria than the financial criteria have emerged. In what way
will a global pandemic impact the spending? With leagues postponing games to eventually
play them without any paying fans in the stands, with other teams in the lower tiers
discussing bankruptcy and financial despair, can Premier League teams keep put winning
trophies ahead of financial security? Or will there be a change in the amount of money spent
on the transfer market when teams lose a big part of their revenue today and perhaps bigger
losses tomorrow?

                                                                                             9(29)
4 Hypothesis
As seen in the previous sections regarding earlier literature, different aspect impacts the way
clubs act on the transfer market, what players they buy, what price they pay and to which
degree they overspend on the players. In this section the hypothesis behind this essay will be
discussed and in which ways the previous literature can be used to answer the research
question at hand.

As described by Depken II and Globan (2020) English teams tend to overspend on players.
They call this overspending a premium which occurs to many of the “top five” leagues but
mostly to the Premier League teams in England. To look at how the spending on the transfer
market has changed after the shock, this thesis will take a similar approach and use the
market values of players to evaluate how these premiums may have been altered. However,
this will be done in a slightly different way with a complete focus on the English teams, using
the same theory about a shock that would change the average prices of football players. In
Depken II and Globans’ (2020) case, the 2012 Premier League broadcasting deal, and in this
case the shutdown and impact of the pandemics initial shock. The revenue loss could in this
circumstance be interpreted as an opposite effect compared to the broadcasting deal, a shock
that would see premiums become lower. However, this shock did not impact only English
teams but other teams across Europe and the world too. Suggesting that the premium that
Depken II and Globan (2020) observed may still exist only slightly altered and thus very
difficult to reassess. Alternatively, then, the usage of one league and the steady market values
from Transfermarkt.de could give more clarity in how spending has changed after the shock.

To see what changes followed the shock of the pandemic, one must keep in mind the theories
of Rottenberg (1956) and Ruijg and Ophem (2015). They state that not only player
characteristics determines how players move from team to team. The size of the club matters
as well as the mind set of putting winning football games ahead of making money in many
cases. This can lead to a smaller effect of the shock since teams still are set on signing new
players to win “on the field”. Especially the top teams of the Premier League who in the
views of Frick (2007) would spend more money since they are the wealthiest clubs.

Pérez-Gonzaléz et al (2015) established that valuations of players on the transfer market go
hand in hand with the revenue of the clubs to a certain extent. Since the average match day
revenue was around 17 %, will there then be an equivalent decrease in the prices for players

                                                                                            10(29)
or is this overshadowed by the desire to win football games? This thesis aims to answer this
question.

Using these theories and hypothesises, the question of how English Premier League teams
respond to the exogenous shock of the pandemic on the transfer market can be answered.

5 Data
The data regarding market values of players used in this article and in most other academic
and popular literature comes from the German webpage Transfermarkt.de. (Depken II and
Globan 2020; Perez-Gonzalez et al 2020) Transfermarkt uses crowd sourced estimated where
community members of the site assign and estimate the transfer values of players. The value
of each player is weighted against its average among the members. The value is then
inspected and determined by a few “judges” who can alter the weight to omit bias and
incorrect values. This system is preferred by scholars as it removes arbitrariness and bias of
experts and judges. Transfermarkt is also the only site to offer these so-called market values
for players outside of the “European top five”. (Depken II and Globan 2020) Something of
great importance in this thesis as players across all continents, countries and leagues can and
have been purchased by different English Premier League clubs. Although, these values can
fluctuate, especially when there are no games, most leagues and teams had a chance to play
before both transfer windows opened meaning that the market values once again reflected
playing players.

Not only market values of players are collected from Transfermarkt.de, but also the actual
values of the transfers. The fees are reported from either the clubs themselves or agents of
players. Both market values and transfer sums are calculated in Euros (€). Additional
information on players were also gathered from Transfermarkt.de, information such as age,
country of birth, previous league, position, contract remaining when signed and team signed
for. This additional data will help determine more closely what effect the pandemic had on
the different transfers.

However, even with their extensive collection of market value and transfer sums, some
values and fees are missing. All players transferred with an undisclosed fee and/or without a
given market value are discarded. The same is done with players on a free transfer as their
transfer fee is equal to zero in the data, while their market value reflect their actual value.

                                                                                                  11(29)
Loan players with or without loan fees are also discarded since they are not transferred
permanently to the club.

The clubs that are researched are the 17 teams that competed in both the 2019/2020 season
and the 2020/21 season. Meaning that the three teams that were relegated to the second
division “The Championship” in 2019/2020 and the three teams that were promoted to the
Premier League in 2020/2021 are not included in the data. This is in order not to confuse
transfer data from the Championship, the second tier of English football, with that of the
Premier League since the differences in revenue between the two is very large. (Deloitte,
Annual Review of Football Finance 2020)

Going into more detail about the explanatory variables used in the thesis to further explain
and motivate the coming results, again looking at data collected from Transfermarkt.de.
There will be a total of 6 extra variables that will explain and help advance the data: age,
origin, position, contract length, player from a “Top Five” league and player joining a “Big
Six” club.

Firstly, considering the age of the player, as mentioned by Frick (2007), the age of the player
transferred plays an important role in when and for how much a player will be sold. Age will
be divided into 4 different dummy categories with “Young” containing players aged 17 to 22
years of age, “Midage” containing players aged 23 to 26, “Prime” 27 to 30 and lastly “Old”
containing player aged 31 and up. These categories were established to distinguish players
based on their age and appeal on the transfer market. The second variable used will consider
the origin of the player if he is from England or not, hence explaining some biased within the
Premier League towards domestic players (Depken II and Globan 2020). Thirdly examining
the players position as it explains the price tag of the player according to both Ruijg and
Ophem (2015) and Frick (2007) since goalkeepers may go for a lower price and those who
score more frequently will have a higher price. Therefore, the players will be divided into
four categories here: Namely goalkeepers, defenders, midfielders, and attackers1. This makes
it possible to differentiate those who are more likely to score and those who are not, as well
as putting goalkeepers into a separate category. Moreover, to determine the hypothesis of
Frick (2007) that after the Bosman ruling the remaining contract of players will play a large
role in the transfer sum, categorizing the remaining contracts of the players bought by
English Premier League teams. According to Frick (2007) players with a longer contract

1
    In the category “attackers” strikers, centre forwards and wingers are included.

                                                                                               12(29)
remaining will be more expensive and vice versa. Thus, the categories here will be: less than
one year remaining, less than two years but more than one year remaining, less than three
years but more than two years remaining and lastly players with more than three years
remaining on their contract, as clubs may become more desperate to sell players before their
contracts runs out and accept a lower transfer fee. The last two variables concern the players’
destination and previous destination. Namely whether the players previous club address
where a “Top Five” club or if their destination where a “Big Six” club. Since players from
bigger leagues such as the “Top Five” leagues demand higher sums or premiums on players
(Depken II and Globan 2020), there might be a higher sum on these players as well as more
of these observations seeing that players with higher fees tend to be sold more frequently than
those who do not (Ruijg and Ophem 2015). Regarding the “Big Six” clubs, the usage of this
variable is important to understand Rottenberg (1956) and Carmichael and Thomas (1993)
who all concluded that the size of the club in question, determine how active one is on the
transfer market. This in hand with the fact that some of the “Big Six” clubs tend to pay a
higher premium for players than the other teams do (Depken II and Globan 2020).

Here some descriptive statistics from these variables are presented in Table 5.1:

Table 5.1.

                                    (1)          (2)     (3)     (4)        (5)
VARIABLES                            N          mean      sd     min        max
Age                                 300        23.58    3.282    17         33
English Player                      300        0.247    0.432     0          1
Top Five                            300        0.407    0.492     0          1
Big Six                             300        0.333    0.472     0          1
Attacker                            300        0.327    0.470     0          1
Defender                            300        0.353    0.479     0          1
Goalkeeper                          300        0.0600   0.238     0          1
Midfielder                          300        0.260    0.439     0          1
Remaining Contract                  280        2.293    1.033     1          4
Contract < 1 year                   280        0.264    0.442     0          1
Contract < 2 years                  280        0.343    0.476     0          1
Contract < 3 years                  280        0.229    0.421     0          1
Contract < 4 years                  280        0.164    0.371     0          1
Young                               300        0.160    0.367     0          1
Midage                              300        0.600    0.491     0          1
Prime                               300        0.160    0.367     0          1
Old                                 300        0.0333   0.180     0          1
Table 5.1 Descriptive statistics of the covariates.

                                                                                             13(29)
From the table it is observable that the average age of players transferred are between 23 and
24 years old. Were 16 % were categorized as Young, 60 % as Midage, 16 % as Prime and 3
% as Old. About 25 % come from England and 40 % had a previous club address within the
top five leagues of Europe. One third of the transfers went to a top six club. 33 % where
attackers, 35 % defenders, 6 % goalkeepers and 26 % midfielders. The average contract
remaining of a player transferred was above 2 years. 26 % had less than one year left on the
contract, 34 % had between one and two years left, 23 % had between two and three years
left and the remaining 16 % had more than three years left on their contracts. The coefficients
of these covariates when added to the difference-in-differences regression, can be found in
Table 10.6 in the appendix.

To make sure there is a common trend of activity of Premier League teams on the transfer
market, previous transfer data from Transfermarkt.de has been used and constructed here in
Figure 5.1. This is of great importance as it reveals that the spending on the transfer market in
the Premier League has been steady over the years. Something that is essential when
performing a difference-in-differences estimation, as the determination that the changes are
due to an interfering factor and not some commonly known fact can be made. The figure in
question also illustrates the increasing importance of the transfer market for the clubs in the
English Premier League.

Figure 5.1.

                                        I N C O M E A N D E X P E N D I T UR E O N T H E
                                           T R A N S F E R M A R K E T 11 / 1 2 - 2 0 / 2 1
                                                      Expenditure          Income            Linjär (Expenditure)

          2500
                                                                                               2180

          2000                                                                                                       1800
 Millions of Euro

                                                                                    1660                  1650
                                                                         1470                                                   1530
          1500                                                                                    1340
                                                              1230

                                                  924.38                              871.87                           923.09
          1000
                                      775.03
                          638.55                                671.99     700.84
                                                                                                            577.42
                             428.95                                                                                               495.35
                    500                  372.53      406.26

                      0
                           11/12.      12/13.       13/14      14/15      15/16      16/17      17/18      18/19      19/20      20/21 Season

Figure 5.1. Income and Expenditure on the transfer market through 2011 to 2020.

                                                                                                                                       14(29)
In Figure 5.1 the overall income and expenditure of the 20 English Premier League teams on
the transfer market is illustrated. On the X-axis, the seasons who each contain two transfer
windows (period 1 and period 2), and on the Y-axis, million in Euros (€). Over the last ten
seasons expenditure has increased rather steady except for the 17/18 season and the 20/21
season. In hand with this increase in expenditure, income from the transfer market has also
increased for the Premier League teams. (Transfermarkt.de 2021)

Figure 5.2

                                M A R K E T VA L U E V S T R A N S F E R F E E S
                                             Market Value    Expenditure

             20/21                                                           1102
                                                                                      1315.5
    Season

             19/20                                                                     1327.925
                                                                                                         1627.99

             18/19                                                 966.425
                                                                                     1284.36

                     0   200    400        600        800       1000          1200     1400       1600        1800
                                                  Millions of Euros

Figure 5.2. Market values and Transfer fees through 2018 to 2020

In Figure 5.2 the expenditure of the 172 clubs in yellow and the combined market values of
the players bought in red throughout the three previous seasons is illustrated. The market
values are relatively lower than the transfer fees over all three seasons.

6 Empirical strategy
Here in this section the methods and strategy of how to use the data will be presented. How
can one showcase the effect the exogenous shock of the pandemic on footballs transfer
market? All regressions and tables have been made using Stata and Excel.

The first wave Covid-19 pandemic hit Europe in early 2020, and by March football leagues
across Europe saw no other option than to postpone play. This postponement caused revenue
from tickets and broadcasting deals to be stopped as well. Implying that teams were losing

2
 In the 2018/2019 only 15 teams were looked at. Since 2 of them were yet to be promoted to the Premier
League.

                                                                                                           15(29)
out on most of their revenue streams during the period of lockdown. Since the English
Premier League restarted, the share of revenue from broadcasting returned. However, the
match day stream remained absent since no fans could attend the games. This in hand with
the uncertainty that the future holds for every club, gives reason to believe that there will be
some changes to how teams act. The pandemic can therefore be used as a natural experiment
on how the revenue losses it caused affected the Premier League team’s interaction with the
transfer market in terms of overspending the market values.

As many have witnessed lately, markets around the globe have seen significant changes in
how business is being made, in which volumes it is being made and for how much. The
transfer market of professional football is no exception to this. With less income and more
uncertainty, there tend to be less activity in spending on markets. The pandemic can also
bring other effects to the transfer market like uncertainty of players’ attributes, with
restrictions on travels and games, team scouts have been forced to find other ways to evaluate
players. Insinuating some transfers could be less thought through or not happen at all.

This thesis uses a difference-in-differences approach to evaluate whether the Covid-19
pandemic caused English Premier League teams to spend less money on players in the two
transfer windows since the start of the pandemic. The difference-in-differences model
facilitate to see the difference between how the English teams spent money on the transfer
market before the initial pandemic shock and how they spent money after. Using data from
Transfermarkt.de and their large database of players’ market values at the time of their
transfers, estimating a difference-in-differences using these market values and the transfer
sums paid by the English teams can be constructed. Obtaining the actual values of players
that were transferred to or within the Premier League as the treatment group and using the
market values of these players from Transfermarkt.de as the control group. The shift of how
teams have spent over the market values after the initial shock of the pandemic is then
visible. Much like Sakai et al (2018) did with the Great East Japan earthquake and Leiter et al
(2009) concerning floods in Europe, taking these groups and examine them before and after
the shock. In the case of the transfer markets, there is a very clear divide as players are only
allowed to be transferred during the month of January and after the season during the summer
in England (FIFA Regulations on the Status and Transfer of Player 2020). Meaning that there
is no coinciding transfer window during the shock. Therefore, one can clearly distinguish
between the transfer periods before the shock and the transfer periods after the shock.

                                                                                             16(29)
To use a difference-in-differences approach, several criteria must be considered and valid. As
seen in the previous section the assumption of common trend holds. Moreover, one must take
into consideration the assumption of parallel trends. However, the extension of data is limited
in the sense that it is difficult to obtain data for every player transferred over the previous
years. Furthermore, all assumptions for an OLS-regression apply to difference-in-differences
models as well and since the difference-in-differences or DID approach has its origin in an
OLS regression, it will take the following form illustrated below in Equation 6.1:

lnOverspendingt = β0 + β1 (Dummy20)t + β2 (Dummytransfer)t + β3                               (1)

(interaction)t +εt

Here the dependent variable lnOverspending is how much the English Premier League teams
tend to overspend the market value of a player at the time t, in percent. Ln is being used to
determine the percent shift in the change. The three independent variables are Dummy20
which is a dummy variable indicating that if it has the value of one, the market value or
transfer value occurred in the 2020/2021 season after the shock. A value of zero would
indicate a value that occurred before the shock. The second independent variable is also a
dummy which determines whether the value is a market value or transfer sum. Zero for
market value and one for transfer sum. The final independent variable is the interaction,
which is the difference-in-differences coefficient, it essentially shows the difference between
before the shock and after the shock. Lastly, the “εt“ is the error term.

To be able to explain more of the regression several covariates are added. This will help in
order to illustrate on which variables the changes depend the most. Therefore, the regression
will look like Equation 6.2 below:

lnOverspendingt = β0 + β1 (Dummy20)t + β2 (Dummytransfer)t + β3 (interaction)t +
                                                                                                           (2 )
β4 (DYoung)t + β5 (DMidage)t + β6 (DPrime)t + β7 (DEngland)t + εt

The newly added variables in Equation 6.2 are firstly regarding age, DYoung, DMidage,
DPrime and DOld. Contain player with age spanning from 17 to 22, 23 to 26, 27 to 30 and 31
to 33, respectively. DOld is removed to avoid the dummy variable trap. These variables are
also dummy variables, meaning they have the value 1 if they are true and 0 if they are not.
Same is for DEngland which is equal to 1 if said player have England as their country of
origin and 0 if not.

                                                                                                  17(29)
Equation 6.3 will have another two variables added and takes the shape of the following:

lnOverspendingt = β0 + β1 (Dummy20)t + β2 (Dummytransfer)t + β3 (interaction)t +
                                                                                                      (3)
β4 (DYoung)t + β5 (DMidage)t + β6 (DPrime)t + β7 (DEngland)t + β8 (DPositionA)t +
β9 (DpositionD)t + β10 (DpositionG)t + β11 (DContract1)t + β12 (DContract2)t + β13
(DContract3)t +εt

In this regression the addition of dummy variables for position and remaining contracts when
signing is made. Positions are A for attacker, D for defender and G for goalkeeper. M for
midfielder is omitted to avoid dummy variable trap. The remaining contract dummy variables
are DContract1 for under one year remaining on the contract, DContract2 for under two years
but above one year and DContract3 for under three years but above two years. Over three
years are omitted to avoid dummy variable trap.

The last regression used in this thesis will look at the important aspects of from where players
are bought, and which teams are buying them. Adding an additional two more variable to the
model in Equation 6.4. below:

lnOverspendingt = β0 + β1 (Dummy20)t + β2 (Dummytransfer)t + β3 (interaction)t +
β4 (DYoung)t + β5 (DMidage)t + β6 (DPrime)t + β7 (DEngland)t + β8 (DPositionA)t +                     (4)

β9 (DpositionD)t + β10 (DpositionG)t + β11 (DContract1)t + β12 (DContract2)t + β13
(DContract3)t + β14 (DTopFive)t + β15 (DBigSix)t +εt

The two variables added above in Equation 6.4 is DTopFive where the player has a value of 1
if they are transferred from a team which plays in one of Europe´s “top five” leagues and 0 if
they are transferred from any other league. DBigSix providing a value of 1 if the player is
transferred to a “Big Six” club and the value 0 if transferred to any other clubs in the Premier
League.

An aspect that is important to take into consideration, is that even though the shock of the
pandemic came in early 2020, between two transfer periods like mentioned earlier, the effect
of the pandemic is still eminent in the time of this thesis, which could lead to limited results
as the consequences are yet to be fully seen. Nevertheless, this thesis will determine the
difference-in-differences in the initial shock of the pandemic, where leagues were in
shutdown completely and not just played in empty stadiums. Another concern which this

                                                                                             18(29)
model does not consider is the changing of market values. As play halted under a few months
no new player statistics were gathered on players, meaning there was no way for a player to
impact his market value on the pitch. However, as Drewers et al (2020) states players can
impact their own market values on social media and other platforms, and in hand with the
relatively short stoppage of time, decreasing market values should not be a major factor.
Especially not since the stoppage of play where relatively short and restarted before the actual
transfer window transpired.

7 Results
In this section the results of the thesis will be discussed. After running the equations from the
previous section, one can establish how the transfers have changed since the pandemics initial
shock. All equations from section six will be included and discussed below.

Firstly, looking at team and league averages. In the Graph 7.1 below, a time series of the total
mean market value (MV) and the total mean expenditure (Exp) per team, for the Premier
League clubs surveyed under the three seasons is shown. Again, market value is the control
group and expenditure is the treatment group. This figure is in line with the estimates
provided below. An effect between the 2019 and the 2020 season where the gap between the
two lines decreases is visible. Implying that there is an effect of the pandemics shock on the
transfer market. The reason for the decrease of the market value line as well lies in the fact
that an overall less amount of money was spent on players suggesting that also the total mean
of market values where down after the pandemic. This graph provides a decent illustration of
a parallel trend using the data that is obtainable.

                                                                                             19(29)
Graph 7.1.

                                          Time series
    100

                                                                 Initial Pandemic Shock
    90
    80
    70
    60

       18
     2018/2019                                 2019/2020                                20
                                                                                   2020/2021
                                               Season

                                    (mean) MV                  (mean) Exp

Graph 7.1 Time series showcasing the decrease in expenditure after March 2020.

Below in Table 7.1 the regressions mentioned in the previous section is illustrated in order of
equation one through four, with lnOverspending as the dependent variable. The nine different
rows consist of the Difference-in-differences value and its standard error in parenthesis,
number of observations, R-squared value, mean for the control variable before the shock,
mean for the treatment variable before the shock, their difference before the shock, mean for
the control variable after the shock, mean for the treated variable after the shock and their
differences after the shock. The standard errors are robust.

                                                                                                20(29)
Table 7.1.
                                    (1)            (2)            (3)            (4)
 VARIABLES                     lnOverspending lnOverspending lnOverspending lnOverspending

 Diff-in-diff                      -0.0892       -0.0892           -0.118            -0.118
                                   (0.273)       (0.267)           (0.261)           (0.227)

 Observations                      300             300               280               280
 R-squared                        0.021           0.077             0.244             0.433
 Mean control t(0)                2.267           2.233             3.040             2.187
 Mean treated t(0)                2.613           2.579             3.380             2.528
 Diff t(0)                        0.346           0.346             0.341             0.341
 Mean control t(1)                2.147           2.178             3.153             2.205
 Mean treated t(1)                2.404           2.435             3.375             2.427
 Diff t(1)                        0.257           0.257             0.223             0.223
 Robust standard errors in parentheses
 *** p
into four categories the first being players with less than one year left on their contract,
second with less than two years and so on.

Looking at column 3 and Equation 3, once again a higher R-squared and slightly lower
standard errors are produced. Meaning once again that the model has been improved. The
loss of observations is caused by some players lacked information of their length of contract
on Transfermarkt.de, hence the difference-in-differences value has gone from about a
negative 9 percent to a negative 11.8 percent. When adding the contracts of players purchased
and their playing positions, using the previous work from Carmichael et al (1999) that
attacking players might go for a larger fee than others and that goalkeepers tend to go for less
(Ruijg and Ophem 2015), this is reconfirmed in this model. Also, the determination that
contracts do matter quite a bit regarding transfer sums can be made. Players with longer
remaining contracts were sold for much larger sums than those who only had one or two
years left, much in line with the hypothesis of Frick (2007). Clubs will agree on selling
players for less instead of risking that a player leaves as a “Bosman” for free when the
contract have ended.

Lastly, column 4 add information about two very important aspects, if the player came from
one of the “top five” leagues in Europe and if the player was transferred to a team within the
“big six”.

These additions have once again providing a decrease in the robust standard errors and an
increase in the R-squared value. Now a standard error of about 22.7 percent and an R-squared
of 0.433 can be seen. While the standard error is still quite high and unsignificant, one can
see that about half of the situation is being explained by this model. The difference-in-
differences still stand on a negative 11.8 percent. The implements of the big six dummy and
the top five dummy provides a clear divide for the more expensive players, since the players
from the “top five” leagues tend to be much more expensive than players from other leagues.
Similarly, those players arriving to the big six clubs of England tend to have a higher fee,
both because of premiums (Depken II and Globan 2020) and that players with higher fees
tend to be transferred more often. (Carmichael et al 1999)

This implies that after the shock, players were sold on average 11.8 percent cheaper
compared to their market values as they did before. Going from a difference of 34.1 percent
prior to the shock to a 22.3 percent difference after the initial shock of the pandemic.

                                                                                               22(29)
Briefly looking at the case of pure numbers, per players on average, and changing the
dependent variable “lnOverspending” and instead shifting to look at “Overspending” in
millions of Euros as the dependent variable, this produces the following result in Table 7.2
below:

Table 7.2

                               (3)
 VARIABLES                Overspending

 Diff-in-diff                -0.951
                             (3.161)

 Observations                280
 R-squared                  0.482
 Mean control t(0)          8.927
 Mean treated t(0)          12.43
 Diff t(0)                  3.501
 Mean control t(1)          8.185
 Mean treated t(1)          10.73
 Diff t(1)                  2.549
 Standard errors in parentheses
 *** p
Table 7.3

                                 (1)              (2)               (3)                (4)                 (5)
  VARIABLES                lnOverspendi lnOverspendi lnOverspendin lnOverspendin                     Overspending
                               ng           ng             g             g

   Dummy20                     -0.120           -0.055             0.113              0.017              -0.461
                              (0.535)           (0.770)           (0.550)            (0.916)             (0.842)
 Dummytransfer                0.346*            0.346*            0.341*            0.341**               3.501
                              (0.070)           (0.064)           (0.057)            (0.029)             (0.109)
Interaction (DID)             -0.089            -0.089            -0.118             -0.118              -0.951
                              (0.744)           (0.738)           (0.650)            (0.602)             (0.764)
    DEngland                                   -0.356**           -0.013             0.266*              3.561*
                                                (0.023)           (0.939)            (0.068)             (0.082)
     DYoung                                     -0.437          -0.820***          -0.740***            -8.443**
                                                (0.135)           (0.005)            (0.004)             (0.019)
    DMidage                                      0.251            -0.005              0.026               0.476
                                                (0.321)           (0.985)            (0.905)             (0.874)
     DPrime                                      0.059             0.048              0.240               1.804
                                                (0.840)           (0.865)            (0.337)              (0.607)
   DPositionA                                                     0.347**           0.479***               2.019
                                                                  (0.047)            (0.002)              (0.347)
   DPositionD                                                      -0.120            -0.106              -4.970**
                                                                  (0.485)            (0.487)              (0.021)
   DPositionG                                                    -0.819**          -0.935***           -14.201***
                                                                  (0.013)            (0.001)              (0.000)
   DContract1                                                   -1.404***          -1.258***           -16.184***
                                                                  (0.000)            (0.000)              (0.000)
   DContract2                                                   -0.855***          -0.598***            -8.363***
                                                                  (0.000)            (0.001)              (0.001)
   DContract3                                                    -0.559**          -0.511***            -7.408***
                                                                  (0.011)            (0.008)            (0.006)
    DTopFive                                                                        0.740***           10.926***
                                                                                     (0.000)            (0.000)
     DBigSix                                                                        0.842***           16.705***
                                                                                     (0.000)            (0.000)
     Constant                2.267***         2.233***           3.040***           2.187***           17.366***
                              (0.000)          (0.000)            (0.000)            (0.000)            (0.000)

                                       Robust P-values in parentheses
                                       *** p
What can be drawn from this table is foremost the important P-values of the Interaction
variable/DID variable. The P-value of Equation 4 is about 0.602 which is only statistically
significant at a 61 percent level, meaning that there is more than a 60 percent probability that
the results are untrue. However, some additions made throughout the equations were
statistically significant such as the additions of contracts remaining, arrivals from the “Top
Five” leagues and the arrivals to the “Big Six” clubs.

Lastly when looking at per team spending where the 17 teams were studied, an on average
5.09 million Euros less were spent over the market value compared to the season of
2019/2020. While considering the season of 2018/2019 as well, the overspending had
decreased with about 6.9 million Euros per team. In-detail tables and estimations of this can
be found in the appendix.

In essence what has been found is, that there has been a change in how Premier League teams
act on the transfer market after the pandemics first initial shock. The uncertainty of the future
and impact of loss of match day revenue have shown to be factors contributing to about an
on-average 11.8 percent lower transfer value per player, when considering player contracts,
positions, country of birth, age, previous leagues, and the teams where they ended up. This is
about 5 percent less of the loss in revenue in match day tickets, implying that some teams
may still have overspent on the transfer market, explaining about 43 percent of the situation
at hand. However, this does not show statistical significance on a reasonable level.

8 Conclusion
In this final section, the conclusion of the main findings will be discussed, also the
determination of what could have been done differently as well as suggestions for further
research.
This thesis has researched the initial shock of the first wave of the covid-19 pandemic and its
effect on the transfer market of English football, using transfer sums and market values from
Transfermarkt.de in the transfer windows prior to the shock and after. Determining not the
change of the transfer sums but rather their changes compared to market values of the players
sold. Statistically unsignificant evidence for an average decrease in transfer sums of about
11.8 percent per player transferred to the Premier League, compared to their market values,
were observed. It is however difficult to determine if this is because teams lack funds or that
players have become cheaper on the market in general, since the pandemic. On the one hand

                                                                                             25(29)
You can also read