What determines transfer values in football?

Erik van den Berg, “The Valuation of Human Capital in the Football Player Transfer Market: An investigation on transfer fees paid and received in the English Premier League”, M.Sc. Thesis, Erasmus School of Economics, Erasmus Universiteit Rotterdam, July 2011. [Link]


I was doing a search for something in preparation for my Advisory Board meeting and came across what is actually a fascinating thesis on one of the more nebulous and controversial elements of modern football business: the transfer market. Erik van den Berg was a M.Sc. student in the School of Economics at Erasmus Universiteit Rotterdam in the Netherlands who received his degree last summer. I don’t know if he decided to continue at the University for a Ph.D. or left to pursue employment — I couldn’t find information about him online — but his thesis presents an excellent exposition of transfer payments in a specialized and differentiated market in which values are determined by either asset characteristics, the context of the transaction, or both. He finds the perfect laboratory for his study in the English Premier League. In the case of the Premier League, asset characteristics are player performance metrics, and transaction contexts are the behaviors of buying and selling clubs.

There are two major parts of the thesis: a discussion of the transfer market in football, and an analysis of player valuations in the English Premier League. van den Berg presents the factors in which a transfer market operates: competition balance, the traditional business model oriented around spectator attendance, regulatory structures, and the player acquisition and selling process which includes the role of agents and compensation schemes. This part of the thesis is background for the second half of the publication, so I’m not going to focus too much on it, but it contains a literature search on sport finance that is worth reading.

I do wish to focus on the two schools of thought between Rottenberg and Sloane, two economists in the 1950s and 60s who were the first to study the economic behaviors of sports teams (baseball and soccer, respectively). Rottenberg claimed that a roughly equal distribution of talent was essential to ensuring unpredictability of sporting outcomes, and that team owners were motivated to maximize profits; the idea of an owner buying a team just for the thrill of owning a sports team made no sense to him. Sloane countered that, to the contrary, sports team owners could and often did assume ownership for the pleasure and ego, and sought to maximize wins or utility (success on the field and at the gate) even if it meant that the team assumed large quantities of debt. The implications of that claim are that football clubs are not run efficiently from a financial point of view, and if that’s the case then player transfers are probably inefficient as well.

The second part of the paper is an analysis of transfer payments made in the 2008-09 and 2009-10 English Premier League seasons. There are two objectives in this phase of the research: assess the viability of models that describe buying/selling behavior, and investigate the significance of performance and contextual variables on the transfer payment. The first model considered is a bargaining approach, in which the value of the player is estimated from the perspective of buyer and seller, a game theory algorithm is applied, and the model tested by a regression analysis. The second model is a non-bargaining approach that considers the value of all players, not just those who have been transferred, and considers the following inputs: innate talent, training/development, and additional revenue streams. The investigation of the significant variables was the most interesting and challenging part of the paper. It is clear that van den Berg is working with a limited data set, as are all researcher who don’t have access to the finely grained data from companies such as Opta or Prozone/Amisco. I believe there were about 20 variables that van den Berg considered in his regression model (age was used twice to model a parabolic relationship), and only four or five of them exhibited significant coefficients. The major result was that the following variables were the largest factors in transfer value:

  • Games played previous season
  • Goals scored
  • Age (up to a peak year, usually later 20s)
  • Domestic transfer
  • Size of buying club, modeled as stadium size
  • Participation of buying club in continental club competitions

van den Berg claims that the influence of buying club size in transfer payments explains why larger clubs who participate in the Champions League pay a premium for players. He explains the behavior as an expression of clubs being risk-averse, which I think means the aversion of clubs to pass up on the chance to buy said player lest he go to a rival. (I could be wrong with his interpretation, I need to convince myself of that again.) The bottom line is that player valuation, in case you weren’t already convinced, is very very messy and inefficient. That’s not a finding that people didn’t know, but it is useful that the numbers point that out. The next step is to find the inefficiencies in the market, which aren’t so obvious.

I would read this publication not just for the analysis, but also for the references therein. There are references to major papers on sport finance in general, and the soccer transfer market in particular. van den Berg progressed very far in his research with imperfect and sparse data. As in-match performance data becomes more widely used at clubs and other end-users, we could start to see more research on the behavior of buyers and sellers in the transfer market and how these actions can be exploited to make smarter choices.