Soccermetrics Interview #1: Erik van den Berg

Erik van den BergErik van den Berg is a recently graduated Economics student from Rotterdam who wrote a fantastic thesis on transfer leagues in soccer.  I read the paper at around the same time that I was thinking about publishing a newsletter, and I knew right then that Erik would be a perfect first interview.

[Interview originally conducted 14 March 2012.]

(Howard) So, tell me about yourself.

(Erik) I’m a 25-year-old graduate in finance from Erasmus University with a huge passion for sports in general and football in particular.  I was born and raised in the Dutch city of Rotterdam, but have also lived a year abroad in Sydney, Australia. I have done a fair amount of travelling to Australia, South America, The United States, Southeast Asia and within Europe and as such am looking to move abroad in the coming years. As such, I am also interested in a great variety of sports worldwide, but football, baseball, F1 racing and golf are my big passions currently.

How did you decide upon this thesis topic?

As a boy I have always followed Dutch and international football religiously and as this passion did not wear off in my years of college, I decided to apply my knowledge in Finance to the subject. I believed that original research topics would pose the most challenges, but would also be the most interesting and enjoyable. As a Finance student, the football transfer market was especially interesting, as on the surface it was full of inefficiencies. I had a natural interest in what drove football clubs to dole out these huge sums for new players, and thankfully my thesis advisor indulged me in letting me attack this subject.

One thing that I’ve observed about sport in Europe, as an American, is that it’s much more difficult to obtain payroll data from teams.  Is that in fact the case?  How easy was it to get your data?

The most obvious problem in my research was indeed the collection of data, as it is mostly proprietary and the public sources that exist are not always reliable. I attempted to tap into private sources of data but unfortunately had no success, so I was left with public transfer and player data which were painfully difficult to collect. Player salaries and team payrolls where an important element of my initial model regarding player transfers, but these turned out to be impossible to find, especially on the level of individual player salaries. For team payrolls, accountants (e.g. Deloitte) and national football associations may be able to provide data, certainly with UEFA’s new financial fair play rules coming into effect in 2012-2013.

Another thing that I learned when I worked on some financial analytics is that there are big differences between the publicly-revealed transfer or salary amounts and the reality.  Did you encounter the same thing in your public transfer data, and how robust are your models to uncertainties in the transfer figures?

Certainly there can be huge discrepancies in public transfer amounts and the reality. These differences are mainly caused by agent fees (which can be quite sizeable) and elevator clauses, which have become more popular in European transfer markets these last couple of years. In my research I chose to ignore this issue as I had no way of solving it. In my experience, these issues are most pronounced in the more expensive transfers. Agent fees for smaller transfers are often modest and elevator clauses are usually not an issue. Consequently, there might be a bias for larger transfers, as these transfer fees might be understated. As I said, I could not find a way to work around this issue, and I think it is one of the more significant problems that exists in this line of research.

You said in your thesis that the transfer market in football presents a perfect laboratory to study specialized economic markets.  Are there other aspects of football that are attractive to a research economist?

The social role and unique competitive environment of professional football make for a very interesting research laboratory so to say, especially with the increased amount of specialized data available on individual employees (i.e., players) . As such, I imagine football salaries to be of particular interest to labour economists, given the amount of data available on individual player performance. Also, competitive behaviour within professional football is of interest (tied to the topic of competitive balance as I touched upon in my thesis) given the unique nature of competition. Winner takes all competitions are repeated yearly, and financial performance of football clubs is increasingly hard to predict given unpredictable team performance and increased significance of other income from competitions like the Champions League.

To what extent does the transfer market fail to make sense from an economist’s point of view?

Of course, many transfer decisions are flawed as imperfect inputs with little predictive power are known to have been used. However, as more resources (i.e. money) flow into professional football, one would expect better transfer decisions to be made (parallel to the Moneyball paradigm shift in baseball from batting average, home runs etc to on-base percentage and such). From a general economic point of view, I don’t think this trend is inconsistent with rational expectations theory. Where transfer markets stop making sense for an economist is when transfer decisions are obviously made because of factors that are hard to incorporate in economic models such as fan pressure or rich individual owners being in ‘win-now’ mode. Another factor to consider herein is that irrational transfer decisions have a tendency to snowball to various other clubs, as they are endowed in unexpected wealth (for example study the effect of the Cristiano Ronaldo transfer to Real Madrid).

I know that you focused on the English Premier League, but do you believe that your findings translate well to other leagues in Europe?  How so? Or if not, how are they different?

I think that the comparability of transfer markets greatly depends on market size: as mentioned, in my mind, transfer decisions improve greatly given the amount of resources available to make them. Hence, I think that my results would translate pretty well to the Spanish, Italian, German and French leagues (and less so to, say, the Dutch market, where football front offices are much, much smaller). On the other hand, many factors differentiate transfer markets, especially ownership (English clubs being pre-dominantly foreign-owned), which will influence the nature of transfer decisions. Furthermore, regulatory differences can have a great impact: the Italian league is renowned for its restriction of foreign (or now, non-EU) players. This would cause Italian clubs to overprice EU-players, as they have become a more scarce commodity to them than to other leagues (however, other leagues can have similar restrictions, including the Premier League).

What do you foresee will be the biggest impact of the Financial Fair Play rules on valuations in the transfer market?

I think that Financial Fair Play, if implemented in an effective way (which is always questionable with UEFA and FIFA regulations), will have a huge impact on transfer fees, bringing them down considerably. These past few years, European football has been allowed to operate at a deficit of several hundreds of millions (UEFA has extensive reports on this). By effectively draining the transfer market of these millions, transfer prices will have to go down. In effect, Financial Fair Play forces owners to be of the ‘profit maximizing’ kind, disallowing for any other incentive to own a football club.

I was particularly interested in your different valuation models for players who had gone through the transfer market and those who remained in the club from academy to first-team (e.g. Messi, Xavi, Scholes). Besides the obvious (lack of a price tag for one-club players who have come up through the academy), what were the differences between the two models in terms of setup and the results?

In my quantitative research I have focused on the determinants of transferred players only. Hence, players such as Messi, Xavi and Scholes (who never transferred from one club to another) did not show up in my research.

The difference in valuation that I mentioned is purely based on accounting standards: transfer fees are included as an asset on the balance sheet and require annual write-offs, and ‘fair player value’ is not included in accounting standards and therefore is not included in a club’s assets. Hence, clubs with many homegrown players on their team (Barcelona FC obviously comes to mind) have a great deal of ‘hidden reserves’ on their balance sheet, as many of their assets are not included in accounting value. Obviously, one could use the (albeit limited) results of my study to give an indication of the value of non-transferred players.

Adaptations that probably should be made in this setting is that homegrown players are generally more popular with a club’s fan base compared to ‘imported’ players. Furthermore, one might imagine a larger performance prediction bias for homegrown players as they have only played at one club and as such have not been able to present their skills in different settings. Consequently, homegrown players might be worth more to their original club than to a prospective new club and as such are less expected to transfer (this is purely my personal speculation).

Could such valuations of non-transferred players be used to estimate the financial value of an academy program within a club?

I think that if you would want to estimate the value of an academy programme, non-transferred homegrown players should definitely be included in the analysis. Obviously, homegrown players that have been transferred should also be included in this analysis. However, I think that it is very difficult to accurately reflect the value of an academy to a club financially. The problem is that clubs in several leagues have some sort of monopoly on the acquisition of young talent. For example in The Netherlands (Spain is a good example as well), the national team consists of players from only 3 or 4 academies. Furthermore, the international competition for young talent these days already starts at the age of 16. Consequently, the designation of ‘homegrown’ players less and less reflects the quality of football academies. However, it is a very interesting topic for discussion.

What was the most surprising result from your thesis research?

In the end, I was most surprised to find that features of the selling club did not factor into the transfer price. Based on the interviews I conducted with Dutch football CFO’s, I expected this to be of huge influence. For example, the resources available to small clubs is very limited (especially in The Netherlands), and I expected their decision-making to be much more imperfect than bigger clubs (i.e. selling very good players for very small transfer fees).

What is the one thing that you want a reader — whether a fan, a journalist, or a club official — to take away from your research?

I think that professional football would benefit greatly from more detailed and specific (public) player statistics, following the American model so to say. As (disputably) ‘the world’s most popular sport’ it is surprising to see that all the public, journalists and front-office employees know about players is how many goals they’ve scored, minutes they have played and assists they have contributed. More specific statistics that more accurately reflect player performance would surely interest the public — and to be sure, various front-offices already employ more advanced player measures — and allow for a more comprehensive fan experience.

That’s especially true in North America, but still a challenge to generate fan interest in Europe, no?

Maybe I’m biased because I’m Dutch and football is our national sport, but as I see the level of fan interest that football generates in my country, I always think that fans would be more effectively served by more advanced statistical analyses. The level of analysis, especially on tv, is poor to say the least, often driven by the opinions of self-proclaimed experts and retired football players who have an ‘intangible’ understanding of the game. Herein, I do think there is a role for more advanced statistics. I think that football fans are not oblivious to these analyses, as they continuously tune in to football-related programmes.

But even in the front offices of the major European clubs, I’d venture that they still don’t know which performance indicators are undervalued in the market.  It seems to me that research on that topic is wide open.

I’ve seen a documentary on, amongst others, the level of statistical analysis at Manchester City FC and I have to say that I was impressed by the quality of analysis. As I understand, especially English clubs have, by now, a relatively strong pedigree in statistical analysis of players. However, the problem is that other European leagues, in my view, have no history in the use of statistical analysis whatsoever. What I’ve heard from front office people is that analyses still are predominantly scout-driven. As such, I agree that many major European leagues are clueless in the field of statistical analysis. That’s why I think it will be an interesting research topic for time to come.

So what are you doing now?

Currently I am looking to get into consulting, although I wouldn’t mind a front-office job in football, and obviously enthusiastically following my favorite sports.  I’m really looking forward to the new Formula 1 and MLB seasons and of course the climax of European football competitions.