Originally written: September 2003
Professional football is big business these days. The creation of the FA Premier League in 1992 marked the start of a sustained period of economic boom in the industry. The engine of growth has been the value of sports TV rights driven by intense competition from competing channels and platforms that regarded Premiership football as the “battering-ram” to win subscribers and advertisers. Fans have flocked back to new or redeveloped all-seater stadiums. The downward trend in gate attendances since the late 1940s has been reversed and league gates have returned to levels last seen in the early 1970s. Increased product demand by fans, TV companies and sponsors has generated revenue growth in excess of 20 per cent annually over the last decade. The principal beneficiaries have been the top players with wage growth over the period in the range of 25 – 30 per cent annually largely due to structural changes in the players’ labour market leading to a greater degree of free agency. Yet despite the increasing economic importance of professional football, it is still far from obvious that the industry is populated by rational economic agents. Vanity often triumphs over sanity in the management of football clubs.
Professional team sports offer an excellent potential research site for economics and other business-related disciplines. It is a highly transparent industry in which key individuals are easily identified and the production process is largely carried out in the public domain. Investment decisions, especially the purchase of new players, are widely reported. Player and team performance is easily tracked. It is a data-rich and news-rich environment. Production statistics (i.e. match results) are reported weekly or even twice weekly. There are enormous opportunities to use professional team sports to investigate theoretical hypotheses concerning production processes and the determinants of organisational performance. For example, Berman et al. (2002) use the National Basketball Association to test the role of tacit knowledge (measured as shared team experience) as a source of sustainable competitive advantage. Publicly-listed sports teams have provided considerable scope for examining how stock markets react to new information, particularly given the diversity and frequency of sporting and financial news. The sports betting market also offers opportunities to investigate the efficiency with which available information is utilised. (Dobson and Goddard, 2001, discuss the efficiency of football share prices and betting odds.)
The economic analysis of sport has a relatively long history. The earliest academic discussions of the economic aspects of sport are to be found in the US legal literature on the relevance of anti-trust laws to professional sports. The analytical basis of the economics of sport was provided by three seminal papers published in top US economics journals: Rottenberg (1956), Neale (1964) and Scully (1974). Rottenberg’s 1956 paper on the players’ labour market in major league baseball was a wide ranging paper touching on many aspects of professional sports including the determinants of gate attendances, and the impact of labour market regulations on competitive balance in sports leagues. Rottenberg took the view that major league baseball is a collusive combination with rules that inhibit competition. His principal contribution has become known as the invariance proposition, and represents an application of the Coase theorem. Rottenberg argued that the equilibrium distribution of playing talent across teams would be invariant to the regulatory structure (i.e. allocation of property rights) of the players’ labour market. The reserve rule in major league baseball creates a monopsony in which teams have exclusive rights to retain their out-of-contract players, thereby preventing players becoming free agents and negotiating with rival teams on the expiry of their current contracts. Rottenberg showed that irrespective of whether a reserve-rule of free-agency regime exists, the pattern of player migration between teams is unaffected so long as teams and players are wealth maximisers. Under both regimes, players will move to those teams that are expected to generate the highest MRP[1] from their playing services. The labour market regime only affects the share distribution of the player’s MRP between the player and the team owners.
A corollary of the invariance proposition is that the equilibrium distribution of playing talent is also invariant to the revenue-distribution regime operated by a sports league. Changing the allocation of revenue property rights to bring about proportionate revenue redistribution between teams has no impact on the relative economic incentives of teams. Again, the only impact is on the share distribution of the player’s MRP.
Neale’s 1964 paper highlights the specificity of sport arising from the peculiar economics of professional sport. The specificity of sport is principally derived from the jointness of production and the Louis-Schmelling (or New York Yankees) paradox. A sports contest is produced by two independent teams; a sports tournament requires many teams. In addition, to the extent that fans are attracted by the uncertainty of outcome, economic profit for a sports leagues as a whole is enhanced by increased sporting competition (i.e. a more equal distribution of playing talent) rather than sporting dominance. Unlike other industries in which monopolistic domination yields supra-normal profits, in the professional sports industry monopolistic domination reduces uncertainty of outcome resulting in a potentially significant revenue loss. This is sometimes referred to as the Louis-Schmelling paradox. Joe Louis, the heavyweight boxing champion, needed a credible contender to maximise his box-office value. The New York Yankees paradox relates to the observation that gate attendances at Yankees games began to decline during their dominance of major league baseball in the immediate post-war era, only to recover when the Yankees became less dominant and their games were no longer foregone conclusions.
The jointness of production in sporting contests necessarily implies that property rights are ill-defined. The value of a single contest depends not only on the two competing teams but also the context of the tournament as a whole. The collective nature of the production process combined with the collective concern with competitive balance creates a peculiar situation in which economic collusion between teams is potentially pro-competitive. This issue is of huge practical concern currently as regulatory authorities increasingly view sports leagues as anti-competitive cartels. The EC is opposed to the existing system of collective selling of TV rights by the FA Premier League and seems to favour a system in which teams individually sell the rights to the games staged in their own stadium. Arguments for the individualisation of TV rights typically ignore the fact that teams are not the sole producers of their home games. Individualisation of TV rights could be accompanied by a league tax system but the enforcement problems and transaction costs are likely to be very significant. At the very least, there is a good case for regulatory economists to give more weight to the specificity of sport and practical implementation problems of rights individualisation in their assessment of the costs and benefits of the collective selling of TV rights.
The final seminal paper is Scully’s 1974 AER paper on pay and performance in major league baseball. This paper represents the first major application of econometric analysis to sports. Scully estimated the MRPs of baseball players using a two-equation estimation procedure. He first estimated a team-production equation to determine the relationship between team win percentages and overall team hitting and pitching performances. He then estimated a team-revenue equation to determine the relationship between team revenue and team win percentages. He used these two equations to infer the gross MRPs of individual hitters and pitchers. Net MRPs were calculated by subtracting training, capital and other related costs. Using estimated salary equations for hitters and pitchers, Scully calculated the degree of monopsonistic exploitation under the reserve rule. He found that average players tended to receive around 20 per cent of net MRP whereas star players received around 15 per cent of net MRP.
Scully’s analysis of the pay-performance relationship in major league baseball became a benchmark study in the economics of sport. Subsequent work has sought to improve the econometric techniques employed by Scully as well as estimating the impact of the introduction of free agency in major league baseball in the mid-1970s. For example, Zimbalist (1992) found that in the period 1986-1989 baseball players under the reserve rule received 17 – 25 per cent of their MRPs as rookies (i.e. new entrants) rising to 50 – 65 per cent of their MRPs after two years of major-league service. Baseball players qualify for free agency after six years in the major leagues. Zimbalist estimated free-agency salaries to have exceeded MRPs by 23 – 39 per cent.
Academic research in the economics of sport over the last 25 years or so has concentrated on four main areas: gate attendance demand, the sporting production function, labour market issues, and competitive balance in sports leagues. (Cairns et al, 1986, provides a very comprehensive if somewhat dated survey of the field. Dobson and Goddard, 2001, is an excellent guide to the economic analysis of the football industry.) My own published research has focused primarily on the football transfer market (e.g. Dobson and Gerrard, 2000) and coaching efficiency (e.g. Dawson et al., 2000). The econometric analysis of transfer fees shows a high degree of predictability prior to the introduction of Bosman free agency in the mid-1990s. Around 80 per cent of the variation in transfer fees can be explained by the player’s quality characteristics (age, experience, goal-scoring record and international recognition), the size and league status of the buying and selling clubs, and the timing of the transfer. After the introduction of Bosman free agency, transfer fees became dependent on the expiry date of the player’s contract. Typically this information is not publicly available rendering more recent econometric studies of transfer fees subject to a serious misspecification problem.
My own research on transfer fees led to two developments. First, the estimated model of transfer fees is effectively a hedonic-pricing model that provides imputed prices for the player quality characteristics. These imputed prices can be used to construct a player (and team) quality index. This approach has been used to measure player inputs in the study of coaching efficiency in English professional football (Dawson et al., 2000).
The second development derived from the academic research on transfer fees has been a player valuation system, SOCCER TRANSFERS, which has been used commercially in the football industry since 1997. This player valuation system utilises econometric estimates of benchmark coefficients of the key drivers of transfer values. It provides a way to combine public and private information to systematically benchmark player transfer values. The valuation system is re-calibrated on a regular basis but to date there remains considerable structural stability as regards relative transfer values. The uncertainty in the transfer market is primarily over the appropriate absolute level of transfer fees. The transfer market in football operates in a manner similar to other asset markets with uncertainty over future cash flows. An “anchor-and-adjustment” process operates. Football clubs normally accept conventional market signals over the absolute (or anchor) level of transfer values and try to adjust the transfer value of individual players to take account of the specific player quality characteristics. However, clubs do occasionally lose confidence in the anchor as has occurred in the last 12 months. The high uncertainty over the future value of TV rights following the collapse of ITV Digital has led a very thin market with considerable volatility as clubs seek to re-assess the fundamental value of players and reset the anchor accordingly.
The player valuation system has been used for several different purposes. The first application was at the request of a stockbroker to investigate the efficiency of a listed club in the transfer market. It was concluded that the club’s manager had lost in excess of £20 million of shareholder value in his transfer dealings. Clubs have used the valuation system to obtain squad valuations to estimate replacement costs for insurance cover, as a guide to current sell-on values, and to report the estimated market value of intangible assets. Current financial reporting standards only allow clubs to include on the balance sheet the amortised value of those player registrations purchased in the transfer market. This can lead to a significant under-reporting of the market value of the club’s asset base. For example, Leeds United plc reported my estimated squad market value of £198 million in September 2001 compared with the book value of purchased player registrations of only £66.5 million. The player valuation system was also very instrumental in the development of a player sale-and-leaseback arrangement by providing lenders with estimated security value. Most recently, SOCCER TRANSFERS has been used to analyse disputed transfers that have become the subject of litigation.
The development of SOCCER TRANSFERS has highlighted the differences between econometric modelling for the purposes of empirical testing of economic theories and that directed towards out-of-sample forecasting for practical purposes. Constructing a congruent econometric model of transfer fees with R2 of around 80 per cent for a sample of 1,350 observations may be a good instrument for the empirical testing of economic theories but it is wholly inadequate as a means of predicting the market value of an individual player. Asset valuation is both a science and an art, involving knowledge of both systematic determinants and context-specific details. The econometric estimate of benchmark coefficients provides a general valuation formula but it is a matter of judgement and experience to determine the appropriate coefficients to use for a specific individual player.
My work as a business consultant in the football industry has covered a variety of projects involving player valuation, wage benchmarking, and investment analysis. Much of the work has involved providing more formal methods for valuing intangible assets. It has been about creating a better dialogue between the “suits” and the “shirts”, facilitating a better relationship between football and finance. Football clubs have traditionally operated close to the financial edge. The demands of sporting competition has created an “arms race” with clubs operating as cost-driven businesses, seeking to maximise the expenditure on playing talent in pursuit of their sporting aspirations. The collapse of ITV Digital and the consequent financial crisis has emphasised the need to adopt a more financially prudent approach. As one club chairman remarked, reality football is replacing fantasy football. There is now a much greater demand by clubs for leading-edge economic and financial analysis that is well grounded in the specific context of the football industry.
There is also considerable demand by the media for informed comment on the economic and financial aspects of professional sport. Financial statements, player transfers, player wages, TV rights deals and take-overs are all regular news items in the football industry with both sports and business journalists keen to get expert analysis. Media enquiries are a daily occurrence. And when a big story kicks off, especially a story involving my local club, Leeds United, things can get really crazy. Some academics are often very dismissive of the media but as university academics we are after all largely publicly funded and that entails a duty to engage with the wider community whenever possible. Economics as a discipline can enhance its standing in the wider community by getting more involved in the public scrutiny of the football industry provided that economic analysis is applied with due respect to the context-specificity of football. Developments in the football industry are highly path-dependent. History matters.
On a final note, much of my time over the last year has been taken up with the financial crisis at Leeds United. It is a classic case of vanity prevailing over sanity as the club borrowed long to invest in short-term playing assets in the hope of achieving sustained sporting success. It was a gamble that failed to pay-off and the club faces a struggle to avoid relegation and administration in the next few years. I have been very actively involved in trying to alert shareholders and fans to the true depth of the financial problems at the club and the need for radical change in the management and governance processes. Football clubs are ultimately social assets and directors have a duty to ensure that the clubs are run in a viable manner for the benefit of all stakeholders. Economic analysis appropriately contextualised can and should make an important contribution to ensuring that sanity prevails over vanity in the running of our national game.
References
Berman, S. L., Down, J. and Hill, C. W. (2002), “Tacit knowledge as a source of competitive advantage in the National Basketball Association”, Academy of Management Journal, vol. 45, pp. 13-31.
Cairns, J., Jennett, N. and Sloane, P. J. (1986), “The economics of professional team sports: a survey of theory and evidence”, Journal of Economic Studies, vol. 13, pp. 3-80.
Dawson, P., Dobson, S. and Gerrard, B. (2000), “Estimating coaching efficiency in professional team sports: evidence from English association football”, Scottish Journal of Political Economy, vol. 47, pp. 399-421.
Dobson, S. and Gerrard, B. (2000), “Testing for monopoly rents in the market for playing talent: evidence from English professional football”, Journal of Economic Studies, vol. 27, pp. 142-16.
Dobson, S. and Goddard, J. (2001), The Economics of Football, Cambridge: Cambridge University Press.
Neale, W. C. (1964), “The peculiar economics of professional sports: a contribution to the theory of the firm in sporting competition and in market competition”, Quarterly Journal of Economics, vol. 78, pp. 1-14.
Rottenberg, S. (1956), “The baseball players’ labour market”, Journal of Political Economy, vol. 64, pp. 242-258.
Scully, G.W. (1974), “Pay and performance in Major League Baseball”, American Economic Review, vol. 64, pp. 915-30.
Zimbalist, A. (1992), “Salaries and performance: beyond the Scully model” in P. M. Sommers (ed.), Diamonds are Forever: The Business Of Baseball, (pp. 109-133), Washington DC: The Brookings Institution.
[1] MRP (i.e. marginal revenue product) represents the economic value of playing talent measured by the incremental impact on team revenue of the player’s contribution both on- and off-the-field.
Winning With Analytics 