Muscle dran versus bran gan n assocaton football: technology transfer through player emgraton and manager mmgraton G. J. Allan a * and J. Moffat b a Correspondng Author: Department of Economcs, Sr Wllam Duncan Buldng, Unversty of Strathclyde, 130 Rottenrow, Glasgow, G4 0GE, Unted Kngdom b Durham Unversty Busness School, Mll Hll Lane, Durham, DH1 3LB, Unted Kngdom * Correspondng Author. E-mal: grant.j.allan@strath.ac.uk Runnng ttle: Muscle dran versus bran gan n assocaton football To test theores of mgraton and economc development, ths artcle examnes whether nternatonal football teams beneft from havng players playng abroad n stronger leagues and from employng managers from countres wth stronger leagues. The results provde evdence n support of the former effect. However, there s a negatve mpact from employng managers from countres wth superor leagues.
I. Introducton To test theores of mgraton and economc development (e.g. Bene et al., 2011; Docquer et al., 2008; Stark and Wang, 2002), a number of recent papers have nvestgated whether natonal assocaton football (soccer) teams beneft from havng players playng club football outsde ther domestc league (Berlnsch et al., 2013 ; Baur and Lehmann, 2007 ; Gelade and Dobson, 2007 ; Frck, 2009 ). Although such muscle dran has the potental to weaken natonal football teams by lowerng the standard of the domestc league, the emprcal evdence tends to support the hypothess that, through competng n hgher qualty leagues and havng access to superor tranng and tactcal methods, players playng abroad mprove the performance of the natonal team (Berlnsch et al., 2013; Baur and Lehmann, 2007) 1. Whle the players market has become an ncreasngly global one, the same s also true of the market for nternatonal managers wth many natonal football teams hrng foregn managers ( bran gan ). We therefore address a gap n the lterature by nvestgatng whether a naton s standng n nternatonal football (measured by ther rankng ponts) s mproved by havng a foregn manager. Followng the lterature on technology transfer through players playng n overseas leagues, we take nto account the qualty of the league of the country from whch the manager comes. Lke much of the lterature to date, we fnd that havng players playng outsde ther domestc league n stronger leagues has a postve mpact on the performance of the natonal football 1 A related lterature (e.g. Alvarez et al, 2011 and Bnder and Fndlay, 2012) looks at whether there s an mpact on the performance of a natonal team from havng non-domestc players wthn those natonal leagues,.e. whether there s a crowdng-out of domestc talent.
team. By contrast, we fnd a negatve and sgnfcant mpact of havng an overseas manager. II. Lterature Revew Four papers have nvestgated emprcally whether the performance of natonal teams s mproved by players playng outsde of the domestc league. 2 Baur and Lehmann (2007), usng data on the countres that qualfed for the 2006 World Cup, fnd that there s a sgnfcant and postve mpact on country rankng of player mports, whle the mpact of player exports s postve but nsgnfcant. Smlarly, Frck (2009) fnds that havng players playng abroad has no systematc mpact on the chance of a team qualfyng for the later stages of an nternatonal tournament. By contrast, Gelade and Dobson (2007) fnd strong support for a postve mpact on natonal team performance from the export of players. Berlnsch et al. (2013) show a postve mpact of player mgraton to European leagues on orgn countres rankng ponts. They are the frst to note that the qualty of the league matters, rather than whether the league s overseas or not. Our approach to capturng league qualty dffers from that taken by Berlnsch et al. (2013) n that, rather than assumng that all leagues outsde of Europe are of the same standard, we take account of the heterogeneous nature of leagues outsde of Europe. III. Data and Econometrc Model The dependent varable s the Fédératon Internatonale de Football Assocaton (FIFA) rankngs ponts that a country accumulates n a year. Ths dffers from the headlne 2 A related paper by Yamamura (2009) fnds that the rankng ponts of developng countres catch-up wth those of developed countres. However, hs model does not explctly take account of player mgraton.
FIFA rankngs ponts whch are a weghted average of ponts accumulated over four years. 3 Because the benefts of exportng players and mportng managers may take tme to materalze, we estmate the model usng FIFA rankng ponts accumulated n 2010, 2011 and 2012 as the dependent varable. 4 Ths approach should also allevate concerns about endogenety as future natonal team performance cannot determne current player exports or manager mports. To construct the player export varable, we follow Berlnsch et al. (2013) by assumng that there s greater scope for technology transfer from stronger leagues. Our player export varable for each country s therefore calculated by subtractng the rankng ponts of the domestc league of player p (.e. the league for the country that they represent n nternatonal football) from the rankng ponts of the league n whch player p plays for hs club. 5 A weghted average of ths fgure s then taken wth the weghts gven by the number of appearances that player p made for ther natonal team n 2010. N leaguepontsp domestcponts p appearances p N appearances p playerexpo rt (1) p p 3 Detals on how the rankngs are calculated are provded here: http://www.ffa.com/worldrankng/procedureandschedule/menprocedure/ndex.html 4 The mgraton ndex used by Berlnsch et al (2013) relates to between 2 and 3 years pror to the rankng ponts used n ther estmaton, acknowledgng that acqurng football sklls s a process that takes tme. We fnd lttle dfference between usng lagged or concurrent player export varables. 5 In our dataset, out of a total of 6049 players, 1634 play n a stronger league than ther home league (and therefore have a postve value (for leagueponts mnus domestc ponts)) whle only 183 play n a weaker league than ther home league (and have a negatve value). Just over 30%, therefore, of nternatonal players n 2010 played outsde ther own natonal league.
The league rankng ponts for 2010 are taken from the Internatonal Federaton of Football Hstory (IFFHS - http://www.ffhs.de/?b6e28fa3002f76b55a76917f7370eff3702bb1c2bb17). Data on clubs and number of nternatonal appearances for 2010 are taken from Natonal Football Teams (NFT http://www.natonal-football-teams.com/). 6 It s unreasonable to expect that players playng n countres wth a stronger top dvson would beneft ther natonal team n the same way f they played n lower dvsons. We therefore experment wth three specfcatons of leagueponts p. Frstly, we set leagueponts p equal to domestcponts p f a player plays n a country wth a superor top league but outsde the top dvson. Secondly, and smlarly to Berlnsch et al. (2003), we dvde leagueponts p by the dvson number. Thrdly, we gve the same value of leagueponts p to all players n a gven country. Results usng the frst approach are presented below but the sgn and sgnfcance of the key coeffcents are nvarant to how ths varable s calculated. 7 The other varable n whch we are prmarly nterested s the foregn manager varable. Because managers from countres wth stronger leagues are lkely to be famlar wth superor tactcs and tranng methods, ths varable s calculated n an analogous manner to the player export varable: managermp ort domestcponts leaguepon ts (2) 6 Substtute appearances are counted as one thrd of a full appearance. 7 Results from alternatve specfcatons are avalable on request from the authors.
To construct ths varable, we collected data on the natonalty of countres managers n 2010. As some natonal sdes changed managers durng the year, t would have been helpful f data on the number of games for each manager were avalable. As t was not, the results presented below use the natonalty of the manager at the end of 2010. We also expermented wth calculatng ths varable under the assumpton that dfferent managers were n charge for an equal share of games n 2010. The results were very smlar to those presented below. Our model s therefore the followng: ponts 1 playerexport 2managermport 3X (3) where ponts s the FIFA rankngs ponts for country n 2010, 2011 or 2012, playerexport and managermport have been dscussed above and X s a vector of control varables. Followng the lterature we nclude n X varables to capture the nfluence of the clmate, the footballng culture of a country (federaton age), and the fnancal and human resources avalable to a country (GDP per capta, total populaton and ther squares). We also nclude a full set of football confederaton dummes. Further detals on these varables are provded n Table 1.
Table 1. Varable Defntons Varable Defnton Source Ponts Rankng ponts n December 2010, 2011 and 2012 FIFA Player Export See text IFFHS and NFT Manager Import See text Wkpeda and IFFHS League Ponts Rankng ponts for 2010 for the league of the country IFFHS n whch the player plays or the league of the country whch the manager manages Domestc ponts Rankng ponts for 2010 for the league of the home IFFHS country of the player or manager Clmate 8 (Average temperature a n captal cty 14) 2 World Bank Federaton Age Number of years snce the country s football NFT federaton was founded GDP per capta GDP per head of country n 2010 ($ thousands) UN Populaton Populaton of country n 2010 (mllons) UN Confederaton dummes Dummy varables for each confederaton (wth Confederaton of Afrcan Football as the baselne category) UN Notes: a Temperature s measured n degrees Celsus IV. Results The results from estmaton of Equaton (3) are provded n Table 2. The coeffcents on the control varables all have the expected sgn: a more extreme clmate s assocated wth poorer performance; countres wth a greater football culture tend to have more rankng ponts; GDP per capta and populaton both have postve but dmnshng effects on natonal team performance. Regardless of the year from whch the FIFA rankngs are taken, the coeffcent on the player export varable s postve and statstcally sgnfcant, ndcatng that havng players playng abroad n better leagues has the expected benefcal mpact on natonal team performance. Ths confrms the results of Berlnsch et al. (2013) and Gelade and Dobson (2007). By contrast, the coeffcent on the manager mport varable s negatve 8 Followng Berlnsch et al (2013), Macmllan and Smth (2007) and Hoffmann et al. (2002) ths specfcaton mples that an annual average temperature of 14 C s the optmal temperature for football.
and statstcally sgnfcant. Ths suggests that cultural dfferences between management and players are dffcult to surmount n nternatonal football and that natonal sdes should employ domestc managers. Table 2. Results from estmaton of equaton (3) usng ordnary least squares Dependent Varable: (1) Ponts 2010 (2) Ponts 2011 (3) Ponts 2012 Player Export 0.332*** 0.335*** 0.218*** (0.088) (0.073) (0.082) Manager Import -0.123*** -0.107*** -0.076** (0.040) (0.033) (0.037) Clmate -0.479** -0.372** -0.357* (0.204) (0.170) (0.190) Federaton Age 0.576 0.678* 1.575*** (0.424) (0.354) (0.396) GDP per capta 4.580*** 1.179 2.222* (1.317) (1.100) (1.230) GDP per capta 2-0.039*** -0.010-0.025** (0.013) (0.011) (0.012) Populaton 1.532*** 0.961*** 0.642** (0.325) (0.272) (0.304) Populaton 2-0.001*** -0.001*** -0.000* (0.000) (0.000) (0.000) Confederaton dummes Yes Yes Yes R-squared 0.549 0.463 0.420 Observatons 179 179 179 Notes: ***/**/* denotes sgnfcance at the 1%/5%/10% level V. Concluson Ths artcle has sought to estmate the mpact of the ncreasngly globalzed marketplace for football playng and management talent on natonal team performance. Specfcally, t s the frst artcle n the lterature to explore the mpact of both the muscle dran of players to non-domestc leagues and the bran gan from overseas managers. Our results ndcate that havng players playng outsde ther domestc league n stronger leagues has a postve mpact on the performance of the natonal team. For overseas
managers, we fnd that the mpact on rankng ponts s negatve and sgnfcant. These results suggest that natonal football sdes should employ domestc managers. References Alvarez, J., Forrest, D., Sanz, I. & Tena, J. D. (2011) Impact of mportng foregn talent on performance levels of local co-workers, Labour Economcs, 18, 287-296. Baur, D. G. & Lehmann, S. (2007) Does the Moblty of Football Players Influence the Success of the Natonal Team?, IIIS Dscusson Paper No. 217, Insttute for Internatonal Integraton Studes, Dubln. Bene, M., Docquer, F. & Oden-Defoort, C. (2011) A Panel Data Analyss of the Bran Gan, World Development, 39, 523-532. Berlnsch, R., Schokkaert, J. & Swnnen, J. (2013) When drans and gans concde: Mgraton and nternatonal football performance, Labour Economcs, 21, 1-14. Bnder, J. J. & Fndlay, M. (2012) The Effects of the Bosman Rulng on Natonal and Club Teams n Europe, Journal of Sports Economcs, 13, 107-129. Docquer, F., Faye, O. & Pesteau, P. (2008) Is mgraton a good substtute for educaton subsdes?, Journal of Development Economcs, 86, 263-276. Frck, B. (2009) Globalzaton and Factor Moblty: The Impact of the ``Bosman- Rulng'' on Player Mgraton n Professonal Soccer, Journal of Sports Economcs, 10, 88-106. Gelade, G. A. & Dobson, P. (2007) Predctng the Comparatve Strengths of Natonal Football Teams, Socal Scence Quarterly, 88, 244-258. Hoffmann, R., Lee, C. G. & Ramasamy, B. (2002) The Soco-Economc Determnants of Internatonal Soccer Performance, Journal of Appled Economcs, 5, 253-272.
Macmllan, P. & Smth, I. (2007) Explanng Internatonal Soccer Rankngs, Journal of Sports Economcs, 8, 202-213. Stark, O. & Wang, Y. (2002) Inducng human captal formaton: mgraton as a substtute for subsdes, Journal of Publc Economcs, 86, 29-46. Yamamura, E. (2009) Technology transfer and convergence of performance: an economc study of FIFA football rankng, Appled Economcs Letters, 16, 261-266.