Collaboration and Homophily in Global Teams
How do barriers related to nationality and language affect collaboration in multinational teams? We address this question by exploiting a newly assembled exhaustive dataset recording all 10.7 million passes by professional European football players from 132 countries fielded by all 154 teams competing in the top five men leagues over eight sporting seasons, together with full information on players’ and teams’ characteristics. We measure collaboration as the average number of passes per minute between a pair of players in a half season. We use a discrete choice model of players’ passing behavior as a baseline to separately identify excess collaboration within nationality or language due to preferences (‘choice homophily’) from collaboration due to opportunities (‘induced homophily’). Our dataset allows us to estimate the model using a rich set of player and play characteristics as well as player fixed effects. We find strong evidence of homophily: conditioning on players’ and teams’ characteristics, player pairs of same nationality exhibit an average number of passes per minute that is 2.5 percent higher than player pairs of different nationality. Same nationality is about as likely to lead to more passes as doubling the player pair’s valuation, which is a consensus measure of players’ skills. Shared language has about half the impact of same nationality. Pairs of same nationality are also more likely to engage in deeper collaboration, disproportionately participating in more complex pass sequences. These findings show that homophily based on nationality and language is pervasive even in teams of very high skill individuals with clear common objectives and aligned incentives and involved in interactive tasks that are well-defined and not particularly language-intensive.