Wearing the Same Jersey? The Impact of Players’ Cultural Diversity and Shared Team Tenure on National Soccer Team Performance

Abstract

In the present paper, we empirically analyze a dataset from national soccer teams between 2004 and 2019 to investigate the impact of team members’ cultural diversity on final team outcomes. Further, we examine the role of shared team tenure as a proxy of tacit knowledge and within team communications in relation to the cultural diversity–team performance relationship. After addressing a potential endogeneity issue, results from multiple instrumental variable estimation methods implied a higher level of cultural diversity enhanced on-field performance. Meanwhile, increased level of shared team tenure further strengthened the effect associated with cultural diversity on team performance.

… when embraced, [diversity] actually delivers practical benefits, since it ensures that a society can draw upon the talents and energy and skill of all its people. And if you doubt that, just ask the French football team that just won the World Cup because not all these folks look like Gauls to me, they are French, they are French. (Bushnell 2018, n.p.)

Introduction

Cultural diversity is a multifaceted concept. Often for cultural diversity nationality is used as a common proxy due to it being a contributing attribute towards individuals’ cultural identities (Godfrey et al. 2020; Richard et al. 2004, 2007). According to Harrison and Klein’s 2007 typology, cultural diversity can be considered a variety diversity, which refers to a “composition of differences in kind, source, or category of relevant knowledge or experience among unit members” (Harrison & Klein 2007 p. 1203). According to this conceptualization, national diversity is a type of variety diversity because members’ nationality is a categorical variable and is associated with country/cultural-specific knowledge (Ely & Thomas 2001; Kearney & Gebert 2009; Lazear 1999). This categorization is important when examining the relationship between cultural diversity and team performance with nationality as a proxy for cultural diversity.

Cultural diversity is a mixed blessing for work groups, as research investigating the relationship between cultural diversity and team performance generally yields mixed results (Stahl et al. 2010; Stahl & Maznevski 2021). To be specific, in a culturally heterogenous team, individuals’ cultural differences are often associated with a greater reservoir of unique information, knowledge, experiences, and skills beneficial to problem solving, decision making, and innovation, which in turn improve team performance. However, the more team members are culturally different, the more they are prone to categorize ingroup (“us”) and outgroup (“them”) within the team based on this dissimilarity, leading to biases as well as impaired communication and collaboration between subgroups (Williams et al. 1998), which ultimately can harm team performance.

One potential way to mitigate the negative impacts of cultural diversity on team performance is through a shared team tenure. Time plays a significant role in relation to the cultural diversity–performance relationship as it can be related to both tacit knowledge and team members’ communications and interactions (Berman et al. 2002; Harrison et al. 1998, 2002). As Gonzalez-Mulé et al. (2020) described, shared team tenure can help develop relationships amongst these members along with the knowledge gained from working together. In addition to examining the main effect of shared team tenure (i.e., time) in terms of affecting team performance, we move a step further to test whether it can moderate the main relationship between cultural diversity and team performance by linking it to tacit knowledge sharing and interpersonal communications within the team. Such an exploration of moderators is an important way of defining boundary conditions for a theory (Busse et al. 2017) and plays a significant role in terms of investigating the relationship between different forms of team diversity and performance (Guillaume et al. 2017; van Knippenberg & Schippers 2007).

The purpose of the present study is to investigate how team heterogeneity with respect to members’ nationalities as a proxy of cultural diversity influences final team outcomes. Furthermore, the present research looks to explore if time as a proxy for tacit knowledge and enhanced team interactions can moderate the cultural diversity and team performance relationship. We utilize a dataset comprised of national soccer teams from the Union of European Football Associations (UEFA) from 2004 and 2019 using Elo Ranking as a measure of team performance. In terms of national soccer, individual teams now are recruiting an increasing number of players from different regions, making nationality a potentially more salient attribute than other demographic traits such as age (Poli et al. 2018; Richard & Miller 2013). These players are either naturalized players or with multicultural origins, and may have only “‘vague’ connections to the countries whose jerseys they wear” (van Campenhout et al. 2019, p. 2). In addition, we argue players from different cultures/countries possess soccer knowledge specific to their national origins (e.g., Brandes et al. 2009) that may have an effect on the diversity and team performance relationship.

We estimate a two-stage regression using an instrumental variable to address the simultaneous causality between cultural diversity and team performance. From our estimation results, we uncover evidence that a greater level of players’ cultural diversity measured in nationality is advantageous to on-field team performance. This finding is in concert with prior work stating a positive relationship between cultural diversity and team performance in the context requiring significant level of cooperation and interdependence among team members (e.g., van Knippenberg & Schippers 2007). Furthermore, we do find that shared team tenure does have a positive effect related to the cultural diversity and team performance relationship. This conclusion speaks to the importance of identifying and testing moderators in terms of explicating the relationship between cultural diversity and team performance, in which equivocal results often emerge.

Theoretical Backgrounds and Hypotheses Development

Cultural Diversity and Team Performance

Compared to other diversity attributes, team heterogeneity in terms of cultural (national) differences, is an area where both positive and negative effects of diversity on performance have been observed (Raithel et al. 2021; van Knippenberg et al. 2013). Building off information/decision-making theory (Williams et al. 1998), which states multinational teams may have greater access to a variety of cultural/country-specific knowledge, skills, and perspectives that are unique or complementary to team tasks associated with cultural heterogeneity, an increase in national diversity within a team is positively related to team performance (Ely & Thomas 2001; Kearney & Gebert 2009). To the contrary, the social categorization perspective suggests people working in teams tend to categorize themselves and others into ingroups and outgroups based on differences and similarities between “us and them” (Tajfel and Turner 1979; Turner 1987). As a result, individuals tend to favor members of ingroups, trust them more, and be more willing to cooperate with those with whom they share similarities (e.g., same nationality). Likewise, the similarity-attraction paradigm predicts individuals are attracted to working and cooperating with those to whom they find similar in terms of values, beliefs, and attitudes (Byrne 1971). Thus, within-team cultural diversity may create a basis for intergroup biases, leading to a disruptive effect on team processes and performance (van Knippenberg et al. 2013).

Although both positive and negative effects of cultural diversity (i.e., heterogeneity in terms of members’ nationalities) are found within the extant literature, for contexts requiring high levels of cooperation and interdependence among team members, the relationship between cultural diversity and team performance is generally a positive relationship (Guillaume et al. 2017; van Knippenberg & Schipper 2007). Through the lens of social categorization perspective, team members originating from different cultures/countries may have an emphasis on the common group identity and be inattentive to subgroup categorizing processes provided the task demands a greater level of cooperation and interdependence among them (Gaertner & Dovidio 2000). Furthermore, in a work setting requiring high level of interdependence, interactions and communications between different subgroups may be enhanced, resulting in more harmonious relations between them (Pettigrew 1998). According to information/decision making theory, the requirement of collaboration may also facilitate information processing, since it may encourage information exchange and discussion among different subgroups, which are essential to reap the benefits associated with a multicultural team (van Knippenberg et al. 2004).

Within the context of professional sport, hockey and soccer are examples of sports requiring a significant level of cooperation and interdependence among players, and empirical works indeed found evidence of a positive connection between cultural diversity and team performance in these settings. For instance, Kahane et al. (2013) analyzed team performance from the National Hockey League (NHL) to investigate potential benefits of franchises signing European players. Controlling for players’ on-ice skills and abilities along with the abilities of the coaching staff, their result indicated increased diversity (i.e., including more European players on the team rosters) was positively associated with sporting success. They indicated two distinct areas of how increasing diversity improved team performance. The first is these players were often more skilled in some areas (e.g., basic skill development) in comparison to their North American counterparts. Second, adding additional players from the same European countries as players currently on the roster improved both communications and any cultural barriers that may exist.

Considering the “Big 5” European soccer leagues,1 Wilson and Ying (2003) asserted a higher share of foreign players was linked to improvement in a club’s on-field performance, especially when players were of Balkan or Brazilian origin. Similarly, Ingersoll et al. (2017) analyzed seasonal data from the UEFA Champions League,2 finding a higher level of inner-team cultural diversity, proxied by language distance, positively associated with sporting success on the field. More recently, Royuela and Gásquez (2019) investigated the impact of including more foreign players on roster on a team’s sporting success by analyzing a cross-section dataset comprised of approximately 1,000 clubs worldwide. Based on their empirical results, the authors claimed a higher share of non-domestic players improved an individual club’s world ranking. Directly connected to the present study, Brandes et al. (2009) examined the effect of national diversity in the German Bundesliga by separating teams into their offensive and defensive position groupings. The estimation results indicated, ceteris paribus, increasing nationality-heterogeneity among defensive players improved the entire team’s performance, while the effect associated with offensive subgroups was insignificant. The bifurcated results implied the importance of exploring moderators in order to explicate the effects related to cultural differences on final team outcomes.

Taken together, with respect to the cultural diversity–team performance nexus, in the context of soccer (which requires a high level of cooperation and interdependence) we hypothesize:

• Hypothesis 1 (H1): An increase in players’ cultural diversity within a national soccer team is positively related to its on-field performance.

The Role of Shared Team Tenure

Hadjimichael and Tsoukas (2019) recently summarized the literature on tacit knowledge and outlined three perspectives where previous research utilized the concept of tacit knowledge: conversion, interactional, and practice. For the present study, our focus is on the conversion perspective. In this perspective, tacit knowledge is defined as “[a] type of knowledge that is subjective, largely inaccessible to consciousness, tied to activity, and awaiting conversion into an explicit form” (Hadjimichael & Tsoukas 2019, p. 678). Building off this definition of tacit knowledge, they discussed how the acquisition of this knowledge comes through shared team tenure in an organization and group setting. Following the conversion perspective, they also outlined two research streams important to our study, performance and knowledge management. In terms of the first one, the authors concluded individuals’ tacit knowledge, which can be acquired through spending more time with their teammates, was linked to improved organizational/group performance. The second stream emphasizes the importance of externalization (i.e., conversion) of tacit knowledge to promote knowledge transfer within and across organizations. To facilitate (tacit) knowledge transfer, individuals are encouraged to spend time with others on the team.

In the context of professional sport, empirical work found players from different origins may possess nation-inherent skills that may pass to other teammates from other nations through shared time together. For instance, Brandes et al. (2009) demonstrated that in the German Bundesliga, compared to German midfielders, Brazilian counterparts scored more goals, had higher passing accuracy, and received fewer yellow cards. By analyzing individual players’ performance from the five prestigious European leagues, Glennon et al. (2021) demonstrated foreign players outperformed their domestic counterparts in several areas critical to team success on the field (e.g., goals scored). Similarly, in the NHL, Kahane et al. (2013) found players from European countries were often highly skilled in some areas such as basic skill development in comparison to their North American teammates. In addition, different countries may have its own style of play on the field. For instance, Italian soccer has a longstanding association with playing Catenaccio—an unspoken system emphasizing defense. Likewise, Tiki-Taka, a playing style characterized by short passes, player movement, and ball possession, is linked to notable success in Spanish soccer (Morley 2014).

These examples and prior research above note the potential importance of a shared team tenure as it is generally referred to tacit knowledge. Prior research on team sports found some supporting evidence regarding a positive tacit knowledge–team performance nexus using the time that players had with a focal team as a proxy of such know-how (Berman et al. 2002; Franck et al. 2011; Humphrey et al. 2009). As Berman et al. (2002) argued “[a]s players interact on the same team over time, they increase team performance and perhaps build a competitive advantage through group-level tacit knowledge.” (p. 18). Therefore, individual members can benefit from others’ tacit knowledge through working together on the same team, which can be shared within the team and enhance team performance. We predict:

• Hypothesis 2 (H2): An increase in players’ shared team tenure is positively related to national soccer teams’ on-field performance.

In addition to the main effect, shared team tenure may play a role in terms of moderating the cultural diversity–team performance relationship. On the one hand, although players from different countries/cultures possess inherent tacit knowledge that is beneficial to overall team performance, within a diverse team it takes some time for such benefits to emerge (Earley & Mosakowski 2000; van Knippenberg et al. 2004, 2020). Consequently, it is possible that heterogeneous teams outperform their homogeneous counterparts in the long run. On the other hand, any potential negative effect associated with cultural diversity on team performance may disappear as time passes since individuals may gain more experience of working and collaborating with each other over time (Harrison et al. 1998, 2002). In other words, when players spend more time together on and off the pitch (i.e., increased shared team tenure), it is plausible they notice their initial stereotypes or biases about “outgroup” members due to different nationalities, which may impair final team outcomes, are incorrect (van Knippenberg & Schippers 2007). Empirically, Georgakakis et al. (2017) demonstrated shared team tenure between a firm’s CEO and top management team members positively moderated the negative relationship between members’ diversity features and team performance.

Given all these arguments, we expect shared team tenure to moderate the relationship between cultural diversity and team performance:

• Hypothesis 3 (H3): Shared team experience moderates the relationship between cultural diversity and team performance such that the positive relationship will be strengthened.

Data and Methods

Data

To test the three hypotheses above, we use a dataset comprised of national teams from all UEFA members running from 2004 to 2019.3 The year 2004 was the first year that FIFA adopted a new rule requiring a player to demonstrate a “clear connection” to any country they wish to represent (The Federation Internationale de Football Association 2021). We end in 2019 due to eliminating any impact the COVID-19 pandemic would have on team performance and availability of players. The unit of observation is a national team-season. During the sample period, there are 853 national team-year observations in total.

Dependent Variable

Consistent with prior work (e.g., Gásquez & Royuela 2016; Peeters et al. 2021), we use Elo scores (Elo_pts) as a proxy for a national team’s performance, where a higher Elo score stands for better on-field performance (Peeters et al. 2021). The Elo rating system4 is constructed as a weighted average of past game results, where weights are defined by the strength of the opposing team, the margin of victory, and the importance of the game.

Independent Variable

Recall from above that cultural diversity can be categorized into variety diversity, which represents individual differences in terms of task-relevant knowledge, experiences, and skills (Harrison & Klein 2007). Following their recommendation as well as prior literature on soccer team diversity (e.g., Haas & Nüesch 2012), the cultural diversity of the team was calculated using the Blau index of players’ second citizenships (Blau). The Blau index, the variable of interest to test H1, is computed as $1-{\sum }_{i=1}^n{p}^2$, where p is the proportion of unit members in the k^th category. The values can range from 0 to (K–1)/K. For example, according to the roster of team France in 2018, 24 out of 34 players have a second citizenship. In terms of these second citizenships, there are 15 different countries (i.e., categories). To calculate the Blau index, we divided the number of players holding the same second citizenship by 24 (i.e., the proportion of unit members in the k^th category), then take the square term before summing all the proportions of these 15 countries. For instance, three players have a second citizenship of Mali, then the proportion of Mali is 3/24.

Compared to a sport that requires less interdependence among players (e.g., baseball), soccer involves a high level of coordination among players fielding different positions, thus, tacit knowledge should play a more important role in terms of promoting on-field performance in this context (H2; Shamsie & Mannor 2013). We use the variable Time to test H2, which we operationalize by the number of times a national team player was called up to the team at the end of the observed year. From there, we take the team averages for the observed national team. This shared team experience or tacit knowledge that we operationalize through the Time variable is consistent with earlier work on this phenomenon in a variety of sport contexts (e.g., Berman et al. 2002; Franck et al. 2011; Humphrey et al. 2009). Information regarding national teams’ rosters and players’ nationalities were obtained from Transfermarkt.

Control Variables

Following prior literature on national soccer team performance determination (e.g., Berlinschi et al. 2013; Hoffmann et al. 2002; Peeters et al. 2021), we control for a number of country-specific demographic, socioeconomic, and climatologic variables. These include the total population (Pop_total), percentage of urban population (Urban_pct), gross domestic product (Gdp_per_capita), and average temperature (the absolute difference from 14C; Temp_14²). All demographic, socioeconomic, and climatologic variables were retrieved from the World Bank (https://www.worldbank.org).5 Except for the climatological variable, we incorporated squared terms for other variables to capture any potential nonlinearity (Pop_total², Urban_pct², Gdp_per_capita²) consistent with other studies (see Gásquez & Royuela 2016).

We also took into consideration a country’s soccer tradition by incorporating the number of years being a member of the UEFA (UEFA_tenure; Leeds & Leeds 2009), and an indicator variable implying whether a specific country had ever hosted either the FIFA World Cup (WC_host; Hoffmann et al. 2002).

Estimation Issues

Equation 1 presents our full model:

$${ELO\_PTS}_{\mathit{it}}={\beta }_0+{\beta }_1{\mathit{Blau}}_{\mathit{it}}+{\beta }_2{\mathit{Time}}_{\mathit{it}}{{+\beta }_3{\mathit{Blau}}_{\mathit{it}}\times Time}_{\mathit{it}}{+\alpha X}_{\mathit{it}}+{\epsilon }_{\mathit{it}}$$ 1

where i indexes national team and t indexes year. The main variables of interest regarding H1–H3 are ${\mathit{Blau}}_{\mathit{it}}$, ${\mathit{Time}}_{\mathit{it}}$, and their interaction term (i.e., ${\mathit{Blau}}_{\mathit{it}}\times {\mathit{Time}}_{\mathit{it}}$), respectively. X is a vector of control variables mentioned previously.

The big issue of using OLS to estimate Eq. 1 is the potential presence of endogeneity resulting from simultaneous causality, which occurs when the causality runs in both directions from the independent variable to dependent variable and vice versa. To be specific, it could be the case with H1 that a higher level of cultural diversity strengthens a national team’s on-field performance. However, we cannot exclude the possibility that a (relatively) soccer-developing country may recruit some naturalized players from (more) soccer-developed destinations to enhance its competitiveness on the field. As a result, the Blau index of players’ nationalities will be affected, since more players with dual citizenship will be included within the team.6 Therefore, in such circumstance (i.e., simultaneous causality), outcomes from OLS estimation are biased and inconsistent.

One way to deal with the issue of endogeneity is to use instrumental variables (IV) with two-stage least squares (2SLS) being the most common IV estimator (Bascle 2008). In the present research, we estimated a first-stage model with the dependent variable being the Blau index (Blau). The IV we chose in the present paper is the number of individual countries’ annual international remittances sent. According to the International Monetary Fund, international remittances refer to the transactions made by migrants when sending home part of their earnings in the form of either cash or goods. For the past decades, the volume of cross-broad remittances grew dramatically, which can be attributed, to a large extent, to the increased flows of international migration (The Global Knowledge Partnership on Migration and Development 2021; United Nations Economic Commission for Europe 2011).7

In addition, prior research demonstrated a positive relationship between the stock of migrant workers and the amount of remittances sent (Elbadawi & Rocha 1992). As an essential component of labor migration, remittances and migration are closely connected (Carling 2008). Therefore, we assume individual teams’ Blau index should be positively related to the amount of remittances flowing out of the country (Remittance_paid). However, the amount of international remittances sent from a country should have no impact on its national soccer team performance. Subsequently, this IV (i.e., annual remittances paid) is potentially correlated with cultural diversity (i.e., independent variable) within the team and not related to team performance (i.e., dependent variable) directly. Relevant data on remittances were collected from the World Bank.

Results

As noted above, there are 853 national team-year observations. The final sample for our analysis is 689, due to the unavailability of related information on temperature, GDP per capita, and annual remittances sent. Table 1 presents the summary statistics and pairwise correlations with the variables used in the present research for the final sample. The mean Elo points in the sample is 1,663, with the Blau index of 0.587, representing the probability of two players chosen at random from the team being of different nationalities. The correlation between cultural diversity and on-field team performance is significantly positive in our sample. Approximately sixteen percent of the observations hosted a World Cup. From the table, one also notices the average of shared team tenure (Time) is 28 (games) and significantly positively related to on-field team success.

Table 1.

Summary Statistics and Pairwise Correlations (n = 689)

Variables	Mean	SD	1	2	3	4	5	6	7	8	9
1.Elo_pts	1663.328	222.073	1.000
2.Blau	0.587	0.287	0.195***	1.000
3.Pop_totala	15.998	21.81	0.571***	0.169***	1.000
4.Gdp_per_capitab	0.027	0.024	0.117***	0.288***	-0.007	1.000
5.Urban_pct	0.698	0.141	0.201***	0.230***	0.149***	0.556***	1.000
6.Temp_142	31.903	35.258	-0.159***	-0.069*	-0.290***	0.393***	0.317***	1.000
7.UEFA_tenure	42.932	19.566	0.443***	0.296***	0.294***	0.529***	0.515***	0.044	1.000
8.WC_host	0.161	0.368	0.528***	0.270***	0.636***	0.326***	0.245***	0.006	0.349***	1.000
9. Time	27.701	8.065	0.417***	0.009	0.256***	0.249***	0.302***	-0.081**	0.323***	0.340***	1.000
Instrument Variable	Mean	SD	Max	Min
Remittances_paid (in million)	2,995	4,988	4.900	28,000

a: in billion; b: in 2015-year US dollar and million. * p < 0.1, ** p < 0.05, *** p < 0.01

In terms of our IV, Remittance_paid, Table 2 presents the first stage results from 2SLS. In concert with our assumption, Blau is positively related to the amount of annual remittances paid. As a robustness check, we included lagged terms in one, two, and three years, along with three- and five-year averages for remittances paid as IVs. In order to examine the relevance and validity of our instrumental variables, we followed the procedure outlined by Bascle (2008) when dealing with the endogeneity issues. First, the Wu-Hausman test rejected the null hypothesis that all variables are exogenous, justifying the application of IV method in our case. Second, the F-statistics from first-stage estimations are greater than 10, which is a “rule-of-thumb” for a strong IV, indicating no presence of a weak IV issue in our study.

Table 2.

First Stage Estimation Results from Instrumental Variables Regression (2SLS)

	Dependent Variable: Blau
Variables	Observed Year	One-Year Lag	Two-Year Lag	Three-Year Lag	Three-Year Average	Five-Year Average
Remittances_paid	0.0000169***	0.0000173***	0.0000173***	0.0000173***	0.0000176***	0.0000175***
	(3.46 $\times$ 10–6)	(3.60 $\times$ 10–6)	(3.71 $\times$ 10–6)	(3.83 $\times$ 10–6)	(3.78 $\times$ 10–6)	(3.99 $\times$ 10–6)
Pop_total	-0.0008083	-0.0006419	-0.000404	0.0000481	-0.0000142	0.0002525
	(0.0020422)	(0.0020479)	(0.0020511)	(0.0020661)	(0.0020658)	(0.0021068)
Pop_total2	4.54 $\times$ 10–07	-1.62 $\times$ 10–6	-3.55 $\times$ 10–6	-7.85 $\times$ 10–6	-7.39 $\times$ 10–6	-1.04 $\times$ 10–5
	(0.0000258)	(0.0000259)	(0.000026)	(0.0000261)	(0.0000261)	(0.0000265)
GDP_per_capita	-2.614432	-2.588887*	-2.476182	-2.388101	-2.392676	-2.050546
	(1.669996)	(1.678706)	(1.674165)	(1.670499)	(1.692496)	(1.71674)
GDP_per_capita2	24.95486	25.54656***	26.01361*	27.02318*	25.61203*	24.4356
	(15.20926)	(15.27348)	(15.28142)	(15.30339)	(15.37228)	(15.56219)
Urban_pct	-2.361622***	-2.336531***	-2.21475***	-2.080314***	-2.16941***	-1.970206***
	(0.7037341)	(0.7052604)	(0.705787)	(0.7060129)	(0.712194)	(0.7236889)
Urban_pct2	1.813269***	1.794006***	1.713174***	1.623282***	1.682547***	1.54772***
	(0.4938154)	(0.4947476)	(0.495452)	(0.4958795)	(0.4990174)	(0.5063678)
Temp_142	-0.0006784*	-0.0006866*	-0.0007155*	-0.0007259**	-0.0006919*	-0.0007309**
	(0.0003655)	(0.0003673)	(0.0003667)	(0.0003668)	(0.0003679)	(0.0003703)
UEFA_tenure	0.002179***	0.0021844***	0.0021082***	0.0019542**	0.0019648**	0.0017883**
	(0.0007385)	(0.0007443)	(0.0007493)	(0.0007553)	(0.0007575)	(0.0007754)
WC_host	0.0878714**	0.088169**	0.0869826**	0.0859493**	0.0809181*	0.0817811*
	(0.0425617)	(0.0427855)	(0.042905)	(0.0430543)	(0.0431801)	(0.0437776)
Wu-Hausman test	4.27**	3.44*	3.78*	4.86**	3.67*	4.13**
F test of excluded instruments	23.86***	23.05***	21.73***	20.32***	21.62***	19.26***
R2	0.205	0.204	0.202	0.199	0.200	0.198

Standard errors in parentheses. * p < 0.1 ** p < 0.05 *** p < 0.01

Table 3 and 4 presents the OLS and IV regression results for Eq. 1 respectively. H1 posited an increase in national team players’ cultural diversity is positively related to teams’ on-field performance. Although the coefficient of Blau is statistically insignificant in Table 3, we find support for this hypothesis using the Blau index as the measure of cultural diversity when looking at the coefficients presented in Table 4. H2 looked to isolate the influence of individual players’ shared team tenure (Time) on national teams’ sporting success. Since the concern of endogeneity surrounds cultural diversity measure (Blau), we estimated an OLS model excluding cultural diversity. The results from various models in Tables 3 and 4 found a positive and significant impact of Time on team performance on team performance. As such, we find support for H2.

Table 3.

Estimation Results for Hypotheses Testing (OLS Estimations)

Dependent Variable: Elo points
Variables	M1	M2	M3
Blau (H1)	15.882	34.302	37.786
	(30.981)	(27.265)	(28.384)
Time (H2)	––	5.201***	5.312***
	––	(0.974)	(0.959)
Blau $\times$ Time (H3)	––	––	2.892*
	––	––	(1.708)
Pop_total	39.106	28.492	30.483
	(24.571)	(19.908)	(19.789)
Pop_total2	-0.303*	-0.226*	-0.240*
	(0.164)	(0.134)	(0.133)
GDP_per_capita	2100.609	3503.456	3373.036
	(2577.571)	(2634.438)	(2584.744)
GDP_per_capita2	-3.8 $\times 10$ 3	-7.5 $\times 10$ 3	-7.5 $\times 10$ 3
	(9462.163)	(9351.190)	(9192.556)
Urban_pct	2370.057	1834.119	1966.639*
	(1574.212)	(1223.351)	(1161.855)
Urban_pct2	-1.7 $\times 10$ 3	-910.129	-1.0e + 03
	(1274.496)	(976.628)	(929.072)
Temp_142	-0.509	-0.320	-0.273
	(0.467)	(0.392)	(0.385)
UEFA_tenure	-1.913	-4.710***	-4.762***
	(1.784)	(1.578)	(1.607)
WC_host	22.410	34.786	31.240
	(29.309)	(25.109)	(25.036)
N	766	766	766
R2	0.048	0.199	0.205

Standard errors in parentheses. * p < 0.1 ** p < 0.05 *** p < 0.01

Table 4.

Estimation Results for Hypotheses Testing (IV Estimations)

Dependent Variable: Elo Points
Variables	IV: Observed Year		IV: One-Year Lag		IV: Two-Year Lag		IV: Three-Year Lag		IV: Three-Year Average			IV: Five-Year Average
	M3	M4	M5	M6	M7	M8	M9	M10	M11		M12	M13	M14
Blau (H1)	285.002**	62.534	263.112**	17.669	282.610**	23.719	324.358**	46.246	278.724**		9.175	305.735**	20.015
	(129.620)	(117.226)	(129.081)	(115.832)	(136.400)	(117.668)	(146.459)	(122.041)	(135.765)		(118.847)	(145.352)	(124.487)
Time	––	7.624***	––	7.458***	––	7.518***	––	7.769***	––		7.415***	––	7.505***
	––	(1.059)	––	(1.071)	––	(1.100)	––	(1.137)	––		(1.118)	––	(1.165)
Blau $\times$ Time (H3)	––	18.227***	––	17.708***	––	17.708***	––	17.710***	––		18.151***	––	17.814***
	––	(3.960)	––	(3.959)	––	(3.840)	––	(3.890)	––		(3.945)	––	(3.935)
Pop_total	12.287***	13.648***	12.134***	13.415***	12.084***	13.388***	11.926***	13.433***	11.965***		13.429***	11.725***	13.343***
	(1.368)	(1.141)	(1.345)	(1.115)	(1.366)	(1.120)	(1.410)	(1.136)	(1.366)		(1.125)	(1.426)	(1.160)
Pop_total2	-0.134***	-0.155***	-0.133***	-0.153***	-0.133***	-0.153***	-0.132***	-0.155***	-0.132***		-0.154***	-0.131***	-0.154***
	(0.016)	(0.014)	(0.016)	(0.014)	(0.016)	(0.014)	(0.017)	(0.014)	(0.016)		(0.014)	(0.017)	(0.014)
GDP_per_capita	1.0 $\times 10$ 4***	8458.767***	1.0 $\times 10$ 4***	8732.395***	1.0 $\times 10$ 4***	8542.235***	9994.061***	8367.749***	1.0 $\times 10$ 4***		8767.457***	1.0 $\times 10$ 4***	8825.437***
	(995.174)	(950.793)	(970.511)	(927.195)	(982.832)	(931.200)	(1008.584)	(923.088)	(989.870)		(945.554)	(1015.860)	(963.076)
GDP_per_capita2	-1.1 $\times 10$ 5***	-9.4 $\times 10$ 4***	-1.1 $\times 10$ 5***	-9.5e $\times 10$ 4***	-1.1 $\times 10$ 5***	-9.4 $\times 10$ 4***	-1.1 $\times 10$ 5***	-9.4 $\times 10$ 4***	-1.1 $\times 10$ 5***		-9.5 $\times 10$ 4***	-1.1 $\times 10$ 5***	-9.6 $\times 10$ 4***
	(1.0 $\times 10$ 4)	(8519.401)	(1.0 $\times 10$ 4)	(8456.926)	(1.0 $\times 10$ 4)	(8467.154)	(1.1 $\times 10$ 4)	(8460.473)	(1.0 $\times 10$ 4)		(8531.873)	(1.1 $\times 10$ 4)	(8538.603)
Urban_pct	2489.698***	1848.585***	2482.756***	1807.231***	2592.068***	1908.122***	2712.796***	1979.716***	2485.574***		1805.968***	2530.145***	1861.301***
	(518.840)	(489.505)	(521.089)	(502.042)	(525.389)	(499.340)	(529.905)	(488.622)	(524.052)		(507.350)	(527.866)	(506.053)
Urban_pct2	-1.9 $\times 10$ 3***	-1.4 $\times 10$ 3***	-1.9 $\times 10$ 3***	-1.4e $\times 10$ 3***	-1.9 $\times 10$ 3***	-1.4 $\times 10$ 3***	-2.0e $\times 10$ 3***	-1.5 $\times 10$ 3***	-1.9 $\times 10$ 3***		-1.4 $\times 10$ 3***	-1.9 $\times 10$ 3***	-1.4 $\times 10$ 3***
	(380.612)	(349.476)	(381.945)	(356.288)	(387.266)	(355.095)	(393.894)	(349.032)	(385.385)		(358.574)	(389.823)	(357.341)
Temp_142	-0.565**	-0.381	-0.639**	-0.507**	-0.625**	-0.505*	-0.583**	-0.464*	-0.633**	-0.524**		-0.617**	-0.521*
	(0.274)	(0.269)	(0.263)	(0.257)	(0.273)	(0.261)	(0.288)	(0.268)	(0.270)	(0.261)		(0.286)	(0.271)
UEFA_tenure	1.362***	1.223***	1.264***	1.137***	1.269***	1.154***	1.276**	1.145***	1.279***	1.118***		1.317***	1.135***
	(0.488)	(0.404)	(0.476)	(0.388)	(0.488)	(0.390)	(0.506)	(0.393)	(0.482)	(0.392)		(0.499)	(0.401)
WC_host	39.592	39.306	43.667	47.098*	44.195	48.809*	41.326	46.155*	45.773	50.573*		46.051	51.457*
	(30.105)	(27.344)	(29.370)	(26.562)	(30.166)	(26.569)	(31.976)	(27.234)	(29.621)	(26.296)		(31.146)	(26.840)
Kleibergen-Paap Wald F statistic	25.30	10.69	24.28	12.50	23.15	13.73	22.09	14.18	22.72	13.18		20.27	12.22
Anderson-Rubin Wald test	7.70***	11.50***	6.35**	10.22 ***	6.71***	10.98 ***	8.23***	11.13***	6.66**	10.58***		7.49***	9.95***
N	696	696	694	694	693	693	691	691	689	689		682	682
R2	0.529	0.644	0.544	0.647	0.533	0.647	0.507	0.650	0.535	0.644		0.518	0.648

Standard errors in parentheses. * p < 0.1 ** p < 0.05 *** p < 0.01

Turning to H3, we posited shared team tenure will strengthen the relationship between team performance and cultural diversity. We focus on the variable coefficient for the interaction term (Blau $\times$ Time) from Table 4. We find a significant and positive variable coefficient for the interaction term. Therefore, H3 is supported.

Discussion

The present study analyzed a dataset comprised of national soccer teams from the UEFA members to investigate how players’ cultural diversity in terms of nationality affected on-field team success. Differing from prior relevant literature, we identified and tested the moderating effect of shared team tenure in relation to the cultural diversity–team performance nexus by connecting it to tacit knowledge sharing and interpersonal interactions.

Conceptualizing cultural diversity within a team, we considered it to be a variety diversity and measured it by the Blau index, as this type of diversity represents individual differences with respect to task-related knowledge, skills, and perspectives (Harrison & Klein 2007). Based on the results from various IV regression models, in terms of H1, we discovered a positive effect of players’ cultural heterogeneity on overall team performance measured by Elo points, which we can explain from following two standpoints. First, soccer is a sport requiring a great level of interdependence and cooperation among players regardless of their cultural backgrounds on the field. These characteristics may make players with different cultural backgrounds to set aside the “us vs. them” ideology (i.e., social categorization) and focus on winning the match as the collective goal for the teams (van Knippenberg & Schipper 2007). Second, the need to collaborate between team members may also facilitate information processing within the team, which is advantageous to team performance (van Knippenberg et al. 2004). To be specific, based on results from Table 4, other things being equal, a one standard deviation increase in players’ cultural diversity would improve team performance by 76 (M5)-93 (M10) Elo points. When looking at the 2019 rankings, this discrepancy could determine the ranks between the 1^st (Belgium with 2084 points) and 6^th (Portugal with 1986 points) ranked teams.

Pertaining to the testing of H2 in Table 3 (M2), our conclusion regarding a statistically significant and positive impact of team members’ shared team tenure on team performance are consistent with some earlier work in sport (Berman et al. 2002; Humphrey et al. 2009). Besides, as increasing number of national soccer teams have recruited players with different cultural backgrounds, we assume these individuals have country-specific soccer tacit knowledge. Thus, our estimation outcome highlights the fact that regardless of the origins of tacit knowledge, it can be synergized to enhance team performance. In addition to confirming its main effect on team performance, in this study, we moved a step further by examining how shared team tenure may moderate the perceived cultural diversity–team performance relationship (H3). While explicating the positive moderating effect of shared team tenure, on the one hand, we argued if players spend more time together on and off the field, they can learn from their culturally different teammates (either multicultural or naturalized) in terms of their ways of interpreting problems and using their skills to solve them, even if they are equally talented (Ingersoll et al. 2017); on the other hand, stereotypes or biases that resulted from subgroups created by social categorization processes based on salient diversity features such as nationality may be reduced as time passes (Katz 1982; van Knippenberg & Schippers 2007).

In Table 5, based on the estimation results from M4 from Table 4, we presented the marginal effects associated with cultural diversity (Blau) on Elo points with shared team tenure (Time) increasing from the minimum to maximum at the interval of 1. In terms of on-field team performance, the marginal effects of players’ cultural diversity changed from significantly negative to significantly positive as shared team tenure increases. In addition, we plotted the marginal effects at the interval of 0.5 in Fig. 1. These results are consistent with our prior expectation that the positive effect associated with heterogeneity of individual’s cultural differences is reinforced when team members’ shared team tenure increases.

Table 5.

Marginal Effects of Cultural Diversity (Centered) for M4

Shared Team Tenure (Time)	Cultural Diversity (Blau)
1	-374.996**
	(177.740)
2	-283.860*
	(162.431)
3	-192.724
	(148.189)
4	-101.588
	(135.351)
5	-10.453
	(124.354)
6	80.683
	(115.723)
7	171.819
	(110.017)
8	262.955**
	(107.701)
9	354.091***
	(108.993)
10	445.227***
	(113.768)
11	536.363***
	(121.618)
12	627.498***
	(131.994)

Shared team tenure increases from the minimum to maximum at the interval of 1

Standard errors in parentheses. * p < 0.1 ** p < 0.05 *** p < 0.01

Fig. 1.

The Moderating Effect of Shared Team Tenure (Centered) for M4

Theoretical and Practical Implications

The theoretical contributions of this research are two-fold. First, using a dataset of national soccer teams, our results emphasized the importance of controlling the nature of a work setting (independent vs. interdependent) as a contextual factor while analyzing the potential impact associated with cultural diversity. To be specific, the ultimate effect of cultural diversity on team performance may vary depending on the level of interdependence required by the nature of the task (Guillaume et al. 2017; van Knippenberg & Schippers 2007). Second, our identification and testing of shared team tenure as a moderator speaks to the significance of exploring various boundary conditions related to the cultural diversity–team performance relationship (Glennon et al. 2021). More importantly, instead of proxying shared team tenure as team collaboration (e.g., Harrison et al. 1998, 2002) solely, it was viewed conceptually as a variable representing tacit knowledge in the present study. This construct follows the information/decision-making perspective, as team members from different countries/cultures may possess specific skills or knowledge that may be beneficial to the national team on-field performance. Thus, the testing of this moderating effect is driven by theory, which is a recommendation by Cunningham and Ahn (2019) to researchers when conducting such analyses.

In the context of sports, the present study may provide some practical implications. As teams continue in their global search for talent (e.g., Schmidt 2021), we found supporting evidence that recruiting players from multiple cultural origins improved teams’ sporting success. More broadly, organizations in general industries should scout suitable employees globally so as to obtain the potential benefits associated with a multicultural team.

Given the nature of our research setting (i.e., soccer), our findings may speak to the significance of valuing and leveraging employees’ cultural differences in the context of interdependent working environments, which applies to most organizations, if not all, as certain degree of interdependence is an imperative to achieve task goals. First, organizations should create and continuously develop an environment that celebrates diversity and inclusion with the aim of enabling all employees at all levels to participate and make meaningful contributions, regardless of any distinguishing factors (e.g., gender, race, ethnicity). Moreover, as the length that team members spend together (i.e., shared team tenure) may positively moderate the relationship between cultural diversity and team performance, implying a more stable team may reap the benefits of a broader range of task-related skills, perspectives, and experiences that generated from a cultural heterogeneous working force. Thus, upper-level managers need to look at the design and implementation policies that can reduce turnover rates, for example, a pay-for-performance strategy while structuring remuneration system (Shaw 2014). However, how to balance managers’ (high) expectations on team performance and the time a team may need to benefit from its culturally heterogenous members is another important issue while designing relevant human resource management policies. For example, while assessing the performance of a culturally diverse team, specific criteria should be in effect to take into consideration how long the individuals have worked together on the same team.

Limitations

As with all other studies, the present study has several limitations highlighting future directions for research. First, although individuals’ nationalities can serve as a proxy for deep-level indicators of cultural differences (Stahl et al. 2010), solely relying on the Blau index may be oversimplified. For example, while calculating the Blau index of players’ second nationality for the French national team, Spain (Lucas Hernández) and Philippines (Alphonse Areola) carry the same weight. Nevertheless, these two countries are very different in terms of a variety features such as language and culture (in a broad way). Therefore, a more nuanced way of modeling cultural diversity should take these factors into consideration. An alternative would be a (mean) Euclidean distance derived from Hofstede’s cultural dimensions (Maderer et al. 2014). However, the premise of operationalizing cultural diversity in this way would be conceptualizing it as a separation diversity, otherwise incorrect results may emerge (Harrison & Klein 2007).

Second, we did not incorporate coach-related factors into our analyses. However, both literature on general industries (e.g., Raithel et al. 2021) and soccer (e.g., Peeters et al. 2021; Szymanski et al. 2019, 2021) demonstrated leaders’ or coaches’ multicultural backgrounds and working experience have significant effects on team performance. In addition, in this paper, while focusing on the final on-field performance of national teams, we overlooked the process through which coaches may work on the practice field, in the locker rooms, and during the games to achieve the benefits associated with culturally different players. Thus, it would be interesting for future studies to see how these attributes interact with team cultural diversity as well as members’ shared team tenure in terms of affecting overall team performance.

The last limitation relates to the generalizability of our research context. Soccer players might share more in common due to their universal possession of certain soccer skills and may not face some of the same limitations “assimilating” as in other settings. For example, players can still understand drills and plays without knowing the local language whereas not knowing the language could be a much more substantive problem in other contexts. To some extent, the generalizability of results generated from the sporting context may be constrained due to its simplicity. However, this vein of research still provides a type of conservative test or “wind tunnel” test (Kahn 2000; Wolfe et al. 2005). In other words, if an expected relationship cannot be observed in sport setting, then it is highly unlikely that it could be detected under more complicated circumstances. Therefore, to further extend the conclusions of our study, future studies may probe the role of tacit knowledge with respect to moderating the cultural diversity-team performance relationship in general industriesy.

Conclusion

By examining a dataset from the UEFA national teams, our research presented evidence that in the context of a work setting requiring high level of cooperation and interdependence among team members, individual differences in cultural backgrounds are conducive to improved final team outcomes. Moreover, we demonstrated team members’ shared tenure can enhance team performance as it may be a representative of a team’s stock of tacit knowledge as well as better communication and interaction among culturally heterogeneous members within the team.

Authors’ Contributions

The first author collected and analyzed data and put together a draft of the manuscript. The second author assisted in the writing, interpretation of the data, and in the conceptualization of the research project. The third and fourth authors assisted in the conceptualization of the research project and in the writing of the manuscript.

Funding

Not applicable.

Data Availability

Not applicable.

Declarations

Ethical Approval

Not applicable.

Competing Interests

Not applicable.