Gravity of attention and affection: Evidence from social media interactions during the World Cup
- with Julian HinzCultural affinity shapes a wide range of cross–border interactions. Countries that are culturally closer exchange more goods, services, capital, and ideas – but also experience more frequent political disputes and conflict (Boisso and Ferrantino, 1997; Melitz, 2008; Spolaore and Wacziarg, 2009, 2016). A central challenge for research is that cultural differences, preferences, and identities are themselves endogenous – that is, affected by social, political, and economic factors and hence subject to powers of convergence (Spolaore, 2022). Understanding whether – and through which channels – cultural affinity persists over time and space is essential for explaining current patterns of interaction and exchange. This paper contributes to this agenda by developing a novel measure of cultural affinity across borders, based on social media activity during a worldwide sporting event. By analyzing both the intensity (attention) and tone (sentiment) of cross–border interactions on social media, we investigate the persistence and channels of cultural affinity, and assess how it complements standard proxies such as language, religion, and migration
The term culture originates in sociology, describing ”the languages, customs, be- liefs, rules, arts, knowledge, and collective identities and memories developed by members of all social groups that make their social environments meaningful. [It] is expressed in social narratives, ideologies, practices, tastes, values, and norms as well as in collective representations and social classifications” (American Sociological Association).2 In economic contexts, culture encompasses both the insti- tutional environment within which transactions occur – such as norms of contract enforcement and work organization – and the preferences, communication patterns, and trust that govern interactions across cultural boundaries. While these features have historical origins, they may weaken or transform over time, or conversely, remain remarkably persistent across generations. Most applications of cultural affinity, however, treat the latter as a static component in explaining some economic rationale
This paper proposes a new measure of cultural affinity that captures both its persistent foundations and its dynamic evolution across time and space. We operationalize cultural affinity as a combination of the frequency (attention) and emotional tone (sentiment) of bilateral interactions. Specifically, we analyze geotagged content from the former social media platform Twitter during the 2018 FIFA World Cup.3 Bilateral attention is measured by the number of Tweets sent from one country about another, while sentiment is classified and identified as positive, neutral, or negative using a large language model.
The World Cup offers an ideal setting for analyzing cultural expression through social interaction. It draws simultaneous engagement from a wide range of countries, generates emotionally charged discourse, and allows individuals to openly express national identity and sentiment in ways rarely observed in everyday life (Tomlinson, 1994; Ehrhart, 2014). Participation is also exogenous for non–qualified countries such as the United States, allowing us to isolate cultural engagement from direct national involvement. This enables us to conceptualize cultural affinity as public, observable engagement shaped jointly by slow–moving structural ties—such as language, religion, and ancestry—and short–term shocks like match participation or home–team status. We use this measure to study how cultural affinity is formed, transmitted, and expressed across borders. First, we link bilateral attention and sentiment during the World Cup to historical, geographic, and cultural relationships between countries, estimating their structural and spatial determinants. Second, we examine whether cultural affinity persists across generations of migrants by drawing on fine–grained U.S. Census data on local ancestry. The United States, which did not participate in the 2018 tournament, offers a unique quasi–experimental setting to identify cultural ties independent of direct national involvement. Finally, we explore the economic relevance of these affinities by connecting our social media–based measures to bilateral trade data. We test whether temporary spikes in attention lead to persistent changes in trade, and whether these effects vary across product types, institutional quality, or consumer relevance.
Our empirical contribution is threefold. First, we provide evidence that cultural ties transmitted through migration and ancestry are highly persistent, even across multiple generations and large geographic distances. The latter remains a robust and persistent determinant of cultural affinity, even in an era of digital communication where physical proximity is no longer a technical constraint. Second, we show that not all historical ties are equally durable: while ancestry–based links remain strong, broader indicators such as colonial history are less consistently correlated with present–day affinity. Third, we demonstrate that cultural affinity influences the composition of trade — particularly in differentiated and consumer products and when institutions are weak — and that attention shocks during the World Cup can generate persistent trade effects. This suggests that cultural affinity is not only persistent, but also evolves in ways that matter for economic outcomes. We contribute to three strands of research regarding culture and economics: cul- ture’s impact on economic outcomes, the origins of culture, and the dynamics of cultural change and persistence.
We contribute to three strands of research regarding culture and economics: culture’s impact on economic outcomes, the origins of culture, and the dynamics of cultural change and persistence.
The existing literature has documented robust associations between cultural affinity and economic behavior. Among the most influential contributions are the studies by Guiso et al. (2008; 2009) showing that culture (trust) increases stock market participation and facilitates international trade and investment. Similarly, using panel data on foreign portfolio investment (FPI) from 2001 to 2007, Aggarwal et al. (2012) show that both cultural traits and cultural distance—measured via Hofstede’s dimensions—significantly influence global FPI patterns. While geographic distance typically deters FPI, certain cultural characteristics can mitigate this effect and promote cross–border investment. In the labor market, cultural norms around gender and family significantly affect female labor force participation. Traditional values and strong family ties are associated with lower participa- tion rates (Fern´andez, 2011; Alesina and Giuliano, 2010), while Mocan (2019) finds that a hereditary leisure culture reduces working hours among second–generation immigrant women. Vietze (2012) furthermore provide evidence that cultural proximity between country of origin and country of destination has positive effects on the tourism flows between those countries.
Despite these advances, the role of culture in economic literature is often conceptualized in static and highly aggregated terms. Culture typically enters empirical models as a time–invariant, symmetric way, broadly proxied by shared language, religion, or ethnicity (Alesina and Giuliano, 2015). This practice reflects the widespread use of Guiso et al.’s (2006) definition of culture as “beliefs and values that ethnic, religious, and social groups transmit fairly unchanged from generation to generation.” As a result, culture is often treated as a black box—assumed to be persistent and reciprocal, but rarely measured dynamically or asymmetrically. Yet, as Guiso et al. argue, culture affects economic behavior through evolving beliefs, preferences, and trust, and need not be stable or mutually shared. We address this shortcoming by focusing on the transmission and salience of cultural affinity over time and space, and by leveraging a novel data source that captures real–time, directional expressions of attention and sentiment across countries.
Given culture’s strong impact on economic outcomes, a separate literature explores how it forms and evolves. While some studies emphasize its long–term persistence across generations (e.g. Grosjean, 2011; Acharya et al., 2016; Alesina et al., 2018), others document rapid change in response to political or economic shifts (e.b Inglehart, 2000; Alesina and Fuchs-Sch¨undeln, 2007). These contrasting findings often hinge on the mode of transmission: vertical (parent to child), horizontal (peer to peer), or oblique (across generations via institutions like schools or media) (Mesoudi, 2019). For example, Di Miceli (2019) shows that vertical transmission dominates in culturally isolated settings, while horizontal and oblique channels prevail in immigrant–dense areas. If vertical transmission fosters cultural persistence, while horizontal channels enable rapid change, the rise of digitally connected, peer–oriented social media may fundamentally alter how culture evolves. We contribute to this emerging area by analyzing cultural transmission through online social networks.
Our paper also contributes to the young but flourishing literature using social media as a source of data for scientific research. Similar to our paper, social media has been used to identify cultural heritage and its role in economic outcomes. Social networks shape many aspects of societal structures and identity formation. These include for example patterns of migration and travel, political preferences and social mobility (Bailey et al., 2020). While understanding that these networks can be informative in various aspects in the social sciences, researchers face serious constraints when it comes to the availability of large–scale representative data on social connections across space. For example, Hellmanzik and Schmitz (2017) explore virtual proximity as a novel proxy for cross–border information flows and cultural proximity. They show that bilateral portfolio investment is significantly affected by the latter, indicating that countries which are more closely connected in terms of web content are more integrated financially. Felbermayr and Toubal (2010) use bilateral score data from the Eurovision Song Contest to construct a measure of cultural proximity between countries. They show that it increases trade volumes even after controlling for conventional cultural indicators. Bailey et al. (2020) use friendship connections on the social network Facebook to investigate social connectedness within Europe. Furthermore, using the same data, Bailey et al. (2021) show that two countries that are more socially connected also trade more with each other. In turn, we exploit an original data set that provides us with a synthetic copy of a real–world social network. We evaluate information about the spatial dimension, the frequency and sentiment of interactions between individuals from different locations around the world on the former social network platform Twitter.
Keywords: gravity equation, cultural economics, international relations, cultural affinity, world cup
JEL-Classification: F02, F50, Z10