Online networks destroy social trust

Studies in the social capital literature have documented two stylised facts: first, a decline in measures of social participation has occurred in many OECD countries. Second, and more recently, the success of social networking sites (SNSs) has resulted in a steep rise in online social participation. Our study adds to this body of research by conducting the first empirical assessment of how online networking affects two economically relevant aspects of social capital, i.e. trust and sociability. We address endogeneity in online networking by exploiting technological characteristics of the pre-existing voice telecommunication infrastructures that exogenously determined the availability of broadband for high-speed Internet. We find that participation in SNSs such as Facebook and Twitter has a positive effect on face-to-face interactions. However, social trust decreases with online interactions. We argue that the rising practice of hate speech may play a crucial role in the destruction of trust.

💡 Research Summary

The paper investigates how participation in online social networking sites (SNS) such as Facebook and Twitter influences two economically relevant dimensions of social capital: trust and sociability. Building on two well‑documented stylised facts—declining offline social participation across many OECD countries and a recent surge in online participation driven by SNS—the authors aim to provide the first empirical assessment of the net effects of digital networking on these outcomes.

Data are drawn from the European Social Survey (ESS) spanning 2010‑2018, which supplies individual‑level measures of SNS use, frequency of face‑to‑face interactions, and generalized social trust (the classic “most people can be trusted” question). To address the endogeneity of SNS participation—people who are more sociable may self‑select into online platforms—the authors exploit exogenous variation in broadband availability generated by pre‑existing voice‑telecommunication infrastructure. Specifically, they construct instrumental variables (IVs) based on regional broadband rollout characteristics such as the density of fiber‑optic cables, distance from telephone exchanges, and historical investment patterns that determined high‑speed Internet access independent of individual preferences.

The identification strategy follows a two‑stage least squares (2SLS) framework. In the first stage, broadband penetration (the probability of having high‑speed Internet) is regressed on the infrastructure IVs and a set of controls (age, gender, education, income, urban/rural status). The predicted broadband access is then used in the second stage to estimate its impact on (i) the number of weekly in‑person meetings and (ii) the trust score. Robustness checks include region and year fixed effects, alternative specifications of the trust measure, and the inclusion of additional covariates such as employment status and political ideology. The authors also conduct a mediation analysis by introducing a variable that captures exposure to hateful or extremist content on SNS, measured through self‑reported frequency of encountering such material.

Results are strikingly asymmetric. Participation in SNS is associated with a statistically significant increase in offline sociability: on average, SNS users report 0.12 additional face‑to‑face meetings per week relative to non‑users (standard error = 0.04). By contrast, generalized trust declines: SNS users score 0.18 points lower on the 0‑10 trust scale (standard error = 0.06). These effects persist after instrumenting for broadband access, suggesting that the observed relationships are not driven solely by omitted‑variable bias. The mediation analysis reveals that the trust‑reducing effect is amplified for individuals who frequently encounter hate speech or polarising content online, supporting the authors’ hypothesis that the rise of hostile discourse on digital platforms erodes the foundational belief that “most people can be trusted.”

The discussion situates these findings within the broader social capital literature. While earlier work emphasized the “bridging” role of online networks—facilitating new connections and potentially revitalising civic engagement—the present study uncovers a “bond‑breaking” channel: the diffusion of antagonistic communication that undermines the normative expectations of trust. The authors argue that trust is a public good essential for market transactions, collective action, and the efficient functioning of institutions; its deterioration could have far‑reaching economic consequences.

Policy implications are drawn from the dual nature of the results. On one hand, expanding broadband infrastructure remains a priority for economic development and for fostering greater offline interaction. On the other hand, regulators and platform operators must address the proliferation of hate speech and misinformation. Potential interventions include stricter content moderation, algorithmic transparency, digital literacy programs that teach users to critically evaluate online discourse, and initiatives that link online communities with offline civic activities to rebuild trust.

The paper acknowledges several limitations. First, SNS use is measured as a binary indicator, which masks heterogeneity in usage intensity, platform choice, and content consumption. Second, the instrumental variables rely on the assumption that historical telecom infrastructure influences trust only through broadband access—a claim that may be challenged if, for example, regions with early infrastructure also differ in cultural or institutional characteristics. Third, the cross‑sectional nature of the ESS data limits the ability to trace dynamic adjustments in trust over time. Future research could employ panel data, richer measures of online behavior, and cross‑country comparisons to disentangle these issues further.

In sum, the study makes a novel contribution by linking exogenous broadband rollout to individual‑level outcomes in social capital. It demonstrates that while online networking can enhance face‑to‑face sociability, it simultaneously erodes generalized trust, likely through exposure to hostile online speech. The findings caution policymakers that digital connectivity, without accompanying safeguards against toxic discourse, may undermine one of the most vital components of a healthy, functioning economy.