Granovetter's strength-of-weak-ties theory predicts that weak social ties are the most useful channel for job finding, yet an observational study of millions of Facebook users found that, per tie, stronger ties are more likely to transmit jobs. The unresolved problem, stated in the source study's limitations, is threefold: (i) identify the causal effect of exogenously changing network formation and tie strength on job finding, since tie formation and tie strength are endogenous to unobserved worker and dyad characteristics; (ii) determine whether jobs found through ties of different strength differ in match quality — wages, tenure, and subsequent career paths — outcomes the original data could not observe; and (iii) establish how labor-market referral networks should be designed when the objective is not only employment but the quality of the worker–job match. A 2022 large-scale experiment on LinkedIn's connection-recommendation algorithm provided the first experimental evidence on part (i) for one platform, but the match-quality and design components, and the platform-general causal question, remain open.

Background

Since Granovetter's classic study, the sociology and economics of labor markets have held that 'weak ties' — acquaintances contacted infrequently, spanning otherwise disconnected social circles — are disproportionately useful for job finding, because one's close 'strong ties' circulate largely the same, redundant information (Granovetter 1973, DOI 10.1086/225469). Testing this claim was long limited by small surveys and self-reported tie strength. Digital platforms changed this. Gee, Jones and Burke (2017, Journal of Labor Economics 35(2):485–518, DOI 10.1086/686225) used data on millions of Facebook users to study 'job transmission': they proxy job help by a sequential job, meaning a user starts working at an employer at least one year after a Facebook friend already worked there (and at least one year after the two became friends). Tie strength is measured by three objective indicators — the share of a user's photo tags and wall posts involving the friend, and the share of mutual friends. In dyad-level regressions with user fixed effects (indicators that absorb all time-invariant traits of the user, such as sociability), stronger ties are significantly more likely to transmit a job, although, because weak ties are far more numerous, most people who find a job through a contact do so through one of their many weak ties. A companion cross-country survey study found self-reported weak-tie job finding to be the dominant pattern across 55 countries (Gee, Jones, Fariss, Burke and Fowler 2017, DOI 10.1016/j.jebo.2016.12.004). The Facebook study is, however, purely observational: tie strength is endogenous — stronger contacts may share workplaces for reasons (for example shared origins or interests) that also drive co-employment, so a tie-strength effect cannot be separated from dyad-level confounding by covariance alone. The study also observes only job starts, not job-match quality — the wage, tenure, and career-path consequences of the job obtained — and it explicitly leaves open how labor-market referral networks (the social structures through which workers learn about and obtain jobs) should be designed when the goal is a good match rather than any job. The only large-scale experimental evidence to date comes from LinkedIn: Rajkumar, Saint-Jacques, Bojinov, Brynjolfsson and Aral (2022, Science 377(6612):1304–1310, DOI 10.1126/science.abl4476) exploited 5 years of randomized variations in LinkedIn's 'People You May Know' recommendation algorithm — the ranking that suggests new connections to members — affecting over 20 million people, and found that moderately weak ties (by mutual connections) and the weakest ties (by interaction intensity) create the most job mobility, with an inverted-U relationship between tie strength and job transmission and strong ties helping in less digital industries. That experiment measures job mobility only; it contains no wage, tenure, or match-quality outcomes, and it varies ties created by algorithmic recommendation on a professional network, not naturally formed ties on a general-purpose platform such as Facebook.

Problem Statement

For labor-market networks mediated by online social platforms, the problem, posed by the limitations of Gee, Jones and Burke (2017) and left open by the single-platform, single-mechanism scope of the 2022 LinkedIn experiment (Rajkumar et al.), is to establish: (i) the causal effect on job finding of exogenous changes in network formation and in the strength of ties within the network — that is, estimates that do not confound tie strength with unobserved worker and dyad characteristics such as homophily, obtained from randomized or otherwise credibly exogenous variation in the network, and valid beyond a single platform and a single algorithmic recommendation mechanism; (ii) whether jobs found through ties of different strength differ in match quality, meaning the wages, tenure, and subsequent career paths of the workers obtaining them, and not merely the probability of employment; and (iii) on the basis of (i) and (ii), how labor-market referral networks and the platforms that mediate them should be designed when the objective is improving the quality of the worker–job match rather than only the rate of job finding. The design conclusions in (iii) are derived from, and consistent with, the causal estimates in (i) and (ii).

References

  1. Gee, L. K., Jones, J., & Burke, M. (2017). Social Networks and Labor Markets: How Strong Ties Relate to Job Finding on Facebook's Social Network. Journal of Labor Economics 35(2), 485–518. DOI: 10.1086/686225. https://doi.org/10.1086/686225
  2. Granovetter, M. S. (1973). The Strength of Weak Ties. American Journal of Sociology 78(6), 1360–1380. DOI: 10.1086/225469. https://doi.org/10.1086/225469
  3. Rajkumar, K., Saint-Jacques, G., Bojinov, I., Brynjolfsson, E., & Aral, S. (2022). A causal test of the strength of weak ties. Science 377(6612), 1304–1310. DOI: 10.1126/science.abl4476. https://doi.org/10.1126/science.abl4476
  4. Gee, L. K., Jones, J. J., Fariss, C. J., Burke, M., & Fowler, J. H. (2017). The paradox of weak ties in 55 countries. Journal of Economic Behavior & Organization 133, 362–372. DOI: 10.1016/j.jebo.2016.12.004. https://doi.org/10.1016/j.jebo.2016.12.004
  5. Brown, M., Setren, E., & Topa, G. (2016). Do Informal Referrals Lead to Better Matches? Evidence from a Firm's Employee Referral System. Journal of Labor Economics 34(1), 161–209. DOI: 10.1086/682338. https://doi.org/10.1086/682338
  6. Ki, N., Yoon, T., & Swanson, J. W. (2024). Do not ask for job information from your close friends or from strangers: strength of moderate ties in finding a satisfying job. Current Psychology 43(12), 10540–10556. DOI: 10.1007/s12144-023-04850-4. https://doi.org/10.1007/s12144-023-04850-4
  7. Ruiz-Palazuelos, S., Espinosa, M. P., & Kovářík, J. (2023). The weakness of common job contacts. European Economic Review 160, 104594. DOI: 10.1016/j.euroecorev.2023.104594. https://doi.org/10.1016/j.euroecorev.2023.104594
  8. Gyetvai, A., & Zhu, M. (2025). Coworker networks and the role of occupations in job finding. Labour Economics 94, 102699. DOI: 10.1016/j.labeco.2025.102699. https://doi.org/10.1016/j.labeco.2025.102699
  9. Jahani, E., Fraiberger, S. P., Bailey, M., & Eckles, D. (2023). Long ties, disruptive life events, and economic prosperity. Proceedings of the National Academy of Sciences 120(28), e2211062120. DOI: 10.1073/pnas.2211062120. https://doi.org/10.1073/pnas.2211062120
  10. Lleras-Muney, A., Miller, M., Sheng, S., & Sovero, V. (2026). Party On: The Labor Market Returns to Social Networks in Adolescence. Journal of Labor Economics 44(1), 25–52. DOI: 10.1086/731308. https://doi.org/10.1086/731308
  11. Fernandez, R. M. (2021). Strength of Weak Ties in the Labor Market: An Assessment of the State of Research. In: Personal Networks (Pescosolido, B. A., & Smith, E. B., eds.), pp. 251–264. Cambridge University Press. DOI: 10.1017/9781108878296.019. https://doi.org/10.1017/9781108878296.019