Garg and Telang (2018) estimated a structural income–leisure job-search model on survey data from 424 recently unemployed, college-educated LinkedIn users, allocating search effort across five modes (Internet postings, online social networks, friends and family, print media, agencies) and tracing three sequential outcomes (job leads, interviews, offers). They found strong ties raise search effort and outcomes while weak ties generate few interviews and are associated with fewer offers, but they explicitly left the causal and econometric foundations of these estimates open: selection and endogeneity were not addressed by experimental designs, only the search and job-leads equations were estimated jointly, reservation wages were assumed equal to last wages, and cross-mode spillovers in which a lead found through one tie is converted through another were not modeled. Later work partially addressed the platform-scale causal question—most notably large-scale randomized experiments on LinkedIn's People You May Know algorithm showing weak ties causally increase job mobility among members generally—but no study identifies the causal effect of creating or mobilizing online social network ties on unemployed workers' search-effort allocation and sequential outcomes, nor jointly estimates the full nonlinear structural system with an independently identified reservation wage. The problem remains open.

Background

Since Granovetter's (1973) work, sociologists and economists have distinguished strong ties (contacts interacted with frequently, such as close friends and family) from weak ties (infrequent acquaintances) in the job-matching process; the classical hypothesis is that weak ties convey non-redundant job information, while strong ties convey redundant but trustworthy information. Online social networks (OSNs) such as LinkedIn let job seekers maintain far more connections—mostly weak—than offline life permits, which revived the question of which ties actually help. Garg and Telang (2018) studied this with a structural economic model in the income–leisure tradition: an unemployed worker divides weekly hours of search effort sijs_{ij} across five job-search modes jj (Internet job postings, online social networking sites, friends and family, print media, and employment agencies), trading leisure against convex search costs, with the expected payoff governed by the probability that each mode yields job leads, which convert to interviews and then to offers. They recovered structural parameters (mode-specific search-cost functions and the value of employment) from survey data on 424 recently unemployed, college-educated LinkedIn users, using a modular estimator: the search-effort equation and the job-leads equation were estimated jointly by bivariate maximum likelihood (to allow correlated shocks), while the interviews and offers equations were estimated separately on progressively restricted subsamples. Their headline results are that LinkedIn strong ties (connections contacted at least monthly) raise search effort and all three sequential outcomes, whereas weak ties produce a small number of leads that fail to convert to interviews and are associated with fewer offers. The authors themselves flag what these estimates cannot support: tie counts are chosen by the worker (selection), unobserved shocks move effort and outcomes together (endogeneity), reservation wages—the lowest wage a seeker would accept, set equal to each respondent's last wage—may be mismeasured, and a lead discovered through one tie might be converted into an interview through another tie, so the mode- and tie-attributed outcomes are not cleanly separable. Later observational work on Facebook found a different marginal pattern (most job help flows through numerous weak ties, but a single additional strong tie is more valuable at the margin), and platform-scale randomized experiments on LinkedIn's connection-recommendation algorithm established that weak ties causally increase job mobility among members generally—but none of these estimates the search-effort allocation or the leads-to-interviews-to-offers sequence of unemployed workers, and none recovers the structural cost and employment-value parameters.

Problem Statement

The problem, as posed by Garg and Telang (2018) for future work, is to establish credible causal estimates of the effect of online social network ties on unemployed workers' job search, together with the structural estimation refinements needed to interpret such estimates. Concretely: (1) identify the causal effects of the creation or mobilization of online social network ties—strong and weak—on unemployed job seekers' allocation of search effort across search modes and on their sequential outcomes (job leads, interviews, and offers attributable to each mode), addressing the selection, endogeneity, and measurement-error concerns that observational tie counts cannot; (2) recover the structural parameters of the search model (mode-specific search-cost functions and the value of employment) by joint estimation of the full nonlinear system—the search-allocation equation together with all three sequential outcome equations across the modes—rather than the modular piecewise estimation used originally, with reservation wages relaxed or independently identified instead of assumed equal to the worker's last wage; and (3) quantify cross-mode spillovers, in particular whether job leads obtained through weak ties are converted into interviews and offers through strong ties. Because no dataset currently available is sufficient to support credible causal and structural estimates for this population, the problem is posed at the level of identification and estimation design rather than realized estimates. The three components are jointly posed parts of one problem, in which the causal claim of (1) rests on explicit identifying assumptions and the mode- and tie-attributed outcomes are well-defined only through an explicit cross-mode spillover specification. The question is posed for unemployed job seekers generally; the original college-educated LinkedIn-using sample is one population on which generalizability, but not the core identification problem, hinges.

References

  1. Garg, Rajiv, and Telang, Rahul (2018). To Be or Not to Be Linked: Online Social Networks and Job Search by Unemployed Workforce. Management Science 64(8):3926–3941. doi:10.1287/mnsc.2017.2784, https://doi.org/10.1287/mnsc.2017.2784
  2. Granovetter, Mark 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. Gee, Laura K., Jones, Jason J., and Burke, Moira (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
  4. Rajkumar, Karthik, Saint-Jacques, Guillaume, Bojinov, Iavor, Brynjolfsson, Erik, and Aral, Sinan (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
  5. Gee, Laura K. (2018). The More You Know: Information Effects on Job Application Rates in a Large Field Experiment. Management Science. doi:10.1287/mnsc.2017.2994, https://doi.org/10.1287/mnsc.2017.2994
  6. Rhee, Kyungsun Melissa, Hwang, Elina H., and Tan, Yong (2018). Social Hiring: The Right LinkedIn Connection that Helps You Land a Job. SSRN working paper. doi:10.2139/ssrn.3261180, https://doi.org/10.2139/ssrn.3261180
  7. Broder, Hanna Rebecca, Förster, Maximilian, Klier, Julia, Klier, Mathias, and Sigler, Irina (2025). Too old to find employment? A novel approach to leverage the power of digital peer groups for older unemployed. Information Systems and e-Business Management 23(4):999–1038. doi:10.1007/s10257-025-00708-3, https://doi.org/10.1007/s10257-025-00708-3
  8. Porter, Caitlin M., Solanelles, Pol, Cullen-Lester, Kristin L., Pearson, Ashley Howell, and Du, Juan (2025). A Tale of Two Job Searches: An Integrative Review of How Job Seeker Characteristics Shape Informal Job Search Effectiveness. Journal of Organizational Behavior. doi:10.1002/job.2864, https://doi.org/10.1002/job.2864
  9. Tumasjan, Andranik (2024). The many faces of social media in business and economics research: Taking stock of the literature and looking into the future. Journal of Economic Surveys 38(2):389–426. doi:10.1111/joes.12570, https://doi.org/10.1111/joes.12570
  10. De Marco, Stefano, Dumont, Guillaume, and Helsper, Ellen Johanna (2026). The reproduction of structural inequalities in online job search strategies and outcomes. New Media & Society 28(3):1125–1147. doi:10.1177/14614448241310497, https://doi.org/10.1177/14614448241310497
  11. Chen, Lei, Gao, Chaoyue, Leung, Alvin, and Wang, Xiaoning (2025). When Assurance Undermines Intelligence: The Efficiency Costs of Data Governance in AI-Enabled Labor Markets. SSRN working paper. doi:10.2139/ssrn.5693142, https://doi.org/10.2139/ssrn.5693142
  12. Matsuda, Norihiko, and Nomura, Shinsaku (2025). The Temptation of Social Networks under Job Search Frictions. World Bank Economic Review 39(1):1–25. doi:10.1093/wber/lhae010, https://doi.org/10.1093/wber/lhae010