While extensive research has shown that worker referrals enhance performance in organizations with stable employment relationships and regular workplace interactions, their effectiveness in the gig economy—characterized by fluid arrangements and physically dispersed workers—remains ambiguous. The conventional mechanisms underlying referral advantages, namely selection effects through screening and treatment effects through social learning, may not apply in gig work given reduced entry barriers and limited physical interaction. Drawing on granular data from over 150,000 delivery workers across 300 cities in India, we examine whether and how referrals shape worker performance and retention in such contexts. Our analysis reveals that referred workers substantially outperform their peers across multiple performance and retention metrics. To disentangle selection and treatment mechanisms, we examine referrer-referee demographic similarities and social group affiliations, analyze systematic differences in task selection strategies, and track performance trajectories over time. Our results reveal patterns consistent with treatment effects: referee productivity systematically correlates with referrer performance metrics, with these effects particularly pronounced among marginalized groups, indicating that referrals facilitate knowledge transfer rather than merely matching high-potential workers. Additionally, referred workers demonstrate superior knowledge in task selection, and their productivity advantages accelerate with experience. This study advances our understanding of how referral relationships shape worker performance in gig work context while offering implications for platform strategy and worker management in the evolving gig economy.
