Strategic Interactions and Peer Learning in Contests

Many educational environments use rank-based rewards that evaluate individuals relative to their peers. While rewarding relative performance can boost effort, it may also discourage cooperation, as students strategically limit knowledge sharing to protect their rank, potentially reducing total learning. This paper studies how competition shapes individual effort, peer interactions, and skill production, using a structural model and a lab-in-field experiment. The model features heterogeneous agents who choose effort and cooperation under competition. The experiment varies competition intensity and peer matching and collects real-time behavioral data through a web-based learning platform I designed, providing the necessary variation and inputs for structural estimation. Experimental results show that moderate competition enhances both individual and peer learning, while intense competition crowds out cooperation and lowers total learning. Students with similar abilities interact less under intense competition, weakening peer learning. LLM-based analysis of chat interactions confirms that cooperative language declines as competition intensifies. I use the structural model to separate the learning technology from the preference and cost components, allowing peer interactions to influence both learning productivity and effort costs. Peer learning entails lower disutility compared to individual learning. Lower-ability students face higher initial costs of engaging in learning, while peer learning helps mitigate them. Counterfactual simulations show that combining rank-based and piece-rate rewards improves learning outcomes by increasing effort while preserving cooperation.