A behavioral model of non-linear peer effects in cognitive achievement

Most work on achievement peer effects focuses on the variables that transmit them, and how their size depends on own and peer characteristics. But these effects are generated by different mechanisms, with potentially different policy implications. Also, effort is a decision variable, and because it is unobserved, it may bias the estimation of peer effects. To address this problem, this paper proposes and estimates a model that explicitly considers the student effort decision and achievement production in a class, where own and peer effort interact in the production function. Three mechanisms for peer effects are considered: a production externality of peer effort, the effect of lagged achievement heterogeneity on production, and the effect of peer characteristics in the cost of effort. The model is estimated using data for Chile. Results are very different from a model with no effort, and are similar to studies that identify the causal effect of policies using experimental methods. Policies are predicted to have a much smaller effect when student effort is considered than when it is not, implying that students compensate changes in their peer group characteristics through their effort