Peer Effects, Social Ties, and Corruption: Evidence From China

In this paper, I provide the first empirical analysis on peer effects in corruption using observational data. In specifications including fixed effects to control for differences across cities and colleges, and using an instrumental variable approach, I find that hometown and college ties with other officials who have been investigated for corruption are associated with significant reductions in investigation probability. Evidence supports the explanation that when anti-corruption efforts are high, being connected to a corrupt official makes one less willing to engage in corruption for fear of being implicated. Such act of ``drawing the line against the `problematic' politicians'' is deeply rooted in contemporary Chinese political culture. I structurally estimate a networked game-theoretic model, and the estimates also indicate that such spillover of investigations is the major mechanism at play. Counterfactual analysis implies that corruption decreases as the network becomes denser, and as the level of corruption detection or punishment increases.