The Causal Effect of Social Networks on Elite Recruitment

Using information on 12,752 passing candidates of the Chinese civil service examinations (CCSEs) between 1400 and 1580, we provide evidence on the causal effect of social networks on recruitment into a particular labor market, the bureaucracy of imperial China. We use a unique institutional feature of the CCSEs as the identification strategy. The candidates with invariant human capital took two otherwise similar exams within one month apart, but were evaluated anonymous only for one exam. This variation in anonymity provides a necessary condition for varying extent of social network activation. The examiners only had opportunity to activate the exogenously formed networks during the non-anonymously evaluated exams. Even the most conservative estimates show that the relative effect of social networks versus school quality (i.e., one human capital measure) on the recruitment outcomes (relative rankings) is between 216% and 327%. Our research has policy implications for network-induced labor-market inequalities and countermeasures by appropriately designed anonymous evaluation.