Changing Wage Effects of Educational Mismatch in China: Evidence from Threshold IV–Selection Models

This study examines the wage effects of educational mismatch in China by jointly addressing sample selection, endogeneity, and nonlinear career-stage heterogeneity within a unified econometric framework. Although educational mismatch has been widely studied, existing evidence largely relies on linear models that overlook experience-dependent wage dynamics and potential selection and endogeneity biases. Using data from the 2020 wave of the China Family Panel Studies (CFPS), this study extends the Duncan–Hoffman model by integrating a sample-selection-corrected threshold regression estimated via instrumental variables. This approach allows the identification of experience thresholds at which the wage effects of overeducation and undereducation differ across regimes. The results reveal pronounced nonlinearities in mismatch-related wage differentials. Overeducation is associated with wage penalties at early career stages, but these penalties weaken and, in some cases, disappear once workers surpass the estimated experience threshold. In contrast, undereducation yields modest wage premiums early in the career but becomes increasingly penalized at higher experience levels. Substantial gender heterogeneity is also observed: male workers are better able to use accumulated experience to offset educational shortfalls, whereas female workers face more persistent penalties, particularly at later career stages.