Using recent rural household survey data, we investigate the effect of education on
earnings in China’s rural industry. OLS estimations present a very low return to
schooling, i. e., 1-2% for women and insignificant for men, lower than that reported in
other studies. In order to assess whether the OLS method causes biases in the estimated
returns, we first investigate the possible attenuation bias caused by measurement error,
and then correct the well-known omitted ability bias using various instrument variable
estimations. Following the natural experiment approach, we identify a unique instrument
based on the Chinese culture. Based on the sample used, we find no significant
attenuation bias, but the omitted ability bias appears to overestimate the rate of return.
The instrumental variable estimations (2SLS and GMM) indicate that education has an
insignificant effect on earnings. The institutional factors behind the rural wage structure
and policy implications are discussed.