Abstract
Regional Chinese infant mortality rates (IMRs) are examined using a stochastic frontier method for the first time. The composite error term method yields estimates of large underreporting of IMRs over time and provinces in China during the past 30 years. China does not follow the standard growth paradigm of more growth leading to lower IMRs. Fiscal decentralization has not alleviated the problem of high IMRs. Both IMRs and the sex ratio at birth suggest reported data constitute a floor or minimal level of demographic distress across provinces with millions of missing females not fully included in the data. China's one-child policy leads to not only underreporting by families but also reporting abuse by local officials who want to be promoted. The hukou system and unbalanced government development policies exacerbate the issue.