In this study, we here examine one of the mostly studied education project in developing countries, the school construction program (INPRES) in Indonesia, on household labor allocation and earning. The effect of (expanding) education on earning has been well- documented in the literature but the results are mixed when it comes to mechanics on how education affects earnings. In addition to revisit its impact on earning, we therefore estimate program effect on allocation of rural household labor supply to establish potential mechanics on how such education project increase earning. In doing so, this study employs the difference- in-difference and instrumental variable methods to utilize the variation resulted from the program. The results show that household labor allocation to non-agriculture sector increased, especially in household where its member(s) born in districts that was exposed to more intensed school construction during the INPRES program period. We also find that the positive effects of higher average of years of education on household earning, indicating a positive return to schooling among these sampled rural households.