This paper investigates the determinants of living standards (measured by per capita consumption expenditure) at the household level, addressing heterogeneity in returns to education and endogeneity of educational status. The estimation results obtained through an instrumental variables quantile regression suggest that the endogeneity of education matters in determining the causal effect of education on living standards, while no evidence of the heterogeneity in the rate of returns to education is found. However, the results also provide evidence that impacts of other determinants vary significantly over the outcome (expenditure) distribution, and consequently a simulation based on the results shows that poverty alleviation impacts of education differs substantially between the instrumental variables quantile regression and standard instrumental variables regression results. The comparison of the two indicates the possibility that the impact on poverty reduction is likely to be overestimated in the standard instrumental variable regression.