This dissertation consists of three essays. In the first essay, I study the long run effect of the Chinese famine on gender difference using a new and high-quality dataset-2004 MEDOW. By comparing the health and education outcomes of pre-famine (1956-1958), famine (1959-1961) and post-famine (1963-1965) cohorts, I find that the long run impact of famine is not gender neutral. Exposure to famine in utero and in early childhood significantly reduces height and years of schooling for adult women, but not for adult men. Thus, famine significantly increases the gender gap. The greater harm of famine for females could be due to son preference or mortality selection, and to disentangle these two effects I use two alternative measures of son preference in ethnic groups. I find that in ethnic groups without strong son preference, the health outcomes for the female famine cohort is relatively better. Although the mortality selection due to the fragile male phenomenon is the main force of the widening gender gap, son preference also appears to play a role. In the second essay, I use the sharp timing of the Great Chinese Famine (1959-1961) to explore the intergenerational transmission of health and education. First, by comparing the health status and education outcomes between children of mothers born before the famine (1956-1958) or during the famine (1959-1961) and children of mothers born after the famine (1963-1965), I find that children of famine-affected women are more likely to be overweight or obese, and are less likely to be underweight. These effects are stronger for children of mothers born in high famine-severity provinces. Second, children of famine-affected women are less educated and have a lower probability of entering middle school. Third, the famine exposure had no significant impact on females' likelihood of marriage, fertility decisions and fertility rate, though it does impact husband's characteristics. In the third essay (with Elaine M. Liu), we conduct a meta-analysis to investigate how changes over time and model specifications could contribute to the differences in estimates of returns to education in China. The results show that approximately 10 percent of the variation can be explained by changes in labor market. Since the reform in 1978, returns to education have increased approximately 0.2 percentage points a year. This increasing trend has reached a stop when the global recession hit China in 2008. We find that returns to education for migrant workers are 2.3 percentage points lower than that of urban workers.