Three essays in empirical labor economics

The first part examines the relatively higher gender earnings gap found in Korea's labor market compared to the US labor market. In contrast to one's perception that there might exist more serious gender discrimination in Korean labor market, most of the differences in the gender gap between these two countries simply stem from the differences in observed characteristics of women among two countries. In particular, relatively lower labor market experience, current job tenure, and educational attainment by Korean female workers play dominant roles in explaining the observed higher earnings gap.;The second part studies on women's rising college enrollment rates during the past three decades. In this paper, I examine several factors that might affect diverging college entrance patterns for young men and women. Using three nationally representative longitudinal data sets of high school students, I find that over the past three decades, women's high school performance---as measured by test scores and the number of math and science courses taken---increased more rapidly then men's. Using a simple theoretical model, I show that the contribution of women's better performance in high school to women's increased college attendance combines both the effects of exogenous changes in how high schools prepare women for college and changes in high school performance induced by women's optimizing responses to increased labor market opportunities.;The third part is an empirical test of the compensating wage differentials. While the model of competitive labor market predicts wage premia in unpleasant jobs, previous empirical studies of compensating wage differentials have not provided a consistent confirmation of the theory. Using a large sample of Displaced Worker Survey, this paper engages in another attempt to test the existence of compensating wage differentials. The paper shows that cross-section estimates of compensating differentials change substantially in the expected direction in differenced displaced-worker data. To explain why using a sample of displaced workers produces more expected results than other studies, the author discusses two factors such as self-selection and measurement error.