The first essay studies the long term trend of internal migration in the United States. Over the last forty years, there has only been a modest change in the overall interstate migration rate in the United States. However, different demographic groups have seen very different patterns of changes. The migration rate for families with two college graduate spouses dropped from 5.66% in 1965-1970 to 2.82% in 2000-2005. As for the families with college-graduate husband, it dropped from 4.05% to 2.15% during the same time frame. Interstate migration rates for other types of families or singles have seen little change. This paper extends Mincer’s family migration model into a search framework and directly estimates the effects of female labor force participation, spousal earnings ratio, correlation of earnings from job offers, and home ownership on the migration propensity by using the Current Population Survey (CPS) data in the period of 1982-2005. Endogeniety issues of these variables are appropriately addressed. According to the Blinder-Oaxaca decomposition analysis, we find that the increasing female labor force participation rate and earnings ratio of wife to husband are the primary determinants for the decline in the interstate migration rate of families with two college-graduate spouses and families with a college-graduate husband in the 1980s-1990s. The rising home ownership accounts for a large portion of the decrease in the migration rate of highly educated families, in the 1990s-2000s. The second essay studies the impact of changing youth cohort size on the unemployment rate. Although an increase in youth cohort size is often found to exert an upward pressure on the aggregate unemployment rate, it has been provided some empirical evidences and a theoretical model to the contrary. We find that the estimated elasticity of unemployment rate is quite sensitive in a fixed effect model, with the inclusion of year dummies, when there is a strong temporal correlation between the youth cohort size and the unemployment rate. Both the sign and magnitude of the estimates vary significantly when using data from different time periods. We propose an alternative way to control for the fixed effects and obtain consistent estimates across the time periods in the United States. Our results support the conventional wisdom of positive correlation between youth cohort size and aggregate unemployment rate. This positive effect of the youth cohort size is strongest for the youngest workers and gradually diminishes for older workers, which implies that the young and the prime age workers are not perfect substitutes to the employers.