This study conducts empirical analysis on the effects of economic incentives on retaining teachers, both in the late-career and early-career, and improving teacher quality in public schools. The first chapter analyzes whether a large pension enhancement helps to retain late-career teachers in an urban district, St. Louis City in Missouri. Many states enhanced pension benefits of public school teachers during the 1990s. St. Louis followed this trend with a major benefit enhancement in 1999. Descriptive analysis of administrative panel data and simple regression analysis suggest that senior teachers were highly responsive: teaching careers were shortened, and the likelihood of retirement increased. To gain further insight into the long run effect of the plan changes, and the differential effects of various components of the change, I estimate a dynamic structural model, which provides good in-sample and out-of-sample fit. Simulations of retirement behavior based on the estimated model under different pension rules imply a large long-run reduction in the labor supply of senior teachers. The expected years of additional teaching for the current cohort of senior teachers would be increased by nearly 27 percent if they were operating under the pre-enhancement pension rules. The second chapter analyzes late-career teacher turnover induced by pension incentives. Using longitudinal data with performance measures for Tennessee public school teachers, we find higher quality teachers are less likely to retire conditional on age and experience. To quantify the effects of pension incentives, we estimate a structural model for retirement and find that high quality teachers have a lower disutility for teaching relative to retirement. We use the structural estimates to simulate the effect of changes in retirement incentives. Enhancements to traditional plans accelerate teacher retirement, whereas targeted retention bonuses delay retirement and retain high quality teachers at a relatively modest cost. The third chapter focuses on the joint dynamics of attrition behaviors and teaching effectiveness of early-career teachers. Using data from a comprehensive evaluation system in Tennessee, I analyze the factors of attrition, the dynamics of teacher quality, and the retention policies for retaining high-quality teachers. The data suggest teacher attrition is negatively correlated with teacher quality. A sequential probit model of binary attrition and binary rating produces good fits for both survival rate and quality distribution. The model shows the quality of novice teachers increases in experience, salary and education level. Increase in salaries of novice teacher could reduce attrition and improve quality.