This dissertation is comprised of three essays related to teacher labor markets. The first essay describes a theoretical model which incorporates an oft overlooked fact of educational production, namely the fact that teachers are asymmetrically well informed about what actions are best for their specific classes. The model shows that to take advantage of teachers' local knowledge, districts should offer contracts with output-based pay for performance coupled with decentralized decision making and support for teachers to help them set locally appropriate goals. I use data from Minnesota's Q-Comp program to empirically test the model. The data, however, do not confirm (or reject) the theory. The second essay investigates the impact of collective bargaining on teacher contracts using the 2003-04 and 2007-08 Schools and Staffng Survey (SASS) and data from a survey that I administered. Contracts negotiated via collective bargaining have greater returns to experience than do districts without collective bargaining. Unions do not appear to be a roadblock to basing compensation on student performance but they do oppose basing compensation on administrator review and basing tenure on student performance. The third essay turns to an analysis of average hourly wages. Using the American Time Use Survey (ATUS), I compare teachers' wages to demographically similar workers in other occupations. First I estimate that teachers work an average of 34.5 hours per week annually. Using the ATUS data, I conclude that high school teachers earn approximately 11% less than full time college educated workers in other occupations; but elementary, middle and special education teachers are not underpaid relative to full time college educated workers in other occupations.