The dissertation studies two specialized labor markets: fashion and science. The essays examine two separate topics in the context of a specialized market. The topics are hiring practice and its consequences for diversity and inclusion and the role of federal investment in a specialized market and its effect on total employment. The first essay provides empirical evidence of race-based hiring quotas in a competitive labor market. Using novel data on the hiring of fashion models for high-end runway shows, I find that Black and Asian models are less likely to be hired if a designer retains more models of the same race from previous shows. The substitution effect between minority newcomers and incumbents of the same race is larger than the substitution effect between a random newcomer and incumbent or a newcomer and incumbent who are White. The substitution effect for minority models is also larger than the same effect derived from a simulation of race-blind hiring. The credentials of Black and Asian newcomers improve with the increase in the percentage of rehired models of the same race in a show. The findings suggest that designers choose a set number of minority models per show and become more selective of minority newcomers when they retain more models of the same race from previous shows.In the second essay, based on the joint work with Margaret C. Levenstein and Jason Owen-Smith, I use the American Recovery and Reinvestment Act (ARRA), a large stimulus package passed into law to combat the Great Recession, to estimate the effect of R&D and science spending on local employment. Unlike most fiscal stimuli, the R&D and science portion of ARRA did not target counties with poor economic conditions but rather was awarded following a peer review process, or based on innovative potential and research infrastructure. We find that, over the program;;s five-year disbursement period, each one million USD in R&D and science spending was associated with twenty-seven additional jobs. The estimated job-year cost is about $15,000.