This dissertation comprises three chapters. The first two address issues within the education system, focusing on strategies to attract and retain high-quality teachers in the public sector. The third chapter, a collaborative effort with Martin Martinez and Miguel Saldarriaga, explores the application of machine learning tools for nowcasting macroeconomic variables. In Chapter 1, I examine the impact of temporary exposure to the public sector on a teacher's pathway to permanent employment. I study exposure effects in Peru, where candidates for temporary positions take a standardized evaluation and then select schools sequentially based on their score. I use administrative data on teacher evaluation performance, school preferences, and public sector trajectory, to estimate differences in employment outcomes between temporary teachers and external candidates. I find that temporary teachers are one percentage point (p.p.) more likely to secure a permanent position in the next competition, representing a 55 percent increase compared to external candidates. Effects are larger for more competitive candidates, who report a 6.8 p.p. increase (56 percent rise) in their hiring probability. I find evidence that an increased interest in the public sector and changes in school preferences drive these effects. In Chapter 2, I explore factors influencing teacher retention in schools in Peru, where candidates for permanent positions go through a standardized evaluation and school-level screenings. These evaluations, combined with teacher preferences, determine school assignments. Using variance decomposition analysis, I identify location characteristics as the most influential factor, particularly for permanent teachers who prioritize district and school location characteristics, low vulnerability to hazards, and proximity to their residence. Teacher attributes, including prior experience in the public sector, also influence retention, while matching variables suggest improved retention when preferences align. Further investigation of matching effects using school fixed effects and instrumental variables models reveals a small positive effect, requiring additional exploration to assess significance. In Chapter 3, my coauthors and I investigate the use of news data for economic forecasting by analyzing a large collection of Peruvian news articles from 2012 to 2020. Employing topic modeling and sentiment analysis, we extract monthly variables reflecting economic themes and news tone changes. Integrating these variables with traditional predictors, we develop prediction models for GDP growth rate and unemployment rate, accurately predicting the direction of change in approximately 9 out of 10 instances. While early 2020 models struggle to fully capture the extent of the downturn in March, subsequent models produce predictions that closely align with actual values. Models for the pre-COVID period also capture directional changes, but they exhibit reduced efficacy in fully representing fluctuation magnitudes, suggesting weaker predictive ability in typical months. [The dissertation citations contained here are published with the permission of ProQuest LLC. Further reproduction is prohibited without permission. Copies of dissertations may be obtained by Telephone (800) 1-800-521-0600. Web page: http://www.proquest.com/en-US/products/dissertations/individuals.shtml.]