This paper analyzes the impact of a recruitment policy aimed at improving the quality of new teachers. The reform introduced a scholarship to incentivize the enrollment of high-achieving high school graduates in teacher training programs and imposed enrollment restrictions on low-achieving high school graduates. The screening device used to define achievement was the national standardized university entry exam. Using rich administrative data, we document that the reform was effective in improving the average test scores of new teachers. We construct teacher value-added (TVA) measures based on standardized test scores of their pupils and find that the reform led to a 30% standard deviation increase in the TVA of mathematics teachers. However, it did not affect the average TVA of Spanish teachers. We show that this heterogeneity can be explained by differences in the correlation of test scores and TVA across subjects. Finally, we show that only 20% of the increase in TVA can be explained by the selection based on the screening device, and the remainder can be attributed to beneficial but unintended effects of the reform. Using survey data, we show that a large share of this remaining effect is driven by the screening out of those low-achieving graduates who were less intrinsically motivated.