Using high-stakes grades to incentivize learning

I use a natural experiment in Norwegian high school to investigate how high-stakes grades affect students’ investment in schooling. By exploiting variation across space and time I compare the performance of students taking the same exit exam in compulsory school, but where the test is high-stakes for only a subset of students. Using a staggered triple-difference framework, I find that exam grades increase in the high-stakes setting if students have a sufficient number of prospective high schools within traveling distance. Results from low-stakes ability assessments suggest actual learning - and not test-taking strategy - could largely explain the effect.