Despite growing evidence on occupational and educational barriers in developing countries, there are few estimates of their effect on the aggregate productivity. This paper measures the magnitude of these barriers and their impact on aggregate productivity using the data on expected occupational choice of students. First, I document striking differences in the impact of students' academic skills on occupational choice across countries. In most developing countries academic skills of students have relatively little effect on skill intensity or earning potential of expected occupations. The observed lower sorting on skills suggests a higher incidence of occupational barriers in developing countries. Next, I evaluate the productivity costs of these sorting patterns by attributing them to latent occupational barriers and calibrate a model of occupational choice based on the Roy (1951) framework. I calibrate the model by combining the data skills and expected occupations from the PISA database with the data from nationally-representative samples of working adults. I find that occupational barriers are particularly high in developing countries in my sample and that their elimination can increase the aggregate output by up to twenty five percent.