Abstract
This article introduces two methods to derive internationally comparable skill and occupational distance measures based on machine learning and natural language processing techniques. We apply these measures to produce descriptive facts about employment transitions and workers’ wage distribution in Brazil using all formal labour contracts registered in the period 2003-18. Our findings indicate that workers who use non-routine cognitive skills intensively are better off in terms of employment, wages and switching occupation. Overall, we observe signs of routine-biased technological change and employment polarization following the Brazilian economic crisis of 2014.