Abstract
Abstract
Outside of Europe and the United States, the knowledge on skills dynamics is scarce due to a lack of data. We therefore assess whether online data on vacancies and applications to a job board can help fill this gap. We propose a taxonomy with three broad categories – cognitive, socioemotional, and manual skills – and 14 commonly observed subcategories, which we define based on unique skills identified through keywords and expressions. The taxonomy is comprehensive but succinct, suitable for developing and emerging economies, and adapted for online data. We then develop a text-mining approach to implement the taxonomy. Based on Uruguayan data from the job board BuscoJobs, we find that our model is able to assign skills to 64% of applicants' employment spells and 94% of vacancies. While online data are usually skewed toward highly qualified work, we show that our data include meaningful numbers of vacancies and applicants of intermediate and even lower qualification levels. Our approach relies on data that are currently available in many countries, thereby allowing for country-specific analysis that does not assume that occupational skills are constant across countries. This is key as we find considerable differences between our findings and those using US O-NET data. Finally, we end with an illustration of how our approach can inform the analysis of skills dynamics. To our knowledge, we are the first to explore this approach in the context of emerging economies.