Abstract

Online job vacancy (OJV) data has become increasingly popular in labor economics over the past 5-10 years. Most researchers use (expensive) preprocessed data, provided by commercial data providers, with limited information on the data-generating process (DGP) and uncertain data quality. An alternative approach is to collect OJV data (free of charge) via cooperations with data providers and conduct the data preparation in-house. This talk provides a detailed overview of necessary steps for this alternative path, comprising data (i) collection, (ii) enrichment & cleaning, (iii) extraction, and (iv) evaluation & quality control. Performing these steps in-house gives researchers more control over the DGP, allowing for greater transparency. While this approach has many advantages, such as creation of novel taxonomies and more data linkage opportunities, it also involves substantial challenges, notably significant time investments and many unforeseen data issues. I conclude this talk with applications of OJV data and promising avenues for future research, based on my own ongoing research projects: (i) task shifts and firm inequality, (ii) identification & diffusion of AI skills, (iii) identification of digital & green skills, and (iv) automation & augmentation potential of general-purpose technologies.