Skill demand versus skill use: Comparing job posts with individual skill use on the job

Skill requirements in a job post reflect an employer’s “wish list,” but do they also reflect skills used on the job by the hired worker? We compare skill measures derived from the text of online job posts with individual-level data from the Danish Labour Force Survey (LFS) in which participants report their main skills used on the job as free text. By identifying individual workers from the LFS who can be matched to a job post, we validate that the extensive margin skills measures derived from job postings data reflect main skills used on the job. Thus, using job postings data to analyze skill use on the job is generally a valid empirical strategy. However, we also show that heterogeneity in returns to skills is missed if only the extensive margin of skill demand is considered. • We construct skill measures from the text of online job posts (JP). • We compare these to individual data from the Danish Labour Force Survey (LFS). • LFS participants report their main skills used on the job as free text. • We find that extensive margin JP skills capture main skills used on the job. • Heterogeneity in skill returns is missed if the intensive skill demand is ignored.