The Worker's Brain Under the Expert's Hat: Tacit Knowledge, Occupation Codes, and Wage-Setting

About this event

While economists have long studied the returns to easy-to-measure skills, such as years of schooling, it is widely recognized that many tasks are difficult to codify, and require embodied intuitions and past on-the-job experiences to perform successfully. We exploit a little-known feature of the occupational codes in O*NET, where experts and workers are quasi-exogenously assigned to code occupational characteristics for the same occupation over time, to construct a measure of occupation-level ``tacit knowledge". For a given occupation, switching from a worker rater to an expert rater leads to significant lower O*NET rating for a variety of skills and task intensities. We document that 18 prior uses of O*NET by economists significantly differ depending on whether worker or expert ratings are used. Comparing to ground truth measures, workers better understand the amenities and experience requirements of the job, while there is no difference in understanding formal education requirements. We then show these perceived differences in occupational characteristics matter for real-world wage setting using H1-B applications. O*NET skill ratings are used to determine whether a H1-B application is sufficiently specialized, and whether the pay exceeds the appropriate prevailing wage for that occupation. We show that when an expert completes the O*NET survey for an occupation, instead of a worker, the experience requirements fall, and the subsequent H1-B applications within that occupation become more likely to be rejected, with the accepted ones exhibiting lower prevailing, as well as offered, wages. We then embed the disagreement between experts and workers over skills and amenities into a structural hedonic model of wages, and study the consequences of making tacit knowledge explicit. Counterfactual analysis reveals that eliminating worker tacit knowledge reduces 90th percentile worker utility (driven by reallocation to lower amenities), while 10th percentile worker utility rises (due to reallocation to higher productivity, higher amenity jobs). We discuss how labor market frictions prevent wages from revealing all the economically-relevant knowledge possessed by workers.