In jobs that require substantial amounts of interpersonal interaction a large beauty premium exists. In jobs where attractiveness seems unlikely to truly enhance productivity (like working with information and data) there is no premium

Beauty, Job Tasks, and Wages: A New Conclusion about Employer Taste-Based Discrimination. Todd R. Stinebrickner, Ralph Stinebrickner, Paul J. Sullivan. NBER Working Paper No. 24479
www.nber.org/papers/w24479

We use novel data from the Berea Panel Study to reexamine the labor market mechanisms generating the beauty wage premium. We find that the beauty premium varies widely across jobs with different task requirements. Specifically, in jobs where existing research such as Hamermesh and Biddle (1994) has posited that attractiveness is plausibly a productivity enhancing attribute—those that require substantial amounts of interpersonal interaction—a large beauty premium exists. In contrast, in jobs where attractiveness seems unlikely to truly enhance productivity—jobs that require working with information and data—there is no beauty premium. This stark variation in the beauty premium across jobs is inconsistent with the employer-based discrimination explanation for the beauty premium, because this theory predicts that all jobs will favor attractive workers. Our approach is made possible by unique longitudinal task data, which was collected to address the concern that measurement error in variables describing the importance of interpersonal tasks would tend to bias results towards finding a primary role for employer taste-based discrimination. As such, it is perhaps not surprising that our conclusions about the importance of employer taste-based discrimination are in stark contrast to all previous research that has utilized a similar conceptual approach.