Limitaciones de las metodologías basadas en opiniones de expertos. El caso de la automatización del empleo
Year of publication: 2021
Keywords: Experts, biases, subjective analysis, automation, Frey and Osborne.
JEL Classification: C53, E24, J31, J62, O33
This study examines the problems of reliability of predictions made by human beings (specifically, those considered experts in the subject matter in question). It is found that these people inevitably suffer from biases, heuristics and overconfidence, which cause them to make systematic errors in their predictions. This perspective was applied to the analysis of the investigation of Frey and Osborne (2013) on job automation risk by occupations in the U.S. The results show how, at least for the moment, the predictions of this study seem not to be coming true, both because the experts' opinions seem not to have been too effective, and because the statistical model used seems not to have improved the results. The errors in the experts' predictions are in line with the expected effects of the biases and heuristics previously described.