Underestimating Learning by Doing

Many economic decisions, such as whether to invest in developing new skills, change professions, or purchase a new technology, benefit from accurate estimation of skill acquisition. We examine the accuracy of such predictions by having experimental participants predict the speed at which they will master an unfamiliar task. The first experiment finds systematic underestimation of learning, even after multiple rounds of performance feedback. Replicating earlier findings by psychologists, we observe an abrupt drop in confidence, from overconfidence to underconfidence, following initial task experience. The second experiment shows that underpredicting learning leads decision makers to make choices that lower average payoffs.