DP17329 Naive Calibration
We develop a model of non-Bayesian decision-making in which an agent obtains an estimate of the state of a relevant economic fundamental but does not know the joint distribution of the two. To make use of the estimate, she relies on an endogenously generated dataset that consists of previous estimates and state realizations. She attributes a systematic difference between the estimates and state realizations in her dataset to a systematic bias in the estimate and naively calibrates it. Her subsequent action affects the probability with which the estimate and the corresponding state realization will be recorded in the dataset that will be used in future decisions. We investigate the steady state of the naive calibration procedure and show that it results in a seemingly pessimistic behavior that is exacerbated by feedback loops. We apply our model to project selection problems and second-price IPV auctions.