Comment on: Reflections on the probability space induced by moment conditions with implications for Bayesian inference

Because assessing an application-specific prior is difficult and, for scientific inference, often not even desirable, quick and standardized approaches to generating priors, like that suggested in Gallant's paper, are useful. However, his argument that his procedure can be given a Bayesian interpretation is flawed. In general, his method requires adjusting the prior and model after seeing the data. It cannot be interpreted as delivering inference based on a prior independent of the data. Variants on his method might have small-sample advantages over the straightforward approach of inverting the frequentist asymptotic distribution to interpret it as a posterior, but it is not at all clear that there are any advantages.