Inferential Expectations

We propose that the formation of beliefs be treated as statistical hypothesis tests, and we label such beliefs inferential expectations. If a belief is overturned through the build-up of evidence, agents are assumed to switch to the rational expectation. Thus, rational expectations is a special case of inferential expectations if agents are unconcerned about mistakenly changing their beliefs (the test size á equals unity), or if there is so much information available about a parameter that it is known with certainty (the sampling distribution of the estimator collapses to a point). We present the results of an individual choice experiment showing preliminary support for inferential expectations in comparison to either rational expectations, or adaptive expectations with one degree of freedom. Depending on how the critical region is determined, either 27-35% or 57-65% of the agents display test size á < 0.9. The impact of inferential expectations is illustrated by showing how it alters a simple model of the exchange rate and a Lucas supply function.