The Wald Problem and the Equivalence of Sequential Sampling and Static Information Costs
10 Pages Posted: 23 Jun 2017 Last revised: 15 Sep 2017
Date Written: September 6, 2017
We consider the sequential sampling problem of Wald (1945). The static cost of attaining any experiment is equal to the expectation of the log likelihood ratio. This equivalence between sequential sampling generalizes to settings where the flow cost depends on the current posterior. The induced cost of any sampling strategy can be represented as a posterior separable cost function. Conversely any posterior separable cost function corresponds to some sequential sampling model with posterior dependent flow cost. As a special case we obtain that Entropy cost (Sims, 2010) corresponds to a flow cost which vanishes when the agent is close to certain about the state.
Keywords: Sequential Sampling, Entropy Cost, Information Acquisition
JEL Classification: D83, C1
Suggested Citation: Suggested Citation