Abstract
In Part I the author was concerned with modeling the monthly logarithmic CHF/USD exchange rate, ''s''. He rejected a model proposed by B.T. McCallum. Searching for a more appropriate framework a story by P. De Grauwe was introduced, bringing complex cognitive processes and social practices into the picture. This paper treats again P. De Grauwe's story, supplements it and embedds it into a broader setting showing its links to a subtle concept of agents' rationality. We derive a testable implication of this approach. Although the test result is negative, it will be helpful for the misspecification analysis of B.T. McCallum's model (applied to ''s''), which will finally be presented in Part III of this paper..