Affiliations: [a] Department of Economics, Box 9732, Hollins College, Virginia 24020-1732, USA | [b] Schiedhaldenstr. 6, CH-8700 Kusnacht-Zürich, Switzerland
Abstract: Forecasters can sometimes exert influence over the events they are predicting. This paper explores the possibilities for such an internal validation of forecasts in the financial markets. Internal forecast validation comes from model multiplicity and from an unequal closeness of actors to the price determination process. The predictive success of a forecaster's model depends on his closeness to traders. A trader's model in turn may differ from the forecaster's model if he does not adopt the forecaster's predictions. In contrast, the economic theoretician assumes that there exists an objective (or true) model of the market that is also adopted by all market participants. But this model may be unidentifiable. The surprisingly low forecast accuracy of experts in the case of the financial markets is usually investigated from the perspective of model monopoly. The recognition of model multiplicity, where some models are held by actors that are more proximate to the price determination process than others, may resolve this conundrum. This approach, however, also points to the inability to determine forecast accuracy when internal validation is involved.
