RATIONAL EXPECTATIONS AND ECONOMIC THEORY:
We assumed above that the role of economic theory is not to provide quantitative predictions about the future. Suppose we assume instead that the primary role economic theory is the prediction of future values of variables of interest. Also, give this role of economic theory, suppose we stipulate that under any particular set of circumstances, the choice amongst alternative theoretical models should be possible on the basis of quantitative historical data alone. It implies that models should provide/predictions in the form of relations between objectively measurable variables, predictions which can be checked using past data on these variables.
Note also that the above assumption implies any one of two things:
i) the same relation between variables have prevailed for the times and places that the past data relates to and will prevail under the present circumstances, or
ii) differences in time and space can be accounted for as different values of quantifiable variables within the model.
Therefore, once a model is chosen on the basis it can be assumed to provide, together with historical data on the variable in the model, objective quantitative predictions. Traditionally, this has meant that models have been used to yield predictions in the form of an objective probability distribution for the future values of dependent variables, given (conditioned on) current and past values of a set of variables.
This objective conditional probability distribution has been derived by econometric methods on the basis of past data on the variables appearing in relationships derived from theory. The basic assumption underlying such an approach is that either there is an objective conditional probability distribution relating these variables which does not change with time and space or that time and span ban themselves be specified in terms of quantifiable variables in the model.