Policy Evaluation in Macroeconometric Doubly Stochastic Models

We consider descriptive macroeconometric models with random coefficients in order to capture the possible relationship between the coefficients in the equations of the variables under the (partial) control of the authorities and those in the behavioral equations as observed in the rational expectation framework. This relationship is the main element of the Lucas Critique. This model family allows us to test for the immunity to this critique and implicitly for the presence of contemporaneous breaks as in the approach proposed by Hendry [1988] and Engle and Hendry [1993]. When the immunity is rejected, we then can also propose a measure of the consequences of some particular changes in the coefficients involved in the policy equation on modeled macroeconomic variables. We illustrate this approach on French data. Starting from descriptive considerations, we specify a system of linear equations with random coefficients to study the response of the French household consumption to the changes in monetary policy linked to the entry in EMS and the Maastricht Treaty.