"A contrast with John Taylor’s approach to investigating the properties of models is instructive. Taylor’s research program includes the use of computational experiments. It is well summarized in his recent book (Taylor, 1993). Like Kydland and Prescott, Taylor relies on fully specified dynamic models and imposes rational expectations when computing stochastic equilibria. However, in fitting linear models he uses all of the information on first and second moments available in the macro data when it is computationally possible to do so. The econometric methods used in parameter estimation are precisely described. Multiple sources of business cycle shocks are admitted into the model at the outset, and rigorous empirical testing of models appears throughout his analyses."