ABSTRACT
Although this method is significantly more tractable than a full solution method, an inherent limitation of this approach is that it cannot admit to the existence of individual-specific unobservables as a component of the state space, generally ruling out forms of serial correlation including permanent unobserved heterogeneity. Furthermore the Emax(·) values calculated from the data are obviously not policy invariant as they depend on the structural parameters of the model. Thus, full solution would be necessary after estimation in order to conduct policy experiments.
