I just have a basic theoretical question to ask, is DSGE model dynamically inefficient when it deviates from steady state? I am incorporating over generation lapping features into my DSGE model.
There is no general answer to this question. The frictionless Ramsey-Cass-Koopmans model and its stochastic variant, the basic RBC model have agents choose consumption optimally and in a Pareto-efficient way. Thus, there is no dynamic inefficiency in those models. If you now start changing the structure and add frictions, things might change.