Is it Worth Refining Linear Approximations to Non-Linear Rational Expectations Models?

We characterize the balanced growth path of the basic neoclassical growth economy using standard, almost linear numerical solution methods, as well as the parameterized expectations approach, which preserves the nonlinearity in the model. We also apply the same methods after adding indivisible labor to the basic model, and to a monetary version of that economy, subject to a cash-in-advance constraint. In a unified framework we tackle the question of how much of the nonlinear structure of the original problem is useful to maintain when using an "almost" linear method. We show that it is possible to find an almost linear method to solve these models as accurately as by parameterizing expectations. Our results show the importance of performing log-linear approximations, as well as the convenience of refining a linear solution method by mixing some structure of the original non-linear problem with structure of the approximated system.div>