This page describes the default algorithm used in RECS to solve for a model's rational expectations equilibrium. It is possible to choose other methods by changing the options.
The numerical algorithm used here is inspired by Fackler (2005) and Miranda and Fackler (1995). It is a projection method with a collocation approach solved by time iteration and approximation of the behavior of response variables.
The method will attempt to find a function that approximates well the behavior of response variables,
where are the parameters defining the approximation. To calculate this approximation, we discretize the state space, and the spline has to hold exactly for all points of the grid. By default, RECS uses a spline approximation of response variables as a function of state variables,
The expectations operator is replaced by a sum over possible realization of shocks, , to which are associated the probability . If shocks are normal, the pairs are calculated by RECS using a Gaussian quadrature. Using this discretization, we can express the equilibrium equation as
For a given approximation, , and a given , this equation is a function of only and can be solved using a mixed complementarity solver.
Once all the above elements are defined, we can proceed to the algorithm, which runs as follows:
Miranda, M. J. and Fackler, P. L. (2002). Applied Computational Economics and Finance. Cambridge: MIT Press.