A Generalized Approach to Indeterminacy in Linear Rational Expectations Models

Modern empirical macroeconomics frequently tests for equilibrium determinacy in linear rational expectations (LRE) models. We propose an augmented representation to solve and estimate LRE models over the entire parameter space, even when the researcher does not know the region of determinacy. The novel approach does not impose restrictions on the parameter space over which the model is estimated and therefore exploits the full informational content of the data. Our methodology tests for determinacy in the wide class of medium- and large-scale models as well as small-scale models for which the region of determinacy is non-trivial. Importantly, the approach is implementable in standard software packages, such as Dynare. We apply our findings by simulating the canonical New-Keynesian model when the Taylor principle is both satisfied and not. We show that our method successfully recovers the true parameter values under the assumption that the researcher does not know the Taylor principle.