Modeling model uncertainty

Recently there has been much interest in studying monetary policy under model uncertainty. We develop methods to analyze different sources of uncertainty in one coherent structure useful for policy decisions. We show how to estimate the size of the uncertainty based on time series data, and incorporate this uncertainty in policy optimization. We propose two different approaches to modeling model uncertainty. The first is model error modeling, which imposes additional structure on the errors of an estimated model, and builds a statistical description of the uncertainty around a model. The second is set membership identification, which uses a deterministic approach to find a set of models consistent with data and prior assumptions. The center of this set becomes a benchmark model, and the radius measures model uncertainty. Using both approaches, we compute the robust monetary policy under different model uncertainty specifications in a small model of the US economy.