Approximately Right?: Global v. Local Methods for Open-Economy Models with Incomplete Markets

Global and local methods widely used to study open-economy incomplete-markets models yield very different cyclical moments, impulse responses, spectral densities and precautionary savings. Endowment and RBC model solutions obtained with first-order, higher-order, and risky-steady-state local methods are compared with fixed-point-iteration global solutions. Analytic and numerical results show that inaccuracies in the autocorrelation of Net Foreign Assets resulting from assumptions used to induce stationarity cause the local solutions’ flaws. DynareOBC solutions of a Sudden Stops model with an occasionally binding collateral constraint yield similar flaws and sharply underestimate the effects of the constraint on financial premia and macroeconomic variables. Local methods yield much larger Euler equation errors and are of comparable speed for the Sudden Stops model.