Oliver De Groot (University of St. Andrews): Global vs. local methods in the analysis of open-economy models with incomplete markets
Co-authors: C. Bora Durdu and Enrique Mendoza
Local and global (“exact”) solution methods are widely used in international macroeconomics to study incomplete markets models in which precautionary savings drive wealth dynamics. Local methods are faster and free from the curse of dimensionality, but can only approximate the exact solution. We evaluate the accuracy of three local methods that assume a debt-elastic interest rate (log-linear, second-order, and risk-adjusted steady-state) by comparing them with the exact solutions for three small-open-economy models: a canonical endowment-economy model, an RBC model, and a model with an occasionally binding collateral constraint. Comparing results in the time and frequency domains yields two main findings. First, local methods are generally inaccurate, unless the local approximations are centred around the “exact” steady state from the global solutions (i.e., accurate local solutions require solving first with a global method). Second, even then the local methods do not capture well all the statistical properties of the global solution. These findings suggest caution in interpreting results from local methods, and favour using global methods except when the curse of dimensionality makes them impractical.
Research seminars organised by the Bank of Finland's research unit are held on Thursdays at 10:30–11:45 in Rauhankatu 19, 3rd floor big meeting room (unless indicated otherwise). Seminars are held in English.
Research seminars are open to all economists (unless indicated otherwise). Please register in advance at research(at)bof.fi by noon of the preceding day. Visitors will be escorted from Rauhankatu 19 B (Kirjasto/Library) entrance to the seminar room 10 minutes before the seminar.