Learning with data (quasi-) differencing

Please click here if you are unable to view this page.

ABSTRACT The paper studies the stability of Rational Expectations Equilibrium (REE) under adaptive learning assuming that agents do not know the econometric specification of the REE and are alert to a potential model misspecification, i.e., serially correlated residuals. Their forecasting model may be under-parameterized with omitting some regressors or correctly specified by coincidence. They recursively apply Feasible Generalized Least Squares (FGLS) estimators to address potential misspecifications. In a general class of models, the condition governing the convergence of FGLS learning of under-parameterized (or correctly specified) models to REE is shown to be no stronger than (or identical to) the usual E-stability condition. The stability results are applied to evaluate alternative monetary policies in New Keynesian models allowing for agents' lack of knowledge of the correct specification.
		PRESENTER Pei Kuang University of Birmingham
		RESEARCH FIELDS macroeconomics, monetary economics,
		DATE: 28 Oct 2016 (Friday)
		TIME: 4pm - 5.30pm
		VENUE: Meeting Room 5.1, Level 5 School of Economics Singapore Management University 90 Stamford Road Singapore 178903