Robust Inflation-Forecast-Based Rules to Shield against Indeterminacy ∗

Content Provider	Semantic Scholar
Author	Batini, Nicoletta
Copyright Year	2004
Abstract	We estimate several variants of a linearized form of a New Keynesian model using quarterly US data. Using these rival models and the estimated posterior probabilities we then design rules that are robust in two senses: ‘weakly robust’ rules are guaranteed to be stable and determinate in all the possible variants of the model, whereas ‘strongly robust’ rules, in addition, use the probabilities to minimize an expected loss function of the central bank subject to this model uncertainty. We find three main results. First, in our two model variants with the highest posterior model probabilities there are substantial stabilization gains from commitment. Second, an optimized inflation targeting rule feeding back on current inflation will result in a unique stable equilibrium and realize at least three-quarters of these potential gains, even if it is used in a variant of the model that is not the one for which it was designed. Third, the performance of optimimized inflation targeting rules perform increasing less well as the forward horizon increases from j = 0 to j = 1, 2 quarters. For j=2, only a rule designed for our most indeterminacy-prone model is weakly robust and yields determinacy across all models. A strongly robust rule can be designed that sacrifices performance in the least probable models for better performance in the most probable models. JEL Classification: E52, E37, E58
File Format	PDF HTM / HTML
Alternate Webpage(s)	http://repec.org/sce2004/up.26953.1081414445.pdf
Language	English
Access Restriction	Open
Subject Keyword	Click fraud Esthesia Indeterminacy in concurrent computation Loss function Probability Projections and Predictions Property (philosophy) Rule (guideline) Shield Device Component cellular targeting
Content Type	Text
Resource Type	Article