CMStatistics 2018: Start Registration
View Submission - CFE
Title: E pluribus, pauca: Measuring different dimensions of slack in the US economy with an agnostic model Authors:  Gianni Amisano - Federal Reserve Board (United States) [presenting]
Matteo Barigozzi - Università di Bologna (Italy)
Matteo Luciani - Federal Reserve Board (United States)
Abstract: Bayesian inference on the US output gap is taken up. This is done by using a model with a large number of macro indicators. The model parses each series into 3 components: 1) a common cyclical component; 2) a common trend component; 3) an idiosyncratic term to be considered as measurement error. Both the cyclical component and the trend components can be either scalars or vectors and we use statistical criteria to determine their dimensions. The cyclical component loading on GDP and GDI with loading normalized to one is what we call ``output gap'', and it is of particular relevance for the conduct of monetary policy, since it synthesize the current state of the US economy. The model consists of five different blocks: 1) a product/income block; 2) a price block; 3) a wage block; 4) a labor/employment block; 5) an interest rates block, containing interest rates at different maturities. Each block loads on a vector of stationary factors. The identification of cyclical factors is obtained by placing restriction on loading coefficients. We have a strong a priori that the number of cyclical factors is equal to one. This prior will be verified against the data. Each block also loads on non stationary factors that are modeled as random walks with stochastic drift affected by linear constraints. The model is estimated by using Bayesian techniques.