Dimitris Korobilis, University of Essex - Essex Business School
ΑΙΘΟΥΣΑ Τ103, 1ος ΟΡΟΦΟΣ, ΝΕΟ ΚΤΙΡΙΟ ΟΠΑ, (ΤΡΟΙΑΣ 2)
"High-Dimensional Macroeconomic Forecasting Using Message Passing Algorithms"
This paper proposes two distinct contributions to econometric analysis with large information sets and structural instabilities. First, it treats a regression model with time-varying coefficients, stochastic volatility and exogenous predictors, as an equivalent high-dimensional static regression problem with thousands of covariates. Inference in this specification proceeds using standard regression tools such as Bayesian hierarchical priors that shrink many irrelevant coefficients towards either zero or time-invariance. Second, it introduces the framework of factor graphs and message passing inference as a means of designing efficient posterior estimation algorithms. In particular, a Generalized Approximate Message Passing (GAMP) algorithm is derived, and is shown to have very low algorithmic complexity and to be trivially parallelizable. The result is a comprehensive methodology that can be used to estimate time-varying parameter regressions with arbitrarily large number of exogenous predictors. In a forecasting exercise for U.S. price inflation this methodology is shown to work very well.