Working Paper: CEPR ID: DP17646
Authors: Niko Hauzenberger; Florian Huber; Massimiliano Marcellino; Nico Petz
Abstract: We develop a non-parametric multivariate time series model that remains agnostic on the precise relationship between a (possibly) large set of macroeconomic time series and their lagged values. The main building block of our model is a Gaussian process prior on the functional relationship that determines the conditional mean of the model, hence the name of Gaussian process vector autoregression (GP-VAR). A flexible stochastic volatility specification is used to provide additional flexibility and control for heteroskedasticity. Markov chain Monte Carlo (MCMC) estimation is carried out through an efficient and scalable algorithm which can handle large models. The GP-VAR is illustrated by means of simulated data and in a forecasting exercise with US data. Moreover, we use the GP-VAR to analyze the effects of macroeconomic uncertainty, with a particular emphasis on time variation and asymmetries in the transmission mechanisms.
Keywords: Bayesian; Nonparametrics; Nonlinear Vector Autoregressions; Asymmetric Uncertainty Shocks
JEL Codes: C11; C14; C32; E32
Edges that are evidenced by causal inference methods are in orange, and the rest are in light blue.
| Cause | Effect |
|---|---|
| Higher unexpected uncertainty (D80) | Stronger negative reactions in real GDP growth (F69) |
| Higher unexpected uncertainty (D80) | Stronger negative reactions in stock market returns (G41) |
| Positive uncertainty shocks (D89) | Stronger responses than negative shocks (E32) |
| Larger uncertainty shock (D89) | Diminished relationship with real activity (E39) |
| Effects of uncertainty (D80) | Varied over time (N52) |
| GPVAR model (C29) | Improved understanding of dynamics (C69) |