Bayesian Neural Networks for Macroeconomic Analysis

arXiv:2211.04752v4 Announce Type: replace-cross
Abstract: Macroeconomic data is characterized by a limited number of observations (small T), many time series (big K) but also by featuring temporal dependence. Neural networks, by contrast, are designed for datasets with millions of observations and covariates. In this paper, we develop Bayesian neural networks (BNNs) that are well-suited for handling datasets commonly used for macroeconomic analysis in policy institutions. Our approach avoids extensive specification searches through a novel mixture specification for the activation function …

abstract analysis arxiv bayesian big contrast data datasets econ.em networks neural networks paper series small stat.ml temporal time series type