Abstract:

This paper represents a collection of econometric analyses used to study the relationship between the industrial production and the yield curve. The approach is based on Logit and Probit models estimated using Bayesian inference. Initially, a univariate analysis has been conducted between a binary variable, indicating the state of the industrial production (positive/negative growth), and different variables derived from the term spread. Later, multivariate models were formed, and the Probit models have improved results in terms of statistical significance and performance. Imposing informative priors based on the univariate settings yield improvements. In the context of structural changes, the yield curve keeps a moderate relevance for predicting industrial “recessions” for Romania.