Real-time Inflation Forecasting Using Non-linear Dimension Reduction Techniques

TL;DR

This paper evaluates the effectiveness of non-linear dimension reduction techniques, like autoencoders, in improving real-time inflation forecasting accuracy compared to linear methods, especially during economic downturns.

Contribution

It demonstrates that non-linear dimension reduction methods can produce highly competitive inflation forecasts and are particularly valuable during recessions and crises.

Findings

Autoencoders and squared principal components improve forecast accuracy.

Non-linear methods outperform linear approaches during recessions.

Controlling for non-linear relations is crucial during economic downturns.

Abstract

In this paper, we assess whether using non-linear dimension reduction techniques pays off for forecasting inflation in real-time. Several recent methods from the machine learning literature are adopted to map a large dimensional dataset into a lower dimensional set of latent factors. We model the relationship between inflation and the latent factors using constant and time-varying parameter (TVP) regressions with shrinkage priors. Our models are then used to forecast monthly US inflation in real-time. The results suggest that sophisticated dimension reduction methods yield inflation forecasts that are highly competitive to linear approaches based on principal components. Among the techniques considered, the Autoencoder and squared principal components yield factors that have high predictive power for one-month- and one-quarter-ahead inflation. Zooming into model performance over time reveals that controlling for non-linear relations in the data is of particular importance during recessionary episodes of the business cycle or the current COVID-19 pandemic.