Forecasting Inflation with Microdata: An Adaptive Machine Learning Approach
Updated
Updated · arxiv.org · Jul 14
Forecasting Inflation with Microdata: An Adaptive Machine Learning Approach
1 articles · Updated · arxiv.org · Jul 14
Summary
Researchers have developed an adaptive machine learning approach to forecast UK inflation using microdata from millions of price changes.
Their pipeline combines high-dimensional price statistics with gradient boosted trees and adaptively integrates micro, macro, and univariate forecasts.
The study finds microdata significantly improve inflation forecasts only after large shocks, such as those seen post-2020, offering new insights for policymakers.