What is it about?
Inflation forecasts are important for households, businesses, investors, and policymakers, but no single forecasting model performs best in every situation. This study examines whether combining multiple forecasting models can improve predictions of consumer price inflation. Using U.S. inflation data, it evaluates a metalearning approach called the Arbitrated Dynamic Ensemble (ADE) and compares it with established forecasting and model-combination methods, including SARIMA, stacking, simple averaging, Fixed Share, and other adaptive combinations. The results show that SARIMA achieved the best overall average rank, while ADE outperformed several widely used combination approaches. The study also finds that ADE performs better when it selectively combines suitable base learners rather than simply using every available model, and that its performance depends on how forecasts are weighted.
Featured Image
Photo by Markus Winkler on Unsplash
Why is it important?
Accurate inflation forecasts affect monetary policy, wage negotiations, investment decisions, pensions, pricing, and many other economic decisions. Because inflation dynamics can change over time, relying on one forecasting model may be risky. Ensemble and metalearning approaches provide a way to combine information from different models and adapt to changing conditions. This study shows that dynamic model combinations can outperform several standard ensemble methods, but also demonstrates that careful model selection and weighting remain essential. The findings therefore help researchers and practitioners understand when sophisticated ensemble forecasting can add value to inflation prediction.
Perspectives
Forecasting is often treated as a competition in which the single best-performing model is selected. However, inflation is a complex and changing economic process, and a model that performs well in one period may not remain the best choice over time. Our results suggest that dynamically combining forecasts can be more effective than relying on simple averaging or fixed combinations. The Arbitrated Dynamic Ensemble is particularly interesting because it attempts to learn which forecasting models are likely to perform well under different conditions. At the same time, the results show that complexity does not automatically guarantee superior forecasts: SARIMA remained the strongest individual performer. Effective ensemble forecasting therefore depends on selecting appropriate base models and designing suitable weighting mechanisms.
Prof. Afshin Ashofteh
Universidade Nova de Lisboa
Read the Original
This page is a summary of: Ensemble Methods for Consumer Price Inflation Forecasting, January 2023, Associacao Portuguesa de Sistemas de Informacao,
DOI: 10.18803/capsi.v23.317-336.
You can read the full text:
Contributors
The following have contributed to this page
