Researchers have developed an adaptive machine learning approach to forecast aggregate inflation using microdata, particularly effective in volatile periods. The method combines a gradient boosted trees algorithm with a high-dimensional vector encoding price change distributions. This adaptive pipeline integrates the micro-forecast with benchmarks, using it only when it demonstrates superior performance. The study found that while the micro-forecast only outperformed a univariate benchmark after 2020 in the UK, the combined forecast showed comparable or better results across different time horizons, highlighting the value of microdata following significant economic shocks. AI
IMPACT This research demonstrates how machine learning can improve economic forecasting by leveraging granular data, especially during periods of economic instability.
RANK_REASON The cluster contains an academic paper detailing a new methodology for inflation forecasting using machine learning.
AI-generated summary · Google Gemini · from 2 sources. How we write summaries →
