Work

Forecasting inflationwith public high-frequency data during periods of elevated uncertainty

Year: 2026
Type: article
Abstract: Inflation spikes during the COVID-19 pandemic and the 2022–2023 energy crisis exposed weaknesses in standard short-term forecasting, with many forecasters underestimating the size and persistence of price increases. This paper examines whether weekly wholesale and retail prices can improve nowcasts of selected HICP components in EU economies. For Germany, France, Italy, Spain, Poland and the Netherlands (2020–2024), we estimate AR(1), bridge equations and MIDAS regressions, including unrestricted variants, and evaluate one-step-ahead accuracy using RMSE, MAE and R². MIDAS provides clear gains for moderately volatile, well-measured categories, most notably fuels, where R² reaches up to 0.95, and, in some cases, meat (R² around 0.30–0.40 for Poland). For highly volatile items such as fruits and vegetables, MIDAS performs similarly to simple benchmarks, matching results from web-scraped retail datasets. Overall, MIDAS offers a cost-efficient tool for real-time inflation monitoring when public high-frequency data are timely, well structured and closely aligned with CPI definitions.
Language: English
Open Access status: diamond
OpenAlex ID: W7234197188
FWCI:
Cites: 37
Cited by:
Related to: 10