Top 6%
in this catalogue (246 scored practices)
Scores cluster high, so position within the catalogue is often more telling than the number alone.
Evidence of impact / measured public value2/3
Peer-reviewed backtests (2014-2018) found NBP's web-scraped eCPI models cut RMSFE by ~27-30% vs ARMA benchmarks, re-confirmed during COVID-19 and the 2022 food-price shock, though no published evidence ties it to a specific rate decision.
Transparency, fairness & accountability2/3
Full methodology, data construction, and benchmark comparisons are published in an official NBP working paper and two peer-reviewed journal articles, open to external scrutiny.
Transferability / demonstrated replication3/3
The original 2019 method was independently extended and re-tested by NBP researchers in 2023 and 2025 studies covering different shocks (COVID-19, war-driven food prices), demonstrating replication over time.
Scalability beyond pilot3/3
The underlying web-scraped price database grew from its initial collection to roughly 250 million prices covering ~640,000 products by the 2025 study, evidencing real scale-up.
Governance, capability & sustainability2/3
The programme has run inside NBP's Economic Analysis and Research Department continuously since at least 2019, with successive published iterations, though no dedicated oversight body is described.
Evidoria. Previsão Imediata da Inflação Alimentar com Preços Recolhidos Online — o Modelo de Big Data do Narodowy Bank Polski para Estimativas de IPC em Tempo Real. Persistent ID: eae15ebb-ed22-4409-aa1b-11987568bb77.