Colombia · Bogotá · See the Colombia profile · See the Bogotá profile
Top 75% 47/100 · Ask Evidence Copilot about this practice
Banco de la República's July 2025 paper shows a Random Forest model combining Google Trends data with macro indicators nowcasts Colombian monthly inflation as accurately as the bank's analyst survey — about 1.5 weeks earlier — beating SARIMA, Ridge and Lasso benchmarks.
Colombia's official monthly inflation figure is released with a lag, and Banco de la República's own Monthly Survey of Financial Analysts (MES) takes time to compile, leaving policymakers without an early read on price pressures between releases.
In Borradores de Economía No. 1318 (July 2025), economists Felipe Roldán-Ferrín and Julián Andrés Parra-Polanía built RF-GT, a Random Forest machine-learning model trained on historical inflation data, macroeconomic indicators and Google Trends web-search activity, with hyperparameters tuned through time-series cross-validation.
Tested out-of-sample over 2023-2024, RF-GT consistently outperformed SARIMA, Ridge and Lasso statistical benchmarks and matched the accuracy of the bank's own MES analyst-survey median forecast — while producing its estimate roughly one and a half weeks earlier than the survey is available.
This remains a research working paper from the bank's economics department rather than a confirmed input to the formal monetary-policy decision process, and its out-of-sample track record covers only a two-year window.
Read the full analysis: https://www.banrep.gov.co/en/publications-research/working-papers-economics/enhancing-inflation-nowcasting-online-search
Implementation detail (cost, timeline, staffing, conditions for success) is not yet available for this practice.
Do you run this practice? Claim it — verified implementers get a public contact pathway and can propose corrections.
Where this practice's information was retrieved from, and when.
South Africa
SARB researchers rigorously tested machine-learning models (LASSO, XGBoost, neural networks) against the bank's own published inflation forecasts: ML modestly beat …
India
Reserve Bank of India economists built a hybrid ML and time-series model — adding financial-market data and an uncertainty index …
Switzerland
SNB researchers tested eight machine-learning methods against 1,100+ time series to nowcast Swiss GDP. For the post-2008 period, every ML …
Netherlands
DNB's DFROG nowcasting model has tracked Dutch GDP growth in real time since 2024, publishing monthly estimates; a decade-long evaluation …
Open full copilot Grounded in cited practices — always check the sources.