This immersive 5-day program equips policy analysts and central bank teams with state-of-the‑art AI tools—LLMs, ensemble ML, and nowcasting methods—for inflation modeling and rate‑path forecasting. Delegates will learn to harness unstructured data, generative AI, and neural forecasting models to refine policy signals, monitor volatility, and improve macroeconomic scenario planning in today’s high‑uncertainty global environment.

TRAINING TOPICS

- LLM-based inflation nowcasting and short‑term forecasting

- Ensemble methods (random forests, XGBoost) vs classical econometrics

- Transformer neural nets for inflation modeling

- Integrating high-frequency and alternative data (payments, satellite, text sentiment)

- Agent‑based and DSGE models augmented with reinforcement learning

- Generative AI workflows in central banking (e.g., BIS Project Spectrum)

- Monitoring polycrisis with big-data LLM classifiers

- Governance, explainability, and AI model risk frameworks

BY THE END OF THIS TRAINING COURSE, DELEGATES WILL BE ABLE TO:

- Implement LLM-based nowcasting pipelines using real-time data

- Develop ensemble and transformer AI models for inflation forecasting

- Incorporate alternative data (payment flows, satellite, text) into forecasts

- Embed agents and RL mechanisms in DSGE or policy simulation frameworks

- Use AI-driven macroeconomic dashboards for volatility and risk monitoring

- Evaluate model performance, bias, explainability, and central bank governance standards

- Design AI-assisted policy scenarios accommodating polycrisis and structural uncertainty

- Build inter-department workflows combining AI with standard econometric tools

TRAINING IS TAILORED TO

- Central bank economists and monetary policy analysts
- Research and forecasting teams in finance ministries and international institutions
- IMF, BIS, OECD, BIS‑IH, SUERF, and financial policy think‑tank professionals
- Financial risk and data science teams in sovereign funds, multilateral banks

TRAINING METHODOLOGY

- Expert-led sessions with live AI model builds and dataset ingest
- Case studies from ECB, Fed, BoE, BIS (Project Spectrum, PaLM nowcasts)
- Hands-on labs: Python, R, TensorFlow, HuggingFace Transformers
- Working groups: scenario design, polycrisis stress-testing
- Model governance workshops on explainability and regulation
- Final capstone: deploy inflation forecasting toolkit with real central bank data

DAY ONE: LLM-DRIVEN INFLATION NOWCASTING

- The evolving central bank data landscape: satellite, transaction, text streams

- Introduction to LLMs (PaLM, GPT) for macroeconomic nowcasting

- Hands-on: PaLM/GPT prompts to forecast inflation

- Benchmarking vs Survey of Professional Forecasters (SPF)

- Governance: hallucination risk and model validation

DAY TWO: MACHINE LEARNING & ENSEMBLE FORECASTING

- Random forests, XGBoost, and neural nets for inflation prediction

- Transformer architectures for macro time series

- Lab: Build and compare forest, transformer, and ARIMA models

- Explainability with Shapley values and LIME

DAY THREE: ALTERNATIVE DATA FOR MACRO NOWCASTING

- Payment system datasets and real-time indicators

- Web-scraped prices and sentiment signals

- Satellite imagery for input‑output dynamics

- Lab: Fuse high-frequency data into inflation models

- Discussion: Data governance and privacy

DAY FOUR: AI IN POLICY SIMULATION & DSGE/ACE MODELS

- Agent-based & reinforcement learning in policy simulations

- Augmenting DSGE models with AI agents

- Lab: Implement a simple policy simulation with RL

- Scenario workshops: polycrisis, supply-chain shocks, energy transitions

DAY FIVE: GOVERNANCE, DEPLOYMENT & FUTURE AI IN POLICY

- Governance frameworks at central banks

- Platforms: HuggingFace, Azure, Fed's ARI tools

- Lab: Build a forecasting dashboard with Python/BI tools

- Case study: BIS Project Spectrum & ECB generative nowcasts

- Plenary: Action plans, integration strategies, collaboration roadmap