Machine learning: Optimising forecasting for better decision-making
About the course
Participants on this course will explore machine learning (ML) applications for central banks, focusing on the technology’s role in supporting decision-making with timely, data-driven insights. The training will address ML as a subset of AI and provide an overview of key methods before delving into its role within central bank forecasting. Primary topics include how ML can be leveraged for monetary policymaking, a comparison of efficiency and effectiveness between ML models and traditional econometric methods, and a study of ML applications for nowcasting. This engaging course will equip participants with the essential knowledge to understand the complexities of ML.
Tutors
Karin Klieber
Economist, monetary policy section
Austrian National Bank
Karin Klieber is a research economist in the Monetary Policy Section at the Austrian National Bank (Oesterreichische Nationalbank, OeNB). She specializes in the intersection of applied econometrics and central banking, bridging academic research and policy needs. Her work focuses on inflation, monetary policy, and machine learning in applied econometrics, and has been published in top field journals such as the Journal of Econometrics and the Journal of Applied Econometrics. She has recently been a visiting researcher at the Federal Reserve Bank of New York and has worked at the European Central Bank. She holds a PhD from the University of Salzburg.
Agenda
Introduction to machine learning for central banks
- What is machine learning (ML) and how does it differ from and complement AI?
- ML applications across central bank functions
- Overview of key ML approaches: supervised, unsupervised, and reinforcement learning
Leveraging ML for monetary policy
- Role of ML in supporting monetary policy decision-making with timely, data-driven insights
- Key applications of ML in monetary policy: forecasting and nowcasting, and economic, sentiment, and policy analysis
- Balancing predictive accuracy of ML-produced models with interpretability
Enhancing economic forecasting with ML
- Overview of ML algorithms for macroeconomic forecasting
- Efficiency and effectiveness: ML models as compared to traditional econometric methods
- Increasing model accuracy with ML and generative AI: exploiting rich data, capturing non-linear and asymmetric shock transmission, and improving interpretability to support decision-making under uncertainty
Nowcasting: Real-time inflation intelligence
- Brief introduction to inflation nowcasting: definition, purpose, and role in supporting monetary policy decision making
- Benefits to and limitations of inflation nowcasting with ML
- How are central banks are leveraging ML and generative AI to enhance inflation nowcasting?
Learning objectives
- Understand how ML can be applied across central bank functions
- Learn the role of ML in supporting decision-making
- Discuss the benefits and challenges of applying ML to forecasting
- Identify the differences between ML and traditional models
- Examine methods of inflation nowcasting with ML
Who should attend
Relevant titles and departments may include, but are not limited to:
- Principal/senior economists
- Heads of forecasting
- Research analysts
- Heads of monetary policy
- Policy advisors/analysts
- Heads of statistics
- Chief data/AI officers
- Data analysts
- Heads of information technology
- IT analysts
- Heads of innovation
- Digital transformation leads
- ML engineers
- Economics department
- Forecasting and policy modelling division
- Business cycle division
- Monetary policy department
- Stress-test modelling division
- Statistics department
- Data office
- Information technology department
- Innovation/digital transformation division
