Projects

New Methods for Emotion-based Economic Uncertainty Modelling

Mgr. Petr Koráb, PhD.

Members of project
prof. Martin Feldkircher, PhD., Ing. Michala Moravcová, PhD.

• Basic informations
• Characteristics

• Duration of the project: 09/2025 - 10/2027
  ID number: GV-2025-14-P04-Z2

• The project develops new methods for economic uncertainty and geopolitical risk inference using advanced Natural Language Processing (NLP) models. The project improves understanding in two related areas: (1) identifying economic uncertainty, building on the EPU index of Baker et al. (2016), and (2) modeling geopolitical risk, introduced with the pivotal GPR index of Caldara and Iacoviello (2022). It employs an innovative methodology, combining semantic similarity identification with word embeddings from BERT and emotion recognition with RoBERTa. It expands geopolitical risk inference by utilizing qualified risk assessment in speech datasets from central banks, NATO, and the International Monetary Fund. The constructed indices will be accessible to the general public on a project website, with a processing architecture securing regular data updates.