Adaptive Neuro-Fuzzy Model Tuning for Early-Warning of Financial Crises

Keywords

Early warning indicators, financial crises, fuzzy logic, adaptive neuro-fuzzy inference systems

Abstract

The paper introduces an early-warning method using multiple financial crises indicators which outputs relevant alerts compared to a precise indication of crisis inception, serving as an effective tool for decision makers. By leveraging fuzzy logic techniques, we design a multi-level fuzzy decision support system based on the evolution of credit growth, housing prices and GDP gap. A neuro-fuzzy approach allows fine tuning of the individual fuzzy sub-systems towards adaptive structures which can follow the particularities of the selected indicators at the individual country level. Simulation results and implementation details are presented, along with conclusion drawn from real economic datasets.

(top)

(back to issue)