Case Study: Georgetown University achieves early-warning prediction of emerging-market financial crises with MathWorks MATLAB

MATLAB Used to Predict Financial Crises in Emerging Markets

Georgetown University economist Dr. Paul McNelis set out to build an econometric early‑warning system to predict and avert financial crises in emerging markets after the 1997 Indonesian turmoil. Facing 13 years of monthly currency and quasi‑money data and large volatility that made traditional linear models inadequate, he used MathWorks' MATLAB and toolboxes (Deep Learning, Statistics and Machine Learning, Optimization, and Spreadsheet Link) to develop a more robust modeling approach.

Using MathWorks' MATLAB platform, McNelis combined linear methods with neural networks (feed‑forward architecture), a genetic algorithm for parameter search, nonlinear optimization, and a time‑varying GARCH proxy for exchange‑rate risk, while Spreadsheet Link supported in‑ and out‑of‑sample evaluation. The MathWorks‑based solution delivered substantially greater predictive accuracy than linear models, helped the Bank of Indonesia forecast money demand and core inflation as an early‑warning system, and produced deployable tools and teaching methods at Georgetown University—yielding measurable improvements in forecasting performance and crisis‑aversion capability.

Georgetown University

Paul McNelis

Economist