Lasso Regressions and Forecasting Models in Applied Stress Testing

IMF Working Papers, May 5, 2017

Source details

Canonical URL: Lasso Regressions and Forecasting Models in Applied Stress Testing

Model selection and forecasting in stress tests can be facilitated using machine learning techniques.
These techniques have proved robust in other fields for dealing with the curse of dimensionality, a situation often encountered in applied stress testing.
Lasso regressions are particularly well suited for building forecasting models when the number of potential covariates is large, and the number of observations is small or roughly equal to the number of covariates.
The paper presents a conceptual overview of lasso regressions, explains how they fit in applied stress tests, describes advantages over other model selection methods, and illustrates application by constructing forecasting models of sectoral probabilities of default in an advanced emerging market economy.

Lasso regressions help address the curse of dimensionality by performing variable selection and regularization within a single estimation framework.
Lasso is advantageous when:
- The number of potential covariates is large.
- The number of observations is small or roughly equal to the number of covariates.
The paper contrasts lasso with other model selection methods, emphasizing its suitability for forecasting in stress-testing contexts.

Illustrative application: constructing forecasting models of sectoral probabilities of default in an advanced emerging market economy.

Lasso Regressions and Forecasting Models in Applied Stress Testing
- Lasso Regressions and Forecasting Models in Applied Stress Testing (Markdown version)
- Lasso Regressions and Forecasting Models in Applied Stress Testing (PDF)