An early-warning system for emerging market crises: a machine-learning approach

dc.contributor.advisorSayed, Ayesha
dc.contributor.authorBarow, Dylan
dc.date.accessioned2026-11-16T13:00:20Z
dc.date.available2026-11-16T13:00:20Z
dc.date.issued2026
dc.date.updated2026-11-16T08:46:06Z
dc.description.abstractThis research report develops an interpretable machine-learning early-warning system for financial crises in emerging market economies. While traditional early-warning system models identify key macro-financial predictors, their reliance on linear specifications often fails to capture the non-linear dynamics and cross-country heterogeneity inherent in modern crises. In contrast, recent machine-learning approaches offer greater predictive flexibility, yet are frequently criticised for lacking policy relevance due to their “black-box” opacity. Although the trigger events surrounding a crisis can materialise rapidly, the predictive content of crises lies primarily in the gradual accumulation of macro-financial vulnerabilities over multiple years, a process that is well captured at quarterly frequency and that aligns with the medium-term horizon over which macroprudential policy operates. Using a quarterly panel of 25 emerging market economies from 1995 to 2024, this study constructs indicators for currency, banking, and sovereign debt crises. Gradient-boosting models are estimated within a walk-forward expanding-window framework to generate probabilistic forecasts at four- and twelve-quarter horizons. Performance is assessed using discrimination, calibration, and policy-usefulness metrics that reflect the asymmetric costs of missed crises relative to false alarms. The results demonstrate that ensemble methods, particularly gradient-boosting classifiers, consistently outperform linear benchmarks, with predictive power improving consistently at longer horizons, supporting the focus on medium-term horizons for policy surveillance. Shapley Additive exPlanations (SHAP)-based interpretability reveals that crisis risk is shaped by non-linear threshold effects in domestic credit conditions, external imbalances, and global financial conditions. Overall, the findings suggest that machine-learning frameworks can significantly enhance macroprudential surveillance by combining improved predictive accuracy with the transparency required for timely policy intervention.
dc.identifier.apacitationBarow, D. (2026). <i>An early-warning system for emerging market crises: a machine-learning approach</i>. (). University of Cape Town ,Faculty of Commerce ,Department of Finance and Tax. Retrieved from http://hdl.handle.net/11427/43946en_ZA
dc.identifier.chicagocitationBarow, Dylan. <i>"An early-warning system for emerging market crises: a machine-learning approach."</i> ., University of Cape Town ,Faculty of Commerce ,Department of Finance and Tax, 2026. http://hdl.handle.net/11427/43946en_ZA
dc.identifier.citationBarow, D. 2026. An early-warning system for emerging market crises: a machine-learning approach. . University of Cape Town ,Faculty of Commerce ,Department of Finance and Tax. http://hdl.handle.net/11427/43946en_ZA
dc.identifier.ris TY - Thesis / Dissertation AU - Barow, Dylan AB - This research report develops an interpretable machine-learning early-warning system for financial crises in emerging market economies. While traditional early-warning system models identify key macro-financial predictors, their reliance on linear specifications often fails to capture the non-linear dynamics and cross-country heterogeneity inherent in modern crises. In contrast, recent machine-learning approaches offer greater predictive flexibility, yet are frequently criticised for lacking policy relevance due to their “black-box” opacity. Although the trigger events surrounding a crisis can materialise rapidly, the predictive content of crises lies primarily in the gradual accumulation of macro-financial vulnerabilities over multiple years, a process that is well captured at quarterly frequency and that aligns with the medium-term horizon over which macroprudential policy operates. Using a quarterly panel of 25 emerging market economies from 1995 to 2024, this study constructs indicators for currency, banking, and sovereign debt crises. Gradient-boosting models are estimated within a walk-forward expanding-window framework to generate probabilistic forecasts at four- and twelve-quarter horizons. Performance is assessed using discrimination, calibration, and policy-usefulness metrics that reflect the asymmetric costs of missed crises relative to false alarms. The results demonstrate that ensemble methods, particularly gradient-boosting classifiers, consistently outperform linear benchmarks, with predictive power improving consistently at longer horizons, supporting the focus on medium-term horizons for policy surveillance. Shapley Additive exPlanations (SHAP)-based interpretability reveals that crisis risk is shaped by non-linear threshold effects in domestic credit conditions, external imbalances, and global financial conditions. Overall, the findings suggest that machine-learning frameworks can significantly enhance macroprudential surveillance by combining improved predictive accuracy with the transparency required for timely policy intervention. DA - 2026 DB - OpenUCT DP - University of Cape Town KW - Early-warning systems KW - financial crises KW - emerging markets KW - machine learning KW - gradient boosting KW - macro-financial vulnerability KW - policy usefulness LK - https://open.uct.ac.za PB - University of Cape Town PY - 2026 T1 - An early-warning system for emerging market crises: a machine-learning approach TI - An early-warning system for emerging market crises: a machine-learning approach UR - http://hdl.handle.net/11427/43946 ER - en_ZA
dc.identifier.urihttp://hdl.handle.net/11427/43946
dc.identifier.vancouvercitationBarow D. An early-warning system for emerging market crises: a machine-learning approach. []. University of Cape Town ,Faculty of Commerce ,Department of Finance and Tax, 2026 [cited yyyy month dd]. Available from: http://hdl.handle.net/11427/43946en_ZA
dc.language.isoen
dc.language.rfc3066eng
dc.publisher.departmentDepartment of Finance and Tax
dc.publisher.facultyFaculty of Commerce
dc.publisher.institutionUniversity of Cape Town
dc.subjectEarly-warning systems
dc.subjectfinancial crises
dc.subjectemerging markets
dc.subjectmachine learning
dc.subjectgradient boosting
dc.subjectmacro-financial vulnerability
dc.subjectpolicy usefulness
dc.titleAn early-warning system for emerging market crises: a machine-learning approach
dc.typeThesis / Dissertation
dc.type.qualificationlevelMasters
dc.type.qualificationlevelMCom
Files
Original bundle
Now showing 1 - 1 of 1
Loading...
Thumbnail Image
Name:
thesis_com_2026_barow dylan.pdf
Size:
2.67 MB
Format:
Adobe Portable Document Format
Description:
License bundle
Now showing 1 - 1 of 1
Loading...
Thumbnail Image
Name:
license.txt
Size:
1.72 KB
Format:
Item-specific license agreed upon to submission
Description:
Collections