期刊名称：Euro Area Balance of Payments and International Investment Position Statistics
印刷版ISSN：1830-3420
电子版ISSN：1830-3439
出版年度：2016
出版社：European Central Bank
摘要：This paper presents rst steps toward robust models for crisis prediction. We conduct a horse race ofconventional statistical methods and more recent machine learning methods as early-warning models.As individual models are in the literature most often built in isolation of other methods, the exercise isof high relevance for assessing the relative performance of a wide variety of methods. Further, we testvarious ensemble approaches to aggregating the information products of the built models, providinga more robust basis for measuring country-level vulnerabilities. Finally, we provide approaches toestimating model uncertainty in early-warning exercises, particularly model performance uncertaintyand model output uncertainty. The approaches put forward in this paper are shown with Europe as aplayground. Generally, our results show that the conventional statistical approaches are outperformedby more advanced machine learning methods, such as k-nearest neighbors and neural networks, andparticularly by model aggregation approaches through ensemble learning
关键词：nancial stability; early-warning models; horse race; ensembles; model uncertainty
