What is it about?
This study examines banking crises for 49 countries over a period of 40 years. 7 Multinomial logistic models are developed to track cyclical crisis formations, end-to-end. Information on the real sector, banking sector and external sector are used in the models.
Featured Image
Photo by ActionVance on Unsplash
Why is it important?
The aim of the study is to better understand the factors responsible for the formation of a banking crises, as well as its recovery. Level of development, regions, periods and severity have an impact on banking crises. Multinomial logit models, which captures the full crisis cycle, outperform commonly used prediction models including several machine learning methods. One machine learning model, namely gradient boost predicts upcoming crises with highest accuracy.
Perspectives
To our knowledge, this is the first attempt to track the full cycle of banking crises, from tranquil times, to a pre-crisis phase, actual banking crisis and post-crisis period. Each phase exhibits unique information on the progression of a banking crises and subsequent recovery.
Emile du Plessis
University of Hamburg
Read the Original
This page is a summary of: Multinomial modeling methods: Predicting four decades of international banking crises, Economic Systems, June 2022, Elsevier,
DOI: 10.1016/j.ecosys.2022.100979.
You can read the full text:
Contributors
The following have contributed to this page
