We use a machine learning approach, namely AdaBoost, to rank the determinants of banking crises over time and across countries. We cover a total of 100 countries, advanced and emerging, over the years from 1970 to 2017. The paper first shows that AdaBoost has a better predictive performance than the logit model, both in-sample and out-of-sample; then, it employs AdaBoost to classify the major macroeconomic factors leading to banking crises. The baseline analysis reveals that the US 10 yr Treasury interest rate and world growth play a key role in anticipating a crisis, and that these two variables explain a growing share of the results over time, for both country groups. Other variables, which have been highlighted as important in the literature on crises - such as inflation, current account, public and external debt and credit - are relevant in the lead up to banking crises, but their role has been decreasing over time compared to the aforementioned variables. We also present extensions of the model, which confirm and add to the main results of the baseline model
A machine learning approach to rank the determinants of banking crises over time and across countries / Casabianca, Elizabeth Jane; Catalano, Michele; Forni, Lorenzo; Giarda, Elena; Passeri, Simone. - In: JOURNAL OF INTERNATIONAL MONEY AND FINANCE. - ISSN 0261-5606. - 129:(2022). [10.1016/j.jimonfin.2022.102739]
A machine learning approach to rank the determinants of banking crises over time and across countries
Catalano, Michele;
2022-01-01
Abstract
We use a machine learning approach, namely AdaBoost, to rank the determinants of banking crises over time and across countries. We cover a total of 100 countries, advanced and emerging, over the years from 1970 to 2017. The paper first shows that AdaBoost has a better predictive performance than the logit model, both in-sample and out-of-sample; then, it employs AdaBoost to classify the major macroeconomic factors leading to banking crises. The baseline analysis reveals that the US 10 yr Treasury interest rate and world growth play a key role in anticipating a crisis, and that these two variables explain a growing share of the results over time, for both country groups. Other variables, which have been highlighted as important in the literature on crises - such as inflation, current account, public and external debt and credit - are relevant in the lead up to banking crises, but their role has been decreasing over time compared to the aforementioned variables. We also present extensions of the model, which confirm and add to the main results of the baseline modelFile | Dimensione | Formato | |
---|---|---|---|
ML.pdf
Open Access dal 08/09/2024
Descrizione: Post-print
Tipologia:
Documento in post-print (versione successiva alla peer review e accettata per la pubblicazione)
Licenza d'uso:
Creative commons
Dimensione
1.06 MB
Formato
Adobe PDF
|
1.06 MB | Adobe PDF | Visualizza/Apri |
1-s2.0-S0261560622001425-main.pdf
Solo gestori archivio
Tipologia:
Versione editoriale (versione pubblicata con il layout dell'editore)
Licenza d'uso:
Tutti i diritti riservati
Dimensione
3.64 MB
Formato
Adobe PDF
|
3.64 MB | Adobe PDF | Visualizza/Apri Richiedi una copia |
I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.