Predicting the Likelihood of Systematic Banking Crises in Selected Developing Countries (An Approach from the Multinomial Logit Model)

The purpose of this article is to predict the likelihood of systematic banking crises in selected developing countries, so that economic policymakers can study and deal with this crisis and reduce the likelihood of its occurrence. For this purpose, 37 developing countries were selected and the probability of a banking crisis for the selected countries during the years 1994-2018 was estimated using the binary and multinomial logit model. The results showed that in the multinomial logit model, the percentage of correctly predicted critical periods is higher than in the binary logit model, and the multinomial logit model is more appropriate. The results of the multinomial logit model indicate the ineffectiveness of exchange rate deviation and the positive effect of inflation, real interest rate and trade relations on the probability of a banking crisis in the countries under study. Economic growth rate, per capita production and capital flow, and the ratio of the ratio of bank redits to private sector to production have had a negative effect on the likelihood of a banking crisis in the countries under study. The results also showed that the ratio of private sector credit to production was not a good predictor of the possibility of a banking crisis in the countries under study