Monetary economy
Seyed Saleh Akbar Mousavi; Behzad Salmani
Abstract
The main purpose of this study is to identify the determinants of banking crisis losses for 49 sample countries over the period 1980-2019. In this regard, two sub-purposes are pursued. In the first preliminary step, we identify and date episodes of banking crises for 49 countries. The graphical analysis ...
Read More
The main purpose of this study is to identify the determinants of banking crisis losses for 49 sample countries over the period 1980-2019. In this regard, two sub-purposes are pursued. In the first preliminary step, we identify and date episodes of banking crises for 49 countries. The graphical analysis of crises showed that about half of the crises were occurred between 2008-2012 in which the share of high-income countries was higher than other country groups. Then, in the second preliminary step, we used the Hodrick-Prescott filter to extract different trends from countries' GDPs to calculate four alternative measures of real output losses. The investigated output losses showed that Angola and Greece had the highest and lowest losses among the four types of losses, respectively. Finally, to achieve the main purpose, we use the Poisson quasi-maximum likelihood (PPML) method to estimate model. The model was estimated without and with currency crisis variable. Our findings show the occurrence of a currency crisis is effective in intensifying output losses following banking crises. Also, the variables of inflation, bank credit to GDP, credit-to-GDP gap, public debt/GDP, with a positive effect and variables of financial openness, discretionary government spending and central bank assets with a negative impact, are important factors in output losses of banking crisis. Therefore, we recommend that the mentioned variables be considered in banking crisis management.
Monetary economy
Seyed Saleh Akbar Mousavi; Behzad Salmani; Jafar Haghighat; Hossein Asgharpour
Abstract
The main purpose of this study is to estimate the probability of banking crisis using the second generation of early warning systems (logit models), for 13 selected high-middle income countries over the period of 1980-2016. In this regard, two types of logit models; binomial and multinomial, are estimated. ...
Read More
The main purpose of this study is to estimate the probability of banking crisis using the second generation of early warning systems (logit models), for 13 selected high-middle income countries over the period of 1980-2016. In this regard, two types of logit models; binomial and multinomial, are estimated. The results of estimated binomial logit model show that three leading indicators of the crisis are broad liquidity ratio, stock price index and inflation, which are the main causes of crisis in the studied countries. These variables account for about 17 percent of the probability of a banking crisis. Then, to avoid post-crisis bias, the multinomial logit model is estimated. The empirical results confirm that above three leading indicators are warning. Also, among the above three variables, only stock price index variable with a probability of 12.68%, causes the economy to exit the banking crisis and change its situation from the crisis/recovery period to the tranquil period. The multinomial logit model exhibit significantly better in-sample predictive abilities than the binomial logit model.