DETERMINANTS OF BANKING CRISES-A SIMULATION ESTIMATION ANALYSIS

We analyse the determinants of banking crises in a large panel of developing and developed countries observed over the 1980-1995 period using a multiperiod probit model. The model allows for random effects and serial correlation in the unobservables and thus is intractable using standard numerical methods. We estimate it using a maximum simulated likelihood method based on the Geweke-Hajivassiliou-Keane (GHK) simulator. We found that banking crises are most likely to develop in countries with high inflation and low growth. Also high past credit growth and high real interest rates can indicate banking sector problems. Results suggest that the presence of deposit insurance schemes seems to increase the probability of a financial crisis. We compare our findings with previous studies that used multivariate logit models to assess the vulnerability to banking crises.