Multinational Banks, Credit Risk, and Financial Crises : A Qualitative Response Analysis

The global financial unrest over the last decade has shifted the attention of banking regulators (Basel II, 2001) in estimating default probabilities for a variety of borrowers. Within a binary choice panel data framework, the current study analyzes various models and cross-examines their performance in identifying financial crises in emerging markets. Using financial ratios, macroeconomic variables, and international factors, the paper identifies a set of warning indicators and discriminates among the three estimators employed. The most important determinants of commercial/official arrears and reschedulings are the debt-to-GDP ratio, inflation, trade liberalization, and the variability of GNP per capita growth. In addition to that, changes in financial flows from foreign investors do affect default frequencies, while external developments are found to be insignificant. Cross-modeling comparison indicates the presence of different exogenous risk factors, depending on the approach employed. Further analysis indicates the presence of heterogeneity, but pertinent estimators fail to perform well. Unlike the fixed- and random-effects estimators, the pooled-logit model yields the minimum number of misclassifications. When past credit performance is taken into account, the significance of some signals is reduced, but the model's misclassification performance is markedly enhanced.