Conclusions

Our domestic bank failure prediction models highlight the importance of monitoring banks’ loan behavior in monitoring bankruptcy or financial weakness. In both Japan and Indonesia ratios such as loans to deposits and loans to total assets entered statistically significantly with odds ratios generally higher than 1. The ratio of non-performing loans also entered significantly statistically and quantitatively for Indonesia. This contrasts with other studies on corporate bankruptcies in the region, which find liquidity and profitability ratios to be the most important indicators of failure (see for example Shirata (1998)).

Surprisingly, we do not find regulatory capital ratios to be important predictors of bank failure for the period under study. This contrasts with other studies and may be due to the inaccuracy of regulatory capital measures during the period, or to unique domestic regulations, such as the inclusion of latent capital gains on equity to count toward tier II capital for Japanese banks.

The main goal of this study was to explore the usefulness of cross-country models in monitoring the health of the banking sector or providing early warning systems of systemic crisis. We find that such models hold promise. Our cross-country bank failure prediction model displays high percentage of outcomes to be correctly classified, good goodness-of-fit, and high specificity. On diagnostics likely to be of most concern to regulators and policy makers - sensitivity, positive predictive power and type II error – the cross-country model actually out-performed the domestic models. We hope these findings will stimulate regional cooperation on this issue.

Download this Discussion Paper [ PDF 286.6KB| 22 pages ].

Comment(s)

There are [0] comment(s) for this entry. Post a comment.