Are capital inflow bonanzas a common precursor to banking crises? A categorical data analysis

The empirical literature that relates capital inflow bonanzas to the risk of a banking crisis has, so far, examined the probability of a crisis conditional on a bonanza episode. We argue that estimating the probability of a prior capital inflow bonanza, given the occurrence of a banking crisis, is at least equally important from a policy perspective to limit the exposure to sudden, big surges in capital inflows and reduce the risk of a systemic banking crisis. We consider two global samples, one consisting of 64 countries for 1970–2008 and the other 111 countries for 1980–2008. In both samples, we show that the latter conditional probability is strikingly large, much higher than the former, and even higher if we consider only "systemic" banking crises in developing countries. We also construct 2 × 2 contingency tables, utilise the odds ratio and conduct statistical tests to show that the association between bonanzas and crises is significant and robust across the two samples. The association is much stronger than commonly understood, most notably for systemic crises in developing countries. The findings provide support for stronger policies to limit or dampen unusual surges in international capital flows.