Reassessing the Use of Aggregate Prudential Ratios to Identify Banking System Problems

Abstract: Building on an early warning baseline model, this paper analyzes the application of macroprudential banking ratios for an economy's five largest financial institutions, constructed from balance-sheet data, on the probability of systemic banking crises occurring relative to the sector-wide aggregates that are commonly used by the International Monetary Fund (IMF) and other authorities. The investigation is motivated by the observation that the distribution of bank assets is highly asymmetric in advanced economies, the fact that an economy's largest banks are often implicated in systemic banking crises, as well as theory and empirical evidence demonstrating the large impact of shocks originating at large banks. For a sample of 25 advanced economies between 1997-2008, a multivariate logit model estimates the effect of a vector of commonly applied macroeconomic indicator control variables, as well as the aforementioned banking ratios for large banks and the banking sector aggregates. The findings support the hypothesis that some characteristics in large financial institutions can be used to identify banking sector turmoil more accurately than their aggregate analogues, however exclusive reliance on these indicators is not advisable.