An Early Warning Model for EU Banks with Detection of the Adverse Selection Effect

by Olivier Brossard of IEP Toulouse & Université Toulouse 1,
Frédéric Ducrozet of Paris Sciences Economiques & Crédit Agricole SA, and
Adrian Roche of Université Paris X & Crédit Agricole SA

April 2007

Abstract: We estimate an early warning model of banks' failure using a panel of 82 EU banks observed between 1991 and 2005. We make two contributions to the literature. Firstly, we construct a distance-to-default indicator and test its predictive power. The tests implemented here are very similar to those realized by Gropp, Vesala and Vulpes (2005), but our time dimension is four years longer and we use a more restrictive definition of banks' "failure". This first part of the paper establishes the accuracy of our data and confirms the robustness of distance-to-default as an early indicator of EU banks' fragility. Our second advance consists in introducing a variable detecting the adverse selection problem that can be caused by rapid growth strategies. A measure of past average growth of assets is shown to be a very significant and powerful predictor of future banks' difficulties. We discuss the origins and implications of such an effect.

JEL Classification: G21, G33, G14, E58.

Keywords: failures, early warning systems, CAMEL ratings, distance to default.

Related reading: Market Discipline and the Use of Stock Market Data to Predict Bank Financial Distress,
Using Securities Market Information for Bank Supervisory Monitoring

Download paper (495K PDF) 24 pages