The financial crises of the past decade, and the IMF’s initiative to build an early warning system against such crises, have stimulated a wave of research into the empirical determinants of banking crises; e.g. Bordo et al. (2001), Borio and Lowe (2004), Caprio and Klingebiel (1996ab, 2002, 2003), Demirgüc-Kunt and Detragiache (1998, 2002), Demirgüc Kunt et al (2000), Drees and Pazarbasioglu (1995), Flannery (1996), Gavin and Hausman (1996), Glick and Hutchison (2001), Goldstein and Turner (1996), Goldstein et al. (2000), Kaminsky and Reinhardt (1996, 1999), Lindgren et al. (1996, 1999). A common methodological challenge facing empirical research in this area is the identification of crises events. Existing studies rely on the observation of exceptional events or very visible policy interventions, such as forced mergers, bank closures, or bailouts by the government. This can be misleading for a number of reasons.

First, such interventions may occur even in the absence of an acute crisis in the banking sector, e.g., when unresolved structural problems in the banking sector have been lingering for some time. Second, deciding whether a particular intervention is large enough to be called a crisis of the banking system and not just an individual institution involves a subjective judgment. Third, policy interventions typically occur when the crisis has already fully developed. Finally, recent literature on currency crises (Eichengreen, Rose, and Wyplosz, 1995, 1996ab) argues that not every crisis leads to a visible policy intervention of this kind, as central banks and regulators may be able to fend off the crisis successfully with less spectacular means. Focusing on crises that trigger policy interventions thus creates a selection bias in the empirical work.

The purpose of this paper is to propose an alternative method to identify banking crises. We follow the ideas of Eichengreen, Wyplosz and Rose (1995, 1996ab) and propose an index of money market pressure. We take extreme values of this index as signals of banking crises. We develop this indicator and discuss its empirical application. Using it to identify the dates of banking crises in a sample of 47 countries covering the period 1980 - 2001, we investigate what are the main empirical determinants of banking crises.

Download
PDF Ebook Money Market Pressure and the Determinants of Banking Crises