Comment on “Costs of banking system instability: Some empirical evidence”

Abstract

We have become accustomed to thinking that it is poor countries that suffer most from banking crises. This view is challenged by Glenn Hoggarth, Ricardo Reis and Victoria Saporta (HRS), who focus on the output loss caused by banking crises. There are several new things in this paper. First, new estimates of output losses during 42 banking crises around the world. Second, analysis of these estimates suggesting inter alia that rich countries may actually do worse out of banking crises on average than poor countries, at least when ‘‘doing worse’’ is measured by the new measure of output losses. Obviously this begs the tough question how much of output fluctuations are explainable by banking crises, and (thirdly) the authors present a brave attempt to quantify this too. Let me take these three points one by one. Pointing out the weaknesses of one standard approach to measuring output losses (the so-called IMF 1998 approach, which sums the annual gap between actual and trend GDP growth rates for the period until the growth rate returns to its previous average), the authors argue that it is desirable not to take ‘‘bygones as bygones’’ but to cumulate the total shortfall of the level of output below the previous trend path of output itself, not just of its rate of growth. The new measure (GAP2) is only weakly correlated with the old (the correlation coefficient being just 0.33), and insignificantly so with fiscal costs, so we are definitely dealing with a new animal here (see Fig. 1). Still, we need to treat it with caution when we see some of the numbers the method turns up. For developing countries the measured range is very wide indeed: from 47% in Ghana (output growth accelerated there while they were addressing the banking problems) to þ112% of GDP during the 1980s in the Philippines. It is instructive Journal of Banking & Finance 26 (2002) 857–860