In Defense of Early Warning Signals.

The 2008 financial crisis has rekindled interest in the issue of early warning signals (EWS) of financial distress. It has also triggered renewed interest in the literature on currency crises, with many countries, especially among emerging market economies, experiencing severe exchange market pressure. While several policy institutions are in the process of developing new early warning systems, there is a lot of skepticism on the ability to predict currency crises or, more generally, any type of financial crises. This skepticism stems from the alleged poor out-of-sample performance of leading models, but also from a more fundamental objection, according to which it is by definition impossible to predict crises – what can be referred to as a new “impossibility theorem”. Moreover, another criticism of early warning systems is that they may contribute to the phenomenon they are supposed to fight (the self-fulfilling prophecies view). The objective of this paper is to challenge this skeptical view. To this aim, the paper discusses the general conditions under which the “impossibility theorem” may fail and self-fulfilling prophecies can be avoided, stemming e.g. from political economy arguments. The ability of a simple currency crisis model to provide useful information on economic vulnerabilities is illustrated by testing its out-of-sample performance in a panel of emerging market economies following the collapse of Lehman Brothers.