The art of averting financial crises
International currency and banking crises have become a fixture of modern life. The most significant recent turmoil, the global credit crunch of 2008 shared many features with global credit collapses of the past. (Reinhart and Rogoff. This time is different: Eight centuries of financial folly. 2009)
Numerous studies on the early warnings of financial distress suggest that crises almost never happen because of a single shock. Typically, several warning signals flair up before the crisis comes to a head.
Given the potential ability to heed the warning signals, what is it that makes crises so prevalent? Several possibilities can be considered. The obvious culprits are various policy implementation lags, which can make timely response impossible or ineffective.
However, it may also be the case that the signals sent by the indicators prior to a realized crisis are qualitatively different (in duration, intensity, timing) than the signals preceding a crisis that has been resolved with or without policy intervention.
The focus of this paper is on identifying such distinguishing features.
As a benchmark, we analyze the instances of “false alarms” as defined by the Early Warnings system (Kaminsky, Graciela. Currency and banking crises-the early warnings of distress. 1999), and compare their signals with the signals of a similar variety leading up to financial crisis.
Since the types of crises are different, it is important to compare the signals, which could potentially lead to the same type of crisis (e.g. currency, banking and the different varieties of thereof). However, since “false alarms” by definition do not result in an actual crisis, we use the Propensity-Score Matching methodology to match the signals from the “hits” with the “false alarms” that are likely to result in the same type of crisis.
We then proceed to systematize the differences between the indicators for “hits” and “false alarms” along several dimensions, such as timing of the signal, initial strength of the signal, types of crises averted, nature of external shocks, policy responses (if any), the timing of intervention.
In addition to “false alarms”, we will also seek to include in our analysis the “moderate turbulences”, the periods when at least one signal was strong, but the event overall failed to qualify as an “alarm” – possibly due to an early and effective policy intervention.
We expect to find evidence that the signals emitted prior to the “non-crises” that have not been realized are qualitatively different than the ones sent prior to the realized crises – namely, the warnings are stronger, earlier, more manageable from the institutional point of view than in the realized crises cases.
Research into the nature of the crisis that “never happened” will provide us with a better understanding on the policy options and the challenges faced by countries in the early stages of financial development. The countries that are intent on avoiding large-scale financial fiascos, in the same time as they are trying to build up the foundations for accelerated development and investment-led growth.