Accounting for Crises

by Venky Nagar & Gwen Yu

Overview — A key endeavor of modern economic theory is to understand the causes of panics. This paper shows empirically that currency investors are more likely to get spooked unnecessarily when they have too much information. This finding accords well with global games models, which argue that self-fulfilling panics—i.e., panics unrelated to fundamentals—are more likely to occur when the quality of public information available to investors is very high. Research was conducted by Venky Nagar (University of Michigan) and Gwen Yu (Harvard). Key concepts include:

  • Because crises in high accounting precision countries are more likely to have no fundamental cause, accounting fundamentals are more likely to predict crises when accounting information has low precision.

Author Abstract

We provide one of the first tests of recent macro global-game crisis models that show that the precision of public signals can coordinate crises (e.g. Angeletos and Werning 2006; Morris and Shin 2002, 2003). In these models, self-fulfilling crises (independent of fundamentals) can occur only when publicly disclosed fundamental signals have high precision; fundamentals are thus the sole driver of crises in low-precision settings. We affirm this proposition on 39 currency crises by exploiting a key publicly disclosed fundamental driving financial markets, namely accounting data. We find that fundamental accounting signals are stronger in-sample predictors of crises in low-precision countries.

Paper Information