We introduce an information-based fragility measure for GMM models that are potentially misspecified and unstable. A large fragility measure signifies a GMM model's lack of internal refutability (weak power of specification tests) and external validity (poor out-of-sample fit). The fragility of...
