Robust Tobin's Q Theory

Abstract

Models of the q theory typically assume that investments are determined by a specific approximating structured q model, hence ruling out perturbations due to a set of statistically nearby unstructured alternatives. This paper formulates a generalized framework, where concern to unstructured q models is admissible. By adopting relative entropy restriction for a set of unstructured alternative models, the model delivers generalized approximation to the ”truth” of the q theory, thereby resolving several puzzles in structured q models. By exploiting polynomial specifications and novel measures of perturbations, I empirically demonstrate the critical importance of the approximation error driven by unstructured models.