A Simple Structural Estimator of Disclosure Costs

We develop structural estimators for disclosure costs (or benefits), and empirical bounds for the probability of information endowment in a generalized disclosure model nesting Verrecchia (1983) and Dye (1985). The baseline estimator of disclosure costs is a closed-form function of (i) the minimum disclosure surprise, (ii) the average disclosure surprise, and (iii) the frequency of disclosure. Furthermore, it can be computed without solving for either the entire equilibrium of the game or knowledge of distributions. We derive the asymptotic properties of the estimator and illustrate, in simulations, its properties in small samples. A new empirical test is derived which can test for samples incompatible with the theory. The probability of information endowment is not point-identified but can be bounded from above in the case of disclosure benefits. Additionally, adapting an argument from conditional choice probabilities, the estimation extends to a multi-period setting with time-varying frictions as a function of observable state variables. As an application, we conduct the estimation using quarterly management earnings forecasts, characterize the magnitude of the disclosure costs, describe covariates associated with disclosure costs and perform various tests to detect the fraction of firms’ behavior inconsistent with the theory or whose probability of information endowment is significantly less than one. The framework offers a simple theory-based approach to estimating voluntary disclosure models.