Do we reject rational expectations models too often?: Interpreting evidence using Nagar expansions

In a recent paper Mankiw and Shapiro presented Monte Carlo evidence assessing the over-rejection of the orthogonality condition which characterizes rational expectations models. The purpose of this note is two-fold — first, to construct a small sample approximation of the expected value of the t-statistic and second, to get a feeling for why the critical values obtained by MS are so disparate.Publicad