We propose a new methodology to test Fama’s [J. Finance 46 (1991) 1575] contention that the present value model (PVM) should be augmented by time-varying expected inflation to more adequately account for actual stock price behavior. Unlike other methods, our testing approach can distinguish between the excess-price movement hypothesis of Shiller [Am. Econ. Rev. 71 (1981) 421] and the dividend-smoothing hypothesis of Marsh and Merton [Am. Econ. Rev. 76 (1986) 483]. We decompose the levels (as opposed to the variances) of stock prices into their fundamental and non-fundamental elements in the context of a multivariate PVM co-integrating framework and utilize the Gonzalo and Granger [J. Bus. Econ. Stat. 13 (1995) 27] procedure to formally test for the statistical significance of the non-fundamental component. Our results from monthly data for the post-WWII period do not support the inflation-augmented PVM since the non-fundamental component continues to achieve significance. This finding persists under alternative model specifications and data frequencies. The apparent failure of the traditional, rational-expectation, PVM model to adequately account for observed market behavior provides another piece of evidence supportive of Shiller’s [Am. Econ. Rev. 71 (1981) 591] belief in some form of market “irrationality”.