Accounting-based valuation with linear information dynamics.

Accounting-based valuation is the process of estimating firm value based on reported accounting numbers. This study documents the necessity and efficacy of using linear models of the stochastic relation between current and future accounting numbers for use in valuation. Linear information models (LIMs) are necessary to provide tractable pricing relations that preserve the assumptions of no arbitrage and clean surplus that underlie the valuation theory. Recent papers that utilize accounting-based valuation theory either implicitly violate one of these assumptions or assume that the specific LIM of Ohlson (1995) or Feltham and Ohlson (1995) is representative of the true stochastic process. However. neither of these models have been tested empirically. I find that the common practice of modeling expectations of future accounting numbers in two parts, the short run and the long run, leads naturally to a violation of the no arbitrage assumption. I also find that while the LIMs of Ohlson (1995) and Feltham and Ohlson (1995) provide better estimates of firm value than a random walk model of relevant accounting numbers, the models imply inappropriate weightings on the relevant information variables for valuation. Furthermore, the conservatism parameter of Feltham and Ohlson (1995) is significantly negative, contrary to the models prediction, for even the most conservative firms. To explain these failures, I estimate a LIM from a more carefully modeled accounting system that provides two parameters of conservatism. However, this model also fails to capture the true stochastic relationship among accounting variables. In fact, the more complex models provide noisier estimates of firm value than the more parsimonious models. This suggests time series LIMs are either too difficult to estimate, due to data constraints, or a more accurate description of the information dynamics is needed for firm valuation.PhDAccountingFinanceSocial SciencesUniversity of Michigan, Horace H. Rackham School of Graduate Studieshttp://deepblue.lib.umich.edu/bitstream/2027.42/130552/2/9732151.pd