Price Models and the Value Relevance of Accounting Information

Abstract

Recent research in empirical accounting has raised questions about the usefulness of price models, and how this tool can be used properly given its economic and econometric problems. This paper first identifies the essential distinctions between the price and return models, and the situations in which a price model is advantageous. An alternative method of estimating price models is then developed through minimizing the symmetrized relative pricing error (LRPE); Compared to the conventional methods of deflating by BVE, Shares, lagged price, or the price itself, the LRPE regression has several advantages: (1) The estimates are economically more meaningful and substantially more accurate. Comparisons using real data show that conventional methods waste 75% of the sample relative to LRPE; (2) the predicted price from the proposed method better captures the intrinsic value of a firm. I find that a hedge portfolio based on LRPE regression gives higher returns; (3) the relative pricing error, as a measure of the value relevance of accounting information, produces intuitive conclusions, while R2 doesn't. These results suggest that LRPE regression can contribute significant refinements to empirical research utilizing the price models.