The explanatory and predictive power of different specifications of the Ohlson (1995) valuation models

Abstract

Accounting-based valuation studies of US firms tend to support Ohlson's proposition that residual income and book value numbers have information content in explaining observed market values. But European evidence also suggests that the conservative/liberal orientation of accounting tradition can produce significant national differences in associations between accounting performance measures and stock prices - in earnings behaviour, coefficient values and parameter sensitivity. We address these issues from an equity valuation perspective using Swedish data to assess the additional information content of Ohlson's information dynamics and analysts' forecasts in relation to market valuations in a more conservative accounting environment than the US. The study compares the explanatory and predictive power of Ohlson's (1995) residual income model (RIV) with a linear information dynamics version (LIM) that specifies both residual income and non-accounting information as autoregressive processes. Both versions are applied with, and without, future performance expectations from non-accounting sources (analysts' forecasts). As with US evidence, we find that the inclusion of analysts' forecasts improves both (i) cross-sectional correlations with current prices for both RIV and LIM models and (ii) the predictive power of RIV models in relation to future annual cross-sectional stock returns. The contribution of linear information dynamics is significant but varies across approaches. We also find significant differences between Swedish and US firms in earnings behaviour and associations between accounting numbers and market equity prices.