An Empirical Evaluation Of The Relationship Between Errors In Analysts Forecasts Of Earnings Per Share And Stock Prices PDF Download

This study tests the Feltham and Ohlson (1996) model by modelling the Feltham and Ohlson other information variables using two different approaches. One approach is developed from the theoretical analysis of Begley and Feltham (2002). Extending their analysis that one-year and two-year-ahead analysts' forecasts of residual income infer the other information about future residual income and future investment growth, I measure the other information variables from the error terms of the regression of analysts' residual income forecasts on accounting numbers. The other approach is the direct implementation of Ohlson (2001) and Liu and Ohlson (2000). These studies demonstrate that one-year-ahead analysts' forecasts of residual income and operating assets can infer the two other information variables, and that other information can be measured as the difference between analysts' forecasts and forecasts from the time-series models. Using both the cross-sectional (in-sample) and the price prediction (out-of-sample) tests, I provide empirical evidence that the Feltham and Ohlson (1996) model is shown to have substantial empirical validity when these two approaches to estimate the other information variables are used. The signs of the estimated valuation regression coefficients are consistent with the theoretical prediction of the Feltham and Ohlson model, and the one-year-ahead price prediction of the Feltham and Ohlson model is more accurate than those of the Ohlson (1995) model. This study uses UK data from year 1999 to 2010.