An empirical assessment of the Feltham-Ohlson models considering the sign of abnormal earnings

Abstract

This paper provides an empirical assessment of the Feltham-Ohlson models, distinguishing between firms with positive and negative abnormal earnings. Abnormal earnings persistence and conservatism parameters differ for these two groups; this implies different earnings prediction models and valuation functions for both profit-making and loss-making firms. The analysis refers to the period 1991-1999 and uses a sample of Spanish firms quoted on the Madrid S.E. The results suggest that our contextual approach is more useful than the non-contextual one to predict future abnormal earnings and explain current prices. Although the Ohlson (1995) model is accurate in forecasting future abnormal earnings and stock prices, the results improve when firms with negative abnormal earnings are valued using a temporary model and firms with positive abnormal earnings using a more permanent one. The Feltham and Ohlson (1995) model generates the lowest forecast errors in the prediction of positive abnormal earnings, but it produces the least accurate results in forecasting prices.