Ebook An Empirical Assessment of the Valuation Accuracy of The Abnormal Earnings Growth Valuation Model
During the last decade, the Edwards-Bell-Ohlson residual income valuation model (Ohlson, 1995; hereafter, the RIV model) has received considerable attention from accounting researchers. The main reason for the RIV model’s widespread acceptance is that it links accounting information to equity value in a more systematic way. According to the RIV model, equity value consists of the reported book value of equity and an infinite sum of the discounted future earnings adjusted by the required returns from the usage of net assets.
However, a recent paper by Ohlson and Juettner-Nauroth (2005) provides an alternative model to mitigate the RIV model’s potential problems, such as deviations from clean surplus relation under the current accounting rules. According to their abnormal earnings growth valuation model (hereafter, the OJ model), equity value consists of capitalized next-period earnings as the first value component and the infinite sum of present values of capitalized expected changes in earnings adjusted for dividends (i.e., abnormal earnings) as the second value component.
Although both the RIV and OJ model are derived from the common dividend discount model and yield the same valuation with infinite horizons, implementations of those models in prior literature differ in their assumptions on the forecast horizon and future earnings growth after the forecast horizon. For example, while implementations of the OJ model use an economy-wide assumption for the earnings growth after the two-year-horizon, those of the RIV model incorporate industry-specific assumptions for the earnings growth after the five-year-horizon. These differences in empirical implementations may cause the valuation accuracy of the resulting equity value estimates to vary. Thus, the relative valuation accuracy of these valuation models is an open empirical question.
This paper examines the valuation accuracy of the OJ model relative to that of the RIV model. Although this is an important research question, existing empirical studies have compared the OJ model with the RIV model only in terms of inferring more reasonable cost of equity. To the best of our knowledge, this is the first study to examine the valuation accuracy of the OJ model relative to that of the RIV model. Given the plethora of possible empirical implementations, we focus on the representative empirical implementation of both models employed in prior research.
Our empirical results indicate that the OJ model generally under performs the RIV model in terms of valuation accuracy, which is measured as the discrepancy between equity value estimates from valuation models and stock prices in the market. Although increasing the forecast horizon for the OJ model from two to five years significantly improves the valuation accuracy of the OJ model, the valuation accuracy of the OJ model using the five-year forecast horizon (hereafter, the five-period OJ model) is still lower than that of the RIV model.
At the core of equity valuation is an analysis of profitability (Nissim and Penman, 2001). Accordingly, we investigate the reason for the low valuation accuracy of the OJ model by analyzing the evolution over time of future expected return on equity (hereafter, ROE) implied by each model. Given that future realized earnings can be an unbiased estimate of market expectations of future earnings, we evaluate the trend of future expected ROE implied by each model in terms of the fit with the trend of future realized ROE. For all the implementations of the models evaluated in this paper, we use the same analysts’ earnings forecasts to proxy for market expectations of future earnings until the forecast horizon, and assume the same cost of equity and the same future dividend payout ratio. Thus, we can focus only on the different assumptions across the models on the forecast horizon and future earnings growth beyond the forecast horizon as the main reason for causing the difference in valuation accuracy.
Posted in :
