Earning Persistence, Earning Expectation And Stock Valuation

Based on the time series property of earnings, this paper builds the theoretical relation between earning persistence, earning expectation and stock valuation. With Ohlson (1995) model as a reference point, this paper tries to answer the following questions: 1) whether we can build a valuation model based on accounting earnings and its persistence directly, without introducing book value as an additional ingredient? 2) Will it be as economically intuitive and empirically friendly as Ohlson (1995)? 3) How it will perform when explaining the cross-sectional variation of stock price? 4) Can it explain the mean-reverting effect of P/E ratios? 5) Can we decompose earning into its components, and build a valuation model based on earnings components and their persistence?Firstly, relying on IMA(1,1), we build the relation between earning persistence,earning expectation and stock value. The model shows that current stock value is the weight average of the capitalization of current earning and lag stock value, and the weights sonly depend on earnings persistence. Earning is more value relevant when it's more persistent, while lag stock value is more value relevant when earning is less persistent. Therefore, our model and Ohlson (1995) have much in common but differs subtly as well. Based on our model, earning response coefficient can easily be derived. Empirically, we find that our model performs well in explaining the cross sectional variation of China's A share stock price.Secondly, our model can directly derive the relation between P/E and transitory earning. It can explain the cross-sectional and temporal variation of P/E, such as the mean reversion phenomenon of P/E towards an intermediate value. The P/E model predicts that transitory earning will lead the P/E to revert, and P/E ratio will negatively correlated with current earning growth, and positively correlated with subsequent earning growth, which is consistent with the empirical findings.Finally, earning is decomposed into its components and differential persistence for these components is allowed for. Again, the relation of earning components'persistence and stock value is derived. The model shows that current stock value is the linear function of current earning component and lag stock value. Earning components are more value relevant when they are more persistent, while lag stock value is more value relevant when earning components are less persistent. This model can interpret the empirical findings of the information content or value relevance studies of earning components.