On the relevance of earnings components in valuation and forecasting

Outline

• Abstract
• 1 Introduction
• 2 Assumptions and Definitions of Informational Relevance
• 3 Informational (ir)relevance of Earnings Components in Valuation and Forecasting
• 3.1 when Earnings Components Are Aggregated
• 3.2 Dirty Surplus Earnings Components
• 4 Discussion and Conclusion
• Acknowledgments
• Appendix
• References

Discussion and conclusion

The analysis of informational relevance of earnings components in valuation and forecasting in Ohlson (1999) and Pope and Wang (2005) can be extended to non-transitory earnings and a nonlinear framework to incorporate gains/losses. Suppose that aggregate (core) abnormal earnings at time t ? 1 is associated with aggregate (core) abnormal earnings at time t via a nonlinear function and in a Markovian system. Then the RIVM will lead to a nonlinear relation between value of equity and aggregate (core) abnormal earnings at time t. The nonlinearity of abnormal earnings and value of equity may be characterized as option valuation components as documented in the prior literature (Burgstahler and Dichev 1997; Yee 2000, 2005; Zhang 2000; Biddle et al. 2001). Although a forecasting irrelevant earnings component is also valuation irrelevant in corresponding type, i.e., FI-j implies VI-j (j = 1, 2, 3), Propositions 1–3 show that the converse is not generally true. I investigate this issue by firstly considering conditions for dividend displacement, and secondly the predictability of an earnings component for itself.

Studies concerned with testing a null hypothesis of irrelevance of earnings components may be informed by the analysis. I show that care is required in defining valuation relevance so as to ensure that coefficient values predicted under the null hypothesis reflect the reduced form relationships. Valuation irrelevance of an earnings component does not imply that the component should necessarily have a zero valuation weight in an unrestricted regression of market value on financial statement variables. One needs to define information irrelevance and the fundamental valuation relevant variables maintained to be sufficient for valuation in order to test the incremental valuation relevance of an earnings component by focusing on the parameter restrictions associated with an irrelevance definition. The analysis also shows that a one-to-one mapping between valuation relevance and forecasting relevance should not be expected. One cannot infer valuation relevance based on evidence of forecasting relevance. Nor can forecasting irrelevance be inferred from evidence of valuation irrelevance.

Finally, my model also appears to provide a basis for understanding some of the features of accounting practice. Although the analysis is presented in terms of two earnings components only, the intuition provides a rationale for the emergence of detailed line item disclosures in GAAP. At least at an anecdotal level, different line items subject to specific disclosure provisions under most GAAP regimes, such as depreciation, financing charges, and research and development expenses, ‘other comprehensive income items’, the gains/losses of financial derivatives qualified and not qualified for hedging, can be expected to have distinct information dynamics properties. GAAP/IFRS developments in relation to line item disclosures are usually not motivated by explicit consideration of the information dynamics. However, it is probable that at least some such disclosure requirements arise from an implicit belief that such items will be valued differently because they have different dynamic properties.