Use of accounting information in the valuation of equity securities

Ballas, A (1992) Use of accounting information in the valuation of equity securities. Doctoral thesis, University of London: London Business School.

Abstract

Over the past two decades numerous studies have shown that capital market participants use accounting information to price equities. In many situations, such as IPOs, sales of divisions, and the valuation of closely held businesses for tax purposes, the only available guide to the value of the equity is accounting data. The aim of this dissertation is to examine which accounting measures of profitability and value of assets are, comparatively, the best predictors of the market value of the firm cross-sectionally. The dissertation starts by looking at simple rules for valuing companies (J)rice earnings, price to book and price dividend ratios). An important contribution of the thesis, is an empirical investigation of the statistical assumptions underlying ratio based valuation models which suggests that the relationship between market capitalization and accounting data is multiplicative. Various definitions of earnings and the book value of assets are used but the price dividends ratio outperforms all of them. The best definitions of earnings and book value of assets as well as dividends are used to construct a combined forecast model which significantly outperforms the univariate rules. Subsequently, alternative growth and risk measures and the time series properties of earnings are utilized to examine empirically the cross sectional distribution of the valuation ratios. Tests for industry effects are also performed and the conclusion is that they are of marginal economic significance. In the second part of the dissertation, more complicated valuation models are used to examine whether all components of earnings or classes of assets in the balance sheet are 'capitalized' at the same rate. A modified version of the Litzenberger Rao model is used to correct for specification error. The conclusion is that the efficiency of the models improves using disaggregated earnings data (but not disaggregated book value data). This is consistent with the view in the literature that the classification of earnings in different classes provides information in excess of that in the aggregate earnings figure. Finally, a theoretical model of the measurement error in accounting data is constructed to examine the biases caused by stylized depreciation schedules in estimates of value based on the book value of assets and capitalized earnings. This has important implications for the construction of optimal forecasts namely, that it allows us to examine the conditions under which the market to book (PB) ratio will outperform the price earnings (PE) ratio and the sign of the error using either of the two rules.