Abstract
The challenge of using accounting numbers for valuation purposes has tempted accounting researchers and professional financial analysts over the years. The choice and measurement of suitable accounting numbers, as well as the specification of the linkage between accounting numbers and stock market prices, have constituted important issues. A wide array of valuation models have been suggested over time, including simple models based only on measures of current earnings, as well as elaborate simulation models based on a multitude of accounting numbers. From a methodological point of view, the proposed valuation models can be divided into two main groups:
• Valuation models that are directly based on the statistical association between accounting numbers and stock market prices.
• Valuation models that are deduced from the theory of capital value.
Models of the first kind – statistical valuation models – often hinge on some simplified assumption about the relationship between accounting numbers and stock market prices, e.g. the simple mathematical relationship of a P/E-ratio valuation model. Hence, such
models are often viewed as particularly easy to use. The price of this simplicity is typically, however, deficiencies in the modelling logic. Furthermore, statistical valuation models can only be estimated when there is some empirical market data to be observed. A prerequisite of such models is then that observed stock market prices are ‘correct’, i.e. that the valuation analysis which investors actually engage in lead to prices that fully reflect all available information. This assumption corresponds to the well-known hypothesis of ‘semi-strong market efficiency ’. Whether this hypothesis is empirically valid is not clear, however.
Models of the second kind – deduced valuation models – have been subject to an increased interest in the academic research since the beginning of the 1990’s. These models do not depend on any assumption about stock market prices being efficient in the semi-strong
sense. In general, they constitute a good foundation for the specification of relationships between accounting numbers and stock market prices. However, statistical problems – in particular concerning the prediction of valuation relevant accounting numbers – cannot be avoided in these models. Such problems can often be analytically isolated though, whereby guidelines for the estimation of statistical forecasting models can be provided. In the following, two deduced valuation models will be specified and discussed – a ‘residual income’ valuation model (section 2) and a ‘value added’ valuation model (section 3). Both models are based on a modelling logic where capital values are determined as the sum of an accounting book measure of capital, the present value of expected future abnormal profitability, and the present value of expected goodwill/badwill at some horizon point in time. Strengths and weaknesses of the two valuation models will be discussed in section 4.