Common Risk Factors vs. Mispricing Factor of Tokyo Stock Exchange Firms: Inquries into the Fundamental Price Derived from Analysts' Earnings Forecasts

Abstract

In this paper we try to find common factors that can explain the stock returns of Tokyo stock exchange firms with multivariate asset-pricing framework. Specifically, we explore the nature of risk contained in the size and the HML factor variables and their information content. For this purpose we utilize Edwards-Bell-Ohlson model to derive accounting based fundamental price and divide this value by the actual market price, whereas we call "value- to-price ratio" as Frankel and Lee did. By comparing the returns of the portfolios ranked by this candidate variable, we find that this value-to-price ratio variable contains new information content that is not yet reflected in the conventional book-to-price ratio, while its explanatory power is slightly inferior to the book-to-price ratio. The portfolio strategy based on the value-to-price ratios can earn abnormal cumulative returns for some subset of the sample portfolios and for a larger fraction of portfolios during a later sub-period of the sampling period. Some of the return differences are significant based on the t-test whose method initially proposed by Daniel and Titman (1997). To test for whether this variable can be a significant explanatory variable in a multivariate asset pricing model, we construct return difference portfolios ranked by the value-to-price ratio, whose method is identical to Fama and French (1993) method in their constructing the HML return difference portfolios. We call this factor variable as "UMO" (underestimating-minus-overestimating) factor variable. We conduct Fama and MacBeth test with this UMO factor as one of the candidate factor variables. Under the null hypothesis of the existence of the risk premium on each possible factor, we find that the risk premium of for UMO factor is significant. However, the degree of the significance is not universal unlike the case for the HML factor. Accordingly, we test for the characteristic hypothesis raised by Daniel and Titman (1997), and we conclude that this factor is rather a characteristic rather than a risk factor and cannot be a good substitute for the book-to-price ratio variable.