INTRODUCTION

Recent research has pointed the way to an exciting new direction for the theory of stock valuation. Moving beyond traditional dividend discounting models, dynamic models have emerged that provide greater structure by assuming explicit stochastic processes for the variables that in°uence stock value (see, e.g., Berk, Green, and Naik (1999), Ang and Liu (2001), and Bakshi and Ju (2002)).

Among these, a distinctive feature of the Bakshi and Chen (2001) stock valuation model (hereafter the BC model) is its focus on earnings. The key underyling valuation variables in this model are the level of earnings, the growth rate of earnings, and the interest rates used to discount earnings. Since earnings performance is a key fundamental underlying stock prices, the BC model has laid out a parsimonious framework of dynamic stock valuation.

Indeed, by modeling the stochastic process for earnings, and by adopting a stochastic pricing-kernel process that is consistent with the Vasicek (1977) term structure of interest rates, the BC model provides a closed-form solution for the stock price. Such structural modeling of the relation between stock value, earnings, and interest rates o®ers the potential for more accurate predictions about stock prices and price movements than have been achieved in previous literature.

Although the BC model makes important headway in modeling how stocks are priced, a property of the model that limits its applicability is that, strictly speaking, it cannot value stocks that have a positive probability of realizing zero or negative earnings.

In reality, stocks that will achieve positive earnings with certainty forever of course do not exist. In this sense the model necessarily provides an approximation that becomes less exact the greater the likelihood that the firm will have zero or negative earnings. Indeed, we will discuss in more depth, the BC model becomes extremely inaccurate when this likelihood is large.

To get a sense for how restrictive the non-negativity constraint is in practice, consider a sample of 6262 I/B/E/S-covered stocks during July 1976 and March 1998. On average, for any stock at any point in time, the chance of having a negative earnings per share number is 11.9%, and 41.3% of the stocks have at least one non-positive earnings outcome. This means that the BC model cannot be applied to over 40% of the stocks for at least some period of time. Of course, even the stocks which did not realize negative earnings ex ante had some probability of negative earnings.

Indeed, this constraint of the model reduces its applicability for some of the stocks that are most interesting from a valuation perspective| those that are generating relatively low earnings now, and might be in deep trouble, or on the other hand may have excellent growth opportunities. ...

Read Full: A Generalized Earnings-Based Stock Valuation Model

PDF format, 312KB, 42Pages.

Ming Donga
David Hirshleiferb

ABSTRACT
This paper provides a model for valuing stocks that takes into account the stochastic processes for earnings and interest rates. Our analysis di®ers from past research of this type in being applicable to stocks that have a positive probability of zero or negative earnings. By avoiding the singularity at the zero point, our earnings-based pricing model achieves improved pricing performance.

The out-of-sample pricing performance of Generalized Earnings Valuation Model (GEVM) and the Bakshi and Chen (2001) pricing model are compared on four stocks and two indices. The generalized model has smaller pricing errors, and greater parameter stability. Furthermore, deviations between market and model prices tend to be mean-reverting using the GEVM model, suggesting that the model may be able to identify stock market misvaluation.

Comments (0)

Name

Email

Comment

smaller | bigger

Add Comment