Application to Investment Decisions
As predicted by Bernard (1995), a large segment of the recent empirical literature
on valuation reflects a shift in research emphasis—away from explaining contemporaneous
stock prices or returns, and toward predicting future stock returns or future
changes in fundamental values. This shift in research objective is consistent with
Penman's (1992) call for more research on fundamental analysis. I believe it represents
a growing opportunity for future research.
What have we learned so far from this research? One emerging theme is that a
better value estimate can lead to better forecasts of future stock returns. Frankel and
Lee (1998) used a simple three-period version of the RIM based on mean I/B/E/S analyst
forecasts to estimate an intrinsic value (V) measure for each firm. They show that
the resulting value-to-price (V/P) ratio is a better predictor of cross-sectional returns
than measures such as book-to-market or firm size. Specifically, high (low) V/P firms
earn higher (lower) future long-term risk-adjusted returns over the next three to five
years. Because of the importance of analyst forecasts to the valuation model, Frankel
and Lee (1998) also examined the predictability of errors in these forecasts. They find
that errors in the mean I/B/E/S analyst forecasts are indeed predictable in the crosssection.
A two-stage trading strategy based on both a V/P filter and an analyst error
prediction filter results in even higher predictive power for future returns.