Blog - Investing Notes

November 6, 2006 - Stock Valuation Indicator Flyoff

Deterioration over the past decade in the forecasting power of traditional indicators (such as price-dividend and price-earnings ratios) have stimulated searches for better ones, with recent emphasis on macroeconomic variables. Which financial and economic variables best predict stock returns over the short, intermediate and long terms? Is "best" good enough for market timing? In her October 2006 paper entitled "How Well Do Financial and Macroeconomic Variables Predict Stock Returns: Time-series and Cross-sectional Evidence", Anne-Sofie Reng Rasmussen evaluates the relative performance of a wide range of variables in forecasting excess stock returns (above the one-month T-bill rate) over horizons from one quarter to eight years. Using annual data for periods as long as 1930-2005 and quarterly data for periods as long as 1926-2005, she concludes that:

• Evidence of stock return predictability is not overwhelming. Only very few indicators have statistical merit. Ignoring the last 10 years of data improves stock return predictability.
• In general, predictability increases with investment horizon, but with some differences across forecasting variables. At very short horizons, few variables explain much of the variation in excess stock returns.
• Over the short and intermediate terms, the consumption-aggregate wealth ratio is the best overall indicator of stock returns, although its predictive power varies with sample period.
• Over the long term, price-normalized variables (price-dividend, price-earnings, price-output and price-consumption ratios) outperform as stock return indicators.
• Multivariate regressions show that very few of the newer macroeconomic variables add any forecasting power beyond that of the price-dividend ratio.

The paper provides an extensive overview of recent efforts to forecast excess stock returns. The discussion, however, is fairly dense for the typical investor.

In summary, over the last 80 years, a few price-normalized variables (price-dividend, price-earnings, price-output and price-consumption ratios) and the approximate consumption-aggregate wealth ratio have been the best stock market indicators.

Skepticism is in order regarding the applicability of very old data to forecasting the current stock market environment. Interrelated changes in the financial services industry (such as the emergence first of mutual funds and then exchange traded funds), the regulatory environment and information technology arguably affect investor risks and behavior. The computing and networking advances of the past decade have enabled dramatic improvements in the timeliness, depth and quality of financial data available to investors. A review of the detailed commentaries of a range of stock market experts (see the links at Guru Grades) suggests that the degree to which they value very old data is an important source of variation in their opinions.

For related research, see Blog Synthesis: Valuation Based on Fundamentals and Blog Synthesis: The Economy and the Stock Market.