Do aggregate earnings (for example, for the S&P 500) predict stock market behavior? In their 2008 paper entitled “Aggregate Earnings, Firm-Level Earnings and Expected Stock Returns”, Turan Bali, Ozgur Demirtas and Hassan Tehranian analyze the predictability of stock returns using market, industry and firm-level earnings. Using monthly stock return data and quarterly fundamentals data for a broad range of U.S. stocks focused mostly on the period from mid-1972 through 2002, they conclude that:
- The level of aggregate earnings does not forecast overall stock market behavior robustly across different stock portfolios and sample periods. The variability in aggregate earnings is noise.
- In contrast to the aggregate-level findings, earnings and earnings yield relate positively to firm-level stock returns and finely segmented (48 industries) industry portfolio returns. However, earnings yield is not predictive of returns for grossly segmented (17 industries) industry portfolios. Only the
unsystematic (idiosyncratic) portion of firm-level earnings is informative about future stock returns. Aggregating earnings across a broad set of stocks diversifies away this informativeness. - At the firm level, both the mean-reversion of stock prices and the correlation of earnings with future stock returns contribute significantly to the forecasting power of earnings yield.
In summary, evidence indicates that aggregate earnings alone are not a useful predictor of overall stock market behavior, but firm-level (highly segmented industry) earnings are useful for predicting returns of individual stocks (highly segmented industries).
These findings are unfriendly to models that attempt to use aggregate earnings to predict overall stock market behavior.
