Is There Alpha in Analyst Forecasts?
A Literature Review of Recent Research
Hi there! It’s time for yet another longer-form article. This one explores potential sources of alpha using analyst forecasts. I hope you find it interesting!
Introduction
Financial analysts play a crucial role in capital markets by providing insights that guide investment decisions. Their forecasts, particularly those related to earnings and stock prices, are widely disseminated and scrutinized by investors. An extensive body of research has explored the information content of analyst forecasts and their ability to predict future earnings and stock returns. This literature review synthesizes key findings on the relationship between analyst forecasts and return predictability, focusing on research that investors can leverage to formulate investment strategies. Overall, the research highlights several intriguing potential sources of alpha.
Table of Contents
Background
Analyst Forecasts and Investment Strategies
Price Targets and Recommendations
Correcting for Analyst Biases
International Alpha
Alpha From Slow Analysts
Machine Learning and Large Language Models
Bottom-Up Forecasts and Aggregate Predictability
Investor Takeaways
References
Background
The quest to determine whether professional forecasters and analysts can consistently outperform the market has long been a focus of financial research. In the early 20th century, skepticism prevailed regarding the value of professional forecasts. Cowles (1933) found that forecasters' recommendations generally underperformed the market, with results often indistinguishable from random chance. This early finding cast doubt on the ability of analysts to provide meaningful predictive insights.
However, the landscape began to shift in the latter half of the 20th century. Research focus pivoted towards examining the information content and accuracy of analyst earnings forecasts. Seminal work by Brown et al. (1987) documented that analyst earnings forecasts outperformed time-series models in predicting future earnings. This finding spurred further investigation into the value of analyst forecasts for predicting stock returns.
Womack (1996) provided early evidence that changes in recommendations were associated with significant post-announcement stock price drifts. Specifically, sell recommendations were found to be associated with a negative price drift, lasting up to six months. These results highlighted the potential for analyst forecasts to predict future returns, setting the stage for a rich body of literature examining various aspects of analyst forecasts and their relationship to stock performance.
Analyst Forecasts and Investment Strategies
Price Targets and Recommendations
The information content of analyst price targets and recommendations has been extensively researched. Barber et al. (2001) find that buying (shorting) stocks with the most (least) favorable recommendations leads to significant abnormal returns but requires very frequent rebalancing, yielding returns close to zero after transaction costs.
Brav and Lehavy (2003) consider sorting stocks based on the change, rather than the level, in price targets. By buying (shorting) stocks with the strongest (weakest) price target revisions, they find a significantly stronger return spread compared to sorting stocks on the level of price targets. The generated returns are also found to be more persistent, requiring less frequent rebalancing. Jegadeesh et al. (2004) reach a similar conclusion, noting that sorting stocks based on the level of consensus recommendations is a poor predictor of returns, whereas changes in recommendations do predict returns.
Boni and Womack (2006) explore a similar strategy by sorting stocks on changes in recommendations within industries. They find that a sector-neutral strategy yields a sizeable increase in risk adjusted returns, suggesting that analyst recommendations are most valuable as relative bets within sectors.
Continuing the sector-neutral theme, Da and Schaumberg (2011) propose a long-short strategy within sectors based on the price target implied expected return, which is the return implied by the analyst's 12-month price target relative to the current market price. Going long (short) stocks with the highest (lowest) expected returns yields a substantial alpha, even after accounting for transaction costs and focusing on S&P 500 stocks. The results suggest that relative price targets are more predictive than absolute price targets. Da et al. (2016) and Hao and Skinner (2023) construct similar strategies, finding stronger results for within-industry sorting.