New research from Robeco identifies and corrects for biases in analyst earnings revisions, says Senior Quantitative Equities Researcher, Joop Huij.
Imagine you are interviewing two candidates for a job opening. Their backgrounds are broadly similar and they are equally qualified for the position. You decide to factor in the personal recommendations of their previous employers. But you find that they both have received glowing reviews.
You notice, however, that the recommendation for Candidate A, who worked for a family firm, comes from a relative, while the recommendation for Candidate B comes from a highly regarded professor at a local university who is known to be very demanding. Instinctively, you regard the recommendation from the professor as more credible and less apt to be biased in favor of the candidate.
In a nutshell, this instinctive adjustment to eliminate bias is what Joop Huij and other researchers in Robeco’s Quantitative Strategies research department say they have quantified and corrected for in quantitative investment strategies using analyst earnings revisions as a factor in stock selection.
Earnings revisions are a common component of both fundamental and quantitative equity strategies. “An earnings-revisions investment strategy is based on empirical evidence that stocks with earnings forecasts that were recently upgraded by analysts tend to outperform, while stocks with negative earnings revisions tend to underperform,” explains Huij.
Investment banking firms typically have research departments that employ analysts to follow companies’ earnings prospects and to make recommendations to clients whether to buy, sell or hold the stocks of these companies. Analysts will typically be responsible for a limited number of stocks, enabling them to follow developments closely and to produce their own estimates of future earnings per share (EPS). Their recommendations can have a significant effect on share prices, with a forecast of higher EPS capable of pushing share prices up.
The problem, however, is that there is increasing evidence that professional analysts are prone to biases. “In particular,” says Huij, “analysts have a tendency to favor expensive stocks with large market caps and low book-to-market ratios. These recommended stocks are often what we define as glamour stocks.”
Glamour stocks are typically trendy stocks that have become widely popular with investors, often owing to favorable publicity in the media. Think IT stocks during the internet bubble. They are typically more expensive than less popular stocks, owing primarily to the high demand for them. “Glamour stocks are also more likely to receive upward revisions from analysts,” notes Huij, “while stocks with the opposite characteristics, typically value stocks, are more likely to be revised downward.”
By favoring stocks with large market caps and low book-to-market ratios, analysts are actually recommending classes of stocks that have been shown to underperform the market. This is in line with research by the well-known financial market economists, Eugene Fama and Kenneth French, who famously showed that two classes of stocks, small caps and stocks with a high book-to-market ratio (value stocks) tend to perform better than the market as a whole.
The book-to-market ratio is an important valuation measure because it compares the value of a company's assets as if it were being liquidated (book value) with the current market value. A low book-to-market ratio signals that a stock is likely overvalued while a higher book-to-market ratio indicates undervaluation.
According to Huij, there are various reasons why analysts might favor glamour stocks. "It could simply be due to human nature and a tendency to favor what other investors already prefer, while ignoring the awkward fact that expensive stocks do not outperform less expensive stocks," he says.
Huij also points to a possible inclination, conscious or not, to favor the stocks of companies that are (or would be) attractive investment banking clients, such as companies with high-growth prospects due to pending mergers. For example, it appears that analysts strategically adjust their earnings forecasts to avoid earnings disappointments for these firms.
The widespread use of earnings revisions in investment strategies despite acknowledged flaws was an investment conundrum. "For the first time, we have quantified the extent of analyst bias toward overvalued stocks, and then developed a more sophisticated earnings-revisions strategy to neutralize this glamour bias," says Huij.
The research consisted of analyzing the returns of a traditional earnings-revisions strategy from 1986 through 2009 using a three-factor model developed by Fama and French. "Using this model, it was possible to attribute the return from a traditional earnings-revisions strategy to three possible factors: the market's return, the size (market-cap) effect and the book-to-market ratio effect."
What were the results? “We found that roughly one-third of the variability in the returns of a typical earnings revisions strategy can be solely attributed to the strategy’s exposure to overvalued glamour stocks," says Huij. While the bias had previously been identified, "the size of the bias was a surprise."
While the exact adjustments that were made to improve the earnings-revision strategy are proprietary, what Huij can reveal is that they are based on finding out that the degree of the bias varies over time and is related to the business cycle. Intuitively, this makes sense. As analysts could be expected to be more bullish toward growth stocks than to value stocks in a bull market, as growth stocks typically perform well in rising markets.
A back-test over the same 23-year time period as the initial analysis, using both the traditional and enhanced earnings-revisions strategies, showed that the enhanced earnings-revision strategy improved the risk/return ratio from 0.6 to 1.1. This means that not only are returns better, there is also less risk.