If Wall Street's stock tips were reliable, portfolio selection would be easy. Investors could simply load up on Apple (AAPL: 427.75, -1.36, -0.32%) because analysts give it more positive ratings like "buy" and "outperform" than any other stock, according to Thomson Reuters (TRI: 28.85, 0.39, 1.37%) data.
Alas, the historical evidence shows stocks with lots of "buys" don't do better than the broad market, on average.
Perhaps that is because so many companies are showered with love. Among those in the Standard & Poor's 500-stock index, there are 10 times as many "buys" as "sells."
But new research suggests a way for investors to tell which "buys" are worth heeding and which ones aren't.
Professional stock-pickers have had an image problem at least since a 1933 study by economist Alfred Cowles confirmed what the market crash of 1929 had amply demonstrated: Stock forecasters can't forecast with any accuracy.
A landmark paper published 16 years ago in the Journal of Finance offered some redemption for analyst recommendations. It divided returns into two components: an initial pop when a new recommendation is announced, and a gradual drift in the months that follow. The distinction matters because ordinary slow-poke investors can take advantage of drifts but not pops.
Two key findings: First, analyst recommendations are like dairy products in that it is best to use them quickly or not at all. Shares tend to drift in the direction of recommendation changes, but for weeks or months, not years.
Second, "sells" tend to be far more prescient than "buys." According to study author Kent Womack, a former Goldman Sachs executive who now teaches finance at the University of Toronto, analysts face little resistance to their "buy" recommendations but risk angering companies and investors with their "sells," so they tend to issue sell calls much more judiciously.
That takes some of the shine off of Apple. It has received no fresh "buys" within the past four weeks among firms polled by Thomson Reuters.
Other members of the S&P 500 index have received multiple "buy" recommendations of late, including upgrades and coverage initiations. Among them are web retailing giant Amazon.com (AMZN: 194.45, 5.01, 2.64%) and Devon Energy (DVN: 64.15, -0.25, -0.39%), an oil-and-gas producer.
There are fresh "sell" ratings, too, received by companies like Pall (PLL: 60.22, 0.47, 0.79%), a Long Island, N.Y., maker of industrial products whose shares have gained in recent weeks, and clothing chain Gap (GPS: 19.37, 0.73, 3.92%), whose stock has sagged. Be warned, however, that betting against stocks -- using, say, options contracts or "short selling" -- carries considerable risk for ordinary long-term investors.
What's needed is a way to find better "buy" ratings. Mr. Womack presents some new thoughts on that in a working paper with Ambrus Kecskes at Virginia Tech and Roni Michaely at Cornell University.
To form their recommendations, analysts often begin with something called discounted-cash-flow analysis, which uses forecasts of revenues, margins and many other factors to determine a fair share price for investors to pay today. Some factors are difficult to measure (like riskiness), others are impossible to know (like distant growth rates) and subtle changes in the assumptions can produce sharply different results.
In other words, with a pinch here and a prod there, analysts can make the math say anything about a stock.
The three authors theorize that the best recommendation changes are ones that stem from concrete new information, and that changes in near-term earnings forecasts are a good sign of such information. In the study, they find that stock prices drift much more when recommendation changes are accompanied by earnings-forecast revisions.
The authors calculate that between 1994 and 2007, a trading strategy of buying stocks following raised ratings and earnings estimates and holding for a month, while doing the opposite (short selling) for stocks following lowered ratings and estimates, would have returned more than 45% a year. That is several times what an S&P 500 index fund would have returned over the same period.
Ross Stores (ROST: 51.13, -0.36, -0.70%), a clothing chain, Broadcom (BRCM: 34.99, 1.64, 4.92%), a chip developer, and Discover Financial Services (DFS: 27.13, 0.21, 0.78%) have gained new analyst endorsements within the past four weeks and seen their earnings forecasts raised. Their shares are off to a strong start this year, up 7.6%, 8.4% and 9.5%, respectively, through Friday.
There is another way investors might be able to improve on analyst picks. That is by using analyst math in reverse, says Julian Koski, co-chief executive of Guggenheim Transparent Value, an investment firm.
"We start with the admission that the future is unknowable, and then we base our math on known measures," Mr. Koski says.
That means starting with the actual stock price rather than constructing a theoretical one. Mr. Koski's method involves calculating the number of widgets a company must sell to justify its current share price, called its required business performance, or RBP. The analyst uses the company's recent results as a guide in determining the probability it will achieve its RBP.
The RBP percentages change daily according to stock price. Mr. Koski points to Netflix (NFLX: 103.46, 4.92, 4.99%) as an example of a recent success. It had an RBP probability of below 5% last summer, when the stock price was over $280, but shares have since plunged below $100, and the stock recently had an RBP probability of nearly 90%.
An index that selects 100 stocks with the highest RBP probabilities, the Dow Jones RBP U.S. Large-Cap Core Index, has returned 10.8% a year in back-testing since 1998, versus 2.1% for its benchmark, the Dow Jones U.S. Large-Cap Total Stock Market Index.
A mutual fund that follows that approach, Transparent Value Dow Jones RBP U.S. Large Cap Market Index, launched in April 2010. It has since returned 6.5%, beating its benchmark by about 0.5 percentage point, despite expenses of $150 a year per $10,000 invested.
Among more than 2,200 stocks Transparent Value covers, Netgear (NTGR: 40.53, 1.54, 3.95%), DuPont (DD: 49.40, -0.05, -0.10%) and Eli Lilly (LLY: 40.17, -0.01, -0.02%) have RBP probabilities in the high 90s. Yahoo (THOO) and Office Depot (ODP: 2.56, 0.08, 3.23%)have probabilities in single digits.
No comments:
Post a Comment