A University of Illinois and University of Florida study recently found that analyst estimates for growth of the companies they cover have, on average, been much higher than the earnings turned out to be. Investors are supposed to use valuation as a key portion of their investing research, but with no real way to predict future growth rates based upon the past, this is a tricky thing to do. That doesn't mean that investors shouldn't try, however.
By
The Motley Fool Otter
You'll take from this story a lesson similar to one Winston Churchill imparted when he said "Democracy is the worst form of government, except for all those others."
Brace yourself, because neither I nor anyone else has much of a solution to the intractable problem of "properly" valuing a company predicting its future cash flows. Anyone who has spent any time investing has seen, heard of, or maybe even produced a complicated model that projects out earnings, dividends and/or cash flows to determine how much a company is worth. It's supposedly what analysts do for a living.
But a recent study by several academics suggests that analysts are not really very good at predicting growth rates. According to a University of Illinois study by Louis Chan, Jason Karceski, and Josef Lakonishok called "The Level and Persistence of Growth Rates," (.pdf file) analysts' predictions for earnings in any five-year period between 1982 and 1998 came in significantly higher than the actual growth rates.
Analysts make the same mistake most investors do: Assuming a company that has previously generated significant rates of growth will continue to do so. In fact, as companies grow bigger they eventually have their rates of growth converge with the mean: No company grows faster than the average indefinitely.
But if the rate of growth is generally unpredictable, is it useful to build these models at all?
On the edge, absolutely. Doing cost models forces you to think about the business, particularly in terms of future growth. If you're going to project out a company's earnings for 10 years, you ostensibly need to consider whether the company's goods and services will even be relevant a decade hence. Unfortunately, these models are quite susceptible to the "garbage in, garbage out" phenomenon. If you go in with the preconceived notion that Yahoo! (Nasdaq: YHOO) is worth $300 per share, odds are your model will reinforce this position.
There's a deeper problem, though: Markets never grow or decline in a straight line. You can't, for example, forecast steady 18% growth per year for Yahoo!. It grew 100% last year but thus far in 2001 is shrinking. Who knows what will happen in the future -- the notoriously cyclical advertising market may make an about face and Yahoo! will be off and running again.
Remember last year, when several research firms forecast that the annual market for business-to-business e-commerce would be several trillion dollars by 2003? Investors bid nearly all of the publicly traded companies in that space to the moon in anticipation of the riches that would roll in. We're more than a year and a half away from 2003, but I think that at this point it is fairly safe to say that those fantastic projections are going to end up being on the high side by a factor of five or so.
This wasn't some mass delusion, and I'm sure that the growth rates that were reported weren't juiced for shock value. And yet many professionals who put their names on such projections will turn out to have been spectacularly wrong.
In investing there is a concept known as the "equity risk premium," which is the added return over the risk-free rate -- usually pegged to the yield on a 30-year U.S. Treasury Note -- an investor expects to get by putting his money in a stock that, at least theoretically, also has the potential for total loss. It would make no sense for an investor to put money into a high-risk stock and expect returns of 2% per year when he could just as easily put it into treasuries and gain returns of 5.5%.
The higher the price-to-earnings multiple, the lower the equity risk premium the investor is willing to receive. Thus, in May of last year, when Cisco (Nasdaq: CSCO) was valued in excess of $500 billion, anyone demanding a risk premium of 9.5% -- which, added to 5.5%, would bring the annual return up to 15% -- was expecting the company to be valued at $2 trillion in 2010 and more than $8 trillion in 2020.
But seeing as the U.S. has a GDP of $11 trillion now and in all likelihood will grow at a maximum sustainable rate of 4%, those numbers work out to be, respectively, fully one-eighth and one-third of the national economy.
The truly absurd part of this scenario was that, to a great number of investors and professional analysts, Cisco was expected to grow at a much higher rate than 15% per year. Even three months ago, the analyst estimate for growth by Cisco exceeded 28% per year for the next five years. This is coming from the professionals, who ostensibly use profit projection models significantly more detailed than mine. According to I/B/E/S, the average analyst growth expectation on the average company over the next five years is in the range of 20% per year.
What the Illinois study found was quite contrary to expectations such as those expressed above. In a universe of more than 9,000 stocks, tracked from 1951 to 1998, only one in 10 achieved a return in excess of 18% over any 10-year period. In fact, the median rate of growth for the companies during this period was about 9%.
This means analysts' average predicted rate of growth -- the 20% I/B/E/S figure noted above -- is higher than the growth more than 90% of companies have been able to achieve in the past. Even with the potential benefits of technological efficiencies and other arguments, that seems quite aggressive.
More scary data shows that, on average, in any five-year period, less than half of all companies provide any return at all from their purchase price. Ostensibly, investors risk their money on stocks in order to achieve a gain higher than that available from risk-free vehicles. What a kick in the teeth to know that more than half of all companies have historically provided no return at all after five years.
This is not to say that cost modeling is futile. In fact, it is essential that an investor go through the test of logic to determine just how a company will have to perform in order to give him a sufficient return. When a company is valued at a price to earnings ratio of 100, the investor who expects a return of 15% and an ending P/E of 25 is essentially predicting that the company will grow in excess of 30% per year during that time. It's possible, but less than one company in 50 manages to grow at that rate.
Might the company you are analyzing be the one? Maybe. But anyone who wishes to pay that kind of multiple should check his assumptions very carefully. And just because an analyst says that a company is going to grow by 150% doesn't mean it will be so: She's guessing, along with you and everyone else.
Fool on!
Bill Mann, TMFOtter on the Fool Discussion Boards
Bill Mann is convinced that one of the seven dwarfs is named "Torpid." At time of publishing, Bill had beneficial interest in Cisco Systems. The Motley Fool is investors writing for investors.