Here's another article, on earnings estimates:
The Problem with Earnings Estimates
26-May-98
Comparing reported earnings numbers to "expectations" has become a
joke. The surveys that purport to measure expectations aren't accurate, as
more and more companies come in "a penny ahead." Whisper numbers,
truer reflections of what is in the price structure, are thus a more common
phenomenon.
Problems with Consensus Estimates
There are a number of reasons why the consensus estimates published by
First Call, Zacks, and others are not good measures of what investors
actually expect. To start with, the survey methods are never going to be
precise, in part because of the following reasons:
No weight to influence: Merrill Lynch's number counts just as much in
the sample as a small regional firm that may not affect trading.
Small samples: sometimes there are just a couple of analysts giving
forecasts, and one that makes a guess counts as much as one that
really studies the company.
Means, not medians: the surveys use average forecasts, which
means that a small firm that no one pays much attention to can have
an outlying forecast and pull the "consensus" in one direction.
The Big Problem
But the most damning problem with the surveys is that Wall Street analysts
simply do not try to accurately forecast earnings.
Simply put, Wall Street plays a game with earnings estimates. It serves no
one's interest when a company misses. It is much better to have earnings
come in at, or above, the published the "consensus." Wall Street has
learned to bias the surveys so that companies are likely to beat estimates.
Here's the proof.
Everyone Beats the Consensus
If analysts were paid based on how close their forecasts were to actual
quarterly numbers, they would try to come as close as possible, and be as
likely to be above the actual figure as below. Best efforts forecasting
would create something like a bell curve distribution of actual data
around consensus estimates.
In fact, every quarter, the number of firms beating estimates is about twice
as high as than the number that come in below estimates. This indicates
bias in the data.
For example, in the first quarter of this year, almost 55% of companies
reported earnings above expectations, while only 27% reported below
(the rest were as expected).
These figures are close to what was reported in the second and third
quarters of 1997 and have become the norm for reporting seasons: twice
as many come in a "penny ahead" or more, compared to those that miss.
Only in the fourth quarter of last year was there a substantial deviation.
Then, when the Asia crisis sprang up quickly, there should have been a lot
of companies that reported below expectations. Analysts quickly revised
numbers lower, but it seemed likely the impact was larger than expected
and hard to gauge, so large number of firms should have missed. Instead,
even under the worst conditions, about 45% beat estimates compared to
35% that missed.
Even When the Company Says So
Here is a clincher. On April 22, Computer Associates said that profits for the
current quarter would reach $0.75 per share.
Yet, when CA reported earnings on May 19 (plenty of time to update those
estimates) at, of all things, $0.75, the consensus estimate was only $0.74. The
press duly reported that Computer Associates reported earnings "a penny
ahead" of expectations! And analysts get paid to make these forecasts.
Whisper Numbers
The bias in the consensus numbers is becoming increasingly recognized. In
fact, it has now come to be expected that companies will beat
expectations. This of course undermines the supposed value of the
surveys, but with a measurable bias, this is how a rational market reacts.
As a result, it is becoming increasingly common for a stock price not to
react positively when earnings come in "above expectations." This is
further evidence that the consensus numbers do not in fact accurately
reflect what is built into the price structure
In fact, for many stocks, it is now expected that earnings will come in well
above expectations. When a company consistently beats the consensus
estimates, which many do (won't those Wall Street analysts ever learn?),
the talk as to what is really expected starts up.
This creates the so-called "whisper numbers."
Expectations for Dell were in the Whispers, not the Consensus
Dell was the most recent example. Prior to this quarter's earnings due last
Tuesday, astute investors noted that Dell had beaten estimates by an
average of 5 cents the past four quarters.
Therefore, it was logical that the talk was Dell would beat this quarter's
estimate of $0.42 by another 4 to 5 cents. Hence, the whisper numbers that
Dell would report as much as $0.47 per share started.
Dell started trading up in advance of the earnings report. Then, when Dell
reported $0.44 which was 2 cents ahead of the published estimates, it was
actually below what was built into the price structure. Dell supposedly did
better than expected, but sold off after the report.
There is no question that the whisper number is what was built into the
price structure by the time the report came out. The consensus estimate
was irrelevant, and so was a comparison to that estimate.
Don't Blame the Whispers
This is not the fault of the whisper numbers. In fact, so called whisper
numbers are rational expectations being built into prices. (Forget the silly
rumors that also fly around, most traders don't buy into them and they
aren't whisper numbers and don't affect prices).
When the supposed consensus expectations show a consistent bias,
rational traders will come to expect that bias.
It is becoming ever more important to understand the bias in these
consensus numbers, and to have a feel for what the market is really
expected. Don't get caught thinking that just because a company reports
earnings above the First Call estimates, that it means the company must be
doing well, or that the market has to react positively to the report.
Unless the method of collected the consensus changes, "beating
estimates" will increasingly have less impact on stock prices.
|
