To print article, click
Print on your browser's File menu.
Go
back
Posted
2/6/2003 |
Related Sites
Trial
offer from Zacks Advisor
The Speculator
Recent articles:
• Investing
lessons you can't refuse, 1/30/2003
• Buy the very
worst stocks you can find, 1/23/2003
• We're buying 8
companies that buy themselves, 1/16/2003
More...
|
|
| Advertisement |
![Click Here!]() | |
The Speculator
A
nice surprise may not mean a nice profit
Companies that surprise on
earnings outperform the market, or so the experts say. We find
that's not necessarily the case.
By Victor
Niederhoffer and Laurel Kenner
The politics of surprise leads through the Gates of
Astonishment into the Kingdom of Hope.
-- Max
Lerner All participants in the human comedy know that
one of the most pleasant things in life is a positive surprise.
Examples are many: the unexpected compliment, the extra appetizer
the chef sends before a fancy dinner, the intriguing guest at a
party you were told was boring, the chance meeting with a good
friend, the unexpectedly lingering goodnight kiss from a significant
other.
The same is
true in stock investing. Nothing pleases more than an earnings
report that beats expectations. Happily, many of them do. Some 62%
of the 388 S&P 500 ($INX)
companies that have reported earnings for the last quarter beat
estimates. Even better, 74 of those 338 companies beat the forecast
by more than 10%.
Surprises can be even more substantial. The
five Dow companies that reported the greatest positive surprises for
the last quarter were J.P. Morgan (JPM,
news,
msgs),
which made 36 cents a share instead of losing 9 cents; MSN Money
parent Microsoft (MSFT,
news,
msgs),
128% better than expected; Caterpillar (CAT,
news,
msgs),
29%; AT&T (T,
news,
msgs),
26%; and International Paper (IP,
news,
msgs),
22%. The only four that fell short of expectations were
Alcoa (AA,
news,
msgs),
by 36%; Kodak (EK,
news,
msgs),
4%, McDonald’s (MCD,
news,
msgs),
1%; and Honeywell (HON,
news,
msgs),
0.4%.
The hunt for
momentum
One of the most venerable techniques for reaping
above-average returns in the market is “earnings momentum” -- to buy
positive earnings surprises and sell negative ones. The big
investment advisers Value Line and Zacks Investment Management
include this basic method in their strategies, as do a horde of
momentum investors in the mutual-fund world. All academics accept
the theory.
The systematic studies on earnings surprises are
so well known that they are discussed in all modern finance books.
For example, the 2002 edition of Zvi Bodie’s classic textbook
"Investments" reports the highest tenth of companies that beat
earnings estimates outperforms the market by some 6 percentage
points in the next month, while those with the worst surprises
perform 2 percentage points worse than the market. (Those results
are based on data from the 1960s.)
Another indication of the
wide acceptance of earnings-momentum strategies is their presence in
glossaries and introductions to investments. Take the answer to
“What is momentum investing?” on the “Just Ask James” Web site:
“Momentum investors watch earnings like a hawk. They even go so far
as to watch the ongoing earnings revisions and inevitable earnings
surprises to determine if a stock has the right stuff.”
Generating forecasts -- and
surprises
Not surprisingly, companies have become
amazingly creative in managing the process of generating the
forecasts.
“A primary way in which firms manage earnings is
by ‘talking down’ analysts’ forecasts,” Dan Bernhardt and Murillo
Campello, both of the University of Illinois, reported in a Dec. 29,
2002, draft of a paper entitled “The Dynamics of Earnings Forecast
Management.” “Quarterly earnings forecasts issued later in the
forecasting cycle are much lower than those early in the cycle.” The
idea is to beat the lowered estimate. (Unfortunately for the
expectations managers, Bernhardt and Campello concluded that stocks
suffer more from negative revisions than they gain from subsequent
positive surprises.)
Mitch Zacks, a portfolio manager at
Zacks Investment Management, a family-run firm founded by his father
and uncle in 1980, includes an updated study in his forthcoming
book, "Ahead of the Market." He found exactly the same disparity.
Using data from 1990 through 2002, he studied companies that
reported positive surprises greater than 1% and found that during
the following month, those companies beat the market by 6 percentage
points on an annualized basis. “Negative surprise” companies
returned 2 percentage points less than the market in the next month.
Results were similar for earnings-forecast revisions, Zacks
found. Companies that had upward revisions over the period of a
month performed 9 percentage points better than the market during
the next month, gaining 9%. Companies that had their estimates
lowered underperformed the market by 8 percentage points, gaining
just 1%.
Combining these two methods, Zacks has an enviable
record in predicting future stock price returns. The actual
companies that Zacks ranked No. 1 (the top 5% in the combined
ranking system) over the previous 22 years returned an average of
32% compounded. The companies in the worst 5% -- some 215 companies
each year -- returned an average of 1% a year compounded. The
S&P 500 returned 13% in the same time period.
(Mitch
Zacks wrote his senior economic thesis at Yale on analyzing the
betting market at Milford Jai-Alai in Connecticut, a favorite Spec
Duo hangout. “I found statistically significant positive returns
could be generated by a 7-8 quinella gamble, but to this day am not
sure if the result is from a rigging of the market or data mining,”
he told us. He received an MBA in analytic finance and statistics in
1999 from the University of Chicago Business School and worked as an
investment banking analyst at Lazard Freres in New York before
joining the family firm.)
Unlike most forecast methods, which
have suffered hits and changes in fortune in 2002, the Zacks system
held up quite well. Its stocks went down only 6%, vs. 27% for the
bottom 5%. As Zacks summarizes the study in his book, “Basically
stocks that are receiving upward earnings estimate revisions tend to
substantially outperform over time.”
Out of the test tube
Zacks has put
theory to practice in the portfolios it manages. The oldest, “Zacks
Rank,” uses both earnings surprises and forecast revisions. The Rank
portfolio has returned 13.3% a year after fees since inception on
Jan. 31, 1995, better than the S&P 500’s 11.6%. The other two
portfolios, begun in May 2001 and January 2002, also have
outperformed the market.
A close look at Zacks Rank’s annual
results provides a nice illustration of changing cycles.
| Difference between Zacks Fund and the
S&P 500 |
| Year |
Zacks Fund (net of fees) |
S&P 500 |
Difference (% pts) |
| 1995 |
32.9% |
34% |
-1.1 |
| 1996 |
24% |
20.3% |
3.7 |
| 1997 |
31.4% |
31% |
0.4 |
| 1998 |
14.6 |
26.7% |
-12.1 |
| 1999 |
16.9 |
19.5% |
-2.6 |
| 2000 |
14.1 |
-10.1% |
24.2 |
| 2001 |
-7.1 |
-13% |
5.9 |
| 2002 |
-19.4 |
-23.4% |
4 | |
Andrew
Zimmer of Zacks attributed the portfolio’s underperformance in 1998
and 1999 to the then-prevailing fashion of according
multibillion-dollar valuations to Web companies with a few hundred
thousand dollars in revenue. Strategies based on earnings momentum
aren’t likely to work if earnings don’t matter. “Historical
valuation models went out the window,” Zimmer said. When dot-coms
lost their appeal, the Zacks system began outperforming.
How,
then, to explain the decline in fortunes of Richard Driehaus? This
earnings-momentum star, who came to our attention when we examined
one of the trend-follower sites, racked up 12 years of 30% returns
employing the above techniques -- and then lost 30% a year in 2001
and 2002.
When universal agreement exists on the efficacy of
a simple method for investment, our antennae start buzzing. “Er,
have you tested that recently?” is our standard question.
We
distrust the inn that offers a free lunch, the advertisement touting
a risk-free way to make a quick 25%, the close relative of the
recently executed Third World finance minister who will share
millions of dollars of plunder with us for a small administrative
fee, paid in cash to a numbered account. Why should things be
different in the treacherous stock market? As we never tire of
reminding ourselves, when a system is widely believed and accepted
by the public, the law of ever-changing cycles unfailingly comes
into play.
Given that there are hundreds of thousands of
articles on earnings surprises, it would be impossible for us to
replicate and evaluate all of the scientific work in this field. Our
main concerns with the studies underlying the “follow the earnings”
strategy are as follows:
- Results are better in the test tube than in real life.
- The techniques are good only for a season.
- Retrospective bias mars the results.
We therefore took
out our pencils and envelopes to see how the strategy has worked
lately. In brief, we found that it pays to buy companies that beat
forecasts, but you may lose if you sell too soon.
First, we
looked at earnings reports and subsequent price moves for each of
the 30 Dow stocks in each of the last six quarters. To calculate the
surprise, we looked at the reported earnings per share relative to
the I/B/E/S consensus forecast the day before the report. (The
so-called consensus forecast is not arrived at by consensus; rather,
it is the mean of all estimates made by the analysts who follow the
company).
We then classified the actual reports into three
categories -- positive surprise, right on target (no surprise) and
negative surprise. We then tracked the performance of each of these
three groups of companies in the first five and 30 days after the
report date’s close.
One might think that by the time a
positive surprise is announced, it’s too late for an investor to
benefit. Our results suggest that buying at the close of the
announcement day and holding for 30 days is not entirely
unprofitable.
| Earnings surprises and stock market
returns |
| (Last six quarters, excluding
reports issued within the past 30 days.) |
|
| 30-day move after: |
|
| Negative surprise |
|
| Average |
-0.6% |
| St dev |
9.2% |
| Count |
34 |
| Probability* |
0.4 |
| No surprise |
|
| Average |
1.6% |
| St dev |
12.3% |
| Count |
11 |
| Probability |
0.44 |
| Positive surprise |
|
| Average |
4.6% |
| St dev |
15% |
| Count |
110 |
| Probability |
0.03 | | *Probability of chance or random
occurrence
The five-day results are consistent
with randomness:
| Earnings surprises and stock market
returns |
| (Last six quarters, excluding
reports issued within the past 30 days.) |
|
| 5-day move after: |
|
| Negative surprise |
|
| Average |
-2.1% |
| St dev |
4.7% |
| Count |
37 |
| Probability* |
0.01 |
| No surprise |
|
| Average |
0.4% |
| St dev |
8.0% |
| Count |
11 |
| Probability |
0.2 |
| Positive surprise |
|
| Average |
-0.45% |
| St dev |
6.66% |
| Count |
126 |
| Probability |
0.22 | | *Probability of chance or random
occurrence
You can use “Advisor FYI-Analyst
Projections” on MSN Money's Stock Screener to find earnings
surprises. To see the results of the search we ran to see how many
Dow Jones Industrial Average companies beat forecasts for the last
two quarters, click
here.
We performed a similar study on revisions of
earnings for Dow companies. In accord with the principle of
ever-changing cycles, the results were completely opposite to the
general wisdom. Those companies with upward revisions actually
performed worse than the Dow, while those with downward revisions
performed significantly better.
| Forecast revisions and stock-market
returns |
| (2001-2002) |
|
| 30-day move after: |
|
| Upward revisions |
|
| Average |
-3.8% |
| St dev |
3.7% |
| Count |
10 |
| Probability* |
0.12 |
| Downward revisions |
|
| Average |
2.5% |
| St dev |
10.1% |
| Count |
43 |
| Probability |
0.84 |
| Dow move in comparable period |
|
| Average |
1.1% |
| St dev |
5.4% |
| Count |
53 |
| Probability |
0.14 | | *Probability of chance or random
occurrence
| Forecast revisions and stock-market
returns |
| (2001-2002) |
|
| 30-day move after: |
|
| Upward revisions |
|
| Average |
-3.5% |
| St dev |
4.7% |
| Count |
10 |
| Probability* |
0.02 |
| Downward revisions |
|
| Average |
1.9% |
| St dev |
5.8% |
| Count |
43 |
| Probability |
0.03 |
| Dow move in comparable period |
|
| Average |
0.5% |
| St dev |
2.9% |
| Count |
53 |
| Probability |
0.19 | | *Probability of chance or random
occurrence
Thus, the market teaches us once again to
test ideas before committing cash.
The workouts of both of
our studies are available on our Web site.
As a
special offer to readers of The Speculator, Zacks is offering a
one-month free trial to the firm’s online premium newsletter,
www.zacksadvisor.com. (See the link at left under "Related
Sites.").
Kindly e-mail
us with your thoughts on surprises, our articles and the market
in care of so we can distill your knowledge of time and place into
our articles for our mutual benefit.
Mitch Zacks kindly has
provided our readers with a 30-day free trial of
the Zacks Advisor newsletter. Usually, it's only for
two weeks.
http://www.zacksadvisor.com/freetrial/session.php?ADID=MSN.
For Dan Bernhardt’s advice on how
investors can use earnings forecast revisions,
click here.
|
