Are Cover Stories Effective Contrarian Indicators?

Financial Analysts Journal
Volume 63 • Number 2
©2007, CFA Institute
Are Cover Stories Effective Contrarian Indicators?
Tom Arnold, CFA, John H. Earl, Jr., CFA, and David S. North
Headlines from featured stories in Business Week, Fortune, and Forbes were collected for a 20-year
period to determine whether positive stories are associated with superior future performance and
negative stories are associated with inferior future performance for the featured company. “Superior”
and “inferior” were determined in comparison with an index or another company in the same
industry and of the same size. Statistical testing implied that positive stories generally indicate the
end of superior performance and negative news generally indicates the end of poor performance.
he covers of business magazines have been
If the information contained in the article is not
considered informative as a potential con-
new, the market can have the following reactions:
Ttrarian signal (e.g., Stalter 2005; Forsyth 1996, • no reaction—that is, all information incorpo-
1997; Queenan 1991). Aside from anecdotal
rated in prices;
evidence and a few factual examples, however, such
•
a reaction that coincides with the popularity
a view has never been validated statistically. Others
of the stock—that is, a possible momentum
have studied specific business analysts or television
indicator;
programs (e.g., Desai and Jain 2004; Ferreira and
•
a reaction that coincides with the stock being
Smith 2003), tested the effect of company fortunes
mispriced—that is, a possible contrarian indi-
statistically, and demonstrated a limited effect on
cator.
future short-term stock prices (many times only on
Most of the anecdotal evidence supports the
the first day after the information was reported). We
idea that cover stories are not informational
statistically tested the effect of business magazine
because of the time needed to gather information
cover stories (as identified by the cover headlines)
for the article and to print it. The availability of
on stock prices/returns for more than a 1,000-day
instantaneous business news and, to a lesser extent,
horizon (500 days prior to and 500 days after publi-
daily business news outlets put weekly and
cation; we excluded the publication date from the
biweekly (every two weeks) news magazines at a
analysis). We investigated effects of cover stories
great disadvantage in providing new information
from Business Week, Fortune, and Forbes for a 20-year
about a company. Furthermore, the anecdotal evi-
period (1983–2002).
dence tends to view the scenario in which the article
To formalize how a magazine cover article
coincides with a stock being mispriced as giving
might affect future returns, one can assume that the
credence to the idea that the magazine cover story
given story reports new pertinent information or
is a contrarian indicator (again, assuming no new
simply reports past information. If the information
information is revealed in the article).
is genuinely new, the market can exhibit the follow-
We carried out tests to determine whether
ing reactions:
cover stories are indicators of future performance
•
an instantaneous and correct reaction—that is,
(momentum or contrarian). We did not consider
a very short term effect occurs but no lingering
whether or not the given article revealed new infor-
future effects;
mation about the company. Our bias, based on the
•
an underreaction—a potentially longer-term
disadvantages of weekly and biweekly business
effect occurs that makes the story a momentum
news magazines in reporting news in a timely man-
indicator;
ner, was toward the hypothesis that cover stories
•
an overreaction—a potentially longer-term
do not reveal new information about a company.
effect occurs that makes the story a contrarian
indicator.
Cover Story Data and Analysis
We analyzed companies that were the subject of
Tom Arnold, CFA, is assistant professor of finance at the
feature (cover) stories in Business Week (weekly),
University of Richmond, Virginia. John H. Earl, Jr.,
Fortune (biweekly), and Forbes (biweekly) maga-
CFA, and David S. North are associate professors of
zines between 1983 and 2002 for stock price perfor-
finance at the University of Richmond, Virginia.
mance 24 months prior to and 24 months after (i.e.,
70
www.cfapubs.org
©2007, CFA Institute

---

Are Cover Stories Effective Contrarian Indicators?
500 business days on either side of) the date the
•
Category 5: Company A is doing very poorly,
feature story was published. For purposes of the
or a scandal has occurred (pessimistic cover,
analysis, the “publication date” was the date on the
sometimes implying a future management
cover minus 7 calendar days for Business Week and
change and/or litigation).
minus 14 calendar days for Fortune and Forbes. In
In addition to the clear tone of the cover head-
other words, the publication date was considered
lines, the large sample should smooth individual
to be equivalent to the first day the magazine was
misinterpretations that may have occurred in cate-
available to readers.
gorizing the feature stories.
We categorized the cover headline of the fea-
There were 593 feature stories during the 20-
ture story according to a five-point scale (1 = very
year time period, for which 4 years of stock returns
positive, 2 = positive/optimistic, 3 = neutral, 4 =
negative, and 5 = very negative). Although the
for 549 subject companies were available for test-
determination of whether a story was positive or
ing.1 Table 1 displays the frequency of positive
negative appears to be quite subjective, the cover
(Categories 1 or 2), neutral (Category 3), and nega-
headlines of the stories clearly followed particu-
tive (Categories 4 or 5) feature stories in the period,
lar patterns:
in total and by periodical. Note that, in general,
•
Category 1: Company A “is” or “has done”
magazine feature stories do not focus on specific
something innovative or profitable (very posi-
corporations. This aspect is demonstrated by the
tive cover).
fact that out of a potential of 2,080 cover stories, only
•
Category 2: Company A “plans to do” or “is in
593 focused on a particular corporation. Also, fea-
the process of doing” something innovative for
ture stories tend to be positive; 350 of the 549 sample
the future, but will it work? (optimistic cover).
feature headlines (63.7 percent) were definitely pos-
•
Category 3: The cover gives no particular opin-
itive, and only 100 of the 549 sample feature stories
ion as to whether Company A is good or bad
(18.2 percent) were definitely negative.
(companies are identified on the cover, but the
cover gives no indication of a positive or neg-
Table 2 provides characteristics of the compa-
ative slant to the feature).
nies featured in the cover stories. The companies
•
Category 4: Company A has experienced
tended to be large, without much variability in size
“poor performance,” but the end of the poor
between the three periodicals. This characteristic is
performance may be near (pessimistic past but
not surprising because the magazines tend to have
a turnaround is predicted).
similar readerships.
Table 1.
Feature Story Frequency, 1983–2002
Neutral
Positive Categories
Category
Negative Categories
Periodical
1
2
Total
3
4
5
Total
Business Week
62
97
159
57
38
36
74
(273 features)
(21%)
(33%)
(54%)
(20%)
(13%)
(12%)
(26%)
Fortune
76
44
120
24
3
9
12
(146 features)
(49%)
(28%)
(77%)
(15%)
(2%)
(6%)
(8%)
Forbes
47
24
71
18
7
7
14
(90 features)
(46%)
(23%)
(69%)
(17%)
(7%)
(7%)
(14%)
Total
185
165
350
99
48
52
100
(549 features)
(34%)
(30%)
(64%)
(18%)
(9%)
(9%)
(18%)
Table 2.
Size of Companies Featured on Covers, 1983–2002
(market capitalization in $ millions)
Company Characteristic
Business Week
Fortune
Forbes
Total
Market-cap mean
43,653.6
52,371.4
38,112.6
45,091.1
Market-cap median
14,365.2
19,781.2
16,771.3
15,112.3
Adjusted market-cap meana
49,532.3
61,312.5
41,651.5
51,401.5
Adjusted market-cap mediana
19,884.4
23,402.2
16,711.1
19,332.0
aAdjusted to 2002 U.S. dollars using the U.S. Consumer Price Index.
March/April 2007
www.cfapubs.org
71

---

Financial Analysts Journal
To analyze the effect of a given feature story on
for testing of cover stories as contrarian or momen-
a company, we calculated prior-period holding-
tum indicators. Specifically, our goal was to mea-
period returns for 1 month (–21 to –1 trading days),
sure something more “predictive” of the future, in
6 months (–125 to –1 trading days), 12 months (–250
the sense of allowing a specific trading strategy to
to –1 trading days), and 24 months (–500 to –1
succeed, than a “quick” adjustment by the market.
trading days) prior to the publication of the story.
Table 3 displays the average holding-period
We calculated similar postpublication holding
return (HPR) for all companies within a category for
periods, with an additional holding period of 1 day
various time horizons before and after publication
to 5 days (to examine short-term effects). We
dates. The “adjusted return” (AR) is the average of
excluded the day on which the feature story was
the holding-period return minus the equivalent
published from the holding-period returns to allow
holding-period return for the value-weighted CRSP
Table 3.
Holding-Period Returns for Three Groups, Data for 1983–2002
(measured in percentages)
Period from Publication (in trading days)
Category/Measure
–500 to –1
–250 to –1
–125 to –1
–21 to –1
1 to 5
1 to 21
1 to 125
1 to 250
1 to 500
Positive categories—1 or 2 (n = 338)a
HPR
111.07
43.68
19.19
2.34
1.25
1.16
9.01
18.23
37.63
t-Statistic
7.47***
9.97***
8.10***
2.97***
3.35***
1.63
4.70***
5.85***
5.90***
Signed rank-sum test
13.26***
11.05***
8.24***
2.35**
3.33***
1.60
4.66***
5.60***
6.79***
AR
79.58
28.17
11.94
1.13
0.72
–0.11
2.97
5.93
12.92
t-Statistic
5.42***
6.75***
5.59***
1.59
2.27**
0.18
1.69*
2.07**
2.13**
Signed rank-sum test
8.66***
7.00***
4.88***
0.49
1.77*
0.47
0.85
0.47
0.24
AHPAR–ISM
24.97
11.11
5.00
0.87
0.39
–0.36
5.13
3.85
11.48
t-Statistic
1.45
2.14**
1.81*
0.97
0.82
0.42
2.51**
1.18
1.65*
Signed rank-sum test
2.75***
2.49**
2.16**
0.34
1.26
1.04
1.78*
1.02
1.05
Neutral Category 3 (n = 98)b
HPR
131.00
56.60
20.13
3.77
0.06
3.94
13.31
26.83
43.37
t-Statistic
2.81***
3.40***
3.67***
2.38**
0.10
1.69*
3.68***
3.58***
4.99***
Signed rank-sum test
6.73***
5.43***
3.81***
2.05**
0.36
1.42
3.87***
3.64***
4.98***
AR
96.90
40.83
13.22
2.01
–0.36
1.85
5.34
14.47
17.87
t-Statistic
2.09**
2.48**
2.50**
1.49
0.66
0.84
1.59
2.04**
2.25**
Signed rank-sum test
2.06**
2.29**
1.66*
1.29
1.11
0.33
1.55
1.01
1.03
AHPAR–ISM
2.11
14.11
9.52
2.23
–0.33
1.24
4.66
8.88
12.61
t-Statistic
0.03
0.83
1.87*
1.46
0.64
0.59
1.04
1.57
1.42
Signed rank-sum test
2.30**
1.40
1.53
1.49
1.04
0.85
1.02
1.52
1.59
Negative categories—4 or 5 (n = 96)c
HPR
6.11
–2.12
–0.72
–0.31
1.53
0.87
2.98
13.18
31.75
t-Statistic
1.08
0.55
0.23
0.21
1.98**
0.69
1.23
3.18***
4.42***
Signed rank-sum test
0.48
0.16
0.40
0.20
2.25**
0.42
1.18
2.83***
4.27***
AR
–23.15
–15.96
–8.67
–2.04
1.11
0.30
–0.50
3.58
8.23
t-Statistic
4.79***
4.70***
2.73***
1.15
1.73*
0.29
0.24
0.94
1.25
Signed rank-sum test
4.80***
4.78***
2.86***
1.46
1.88*
0.08
0.53
0.04
0.24
AHPAR–ISM
–71.22
–34.62
–13.54
–3.15
0.44
–0.17
0.08
3.22
7.91
t-Statistic
3.34***
4.36***
3.66***
1.68*
0.57
–0.11
0.27
0.75
1.07
Signed rank-sum test
4.85***
4.86***
3.41***
1.49
0.17
0.17
0.64
0.20
0.37
Notes: HPR = holding-period return, AR = adjusted return, and AHPAR-ISM = average holding-period abnormal return. The t-statistics
are for differences from zero. Signed rank-sum statistics are for Wilcoxon signed rank-sum tests for differences from zero.
a Twelve cover headlines were redundant because the companies were in Category 1 or 2 within the prior three months.
b One cover headline was redundant because the company was in Category 3 within the prior three months.
c Four cover headlines were redundant because the companies were in Category 1 or 2 within the prior three months.
*Significant at the 10 percent level.
**Significant at the 5 percent level.
***Significant at the 1 percent level.
72
www.cfapubs.org
©2007, CFA Institute

---

Are Cover Stories Effective Contrarian Indicators?
stock index. The “average holding-period abnormal
egories 4 and 5 are presented separately in Table 4,
return—industry and size matched” (AHPAR-ISM)
almost all statistical significance on an adjusted
is the average of the HPR of a given company minus
basis disappears. The holding-period return for the
the equivalent HPR of another company that was
one-month period prior to publication for Category
matched to it by size and industry (i.e., a non-cover-
5 companies, however, is significantly negative on
story equivalent company).2 These two measures
an adjusted basis.
provided adjustments to the initial HPR calculation
In summary, we found no signs that negative
to determine whether a positive or negative abnor-
stories are a momentum or contrarian indicator. If
mal return was present.
a trader was shorting one of these stocks, however,
For all return calculations, we attempted to
the trader should have considered covering the
eliminate duplicate observations, defined as a fea-
short position after seeing a negative cover head-
ture story happening fewer than three months after
line on the company because the stock hit its “bot-
another feature for a given company and having
tom” level at that time.
the same categorization. We included the first
To test the robustness of these results, we per-
instance and dropped the subsequent observa-
formed a calendar-time portfolio analysis. For each
tion(s).3 Additionally, when a company was
category and subperiod, we formed an equally
delisted after publication of the issue in which it
weighted portfolio of companies; we rebalanced it
was featured, we carried out the return calculation
every month for the entire sample period, January
through the delisting date and included it in all
1983 through December 2002. Over time, some com-
calculations of holding-period returns.
panies dropped out of the portfolio as they were
Table 3 indicates that for companies with pos-
delisted. The results were consistent with Table 4.4
itive cover stories, positive holding-period returns
existed before and after the publication date that are
Conclusion
generally statistically significant (based on a t-test
and a Wilcoxon signed rank-sum test). When the
As one might expect, positive feature stories head-
holding-period returns were adjusted for an index
lined on business magazine covers follow extremely
or for size/industry, however, most positive abnor-
positive company performance and negative head-
mal returns dissipated after the publication date.
lines follow extremely negative performance. In
This result is not always the case; the AR measure
both cases, however, the appearance on a cover of
displays some positive adjusted returns based on
Business Week, Fortune, or Forbes tends to signal the
t-tests (a mean-based test) but not always based on
end of the extreme performance. Going forward
the signed rank-sum test (a median-based test). The
from publication dates, we found only weak evi-
AHPAR-ISM measure displays a statistically signif-
dence that optimistic cover stories (Category 2) are
icant positive adjusted return based on agreement
an indicator for momentum for a six-month horizon
by both statistical tests only for the six-month post-
after publication. Negative cover headlines appar-
publication horizon.
ently do not provide a good signal for momentum
In Table 4, where Category 1 and Category 2
or contrarian strategies when performance is mea-
companies are separately examined, much of the
sured against an index or measured on a size/
statistical significance vanishes except for the six-
industry-adjusted basis—despite a popular belief
month horizon AHPAR-ISM measure for Category
that such stories are a contrarian signal. Indeed,
2. Consequently, for companies that are the subject
companies in the study that were the subjects of
of optimistic cover stories, we found, at best, weak
negative feature stories tended to have positive
evidence of a momentum indicator over the six-
holding-period returns after publication of the mag-
month horizon.
azine, but the positive return was not abnormally
For the companies with negative cover stories,
positive when adjusted for an index or for size and
Table 3 reports statistically negative adjusted
industry. Consequently, we conclude that if an
holding-period returns for the two years (and peri-
investor is short the stock of a company that is the
ods of less than two years) prior to publication,
subject of a negative cover story, the publication of
except for one month prior to publication. After
the story indicates it is time to cover the short posi-
publication, the holding-period returns are no
tion because the stock has hit bottom.
longer negative and, on an adjusted basis, tend to
This article qualifies for 1 PD credit.
not be significantly different from zero. When Cat-
March/April 2007
www.cfapubs.org
73

---

Financial Analysts Journal
Table 4.
Holding-Period Returns: Positive and Negative Categories, Data for 1983–2002
(measured in percentages)
Period from Publication (in trading days)
Category/Measure
–500 to –1
–250 to –1
–125 to –1
–21 to –1
1 to 5
1 to 21
1 to 125
1 to 250
1 to 500
Positive Category 1 (n = 182)a
HPR
119.41
56.92
19.73
1.54
1.15
1.45
8.13
17.37
29.81
t-Statistic
5.70***
7.41***
5.77***
1.78*
2.20**
1.69*
3.66**
3.86***
5.16***
Signed rank-sum test
10.50***
9.00***
6.53***
1.30
2.24**
1.70*
3.21***
3.58***
4.87***
AR
88.82
32.01
13.02
0.64
0.79
–0.03
1.82
4.63
5.07
t-Statistic
4.29***
5.31***
4.19***
0.78
1.56
0.06
0.87
1.09
0.93
Signed rank-sum test
8.00***
6.575***
4.62***
0.04
1.20
0.05
0.04
0.38
0.77
AHPAR-ISM
42.71
17.63
8.25
1.03
0.37
–0.43
2.87
2.75
4.18
t-Statistic
1.87*
2.40**
2.05**
0.93
0.50
0.51
1.26
0.62
0.59
Signed rank-sum test
3.91***
3.36***
2.92***
0.51
0.38
0.91
0.19
0.06
0.34
Positive Category 2 (n = 162)b
HPR
105.61
39.92
18.65
3.23
1.22
0.51
9.57
19.59
45.76
t-Statistic
5.02***
6.89***
5.84***
2.39**
2.55**
0.51
3.04***
4.59***
3.92***
Signed rank-sum test
8.23***
6.71***
5.27***
2.08**
1.97**
0.21
3.24***
4.32***
4.79***
AR
73.37
23.78
10.93
1.68
0.51
–0.21
4.11
7.92
21.43
t-Statistic
3.54***
4.32***
3.81***
1.35
1.48
0.27
1.44
2.11**
1.93*
Signed rank-sum test
4.28***
3.42***
2.45**
0.69
0.94
0.73
1.07
1.08
0.38
AHPAR-ISM
4.12
1.72
0.03
0.22
0.37
–0.16
7.46
5.15
18.54
t-Statistic
0.16
0.25
0.01
0.23
0.75
0.10
2.16**
1.11
1.53
Signed rank-sum test
0.05
0.10
0.16
0.17
1.45
0.51
2.30**
1.39
0.99
Negative Category 4 (n = 47)c
HPR
7.11
1.61
4.44
4.27
0.78
0.285
0.12
6.89
23.66
t-Statistic
0.88
0.36
1.06
1.75*
1.01
0.19
0.02
1.50
3.29***
Signed rank-sum test
0.11
0.27
0.23
1.37
0.92
0.27
0.11
1.39
2.76***
AR
–21.13
–11.40
–3.17
2.14
0.81
–0.20
–2.58
0.20
4.01
t-Statistic
2.90***
2.78***
0.78
0.92
1.23
0.15
0.39
0.05
0.60
Signed rank-sum test
3.20***
2.83***
1.11
0.26
1.06
0.25
0.83
0.50
0.21
AHPAR-ISM
–111.03
–46.10
–12.67
0.645
–0.58
–2.88
–2.18
0.72
5.91
t-Statistic
2.65***
3.14***
2.22**
0.25
0.70
1.89*
0.48
0.14
0.63
Signed rank-sum test
3.54***
3.23***
1.84*
0.20
1.01
1.40
0.15
0.02
0.50
Negative Category 5 (n = 52)
HPR
4.62
–6.33
–6.21
–4.73
1.81
1.14
5.27
18.78
39.63
t-Statistic
0.60
1.10
1.21
1.89*
1.56
0.65
1.57
2.92***
3.39***
Signed rank-sum test
0.41
0.72
1.05
1.09
1.93*
0.12
1.51
2.65***
3.34***
AR
–25.22
–20.44
–14.57
–5.85
1.38
0.79
0.79
5.98
12.03
t-Statistic
4.09***
4.03***
3.17***
2.44**
1.17
0.44
0.22
1.00
1.12
Signed rank-sum test
3.63***
4.01***
2.93***
2.53**
1.37
0.13
0.32
0.24
0.24
AHPAR-ISM
–34.63
–23.72
–15.41
–6.78
1.25
2.22
2.98
5.37
12.42
t-Statistic
3.77***
3.76***
3.31***
2.76***
1.12
1.25
0.79
0.81
1.11
Signed rank-sum test
3.50***
3.71***
3.17***
2.53**
0.95
1.51
0.99
0.17
0.40
Notes: See notes to Table 3.
a Three cover headlines were redundant because the companies were in Category 1 within the prior three months.
b Three cover headlines were redundant because the companies were in Category 2 within the prior three months.
c One cover headline was redundant because the company was in Category 1 within the prior three months.
*Significant at the 10 percent level.
**Significant at the 5 percent level.
***Significant at the 1 percent level.
74
www.cfapubs.org
©2007, CFA Institute

---

Are Cover Stories Effective Contrarian Indicators?
Notes
1.
For hypothesis testing, we used the 24 months prior to the
3.
The results remained qualitatively similar whether these
feature story.
observations were deleted or not.
2.
Size and industry match was based on market value of
4.
These tabulations are available through a link from this
equity and two-digit SIC classification and was set as of the
issue’s contents page at www.cfapubs.org.
day of publication.
References
Desai, H., and P. Jain. 2004. “Long-Run Stock Returns Following
———. 1997. “Beware: Magazine Cover Stories about Fed
Briloff’s Analyses.” Financial Analysts Journal, vol. 60, no. 2
Chairman Tend to Be Contrarian Indicators for Bonds.” Barron's
(March/April):47–56.
National Business and Financial Weekly, vol. 77, no. 28 (14
Ferreira, E., and S. Smith. 2003. “‘Wall $treet Week’: Information
July):MW10.
or Entertainment?” Financial Analysts Journal, vol. 59, no. 1
Queenan, J. 1991. “Exotic Indicators—Give Paul Macrae
(January/February):45–53.
Montgomery a Forecasting Edge.” Barron's National Business and
Forsyth, R. 1996. “On Borrowed Time: Why Investors Should
Financial Weekly, vol. 71, no. 22 (3 June).
Worry about Magazine’s Cover on Fidelity.” Barron's National
Stalter, K. 2005. “Magazine Covers Often Signal the Top of a
Business and Financial Weekly, vol. 76, no. 40 (30 September):18.
Great Stock’s Rally.” Investor’s Business Daily (22 June):B07.
[ADVERTISEMENT]
March/April 2007
www.cfapubs.org
75

---