You are on page 1of 17

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/227430029

The Tiger Woods scandal: A cautionary tale for event studies

Article  in  Managerial Finance · April 2012


DOI: 10.1108/03074351211217850 · Source: RePEc

CITATIONS READS
15 1,612

1 author:

Matthew Hood
Texas State University
18 PUBLICATIONS   405 CITATIONS   

SEE PROFILE

All content following this page was uploaded by Matthew Hood on 29 July 2016.

The user has requested enhancement of the downloaded file.


The current issue and full text archive of this journal is available at
www.emeraldinsight.com/0307-4358.htm

The Tiger
The Tiger Woods scandal: Woods scandal
a cautionary tale for event studies
Matthew Hood
Department of Finance and Economics, Texas State University-San Marcos, 543
San Marcos, Texas, USA
Received April 2011
Abstract Revised October 2011
Purpose – Tiger Woods suffered minor injuries and major scrutiny into his personal life due to a Accepted December 2011
suspicious car crash. Previous research suggests that his sponsors would be expected to suffer a
Downloaded by Texas State University San Marcos At 10:23 09 June 2016 (PT)

significant negative shock. The purpose of this paper is to determine if there was such a shock and to
clarify the role that the estimation and event windows have on measuring its significance.
Design/methodology/approach – The event study methodology is used for Tiger’s core sports
related sponsors: Electronic Arts (Tiger Woods PGA TOUR), Nike (Nike Golf), and Pepsi (Gatorade)
and his sponsors that are unrelated to sports: AT&T, Accenture, and Procter & Gamble.
Findings – The sponsors did not suffer a statistically significant negative cumulative abnormal
return. Even under liberal standards and for any event or estimation window that was considered.
However, the event windows and estimation windows greatly affected the test statistics.
Originality/value – Tiger Woods’ situation is, perhaps, the perfect case study for many papers that
have found a negative stock market reaction to a scandal. Yet, there sponsor’s stocks did not have a
significantly negative response under any condition studied. Also, the estimation window is found to
have a very large impact on the test statistics.
Keywords Stock returns, Golf, Celebrities, Sponsorship, Tiger Woods, Event study,
Cumulative abnormal returns
Paper type Research paper

Introduction
On Thanksgiving in 2009, Eldrick Tont “Tiger” Woods had much to be thankful for.
At 33 he had accumulated 14 professional majors in pursuit of Jack Nicklaus’s record
18. Earlier that year Sports Illustrated had estimated his annual earnings to be nearly
$100 million (Freedman, 2009) and Forbes speculated he might soon become a
billionaire (Pendleton, 2009). He was also married to a former model with two young
children. All of this success was expected of Tiger. His father, Earl Woods, had been
quoted years earlier: “Tiger will do more than any man in history to change the course
of humanity [. . .] He is the Chosen One” (Moskowitz and Wertheim, 2011). His
dominance on the course, his unique heritage, and careful control over his image
brought him unprecedented sponsorships.
Before the sun rose the next morning, a mysterious car crash led to a series of rapid
fire scandals that shredded his private, guarded persona. The scandals brought an end
to some of his endorsements and, based on an early version of a study by Knittel and
Stango (2010), news reports estimated his sponsors lost up to 12 billion dollars
(see, for example, the Huffington Post, 2009).
Managerial Finance
Vol. 38 No. 5, 2012
JEL classification – G14 pp. 543-558
q Emerald Group Publishing Limited
The author gratefully acknowledges Paul Brann, former colleagues at the University of 0307-4358
Southern Mississippi, and the referees for their curiosity and thoughtfulness. DOI 10.1108/03074351211217850
MF Mr Woods’ injuries turned out to be minor, but he refused to speak directly with the
38,5 police and the media in the wake of the accident. This led journalists to find other
avenues to uncover a story. They started with a recently published claim in The National
Enquirer that Tiger had an affair with Rachel Uchitel, which she denied. The next
woman linked to Tiger Woods was Jaimee Grubbs, who produced evidence of their affair
for US Weekly on December 2. Then, over the next two weeks more than a dozen women
544 came forward and his marriage to Elin was on the rocks.
Golf is a game that requires great mental concentration to make competitive decisions
under pressure. Although there was little to fear for Mr Woods’ physical ability to play,
the mental and emotional toll of the scandals on his legendary drive to practice and his
ability to focus was expected to have a negative effect on his 2010 season. Then, seeking
more privacy and attempting to straighten out his personal and family life, Tiger
Downloaded by Texas State University San Marcos At 10:23 09 June 2016 (PT)

announced his career would be temporarily interrupted. During this time, the public
perception was that his play would be affected to some degree and that the damage to his
image was monumental. It appeared to be mostly because of the damage to his image
that many of his sponsors reduced his role or ended their relationship with him.
We seek to understand the statistical significance of the stock market response for
his sponsors. His sponsorships at that time included three that were sports-related and
even used his creative contributions for Tiger-specific products. These three are
Electronic Arts (Tiger Woods PGA TOUR), Nike (Nike Golf’s TW), and Pepsi
(Gatorade’s Tiger Focus). Another three companies used Tiger heavily in their
marketing campaigns: Procter & Gamble (Gillette), Accenture, and AT&T. Other small
and foreign companies also had endorsement deals with Mr Woods: Tag Heuer, Upper
Deck, TLC Laser Eye Centers, and NetJets.
We research the stock market response on the six large companies trading on the
prominent American stock exchanges from the Tiger Woods scandal. Further, we
examine the importance of seemingly innocuous decisions made by researchers using
the event study methodology – stock selection, the event window, and the estimation
window. We unexpectedly find that the stock market response is insignificant for any
set of these decisions and that the p-values are very different for the statistical tests.
This leads us to draw two conclusions, the initial reports about the lost value by
Tiger’s sponsors should not have reached such large conclusions and the event study
methodology is sensitive to issues that are at the discretion of the researcher.

Celebrity endorsements
To understand the impact of celebrity endorser’s behavior on stock performance, we
turn to a large body of research primarily using event study methodologies. Agrawal
and Wagner (1995) report a positive stock market response when a company initially
hires a famous spokesperson. They also show that the use of celebrities to endorse
products is quite old; the first celebrity endorser was Lillie Langtry, an English actress,
in 1893 for Pears Soap.
Celebrity endorsers occasionally find themselves in troubling circumstances.
Sometimes it is their own fault, sometimes not. Till and Shimp (1998) find that negative
information about a spokesperson can damage product evaluation and that physical
injuries can limit an endorser’s effectiveness. They find that the stock market responds
quickly to these news stories because these events are truly unanticipated and very
newsworthy. They find statistically significant market reactions to undesirable events
and that the reactions are worse when the blame is high. Louie et al. (2001) find that high The Tiger
levels of culpability cause sponsors to lose value in the stock market. However, if the Woods scandal
unfortunate circumstance is not the result of poor behavior, for example an illness –
Lance Armstrong – or a victim of a crime – Nancy Kerrigan, the sponsors do not lose
value in the stock market. Sympathy for the athlete overrides the lost value in future
successful endeavors in competition. There is a study with the opposite conclusion in a
clear case of high culpability. Leeds (2010) studies the impact of Floyd Landis’s 545
surprising 2006 Tour de France victory and then equally surprising failed drug test on
his team’s sponsor, Phonak AG. He concludes that Phonak AG’s stock rose throughout
the news cycle. This leads credence to the old adage that all publicity is good publicity.
The stock market reaction is not the same for all types of businesses. Mahar et al.
(2005) examine stock market response to NASCAR race performance with an event
Downloaded by Texas State University San Marcos At 10:23 09 June 2016 (PT)

study methodology. They find that there is a positive relationship between stock car
performance and stock market performance for sponsors that market to consumers but
no such relationship for firms that market to businesses. They also find positive
returns for firms in the auto industry regardless of the sponsors’ customer group. The
finding that companies in the auto industry benefit more from racing performance is
similar to some of the findings of the papers by John M. Clark, T. Bettina Cornwell, and
Stephen W. Pruitt regarding sponsorships and stock prices. Pruitt et al. (2005) show
that companies in the consumer automobile industry benefit more than those that are
in unrelated industries from NASCAR sponsorships. In works studying several
popular American sports, Cornwell et al. (2005) and Clark et al. (2008) find that
congruent sponsors, those where the product is related to the activity on the field or
related to the activities of the viewing audience, benefit more from the announcement
of sponsorship arrangement.
From Mr Woods’ list of six sponsors at the time of the accident, three of them are
congruent. He uses and wears Nike products on the golf course, Gatorade (Pepsi) is a
sports drink, and Tiger Woods PGA TOUR (Electronic Arts) is a golfing video game
with more than 25 million copies sold at the accident. The stock market response
should be stronger for these types of companies than those that only rely on his image.
This branch of literature suggests, but not unequivocally, that the stock market
reaction to news of Tiger Woods’ marital infidelity will be negative. The news revealed
a high level of culpability in combination with some degree of reduced expected
winnings for 2010. Tiger was not a victim of circumstances and the sponsors market
directly to consumers. The news was dramatic, unanticipated, and continued for
weeks. Therefore, we expect that the response will start on the day of the accident,
November 27, 2009 and continue for a period of weeks. Even in an efficient market
there will be continuing downward pressure when embarrassing news follows
embarrassing news, as it did for Tiger.

Event study methodology


An event study methodology allows researchers to determine the abnormal return (AR)
and then determine if the AR is statistically significant. The event study methodology
originated in 1969 with a work by Fama et al. (1969) on the impact of stock splits on
stock prices. The cumulative abnormal return (CAR) is measured by first determining
the AR – that part of return which is not related to the normal movement of the stock
market – and then building the return from one period to the next (Binder, 1998):
MF ARt ¼ Rt 2 E½Rt jX t 
38,5 The AR, is the difference between the return, R, and the expected return, given the
informational content in independent variables, X. Most commonly, X is the return for
the general stock market, which is approximated by the S&P 500, M, and an intercept.
The expected return for the stock is determined by an ordinary least squares
546 regression using a predetermined period before the event:

R t ¼ a þ b M t þ 1t
dt , is the difference between the stock’s return and what has
Therefore, the estimated AR
been the stock’s typical response to the move in the stock market, Mt:
Downloaded by Texas State University San Marcos At 10:23 09 June 2016 (PT)

dt ¼ Rt 2 a^ 2 b^M t
AR

The estimated CAR for a date T days after the event is computed by multiplying all of
the daily gross returns up to that date to determine their cumulative effect on the stock
price:

Y
T
dT ¼
1 þ CAR dt
1 þ AR
t¼1

To determine the statistical significance of the ARs and CARs the standard error of the
regression, s^, for the estimation window is necessary. The standard error for the
estimated AR and estimated CAR are just as they would be for any out-of-sample point
estimate. We use N for the sample size of the regression from the estimation window,
MT for the stock market’s net return T days after the event, M  for the stock market’s
average daily return during the estimation window, and SM for the standard deviation
of the stock market’s daily returns during the estimation window. The mean and
standard deviation of the market’s return are estimated from the estimation window:
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
 2
SEðARdt Þ ¼ s^ £ 1 þ 1 þ ½M t 2 M
N N £ SM 2
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
2 2  2
dT Þ ¼ s^ £ T þ T þ T £ ½ðM T =TÞ 2 M
SEðCAR
N N £ SM 2

A test of statistical significance for an event study uses the CAR and its standard error.
Some of the variables are subject to the choices of the researcher. Foremost among
them are the sample size for the estimation window, N, and the event window, T. The
coefficients for the remaining variables are determined by the estimation window and
the event window. However, it should be noted that if N increases, ceteris paribus, the
standard error decreases and statistical significance increases. Also, that as T
increases, ceteris paribus, the standard error increases and statistical significance
decreases. It is clearly possible for an unscrupulous, unlucky, or uninformed research
to affect the statistical significance from an event study with these choices.
Hypotheses and data The Tiger
The purpose of this study is to determine if there is a statistically significant response Woods scandal
in the market value of Tiger Woods’ sponsors due to the infidelity scandal during the
2009 holiday season. His sponsors at that time included: AT&T, Accenture, Electronic
Arts (Tiger Woods PGA TOUR), Nike, Pepsi (Gatorade), and Procter & Gamble
(Gillete). Three of which (Electronic Arts, Nike, and Pepsi) are considered by Knittel
and Stango (2010) to be core sports related, or in the vernacular of Cornwell et al. (2005), 547
congruent:
H1. The sponsors of Tiger Woods will have a significantly negative CAR during
the infidelity scandal, with the strongest response for the congruent,
sports-related, companies.
Downloaded by Texas State University San Marcos At 10:23 09 June 2016 (PT)

The purpose of the event window is to capture the time frame the stock market
responds to the new information. In studies with the possibility of rumors leaking
before the announcement (Hood and Nofsinger, 2008) the event window begins before
the announcement. Frequently, news might percolate through the system and cautious
researchers continue the event window several days past the announcement. In the
case of Tiger Woods, the news all appears to be a surprise to the public, however the
stock market response might have started with the story in The National Enquirer
(Wednesday, November 25, 2009); the accident (Friday, November 27, 2009); or as other
researchers (Knittel and Stango, 2010) have suggested, the first full day of trading
(Monday, November 30, 2009)[1]. We consider that most of the details of Tiger’s affairs
surfaced in the first few weeks after the accident. Therefore, we will examine the stock
response for his sponsors from November 25, 2009 until December 18, 2009 to
incorporate the entire time frame of the scandal. Our standard is to begin the event
window with the accident, November 27, and end it when Tiger held a news conference
that admitted his infidelity and that he would be taking a hiatus from professional golf
on December 11:
H2. The event window will have a noticeable impact on the test statistics.
The purpose of the estimation window is to determine what the normal response is for
the stocks in question. Different windows will provide different estimates for beta and
different ARs. Shorter windows place a larger weight on more recent performance
which might be more relevant. However, longer windows might show more stability
with a more diverse sample. In this research our primary estimation window will use
the 250 trading days, roughly one year, to best fit with existent event study research
and we will explore windows that are one-fifth of this size, 50 trading days, and up to
five times this size, 1,250 trading days, to determine what impact the length of the
estimation window has on the test statistics:
H3. The estimation window will have a noticeable impact on the test statistics.
The closing stock price and dividend data for these six companies, plus the S&P 500
(ticker symbol: SPY) from November 2, 2004 to December 31, 2009 are gathered from
Bloomberg. The daily return for each of these seven entries is computed by dividing
the net change in the stock price plus any dividends by the previous close. When
sponsors are grouped together to form a portfolio, the portfolio’s return is an
equally-weighted average of the net return of the members of the portfolio.
MF Results: sponsor stock response
38,5 On the day of the accident the stock prices of all six of the sponsors fell. However, since
the average decline was only 1.1 percent and the S&P Index, SPY, fell 1.6 percent it was
not a particularly bad day for Mr Woods’ sponsors. The AR on that day is only negative
for Electronic Arts (Tiger Woods PGA TOUR video game) and Pepsi (Gatorade),
Figure 1 and Table I. An equally-weighted portfolio of the six sponsors had a positive
548 AR of 0.2 percent. Two weeks later, on the date that Tiger announced his temporary
break from professional golf, four of the six sponsors CAR was negative, all three
congruent sponsors – Electronic Arts, Nike, and Pepsi – and Procter & Gamble.
However, none of the moves were statistically significant. The closest any of them come
to a significantly negative CAR is Pepsi and Electronic Arts. Pepsi’s CAR is 2 5.3 percent
on December 18 which creates a p-value of 15 percent (15 percent of the time the stock
Downloaded by Texas State University San Marcos At 10:23 09 June 2016 (PT)

price would be expected to fall more than 5.3 percent over a three-week period).
The equally-weighted portfolio has a positive CAR until December 9. December 9 is
important because it is the date that Pepsi made a negative earnings announcement
(Knittel and Stango, 2010) and its stock price falls 2.7 percent. In the interest of
determining the range of values a researcher might find, we also consider limiting the
study to the three congruent, or core-sports-related firms. The CARs for an
equally-weighted portfolio of these three stocks is negative for the entire three-week
period. Yet, the closest it ever reaches is 21 percent on December 11. The first part of
H1 cannot be substantiated.
The sponsors that are not sports related – AT&T, Accenture, and Procter
& Gamble – have positive ARs and could have significantly outperformed the
congruent sponsors – Electronic Arts, Nike, and Pepsi – as previous research suggests
likely. The proper test for this question uses a t-test for the equality of two means with
independent samples[2]. Table II shows the results of this test when the variances are
assumed to be equal and when that assumption is relaxed.
On the day of the accident the congruent sponsors had an average AR
of 2 0.3 percent and the unrelated sponsors had an average of 0.7 percent. This
difference of 1.0 percent is significant at the 10 percent level for a lower-tail test of
significance. The CAR from this date through December 11 for the congruent sponsors
is 2 3.7 percent and it is 2.3 percent for the unrelated sponsors. This difference of
6.0 percent is significant at the 5 percent level. Although the congruent sponsors did
not statistically underperform expectations during the scandal, they did underperform
those of the unrelated sponsors.
It is clear that there is no statistically significant move from Tiger’s sponsors
coinciding with the scandal. However, the stock sample does make a significant
difference. In fact, one set of sponsors has a significantly worse return than another set.
Researchers should be well aware of the importance of proper stock selection in event
studies.

Results: event window


Two critical factors in all event studies is the determination of when an efficient market
begins to respond to news and how long it takes to accurately reflect that new
information. In MacKinlay’s (1997) review of the event study literature, he states that
the information for an earnings announcement could be analyzed with daily data and
that the event window must include the date of the announcement, customarily includes
7.0%
The Tiger
Woods scandal
Electronic Arts
Nike Accenture
Pepsi
AT & T 549
Accenture
Procter & Gamble

3.5%
Downloaded by Texas State University San Marcos At 10:23 09 June 2016 (PT)

AT & T

Pepsi

Procter & Gamble


0.0%

Nike

–3.5%

Electronic Arts

Figure 1.
–7.0% The CARs for all six
Nov. 27 Dec. 04 Dec. 11 publicly traded sponsors
Notes: The three sports-related firms – Electronic Arts, Nike, and Pepsi – are shown with of Tiger Woods from the
date of the accident
dotted lines and the other three – AT&T, Accenture, and Procter & Gamble – are shown through the next
with solid lines; the estimation window is 250 trading days two weeks
Downloaded by Texas State University San Marcos At 10:23 09 June 2016 (PT)

MF
38,5

550

sponsors
Table I.

Tiger Woods’ six


The CARs and the
p-values for the CARs for
November November December December December December December December December December December December December December December December
27 30 1 2 3 4 7 8 9 10 11 14 15 16 17 18

CARs for the Sponsors of Tiger Woods


Electronic
Arts (%) 20.9 21.9 2 2.6 2 3.1 23.8 26.3 26.0 23.8 2 4.1 24.1 26.3 25.1 2 4.3 22.4 22.6 21.8
Nike (%) 0.6 0.0 2 0.4 2 0.4 20.3 21.4 20.9 21.4 2 4.2 23.6 22.2 22.4 2 1.7 22.3 22.3 21.1
Pepsi (%) 20.7 21.0 1.0 1.4 0.3 1.8 2.4 1.7 2 1.1 21.4 22.5 23.3 2 3.1 23.7 24.1 25.3
AT&T (%) 0.9 0.5 0.6 1.3 2.6 2.5 4.0 3.6 3.2 3.6 4.3 4.0 2.7 2.4 2.2 2.2
Accenture
(%) 0.7 1.8 2.1 3.4 3.4 4.6 5.3 6.0 4.2 4.9 2.9 2.3 1.2 2.0 2.2 1.1
Procter
& Gamble
(%) 0.4 0.1 0.2 0.8 0.3 0.1 0.0 20.3 0.2 20.1 20.2 0.5 2 0.6 20.6 20.7 21.0
Portfolios
All six
sponsors
(%) 0.2 20.1 0.2 0.5 0.4 0.1 0.7 0.9 2 0.3 20.1 20.7 20.7 2 0.9 20.7 20.9 21.0
Congruent 3
sponsors
(%) 20.3 20.9 2 0.6 2 0.7 21.3 22.0 21.6 21.2 2 3.1 23.0 23.6 23.6 2 3.0 22.7 23.0 22.7
P-values for the CARs
Electronic
Arts (%) 37 32 30 29 27 18 21 32 32 33 26 31 34 41 41 44
Nike (%) 45 45 47 37 42 39 22 26 36 35 40 37 37 44
Pepsi (%) 30 29 38 37 28 23 25 22 20 15
AT&T (%)
Accenture
(%)
Procter
& Gamble 50 47 49 48 45 45 44 42
Portfolios
All six
sponsors
(%) 47 45 48 40 41 38 41 39 39
Congruent 3
sponsors
(%) 40 31 39 39 33 27 33 38 22 24 21 22 27 29 29 31

Notes: The Congruent 3 sponsors that market sports-related products are: Electronic Arts, Nike, and Pepsi; only p-values less than 50 percent are reported, which result from CARs less than
0 percent
a few days following the announcement, and might consider the day before the The Tiger
announcement to capture the effect of information leaking out before the public Woods scandal
announcement.
The beginning of the event window for the impact of the Tiger Woods scandal must
begin around the time of his car accident in the early morning hours of Friday,
November 27, 2009. Unlike an earnings announcement, the possibility of information
leaking out of an imminent wreck is not possible. However, two days earlier The 551
National Enquirer ran a story claiming he had an affair with Rachel Uchitel. This
suggests the stock market response on Wednesday may also be of interest. The date of
Tiger’s accident is also a light day for the market because of Thanksgiving. The
market closes early on the Friday after Thanksgiving and Knittel and Stango (2010)
surmise that the reaction to the accident began on Monday, November 30, 2009.
Downloaded by Texas State University San Marcos At 10:23 09 June 2016 (PT)

In the days following the accident, many of his affairs became public. He also pulled
out of golf tournaments and The New York Times reported that a sports doctor who
had treated Tiger was under investigation for providing athletes with human growth
hormones. Clearly the news swirling around Tiger Woods over a period of weeks
damaged his image and career. Therefore, we consider the daily ARs and CARs for
portfolios of Tiger Woods’ sponsors from Wednesday, November 25 to Friday,
December 18 in Table III.
Both of the portfolios, the three congruent sponsors and all six sponsors,
outperformed the market on the date of The National Enquirer story. The AR was
1.6 percent for the smaller portfolio and 0.7 percent for the larger portfolio. On the date of
the accident the portfolio of the three congruent stocks had an AR of 2 0.3 percent, but
the portfolio of all six stocks had another relatively good day. The first full day of trading
saw the first poor day for the portfolio of all six sponsors, its AR was 2 0.3 percent, and
the portfolio of congruent stocks was again negative, 2 0.6 percent. None of these daily
ARs are significant and both portfolios were actually in the top quintile for ARs on the
date of the story in The National Enquirer.
The only date with a significantly negative shock to these portfolios is December 9,
2009. The portfolio of three congruent stocks lost 2 percent and the portfolio of all six
sponsors lost 1.3 percent; each of these moves are significant for a one-tailed test at the
10 percent level. However, it appears that the cause of the significant moves is a poor

AR CAR
November 27
November 27 December 11

Averages Table II.


Congruent stocks (%) 20.3 23.7 The results from testing
Unrelated stocks (%) 0.7 2.3 the returns for Tiger
p-values Woods’ three congruent
Equal variances (%) 5.3 1.6 sponsors – Pepsi, Nike,
Unequal variances (%) 7.6 1.6 and Electronic Arts – are
less than that of his
Notes: The two days shown are the ARs on the first trading day after the accident, November 27, and non-sports-related
the CARs, November 27 to December 11; the returns are tested for the possibility that the congruent sponsors – AT&T,
stocks significantly underperformed the unrelated stocks assuming the variances are equal, even Accenture, and Procter
though the means are not, and relaxing this assumption & Gamble
MF
ARs and CARs
38,5 CAR – Wednesday
ARs (%) (%) CAR – Friday (%) CAR – Monday (%)
Date Congruent 3 All six Congruent 3 All six Congruent 3 All six Congruent 3 All six

November 25 1.6 0.7 1.6 0.7


552 November 27 20.3 0.2 1.2 0.9 20.3 0.2
November 30 20.6 20.3 0.6 0.6 20.9 20.1 20.6 2 0.3
December 1 0.3 0.2 0.9 0.9 20.6 0.2 20.3 2 0.0
December 2 20.1 0.4 0.8 1.3 20.7 0.5 20.4 0.4
December 3 20.6 20.2 0.3 1.1 21.3 0.4 21.0 0.2
December 4 20.8 20.2 20.5 0.9 22.0 0.1 21.7 2 0.0
Table III. December 7 0.5 0.6 20.0 1.4 21.6 0.7 21.2 0.6
The daily ARs and CARs
Downloaded by Texas State University San Marcos At 10:23 09 June 2016 (PT)

December 8 0.4 0.2 0.4 1.7 21.2 0.9 20.8 0.8


for an equally-weighted December 9 22.0 * 21.3 * 21.6 0.4 23.1 20.3 22.8 2 0.5
portfolio of Tiger Woods’ December 10 0.1 0.2 21.5 0.6 23.0 20.1 22.7 2 0.3
three congruent sponsors, December 11 20.7 20.6 22.1 20.0 23.6 20.7 23.4 2 0.9
Congruent 3 – Pepsi, December 14 0.1 0.0 22.0 0.0 23.6 20.7 23.3 2 0.9
Nike, and Electronic Arts December 15 0.6 20.3 21.4 20.2 23.0 20.9 22.7 2 1.1
and a portfolio with all December 16 0.2 0.2 21.2 20.0 22.7 20.7 22.4 2 0.9
six of his sponsors, all December 17 20.3 20.2 21.4 20.2 23.0 20.9 22.7 2 1.1
six – adding AT&T, December 18 0.3 20.1 21.2 20.3 22.7 21.0 22.4 2 1.2
Accenture, and Procter
& Gamble, in the days Notes: Returns that are statistically significant, for a lower-tail test are: *10, * *5 and * * *1 percent
and weeks after his car levels; the CARs are shown given different starting dates for the event window: the date of The
crash and subsequent National Enquirer story, November 25; the date of the accident, November 27; and the first full day of
scandals trading after the accident, November 30

earnings announcement from Pepsi, which fell 2.7 percent that day, not from news
related to Tiger Woods. With just this one exception, no other day had a significantly
negative shock for either portfolio.
Table III also reports the CARs considering the start date as Wednesday,
November 25; Friday, November 27; and Monday, November 30. Considering the most
likely starting date as November 27, the CAR for the more restrictive sample reached
its lowest point on December 11 at 2 3.6 percent. This is the date that Mr Woods
admitted his infidelity and announced he would be taking an indefinite leave from
professional golf. The CAR to that point has a p-value of 21 percent in a one-tailed test
of significance. Although the CARs are negative for the congruent portfolio throughout
the scandal, they are never significantly so. Therefore, statistical evidence that the
Tiger Woods scandal drove down the stock prices of his sponsors is lacking.
The most likely starting date of the Tiger Woods scandal begins with the car accident
early Friday morning. The National Enquirer story might have indirectly led to Tiger’s
accident, but the stock market response on that day is actually positive. Waiting until
Monday to begin the event window also yields less significant results. An efficient
market does not require a full day of trading to respond to news as public and pervasive
as an accident involving the richest athlete in the world, the p-values for the portfolio’s
CAR is most significant when it begins with the first trading day after the accident.
An equally-weighted portfolio of Tiger Woods’ core-sports-related sponsors –
Electronic Arts, Nike, and Pepsi – performed poorly in the days and weeks following
his car accident; however it was not a statistically significant move. The event window
dramatically influences the p-values of the test statistics. In fact, if a researcher were to The Tiger
isolate the date of The National Enquirer story, they could nearly reach the conclusion Woods scandal
that the scandal had a significantly positive impact on the performance of these stocks.

Results: estimation window


The choice of the estimation window is fundamental, but rarely discussed in detail, for
an event study. Event studies using daily data generally use an estimation window of 553
250 trading days before the event to 30 trading days before the event and this period
generally appears to be chosen arbitrarily (Aktas et al., 2007). The purpose of the
estimation window is to determine the typical response of the stock or portfolio
to the market during the event window. Trading days that occur several years before
the event are generally considered to have less informational content than those near
Downloaded by Texas State University San Marcos At 10:23 09 June 2016 (PT)

the event. Yet, longer estimation windows can create smaller standard errors and more
significant results.
Ending the estimation 30 trading days before the event is chosen to avoid
contaminating the estimation window with possible early effects coming from the
event. For the Tiger Woods scandal, there is no need to dismiss the 30 trading days
prior to the accident because it was not any more imaginable one week before the
accident as it was one year before the accident. All estimation windows in this study
end on Tuesday, November 24, 2009.
The estimation window for the previous sections of this study uses 250 trading
days; the daily returns from November 28, 2009 to November 24, 2009. The chosen
estimation window can have a nontrivial, but unintended, effect on the significance of
the returns. Therefore, we explore estimation windows that are one-fifth as long and
five times as long (50 trading days to 1,250 trading days) and find an interesting
pattern.
Arguably the two most important CARs are for the day of the accident – November 27
(A6-01 and C3-01), and the date of the minimum p-value, December 11 – the 11th trading
day (A6-11 and C3-11). Figure 2 shows the p-values of the CARs at these points for the
equally-weighted portfolios of all six sponsors and for only the three congruent
sponsors. No estimation window yields a significant result at even the 10 percent level of
significance. The p-values for the AR on the date of the accident range from 25.5 to
43.0 percent for the congruent sponsors and from 42.2 to 60.4 percent for all six sponsors.
The p-values for the CAR for the two weeks after the accident range from 10.4 to
27.3 percent for the congruent sponsors and from 18.5 to 43.1 percent for all six sponsors.
The minimum p-values for the AR on the date of the accident occur with a short
estimation window. In every case the p-values begin near their most significant levels
and move up erratically to their least significant levels shortly after one year. The
p-values for all six sponsors for the date of the accident then remain steady near
60 percent. The other three sets of p-values, though, steadily decline down near their
original levels. In fact, the p-value for the three congruent sponsors after 11 trading
days with an estimation window of 1,250 trading days is 12.0 percent – much closer to
the p-value of 50 trading days (10.4 percent) than it is to the p-value of 250 trading days
(20.6 percent).
This general pattern of inconsistent p-values when the estimation windows are
small continues further than would seem unlikely. Consider estimation windows
between 200 and 300 trading days. The smallest range in the p-values is 5.2 percent for
MF
38,5
60%

C3-01
554 C3-11
A6-01
A6-11
Downloaded by Texas State University San Marcos At 10:23 09 June 2016 (PT)

40%

20%

Figure 2.
The p-values for tests that 0%
the AR on the date of the 0 250 500 750 1,000 1,250
accident (A6-01 and C3-01) Notes: The two portfolios are all six of the major sponsors of Tiger Woods (AT&T,
and the CAR 11 trading Accenture, Electronic Arts, Nike, Pepsi, and Procter & Gamble) and the Congruent 3 of
days later (A6-11 and the major sponsors (Electronic Arts, Nike, and Pepsi); the number of trading days
C3-11) are significantly ranges from 50 to 1,250, approximately five years
negative
all six stocks for the first day of trading after the accident. The largest range, from The Tiger
13.8 to 27.3 percent, is for the three congruent stocks 11 trading days after the accident. Woods scandal
The other two have ranges between 9 and 10 percent. Dropping off a month from the
estimation window or adding a month can make a very large difference in the p-values
that are obtained.
Examining the p-values for the AR on the day of the accident and the CAR through
December 11, we find that no estimation window will yield statistically significant 555
results to conclude that the scandal negatively affected Tiger’s sponsors. However, we
do find reason to believe that the results from event studies are dependent upon the
estimation window.
The significance (or insignificance) of the test statistic is dependent upon the AR and
the standard error. Figure 3 shows the ARs for the day of the accident (A6-01 and C3-01)
Downloaded by Texas State University San Marcos At 10:23 09 June 2016 (PT)

and the CARs for the two weeks after the accident (A6-11 and C3-11). The ARs on the day
of the accident are roughly steady around 20.5 percent for the congruent stocks and 0.0
percent for all six stocks but the CARs for the 11 trading days following the accident are
inconsistent, even out to 500 trading days. They range from 2 4.7 to 2 2.9 percent and
from 22.0 to 2 0.5 percent for the three congruent stocks and all six sponsors,
respectively, before settling around 24.0 and 2 1.0 percent.
The shape and levels of the p-values (Figure 2) seem much more connected with that
of the standard error of the ARs (Figure 4). The standard errors for the ARs on Friday,
November 27, 2009 and for the CARs from that day until Friday, December 11, 2009
generally rise as the estimation window increases to their maximum levels at around
300 trading days. From that point onward they decay slowly. In the case of the CAR
over 11 trading days, the decline is from 5.3 to 3.4 percent for the portfolio of the three
congruent stocks and from 3.2 to 2.2 percent for the portfolio of all six sponsors.
The declines in the standard errors are just over 30 percent for the two portfolios from

1%

0 250 500 750 1,000 1,250

–1%

C3-01
C3-11
A6-01
A6-11
–3%

Figure 3.
The CARs for the day of
the accident and the 11th
day after the accident
–5%
MF 6%
38,5 C3-01
C3-11
A6-01
A6-11
556
4%
Downloaded by Texas State University San Marcos At 10:23 09 June 2016 (PT)

2%

Figure 4.
The standard errors for
the CARs for the day of the
accident and the 11th day
after the accident 0%
0 250 500 750 1,000 1,250

300 trading days to 1,250 trading days and cause the bulk of the declining p-values for
longer estimation windows.
After examining the AR and CAR and their standard errors, it is clear that the size
of the estimation window has a dramatic impact on the p-values, but there is no
estimation window that produces a significant result – even in the very open case of a
10 percent level of significance and a lower-tail test. However, there is evidence that the
choice of the estimation window can have dramatic effects on the statistics, even after
more than 200 trading days are considered.

Conclusions
Celebrity endorsements have been a popular mode of marketing in many industries,
especially for goods that are purchased by consumers. The sponsor’s objective is to tie
their name to that of the celebrity and put the good that is known about the celebrity
into the name of the sponsor. When a celebrity athlete behaves poorly off the field or
performs poorly on the field, the reputation of the sponsor is also on the line. Prior
research suggests that this response will be stronger in three scenarios:
(1) if the business markets directly to consumers rather than other businesses;
(2) if the product is related to the athlete’s performance or their fan’s activities; and
(3) if the athlete’s struggles are self-inflicted.

With all three of these conditions met for in the case of Tiger’s Woods infidelity
scandal we are surprised to find the stock market response is not significant. Since
Tiger Woods was the richest and one of the most respected athletes in the world this
finding is more surprising. Even more surprising is that different, but all reasonable,
event and estimation windows are capable of producing a wide range of p-values but The Tiger
still not capable of producing significant test statistics. Woods scandal
The only conclusion with statistically significant results in the examinations is that
the congruent stocks – Electronic Arts, Nike, and Pepsi – performed worse than the
unrelated stocks – AT&T, Accenture, and Procter & Gamble. (A portfolio of
the congruent stocks was insignificantly negative on its own and a portfolio of the
unrelated stocks was insignificantly positive on its own). 557
Early news reports circulated stories that the investors in these stocks lost up to
$12 billion because of the scandal. We conclude that the insignificance of these results
should cause future research to be more skeptical of large claims and more cautious in
valuing the endorsement of celebrity athletes.
Another noteworthy discovery in this work is the variability in the p-values because
Downloaded by Texas State University San Marcos At 10:23 09 June 2016 (PT)

of the estimation window. In one case the p-values ranged from 13.8 to 27.3 percent
over an estimation window of 200 to 300 trading days. Estimation windows from
50 trading days to more than 1,250 produce even more variety in the p-values but still
failed to generate any statistically significant results. However, the shape of the results
suggests that extrapolating out beyond 1,250 trading days might have led to a
statistically significantly result. Increasing the number of stocks in the portfolio might
be expected to dampen the effects of other decisions such as the length of the
estimation window, but it does not. The standard errors fall just over 30 percent for the
portfolio of six and the portfolio of three.

Notes
1. Thursday, November 26, 2009 was Thanksgiving Day. Holidays are generally light for news
and market activity. The New York Stock Exchange is closed all day Thursday and closes at
1:00 on Friday.
2. If the variances of the two groups are assumed to be identical the results are the same as
those coming from a regression using an indicator variable to distinguish between congruent
and incongruent stocks.

References
Agrawal, J. and Wagner, A.K. (1995), “The economic worth of celebrity endorsers: an event study
analysis”, The Journal of Marketing, Vol. 59 No. 3, pp. 56-62.
Aktas, N., de Bodt, E. and Cousin, J.-G. (2007), “Event studies with a contaminated estimation
period”, Journal of Corporate Finance, Vol. 13 No. 1, pp. 129-45.
Binder, J. (1998), “The event study methodology since 1969”, Review of Quantitative Finance and
Accounting, Vol. 11 No. 2, pp. 111-37.
Clark, J.M., Cornwell, T.B. and Pruitt, S.W. (2008), “The impact of title event sponsorship
announcements on shareholder wealth”, Marketing Letters, Vol. 20 No. 2, pp. 169-82.
Cornwell, T.B., Pruitt, S.W. and Clark, J.M. (2005), “The relationship between major-league
sports’ official sponsorship announcements and the stock prices of sponsoring firms”,
Academy of Marketing Studies, Vol. 33 No. 4, pp. 401-12.
Fama, E.F., Fisher, L., Jensen, M.C. and Roll, R. (1969), “The adjustment of stock prices to new
information”, International Economic Review, Vol. 10 No. 1, pp. 1-21.
Freedman, J. (2009), “The 50 highest-earning American Athletes”, Sports Illustrated, Vol. 111
No. 1, p. 20.
MF Hood, M. and Nofsinger, J.R. (2008), “Corporate PACs and the stock market: the case of the 2004
presidential election”, The Journal of Wealth Management, Vol. 11 No. 2, pp. 93-103.
38,5 Huffington Post (2009), “Tiger Woods Scandal: $12 billion fallout”, Huffington Post, December 28,
available at: www.huffingtonpost.com/2009/12/28/tiger-woods-scandal-12-bi_n_405228.
html (accessed March 2011).
Knittel, C.R. and Stango, V. (2010), “Celebrity endorsements, firm value and reputation risk:
558 evidence from the Tiger Woods scandal”, working paper, October 28, pp. 1-27.
Leeds, M.A. (2010), “Is bad news always bad? The impact of Floyd Landis’s rise and fall on
Phonak”, Applied Economic Letters, Vol. 17 No. 8, pp. 805-8.
Louie, T.A., Kulik, R.L. and Jacobson, R. (2001), “When bad things happen to the endorsers of
good products”, Marketing Letters, Vol. 12 No. 1, pp. 13-23.
MacKinlay, A.C. (1997), “Event studies in economics and finance”, Journal of Economic
Downloaded by Texas State University San Marcos At 10:23 09 June 2016 (PT)

Literature, Vol. 35 No. 1, pp. 13-39.


Mahar, J., Paul, R. and Stone, L. (2005), “An examination of stock market response to NASCAR
race performance”, Marketing Management Journal, Vol. 15 No. 2, pp. 80-6.
Moskowitz, T.J. and Wertheim, L.J. (2011), Scorecasting, Crown Archetype, New York, NY.
Pendleton, D. (2009), “Fame and Fortune”, Forbes, Vol. 184 No. 7, p. 44.
Pruitt, S.W., Cornwell, T.B. and Clark, J.M. (2005), “The NASCAR phenomenon: auto racing
sponsorships and shareholder wealth”, Journal of Advertising Research, Vol. 44 No. 3,
pp. 281-96.
Till, B.D. and Shimp, T.A. (1998), “Endorsers in advertising: the case of negative celebrity
information”, Journal of Advertising, Vol. 27 No. 1, pp. 67-82.

Corresponding author
Matthew Hood can be contacted at: mh91@txstate.edu

To purchase reprints of this article please e-mail: reprints@emeraldinsight.com


Or visit our web site for further details: www.emeraldinsight.com/reprints

View publication stats

You might also like