Most event studies rely on cumulative abnormal returns, measured as percentage changes in stock prices, as their dependent variable. Stock price reflects the value of the operating business plus non-operating assets minus debt. Yet, many events, in particular in marketing, only influence the value of the operating business, but not non-operating assets and debt. For these cases, the authors argue that the cumulative abnormal return on the operating business, defined as the ratio between the cumulative abnormal return on stock price and the firm-specific leverage effect, is a more appropriate dependent variable. Ignoring the differences in firm-specific leverage effects inflates the impact of observations pertaining to firms with large debt and deflates those pertaining to firms with large non-operating assets. Observations of firms with high debt receive several times the weight attributed to firms with low debt. A simulation study and the reanalysis of three previously published marketing event studies shows that ignoring the firm-specific leverage effects influences an event study’s results in unpredictable ways.
The Art of Decision-Making: Navigating Complexity and Uncertainty
Skiera bayer-schoeler-event-study-2018-05-20
1. 20.05.2018
What Should be the Dependent Variable in Marketing-
Related Event Studies?
Bernd Skiera, Emanuel Bayer, and Lisa Schöler
International Journal of Research in Marketing,
Vol. 34, Issue 3, 641-659,
(http://dx.doi.org/10.1016/j.ijresmar.2017.01.002)
2. 120.05.2018
Summary of Paper published in International
Journal of Research in Marketing
Open Access Availability: https://www.sciencedirect.com/science/article/pii/S0167811617300046?via%3Dihub
4. 320.05.2018
Basic Idea of Event Study
• Pioneered in accounting and finance
• Later used in other areas such as marketing
• Basic concept
• Stock price reflects true value of firm (i.e., discounted future cash flows)
• Financial markets are efficient
• Stock prices react to new (valuable) information
• Changes in stock price reflect value of unanticipated event (Fama 1970; 1991)
• Procedure: Collect sample of observations of one event type
• Step 1: Average value of event (compared to benchmark)
• Percentage return in stock price (usual approach but problematic)
• Absolute change in market capitalization (hardly used, not problematic)
• Step 2: Determinants of value of event
5. 420.05.2018
Current Practice in Event Studies
• Step 1: Calculate percentage return (CARSHV) due to arrival of new information (i.e.,
event)
• Derive percentage change in stock price
• Step 2: Use CARSHV as dependent variable and regress on characteristics of events
t
Stock
Price
Difference reflect value of event
(usually measured as percentage change in stock price)
Event
k
SHV
i it
t=1
CAR = ARSHV
it it itAR = R E(R )
Actual Stock Price
Expected Stock Price
(1) (2)
J
SHV
i j ij i
j=1
CAR = α + β CHAR + ε .(3)
7. 620.05.2018
Valuation Theory
Non-operating assets (NOA) and Debt
Purpose of NCB activities
• Realization of additional returns
• Employment of excess cash resources
NCB is responsible for
• Other income and expense
Operating Business (OB)
Purpose of OB activities
• Selling products (goods & services)
OB is responsible for
• Revenues
• Costs of goods sold (COGS)
• Operating expense $
• Operating assets = assets used by
firms in core business operations
• Property
• Plant & equipment
• Natural resources
• Intangible assets
• Non-Operating assets = play no
role in firm’s operations
• Excess cash
• Marketable securities
(commercial papers, money
market instruments)
• Debt
• Corporate bonds
Damodaran, A. (2006), “Damodaran on Valuation: Security Analysis for Investment and Corporate Finance”, New York: John Wiley & Sons.
Schulze, C., Skiera, B., & Wiesel, T. (2012), “Linking Customer and Financial Metrics to Shareholder Value: The Leverage Effect in Customer-Based
Valuation”, Journal of Marketing, 76(2), 17–32.
8. 720.05.2018
Different Parts of Shareholder Value (SHV)
Value of operating
business
• Value of assets
used in core busi-
ness operations
• Examples are
property, plant,
equipment, natural
resources and
intangible assets
+ NOB
Value of Non-
Operating Business
• Value of excess
cash and short-
term investments
• Not needed to run
operating business
Debt / Non-Equity
Claims
• Money generating
value for firm
• But not owned by
shareholders
• Borrowed at
interest
− Debt
* for simplicity, before applicable tax which would affect both sides of the equation similarly
Value to stock
owners*, also called
• Market
capitalization
• Value of Equity
=OB
Damodaran, A. (2006), “Damodaran on Valuation: Security Analysis for Investment and Corporate Finance”, New York: John Wiley & Sons.
Schulze, C., Skiera, B., & Wiesel, T. (2012), “Linking Customer and Financial Metrics to Shareholder Value: The Leverage Effect in Customer-Based
Valuation”, Journal of Marketing, 76(2), 17–32.
9. 820.05.2018
Examples of Value of Non-Operating Business
• Apple (end of 2015)
• $216 billion
• Microsoft (end of 2016)
• $103 billion
11. 1020.05.2018
Impact of Event on Different Parts of SHV
Operating Business Non-operating assets Debt+ -
SHV
=
Event
? ?!
12. 1120.05.2018
Different Types of Events
• Influence all part of SHV
• Ability to earn money with core
business operations
• Lead to re-evaluation of (relation
of) non-operating assets and debt
• Examples:
• Regulatory changes (e.g., change
in tax level, introduction of new
tax)
• Natural disasters (e.g., nuclear
accident, earthquake)
Typical Marketing-Related EventsTypical Events in Finance and Economics
• Only influence value of operating
business
• Examples:
• Firing of advertising agency
• New product’s visual appearance
• Change of name of firm
31. 3020.05.2018
New Dependent Variable for Marketing-Related
Event Studies
• Cumulative Abnormal Return on Value of Operating Business (CAROB)
• Link between CARSHV and CAROB
where
Example:
• LE = Leverage Effect
• Captures financial structure of firm
• Information on NOA and Debt is available in firms‘ financial statements
• 1% change in OB leads to LE% change in SHV
before
SHV
OB i
i
i
CAR
CAR
LE
OB SHV-NOA+DEBT DEBT-NOA
LE = = = 1+
SHV SHV SHV
100
OB=100; DEBT=30: LE = = 1.43
70
100
OB=100; NOA=40: LE = = =0.71
140
100 100
OB=100; NOA=10; DEBT=45.5: LE = = = 1.55
100+10-45.5 64.5
33. 3220.05.2018
Calculation of Leverage Ratio
• Compustat Data Items (CDI):
• DEBT = total debt
long-term debt payable within a year (dd1) +
long-term debt (dltt) +
value of preferred stock (pstk)
• NOA = non-operating assets
short‐term investments (ivst)
• SHV = shareholder value
price close - annual - fiscal (prcc_f) ×
common shares outstanding (csho)
OB SHV-NOA+DEBT DEBT-NOA
LE = = = 1+
SHV SHV SHV
34. 3320.05.2018
Data to Calculate Leverage Effect
VariablesData
• S&P Total Market Index
• 4,903 firms
• 17 years (1998-2014)
• 62,012 firm-year observations
• DEBT = total debt
• Compustat Data Item (CDI):
• long-term debt payable within a year
(dd1) +
• long-term debt (dltt) +
• value of preferred stock (pstk)
• NOA = non-operating assets
• short‐term investments (ivst)
• SHV = shareholder value
• price close - annual - fiscal (prcc_f) ×
• common shares outstanding (csho)
OB SHV-NOA+DEBT DEBT-NOA
LE = = = 1+
SHV SHV SHV
36. 3520.05.2018
Descriptive Analysis of Leverage Effect
• Average leverage effect across firms: 1.56 (SD=1.67)
• Value of Operating Business 56% higher than shareholder value
• Example:
• Value of Operating Business (OB): 100
• Debt: 35.7
• Non-Operating Assets (NOA): 0
• Ratio of 100 and 64.3 (=100-35.7): 1.56
• Cross-firm analysis
• Average leverage effect in highest ventile (top 5%): 2.67
• Average leverage effect in lowest ventile (bottom 5%): .84
• CARSHV of highest ventile firm receives three times (2.67/0.84 ≈ 3.2) higher weight
in calculation of average CARSHV than lowest ventile firm
37. 3620.05.2018
a) Histogram of Within-firm Variations of
Leverage Effects over Time
b) Histogram of Cross-Firm Variations of
Leverage Effects (1998-2014)
05
10152025
Percent
0 .5 1 1.5 2 2.5
Coefficient_of_Variation
0
10203040
Percent
.6 .8 1 1.2 1.4
Coefficient_of_Variation
Average value of variation coefficient: .25
SD of variation coefficient: .28
N= 4,461 firms (442 firms were dropped
because only 1 year of data was available)
Average value of variation coefficient: 1.21
SD of variation coefficient: .55
N=17 years
Within-Firm versus Cross-Firm Variation of
Leverage Effects
Distribution: Log-Logistic (scale parameter = 4.41, shape parameter = 1.22)
39. 3820.05.2018
Design of Simulation Study
Experimental factors Number of factor levels Factor levels
Leverage ratio 1 Random draw from log-
logistic distribution fit-
ted on real data (scale
parameter = 4.41, shape
parameter = 1.22)
Coefficient α, β1, β2 1 α=.1, β1=.2, β2=-.4
Variable x1 1 Uniform Distribution
[-10; +10]
Variable x2 1 Uniform Distribution
[-10; +10]
Number of Firm-instance
Observations (i.e., sample size)
2 Small: 100
Large: 500
Number of Experimental
Settings
2
Number of Replications 100
Number of Event Studies 1002=200
OB
i 1 i 2 iCAR = α + β x1 + β x2
40. 3920.05.2018
Results of Simulation Study
ResultsAims
• Step 1 of event study: Sign of CAROB and
CARSHV
• Step 2 of event study: Sign of coefficients
of regression
• Determinants of deviations between
event studies with CAROB and CARSHV
• Step 1 of Event Study:
• Difference in sign: 10.5%
• Type I error: 16.7% (Model with CARSHV
erroneously finds sig. difference from zero)
• Type II error: 28.6% (Model with CARSHV
fails to detect sig. difference from zero)
• Step 2 of Event Study:
• Type I error: 1.7% (Model with CARSHV
erroneously finds significant coefficient)
• Type II error: 9.0% (Model with CARSHV
fails to find significant coefficient)
• Deviations are particularly pronounced if
• Correlation between CAROB and
leverage effect is high
• Sample size is small
46. 4520.05.2018
Overview about Reanalyzed Event Studies
• Bornemann, Torsten, Lisa Schöler, and Christian Homburg (2015), "In the Eye of
the Beholder? The Effect of Product Appearance on Shareholder Value," Journal
of Product Innovation Management, 32 (5), 704-715
• Aim of event study: Analysis of Importance of product design decisions
• Aesthetic value (perceptions of the visual attractiveness of product)
• Ergonomic value (ability of product to communicate its utilitarian function)
• Symbolic value (ability of product to reflect the (desired) identity of its owner)
• Kulkarni, Mukund S., Premal P. Vora, and Terence A. Brown (2003), "Firing
Advertising Agencies: Possible Reasons and Managerial Implications," Journal of
Advertising, 32 (3), 77-86.
• Aim of event study: Analysis of Effect of firing of advertising agency
• Karpoff, J. M., & Rankine, G. (1994). In Search of a Signaling Effect: The Wealth
Effects of Corporate Name Changes. Journal of Banking & Finance, 18(6), 1027–
1045
• Aim of event study: Analysis of Effect of a firm’s name change
47. 4620.05.2018
Differences in Determinants of Cumulative
Abnormal Returns in Bornemann et al. (2015)
Variables Regressions
Model I Model II
CARSHV CAROB
Functional Product Advantage .41 *** .27 **
Aesthetic Value .33 .53 ***
Ergonomic Value .26 ** .29 **
Symbolic Value -.43 ** -61 ***
Aesthetic Value × Functional Product Advantage .32 ** .49 ***
Ergonomic Value × Functional Product Advantage .00 -.01
Symbolic Value × Functional Product Advantage -.28 ** -.42 ***
CV: Industry -.13 -.08
CV: Firm Size -.19 * -.28 ***
CV: Brand Familiarity -.41 *** -.25 **
R2
.49 .52
Adj. R2
.41 .45
N 83 83
• CAROB allows for detecting the expected (given accepted theory) value-creating effect of
a product’s aesthetic value
48. 4720.05.2018
Variables/
Fit Measures
Regressions
Model I Model II
CARSHV CAROB
Intercept -.01 -.01
SG4-IA .09 * .08
ROE-IA .02 .02
ROESG4 -.03 -.03
R2 .21 .20
Adj. R2 .14 .13
N 36 36
Differences in Determinants of Size of Abnormal
Return in Kulkarni, Vora, and Brown (2003)
Notes: * Significant at 5% level; ** Significant at 1% level.
SG4-IA is the industry-adjusted growth in sales from quarter 4 to quarter 1;
ROE-IA is the industry-adjusted return on equity;
ROESG4 is the interaction between ROE-IA and SG4-IA.
49. 4820.05.2018
Differences in Descriptive Statistics for CARSHV
and CAROB in Karpoff and Rankine (1994)
Mean SD n
Comparison
Value
t df
Total Sample
CARSHV .58% .05 110 0 1.33 109
CAROB -.01% .04 110 0 -.03 109
Name change announced in WSJ for the first time
CARSHV† 1.20% .04 57 0 2.16* 56
CAROB† .85% .03 57 0 2.01* 56
Name change announced in proxy statement before WSJ
CARSHV -.09% .05 53 0 -.14 52
CAROB -.93% .04 53 0 -1.58 52
* p < .05, ** p < .01, *** p < .001. SD = Standard deviation. df: degrees of freedom. † Non-parametric Wilcoxon signed-rank test rejects the null
hypothesis of zero abnormal returns at 10% level for both CARSHV (Z = 1.88, p < .10) and CAROB (Z = 1.67, p < .10); a generalized sign test also
rejects the null hypothesis of zero abnormal returns at the 10% level.
50. 4920.05.2018
Summary of Reanalysis of Previously Published
Marketing Event Studies
Bornemann,
Schöler, &
Homburg (2015)
Kulkarni, Vora,
& Brown (2003)
Karpoff &
Rankine (1994)
Average leverage effect 2.19 1.08 1.80
Average cross-firm variation coefficient
of leverage effect
.76 .24 .57
Share of firms with leverage effect <1 17% 24% 13%
Lowest value of leverage effect .66 .86 .40
Highest value of leverage effect 9.37 2.12 6.78
Ratio of highest to lowest value
of leverage effect
14.20 2.47 16.95
Percentage difference between
average CARSHV
and average CAROB
CARSHV
is
13.27% higher
than CAROB
CAROB
is 12.12%
lower than
CARSHV
CAROB
and
CAROB
differ in
sign
Correlation between CAROB
and
leverage effect
-.27 .14 .15
Substantive insight Insignificant ef-
fect turns into
significant effect
(second step of
event study)
Significant effect
turns into insig-
nificant effect
(second step of
event study)
CAR remains
insignificant but
differs in sign
(first step of
event study)
53. 5220.05.2018
Authors of the Study
Bernd Skiera
skiera@wiwi.uni-frankfurt.de
Emanuel Bayer
embayer@wiwi.uni-frankfurt.de
Lisa Schöler
lisa.schoeler@strategyand.de.pwc.com