Industrial Marketing Management 99 (2021) A1–A12
Available online 22 March 20210019-8501/© 2021 Elsevier Inc. All rights reserved.
Editorial
How to apply the event study methodology in STATA: An overview and a step-by-step guide
for authors
A R T I C L E I N F O
Keywords
Event study
Research methodology
Corporate announcement
Covid-19
STATA commands
A B S T R A C T
The event study methodology, which is gaining recognition in the business and marketing disciplines, is a
technique used to capture the impact of significant events and announcements at the firm level and country level.
Originating from the finance and economics disciplines, and being widely used in the finance literature, the
method has recently attracted the attention of business and marketing researchers, particularly in the aftermath
of Covid-19, which has adversely affected all kinds of businesses across the world. The event study methodology
can be implemented to measure the impact of a major corporate announcement (e.g. new product development)
or a significant event on corporate financial performance, profitability, and market valuation over a specific
event window, such as a few days (a short window) or a few years (a long window). In this article, we provide a
detailed explanation of the step-by-step procedure for implementing the event study methodology in STATA,
using Covid-19-related death announcements from the United States, France, Spain, Italy, China, and the United
Kingdom. We also provide STATA commands that can be used by researchers when implementing the event
study methodology.
1. Introduction
The classical event study methodology (hereafter ‘ESM’) is rooted in
finance, and its basic premise is based on the fundamental view that
capital markets reflect publicly available information on the firms’ stock
prices. Thus, the ESM measures the effects of particular corporate events
on a firm’s prospects and stock price movements by calculating the
abnormal returns. However, the ESM is rarely used outside the ac-
counting and finance disciplines. In view of this observation, the present
article extends the previous methodological papers published in the
Industrial Marketing Management journal (e.g. Ullah, Akhtar, & Zae-
farian, 2018; Ullah, Zaefarian, & Ullah, 2020; Lim, Ahmed, & Ali, 2019).
The purposes of the present article are twofold: (a) to provide an over-
view of the ESM as applied within the business-to-business (B2B) mar-
keting literature; and (b) to equip non-specialists with an understanding
of the ESM and its application in marketing research by providing a step-
by-step guidance on how to apply the method.
In the marketing discipline and at the micro level, ESM-related
corporate events might include firm announcements about new prod-
uct launches, mergers and acquisitions, new market entries, etc. or an-
nouncements made by other entities, such as governments, regulatory
bodies, and competitors (Sorescu, Warren, & Ertekin, 2017). A fine
example of a macro-level event in recent times is the Covid-19 global
pandemic, which has adversely affected many businesses around the
world. Business researchers need to understand the short-term and long-
term impact of major events, including macro-level events (e.g. Covid-
19) and micro-level (firm-specific) events.
A close analysis of recent articles in the Industrial Marketing Man-
agement journal reveals that authors have predominantly used survey-
based econometrics, time series, cross-sectional and panel data, quali-
tative interviews, case study methodologies, and standard regression
estimations to evaluate the relationships between variables of interest.
However, the field of business marketing could significantly benefit
from the application of the ESM. This approach enables researchers to
more accurately capture the financial impacts of firm-specific marketing
initiatives than conventional customer-oriented measures, such as
satisfaction. In addition, event studies are based on ‘objective’ forward-
looking financial market data that is free from the influence of man-
agers, as opposed to ‘subjective’ performance measures, which are prone
to biases that stem from management perceptions.
The ESM offers several advantages to researchers, including, firstly,
the ability to examine the impact of specific events on corporate
financial performance (Brown & Warner, 1980). In this regard, scholars
can empirically isolate and measure the impact of various events,
whether internal (e.g. new product announcements, major R&D in-
vestment announcements, appointments of senior executives, dividend
announcements, corporate press releases, etc.) or external (e.g. entries of
direct competitors, introductions of new laws, etc.), on the firm(s) under
observation (Lubatkin & Shrieves, 1986; de Mortanges & Rad, 1998;
Delattre, 2007; Sorescu et al., 2017). Secondly, by focusing on stock
prices, the ESM provides both an objective measure of firm performance
(Fama, Fisher, Jensen, & Roll, 1969) and an unambiguous assessment of
the impact of different corporate events on shareholder value (McWil-
liams & Siegel, 1997). Lastly, the ESM is a versatile analytical technique
Contents lists available at ScienceDirect
Industrial Marketing Management
journal homepage: www.elsevier.com/locate/indmarman
https://doi.org/10.1016/j.indmarman.2021.02.004
Industrial Marketing Management 99 (2021) A1–A12
A2
that permits authors to estimate the impact of corporate announcements
and events over short (Cowan, 1992) or long event windows (Brown &
Warner, 1985). The ESM thus makes it possible for researchers to un-
derstand the impact of specific corporate events on stock prices, market
valuation, and profitability over time periods ranging from just a few
days to several years.
In the B2B marketing research published in the Industrial Marketing
Management journal, only a handful of studies have applied the ESM.
These studies have used the ESM to assess the impact of various firm-
level announcements, such as: (a) assessing the impact of announce-
ments of additional internet-based channels of distribution (i.e. eChan-
nels) on the economic value added (EVA) and market value added
(MVA) (Cheng, Tsao, Tsai, & Tu, 2007); (b) assessing the impact of
merger announcements on marketing performance (Rahman & Lamb-
kin, 2015); (c) evaluating the impact of media announcements relating
to firms’ outsourcing on abnormal stock returns (Lee & Kim, 2010); (d)
assessing market reactions to brand alliance announcements (Cao &
Yan, 2017); (e) examining the impact of marketing alliance announce-
ments on the focal firm’s abnormal stock returns (Oh, Lee, & Kim, 2018);
(f) evaluating the impact of CEO endorsements (measured by the pres-
ence of a CEO quotation in a press release) of sales and marketing
leaders on firm performance (Vaid & Ahearne, 2018); and (g) measuring
the impact of announcements of new executives taking up marketing
and sales positions (Vaid, Ahearne, & Krause, 2020). The growing
number of B2B studies that used ESM in recent years signals the rele-
vance of this methodology for business marketing research. Neverthe-
less, very few scholars have attempted to introduce the ESM to the
marketing and business research communities. Sorescu et al. (2017)
carried out a comprehensive literature review on the ESM, and they offer
a good conceptual understanding of how the ESM can be implemented in
marketing research. We extend the work of Sorescu et al. (2017) by
introducing specific STATA commands that can be used when applying
the ESM in different research settings.
We identify the steps that can be used by researchers to implement
the ESM, and we demonstrate STATA commands that can be used by
researchers to compute the abnormal returns before and after the event
date. We also discuss various aspects of STATA codes that can be used to
determine the event window. The commands reported in this paper can
be applied to assess the impact of different firm-level and other macro-
level events on corporate performance.
Accordingly, we focus on the Covid-19 outbreak as a major macro-
economic event and demonstrate how to statistically capture the
impact of such an event on major markets around the world. The Covid-
19 outbreak serves as a useful reference point in this ESM paper, as it is a
significant event that has had a considerable impact on the performance
of businesses around the globe. Covid-19 has resulted in high volatility
in the financial and commodity markets on a scale that has not been
witnessed in recent history (Wigglesworth, 2020). The strict population
lockdowns introduced by many governments around the world have
also had unimaginable consequences for the consumer markets. Many
businesses, ranging from small retail enterprises to large high-street
stores, have experienced sudden losses of market shares (Romei,
2020). Millions of people have been put out of work whilst many busi-
nesses have undergone temporary or permanent closure, severely
threatening the survival of many economies (Carlsson-Szlezak, Reeves,
& Swartz, 2020; Gopinath, 2020). These consequences of the Covid-19
crisis make the ESM even more relevant in current marketing research.
Previously, the ESM has been more widely used in accounting and
finance research to examine the impact of various events on corporate
stock prices (e.g. Binder, 1998; Boyd, Chandy, & Cunha, 2010; Corrado,
2011; Ball & Brown, 2013). Although the use of the ESM in marketing
research has increased over the years (Beckers, van Doorn, & Verhoef,
2017), it is still conspicuously low when compared with the fields of
accounting, finance, and management (Sorescu et al., 2017; Das,
McNeil, Pouder, & Daly, 2020). In this regard, the objective of the
present paper is to encourage marketing researchers to consider
applying the ESM to analyse how various firm performance measures (e.
g. revenue, profitability, and customer perceptions) are impacted by
firm-level events such as news of R&D investment, new product releases,
or even company branding. We attempt to do this by providing readers
with a step-by-step account of the procedure to follow when conducting
ESM research. For demonstration purposes, we utilise a sample of 18
major global companies. These companies are drawn from six countries
that comprise some of the largest economies in the world: China, Italy,
Spain, France, the United Kingdom (UK), and the United States (US).
Finally, we believe that this article will serve as a handy manual for
researchers wishing to take advantage of the benefits that the ESM has to
offer. The contributions of this article are interdisciplinary in nature.
Thus, postgraduate students and early career researchers from a range of
social sciences backgrounds, as well as other non-specialists, could find
it useful. Journal reviewers assessing papers on studies that have
employed the ESM and contributors to the Industrial Marketing Man-
agement journal could also find it beneficial.
Although the ESM is suitable in interdisciplinary research, there are
some caveats, as with any other econometric approach. Firstly, the as-
sumptions behind the ESM may not fit all situations. For example, we
live in an imperfect world where stock prices may not always fully and
precisely reflect all of the available information pertaining to a com-
pany, whilst the ESM assumes that markets are always efficient. Sec-
ondly, estimating an ideal event window can be daunting initially,
although it becomes easier once an individual has become more
acquainted with the ESM. Thirdly, researchers must consciously choose
the most appropriate model (i.e. perform model selection) to estimate
the expected returns, as the choice of model has the potential to affect
the results, in terms of the size and significance of the abnormal returns.
Finally, event studies has limitation for event date clustering, which is
an extensively discussed problem in ESM literature (for example see,
Kolari & Pynnonen, 2011) for which researchers have developed test
adjustments. These test statistics have been developed and implemented
in the user-written Stata commands eventstudy2 and estudy (for e.g. see
Kaspereit, 2020).
The rest of the paper is organised as follows: Section 2 performs a
review of the existing ESM literature, Section 3 presents the step-by-step
procedure for implementing the ESM in STATA, and, finally, we provide
a succinct overview of the ESM in the conclusion.
2. The ESM
2.1. Origins, history, and development over the years
The origin of the ESM can be traced to the work of James Dolley in
the early 1930s, which sought to understand how securities prices
behaved following corporate announcements relating to stock splits (i.e.
when a company divides its existing shares into multiple shares) (see
Dolley, 1933). The ESM was adopted by various scholars over the next
few decades, when the technique underwent further refinement or (as in
MacKinlay, 1997) sophistication. Some of the notable contributors to
the development of the ESM include Myers and Bakay (1948), Barker
(1956, 1957, 1958), Ashley (1962), Ball and Brown (1968) and Fama et
al (1969) (see also MacKinlay, 1997; Corrado, 2011). For instance, the
work of Ray Ball and Philip Brown, which examined the impact of an-
alysts’ earnings forecasts on corporate income (see Ball & Brown, 1968),
is credited for suggesting the splitting of events into ‘good news or bad
news’ as a way of controlling the challenge of high variance in ESM
studies (Brown & Warner, 1985). Fama et al.’s (1969) influential ESM
work examined the adjustment of stock prices to stock splits, where they
observed the behaviour of security prices before and after releases of
new information concerning stock splits. Following their work, Fama
et al. (1969) came to be viewed in the literature as the originators of the
concept of the event window in the ESM (see, for instance, Ball & Brown,
2013). The studies by Ball and Brown (1968) and Fama et al. (1969) are
also observed in the literature as having the most significant influence
Editorial
Industrial Marketing Management 99 (2021) A1–A12
A3
on the contemporary ESM (see Brown & Warner, 1980; MacKinlay,
1997; Corrado, 2011; Ball & Brown, 2013). Over the years, the ESM has
been widely applied in the accounting, economics, and finance realms,
owing mainly to the ease of accessing financial data from databases such
as, most notably, the University of Chicago’s Centre for Research in
Security Prices (CRSP) (Binder, 1998; Corrado, 2011). Other notable
contributors to the ESM include Brown and Warner (1980, 1985), who
provide useful directions on how to conduct estimations using monthly
and daily data, respectively, and McWilliams and Siegel (1997), who
recommend a short event window of 1–2 days for unanticipated events.
Although we use Covid-19, an enduring and unprecedented event, as
the catalyst occurrence in our analysis, the event window that we have
selected for each country is based on the days when each of the studied
countries reported the highest number of Covid-19-related deaths dur-
ing the early stages of the first wave of the pandemic in 2020. This de-
cision was informed by prior research that has used enduring events
such as sponsorship deals (Tsiotsou & Lalountas, 2005) or product
placement in films (Wiles & Danielova, 2009). In these studies, the au-
thors judiciously used the date of announcement of the events as the
basis for their chosen event windows.
2.2. Key features, procedure, and properties
Before outlining the step-by-step procedure, we provide a discussion
of the underlying concepts and terminology used in the ESM.
2.2.1. Event definition
As pointed out in the preceding sections, the ESM focuses on exam-
ining the impact of corporate events (comprising new information and/
or announcements) on a range of firm performance measures, such as
stock prices and firm earnings (Ball & Brown, 1968; Fama et al., 1969;
Brown & Warner, 1980, 1985; McWilliams & Siegel, 1997). Examples of
such events, to mention a few, include announcements about stock splits
(Fama et al., 1969), mergers and acquisitions (MacKinlay, 1997), and
corporate bankruptcies (Jayanti & Jayanti, 2011). Other marketing-
related events that have been studied in prior literature include news
about launches / bad publicity / recalls / modifications of products (de
Mortanges & Rad, 1998), sponsorship announcements (Tsiotsou &
Lalountas, 2005), product placement in films (Wiles & Danielova, 2009),
and “appointment of a new CMO […] or an announcement made by a
competitor or a regulatory body that can impact the focal firm’s value (e.
g., an FDA drug approval)” (Sorescu et al., 2017, p. 186). Some useful
sources of information about events include “daily financial press, legal
publications, professional databases, company press releases/confer-
ences, publications by stock exchange authorities, and news agency
stories” (Delattre, 2007, p. 59).
2.2.2. Event date identification
Following the identification of an appropriate event, the next step
involves selecting the date when the event occurred. The event date is an
important feature of the ESM, as it forms the basis for evaluating the
impact of the observed event on firm value/returns (Brown & Warner,
1985; MacKinlay, 1997). The event date allows researchers to compare
firm returns before an event with returns subsequent to the news
reaching the market, in order to measure the abnormal returns earned
due to the analysed event (Armitage, 1995; Binder, 1998). Abnormal
returns refer to the actual ex-post return of a security (stock) over the
event window minus the normal return of the firm over the event win-
dow (MacKinlay, 1997). Thus, it is extremely important to ensure that
the precise date of the analysed event is identified to avoid the flawed
estimation of the associated abnormal returns. It is also not unusual for
some events to ostensibly exhibit multiple dates, such as where “an
executive conveys relevant information in an interview reported in the
business press or at a trade show” compared to when the firm formally
“issues a press release through services such as Dow Jones Newswires”
(Sorescu et al., 2017, p. 191). The presence of such leakages, as
McWilliams & Siegel (1997, p. 634) observe, makes it “difficult to
determine when traders became aware of the new information.” To
overcome this problem, it is often advised that users of the ESM should
use the first date when information about the analysed event reached the
market (Fama et al., 1969; McWilliams & Siegel, 1997; Sorescu et al.,
2017).
2.2.3. Data/sample selection
When selecting the data and sample for ESM analysis, it is important
to ensure that the data covers the entire event timeline (i.e. the esti-
mation window, event window, and post-event window). An illustration
of the event timeline is shown in figure 1.
Researchers can select the sample based on a range of criteria, such
as data availability (Brown & Warner, 1980) and membership of a
specific industry (MacKinlay, 1997). Lubatkin and Shrieves (1986) call
for judicious selection of the sample to ensure no other events fell within
the event timeline under consideration whilst cautioning that this could
decrease the sample size. When a researcher is confronted by the chal-
lenge of inadequate sample size subsequent to cleaning the data, they
can avoid this problem by using daily returns (Brown & Warner, 1985)
or weekly returns (MacKinlay, 1997), instead of monthly returns (see
also Brown & Warner, 1980; Lubatkin & Shrieves, 1986).
2.2.4. Event window
Depending on the sample size and the length of the event timeline,
the event window may comprise a few days, weeks or months before and
after the event date. The event window is an important feature of the
ESM, as it permits researchers to measure the impact of the analysed
event on firm returns. Whilst there exists no fixed number of days/
weeks/months that should form the length of an event window, it should
be kept relatively short to avoid the impact of unrelated events on the
post-event returns (Armitage, 1995; McWilliams & Siegel, 1997; Delat-
tre, 2007). Accordingly, it is important for researchers to use good
judgement in selecting a suitable event window.
2.2.5. Measuring abnormal returns
Abnormal returns represent the earnings that investors make over
and above their otherwise normal returns in the absence of the analysed
event (Lubatkin & Shrieves, 1986; Boehmer, Musumeci, & Poulsen,
1991). The analysis of abnormal returns (or ‘unexpected returns’, as per
Lubatkin & Shrieves, 1986) may be performed based on the daily
earnings or aggregated earnings realised during the selected event
window (McWilliams & Siegel, 1997). The aggregated earnings
approach is particularly useful where researchers wish to estimate the
abnormal returns for multiple securities (stocks) over time or where a
multiple-period event window is analysed (MacKinlay, 1997). It is not
the intention of the present article to go into details about the modelling
of abnormal returns, as this has been adequately covered in various
studies (see, for instance, Armitage, 1995; MacKinlay, 1997; McWil-
liams & Siegel, 1997; Binder, 1998). The following section discusses the
application of the ESM in prior marketing literature.
2.3. Use of the ESM in prior marketing literature
Despite the huge potential of the ESM in analysing the impact of
corporate events and news releases on firm performance, this technique
is still considerably underexploited by researchers outside the account-
ing and finance realms. Nonetheless, there have been a few encouraging
efforts of marketing studies using the ESM in the past. A select number of
these studies are reviewed below. The reviewed articles are restricted to
those focusing on B2B marketing research that have been published in
high-quality journals listed in the Australian Business Deans Council
(ABDC) and Academic Journal Guide (AJG) journal quality lists.
Bobinski and Ramirez (1994), using data obtained from the Wall
Street Journal (WSJ), applied the ESM to examine stock market reactions
to corporate advertising aimed at financial institutions. The authors
Editorial
Industrial Marketing Management 99 (2021) A1–A12
A4
established that financial-relations-inspired advertising led to height-
ened share turnover on the day preceding and during the introduction of
advertisements, although no rise in stock prices was documented. Also
employing WSJ data and the ESM, Hozier and Schatzberg (2000) report
that termination of contract with advertising agencies resulted in
decreased corporate accounting performance and share prices in the two
days leading to the event. In a study looking at the value of sponsorship
of motoring events to investors, Cornwell, Pruitt, and van Ness (2001)
examined data drawn from the CRSP, newspapers, and archival records.
They found that share prices increased during the post-sponsorship
announcement period, with no significant rises in prices before the
event announcement. This contradicts the findings of Cornwell, Pruitt,
and Clark (2005) who, using data obtained from the CRSP, LexisNexis,
and Factiva databases, report evidence of stock price increases around
the announcement of sponsorship of major sports leagues in the US.
Furthermore, some authors have documented a neutral impact of sports
sponsorships on corporate value, for instance Clark, Cornwell, and Pruitt
(2009), whose data was obtained from the CRSP, various US stock ex-
changes, and sports associations. The above studies show the variety of
databases used and the issues examined in prior literature, as well as the
mixed nature of the results reported.
Accordingly, Swaminathan and Moorman (2009) investigated the
value emanating from alliance marketing and report that the practice
had a positive impact on shareholder value around the period of the 230
announcements studied. They used both the market model and the (four-
factor) Fama–French model to calculate the abnormal returns. Boyd
et al. (2010), on the other hand, used the market model to measure the
impact of CMO appointments on firm value and report that 46% of firms
had positive responses as reflected in their share prices, whilst 54% of
firms exhibited negative share price reactions. Similarly, Wiles, Morgan,
and Rego (2012) utilised the market model to analyse stock market re-
actions to brand acquisition and disposal, and their findings suggest a
positive impact on shareholder value. They further explain that they
were constrained from using the Fama–French model due to the lack of
Fama–French three-factor or four-factor data for the non-US listed firms
included in their sample. Researchers conducting studies in non-US
contexts, such as in emerging markets, may also find the market
model easily applicable for measuring abnormal returns in similar ESM
studies. It is also important to bear in mind that studies seeking to
measure abnormal stock returns can only do so using publicly listed
firms (see Fang, Lee, & Yang, 2015).
Besides, Kalaignanam and Bahadir (2013) examined stock market
reactions to corporate brand name changes and business restructuring.
Their results show that the two events had the greatest positive impact
on firm value during the time period of two days before and two days
after the announcements of the events (i.e. t1 ˆ 2 to t2 ˆ ‡2). In
addition, Homburg, Vollmayr, and Hahn (2014) studied the value
relevance of corporate distribution channel expansions, where they
observed significant abnormal returns occurring one day before the
event announcement, as well as on the day of the announcement (i.e. t1
ˆ 1 to t2 ˆ 0). Lastly, Fang et al. (2015) analysed the impact of an-
nouncements concerning product co-development in the biotech and
pharmaceutical industries and report evidence of abnormal returns in
the period between two days before and one day after event an-
nouncements (i.e. t1 ˆ 2 to t2 ˆ‡1). These studies show that abnormal
returns may occur at any point around the analysed event, and it is thus
the duty of researchers to apply good judgement in identifying an
appropriate event window. This can be done by first selecting various
(longer) windows (e.g. -10/15/20/25 days to ‡10/15/20/25 days) and
then testing the significance of the selected windows with the t-test and
z-test statistics (Brown & Warner, 1985). Where researchers face chal-
lenges specifying their models, they can substitute the parametric tests
(t- and z-statistics) with nonparametric rank procedures for assessing the
statistical significance of the observed stock price reactions (see Cor-
rado, 1989; Cowan, 1992).
From the above discussion, we can see how various scholars have
used the ESM in B2B marketing research to understand the impact of
various B2B marketing activities on stock prices. Whilst these studies
provide interesting findings concerning the potential benefits that
managers can bring to their firms in terms of increased shareholder
value, very little is known about the potential impact of B2B marketing
activities for firms that are not formally listed in organised stock mar-
kets. Accordingly, and as the majority of businesses in many countries
are privately owned, researchers need to consider how the ESM could be
applied in the context of unlisted firms. In the absence of listed share
prices for privately owned companies, an alternative approach would be
to compute the market value for shares of unlisted firms. Several
methods have been proposed in the literature for determining the value
of shares of privately held firms (see, for instance, Kantor & Pike, 1987a,
1987b). Besides, researchers may also combine the ESM with primary
data collection methods such as surveys and interviews with managers
in order to understand whether and/or how the timing of major
corporate announcements is predetermined. This is especially important
considering the scant literature on whether managers really consider
factors relating to the prevailing market value of their firms’ equity (i.e.
market capitalisation) before disclosing news of major corporate events.
In the following section, we report the step-by-step procedure with
generic STATA commands that can be used by researchers when
implementing an event study approach. We also provide a succinct
overview of the procedures and relevant STATA codes in the Appendix.
3. Step-by-step procedure for the ESM
3.1. Step 1: Identifying the event
The first step is to determine the event to examine and to collect the
required data about firms that have been affected by the event. The
other requirements include the collection of data to capture the
announcement date (day 0) of the main event (e.g. announcements of
dividends, announcements of earnings, and product launch events) and
the stock prices of all the affected companies before and after the event
(e.g. from 90 days to ‡90 days).
In this study, we consider Covid-19 as the main global event and
explore the economic impact of this event on the major capital markets
around the world. We provide actual examples in each step and consider
the highest number of deaths in a day related to Covid-19 as the core
event (day 0) under examination. Table 1 illustrates the six countries
sampled, which have been strongly affected by Covid-19, along with
overall data from Europe and the rest of the world.
Table 2 shows that the data was collected from two different sources.
Information on event dates was collected from the Our World in Data
website (https://ourworldindata.org), whilst stock data was derived
from the Bloomberg database. We used two databases because Our
World in Data reports relevant event dates (e.g. the highest number of
deaths in a day). The website is managed by the University of Oxford,
Table 1
List of countries and event dates considered in our event study analysis.
Country name Event date Highest number of deaths in a day
related to Covid-19
United States 16/04/2020 4920
France 04/04/2020a 2004
Spain 03/04/2020 950
Italy 28/03/2020a 971
China 17/04/2020 1290
United Kingdom 22/04/2020 1172
World 16/04/2020 10,520
Europe 04/04/2020 5139
a These dates fell on weekends; therefore, we have used the next trading day to
calculate the stock market returns. The event date is when the highest number of
deaths in a day was reported. The table also provides the highest number of
deaths reported on the event date.
Editorial
Industrial Marketing Management 99 (2021) A1–A12
A5
which reports on Covid-19 cases. The Bloomberg database was used to
collect stock return dates, stock returns, and stock indices from all the
countries.
After the identification of an event date for each country (i.e. the
highest number of deaths in a day related to Covid-19), we selected three
companies from each country to demonstrate how to apply the ESM.
Table 3 presents the names of the companies selected from each country.
We selected the three companies from each country based on highest
market capitalisation because these companies are highly valued listed
companies in their respective stock markets. Further, the chosen coun-
tries’ economies have been strongly affected by the pandemic, and the
stock market data was conveniently accessible for these economies.
3.2. Step 2: Selection of estimation, event, and post-event windows
Fig. 1 illustrates the estimation window, which helps to examine the
normal returns in an event study. The market model is useful in event
study analysis, as it observes the abnormal returns on the event day,
examines the stock returns, and compares them to the average returns
(MacKinlay, 1997). The stock returns are regressed on the market
returns to measure the association between the stock price and the stock
index. The second step requires us to determine the period over which
stock prices of the sample companies included in this event. This is
called the ‘event window’ (shown in Fig. 1).
Our event window is based on the unique event dates reported in
Table 1. Fig 2 provides an overview of the number of deaths and the
cummulative abnormal return in sample countries.
3.3. Step 3: Estimation of parameters
Our focus is now based on estimations of the main parameters that
will provide us with the expected returns during the event period. For
instance, estimations of expected returns through the market model
require the alpha (y-intercept) and beta (slope) of the stock prices over
the estimation window (e.g. for 120 to 31 days). Researchers can
amend the number of days based on their requirements, and both short
and long event windows are commonly used in the ESM literature (see,
for example, MacKinlay, 1997). The estimation window is based on
earlier days than the actual event window. It is useful to calculate the
systematic risk of the stock market to help us with the regression
analysis.
3.4. Step 4: Data cleaning and computing the event and estimation
windows
We expect that users will already have data with the event date,
which in our analysis we called “date”, and company identifier, which
we called “company_id”. Next, we must make sure that our estimation
window is conducting analyses on accurate observations. Therefore, we
form a variable, “dif”, that will count the number of days from the event
date (day 0).
We use the generic STATA codes relating to the “dif” command to
calculate the trading days or calendar days. Please refer to Table 4 (a-i).:
Table 4
(a-i): Event study STATA code for the trading days.*
Code Explanation
sort company_id date This command is used to sort our data by
“company_id” and “date”
by company_id: gen datenum ˆ _n This command creates a new column called
“datenum” to number all dates (e.g. 01/01/
2020 as 1, 02/01/2020 as 2, and so on). It is
useful to further identify the event dates from
the date column
by company_id: gen target ˆ
datenum if dateˆˆevent_date
egen td ˆ min(target), by
(company_id)
drop target
This command generates the new column
“target” to match with the event date. It
identifies the target date of our event as 1 and
the remaining dates as 0. The command targets
the event date with “datenum” column
gen dif ˆ datenum-td Finally, the “dif” command calculates the
number of days from the event date (i.e. the
pre-event date difference in days from the
actual event date and the post-event date
difference in days from the actual event date)
Table 2
List of variables (datasets used for the event study).
Event date Date of event https://ourworldindata.org/coronavirus
Stock data Stock return date Bloomberg
Stock returns Bloomberg
Stock index return Bloomberg
The two datasets were collected from different sources. The Covid-19 event
dates (the day with the highest number of deaths) were collected from the
website Our World in Data. The stock data, including stock return dates, stock
returns, and stock indices from all the countries, were collected from the
Bloomberg database.
Table 3
Information about the sample countries and companies used in the event study
analysis.
List of companies selected for the Covid-19 event study
United States Amazon
Facebook
Apple
United Kingdom
BHP Group Plc
Tesco Plc
Unilever Plc
China
Bank of China Co. Limited
Agricultural Bank of China
Industrial and Commercial Bank of China
France
BNP Paribas
L’Oreal
Sanofi
Italy
Ferrari
ENEL
ENI
Spain
Banco Santander
Iberdrola
Industria de Diseno Textil
* The general STATA codes are inspired by the Princeton University website i.
e. https://dss.princeton.edu/online_help/stats_packages/stata/eventstudy.ht
ml.
Editorial
Industrial Marketing Management 99 (2021) A1–A12
A6
Or Please refer to Table 4 (a-ii)
Table 4
(a-ii): Event study STATA code for the calendar days.
Code Explanation
gen dif ˆ date-
event_date
We can apply the calendar days command only when we are
using the calendar days in a month instead of trading days.
This is an alternative “dif” command that is used to provide us
with the number of days from the event date (i.e. the pre-event
date difference in days from the actual event date and the post-
event date difference in days from the actual event date)
In the next step, we identify the minimum number of observations
required for the pre- and post-event dates, along with the minimum
number of observations before the event window for measuring the
estimation window. For example, we are using two days for the pre- and
post-event dates (five days in the event window) and 30 to 60 days
for the estimation window.
We apply the STATA codes to identify the minimum number of ob-
servations required for the estimation window and the event window.
Please refer to Table 4 (b):
Table 4
(b): Event study STATA codes for the pre- and post-event windows.
Code Explanation
by company_id: gen event_window ˆ 1 if
dif 2 & dif 2
This command is used to create the pre-
and post-event windows. The pre-event
window is 2 days and the post-event
window is ‡2 days from the event date
(0)
egen count_event_obs ˆ count
(event_window), by(company_id)
This command is used to count the total
number of observations for the event
window. Overall, five days are used here:
2 days (pre-event), 0 (the event day)
and ‡ 2 days (post-event)
by company_id: gen estimation_window ˆ
1 if dif < 30 & dif 60
We also need an estimation window for
the regression analysis. The estimation
window is important to calculate the
systematic risk of the market and help us
to run the regression analysis. We
develop an estimation window of 30
days (i.e. outside the pre-event window
days)
egen count_est_obs ˆ count
(estimation_window), by(company_id)
replace event_window ˆ 0 if
event_windowˆˆ.
replace estimation_window ˆ 0 if
estimation_windowˆˆ.
This command is used to count the total
number of observations for the
estimation window
The method for measuring the event and estimation windows is
identical. The STATA code recognises which companies are lacking an
adequate number of observations. Therefore, first, we develop a variable
that equals 1 if the observation is within the specified days. Second, we
construct another variable that counts the number of observations
within each “company_id”. Lastly, missing values get a value of ‘0’ in our
data analysis. Please refer to Table 4(c-i) and Table 4 (c-ii).
Table 4
(c-i): Event study STATA code for determining companies with insufficient
observations in the event window.
Code Explanation
tab company_id if
count_event_obs < 5
We use the tab command to check for any inadequate
numbers of observations within the event window
before performing the regression analysis. In our
model, where the event window is five days, the tab
command considers the number of observations to be
(continued on next column)
Table 4 (continued )
Code Explanation
inadequate if there are fewer than five. It is important
to have a sufficient amount of data to run the
regression analysis
Table 4
(c-ii): Event study STATA code for determining companies with insuf-
ficient observations in the estimation window.
Code Explanation
tab company_id if
count_est_obs < 30
We use the tab command to check for any inadequate
numbers of observations within the estimation window
before performing the regression analysis. The criterion
to determine inadequate numbers of observations is
fewer than 30 (based on the estimation window of 30
days). It is important to have a sufficient amount of data
to run the regression analysis
After using the “tab” command, we have identified a list of com-
panies (“company_id”) that do not have adequate numbers of observa-
tions within the event and estimation windows and thus lack the total
number of observations required. Therefore, we will exclude these
companies by using the STATA command. Please refer to Table 4 (d-i)
and Table 4 (d-ii):
Table 4
(d-i): Event study STATA code to drop any companies with insufficient obser-
vations in the event window.
Code Explanation
drop if count_event_obs
< 5
This command will drop any companies with fewer than
five observations within the event window
Table 4
(d-ii): Event study STATA code to drop any companies with insufficient
observations in the estimation window.
Code Explanation
drop if count_est_obs
< 30
This command will drop any companies with fewer than 30
observations in the estimation window
3.5. Key features of the ESM
The main goal of an event study is to examine stock price reactions to
event announcements. In practice, the ESM has been applied for two
main objectives. First, it has been used to measure the null hypothesis
that the market efficiently integrates information (e.g. Fama et al.,
1969). Second, based on the efficient market hypothesis, the ESM has
been applied to measure the influence of events on firm value with
respect to publicly available information. The ability to measure swift
reactions in stock prices is possibly the most attractive feature of the
ESM. Another advantage of the methodology is that it can be used to
measure the expected value of a firm after public corporate
announcements.
3.6. Regression analysis
After cleaning the data, we carried out appropriate regression anal-
ysis. Initially, we analysed and estimated the normal performance. We
computed the regressions for individual companies separately by uti-
lising the data within the estimation window, and we measured the
alpha (the intercept) and beta (the coefficient of the independent vari-
able). Moreover, we also applied these regression equations to predict
the normal performance during the event window. We used the STATA
commands for the regression analysis. Please refer to Table 4 (e):
Editorial
Industrial Marketing Management 99 (2021) A1–A12
A7
Table 4
(e): Event study STATA code for the regression analysis.
Code Explanation
gen predicted_return ˆ . After cleaning our dataset, we are ready to
run the regression analysis. Therefore, in
the initial stage, we use this command to
evaluate the predicted returns.
The predicted returns (or expected returns)
are useful for the calculation of the
abnormal returns
egen id ˆ group(company_id) This command is useful if more than one
event is considered for one company. In our
dataset, we have considered only one event
for one company. However, researchers can
consider multiple events for one company
and group the events by using this
command
reg ret market_return if idˆˆ`i’ &
estimation_windowˆˆ1
predict p if idˆˆ`i’
replace predicted_return ˆ p if
idˆˆ`i’ & event_windowˆˆ1
drop p}
This regression command is applied to
predict the normal performance during the
event window. We also use this regression
command to measure the alpha and beta.
We regress the stock returns and market
returns within the estimation window
3.7. Calculating abnormal and cumulative abnormal returns around the
event dates
In the next step, we measure the abnormal and cumulative abnormal
stock returns. The daily abnormal return is calculated by subtracting the
predicted normal return from the actual return for each day in the event
window. Moreover, the cumulative abnormal returns are the sum of the
abnormal returns from the event window.
To evaluate the event’s impact, we need to examine the abnormal
returns. Therefore, it is relevant to calculate the abnormal returns in an
event study by calculating the difference between the actual returns and
the predicted returns. Cumulative returns are basically the summation of
abnormal returns in the event window, or an accumulation of abnormal
returns that allows us to observe the impact of the event. We used the
STATA commands to calculate the abnormal and cumulative abnormal
returns. Please refer to Table 4 (f).
Table 4
(f): Event study STATA code for calculating the abnormal returns and cumula-
tive abnormal returns.
Code Explanation
sort id date
gen abnormal_return ˆ ret-
This command is used to calculate the
abnormal returns and cumulative
(continued on next column)
Table 4 (continued )
Code Explanation
predicted_return if event_windowˆˆ1
by id: egen cumulative_abnormal_return ˆ
total(abnormal_return)
abnormal returns. It is important to
recognise any abnormal returns during
the event day. Abnormal returns are
calculated as the difference between the
actual returns and the predicted
returns. Cumulative returns are
calculated by summing the abnormal
returns during the event window
In the following step, we check the level of significance of the
abnormal returns.
3.8. Level of significance – Testing
We calculate the t-test statistic to confirm the following:
H0 ˆ abnormal return for each stock ˆ 0.
TEST ˆ ……ΣAR†=N †=…AR SD=sqrt…N† † (1)
AR ˆ abnormal return.
AR_SD ˆ abnormal return standard deviation.
Our conclusion is based on the absolute value of the t-test statistic.
For example, at the 5% significance level, if the absolute value of the t-
test statistic is greater than 1.96, then we reject the null hypothesis.
Please refer to Table 4 (g) for measuring t-test results
Table 4
(g): Event study STATA code for measuring the t-test results.
Code Explanation
sort id date
by id: egen ar_sd ˆ sd(abnormal_return)
gen test ˆ (1/sqrt(number of days in event
window)) * (cumulative_abnormal_return
/ar_sd)
list company_id cumulative_abnormal_return
test if difˆˆ0
This command is used to examine
the t-test results and identify the
significance level. We use eq. 1 to
calculate the t-test statistic. We
reject the null hypothesis if the
results are significant
4. Findings and discussion
We have implemented the ESM to analyse the impact of the Covid-19
pandemic by using the above-mentioned step-by-step STATA procedure.
Our sample includes six countries: the US, the UK, China, France, Italy
and Spain, with the three companies selected from each country shown
in Table 3 (analysed in Panel A of Table 5). Moreover, we have analysed
overall world data and European data using the ESM (Panel B of
Table 5).
As shown in Table 5, our event date is ‘day 0’, which represents the
Fig. 1. Illustration of an event timeline (adapted from De Jong, 2007).
Editorial
Industrial Marketing Management 99 (2021) A1–A12
A8

0
2000
4000
6000
0
0.1
0.2
0.3
0.4
0.5
-10 -8 -6 -4 -2 0 2 4 6 8 10 N
u
m
b
er
o
f
D
ea
th
s
C
u
m
u
la
ti
v
e
A
b
n
o
rm
al
R
et
u
rn
s
Event Window
Cumulative Abnormal Returns of
Amazon and the Number of
Deaths in the US
Cumulave Abnormal Returns Number of Deaths
0
500
1000
1500
0
0.1
0.2
0.3
-10 -8 -6 -4 -2 0 2 4 6 8 10 Nu
m
be
r o
f D
ea
th
s
Cu
m
ul
a
ve
A
bn
or
m
al
R
et
ur
n
Event Window
Cumulative Abnormal Returns of
Tesco and the Number of Deaths
in the UK
Cumulave Abnormal Return Number of Deaths
0
200
400
600
800
1000
1200
1400
-0.05
-0.04
-0.03
-0.02
-0.01
0
0.01
-10 -8 -6 -4 -2 0 2 4 6 8 10
Nu
m
be
r o
f D
ea
th
s
Cu
m
ul
a
ve
A
bn
or
m
al
R
et
ur
n
Event Window
Cumulative Abnormal Returns of
Bank of China Co. Limited and
the Number of Deaths in China
Cumulave Abnormal Return Number of Deaths
0
500
1000
1500
2000
2500
-0.12
-0.1
-0.08
-0.06
-0.04
-0.02
0
0.02
0.04
-10 -8 -6 -4 -2 0 2 4 6 8 10
Nu
m
be
r o
f D
ea
th
s
Cu
m
ul
a
ve
A
bn
or
m
al
R
et
ur
ns

Event Window
Cumulative Abnormal Returns of
Sanofi and the Number of Deaths in
France
Cumulave Abnormal Returns Number of Deaths

Note: all the above figures report the number of deaths and abnormal returns for sample countries
0
200
400
600
800
1000
1200
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35
0.4
-10 -8 -6 -4 -2 0 2 4 6 8 10
Nu
m
be
r o
f D
ea
th
s
nruteR la
mronbA evitalu
muC
Event Window
Cumulative Abnormal Returns of ENI and
the Number of Deaths in Italy
Cumulave Abnormal Return Number of Deaths
0
200
400
600
800
1000
-0.2
-0.15
-0.1
-0.05
0
-10 -8 -6 -4 -2 0 2 4 6 8 10
Nu
m
be
r o
f D
ea
th
s
Cu
m
ul
a
ve
A
bn
or
m
al
R
et
ur
n
Event Window
Cumulative Abnormal Returns of
Iberdrola and the Number of Deaths in
Spain
Cumulave Abnormal Return Number of Deaths
Fig. 2. Graphical illustrations of the number of deaths alongside the cumulative abnormal returns in the sample countries.
Note: all the above figures report the number of deaths and abnormal returns for sample countries
Editorial
Industrial Marketing Management 99 (2021) A1–A12
A9
Table 5
Event study analysis on the day with the highest number of deaths recorded (day
0).
Panel A – Event study (country-level analysis) – day 0
United States (US) Event
Date
Abnormal
Return
Cumulative
Abnormal
Return
t-test
Amazon 16/04/
2020
0.0538 0.1360 4.8739**
Facebook 16/04/
2020
0.0156 0.0990 10.0143**
Apple Inc. 16/04/
2020
0.0190 0.0843 2.9856**
United Kingdom
(UK)
Event
Date
Abnormal
Return
Cumulative
Abnormal
Return
t-test
BHP Group Plc 22/04/
2020
0.0449 0.0875 2.5106**
Tesco Plc 22/04/
2020
0.0341 0.0378 2.3053**
Unilever Plc 22/04/
2020
0.0087 0.0523 1.8151
China Event
Date
Abnormal
Return
Cumulative
Abnormal
Return
t-test
Bank of China Co.
Limited
17/04/
2020
0.0049 0.0099 4.8605**
Agricultural Bank of
China
17/04/
2020
0.0004 0.0026 2.4050**
Industrial and
Commercial Bank
of China
17/04/
2020
0.0041 0.0078 2.0790**
France Event
Date
Abnormal
Return
Cumulative
Abnormal
Return
t-test
BNP Paribas 06/04/
2020
0.0130 0.0188 1.1842
L’Oreal 06/04/
2020
0.0484 0.0469 1.8346
Sanofi 06/04/
2020
0.0113 0.0783 6.5515**
Italy Event
Date
Abnormal
Return
Cumulative
Abnormal
Return
t-test
Ferrari 30/03/
2020
0.0098 0.0102 1.0600
ENEL 30/03/
2020
0.0338 0.0023 0.0859
ENI 30/03/
2020
0.0472 0.1432 8.1881**
Spain Event
date
Abnormal
Return
Cumulative
Abnormal
Return
t-test
Banco Santander 03/04/
2020
0.015 0.029 1.208
Iberdrola 03/04/
2020
0.024 0.016 1.047
Industria de Diseno
Textil (ITX)
03/04/
2020
0.005 0.001 0.070
Panel B – Event Study (World and European Analysis) – Day 0
Event
Date
Abnormal
Return
Cumulative
Abnormal
Return
t-test
World 16/04/
2020
0.0538 0.1360 4.8739**
Europe 06/04/
2020
0.0354 0.0442 2.1665**
Panel A considers six countries that have been significantly affected by Covid-
19. Moreover, we selected three companies within each country based on
highest market capitalisation in their respective stock markets. Panel B illus-
trates the dataset for the world and Europe. The event date represents the
highest number of deaths on a particular day.
Table 6
Event study analysis on the day prior to the day with the highest number of
deaths recorded (day 1).
Panel A – Event study (country-level analysis) – day-1
United States (US) Event
Date
Abnormal
Return
Cumulative
Abnormal
Return
t-test
Amazon 15/04/
2020
0.0169 0.1960 6.7731**
Facebook 15/04/
2020
0.0055 0.0797 7.5505**
Apple Inc. 15/04/
2020
0.0086 0.0824 3.1303**
United Kingdom
(UK)
Event
Date
Abnormal
Return
Cumulative
Abnormal
Return
t-test
BHP Group Plc 21/04/
2020
0.0460 0.0725 1.8072
Tesco Plc 21/04/
2020
0.0003 0.0182 0.9992
Unilever Plc 21/04/
2020
0.0175 0.0493 1.5024
China Event
Date
Abnormal
Return
Cumulative
Abnormal
Return
t-test
Bank of China Co.
Limited
16/04/
2020
0.0010 0.0066 1.7158
Agricultural Bank of
China
16/04/
2020
0.0015 0.0029 2.1440**
Industrial and
Commercial Bank
of China
16/04/
2020
0.0020 0.0062 1.6672
France Event
Date
Abnormal
Return
Cumulative
Abnormal
Return
t-test
BNP Paribas 03/04/
2020
0.0301 0.0325 2.085**
L’Oreal 03/04/
2020
0.0050 0.0342 6.8601**
Sanofi 03/04/
2020
0.0352 0.0577 3.7800**
Italy Event
Date
Abnormal
Return
Cumulative
Abnormal
Return
t-test
Ferrari 27/03/
2020
0.0053 0.0001 0.0102
ENEL 27/03/
2020
0.0062 0.0232 2.052**
ENI 27/03/
2020
0.0279 0.0008 0.0296
Spain Event
Date
Abnormal
Return
Cumulative
Abnormal
Return
t-test
Banco Santander 02/04/
2020
0.0194 0.0360 1.8354
Iberdrola 02/04/
2020
0.0064 0.0212 1.7559
Industria de Diseno
Textil (ITX)
02/04/
2020
0.0263 0.0182 0.9839
Panel B – Event Study (World and European Analysis) – Day 1
Event
Date
Abnormal
Return
Cumulative
Abnormal
Return
t-test
World 15/04/
2020
0.0169 0.1960 6.773**
Europe 03/04/
2020
0.0081 0.0160 1.2821
Panel A considers six countries that have been significantly affected by Covid-
19. Moreover, we selected three companies within each country based on
highest market capitalisation in their respective stock markets. Panel B illus-
trates the dataset for the world and Europe. The event date represents the
highest number of deaths on a particular day.
Editorial
Industrial Marketing Management 99 (2021) A1–A12
A10
highest number of deaths recorded in a specific day. Table 5 also shows
the abnormal returns, cumulative abnormal returns, and t-test results.
The results indicate that the cumulative abnormal returns were different
from 0 and that the market behaved abnormally in response to the
announcement of the highest number of deaths on the event date. The
US and China had similar results, and the findings for all the companies
in these countries were statistically significant on the event day (US –
Amazon: 4.8739**, Facebook: 10.0143**, and Apple Inc.: 2.9856**, and
China – Bank of China Co. Limited: 4.8605**, Agricultural Bank of
China: 2.4050**, and Industrial and Commercial Bank of China:
2.0790**). The UK stock performance indicated statistically significant
results for BHP Group Plc (2.5106**) and Tesco Plc (2.3053**) on day 0.
France and Italy had significant results (see Sanofi: 6.5515** and ENI:
8.1881**, respectively). The significant values (**) indicate abnormal
stock returns on day 0. Therefore, higher abnormal returns had a rela-
tionship with the announcement of the highest number of deaths on day
0. On the day of announcement, cumulative abnormal returns were
equal to 0 in the Spanish stock market (Banco Santander: 1.208,
Iberdrola: 1.047, and Industria de Diseno Textil (ITX): 0.070). These
insignificant results indicate that corporate returns in the Spanish stock
market were unaffected by the announcement of the highest number of
deaths.
Panel B shows the combined results for Europe ( 2.1665**) and the
world (4.8739**). From this analysis, we can conclude that the cumu-
lative abnormal returns were different from 0. In other words, the stock
markets reacted abnormally to announcements of the highest number of
deaths.
To summarise the results, the US, the UK, and China had significant
results in terms of their stock returns on day 0. This indicates abnormal
returns due to the highest number of deaths being observed on day 0.
Table 6 shows the event study results for before the announcement
day (day 1). Panel A shows the abnormal movements in the stock
returns before the event day. The findings show significant results before
the announcement day (US – Amazon: 6.7731**, Facebook: 7.5505**,
and Apple Inc.: 3.1303**) and (France – BNP Paribas: -2.085**, L’Oreal:
6.8601**, and Sanofi: 3.7800**). This indicates that the cumulative
abnormal returns were different from 0 and that stocks reacted abnor-
mally before the announcement date. In China and Italy, the results are
statistically significant (Agricultural Bank of China: 2.1440** and ENEL:
2.052**, respectively). The significant values (**) indicate abnormal
stock returns prior to the date when the highest number of deaths owing
to Covid-19 was reported, which implies that the stock markets had
inside information on the trend of deaths before the event day (day 0).
However, in some cases, the results are insignificant before the event
date (UK – BHP Group Plc: 1.8072, Tesco Plc: 0.9992, and Unilever Plc:
1.5024 and Spain – Banco Santander: 1.8354, Iberdrola: 1.7559, and
ITX: 0.9839). All the insignificant results imply that corporate returns
were unaffected on day 1.
In Panel B, the findings for the aggregate ‘world’ show significant
results (6.773**) on the pre-event date. However, there are insignificant
results ( 1.2821) for the European data.
The results for the pre-event day ( 1) indicate that US and French
companies observed abnormal returns before the event date, indicating
that the markets started exhibiting abnormal reactions before the
countries reached the highest number of deaths from Covid-19.
Table 7 shows the results relating to the impact of the event after the
announcement day (day ‡1). In Panel A, we can see that France (BNP
Paribas: 3.2365**, L’Oreal: 2.4330**, and Sanofi: 7.4528**) and
Spain (Banco Santander: 2.5352**, Iberdrola: 6.6132**, and ITX:
4.0920**) had significant abnormal returns after the announcement day
(day ‡1). This indicates that the cumulative abnormal returns were
different from zero and that stocks reacted abnormally after the
announcement date. The abnormal stock returns for Amazon (US)
(2.4284**), Facebook (US) (6.1185**), Tesco Plc (UK) (2.2085**), In-
dustrial and Commercial Bank of China (China) ( 2.3674**), and ENI
(Italy) (13.0963**) were also statistically significant for day ‡1. All the
Table 7
Event study analysis on the day after the day with the highest number of deaths
recorded (day ‡1).
Panel A – Event study (country-level analysis) – day ‡ 1
United States (US) Event
Date
Abnormal
Return
Cumulative
Abnormal
Return
t-test
Amazon 17/04/
2020
0.0066 0.0627 2.4284**
Facebook 17/04/
2020
0.0337 0.0619 6.1185**
Apple Inc. 17/04/
2020
0.0039 0.0120 1.1864
United Kingdom
(UK)
Event
Date
Abnormal
Return
Cumulative
Abnormal
Return
t-test
BHP Group Plc 23/04/
2020
0.0534 0.0646 1.6089
Tesco Plc 23/04/
2020
0.0128 0.0411 2.2085**
Unilever Plc 23/04/
2020
0.0050 0.0066 0.6427
China Event
Date
Abnormal
Return
Cumulative
Abnormal
Return
t-test
Bank of China Co.
Limited
20/04/
2020
0.0109 0.0158 1.7381
Agricultural Bank of
China
20/04/
2020
0.0078 0.0060 1.3265
Industrial and
Commercial Bank
of China
20/04/
2020
0.0067 0.0079 2.3674**
France Event
Date
Abnormal
Return
Cumulative
Abnormal
Return
t-test
BNP Paribas 07/04/
2020
0.0232 0.0641 3.2365**
L’Oreal 07/04/
2020
0.0026 0.0510 2.4330**
Sanofi 07/04/
2020
0.0337 0.0831 7.4528**
Italy Event
Date
Abnormal
Return
Cumulative
Abnormal
Return
t-test
Ferrari 31/03/
2020
0.0094 0.0128 0.9044
ENEL 31/03/
2020
0.0171 0.0022 0.0991
ENI 31/03/
2020
0.0673 0.1995 13.0963**
Spain Event
Date
Abnormal
Return
Cumulative
Abnormal
Return
t-test
Banco Santander 06/04/
2020
0.0696 0.0863 2.5352**
Iberdrola 06/04/
2020
0.0492 0.1034 6.6132**
Industria de Diseno
Textil (ITX)
06/04/
2020
0.0265 0.0601 4.0920**
Panel B – Event Study (World and European Analysis) – Day ‡ 1
Event
Date
Abnormal
Return
Cumulative
Abnormal
Return
t-test
World 17/04/
2020
0.0066 0.0627 2.4284**
Europe 07/04/
2020
0.0085 0.0423 2.1922**
Panel A considers six countries that have been significantly affected by Covid-
19. Moreover, we selected three companies within each country based on
highest market capitalisation in their respective stock markets. Panel B illus-
trates the dataset for the world and Europe. The event date represents the
highest number of deaths on a particular day.
Editorial
Industrial Marketing Management 99 (2021) A1–A12
A11
significant (**) results indicate that stock prices continued to react
abnormally on the day after (day ‡1) the highest number of deaths was
announced (day 0).
Panel B shows that the stock returns for the world (2.4284**) and
Europe ( 2.1922**) were statistically significant on the post-event day
(day ‡1). It shows that the cumulative abnormal returns were different
from 0. Further, the US and Spain continued to experience abnormal
returns on the post-event day (day ‡1).
The Fig. 2 present the number of deaths alongside the cumulative
abnormal returns for an example company in each country. The dia-
grams demonstrate that, for the US, the UK, France, and Italy, the higher
the number of deaths, the greater the cumulative abnormal returns.
However, the opposite trend was observed for China and Spain.
5. Conclusion
The ESM is a commonly used methodological approach in the
accounting and finance literature. Researchers in the multi-disciplinary
fields of marketing and management have not adequately utilised this
methodological approach to assess the impact of major marketing-
related events on corporate stock returns.
This paper extends previous methodological papers published in the
Industrial Marketing Management journal. We aimed to use Covid-19 as an
example of a major event, and we investigated the impact of Covid-19
(highest death rate in a day) on stock returns in six major capital mar-
kets around the world. We have reported the step-by-step procedure that
can be utilised by researchers to understand the pre-announcement and
post-announcement impact of a major external shock. The ESM can also
be used to model investment behaviour and sentiments surrounding an
event date. Pre-announcement and post-announcement stock returns are
calculated to make judgements about the economic significance of an
event. We have also provided STATA commands that can be used by
non-technical users to understand and apply the ESM in marketing
research.
Appendix A: Mapping of event study methodology
References
Armitage, S. (1995). Event study methods and evidence on their performance. Journal of
Economic Surveys, 9(1), 25–52. https://doi.org/10.1111/j.1467-6419.1995.tb00109.
x.
Ashley, J. (1962). Stock prices and changes in earnings and dividends: Some empirical
results. Journal of Political Economy, 70(1), 82–85.
Ball, R., & Brown, P. (1968). An empirical evaluation of accounting income numbers.
Journal of Accounting Research, 6(2), 159–178.
Ball, R., & Brown, P. R. (2013). Ball and Brown (1968): A retrospective. SSRN Electronic
Journal, 1–44. https://doi.org/10.2139/ssrn.2304409.
Barker, C. A. (1956). Effective stock splits. Harvard Business Review, 34(1), 101–106.
Barker, C. A. (1957). Stock splits in a bull market. Harvard Business Review, 35(3), 72–79.
Barker, C. A. (1958). Evaluation of stock dividends. Harvard Business Review, 36(4),
99–114.
Beckers, S. F. M., van Doorn, J., & Verhoef, P. C. (2017). Good, better, engaged? The
effect of company-initiated customer engagement behavior on shareholder value.
Editorial
Industrial Marketing Management 99 (2021) A1–A12
A12
Journal of the Academy of Marketing Science, 46(3), 366–383. https://doi.org/
10.1007/s11747-017-0539-4.
Binder, J. J. (1998). The event study methodology since 1969. Review of Quantitative
Finance and Accounting, 11, 111–137.
Bobinski, G. S., & Ramirez, G. G. (1994). Advertising to investors: The effect of financial-
relations advertising on stock volume and price. Journal of Advertising, 23(4), 13–28.
https://doi.org/10.1080/00913367.1943.10673456.
Boehmer, E., Musumeci, J., & Poulsen, A. B. (1991). Event-study methodology under
conditions of event-induced variance. Journal of Financial Economics, 30(2),
253–272. https://doi.org/10.1016/0304-405X(91)90032-F.
Boyd, D. E., Chandy, R. K., & Cunha, M. (2010). When do chief marketing officers affect
firm value? A customer power explanation. Journal of Marketing Research, 47(6),
1162–1176. https://doi.org/10.1509/jmkr.47.6.1162.
Brown, S. J., & Warner, J. B. (1980). Measuring security price performance. Journal of
Financial Economics, 8(3), 205–258. https://doi.org/10.1016/0304-405X(80)90002-
1.
Brown, S. J., & Warner, J. B. (1985). Using daily stock returns: The case of event studies.
Journal of Financial Economics, 14(1), 3–31.
Cao, Z., & Yan, R. (2017). Does brand partnership create a happy marriage? The role of
brand value on brand alliance outcomes of partners. Industrial Marketing Manage-
ment, 67, 148–157.
Carlsson-Szlezak, P., Reeves, M., & Swartz, P. (2020). Understanding the economic shock
of coronavirus. Harvard Business Review, 3, 1–10, 27 March. Retrieved from https
://hbr.org/2020/03/understanding-the-economic-shock-of-coronavirus (Accessed 5
February 2021).
Cheng, J. M. S., Tsao, S. M., Tsai, W. H., & Tu, H. H. J. (2007). Will eChannel additions
increase the financial performance of the firm? The evidence from Taiwan. Industrial
Marketing Management, 36(1), 50–57.
Clark, J. M., Cornwell, T. B., & Pruitt, S. W. (2009). The impact of title event sponsorship
announcements on shareholder wealth. Marketing Letters, 20(2), 169–182. https://
doi.org/10.1007/s11002-008-9064-z.
Cornwell, T. B., Pruitt, S. W., & Clark, J. M. (2005). The relationship between major-
league sports’ official sponsorship announcements and the stock prices of sponsoring
firms. Journal of the Academy of Marketing Science, 33(4), 401–412. https://doi.org/
10.1177/0092070305277385.
Cornwell, T. B., Pruitt, S. W., & van Ness, R. (2001). The value of winning in motorsports:
Sponsorship-linked marketing. Journal of Advertising Research, 41(1), 17–31. https://
doi.org/10.2501/JAR-41-1-17-31.
Corrado, C. J. (1989). A nonparametric test for abnormal security-price performance in
event studies. Journal of Financial Economics, 23(2), 385–395.
Corrado, C. J. (2011). Event studies: A methodology review. Accounting and Finance, 51
(1), 207–234. https://doi.org/10.1111/j.1467-629X.2010.00375.x.
Cowan, A. R. (1992). Nonparametric event study tests. Review of Quantitative Finance and
Accounting, 2(4), 343–358. https://doi.org/10.1007/BF00939016.
Das, N., McNeil, C., Pouder, R. W., & Daly, J. P. (2020). Is cause-related marketing a
winning strategy? An event study analysis. Journal of Marketing Theory and Practice,
28(1), 43–55. https://doi.org/10.1080/10696679.2019.1662313.
Delattre, E. (2007). Event study methodology in marketing. Recherche et Applications en
Marketing (English Edition), 22(2), 57–75. https://doi.org/10.1177/
205157070702200204.
Dolley, J. C. (1933). Open market buying as a stimulant for the bond market. Journal of
Political Economy, 41(4), 513–529.
Fama, E. F., Fisher, L., Jensen, M. C., & Roll, R. (1969). The adjustment of stock prices to
new information. International Economic Review, 10(1), 1–21. https://doi.org/
10.2307/2525569.
Fang, E., Lee, J., & Yang, Z. (2015). The timing of codevelopment alliances in new
product development processes: Returns for upstream and downstream partners.
Journal of Marketing, 79(1), 64–82. https://doi.org/10.1509/jm.13.0490.
Gopinath, G. (2020). The great lockdown: Worst economic downturn since the great
depression. IMFBlog, 14 April. Retrieved from https://blogs.imf.org/2020
/04/14/the-great-lockdown-worst-economic-downturn-since-the-great-depression/
(Accessed 5 February 2021).
Homburg, C., Vollmayr, J., & Hahn, A. (2014). Firm value creation through major
channel expansions: Evidence from an event study in the United States, Germany,
and China. Journal of Marketing, 78(3), 38–61. https://doi.org/10.1509/jm.12.0179.
Hozier, G. C., & Schatzberg, J. D. (2000). Advertising agency terminations and reviews.
Journal of Business Research, 50(2), 169–176. https://doi.org/10.1016/s0148-2963
(99)00029-6.
Jayanti, R. K., & Jayanti, S. V. (2011). Effects of airline bankruptcies: An event study.
Journal of Services Marketing, 25(6), 399–409. https://doi.org/10.1108/
08876041111160998.
de Jong, F. (2007). Event studies methodology. Tilburg University. Retrieved from
https://www.academia.edu/35248600/Event_Studies_Methodology (Accessed 20
May 2020).
Kalaignanam, K., & Bahadir, S. C. (2013). Corporate brand name changes and business
restructuring: Is the relationship complementary or substitutive? Journal of the
Academy of Marketing Science, 41(4), 456–472. https://doi.org/10.1007/s11747-
012-0321-6.
Kantor, J., & Pike, R. (1987a). The determinants of the value of unlisted shares: Opinions
of professional valuers in Canada. Accounting and Business Research, 17(66),
109–115.
Kaspereit, T. (2020). Event studies with daily stock returns in Stata: Which command to
use? Stata Journal.
Kolari, J. W., & Pynnonen, S. (2011). Nonparametric rank tests for event studies. Journal
of Empirical Finance, 18(5), 953–971.
Lee, R. P., & Kim, D. (2010). Implications of service processes outsourcing on firm value.
Industrial Marketing Management, 39(5), 853–861.
Lim, W. M., Ahmed, P. K., & Ali, M. Y. (2019). Data and resource maximization in
business-to-business marketing experiments: Methodological insights from data
partitioning. Industrial Marketing Management.. https://doi.org/10.1016/j.
indmarman.2018.08.007.
Lubatkin, M., & Shrieves, R. E. (1986). Towards reconciliation of market performance
measures to strategic management research. Academy of Management Review, 11(3),
497–512.
MacKinlay, A. C. (1997). Event studies in economics and finance. Journal of Economic
Literature, 35(1), 13–39.
McWilliams, A., & Siegel, D. (1997). Event studies in management research: Theoretical
and empirical issues. Academy of Management Journal, 40(3), 626–657. https://doi.
org/10.5465/257056.
de Mortanges, C. P., & Rad, A. T. (1998). Marketing strategy and market value: An event-
study analysis. European Management Journal, 16(3), 365–371. https://doi.org/
10.1016/S0263-2373(98)00013-9.
Myers, J. A., & Bakay, A. (1948). Influence of stock split-ups on market price. Harvard
Business Review, 26(2), 251–255.
Oh, Y., Lee, J., & Kim, N. (2018). The contingency value of the partner firm’s customer
assets in a business-to-business relationship. Industrial Marketing Management, 73,
47–58.
Rahman, M., & Lambkin, M. (2015). Creating or destroying value through mergers and
acquisitions: A marketing perspective. Industrial Marketing Management, 46, 24–35.
Romei, V. (2020). Household deposits soar by record amount as lockdown hits spending.
Financial Times, 29 June. Retrieved from https://www.ft.com/content/f3640c3
a-1a4e-4d37-8226-b773db49ddea (Accessed 5 February 2021).
Sorescu, A., Warren, N. L., & Ertekin, L. (2017). Event study methodology in the mar-
keting literature: An overview. Journal of the Academy of Marketing Science, 45(2),
186–207. https://doi.org/10.1007/s11747-017-0516-y.
Swaminathan, V., & Moorman, C. (2009). Marketing alliances, firm networks and firm
value creation. Journal of Marketing, 73(5), 52–69. Retrieved from http://www.jstor.
org/stable/20619046?seqˆ1&cidˆpdf-reference%0Ahttps://faculty.fuqua.duke.
edu/%7B~%7Dmoorman/Publications/JM2009.pdf (Accessed 5 February 2021).
Tsiotsou, R., & Lalountas, D. (2005). Applying event study analysis to assess the impact
of marketing communication strategies: The case of sponsorship. Applied Financial
Economics Letters, 1(4), 259–262. https://doi.org/10.1080/17446540500143764.
Ullah, S., Akhtar, P., & Zaefarian, G. (2018). Dealing with endogeneity bias: The
generalized method of moments (GMM) for panel data. Industrial Marketing Man-
agement., 71, 69–78. https://doi.org/10.1016/j.indmarman.2017.11.010.
Ullah, S., Zaefarian, G., & Ullah, F. (2020). How to use instrumental variables in
addressing endogeneity? A step-by-step procedure for non-specialists. Industrial
Marketing Management.. https://doi.org/10.1016/j.indmarman.2020.03.006.
Vaid, S. S., & Ahearne, M. (2018). When does CEO endorsement of sales & marketing
leaders help firms? The role of heavy marketing emphasis. Industrial Marketing
Management, 69, 185–197.
Vaid, S. S., Ahearne, M., & Krause, R. (2020). Joint marketing and sales appointment:
Uncertainty from intertwining of marketing and sales in one position. Industrial
Marketing Management, 85, 221–239.
Wigglesworth, R. (2020). In the coronavirus crisis investors cannot rely on the old
playbooks. Financial Times, 31 March. Retrieved from https://www.ft.com/content
/4d2561da-ed80-4204-916e-8ffbde31e0fe (Accessed 5 February 2021).
Wiles, M. A., & Danielova, A. (2009). The worth of product placement in successful films:
An event study analysis. Journal of Marketing, 73(4), 44–63. https://doi.org/
10.1509/jmkg.73.4.44.
Wiles, M. A., Morgan, N. A., & Rego, L. L. (2012). The effect of brand acquisition and
disposal on stock returns. Journal of Marketing, 76(1), 38–58. https://doi.org/
10.1509/jm.09.0209.
Subhan Ullaha,*, Ghasem Zaefarianb, Rizwan Ahmedc, Danson Kimanid
a Nottingham University Business School, University of Nottingham,
Nottingham, UK
b Leeds University Business School, University of Leeds, Leeds, UK
c Birmingham University Business School, University of Birmingham,
Birmingham, UK
d Essex Business School, University of Essex, Colchester, UK
* Corresponding author.
E-mail address: subhan.ullah@nottingham.ac.uk (S. Ullah).
Editorial