Content uploaded by Vinod R. Singhal
Author content
All content in this area was uploaded by Vinod R. Singhal on Aug 06, 2022
Content may be subject to copyright.
RESEARCH ARTICLE
Stock market reaction to global supply chain disruptions
from the 2018 US government ban on ZTE
Brian W. Jacobs
1
| Vinod R. Singhal
2
| Xinrui Zhan
3
1
Graziadio Business School, Pepperdine
University, Malibu, California, USA
2
Scheller College of Business, Georgia
Institute of Technology, Atlanta,
Georgia, USA
3
Institute of Finance and Public
Management, Anhui University of
Finance and Economics, Bengbu, China
Correspondence
Xinrui Zhan, Institute of Finance and
Public Management, Anhui University of
Finance and Economics, Bengbu 233030,
China.
Email: zhanxinrui@aufe.edu.cn
Handling Editors: Di Fan, Chris
K.Y. Lo, Christopher S. Tang, Andy
Yeung, and Yi Zhou.
Abstract
Government trade actions are an increasing source of supply chain risk. This
research provides empirical evidence of the stock market reaction to trade actions
against a targeted firm on other firms in the targeted firm's supply chain
eco-system. We test our hypothesized stock price effects using the case of the 2018
US government ban on US firms from supplying to ZTE, a Chinese telecommuni-
cations manufacturer. We estimate the ban's effects on ZTE's tier-one US and
non-US suppliers, as well as the upstream and downstream supply chain propaga-
tion effects by considering ZTE's tier-two suppliers and business customers. We
also estimate impacts to ZTE's competitors. We find that tier-one US suppliers
experienced a stock price effect of 3.33% following the ban, and the reaction was
more negative for those suppliers more dependent on ZTE for revenues. We find a
stock price effect on tier-two suppliers of 0.40%, but an insignificant effect on
non-US tier-one suppliers. Business customers experienced a stock price effect of
0.66%, and the competitors' stock price effect was 1.34%. The reversal of the ban
4 weeks later resulted in a stock price effect of 1.56% for tier-one US suppliers,
1.72% for tier-one non-US suppliers, and 1.35% for competitors.
KEYWORDS
empirical research, global operations management, government sanctions, supply chain
disruption, trade actions, trade policy
Highlights
•The 2018 ZTE trade ban by the US government resulted in significant
market value losses (median 3.33%) for ZTE's US suppliers, but not for its
non-US suppliers.
•The reversal of the ban 4 weeks later resulted in significant market value
gains (median 1.56%) for ZTE's US suppliers, but the gains were not suffi-
cient to offset the losses incurred by the ban.
•Policymakers and regulators need to be sensitive to the potential market
value gains and losses due to government trade actions for both domestic
and non-domestic firms, and supply chain managers and investors need to
be aware of the magnitude of the impacts.
Received: 3 June 2021 Revised: 14 April 2022 Accepted: 29 April 2022
DOI: 10.1002/joom.1197
J Oper Manag. 2022;1–25. wileyonlinelibrary.com/journal/joom © 2022 Association for Supply Chain Management, Inc. 1
1|INTRODUCTION
In recent years, geopolitical issues such as nationalism,
national security, protectionism, and self-sufficiency have
increased the vulnerability of supply chains (e.g.,
McKinsey, 2020; Sodhi & Tang, 2021). Governments are
using policy tools such as regulations, restrictions, tariffs,
and bans to address these geopolitical issues. The deploy-
ment of these tools can often be unilateral and unantici-
pated, surprising firms and causing supply chain
disruptions. Recent examples include Russian govern-
ment restrictions on agricultural imports from the
United States and the EU (Liefert et al., 2019); Japanese
government restrictions of chemical exports needed for
semiconductor manufacturing in South Korea
(Martin, 2019); and US government actions to curb sales
of telecommunications equipment by Huawei (Kaska
et al., 2019). The use of such policy tools is increasing.
For example, the entity list of export restrictions
maintained by the US Department of Commerce has
grown from 300 in 2010 to almost 1200 in 2020
(Ney, 2021). The length of the disruptions from these
trade actions and their ultimate resolutions are typically
uncertain and beyond the control of firms. Politicization
of supply chains is a new source of risk that can disrupt
supply chains and can harm not only foreign firms but
also domestic firms. Governments need to consider these
costs and make tradeoffs between the potential gains
from implementing policies that address the geopolitical
issues, versus the potential costs of supply chain disrup-
tions that result from these policies. Although economists
have long studied the macro-level financial impacts of
global trade frictions (e.g., Brander & Spencer, 1992;
Lee & Swagel, 1997; Li et al., 2018), there is limited
research that explores the financial consequences of sup-
ply chain disruptions at the firm-level due to geopolitics.
We address this gap by analyzing the effect of a spe-
cific US government action on the financial performance
of firms in a multinational supply chain. Examining the
impact on supply chains is important since most firms no
longer compete with other firms as individual entities but
instead competition is more supply-chain-to-supply-chain
(Christopher, 2000). To estimate the financial effects of
trade actions, we leverage the case of ZTE, a Chinese tele-
communications manufacturer. On April 16, 2018, the
US Department of Commerce banned US firms from sell-
ing products to ZTE for 7 years. The ban was imposed
following ZTE's violation of the terms of a 2017 agree-
ment between the US and ZTE that settled allegations of
sanctions violations by ZTE involving North Korea and
Iran. The ban disrupted ZTE's supply chain and raised
concerns about the effect of the ban on ZTE and its sup-
pliers and customers, and on other global supply chains
for high technology products. In the immediate aftermath
of the ban, ZTE delayed its quarterly earnings release to
assess the consequences of the ban and trading in its
stock was suspended. On May 13, 2018, 4 weeks after the
ban, the US government announced that it was working
with ZTE and the Chinese government on a plan to lift
the ban. The ban was ultimately lifted on July 15, 2018,
nearly 3 months after it was imposed.
This paper provides empirical evidence on the effect
of supply chain disruptions from the ban on the financial
performance of ZTE, firms in ZTE's supply chain, and
ZTE's competitors. We measure financial performance by
stock market reaction (i.e., stock price effects). Our analy-
sis considers both the suppliers and business customers
of ZTE. The sample comprises tier-one suppliers, tier-two
suppliers, and business customers. We also estimate the
financial impact of the ban on ZTE's competitors. Our
analysis focuses on the stock price performance of sup-
pliers, customers, and competitors in the immediate
aftermath of the announcement of the ban as well as its
subsequent reversal. Estimating the economic impacts of
trade actions such as the ZTE ban can help managers to
decide the appropriate scope and course of action, both
preventive and corrective. Understanding the magnitude
of economic impact to supply chain partners can also
help policymakers judge whether the intended policy
aims of a trade action targeted against a firm are worth
the costs incurred by firms.
Researchers have studied the effect of trade actions
such as tariff announcements by the United States and
China on the stock market returns of nearly all the firms
traded on certain stock exchanges. For example, Egger
and Zhu (2020) examine the stock returns for 40 different
exchanges; Huang et al. (2020) focus on the United States
and China stock exchanges; Wang, Wang, et al. (2021)
restrict their analyses to Shanghai and Shenzhen stock
exchanges; and He et al. (2021) focus on the US S&P
500 Index and Shanghai Composite Index. These papers
either focus on the overall stock market or all the firms
that are part of these exchanges. They do not look at sam-
ples of firms that are directly affected by the tariffs.
Research that estimates the financial effects of trade
actions at the firm-level is limited (Webb, 2020), and even
more limited when considering the entire supply chain.
In a related paper, Allen (2021) examines the stock mar-
ket reaction to three cases of United States–China trade
sanctions on only US suppliers. Our paper builds on this
research by examining a different case, the 2018 ZTE
trade ban, adding to the limited evidence on how targeted
trade actions against firms affect the stock market reac-
tion to the firm's supply chain partners. In contrast to
Allen (2021), we take a supply chain-centric approach,
considering not only tier-one suppliers in the targeting
2JACOBS ET AL.
country but also other tier-one suppliers, tier-two sup-
pliers, customers, and competitors in all countries. We
also estimate the effects of the ban reversal whereas Allen
(2021) studies the imposition of trade sanctions but not
their removal.
There are four reasons why the ban on ZTE is a good
context to study the financial consequences of supply
chain disruptions due to trade actions. First, ZTE is one
of the world's leading telecommunications equipment
and smartphone manufacturers, and one of China's
global firms in the high technology sector. It had 2017
revenues of about $16 billion (109 billion RMB). It is the
fourth-largest telecommunications equipment manufac-
turer in the world and the fourth-largest seller of
smartphones in the United States. ZTE is highly depen-
dent on US firms for many critical and high technology
components. Analysts estimate that US firms provide
about 25%–30% of the components used in ZTE's prod-
ucts (Stecklow et al., 2018). The ban on ZTE escalated the
ongoing trade tensions between the United States and
China, the world's two largest economies. The event
received widespread press coverage, enabling investors
across the globe to access information about the ban. Sec-
ond, ZTE requires high tech and complex components to
build its products. To achieve this, ZTE depends on a
highly interconnected, interdependent, and global supply
chain. Our sample comprises firms headquartered in sev-
eral countries with shares traded on various national
markets, which enables us to consider the effects across
complex, global supply chains. Third, unlike the trade
actions studied by Allen (2021) and Wang, Wang, et al.
(2021), the ZTE action was an outright ban rather than a
sanction or tariff. With sanctions and tariffs, licenses are
often issued by the US government to permit otherwise-
affected US firms to be exempted. Last, the sudden and
unexpected reversal of the ZTE ban enables us to deter-
mine whether stock prices of the affected firms fully
recovered from the effects of the trade action.
Our analysis provides evidence of the direct effects of
the ban on the 53 ZTE tier-one suppliers headquartered
in the US, as well as the indirect effects of the ban on the
149 ZTE tier-one suppliers headquartered outside the
US. We also estimate the degree and extent with which
the effects of trade actions propagate through supply
chains. We document the upstream propagation of the
ban's financial effects through the supply chains of ZTE's
tier-one US suppliers by examining the impact on their
suppliers. This analysis is based on a sample of 295 tier-
two suppliers of ZTE that supply to the tier-one US sup-
pliers of ZTE. While there are papers in the literature that
examine the effect of major corporate events on tier-one
suppliers, there is limited research that considers the
propagation of effects to tier-two suppliers. Further, we
examine the downstream supply chain propagation of
the ban's effects based on a sample of 70 business cus-
tomers of ZTE. In addition to the direct effects and supply
chain propagation effects of the ban, we also estimate its
effects on a sample of 90 competitors of ZTE.
To summarize our main results, we find that tier-one
US suppliers of direct material to ZTE had a significant
median market reaction of 3.33%. Further, the market
reaction was more negative for those suppliers more
dependent on ZTE for revenues. Tier-two suppliers of
ZTE had a significant median market reaction of 0.40%.
The evidence indicates insignificant effects for ZTE's tier-
one non-US headquartered suppliers. Business customers
of ZTE had a marginally significant median market reac-
tion of 0.66%, and competitors experienced a significantly
positive median market reaction of 1.34%.
The reversal of the ban had a positive and significant
impact on ZTE's tier-one US headquartered, tier-one non-
US headquartered suppliers, and competitors. The median
market reaction was 1.56% for tier-one US headquartered
suppliers, less than the 3.33% loss they experienced from
the ban's imposition. For tier-one suppliers not head-
quartered in the US, the median market reaction was
1.72%. Competitors experienced a 1.35% median market
reaction. The market reaction to the ban reversal was insig-
nificant for tier-two suppliers, and customers of ZTE.
Section 2presents a literature review and develops
our hypotheses. Section 3describes the data collection
process and provides details about the various samples of
firms used in the analyses. Section 4discusses our
approach and method for estimating the stock price reac-
tion. Section 5presents our results. Section 6summarizes
our results and discusses the implications of our work.
2|LITERATURE REVIEW AND
HYPOTHESIS DEVELOPMENT
Although the study of supply chain risk management has
been prevalent in the operations management literature
since the mid-1990s (see Sodhi et al., 2012; Ho et al., 2015
for overviews of this research), examination of geopoliti-
cal risks is nascent. Kleindorfer and Saad (2005) note that
specifying risk sources is the requisite first step in manag-
ing supply chain risk. However, risk sources change and
evolve over time. As evidence, many earlier categoriza-
tions of risk sources (e.g., Chopra & Sodhi, 2004;
Christopher & Peck, 2004) do not include geopolitical
issues even though they are commonly cited in the recent
business press. For example, Jaeger (2019) lists global
trade wars as the most important source of supply chain
risk to monitor, and McKinsey (2020) includes trade dis-
putes among their list of costliest supply chain shocks.
JACOBS ET AL.3
The continued globalization of supply chains is increas-
ing geopolitical risks. Bode and Wagner (2015)definespa-
tial complexity as the geographic distance between firms
and suppliers, a measure which is increasing in globaliza-
tion. They argue that growth in spatial complexity increases
the exposure of firms to risks such as trade restrictions and
geopolitical events. Given that a key part of risk manage-
ment is first understanding the financial implications of
similar prior events (Kleindorfer & Saad, 2005), it is impor-
tant to give in-depth consideration of emerging risk sources
such as geopolitical events.
Many researchers have noted the tendency of man-
agers to plan for risks from high probability–low conse-
quence events that are commonly found in daily
operations, but they also note this cannot substitute for
the risk planning required for low probability–high con-
sequence (LP–HC) disruptions (e.g., Kleindorfer &
Saad, 2005; Knemeyer et al., 2009). Like other supply
chain disruptions from natural disasters, industrial acci-
dents, and pandemics, most geopolitical risks fall into the
LP–HC category. Ellis et al. (2010) explain that percep-
tions of risk, and subsequent firm mitigation efforts, are
driven partly by the degree to which the risk can be con-
trolled. Uncontrollability, which typifies many LP–HC
events, tends to elevate risk perceptions, and can con-
sume an outsized portion of risk mitigation efforts.
Geopolitical events such as trade actions can present
many of the same challenges as other LP–HC supply
chain risks: demand can suddenly shift; access to supplies
and/or markets can be curtailed; and internal operations
can be shut down or restricted. However, trade actions
also present challenges that can be somewhat unique.
The political issues that result in supply chain disrup-
tions, especially those in foreign countries, are often less
monitored by managers despite the advocacy of
researchers (e.g., Christopher & Peck, 2004) for continu-
ous analysis of political, economic, social, and technologi-
cal issues. Short-term recovery efforts following most LP–
HC disruptions are often aimed at repairing the existing
supply network, but that might not be feasible for trade
actions. Rather than restoring network flows, the struc-
ture of the network might need to change if the trade
actions are long-lasting; the solution often involves find-
ing new supply chain partners in countries or regions
that are unaffected by the trade actions. And while sup-
ply chain managers often work extensively with supply
chain partners to develop capabilities such as redundancy
and agility to withstand LP–HC disruptions, their capa-
bility to work with governments and trade regulators is
likely not as developed. Another unique feature of trade
actions is that they can sometimes be altered or reversed
with little warning, creating uncertainty for managers
regarding whether substantial recovery efforts are
warranted. Thus, it is important to learn more about the
effects of trade actions as an increasing source of supply
chain risks and disruptions.
Research on mitigating the supply chain risks from
trade actions is scarce. Using stylized models, Wang et al.
(2011) examine sourcing strategies to deal with regula-
tory trade risks such as export constraints or anti-
dumping measures. They find that the best strategy is
dependent on the volatility of the trade barriers. In an
experimental study, Phadnis and Joglekar (2021) advo-
cate for the use of multi-party scenario planning as a tool
to better prepare for regulatory disruptions such as trade
wars and tariffs. Charpin et al. (2021), in a multiple case
study of foreign-owned subunits operating in China, sug-
gest that firms can mitigate trade risks by seeking to gain
legitimacy from host governments. Legitimacy can be
built through activities such as corporate political activ-
ity, corporate social responsibility (CSR), and capability
development of local suppliers.
2.1 |Economic consequences on
suppliers in the targeting country
An important consequence of supply chain disruptions due
to trade actions is their economic impact not only on the
targeted firm(s) but also on their supply chain partners,
including suppliers. The suppliers most affected by trade
actions are those headquartered in the targeting country
since governments in targeting countries can only directly
regulate domestic firms, that is, firms headquartered in the
targeting country. Governments in targeting countries have
no legal authority to regulate the operations of suppliers
headquartered in other countries. When trade actions by a
targeting country—whether they are tariffs, quotas, import
or export controls, sanctions, bans, and so forth—restrict
transactions with targeted firms, the expected future finan-
cial performance of suppliers to those targeted firms is
affected. For suppliers in the targeting country, this can
happen in at least three ways.
First, suppliers could face future revenue losses. Trade
actions targeted against a firm can severely hamper the
firm's operations, reducing their business prospects and
hence their orders to suppliers. In the case of trade bans,
suppliers of the targeting country are forbidden from sup-
plying the targeted firm, thereby losing revenues from
the targeted firm. In addition to lost orders, suppliers of
targeted firms might lose revenue through reduction
of their pricing power. Existing and potential customers
of the affected suppliers are aware of the trade action and
its impact on the suppliers. Accordingly, customers might
demand price reductions or refuse price increases from
suppliers in weakened market positions. Suppliers might
4JACOBS ET AL.
also lose revenue by reducing prices to attract new cus-
tomers and replace the lost volume of the targeted firm.
For suppliers in a targeting country, trade actions can
also damage their reputation as a reliable and dependable
supplier (Allen, 2021). Customers in other countries
might be hesitant to source from suppliers in the
targeting country for fear of future trade actions and sub-
sequent supply chain disruptions.
Second, the loss of business with targeted firms could
result in future added operating costs to suppliers. Any
excess capacity at the supplier due to lost orders from the
targeted firm can be costly. Excess capacity means not
only that fixed costs are spread over fewer units, but
expenses will increase if there is a need to mothball
equipment or facilities. If the volumes lost at a targeted
customer are significant, workforce reductions might be
warranted. These can create costs for separation packages
and worker relocations. Further, if the effects of the trade
action are immediate, suppliers might be burdened with
unsold inventories of work-in-process and finished goods
destined for the targeted firms. Excess inventories are
likely costlier for short-life-cycle products such as high-
tech goods as they are more challenging to quickly repur-
pose. Also dependent on the nature of the products, asset
specificity at the supplier might be high. Specific assets
are difficult and expensive to adapt for alternate uses
(Grover & Malhotra, 2003). Additionally, if new products
must be developed to offset the lost volume, operating
expenses for research and development will increase.
Third, future selling, general, and administrative costs
of suppliers could increase from trade actions targeted at
a customer. Increased costs and efforts in sales and mar-
keting will be needed for suppliers to find customers
and/or markets to replace the volumes lost at the targeted
firm. Suppliers might also opt to increase their spending
on lobbying to influence future government trade actions.
Such efforts require developing a lobbying network to
interact with politicians and trade regulators. Not only
are such efforts costly, but they require significant top
management attention. Trade actions also spur added
costs for suppliers to monitor their trade compliance.
For these reasons, trade actions are expected to nega-
tively impact the future financial performance of the
targeted firm's suppliers. Efficient markets theory
(Fama, 1970) posits that investors quickly incorporate
available information about expected future changes to
financial performance into the stock price of a firm.
Thus, when trade actions targeting a firm are announced,
investors will anticipate the expected future negative
financial impacts for the targeted firm's suppliers and
react accordingly by lowering the stock price of the
targeted firm's suppliers. See Figure 1for a conceptual
model of this process.
Several studies document the negative effect of other
types of supply chain disruptions on stock prices. For
example, Hendricks and Singhal (2003) study the effects
of endogenous supply chain disruptions that occur within
firms. Drakos (2004) and Carter and Simkins (2004) esti-
mate the effects of the September 11, 2001 terror attacks
on airlines. Barrot and Sauvagnat (2016) study the effects
of natural disasters on supply chains, and Hendricks
et al. (2020) estimate the effects of the 2011 Japanese
earthquake and tsunami on supply chains.
Given the above theoretical arguments and empirical
evidence, we hypothesize:
H1. Trade actions targeted at a firm negatively
affect the stock prices of the targeted firm's sup-
pliers in the targeting country.
Suppliers in the targeting country that are more reve-
nue dependent on a targeted firm are likely to suffer
greater impacts from trade actions than suppliers with
less revenue dependence. The increased impact can occur
through any of the three mechanisms described above
and depicted in Figure 1: revenues; operating costs; and
selling, general, and administrative costs. Since the
effects of revenue loss from any lost customer should be
at least proportional to a supplier's revenue dependence
on the customer, revenue losses will be greater for more
dependent suppliers. Revenue effects due to loss of pric-
ing power should also be exacerbated for suppliers with
greater revenue dependence.
Similarly, potential increases in operating costs are
also likely greater for more dependent suppliers. Excess
capacity and any required workforce reductions should
be proportional to the volume lost. As Banerjee et al.
(2008) explain, when suppliers have relatively few cus-
tomers, the goods they provide are also likely to be more
specific and less standardized. Thus, any excess invento-
ries are not only likely greater in volume due to greater
revenue dependence, but also more difficult to repurpose.
Further, asset specificity tends to be greater when sup-
pliers depend on customers for a large portion of their
revenues (Banerjee et al., 2008; Holcomb & Hitt, 2007).
As mentioned above, specific assets are difficult and
expensive to adapt for alternate uses. In highly dependent
relationships, customers often give suppliers preferential
financial treatment including more favorable terms, con-
ditions, or trade credits (Banerjee et al., 2004). Thus, the
loss of such a relationship could create cost increases for
the supplier.
Highly dependent suppliers will also need increased
sales and marketing efforts to find more replacement cus-
tomers and/or markets. Given the severity of the loss they
are facing, dependent suppliers are also more motivated
JACOBS ET AL.5
to increase their lobbying efforts and costs to influence
trade actions. Thus, the selling, general, and administra-
tive costs for more dependent suppliers should be greater
than those for less dependent suppliers.
In light of the increased magnitudes of negative
changes in revenues and costs that more dependent sup-
pliers face, we hypothesize:
H2. The negative effect of trade actions on the
stock prices of a targeted firm's suppliers in the
targeting country is increasing in the suppliers'
revenue dependence on the targeted firm.
Suppliers in the targeting country that lose business
from a firm targeted by trade actions will in turn reduce
orders from their suppliers (tier-two suppliers). In a man-
ner similar to tier-one suppliers of targeted firms, tier-
two suppliers can experience future lost revenues and
increased costs. Revenue losses can result from either
reduced order volumes from tier-one suppliers, and/or
potential impacts on the pricing power of tier-two sup-
pliers. Operating cost increases can result at tier-two sup-
pliers due to excess capacity and inventories that must be
dealt with, specific assets that must be repurposed, and
even unfavorable changes in terms and conditions due to
belt-tightening at their tier-one suppliers. The selling,
general, and administrative costs of tier-two suppliers can
also increase due to the need for greater sales, marketing,
and lobbying efforts that likely result from reduced order
volumes from tier-one suppliers impacted by trade
actions.
Propagation of financial effects from major events
through the supply chain can be expected due to the con-
tractual links between customers and suppliers. Studies
have established supply chain propagation effects on
upstream partners. Hertzel et al. (2008) examine the effect
of bankruptcy announcements by firms on their tier-one
suppliers, and Hendricks et al. (2020) study the effect on
tier-one suppliers of firms directly affected by the 2011
Japanese earthquake and tsunami. Studies that examinee
deeper propagation effects include Jacobs and Singhal
(2020) who examine the stock price effects of the 2005
Volkswagen diesel emissions scandal on tier-two suppliers,
and Wang, Li, et al. (2021) who examine how risk events
at tier-two suppliers affect the focal firm (tier-zero).
Given the above theoretical arguments and empirical
evidence related to propagation, we hypothesize:
H3. The negative effect of trade actions on the
stock prices of a targeted firm's suppliers in the
targeting country propagates to the stock prices
of their suppliers.
2.2 |Economic consequences on other
supply chain partners
To this point, we have considered the effects that trade
actions against targeted firms can have on suppliers in
the targeting country (and their suppliers). We now con-
sider the potential impacts for suppliers in countries
other than the targeting country. There are factors that
can negatively affect these suppliers, and other factors
that can positively affect these suppliers.
Even though governments that undertake trade
actions targeted against a firm have no direct legal or reg-
ulatory authority over suppliers headquartered outside of
their country, those suppliers can be negatively affected.
Many of the same mechanisms described in our develop-
ment of H1 (and depicted in Figure 1) apply to all sup-
pliers of the targeted firm, not just those in the targeting
country. To the extent that the overall business prospects
of a targeted firm are reduced, all of the firm's suppliers
could be subject to potential revenue losses and increases
in costs, both operating costs as well as selling, general,
and administrative costs.
On the positive side, suppliers not located in targeting
countries are free to continue their business relationships
with targeted firms. Even if the trade action reduces the
business prospects of the targeted firm, the targeted firm
might be able to find a work-around and continue con-
ducting business with suppliers in countries other than
the targeting country. Further, firms targeted by trade
actions may source some of their critical parts and com-
ponents from multiple suppliers in multiple countries.
Hence, targeted firms might be able to shift orders from
FIGURE 1 Conceptual model of
the mechanisms by which a trade action
affects the financial performance and
stock price of the targeted firm's
suppliers
6JACOBS ET AL.
suppliers in the targeting country to alternate suppliers in
other countries, potentially benefiting those suppliers in
other countries. In fact, if the trade action is expected to be
long-lasting, the targeted firm will likely change its sourcing
strategy to only deal with suppliers in countries other than
the targeting country. Anecdotal evidence for such strate-
gies is offered by Fitch and Strumpf (2019) who report that
the Chinese telecom company Huawei shifted it sourcing
strategy away from US suppliers after it was the target of US
government trade actions. Jeong (2019)reportsthat
South Korean chipmakers Samsung and SK Hynix are
developing domestic sources of supply after they were
affected by Japanese government trade actions.
Given that suppliers in other countries face a mix of
potential negative and positive effects from trade actions
targeted at a firm, we cannot predict which effects will
dominate. Hence, we determine the directionality empiri-
cally by hypothesizing:
H4. Trade actions targeted at a firm affect
the stock prices of the targeted firm's suppliers
in other countries beyond the targeting
country.
We now shift our focus from upstream supply chain
partners of a firm targeted by trade actions to its down-
stream supply chain partners. For all business customers
regardless of their country location, if the targeted firm is
forced to curtail its operations due to lack of supplies, it
may not be able to fill its contractual obligations to them.
This creates a supply chain disruption for the customer.
If the customer loses all or partial supply from the
targeted firm, it will need to quickly shift its procurement vol-
ume to alternate suppliers. As Friedl and Wagner (2012)
explain, customers incur costs to switch suppliers due to
prior commitments to the incumbent supplier such as spe-
cific physical or informational assets, as well as the adminis-
trative costs of onboarding a new supplier. Further, if we
assume that customers are already sourcing from their best
value supplier, switching to an alternate supplier will likely
result in less value for the customer, either in the form of
reduced quality or service, increased price, or increased
uncertainty (Cannon & Homburg, 2001). The effect of value
changes and switching costs could be greater if the customer,
in fear of further trade actions, chooses to avoid suppliers in
either the targeted or targeting country. Limitations in the
number of potential suppliers not only reduces the buyer's
consideration set but can harm supplier diversification efforts
and subject the buyer to increased risk (Chod et al., 2019).
Of course, if the targeted firm is a customer's sole sup-
plier for an item, the customer's operations might be
severely hampered, and its costs greatly inflated. Con-
versely, if the targeted firm is one of multiple suppliers
for similar items, the disruption and costs to the customer
from the trade action might be mitigated to some extent.
A few studies have considered the stock price effects
of supply chain disruptions on downstream firms. As
examples, Barrot and Sauvagnat (2016) consider the
impact of natural disasters at suppliers on their cus-
tomers, and Jacobs and Singhal (2017) study the stock
price effects of the Rana Plaza building collapse in
Bangladesh on customers in the apparel industry.
Our hypothesis of the stock price effects of trade
actions on downstream firms is:
H5. Trade actions targeted at a firm negatively
affect the stock prices of the targeted firm's busi-
ness customers.
Last, we examine the potential impacts of trade
actions against a targeted firm on the targeted firm's com-
petitors. Although competitors are not directly affected
by trade actions unless they are targeted at them, they
might be indirectly affected either through contagion
effects or competitive effects.
Contagion effects stem from negative events affecting a
firm or firms in an industry that cause stakeholders (includ-
ing suppliers, customers, and investors) to become wary of
other firms in the same or similar industries. Contagion
effects have been found for accidents in the nuclear indus-
try (e.g., Bowen et al., 1983; Kalra et al., 1993), bankruptcies
(Lang & Stulz, 1992), and the 2015 VW diesel emissions
scandal (Jacobs & Singhal, 2020), among others. The likeli-
hood and magnitude of a contagion effect are affected by
the idiosyncrasy of the event (Aharony & Swary, 1983). Idi-
osyncratic events, such as trade actions targeted at a specific
firm, have limited impact on other firms.
On the other hand, negative events could also result in a
competitive effect. A competitive effect can occur when
negative events at a firm (such as a trade action against a
firm) create positive business opportunities for its competi-
tors. Such opportunities might include increased revenues
from market share gains, and/or improved pricing power
because of the inability of the affected firm to maintain its
sales. Other potential benefits could include reduced costs
due to greater volumes, or enhanced brand value. Firms
might also benefit from better terms and conditions offered
by former suppliers or customers of the affected firm seek-
ing new contracts. Competitive effects resulting from sup-
ply chain disruptions, and their subsequent impact on stock
prices, are studied by Lang and Stulz (1992), Jacobs and
Singhal (2020), and Hendricks et al. (2020).
Lang and Stulz (1992) argue the ability of competitors
to capitalize on the affected firm's negative event depends
partially on the degree of competition. If competition is
perfect, competitors would be unable to substantially raise
JACOBS ET AL.7
prices or capture market share. But in less competitive set-
tings, competitors can often gain revenues from increased
prices and/or volumes. We note that trade actions are gen-
erally counter to perfect competition, increasing our
expectation of a competitive effect from trade actions.
Given that targeted trade actions are aimed at a specific
firm or group of firms for specific reasons (e.g., violation of
previous agreements or policies), it seems unlikely that com-
petitors of the targeted firm would face similar actions. This
can lead to a competitive effect rather than a contagion
effect.
Given the above discussion, our hypothesis is:
H6. Trade actions targeted at a firm positively
affect the stock prices of the targeted firm's
competitors.
3|SAMPLE SELECTION AND
DESCRIPTION
To test our hypotheses, we develop our samples of ZTE tier-
one suppliers, tier-two suppliers, and customers using the
BloombergSPLC(SupplyChain)databasethatprovides
information on the nature and magnitude of business rela-
tionships between buying firms and their suppliers and cus-
tomers. Bloomberg uses multiple sources, including SEC
filings, annual reports, webpages, earnings calls, and so forth.
Each supplier is classified as either direct material (Cost of
Goods Sold); indirect material and services (Sales, General,
and Administrative); capital equipment (Capital Expendi-
ture); or research and development (R&D). When sufficient
information is available, Bloomberg (2011) reports the per-
cent of the supplier's revenues obtained from each buyer.
Bloomberg SPLC is increasingly used by academic
researchers as a source of business relationships within sup-
ply chains (e.g., Bellamy et al., 2020;Osadchiyetal.,2015;
Steven et al., 2014; Wang, Li, et al., 2021). We assembled the
data from Bloomberg SPLC in February 2019.
Because we are estimating changes in stock prices, we
require firms in our sample to be publicly traded and covered
in Global Compustat, North American Compustat, and/or
the CRSP (Center for Research in Security Prices) databases.
We also require that firms have sufficient stock price infor-
mation in the periods immediately surrounding the ban.
To study the direct effects of the ban, we develop a
sample of tier-one US suppliers to ZTE. Our sample
includes 53 suppliers classified as direct material, and
one supplier each classified as capital equipment and
R&D. Given this, we restrict our attention to direct mate-
rial suppliers. Table 1presents the descriptive statistics of
this sample. The majority of the tier-one US suppliers
(56.6%) are headquartered in California, which is not
surprising given that they are mostly in the high-tech sec-
tor. About 58.5% of the firms are semiconductor manu-
facturers. The median tier-one US supplier has annual
sales of $864 million, total assets of $1.54 billion, market
value of $2.62 billion, and 3000 employees.
To consider upstream supply chain propagation effects
of the ban, we obtain a sample of tier-two suppliers. For
each of the 53 tier-one US suppliers of ZTE, we search
Bloomberg SPLC to obtain their suppliers (tier-two sup-
pliers to ZTE). Again, we restrict our attention to only
direct material suppliers. Our search yields 295 unique
firms that supply direct material to at least one of the
53 tier-one US suppliers. Table 2Panel A gives the descrip-
tive statistics of this sample. The sample of tier-two sup-
pliers includes firms headquartered in 20 different
countries. Over 32% of the firms have corporate headquar-
ters in the United States, 29% in Taiwan, and 13% in
Japan. The most common industry for tier-two suppliers is
the same as that for tier-one US suppliers, semiconductor
manufacturing. The median tier-two supplier has annual
sales of $1.34 billion, total assets of $1.71 billion, market
value of $1.46 billion, and 4000 employees.
We assemble a sample of ZTE's tier-one suppliers of
direct material that are not headquartered in the US by
searching Bloomberg SPLC to identify a sample of 149 sup-
pliers. Table 2Panel B presents the descriptive statistics of
TABLE 1 Sample statistics for the 53 tier-one US suppliers
to ZTE
Panel A: Most frequently occurring corporate headquarters
locations (states)
State Frequency (%)
California 30 (56.6%)
Massachusetts 4 (7.6%)
Arizona 3 (5.7%)
Other states (12) 16 (30.1%)
Panel B: Most frequent industries
SIC code Frequency (%)
3674 Semiconductors 31 (58.5%)
3663 Radio & TV equipment 4 (7.6%)
6794 Patent owners 4 (7.6%)
Panel C: Statistics at the fiscal year ending prior to
April 2018
Mean Median Std. dev
Sales ($M) 6453 864 15,975
Total assets ($M) 15,261 1535 41,609
Market value ($M) 30,214 2618 106,974
Employees (000s) 14 3 29
8JACOBS ET AL.
this sample. These suppliers are headquartered in 17 coun-
tries. About 62% of the firms are headquartered in China,
followed by Taiwan and South Korea with 9% and 7%,
respectively. As with tier-one US suppliers, the most fre-
quent industry is semiconductor manufacturing. The
median tier-one non-US supplier has sales of $405 million,
total assets of $592 million, market value of $908 million,
and 4000 employees.
To estimate the downstream supply chain propagation
impacts of the ban, we develop a sample of ZTE's business
customers. Our examination of the Bloomberg SPLC data-
base results in 70 firms. Table 3Panel A presents the
descriptive statistics of this sample. Sample customer firms
are headquartered in 31 different countries, with China
accounting for over 21% of the firms, the US accounting for
over 11%, and Indonesia accounting for about 7%. Most
business customers are in the wireless or telephone com-
munications industries. The median customer has annual
sales of $5.14 billion, total assets of $9.30 billion, market
value of $6.62 billion, and 24,000 employees.
To determine ZTE's competitors, we consider all firms
in ZTE's industry, radio and television equipment
manufacturing (SIC code 3663). We include all such firms
that are not otherwise included in our samples of suppliers
or customers, and that have at least $100 million in sales
for the fiscal year ending prior to April 2018. Table 3Panel
B provides descriptive statistics for the 90 competitors.
Sample competitors are headquartered in 19 countries—
over 24% in China, almost 16% in South Korea, and over
14% in Taiwan. The median competitor has sales of $285
million, total assets of $452 million, market value of $353
million, and 1000 employees.
4|METHODOLOGY
This paper uses event study methodology to estimate the
stock market reaction to the ban. This methodology is used
to estimate the abnormal stock market returns linked with
specific events, while controlling for variables known to
influence stock prices (Brown & Warner, 1985). Abnormal
stock market returns represent the changes in stock price
due to a specific event. The Efficient Market Hypothesis
(EMH) is the methodological basis of event studies; if an
event has an effect on shareholder value, it will be reflected
quickly in the stock price.
In designing an event study, there are three important
issues: (1) selecting the event period during which to esti-
mate abnormal returns (ARs); (2) choosing the appropri-
ate method to estimate the ARs; and (3) selecting the
appropriate statistical tests to determine the significance
of the ARs. We next discuss these three issues.
4.1 |Event period selection
The ZTE ban was first reported on Monday, April
16, 2018, at 5:36 a.m. EDT (9:36 a.m. GMT). Stock
TABLE 2 Sample statistics for the 295 tier-two suppliers to ZTE, and 149 tier-one suppliers of ZTE that are not headquartered in the US
Panel A: Tier-two suppliers
Country Frequency (%) SIC code Frequency (%) Mean Median Std. dev
United States 95 (32.2%) 3674 Semiconductors 60 (20.3%) Sales ($M) 7763 1337 19,083
Taiwan 85 (28.8%) 3670 Electronic
components
20 (6.8%) Total assets
($M)
10,560 1708 28,069
Japan 38 (12.9%) 3672 Printed circuit
boards
15 (5.1%) Market value
($M)
9621 1462 22,507
Other countries
(17)
77 (26.1%) Other industries (83) 200 (67.8%) Employees
(000s)
29 4 66
Panel B: Tier-one non-US suppliers
Country Frequency (%) SIC code Frequency (%) Mean Median Std. dev
China 93 (62.4%) 3674 Semiconductors 26 (17.5%) Sales ($M) 2036 405 7264
Taiwan 14 (9.4%) 3679 Electronics, other 14 (9.4%) Total assets
($M)
4035 592 23,665
South Korea 10 (6.7%) 3670 Electronic
components
13 (8.7%) Market value
($M)
3866 908 9330
Other countries
(14)
32 (21.5%) Other industries (43) 96 (64.4%) Employees
(000s)
12 4 22
Note: Tier-two firms supply to tier-one US suppliers to ZTE.
JACOBS ET AL.9
markets in Europe, South America, and North America
were open on that day, making April 16, 2018 the first day
that investors in these markets could react to the ban. Stock
markets in Asia were already closed at the time the ban
was announced. Thus, Tuesday, April 17, 2018 is the first
day that the Asian markets could react. Using the nomen-
clature common to event studies, we translate calendar days
into event days so that the day on which the event occurred
(April 16, 2018 for the Americas and Europe; April 17, 2018
for Asia) is Day 0, the trading day immediately preceding
the event is Day 1, the trading day immediately subse-
quent to the event is Day 1, and so forth.
Stock market reaction is typically estimated in event
studies on the day of the event and the days around it
(e.g., Hendricks & Singhal, 2003; Lo et al., 2018). Using
short event periods (1 or 2 days) is usually preferred
because stock prices quickly incorporate new informa-
tion, and noisiness in stock price data can obscure results
across longer event periods. For the ZTE ban, more infor-
mation about the event was gradually revealed over sev-
eral days subsequent to Day 0. Press coverage about
potentially lifting the ban continued over several days,
with some debate within the US Congress whether the
lifting might be blocked. After 20 trading days, it was
announced that US and Chinese leaders were in discus-
sions to remove the ban.
To estimate the effects of the ban, we use a 7-day event
period (Day 0–Day 6) as our main result. Multiple factors
prompt us to focus on a 7-day event period. Our first consid-
eration is the time sequence of important milestones as the
story of the ban unfolded. On Day 0 (April 16), the US
Department of Commerce announced that it was banning
US firms from supplying ZTE for a period of 7 years. On
the same day, the UK National Cyber Security Center
warned phone carriers to not purchase ZTE goods or ser-
vices. Stock trading for ZTE was immediately suspended.
On Day 2, ZTE announced that it was delaying its earnings
report due to the ban. It was revealed on Day 4 that ZTE
did not have any appeal rights as per US Department of
Commerce regulations. To consider business press coverage
during this time, we searched the Factiva database that
compiles over 8000 business press sources. Our search of
Factiva reveals a surge in articles with ZTE in the headline
or lead paragraph for the first 7 days. Similarly, Google web
search intensity for ZTE also peaked in the first 7 days of
the ban period. Given the sequence of events, and the
timing of press coverage of the ban, our primary focus
within the ban period will be on explaining the ARs over
the first 7-day event period (Day 0–Day 6).
Since March–April 2018 was a period of increasing trade
tensions between the United States and China, there may
have been other trade-related events that potentially
TABLE 3 Sample statistics for the 70 business customer of ZTE, and 90 competitors of ZTE
Panel A: Customers
Country Frequency (%) SIC code Frequency (%) Mean Median Std. dev
China 15 (21.4%) 4812 Wireless
communications
27 (38.0%) Sales ($M) 18,892 5136 31,652
United States 8 (11.4%) Total assets
($M)
54,716 9295 143,382
Indonesia 5 (7.1%) 4813 Telephone
communications
16 (22.5%) Market value
($M)
38,645 6622 86,054
Other
countries
(28)
42 (60.1%) Employees
(000s)
64 24 98
Panel B: Competitors
Country Frequency (%) SIC code Frequency (%) Mean Median Std. dev
China 22 (24.4%) 3663 Radio & TV
equipment
manufacturing
90 (100%) Sales ($M) 3795 285 24,351
South Korea 14 (15.6%) Total assets
($M)
6040 452 39,855
Taiwan 13 (14.4%) Market value
($M)
10,802 353 83,196
Other countries
(16)
41 (45.6%) Employees
(000s)
61 20
10 JACOBS ET AL.
confound our analysis. Egger and Zhu (2020) and Wang,
Wang, et al. (2021) both chronicle major events affecting
United States–China trade tensions during that timeframe,
and neither of them list any dates during our event period.
The closest dates they identify are April 4, 2018 (Day 8rela-
tive to April 16, 2018) when the Chinese government
announced import duties on US products, and June 15, 2018
(Day 43 relative to April 16, 2018) when the US Trade Repre-
sentative released a revised list of Chinese products subject to
tariffs. To verify this, we searched the business press through-
out the event period to look for other major trade-related
events but found none. The lack of any other significant
trade-related events during our event period suggests con-
founding events are not a concern.
We also consider the effects of the ban reversal. The first
announcement about the potential reversal was released on
Sunday, May 13, 2018, at 11:46 p.m. EDT (3:46 a.m. GMT)
when the US President announced that he was working
with the President of China to reverse the ban. All global
markets could react to this announcement and trade on
this information beginning on Monday, May 14, 2018.
Thus, May 14, 2018 is designated as Day 0 for the ban
reversal period. Over the next few days, more information
was released about the reversal. On May 22, 2018, the
United States and China agreed on a broad outline of a
deal that would allow ZTE to resume buying from US
firms. May 22, 2018 is Day 6, the sixth trading day after
May 14, 2018. To estimate the effects of the ban reversal,
we use a 7-day event period (Day 0–Day 6).
4.2 |Estimation of abnormal stock
market returns
Event studies (e.g., Brandon-Jones et al., 2017;
Hendricks & Singhal, 2003) commonly utilize the market
model as a means to estimate ARs. The relationship over
time between the stock return of a firm and the stock
return of the market is expressed by the market model as:
Rit ¼αiþβiRmt þεit ð1Þ
where, R
it
is the return of stock ion Day t;α
i
is the inter-
cept; β
i
is the systematic risk of stock i, commonly
referred to as beta; R
mt
is the market return on Day t; and
ε
it
is stock i's error term on Day t. The return of a stock is
divided in two portions: β
i
R
mt
, the portion explained by
the stock market movement; and ε
it
, the portion that can-
not be explained by the stock market movement. The ε
it
term captures the effect of information that is specific to
the firm.
Given that the firms in our sample are traded in mul-
tiple stock markets, we generate each stock i's R
mt
by
using the appropriate market index. We use the domi-
nant stock market index in each sample firm's country to
estimate Equation (1). For firms headquartered in the
US, we employ the CRSP index (Value Weighted). For
the non-US firms in our sample, some examples of indi-
ces that we employ include the SSE Composite for the
Shanghai Stock Exchange, the SZSE 1000 for the
Shenzhen Stock Exchange, and the TAIEX (Taiwan Capi-
talization Weighted Stock Index) for the Taiwan Stock
Exchange.
Ordinary least squares (OLS) regression is used to
estimate the coefficients b
αiand b
βiover a period that just
precedes the event date. This estimation period comprises
200 trading days, starting on Day 210 and ending on
Day 11. For robustness, we also employ estimation
periods of 253 days (a full trading year) ending either on
Day 11 or Day 20. The results from using these differ-
ent estimation periods are similar and so we do not
report them for brevity. For firm ion Day t, its AR is den-
oted as A
it
; it is calculated by subtracting the expected
return from the actual return. For firm ion Day t, the
expected return is estimated as b
αiþb
βiRmt
. Thus,
Ait ¼Rit b
αiþb
βiRmt
ð2Þ
For Day t, the mean AR is:
At¼X
N
i¼1
Ait
Nð3Þ
where, Nis the quantity of firms in the Day tsample.
To determine the effects of the ZTE ban, we must esti-
mate the abnormal stock returns for our sample firms
during the same trading days. This has the potential to
result in cross-sectional dependence between the sample
firms' returns. To compensate for any potential cross-
sectional dependence, we employ the cross-sectional
dependence adjustment test (see, for example, Brown &
Warner, 1985). The first step in the procedure is to com-
pute the mean AR ðAÞduring the estimation period as:
ðAÞ¼Pt¼11
t¼210At
200 ð4Þ
Next, we calculate the estimation period standard devia-
tion of the mean daily ARs as:
b
SAt
¼ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
Xt¼11
t¼210 AtA
2=199
rð5Þ
The test statistic TS
t
is estimated for Day tas:
JACOBS ET AL.11
TSt¼At=b
S At
ð6Þ
We employ t-tests to establish whether the mean ARs are
statistically significant. For any given j-day event period
(t
1
,t
2
), the cumulative abnormal return (CAR) is:
CAR t1,t2
ðÞ¼
X
t2
t¼t1
Atð7Þ
For the j-day event period, the test statistic TS
j
, is esti-
mated similarly to the test statistic for a single day:
TSj¼Pt2
t¼t1At
b
SAt
ffiffij
pð8Þ
Since the AR results are often skewed, the means might
be influenced by outliers. Thus, we also report medians
and the percentage of negative ARs in our results tables.
We utilize the Wilcoxon signed-rank test to conclude if
the median ARs significantly differ from zero. For the
percent negative ARs, we use the binomial sign test to
determine if their difference from the null value of 50% is
statistically significant. In all cases, two-tailed p-values
are reported for our tests of mean ARs, median ARs, and
percent negative ARs.
5|RESULTS
In our analyses, the market reaction for the event period
is the cumulation of the market reaction over 7 days that
make up the event period (Day 0–Day 6). Although our
primary focus is on the event period, we report the mar-
ket reaction for some of the early days in our event period
to yield insights about the evolution of the initial market
reaction. However, we note that since the information
about the event is released over a few days, the daily mar-
ket reaction can fluctuate. Daily market reactions may
not be significant for individual days but could be signifi-
cant when cumulated over the days that make up the
event period. The reverse can also happen where market
reaction on some days could be significant but may not
be significant when cumulated over the event period.
Our theorizing and testing of hypotheses are with respect
to the event period and not for the individual days that
makes up the period of interest. Thus, the relevant results
for hypothesis testing are the market reaction
cumulated over the days that make up the event period
(Day 0–Day 6).
The strongest support for a hypothesis of a significant
market reaction is when the parametric and non-
parametric tests are all significant (MacKinlay, 1997).
Campbell and Wasley (1993) find evidence that non-
parametric tests provide more reliable inferences for daily
AR data on the NASDAQ. Thus, if the non-parametric
tests support a hypothesis but the parametric tests do not,
this is considered to be some evidence to support the
hypothesis. The weakest support for a hypothesis is when
only the parametric test is significant, but the non-
parametric tests are not. This case typically results
because ARs are skewed, and it is possible that outliers
are driving the result. Given this, McWilliams and Siegel
(1997) note that that relying solely on parametric tests is
problematic Accordingly, although we present both the
parametric and non-parametric tests, we put more weight
on the results of the non-parametric tests to test whether
a hypothesis is supported or not.
5.1 |Direct effects of the ban
We begin our empirical analyses by estimating the effect
of the ban on ZTE's stock price. ZTE's major stock issue
is traded on the Shenzhen Stock Exchange. To check for
any leakage of the news about the ban, we estimated
ZTE's AR for the 2 days before April 17, 2018. The ARs
for these 2 days were 0.29% and 0.86%, both insignifi-
cantly different from zero. This lack of a significant result
in the days prior to Day 0 indicates that investors were
not anticipating news about the ban.
The Shenzhen Stock Exchange halted trading on ZTE
stock beginning on April 17, 2018 and did not resume
trading until June 13, 2018, after the ban was reversed.
Thus, it is not possible to directly compute the impact on
ZTE's stock price for Days (0, 6). However, ZTE stock
price dropped on resumption of trading and continued to
drop over the next several days. After eight trading days,
the stock price largely leveled off after suffering a CAR of
64.98%. However, we note that since the ban had
already been reversed before trading resumed, the impact
of the ban imposition would likely have been even more
severe.
5.1.1 | Stock price reaction for ZTE tier-one
US suppliers
To test our first hypothesis that trade actions against a
targeted firm have a negative effect on the stock prices of
the firm's suppliers in the targeting country, we examine
the stock price performance of ZTE's 53 tier-one suppliers
of direct material that were headquartered in the
US. These suppliers were directly affected by the US gov-
ernment ban on ZTE. Table 4presents these results.
To check for any leakage of the news about the ban,
Panel A of Table 4reports ARs for the 2 days before April
12 JACOBS ET AL.
16, 2018. The ARs for these 2 days are insignificantly dif-
ferent from zero. The mean and median CAR over Days
(2, 1) are 0.30% and 0.42%, respectively, both insignifi-
cantly different from zero. About 40% of firms experience
a negative CAR, insignificantly different from the null of
50%. Insignificant abnormal performance on the days
prior to Day 0 suggest no leakage of the news about the
ban for ZTE's tier-one US suppliers, and that the ban was
a surprise to the US market. Our subsequent analyses
confirm no significant abnormal market reactions prior
to the ban announcement for any of our samples. Hence,
hereafter we report ARs and CARs with results beginning
on the day of the ban announcement (Day 0).
Panel B of Table 4presents the results from the ban.
On Day 0, the mean AR is 2.69% and the median AR is
1.29%; both are significant at the 1% level. Almost 87%
of the firms experience a negative AR, which significantly
differs from 50% at the 1% level. On Day 1, the mean and
median ARs are positive but only the median is signifi-
cant at the 5% level. Over 66% of the firms experience a
positive AR, which significantly differs from 50% at the
5% level. The results suggest that on Day 1 there is some
reversal of the negative ARs observed on Day 0. On Day
2, the mean and median ARs are negative but only the
median is significant at the 5% level. Over 64% of the
firms experience a negative AR, which differs signifi-
cantly from 50% at the 5% level.
The CARs are clearly negative and significant over
Days (0, 6), the event period. For Days (0, 6), the mean
(median) CAR is 4.13% (3.33%), significant at the 10%
(1%) level, and more than 81% of firms experience a nega-
tive CAR, significantly different from 50% at the 1% level.
These results support H1.
To better evaluate the magnitude of the ARs from
supply chain disruptions due to the ZTE ban, we com-
pare our results with other event studies. Allen (2021)
finds a mean stock market reaction of 2.20% for US sup-
pliers affected by US trade sanctions against China. The
more negative reaction that we find could be because the
ZTE action was an outright ban rather than sanctions as
in Allen (2021). With sanctions, licenses are often issued
by the US government to permit otherwise-affected US
firms to be exempted. Sanctions are typically not as
severe as outright bans.
We also compare our results with event studies that
have examined other types of supply chain disruptions.
Hendricks and Singhal (2003,2014) measure the market
reaction from endogenous, firm-specific supply chain dis-
ruptions from production problems, parts shortages, and
so forth. We obtained the original sample from the
authors and we used the same process as in Hendricks
and Singhal (2003,2014) to update it by obtaining addi-
tional announcements from 2004 to 2014. Using the same
event study methodology as outlined in Hendricks and
Singhal (2003), we find that the mean market reaction
based on the expanded sample of 1868 firms is 7.83%,
more negative than the market reaction to the ZTE ban.
We conjecture that the stock price effects of endogenous
supply chains disruptions are typically more negative
than those of trade actions because the firm itself is
responsible for these disruptions, and these disruptions
suggest some critical operational issues in the supply
chains that need to be addressed.
Event studies have also examined the stock price
effects of exogenous supply chain disruptions. Hendricks
et al. (2020) estimate that firms experiencing direct
TABLE 4 Abnormal returns for the 53 tier-one US suppliers to ZTE
Panel A: Two days prior to the ban announcement
Event day(s) NMean tMedian Z
a
% Negative Z
b
2 53 0.13% (0.17) 0.19% (1.21) 37.74% (1.79)†
1 53 0.16% (0.20) 0.12% (0.75) 47.17% (0.41)
(2, 1) 53 0.30% (0.14) 0.42% (1.46) 39.62% (1.51)
Panel B: Ban announcement
Event day(s) NMean tMedian Z
a
% Negative Z
b
0532.69% (3.32)** 1.29% (5.09)*** 86.79% (5.36)***
1 53 0.57% (0.70) 0.36% (2.45)* 33.96% (2.34)*
2530.50% (0.61) 0.37% (2.53)* 64.15% (2.06)*
(0, 6) 53 4.13% (1.92)†3.33% (5.27)*** 81.13% (4.53)***
Note: Event Day 0 is April 16, 2018. Two-tailed tests: †p< .10; * p< .05; ** p< .01; *** p< .001.
a
Obtained from Wilcoxon signed-rank tests.
b
Obtained from binomial sign tests.
JACOBS ET AL.13
supply chain disruptions due to the 2011 Japanese earth-
quake and tsunami had a mean market reaction of
7.19%. Jacobs and Singhal (2020) estimate that tier-one
suppliers to Volkswagen (VW) experienced a mean mar-
ket reaction of 2.69% from the 2015 VW diesel emis-
sions scandal. Hertzel et al. (2008), in their study of firms
filing for bankruptcy, find that the mean market reaction
was 1.96% for suppliers of those firms. The mean mar-
ket reaction of 4.13% that we find for US suppliers
directly affected by the ZTE ban is more negative than
the reaction reported for bankruptcies or the VW scandal,
but less negative than the reaction due to the Japanese
tsunami. The 2011 Japanese earthquake and tsunami was
characterized as the most significant disruption ever for
global supply chains. This event caused significant dam-
age to the physical assets and infrastructure of firms,
requiring significant resources and time to fully recover.
Like other exogenous disruptions, individual firms are
not responsible for trade actions, and trade actions do not
indicate that supply chains are physically broken and need to
be fixed. Further, trade actions can sometimes be quickly
reversed (as in the case of ZTE), and supply chains can rap-
idly get back to the original state. Given this, we conjecture
that supply chain disruptions from trade wars have a less
negative stock market reaction than either the endogenous
supply chain disruptions examined by Hendricks and Sin-
ghal (2003,2014) or the exogenous supply chain disruptions
from the 2011 Japanese earthquake and tsunami (Hendricks
et al., 2020).
5.1.2 | The effect of revenue dependence of
tier-one US suppliers on stock market reaction
As explained in our development of H2, we expect that
US suppliers that depend on ZTE for a greater portion of
their revenues suffer greater negative effects from the
ban than those US suppliers that are less dependent on
ZTE for their revenues. We analyze this by testing for rev-
enue dependence effects using multivariate regression
analyses. We use the CAR for Days (0, 6) as our depen-
dent variable. The independent variable we test is the
suppliers' revenue dependence on ZTE (Dependence). We
construct Dependence as the percentage of the supplier's
2018 sales revenues accounted for by ZTE.
To control for factors found by previous researchers to
influence abnormal stock market reactions (e.g., Flam-
mer, 2013; Hendricks & Singhal, 2003), we incorporate four
firm-level control variables into our analyses: firm size; prior
financial performance; debt-to-equity; and market-to-book.
Firm Size (Size) can potentially influence stock price reac-
tions because large firms can sometimes better withstand
impacts of negative events. We operationalize Size as natural
log-transformed sales in the fiscal year that ends prior to
April 2018. Prior Financial Performance (PFP) can affect
investor reactions as the investors could be more (or less)
concerned about how negative events impact firms with poor
(or strong) financial performance. We industry adjust PFP by
taking the difference between the focal firm's return-on-
assets (the ratio of operating income before depreciation over
total assets) and the median return-on-assets of its industry
(all US firms with the same four-digit SIC code). Return-on-
assets are calculated for the fiscal year that ends prior to April
2018. Debt-to-Equity (D_E) can potentially influence the
reaction of bondholders versus shareholders. D_E is
operationalized as the ratio of debt book value over the sum
of equity market value and debt book value. Debt book value
is calculated as of the most recent fiscal year that ends prior
to April 2018, and equity market value is calculated as of Day
11. Market-to-Book (M_B) represents the potential for firm
growth, and could influence the reaction of value-investors
relative to growth-investors. M_B is operationalized as the
ratio of equity market value to equity book value. We com-
pute this using the equity market value on Day 11, over the
equity book value that was reported in the fiscal year that
ends prior to April 2018.
We also include media coverage (Media)asan
additional control variable. To measure Media, we iden-
tify how many times each of the 53 tier-one US suppliers
of ZTE was mentioned in the media together with ZTE
from April 16, 2018 to April 24, 2018, which represents
the event period over Days (0, 6). The media sources
include the following major newspapers: Chicago Tri-
bune, Denver Post, Financial Times, Houston Chronicle,
Los Angeles Times, New York Times, New York Daily
News, Philadelphia Inquirer, Wall Street Journal,
Washington Post, and USA Today. Sources also include
the following newswires: Businesswire,Dow Jones Busi-
ness News, and PR Newswire (US). For each supplier, we
search these sources to obtain a count of the number of
media mentions for that supplier together with ZTE in
the same article. The mean (median) supplier had 4.2
(1.0) mentions in the media. Media is defined as the
count of total media mentions during Days (0, 6).
Since firms in our sample self-selected to supply ZTE,
our sample is non-random. Given the non-randomness, it
is possible that the error term in our regression model
might be correlated with some of our independent vari-
ables, potentially creating a bias in our regression esti-
mates. To control for this, we employ the two-step
Heckman procedure (Heckman, 1979). In the first step,
we use a probit selection model that includes variables
related to whether a firm would be a ZTE supplier or not.
This enables us to estimate the inverse Mills ratio (IMR)
for each sample firm. In the second step, we include IMR
as an additional variable in our regression model.
14 JACOBS ET AL.
For each of the 53 suppliers in our sample, we use a
three-stage process to identify a unique matching firm not
in our sample. In stage 1, we find US firms in the same
4-digit SIC code as the sample firm that are closest in size to
the sample firm. For each pair of firms identified, we com-
pute the absolute percent difference in size between the
sample firm and the matched firm. If the size difference is
less than 25%, we accept the firm identified in stage 1 as a
match. For firms not matched in stage 1, we expand our sea-
rch in stage 2 from the same 4-digit SIC code to the same
3-digit SIC code as the sample firm. We then designate the
potential matching firm as the one closest in size to the sam-
ple firm. Again, for each pair of firms identified, we com-
pute the absolute percent difference in size. If the size
difference is less than 25%, we accept the firm identified in
stage 2 as a match. For each remaining firm not matched in
stage 2, we expand our search in stage 3, this time to the
same 2-digit SIC code as the sample firm. We designate the
matching firm as the one closest in size to the sample firm.
Using this three-stage process, we match 30.2% of the firms
in stage 1, 26.4% in stage 2, and 43.4% in stage 3. The aver-
age size difference between sample firms and matched
firms is 19.3%.
For the probit selection model, we include five vari-
ables to predict whether firms would choose to be a ZTE
supplier. Since older, more established firms are likely to
have greater numbers of global customers, we control for
firm age (Age). We operationalize Age as the number of
years from the first Compustat record until the event year
of 2018. The growth versus value-orientation of firms
likely affects the customers they choose. To measure this,
we define market-to-book (M_B) as described above.
Firms' prior financial performance (PFP) can factor into
their decisions to supply more (or less) profitable cus-
tomers. We operationalize and industry-adjust PFP as
described above. Firms that are more (or less) dependent
on R&D spending might be more (or less) likely to choose
customers in the high-tech sector. We operationalize RDI
as the industry-adjusted ratio of R&D spending over total
sales at the fiscal year-end preceding April 2018. Industry
is defined as all US firms with the same 4-digit SIC code.
Last, firms with greater sales in the Asia-Pacific region
are more likely to supply Chinese-based customers such
as ZTE. We define AP as the percentage of the firms' total
sales that are reported in Asia-Pacific countries in the
Compustat Segments database for the fiscal year preced-
ing April 2018.Our selection model is:
Pr Sample Firm ¼1ðÞ¼Φβ0þβ1Ageiþβ2M_Bi
þβ3PFPiþβ4RDIiþβ5APiþεi
ð9Þ
The resulting Likelihood Ratio χ
2
is 52.63, significant at
the 1% level. We find the coefficient for M_B to be signifi-
cantly negative, and the coefficient for AP to be signifi-
cantly positive. This indicates that more value-oriented
firms and firms with greater Asia-Pacific sales are more
likely to be ZTE suppliers. The coefficients for Age,PFP,
and RDI are insignificant. Using the results of (9), we
estimate IMR for each sample firm i. Our second-step
regression model is:
CAR 0,6ðÞ¼β0þβ1Sizeiþβ2D_Eiþβ3M_Biþβ4Mediai
þβ5IMRiþβ5Dependenceiþεi
ð10Þ
As recommended by Kutner et al. (2005), we use the
WLS (weighted least squares) regression methodology
since the variances of the CAR errors are heterogeneous.
Small variance CARs produce more reliable estimates
about the regression relationships than large variance
CARs. As in Jacobs (2014), we utilize the inverse of the
market model residual standard deviations as the weights
for each case.
In Table 5, we present the results with the CAR for
Days (0, 6) as the dependent variable. We note that, of
the 53 tier-one US direct material suppliers to ZTE, only
50 firms have revenue dependence data available in
Bloomberg SPLC. Model 1 contains only control vari-
ables. Of the control variables, PFP is the most significant
(at the 1% level), indicating that better performing firms
suffer a more negative stock market reaction. Given that
changes in stock price reflect investor beliefs regarding
changes in future cash flows, investors believe that better
performing firms have more to lose from the ZTE ban.
Future cash flows comprise both future revenues and
future costs. This means that investors anticipated better
performing firms had a better chance to grow their future
business (and revenues) with ZTE and were also more
capable of reducing future costs associated with ZTE.
Hence, investors believed that better performing firms
were more harmed by the ZTE ban than their lesser-
performing counterparts. Media is marginally significant
(at the 10% level), suggesting that suppliers subjected to
greater media coverage suffer a more negative market
reaction. We note that the insignificance of IMR suggests
that selection bias is not a major concern in our analysis.
Model 2 contains both the control variables and the
independent variable of interest, Dependence. Model
2 has better fit than Model 1 (adjusted R
2
values of
37.83% and 17.28%, respectively). PFP remains significant
(at the 1% level) but Media is no longer significant. In
Model 2, the coefficient for Dependence is negative and
significant (at the 1% level), indicating that US suppliers
JACOBS ET AL.15
more dependent on ZTE experienced a more negative
market reaction. Each percent increase in ZTE revenue
dependence results in an additional 0.54% abnormal
market reaction. Our results strongly support H2.
5.2 |Effects of the ban on ZTE on other
suppliers
H3 posits that the effects of trade actions propagate
through the supply chain to not just the suppliers of a
targeted firm, but also to their suppliers. We test H3 by
examining the tier-two suppliers of ZTE that supply
direct material to the tier-one US suppliers of ZTE. As
mentioned earlier, we use the Bloomberg SPLC database
to identify suppliers for each of the 53 tier-one US sup-
pliers of ZTE to generate a sample of 295 tier-two sup-
pliers of ZTE from 20 different countries (see Table 2
Panel A).
Table 6presents the results for the tier-two suppliers
of ZTE. On Day 0, the mean and median AR are both
negative but insignificant. About 52% of the suppliers
experience a negative AR, insignificantly different from
50%. On Day 1, the median AR is again negative but
insignificant, and over 56% of the firms experience a neg-
ative AR, which differs significantly from 50% at the 5%
level. On Day 2, the results are insignificant. The mean
CAR and median CAR for Days (0, 6) are 0.84% and
0.40%, respectively, with the median significant at the
1% level. About 55% of the suppliers experience a nega-
tive CAR, which differs significantly from 50% at the 10%
level. These results provide some support for H3.
There is very limited empirical evidence on how the
negative effect of supply chain events propagates to tier-
two suppliers. An exception is Jacobs and Singhal (2020)
who study the effect of the 2015 VW scandal on tier-two
suppliers of VW. They do not find any evidence of signifi-
cant stock market reaction for a sample of about 300 tier-
two suppliers of VW. In contrast, ZTE tier-two suppliers
experience a negative market reaction from the supply
chain disruption due to the ban on ZTE, again suggesting
the importance of trade actions as a supply chain disrup-
tion with effects that can propagate upstream beyond the
first tier of suppliers.
We note that the propagation effects of the trade
action should diminish in each successive tier of the sup-
ply chain due to decreasing revenue dependence. As we
posited in H2, we expect the negative effect of trade
actions is increasing in revenue dependence. Since tier-
two suppliers are generally less revenue dependent on
the targeted firm than tier-one suppliers, the negative
effects of trade actions on tier-two suppliers should be
less than those on tier-one suppliers. As expected, the
mean and median differences in CARs for Days (0, 6)
between the tier-one US suppliers and their tier-two sup-
pliers are 2.63% and 3.30%, respectively, both signifi-
cantly different from zero at the 1% level. This suggests
that although the negative effects of the trade ban propa-
gate through the supply chain, the magnitude is greatly
reduced in lower tiers.
As described in our development of H4, we cannot
predict whether trade actions positively or negatively
impact suppliers in countries other than the targeting
country. We empirically determine this using the sample
of tier-one suppliers of ZTE not headquartered in the
US. The ban on ZTE by the US government did not apply
to these suppliers but they could be potentially affected.
As indicated in Table 2Panel B, this sample consists of
149 firms from 17 different countries. Table 7presents
the results for these suppliers. The results indicate that
on Day 2 these suppliers experience a negative and statis-
tically significant stock market reaction. However, this
TABLE 5 Regression results (WLS) for the 50 tier-one US
suppliers to ZTE with Bloomberg SPLC revenue dependence data
Variable Model 1 Model 2
Constant 10.798 8.212
(3.69)*** (2.95)*
Firm Size (Size) 0.793 0.513
(1.74) (1.26)
Prior Financial Performance (PFP) 0.224 0.227
(2.75)** (3.25)**
Debt-to-Equity (D_E) 6.917 3.099
(1.16) (0.58)
Market-to-Book (M_B) 0.272 0.212
(0.86) (0.77)
Media Coverage (Media) 0.110 0.040
(1.93)†(0.77)
Inverse Mills Ratio (IMR) 0.755 1.626
(0.427) (1.02)
ZTE Revenue Dependence
(Dependence) 0.537
(3.80)***
Observations 50 50
Fvalue 2.810** 5.259***
R
2
26.82% 46.71%
Adjusted R
2
17.28% 37.83%
Maximum VIF 2.692 2.705
Note: Event Day 0 is April 16, 2018. The dependent variable is CAR (0, 6).
Two-tailed tests: †p< .10; * p< .05; ** p< .01; *** p< .001. t-statistics
shown in parenthesis. WLS case weights are the inverted market model
residual standard deviations.
16 JACOBS ET AL.
negative reaction dissipates when we measure the market
reaction over a longer period. The CAR over Days (0, 6)
is not statistically significant. Thus, our findings fail to
support H4.
As detailed in our arguments for H4, there are a mix
of both negative and positive factors potentially affecting
suppliers in other countries. From a negative perspective,
we expected that non-US suppliers would experience lost
revenues and increased costs associated with ZTE's
reduced business prospects. From a positive perspective,
ZTE might switch its remaining business from US sup-
pliers to non-US suppliers. Non-US suppliers could also
be better positioned to gain sales to customers other than
ZTE, and particularly to other potential Chinese cus-
tomers. Such Chinese customers might be leery of sourc-
ing from US suppliers considering the US government
action. Another possibility is that Chinese suppliers,
which comprise the majority (62.4%) of the tier-one non-
US supplier sample, might receive some type of assis-
tance from their government to offset the impacts of US
government actions. Our finding of an insignificant effect
for non-US suppliers suggests that the negative and posi-
tive factors might be offsetting.
Comparing results between the tier-one US suppliers
and the tier-one non-US suppliers, the mean and median
differences in CARs for Days (0, 6) are 4.04% and
3.67%, respectively, both significant at the 1% level. The
impact of the trade ban was significantly more negative
for suppliers in the targeting country than for suppliers
in other countries.
5.3 |Downstream effects of the ban
on ZTE
H5 predicts that the effects of a trade action propagate
not only upstream to suppliers but also downstream to
customers. We test this using our sample of the 70 busi-
ness customers of ZTE. Table 8presents the results. Most
of the results for Days 0, 1, and 2 are insignificant, except
for the means on Day 0 (0.56%) and Day 2 (0.56%),
which are marginally significant at the 10% level. The
mean CAR and median CAR for Days (0, 6) are 1.07%
and 0.66%, respectively, with the mean insignificant and
the median marginally significant at the 10% level. About
65% of the customers experience a positive CAR, which
differs significantly from 50% at the 5% level. These find-
ings fail to support H5.
The arguments for H5 were that customers would
incur switching costs to find new suppliers and be forced
to switch to alternate suppliers that offered less value.
The marginally positive result that we find suggests that
there are other factors that could be driving the market
reaction to customers. To examine this possibility further,
TABLE 6 Abnormal returns for the 295 tier-two suppliers to ZTE
Event day(s) N
a
Mean tMedian Z
b
% Negative Z
c
0 295 0.07% (0.25) 0.03% (0.95) 52.54% (0.87)
1 294 0.07% (0.26) 0.21% (1.47) 56.46% (2.22)*
2 293 0.02% (0.06) 0.09% (0.66) 52.90% (0.99)
(0, 6) 292 0.84% (1.19) 0.40% (2.68)** 55.14% (1.76)†
Note: These firms supply to ZTE tier-one US suppliers. Event Day 0 is April 16, 2018. Two-tailed tests: †p< .10; * p< .05; ** p< .01.
a
Numbers vary since some firms have insufficient stock price data to estimate ARs or CARs for the different event periods.
b
Obtained from Wilcoxon signed-rank tests.
c
Obtained from binomial sign tests.
TABLE 7 Abnormal returns for the 149 ZTE tier-one suppliers not headquartered in the US
Event period N
a
Mean tMedian Z
b
% Negative Z
c
0 147 0.08% (0.28) 0.00% (0.34) 49.66% (0.08)
1 149 0.39% (1.46) 0.00% (0.84) 51.01% (0.25)
2 149 0.80% (2.98)** 0.90% (4.87)*** 69.13% (4.67)***
(0, 6) 146 0.09% (0.13) 0.34% (0.29) 48.63% (0.33)
Note: Event Day 0 is April 16, 2018. Two-tailed tests: †p< .10; * p< .05; ** p< .01; *** p< .001.
a
Numbers vary since some firms have insufficient stock price data to estimate ARs or CARs for the different event periods.
b
Obtained from Wilcoxon signed-rank tests.
c
Obtained from binomial sign tests.
JACOBS ET AL.17
we consider the two main product lines of ZTE,
smartphones and telecommunications equipment. In the
fourth quarter of 2017, ZTE's global market share was 3%
in smartphones and 10% in telecommunications equip-
ment. ZTE's major customers are consumer electronics
retailers who sell smartphones, and mobile network oper-
ators who build and operate networks, some of which
also sell smartphones.
A disruption in the supply of ZTE smartphones might
not affect the sales of smartphones by ZTE's customers as
they are likely to sell smartphones from multiple manu-
factures. Further, given that ZTE's smartphone market
share was only 3%, any shortfall in production of ZTE
smartphones could be made up by other smartphone
manufacturers, meaning that switching costs would be
minimal. Alternate smartphone suppliers perhaps pro-
vide greater value to their customers than ZTE. Increased
value could result from greater brand recognition or fea-
tures of other smartphones, superior customer service, or
perhaps reduced prices by suppliers to win the business
lost by ZTE.
On the telecommunications equipment side, many
major mobile network operators buy and operate equip-
ment from multiple telecommunications equipment sup-
pliers. This is a prudent strategy to: ensure competition
among the major telecommunications equipment sup-
pliers; incentivize them to invest in R&D and technology
development; offer competitive pricing; and provide good
service. From a risk management <