Socially Responsible Corporate Customers

Rui Dai, Hao Liang, and Lilian Ng∗

Current Version: April 15, 2018

∗Dai is from WRDS, The Wharton School, University of Pennsylvania, Philadelphia, USA; Liang from Sin-gapore Management University; Ng from the Schulich School of Business, York University, Toronto, Canada.We thank Ekkehart Boehmer, Simba Chang, Zhenhui Chen, Tao Chen, Travers Child, Hyunsoo Doh, ItamarDrechsler, Ying Duan, Caroline Flammer, Aurobindo Ghosh, Olga Hawn, Jie (Jack) He, Johan Hombert,Dashan Huang, Chuan Yang Hwang, Oguzhan Karakas, Andy Kim, Lloyd Kurtz, Mauricio Larrain, Long(Jason) Li, Weikai Li, Karl Lins, Roger Loh, Christopher Marquis, Ron Masulis, Clemens Otto, HoonsukPark, Hyeong-sop Shim, Victor Song, Choi Hyun Soo, Neal Stoughton, Alexander Verdarushko, Cong Wang,Heli Wang, Chishen Wei, Frank Yu, Xiaoyun Yu, Chendi Zhang, and seminar participants at the China Eu-rope International Business School, China International Conference in Finance (CICF), City University ofHong Kong, Deakin University, Nanyang Technological University, the Schulich School of Business, MonashUniversity, Singapore Management University, Simon Fraser University, Sungykunkwan University, TelferAnnual Conference on Accounting and Finance at the University of Ottawa, Third International Conferenceon Corporate Finance (Tokyo), and WRDS of The Wharton School, for their many helpful comments andsuggestions. Authors’ contact information: Dai: rui.dai.wrds@outlook.com; Liang: hliang@smu.edu.sg; Ng:lng@schulich.yorku.ca.

Socially Responsible Corporate Customers

Abstract

Using corporate social responsibility (CSR) ratings of 34,117 corporate customer-supplier rela-

tionships from 50 countries worldwide, we find that customers’ CSR ratings are associated with

suppliers’ subsequent CSR performance, but not vice versa and that their locations matter. Results

from quasi-natural experiments and M&A acquisitions suggest that customer activism for promot-

ing CSR in suppliers is causal. The bargaining power of firms and their network connectedness are

channels through which customers affect suppliers’ CSR practices. Finally, increasing collaborative

CSR efforts between customers and suppliers helps improve their operational efficiency and firm

valuation but increase only the customers’ future sales growth.

Keywords: Corporate Social Responsibility, Corporate Customers, Global Supply Chains, Eco-

nomic Benefits.

JEL Classification Number: G23, G30, G34, M14

More than 1,000 of the world’s largest companies ... have emissions-reduction targets

for their own operations. Now, they want the thousands of companies that supply them

with goods and services to reduce their own emissions.1

1. Introduction

Today’s changing global businesses and corporate environments have brought about a new

wave of corporate social responsibility (CSR) activities. Corporations worldwide are increasingly

integrating CSR into their business operations as consumers place growing importance on ethical,

safe and sustainable business practices, and hold corporations to high standards.2 As the above

opening quote suggests, there is mounting anecdotal evidence that many corporate customers are

concerned about not only their own CSR standards but also those of their suppliers. Some scholars

argue that the increasing popularity of CSR activities around the world is, in part, in response to the

repeated failures of laws and regulations protecting stakeholders, raising the need from stakeholders

to protect their own interests through pushing the company to engage in CSR (Benabou and Tirole,

2010; de Bettignies and Robinson, 2017). However, it is not apparent whether corporate customers,

one of the most important stakeholders, are really taking actions to push suppliers to engage in

socially responsible business practices, or whether their public mention of CSR is simply a sideshow.

Furthermore, there is limited academic research that shows that corporate customers do play an

active role in suppliers’ CSR engagements. Thus, the goal of our study is to explore whether

corporate customers are an important driver for CSR practices in global supply chains around the

world and possible channels that customers employ to influence CSR in suppliers.

Increases in economic globalization, advancements in production and information technologies,

and improvement in logistics have facilitated a dramatic growth in global supply chains in many

industries and across countries. However, as corporations exploit these expanding opportunities in

supply chains, they face new challenges to ensuring that these economically-linked firms commit

to CSR practices which are consistent with the corporations’ values. For example, Nike has long

1https://www.greenbiz.com/article/how-get-suppliers-act-climate2As of 2017, more than 9,500 corporations from 160 developed and developing countries have become participants

of the United Nations Global Compact program, a global initiative to encourage “companies to align strategies andoperations with universal principles on human rights, labour, environment and anti-corruption, and take actions thatadvance societal goals.”https://www.unglobalcompact.org/what-is-gc/participants.

1

been accused of abusive labor practices since the early 1970s when outsourcing its manufacturing to

developing economies with labor practices which would be considered illegal in the U.S. In light of

these allegations, one Nike director admitted that it is difficult for corporations to pressure suppliers

to assume responsibility for the ways their operations affect societies and the environment.3 In 2012-

2013, devastating workplace disasters (i.e., the factory fires in Pakistan and the collapse of a factory

in Bangladesh)4 together claimed more than 1,300 lives.5 Activists have raised concerns and said

international retailers, such as the Gap, Walmart, among others, need to take responsibility for

the working conditions in these factories that produce their clothes. These are few examples of

numerous cases that have drawn public outcry and questioned the role of corporate customers in

CSR across global supply chains.

To begin, we test whether socially responsible corporate customers can infuse similar socially

ethical business behavior in suppliers. Our tests exploit two unique international databases: (1)

a newly available FactSet Revere database that provides information on firm-level networks of

customers and suppliers around the world, and (2) Thomson Reuter’s ASSET4 Environmental (E),

Social (S), and Corporate Governance (G) database (ASSET4) that contains ASSET4 ratings (i.e.,

a composite firm-level CSR rating) as well as more than 750 constituent ESG ratings of global

publicly listed firms. After merging these two databases, our sample consists of 34,117 unique

corporate customer-supplier pairs from 50 countries worldwide for the period from 2003 to 2015.

Using this large international sample of corporate customer-supplier relationships, we find evidence

of uni-directional effects of customers’ socially responsible behavior on their suppliers’. Specifically,

customer CSR ratings are strongly and positively correlated with the suppliers’ subsequent CSR

performance. In terms of economic significance, a one standard-deviation-change in customer CSR

rating will generate a 4% increase in CSR performance of suppliers in the customer’s direct network.

Our key results are robust to the inclusion of a multitude of firm-level control variables, the country’s

gross domestic product per capita (GDPC), as well as different combinations of supplier-customer,

country, industry, and year fixed effects. Furthermore, the one-directional CSR effect is statistically

robust to (i) social and environmental aspects of CSR, and (ii) alternative CSR databases (i.e.,

3https://en.wikipedia.org/wiki/Nike sweatshops.4http://www.nytimes.com/2012/12/07/world/asia/bangladesh-fire-exposes-safety-gap-in-supply-chain.html.5http://www.nytimes.com/2012/09/13/world/asia/hundreds-die-in-factory-fires-in-pakistan.html.

2

MSCI Intangible Value Assessment and Sustainalytics).

However, we find that locations of customers and suppliers matter for the working of CSR

in supply chains. Customers play a crucial role in improving CSR standards at their suppliers

when the latter are from different countries, but not when they are from the same country. Firms

from the same country are typically subject to the same regulatory requirement, and perhaps they

voluntarily engage in similar CSR practices. Our findings also suggest that customers are unable to

drive CSR of suppliers in developing countries, unless catastrophic events elicit customers’ responses

to offer a platform of support for these suppliers, as implied by our results from quasi-natural

experiments. Nor do our results show that corporate customers in emerging markets can affect

suppliers’ CSR. In emerging markets, business practices are typically shaped by socio-economic

and political conditions, which have aggravated many environmental and social problems. Hence,

our results are not unexpected.

One may contend that our baseline result of CSR effects from customers to suppliers simply

captures the endogenous relationships in our sample of economically-linked firms. It is possible that

customers selectively choose suppliers that are likely to align their CSR engagements with theirs.

Alternatively, it is probable that suppliers may want to win customers’ businesses by conforming

to their customers’ CSR practices prior to the link. To address this potential endogenous issue and

examine whether customers can exert pressure on supplier’s CSR beyond selection, we conduct a

multitude of tests to establish causality. We employ several quasi-natural experiments of exogenous

shocks to societal demand for CSR to test the causal relationship. Specifically, we focus on a number

of unexpected global corporate scandals and disasters (i.e., Nike’s first admission to worker abuse

scandal in 2005, the deadly collapse of a garment factory in Bangladesh in 2013, the Chinese

milk scandal in 2008, and Takata/Toyota recalls in 2013). Our results show stronger corporate

customer effects of CSR on suppliers following these unexpected global shocks. In terms of economic

significance, a one-standard-deviation increase in the customer’s product responsibility rating in the

scandal year will generate a 9.1%-10.9% rise in the supplier’s mean product responsibility rating.

Similarly, a one-standard-deviation rise in the customer’s human rights rating associated with

workplace disaster and worker abuse scandal will lead to increases of 2.7% and 19.8%, respectively,

in the supplier’s mean human rights rating. The large increase in the latter perhaps suggests that

3

the labor abuse scandal must have greatly shocked the world as customers put substantial efforts

to improve CSR practices along the supply chain. Finally, we implement an identification strategy

using a corporate event, particularly M&As, where target firms are a source of exogenous variation

of the CSR effect. When a supplier or a customer is targeted (i.e., such firms are targeted not

by their own choice) and successfully acquired by another firm which is not part of the supplier

chain, the stakeholder effect of CSR becomes weaker or statistically insignificant. Combined, these

findings allay potential endogeneity concerns on the impact of corporate customers on the socially

responsible behavior of suppliers.

Next, we investigate two potential channels through which customers drive CSR in suppliers.

The first possible channel is through the bargaining powers of customers and suppliers. We argue

that the bargaining power of a customer depends on its reliance of relationship-specific investment

(RSI) made by its supplier and the competition intensity of an industry. When the customer

depends heavily on its supplier’s RSI, it has less power to impose greater, typically costly, CSR

commitment on the supplier. Prior literature suggests that customers from research-intensive in-

dustries tend to involve in specialized inputs that require their suppliers to make investments

consistent with their own (e.g., Armour and Teece, 1980; Levy, 1985; Allen and Phillips, 2000;

Dhaliwal, Shenoy, Williams, 2016; Chu, Tian, and Wang, 2017). Following this strand of literature,

we employ a supplier’s level of R&D and number of patents registered as measures of its RSI.

Similarly, we expect a customer to be powerful when its industry is more concentrated, or when

its supplier’s industry is highly competitive. The results suggest that customers are less inclined to

affect their supplier’s CSR performance when the supplier is highly innovative, or when the sup-

plier’s Herfindahl-Hirschman index (HHI, a measure of industry competitiveness) is low, or when

the customer’s HHI is high.

The other mechanism is network connectedness. Existing studies suggest that common own-

ership produces positive externalities as shareholders aim to maximize the value of firms in their

portfolio as opposed to individual firm value. For example, Freeman (2017) provides evidence that

common institutional ownership strengthens customer-supplier links and have synergistic effects on

the related firms. Therefore, we expect that common ownership in both the customer and supplier

would promote or facilitate CSR propagation from the customer to the supplier. Furthermore,

4

our analysis also looks at the impact of interlocking directorates, where a member(s) of a cus-

tomer’s board of directors also serves on its supplier’s board. We expect interlocking directorates

to facilitate greater alignment of CSR practices of the customer and supplier. Our evidence sup-

ports our expectation of greater CSR effects in economically-linked firms with common owners and

interlocking directorates.

Finally, we examine the economic implications of collaborative CSR efforts between customers

and suppliers that arise from the customer effect of CSR on suppliers. Previous studies show value

enhancements in corporations that implement CSR initiatives, such as issues related to human

rights, the community, the environment, and the treatment of employees (e.g., Dowell, Hart, and

Yeung, 2000; Gillan et al., 2010; Edmans, 2011; Kruger, 2015; Ferrell, Liang, and Renneboog,

2016). However, implementing these CSR initiatives is costly and has negative financial implications

(e.g., greater cost structure and agency problems) for their corporations (Balotti and Hanks, 1999;

Masulis and Reza, 2015). Unlike these studies that focus on corporations’ own CSR activities and

performances, our analyses look at the economic impact of collaborative CSR efforts of customers

and suppliers through their alignment of CSR standards. The increase of collaborative efforts helps

improve operational efficiency and firm valuation for both the customer and supplier but enhance

only the customer’s future sales growth.

Our research makes two significant contributions to the literature. First, our paper represents

the first to examine the role of a specific group of stakeholders – corporate customers – in propa-

gating CSR along global supply chains, and to show that the propagation is uni-directional from

customers to suppliers only. While this evidence is interesting on its own, our study further ad-

dresses a more important question: What are the economic forces that drive this behavior?6 Such

analyses contribute to our understanding of how CSR gets transmitted around the world. Existing

studies attribute CSR to a firm’s strategic pursuit for superior financial performance (Flammer,

2015a), or a manifestation of agency problems (Masulis and Reza, 2015; Cheng, Hong, and Shue,

2016). Recently, researchers begin to investigate how a firm’s surrounding environment, such as

national institutions (Ioannou and Serafeim, 2012; Liang and Renneboog, 2017) and interactions

6A contemporaneous study by Schiller (2018) shows that the global supply chain acts as a mechanism throughwhich CSR spills over from customers to suppliers. Instead, our analysis shows that the stakeholder bargainingpower and the network connectedness in customer-supplier relationship are two important channels through whichcustomers drive CSR in suppliers.

5

with other firms (Flammer, 2015b; Cao, Liang, and Zhan, 2018), plays a role in CSR. However,

little is known about how CSR is influenced by economically-linked stakeholders. Our focus on an

important type of stakeholders, namely corporate customers, helps to reconcile some puzzles in the

emerging CSR literature, especially why firms often engage in costly CSR activity. The fact that

such activity is increasingly prevalent worldwide may be a result of forces by other market players,

such as powerful customers. This is especially the case when societal demand for CSR becomes

greater following numerous CSR-related scandals in recent years. Our findings not only enhance our

understanding on what drives CSR but also, more generally, shed light on non-economic incentives

and practices of modern corporations around the world.

Second, our research contributes to the understanding of how corporate policies and behavior

spillover along global supply chains and the value implications of such spillovers. It also expands

the supply chain literature, such as the spillover of corporate tax avoidance (Cen et al., 2017),

innovation knowledge transfers (Chu, Tian, and Wang, 2017), and information diffusion along

supply chains (Cen, Doidge, and Schiller, 2016; Cen, Hertzel, and Schiller, 2017). By focusing

on international corporate customer-supplier relationships, our study joins this strand of literature

and further demonstrates that some corporate behaviors, such as CSR, propagate uni-directionally

across countries, except those emerging ones. These institutional and firm-level nuances are often

overlooked in the extant literature.

2. Data and Summary Statistics

This study employs data from several different sources: (i) information on the global network of

customer-supplier relationships from the FactSet Revere global supply chain data obtained through

the Wharton Research Data Services (WRDS); (ii) information on firm-level CSR ratings provided

by Thomson Reuters ASSET4 ESG (i.e., Environment, Social, and Governance) database, together

with alternative ratings information from MSCI Intangible Value Assessment, and Sustainalytics;

(iii) M&A information from SDC Platinum from Thomson Reuters; (iv) R&D and sales informa-

tion for computing a firm’s industry concentration intensity from Worldscope, and patent data

from the European Patent Office’s worldwide Patent Statistical Database (PATSTAT); (v) interna-

6

tional ownership information from the FactSet Global Ownership data, available from WRDS; (vi)

records of interlocking directorates from BoardEx company-level networks data from WRDS, which

covers over 550,000 interlocking individuals worldwide; and (vii) control variables from Datastream

Worldscope. The definitions of all key variables are depicted in Appendix Table A.3.

2.1. Global economic links

FactSet Revere offers a unique database of supply chain relationships that identifies companies’

interrelationships and a comprehensive geographic revenue exposures, starting from April 2003.

It covers about 23,400 global companies, whose information is culled from company regulatory

filings, websites, and daily updates based on new filings, press releases, and corporate actions

releases. FactSet Revere gathers information on corporate direct relationships disclosed by the

reporting company and on indirect relationships not disclosed by the reporting company but by

companies doing business with the reporting company. For example, their public sources of US

firms include regulatory filings (e.g., 8-K, 10-Q, and 10-K), investor presentations, websites, and

press releases. One advantage of Factset Revere data is that they contain information of both

major and minor private and publicly-listed customers, as well as their identities. To illustrate the

information contained in the FactSet Revere database, Figure 1 shows in 2013, a snapshot of BMW

with some examples of its suppliers from the U.S., the Euro markets, Canada, China, Japan, South

Korea, Mexico, and 87 other suppliers worldwide. For example, Alfa is BMW’s supplier in Mexico,

Hankook, Hyundai, and Mobiis are its suppliers in South Korea, and Baosteel in China. Under

Regulation SFAS No. 131, firms are required to disclose any major customer that represents at

least 10% of the firms’ total reported sales. Unlike FactSet Revere, the Compustat segment data,

which are commonly employed in existing studies, obtain the supply chain relationship information

only from companies’ annual 10-K filings and hence, contain a revenue distribution of firms’ major

customers. A critical limitation of Compustat segment data to a worldwide CSR research is that

it only collects information of significant customers for suppliers that are required to file 10-K (or

equivalent) to the SEC.

We merge FactSet Revere data with other sources of data, mentioned above, and our final

sample consists of 34,117 unique corporate customer-supplier pairs, with customers and suppliers

7

from 50 different countries worldwide. Columns (1)-(2) of Appendix Table A.1 list the numbers of

suppliers and customers in our final sample, by year, and Appendix Table A.2 presents the same

by country. In Appendix Table A.1, the numbers of suppliers and customers are increasing over

time from 5,484 and 5,388 in 2003 to 23,556 and 20,749 in 2015, reflecting the growing coverage

of firms by FactSet Revere and the expanding network of supply chain relationships. As seen in

Appendix Table A.2, there are 78,823 suppliers and 82,113 customers in our sample. Morocco has

the smallest number of suppliers with CSR ratings (i.e., 4), whereas the U.S. has the largest (i.e.,

38,546). On the other hand, the number of customers is the smallest in Egypt and Kuwait (i.e., 2)

and is the largest in the U.S. (i.e., 43,741).

2.2. Global CSR Ratings

Thomson Reuters’s ASSET4 database provides ESG ratings of more than 6,000 publicly-listed

companies worldwide and are employed in studies such as Ferrell, Liang, and Renneboog (2016),

Dyck et al. (2017), Hsu, Liang, and Matos (2017), among others. The database focuses on global

companies whose stocks are members of the S&P 500, Russell 1000, NASDAQ 100, MSCI Europe,

FTSE 250, ASX 300, STOXX 600, the MSCI World Index, the MSCI Emerging Market index,

among other major equity indices. The ratings consist of more than 750 environmental, social, and

corporate governance data points, including all exclusion (ethical screening) criteria and all aspects

of sustainability performance. Every data point goes through a multi-step verification process,

including a series of data entry checks, automated quality rules, and historical comparisons. These

data points reflect more than 280 key performance indicators and are rated as both a normalized

score (0 to 100) and the actual computed value. Note that the ASSET4 score assigned to each firm

is evaluated based on the firm’s CSR performance relative to the CSR performance of all firms in

the same industry around the world; in other words, the ratings are industry-benchmarked. All

ratings are provided at the corporate subsidiary-level and on a yearly basis. The information is

available from 2002 onwards, and our sample period covers from 2003 to 2015.

For all companies, at least three years of historical information are available, and most companies

have coverage for at least 10 years. It is worth mentioning that firms in the ASSET4 sample

are rated based on both their ESG compliance (regulatory requirements) and ESG engagement

8

(voluntary initiatives), and the effectiveness of their endeavor. Therefore, a firm’s CSR rating

or score reflects a comprehensive evaluation of how the firm engages in stakeholder issues and

complies with regulations. Columns (3)-(4) of Appendix Table A.1 show average composite values

of supplier (CSRS) and customer CSR (CSRC) scores, respectively, by year. These statistics suggest

that customers, on average, are socially more responsible than suppliers. In Table 1, we present

a more detailed distribution of CSRS and CSRS , with their means, medians, standard deviations,

minimum and maximum values, and their values at 25th and 75th percentiles. In addition, we

also report ESG component scores, particularly product responsibility and human rights ratings.

Except for product responsibility ratings, CSR and ESG component scores are generally greater for

customers than for suppliers, further confirming that suppliers tend to be socially less responsible,

compared to customers.

For robustness, our study also employs two alternative firm-level ESG databases, namely MSCI

ESG Research Intangible Value Assessment (IVA) and Sustainalytics’ ESG Research & Ratings

(Sustainalytics). Both databases provide research, ratings, and analysis of companies’ risks and

opportunities arising from ESG factors. IVA ratings are between 0 and 10 and Sustainalytics’ are

between 0 and 100. Both ratings gauge how well companies manage CSR issues that are related

to their businesses and provide an assessment of firms’ ability to mitigate risks and capitalize on

opportunities. Similar to ASSET4, these two alternative ratings are also industry-adjusted. In a

subsequent section, we show that our baseline evidence is robust to CSR ratings provided by the

two alternative databases.

2.3. Channel variables

Our study explores whether the stakeholder bargaining power and network connectedness are

two possible channels through which customers can drive CSR in suppliers. To assess the de-

gree of bargaining power of customers, we look at a supplier’s investment specific to the needs

of its customers and the intensity of industry competitiveness. Our study measures the sup-

plier’s relationship-specific investment by the supplier’s R&D and number of patents registered

to measure the level of innovation. A firm’s R&D information is obtained from Datastream World-

scope, whereas patent information is available from PATSTAT, which contains information on

9

patents awarded to companies, individuals, and other institutions. We employ a firm’s Herfindahl-

Hirschman index (HHI) to measure the degree of industry competitiveness. Network connectedness

is measured by the network of institutional ownership and directorates common in both the supplier

and customer. Common ownership holdings information is obtained from the FactSet ownership

database. FactSet gathers US institutional holdings from mandatory quarterly 13F filings with the

Securities and Exchange Commission (SEC) and holdings of non-US equities from sources such as

national regulatory agencies or stock exchange announcements (e.g., the Regulatory News Service

in the U.K.), local and offshore mutual funds, mutual fund industry directories (e.g., European

Fund Industry Directory), and company proxies and annual reports. Our analysis employs the

percentage of shares held in customer and supplier firms by common owners and number of com-

mon owners in a customer-supplier pair. Our interlocking directorates (and executives) records

are obtained from BoardEx company-level networks data from WRDS, which covers over 550,000

interlocking individuals worldwide.

Summary statistics of these channel variables are shown in the second panel of Table 1. The

statistics suggest that suppliers tend to invest more in R&D than their customer counterparts. The

mean (median) ratio of R&D relative to total assets is 0.5 (0.04) for suppliers and 0.04 (0.02) for

customers. While there are significantly fewer firm-year observations for the number of registered

patents, suppliers have slightly larger R&D and number of registered patents than customers. The

average percentage of common ownership held is 1% and the average number of common owners is

135.34. The mean number of common directors is 0.06 with the mean number of board positions

held by common directors is 0.07.

2.4. Control Variables

Our analysis controls for firm-level covariates commonly employed in the existing literature,

such as leverage, return on assets (ROA), Tobin’s Q, total assets (TAssets), closely held shares

(CloseShares), and sales growth (SalesGrowth). Given the cross-country variation of our data,

we control for country-level GDP per capita (GDPC) obtained from World Bank Indicators. The

distribution of these control variables is displayed in the bottom panel of Table 1. The descriptive

statistics show that on average, suppliers have marginally lower ROA and total assets, while greater

10

Tobin’s Q, the level of closely-held shares, sales growth, and GDPC, compared with those of their

customers.

3. CSR and Customer-Supplier Relationships

In this section, we examine whether corporate customers affect social and environmental engage-

ments at their economically linked firms – the suppliers. Specifically, we investigate the CSR effect

along the supply chain and determine whether any evidence of such effects is robust to a specific

component (i.e., environmental, social, or governance aspect) of CSR, and alternative databases.

Finally, we analyze whether locations of the customer and supplier play a role in the customer

effect of CSR and whether our baseline evidence reflects how customers choose and terminate their

suppliers.

3.1. Stakeholder (‘Corporate Customer’) effects of CSR

To examine for evidence of stakeholder effects of CSR in suppliers’ CSR practices, we regress

supplier CSR (CSRS) Score on the lagged customer CSR (CSRC) Score while controlling for various

firm- and country-level characteristics, as denoted by Xk, that have previously found to affect CSR.

In our key regression below, we also control for different combinations of fixed effects (FE), namely

(1) customer industry x supplier industry fixed effect, customer country x supplier country fixed

effect, and year fixed effect, or (2) customer, supplier, industry, country, and year fixed effects

separately. Given that the results from controlling different fixed effects are qualitatively similar,

in most of our regressions following Table 2, we report only results that control for fixed effects

described in (1). The baseline regression is given by

CSRS Score(t+ 1) = a0 + a1CSRC Score(t) +K∑k=1

bkXk(t) + FE(t) + ε(t+ 1). (1)

In regression model (1), Xk include the supplier’s leverage (Leverage), return on assets (ROA),

Tobin’s Q, total assets (TAssets), closely-held shares (CloseShares), sales growth (SalesGrowth),

as well as country’s gross domestic product per capita (GDPC). All these variables are defined in

11

Appendix Table A.3. We also evaluate whether customer characteristics play a role in explaining

next period’s supplier CSR Score by expanding (1) to include the customer’s firm-level character-

istics and country’s GDPC as well. While (1) allows us to investigate the stakeholder effect on

the supplier’s CSR, we cannot rule out the possibility that there may be many large, reputable

suppliers that are socially responsible and want their customers to behave responsibly. To examine

this possible supplier effect of CSR, we estimate (1) by reversing the roles of supplier and customer

CSR ratings. Results are shown in Table 2 with reported standard errors clustered at the customer-

supplier-pair level. Columns (1)-(4) show estimates of (1), while columns (5)-(8) report those with

CSRC Score as the dependent variable and CSRS Score as the key independent variable of interest.

We employ the random-effects GLS approach to estimate the different model specifications in the

table and throughout the study.

Several interesting results emerge from the table. First, there exists strong evidence of a uni-

directional CSR effect from the customer to the supplier only, an indication that suppliers do not

have a similar influence on customer CSR score. For example, the coefficient of CSRC Score, as

shown in columns (1)-(4), is positive and statistically significant, while the coefficient of CSRS

Score, as indicated in columns (5)-(8), is small and statistically insignificant. In terms of economic

significance, a one standard deviation increase in a customer’s CSRC Score will generate, on average,

a 4% (= 0.021 × 26.73 × 4.55/64.21) improvement in the performance of its suppliers’ total CSRS

Score relative to the mean supplier CSR rating in the customer’s direct network.7 We also test

whether customers in general, or a specific group of customers have the ability to influence suppliers

to align their CSR activities and practices with theirs. To address this issue, we split the sample

into two sub-groups: major versus non-major customers. As mentioned above, under SFAS 131, it

is mandatory that firms disclose the existence and sales to principal customers representing more

than 10% of total revenues. Following the SFAS definition, we define major customers as those that

are considered important customers, hence accounting for at least 10% of a supplier’s sales. Our

Online Appendix Table OA.1 suggests that only major customers are able to influence suppliers

to conform to their standard of CSR practices. Combined, the results suggest that customers,

primarily major customers, are able to exert pressure on suppliers to commit to higher standards

7For the entire sample period, a customer has, on average, about 4.6 suppliers.

12

of CSR but not the supplier on the customer.

Second, including the customer’s control variables in column (4) has no material impact on the

coefficient of CSRC Score,8 nor does it qualitatively affect the coefficients of the supplier’s control

variables. Except for a marginally significant customer’s leverage effect, the other customer-specific

variables are all statistically insignificant. Our untabulated results produce consistent evidence

that when estimated without the presence of supplier-specific characteristics, the customer-specific

control variables still display no effect on a supplier’s CSR rating. In light of these results, adding

the set of mostly insignificant customer-level control variables to the model would simply increase

noise in the estimation, thereby mitigating the significance of our key variable. Hence, in subsequent

analyses, we do not report such estimations.

Finally, the effects of supplier-specific characteristics on supplier CSRS Score are broadly con-

sistent with those obtained in the existing literature (e.g., Ferrell, Liang, and Renneboog, 2016,

Liang and Renneboog, 2017). For example, the coefficients of leverage, closely-held shares, and

sales growth are negative and significantly associated with CSRS Score, whereas those of return on

assets, Tobin’s Q, and total assets are positive. GDPC is statistically and negatively significant in

column (1), but becomes insignificant after we include supplier-country and customer-country fixed

effects in the model, suggesting that GDPC captures country varying and fixed effects on CSRS

Score.

Furthermore, we also conduct robustness tests to verify that our key finding does not necessarily

imply that customers with weak CSR performance will induce suppliers to become less socially

responsible. To carry out our tests, we construct two additional measures of the CSR Gap; one

measures the difference between CSRC and CSRS ratings, and another is a binary indicator that

equals one if CSRC Score is greater than CSRS Score and zero otherwise. The results reported in

Online Appendix Table OA.2 show that the coefficients of both the CSRC Score and the interaction

variable (CSR Gap × CSRC Score) are positive and strongly significant at the 1% level. These

findings further corroborate our baseline evidence of a positive customer effect of CSR on supplier

CSR.

8We have also included the lagged supplier CSR in our model, and our untabulated results show that the correlationbetween the lagged customer CSR and future supplier CSR does not capture the autocorrelation in supplier CSR.

13

In summary, we find that, in aggregate, corporate customers, as a group of influential stake-

holders, play an important role in improving CSR performance of suppliers across the world. This

evidence also reflects customer activism as a disciplining mechanism in suppliers’ corporate social

responsible behavior.

3.2. Environmental, Social, and Governance Aspects of CSR

In this subsection, we investigate whether the customer effect of CSR is concentrated in a

specific aspect of corporate social initiatives, such as those pertaining to environmental (Env),

social (Soc), and governance (Gov) issues. We re-estimate model (1) by replacing the composite

CSR score by each component score (i.e., Env, Soc, or Gov rating). The results are presented in

Table 3, with the format similar to that of Table 2. In columns (1)-(3), the dependent variable

is supplier CSR component, and hence, the regression analysis also controls for supplier-specific

characteristics (Supplier Controls), as well as customer-supplier, industry, country, and year fixed

effects. However, in columns (4)-(6), the dependent variable is customer CSR component, and

the results also controls for customer-specific characteristics (Customer Controls), as well as the

different fixed effects. However, to conserve space, we only report the key coefficients with their

robust standard errors in parentheses in the table as well as the remaining tables of our study.

The findings underscore the robustness of uni-directional CSR effects in environmental and social

issues only from customers to suppliers, but show bi-directional effects in governance. Specifically,

columns (1)-(2) indicate that the greater the customer’s environmental and social ratings, the

higher is its supplier’s subsequent environmental and social ratings. The coefficients of EnvC and

SocC Scores are 0.032 and 0.013, respectively, and both coefficients are statistically significant at

conventional levels. In contrast, columns (4)-(5) show no statistically significant relation between

the supplier’s environmental and social ratings and its customer’s. Interestingly, the bi-directional

governance effects (columns (3) and (6)) imply that both the customer and supplier are concerned

about this aspect of CSR and perhaps consider the importance of governance relative to social and

environmental issues. The evidence of governance effects is consistent with the results reported in

Aggarwal et al. (2011) that governance “travels” around the world through institutional investors’

efforts to promote good governance.

14

3.3. Alternative CSR databases

Oftentimes, one contend that the key evidence of a study is specific to the sample of data it

employs.9 In other words, it is plausible that our baseline finding of uni-directional CSR effects of

customers to suppliers is attributed primarily to the ASSET4 data used in our study. The coverage

of ASSET4 data is fairly extensive, and the database is also employed in a number of important

studies, as cited in the Introduction section. Still, it is arguable that the assignment of individual

firm ratings may be biased toward the methodology ASSET4 adopts. To rule out this possible bias,

we employ two alternative CSR ratings databases, namely the MSCI IVA and Sustainanalytics

databases. We therefore replicate the baseline results of columns (2) and (6) in Table 2 using firm-

level ratings assigned by the two databases. Results in columns (1) and (2) of Table 4 are based

on MSCI IVA, whereas those of columns (3) and (4) are based on Sustainanalytics. However, the

results remain materially unaffected; the coefficients of CSRC in columns (1) and (3) are positive

and strongly significant, whereas those of CSRS in the other two columns yield inconsistent signs

and are insignificantly different from zero. Thus, these findings further underscore the robustness

of our baseline evidence that the uni-directional CSR effect along the supply chain is not specific

to the ASSET4 ratings employed in our earlier analyses.

3.4. Locations of customers and suppliers

CSR efforts along the supply chain may vary, depending on the country in which customers

and suppliers do their businesses. Such variations arise due to disparate costs and benefits faced

by firms when implementing a CSR initiative. For example, doing business in emerging economies

is probably challenging as many are characterized by “either bad or weak public governance and

administration, lack of public transparency, high levels of bribery and corruption, poor records of

human rights, inadequate environmental, safety and labor standards, and high levels of poverty and

inequality” (Nelson, 2004, p. 31). It is therefore apparent that a country’s socio-cultural orientation

with the business cultures customer firms embody will determine the level of difficulty it is for them

to influence their suppliers in emerging economies to be good corporate citizens. We argue that a

9Chatterji et al. (2016) suggest that for every CSR research, one has to cross-validate the results using severaldifferent ESG samples/data sources.

15

customer’s attempt to create certain CSR standards at their suppliers is determined by their local

socio-cultural contexts and hence, their location. Thus, we test this conjecture by examining CSR

propagation along the supplier chain, where suppliers and customers are from the same country,

or from different countries, such as developed (DEV) versus emerging (EMG) markets. We also

replicate the results of column (3) in Table 2 on a subsample of customer-supplier pairs from non-

US countries. The test is to ensure that our baseline evidence is not driven by US firms, which

form the largest portion of both suppliers and customers in the full sample (see Appendix Table

A.2). Results are shown in Table 5.

The table reveals a number of striking findings. First, the stakeholder effect of CSR is positive

and statistically significant when both the customer and supplier are from different countries, but

not statistically significant when they are from same country. Most firms from the same country

typically take regulatory and voluntary approaches to CSR issues, as well as commit to a similar

CSR standard. Hence, expectedly, there ought to be substantially less CSR variation between

customers and suppliers from the same country than from different countries. Second, our key

evidence remains qualitatively unaffected even after removing US firms from the sample (column

(3)). For instance, the coefficient of customer CSRC Score is 0.023 with standard error of 0.009,

suggesting a more pronounced stakeholder effect of CSR in global supply chains with no U.S.

customer or supplier.

In columns (4)-(5), we test whether the location of suppliers from emerging and developed

markets matter, and in columns (6)-(7), we look at the location of customers. Consistent with our

expectations, we find no evidence that customers are able to push for better CSR at their suppliers

from emerging markets, and that customers from emerging markets can significantly influence their

suppliers from other markets. The results therefore suggest that it is indeed difficult for customers

or suppliers to implement certain CSR initiatives when they are located in countries with poor

records of human rights, or inadequate environmental, safety and labor standards.

Thus, the overall evidence suggests that the locations of customers and suppliers matter for the

effect of CSR propagation along global supply chains.

16

3.5. Linked and delinked relationships

Our baseline evidence may also give rise to an alternative interpretation that customers prefer

suppliers that are more likely to engage in better socially responsible practices. To test this po-

tential selection channel, we construct a sample of customer-supplier pairs that have experienced

a change in relationships at any point in our entire sample period. For example, if a supplier

(customer) reports a relationship with a customer (supplier) for the first time in the sample period,

we classify the two firms as linked. Alternatively, if a supplier and a customer no longer report

their relationship which has previously existed in the sample period, we consider the chain to be

delinked. It is important to highlight that customer-supplier pairs that do not experience a change

in the relationship during our sample period (i.e., always linked or never linked firms) do not enter

into the sample.

To capture the change-status effect, we use two binary indicators, namely Post Link and Post

Delink. Post Link (Delink) is a binary indicator that equals 1 if the customer and supplier first

establish (severe) the relationship and 0 otherwise. We then run the following regression.

CSRS Score(t+ 1) = a0 + a1CSRC Score(t) × Post Link/Delink(t) + a2CSRC Score(t)

+ a3Post Link/Delink(t) +

K∑k=1

bkXk(t) + FE(t) + ε(t+ 1),(2)

In (2), the key independent variable is the interaction between CSRC Score and Post Link (or

Post Delink). Results shown in Table 6 reveal that the customer’s influence on newly established

supplier’s CSR becomes positive and significant after the two firms establish an economic link. As

seen in column (1), the overall customer CSR effect on supplier CSR is 0.013 (=0.020-0.007), sug-

gesting that there is selection evidence associated with the initial establishment of the relationship.

On the other hand, Column (2) shows that when supply chains are terminated, customers exhibit

no statistically significant influence on suppliers. The interaction term of CSRC Score and Post

Delink is statistically insignificant in the post delinked period, even though Post Delink itself is

still significant at conventional levels. The latter perhaps indicates that suppliers do not necessar-

ily eliminate their existing CSR activities immediately following the termination of economic links

17

with customers.

While the baseline evidence could be driven by customers having preferences for suppliers who

would be willing to assume greater corporate social responsibilities, our unreported results suggest

that the customer effect of CSR still holds even after we remove the observations when customers

and suppliers first establish their relationship. Nevertheless, in the subsequent section, we conduct

several causality tests to further corroborate the baseline finding that the CSR effect from customers

to suppliers is causal – customers do push suppliers for greater CSR performance.

4. Evidence from Global Scandals and Acquisitions

We have, thus far, established evidence of significant stakeholder effects in CSR and that cus-

tomer CSR has a statistically significant effect on future supplier CSR. However, such stakeholder

effects might be due to either customers’ ability to induce their suppliers to behave socially re-

sponsibly or the deliberate decisions of customer firms to selectively pick suppliers with a greater

likelihood to align their CSR practices with the customers’. In this section, we conduct several tests

to establish the direction of causality consistent with our baseline evidence. Specifically, we ex-

plore a number of exogenous shocks related to unexpected workplace disasters and product safety

scandals that have created global shocks to consumerism and the general public on our sample

of existing linked customer-supplier pairs. We also implement an identification strategy using a

corporate event, particularly M&As, to further evaluate the stakeholder effect in CSR.

4.1. Evidence from quasi-natural experiments

Our analysis focuses on some of the worst human rights and product safety issues or scandals in

history that have drawn global attention and heightened societal awareness and activism demanding

for more socially responsible practice along supply chains. Specifically, for human rights issues, we

look at the impact of Nike’s labor abuse and the collapse of a textile building in Bangladesh. For

product safety scandals, we focus on the Chinese milk scandal and Toyota car/Takata airbag recalls.

It is important to point out that one can alternatively use government-mandated initiatives or

18

country-level ESG regulations as an instrument variable for firm-level CSR.10 However, a critical

problem with such approaches is that these regulatory outcomes typically bear different implications

from those arising from voluntary CSR engagements. While these mandates are usually targeted

at specific aspects of CSR (e.g., the 2008 emissions reporting scheme mandated by the Australian

government) and even if it is a broader mandate that applies to the whole CSR spectrum, the

compliance information is often missing for most firms due to cross-country regulatory differences.

As Li and Wu (2017) point out, such before-and-after comparisons have limited causal power than

difference-in-differences analyses. Due to the lack of power, we avoid using government-mandated

CSR initiatives or mandatory ESG regulations as an instrument variable for firm-level CSR.

4.1.1. Human rights issues

We begin by evaluating the first lengthy 108-page report released by Nike in 2005, detailing

the company’s admissions to the widespread use of sweatshops in various Asian factories that

produce Nike’s products.11 Prior to this report, there have been numerous allegations of Nike’s

worker abuse. Human rights activists and aid groups have long criticized Nike for relying on

sweatshops and child labor to keep production costs low, and for not doing enough to improve

the poor working conditions in its supply chain. Nike’s 2005 report confirming the longstanding

allegations causes a global shock to all industries that outsource their production and operations in

developing economies or the Third World countries, and has raised corporations’ awareness about

their responsibility to monitor factory standards aimed at improving working conditions in supply

chains.

We also assess the 2013 garment factory collapse in Bangladesh that has brought the world’s

attention to worker safety issues and human costs of cheap fashion. The disaster, which killed 1,138

workers, has sparked serious questions about the fashion industry’s use of sweatshops and cheap

labor within the supply chain. Many workers were found to work long hours in overheated factories

without proper fire exits. Among the various textile businesses in this garment factory include

Phantom Apparels Ltd., New Wave Style Ltd., New Wave Bottoms Ltd., and New Wave Brothers

10Schiller (2018) employs mandatory ESG regulations that affect customers’ countries.11https://www.theguardian.com/business/2005/apr/14/ethicalbusiness.money.

19

Ltd., which produce clothing for major brands including H&M, Gap, Walmart, The Children’s

Place, Dress Barn, Primark, among others.

To test whether the above shocks strengthen customer CSR effects on supplier CSR, we follow

Liang and Renneboog (2017) and use the following difference-in-differences approach.

CSRS ComScore(t+ 1) = a0 + a1CSRC ComScore(t) × Event + a2CSRC ComScore(t)

+ a3Event +K∑k=1

bkXk(t) + FE(t) + ε(t+ 1),(3)

where CSRS ComScore is the supplier’s human rights component score,12 CSRC ComScore is its

customer’s human rights component score, Event is a binary indicator that equals 1 if it is the

specific event year (i.e., the event year for Nike’s worker abuse case is 2005, and for Bangladesh’s

garment factory collapse is 2013) and 0 otherwise,13 Xk(t) is the vector of supplier firm charac-

teristics employed as control variables, and FE represents the set of fixed effects. We conduct our

regression of (3) on a subsample of already linked suppliers in developing countries and customers

from both developed and developing countries. The results are contained in Table 7, with columns

(1)-(2) showing estimates of (3) associated with 2005 and 2013 events, respectively.

The variable of interest is the interaction between CSRC ComScore and Event indicator, and

its coefficient is positive and highly statistically significant in both cases. Their coefficient is 0.343

(standard error= 0.093) in column (1) and 0.031 (standard error= 0.012) in column (2). These

findings suggest that the effect of a customer’s human rights rating on the supplier’s next-period

human right rating is more pronounced in the event year, an indication that suppliers pay increasing

attention to human rights issues following the workers abuse report and workplace disaster. In

terms of economic significance, a one-standard-deviation increase in the customer’s human rights

rating (i.e., 31.43, Table 1) in the event year will lead to a 2.7% (= (0.031 + 0.019) × 31.43/58.56)

to 19.8% (= (0.343 + 0.025) × 31.43/58.56) increase in the supplier’s mean human rights rating

in columns (2) and (1), respectively. Clearly, corporate customers have responded strongly to

12The human rights component rating gauges a company’s management commitment and effectiveness towardsrespecting the fundamental human rights conventions. It reflects the company’s capacity to maintain its license tooperate by guaranteeing the freedom of association and excluding child, forced or compulsory labor.

13Note that the Event indicator does not capture post-event years because we expect the CSR gap would not belong-lasting after such large-scaled scandals and media exposure.

20

the labor abuse scandal in 2005 more than to the workplace disaster in 2013 and are able to

influence their suppliers to improve their CSR performance in terms of employee treatment and

workplace conditions. It is worthwhile to mention that when we interact CSRC ComScore with

other year dummies, our unreported results show that none of the coefficients of the interaction term

is statistically significant. Such placebo tests provide further reinforcing evidence of stakeholder

effects in CSR activities along the global supply chain.

4.1.2. Product responsibility

We select two major product safety events for our analysis. The first is the 2008 Chinese

milk scandal involving infant milk formula adulterated with melamine exploded when 16 infants in

China’s Gansu Province were diagnosed with kidney stones after consuming the melamine-tainted

milk powder. Subsequently, an estimated 300,000 babies in China were sick from consuming the

contaminated milk, and the kidney damage led to six deaths. The World Health Organization

claimed this incident was one of the largest food safety events it has had to deal with in recent

years, and the incident raised serious concerns about food safety, not only in China but also across

the world.

The other event is the recalls issued by Takata and Toyota in 2013. At least 7 million cars

worldwide had been recalled for defective airbags made by Takata, and potentially 42 million

cars in the United States were affected. The problem was the defective metal airbag inflators.

When deployed, the huge explosive force may disintegrate the canister of the inflator sending metal

shrapnels into the passenger cabin and possibly injuring or killing the occupants in the vehicles.

According to the Consumers Report,14 this problem resulted in 16 deaths and about 180 injuries

reported worldwide, and affected 19 automakers. Toyota recalled more than one million vehicles

sold in the United States over faulty airbags and windshield wipers. The National Highway Traffic

Safety Association has called this “the largest and most complex safety recall in U.S. history.”15

Similar to the analysis of the human rights issue in the preceding subsection, we also employ

model (3) to examine exogenous shocks arising from product safety scandals. In (3), ComScore

14https://www.thestar.com/business/2017/02/27/attorneys-say-five-automakers-knew-takata-airbags-were-dangerous.html.

15https://www.theverge.com/2016/5/4/11591724/takata-air-bags-largest-recall-nhtsa.

21

denotes a firm-level rating on product responsibility. The Event indicators are 2008 for the China

milk scandal and 2013 for the Takata and Toyota recalls. Column (3) of Table 7 shows estimates

of model (3) on a subsample of existing linked firms in the global food-related industry, whereas

column (4) reports those using a subsample of firms in the global auto industry with suppliers from

Japan and corporate customers from the auto industry in both developed and developing countries.

Consistent with those of columns (1)-(2), the coefficients of the interaction variable (i.e., CSRC

ComScore x Event) are positive and statistically significant. For example, the coefficient of CSRC

ComScore x Event in column (4) is 0.268 and is statistically significant at the 5% level. In terms of

economic significance, a one-standard-deviation increase in the customer’s product responsibility

rating (i.e., 28.04, Table 1) in the scandal year will lead to a 9.1% (= (0.268−0.082)×28.04/57.43)

to 10.9% (= (0.250− 0.026)× 28.04/57.43) rise in the supplier’s mean product responsibility rating

in columns (4) and (3), respectively. This time-series evidence suggests that the pressure from

corporate customers driving suppliers to improve their CSR becomes more intense when public

demands for responsible business practices are greater following major product safety scandals.

This evidence lends further credence to our causal inferences of our baseline finding (based on cross-

sectional analyses) that the stakeholder orientation of CSR is only from customers to suppliers.

It is important to draw some comparison of the contradicting results between columns (1)-(2) of

Table 7 and column (4) of Table 4. The former suggest that customers are able to coerce suppliers

from developing countries to improve their CSR standards, following major labor abuse scandals

and workplace disasters. In contrast, column (4) of Table 4 shows that customers are, on average,

unable to pressure suppliers from emerging economies to better their CSR performance. Given that

corporate customers are exploiting cheap workforce and low production costs in these economies,

it is probably unlikely that the margins given to suppliers are large enough to cover the latter’s

costs of implementing certain corporate responsibility initiatives. Furthermore, it is challenging

to implement for these suppliers operating in developing economies with inadequate infrastructure

and institutions and with poor education, widespread corruption, and little or no accountability.

On the other hand, it may not be surprising that in response to the spate of major business ethics

scandals, customers are more willing to put more resources for assisting their suppliers to meet

increased pressures and demands for corporate responsibility. The customers may evaluate the cost

22

of such resources and efforts to be significantly lower than the cost of a bad reputation or brand.

4.2. Identification issue – M&As

In this subsection, we resort to using an identification strategy to further verify the direction of

causality associated with our key result. We focus on target firms of M&As as a source of exogenous

variation of CSR propagation from customers to suppliers along the supply chain. The reason why

we examine only targets, and not acquirers, is that target firms are typically not being acquired

by their own choices. For example, when a supplier or a customer is acquired by a third party, the

economic link between the supplier and customer could probably be weakened after either of the

firm gets acquired by a firm with no economic link prior to the acquisition. Such an acquisition

would be different from an M&A, where a customer (supplier) intentionally merges with a supplier

(customer) to establish an upstream (downstream) integration. As a result, these M&A cases would

be more appropriately considered as an endogenous choice of upstream/downstream integration.

Our sample contains 839 customer or supplier firms with ASSET4 ratings that are M&A targets

and eventually acquired by another firm. To implement our test, we define a binary indicator, Post

M&A, which captures the supply chain relationship following the completed acquisition of a target

supplier or customer and run the following regression.

CSRS Score(t+ 1) = a0 + a1CSRC Score(t) × Post M&A S or C(t) + a2CSRC Score(t)

+ a3Post M&A S or C(t) +K∑k=1

bkXk(t) + FE(t) + ε(t+ 1).(4)

Results are shown in Table 8. In columns (1)-(2), we estimate the impact of all acquisitions on

the supplier’s subsequent CSR performance. Columns (3)-(4) and (5)-(6) show separate results

of suppliers as targets and customers as targets, respectively. Columns (2), (4), and (6) report

the effect of customers whose CSR ratings are above the median level of the sample customer

CSR rankings. If a customer or supplier is acquired by a third-party firm with no prior link to

either the customer or supplier, the CSR propagation effect becomes weaker after the supplier or

customer is being taken over. For instance, in column (1), the combined customer’s CSR effect on

the supplier’s next period CSR Score is -0.038 (= −0.051 + 0.013). A closer analysis suggests that

23

the negative effect is driven mainly by acquisitions of suppliers. As seen in columns (3)-(4), when

a supplier is acquired by another firm, its subsequent CSR practice becomes negatively associated

with its customer’s, especially when the latter is above the median level. It is plausible that the

supplier pays less attention to its CSR practice during and immediately after the acquisition process.

Alternatively, this may imply that the supplier’s acquirer is unwillingly to expend more resources

to maintain the sustainability of the supplier’s CSR standard, or that the supply chain is severed

following the acquisition. In contrast, any acquisition of a customer has no statistically significant

effect on the supplier’s subsequent CSR performance. The implication is that such acquisitions do

not break down the supply chain and hence, bear no effect on its existing CSR alignment.

Collectively, all the above tests bolster a causal interpretation of our baseline evidence that

customer CSR exhibits a significant impact on the subsequent supplier CSR.

5. The Channels

The preceding sections show that customers have the ability to influence or coerce suppliers

to commit to higher CSR standards as they do. Thus, a natural question that arises is: How do

corporate customers make their suppliers improve CSR practices? In this section, we explore two

possible channels through which corporate customers are able to influence suppliers to act socially

responsibly, and they are (i) the degree of stakeholder bargaining power, and (ii) the network

connectedness in customer-supplier relationships.

5.1. Degree of stakeholder bargaining power

The extent to which customers can exert considerable pressure on suppliers and demand for

improved CSR depends on the customers’ bargaining power. We argue that when customers have

low bargaining power, suppliers can afford to withhold and avoid incurring cost of concession to

meet the customers’ demand for better CSR practice. To test this bargaining power channel of

CSR effect along the supply chain, we look at suppliers’ investments specific to customers and the

intensity of industry competition for customers and suppliers.

24

5.1.1. Relationship-specific investments

Existing studies have established that customers from research-intensive industries tend to in-

volve in specialized inputs that require their suppliers to make transaction-specific investments,

consistent with their own investments (e.g., Armour and Teece, 1980; Levy, 1985; Allen and Phillips,

2000; Dhaliwal, Shenoy, Williams, 2016; Chu, Tian, and Wang, 2017). Extending this strand of

research, we argue that suppliers with more investments in innovation are more likely to engage

in customer-specific investments. We therefore hypothesize that the greater the supplier’s innova-

tion capacity, the customers would have less power to influence their suppliers to be more socially

responsible.

Following Chu, Tian, and Wang (2017), we employ the amount of R&D relative to total assets

and the log of the number of registered patents as proxies for a supplier firm’s investments in

innovation specific to the customer’s needs and conduct the following regression.

CSRS Score(t+ 1) = a0 + a1CSRC Score(t) × InnovS(t) + a2CSRC Score(t)

+ a3InnovS(t) +K∑k=1

bkXk(t) + FE(t) + ε(t+ 1).(5)

In (5), we evaluate the impact of a supplier’s innovation (InnovS) on the supplier’s CSR Score.16

We expect the coefficient of the interaction between InnovS and the customer’s CSR Score (i.e.,

CSRC Score) to be negative if the customer relies heavily on the supplier’s innovation. In other

words, when suppliers make large investments specific to customers’ needs, they would be in a

better bargaining position to decide whether or not to to align their CSR practices with those of

customers. The results reported in columns (1)-(2) of Table 9 are consistent with our expectations.

We find that the interaction effect of CSRC Score and InnovS is negative and statistically

significant for both proxies of the customer-specific investment. For instance, the coefficient of

CSRC Score x R&DS is -0.323 and is statistically different from zero. This finding suggests that

when a supplier’s customer-specific investment is low, the supplier is more inclined to meet the CSR

16We have also examined whether a customer’s innovation activity has any effect on a supplier’s CSR performance.Unreported results suggest that investment in innovation does not give customers strong bargaining power to driveCSR in suppliers.

25

standard of its customer. Alternatively, when there is less resource dependence in the customer-

supplier relationship, the customer would have more power to demand for a higher CSR standard

at its supplier. Thus, the evidence suggests that the extent of the customer-specific investment

is one mechanism that drives the supplier’s desire to align its CSR with that of its more socially

responsible customer.

We recognize that our findings are also consistent with a learning channel. A significant part

of CSR initiatives relates to product innovation and production process by using environmental-

friendly technologies and engaging in R&Ds that enhance product safety and responsibility. Such

processes involve learning from customers and feedback to suppliers; more innovative suppliers

would have less of a need to learn from their customers. Hence, in line with the evidence, it would

be harder for customers to influence highly innovative suppliers.

5.1.2. Intensity of industry competition

The intensity of competition in an industry is one determining force that drives the bargaining

power of suppliers and customers. A supplier has strong bargaining power if there are barriers to

entry, fewer threats of substitutes, the industry is highly concentrated or low intensity of industry

rivalry, and customers have weak or no power. Conversely, a customer has high bargaining power if

it has fewer competitors, substitutes are available, little product differentiation, and its purchases

comprise a large portion of the supplier’s sales. Thus, the degree to which customers are able

to pressure suppliers to achieve high CSR standards is determined by the intensity of industry

competition of the supplier and customer. We measure the intensity of competition using the

Herfindahl-Hirschman Index (HHI). A customer’s HHI is measured by the summation of squared

market share (based on sales) of each firm within the same industry, whereas a supplier’s HHI

is measured within the same global industry to approximate the level of competition suppliers

encounter globally by squaring the market share of the firm’s sales in the industry and then summing

the squares.17 We then estimate (5) by replacing the innovation variable with HHI and present the

results in columns (3)-(4) of Table 9.

Column (3) evaluates the effect of a supplier’s HHIS on the supplier’s CSR Score, whereas col-

17An industry is defined using Fama and French’s 48 industry classifications.

26

umn (4) assesses the impact of a customer’s HHIC . When the supplier is in a highly competitive

industry, it tends to align its CSR practice with that of its customer. The coefficient of the inter-

action variable, CSRC Score x HHIS is negative and statistically significant at conventional levels.

This finding probably suggests that the supplier may lose its customer if it does not improve its

CSR standard. On the other hand, column (4) shows that when a customer is in a less competitive

industry, it has strong bargaining power to influence its suppliers to be more socially responsible.

The coefficient of CSRC Score x HHIC is positive and statistically significant at the 5% level (i.e.,

0.048 with robust standard error of 0.023). Combined, these results suggest that the stakeholder

bargaining power plays a critical role in propagating CSR along the supply chain.

5.2. Network connectedness

Extant literature has shown that corporate behavior propagates through networks of decision-

making bodies, namely owners and directors. Common ownership exists between two firms when-

ever an investor owns shares of both firms, and two firms are board-linked whenever they have

shared directors on their boards. Such network structures are one potential mechanism for spread-

ing corporate policies from firm to firm. This subsection examines whether shared ownership and

interlocking corporate boards affect CSR efforts along the supply chains.

5.2.1. Common ownership

There is substantial robust evidence that institutional investors, who are large equityholders,

play a critical role in corporate policies of the firms they invest in. In recent years, several studies

find that institutional cross-ownership influences the outcomes of mergers and acquisitions (e.g.,

Matvos and Ostrovsky, 2008; Harford, Jenter, and Li, 2011), industry competitiveness (e.g., Azar,

Raina, and Schmalz, 2016; Azar, Schmalz, and Tecu, 2017; He and Huang, 2017), return correlations

(Anton and Polk, 2014), CEO pay incentives (Anton et al., 2016), and customer-supplier links

(Freeman, 2017). In particular, Freeman shows that common institutional ownership in both the

customer and supplier strengthens their supply chain relationship. Based on her finding, it is

plausible that common ownership is a channel through which a supplier’s CSR is aligned with its

customer’s.

27

Our analysis employs two different measures of common institutional ownership. Our first

measure computes the sum of the maximum percentage of ownership held by all common owners

in the supplier (customer) in a given year, and this measure is labeled, “% of Shares HeldS”. We

compute the same for the customer firm and label it “% of Shares HeldC”. The other measure is

based on the number of common institutional investors that invest in both firms in a given year

and is represented by “# of Common Owners”. Using these measures, we estimate the following

model.

CSRS Score(t+ 1) = a0 + a1CSRC Score(t) × OwnershipS or C(t) + a2CSRC Score(t)

+ a3OwnershipS or C(t) +K∑k=1

bkXk(t) + FE(t) + ε(t+ 1),(6)

where Ownership represents either % of Shares Held or # of Common Owners in a supplier firm

or customer firm. If common institutional investors can influence their supplier firm to engage in

socially responsible activities as their customer firm does, we expect the interaction between CSR

score and measures of common institutional ownership to be positively correlated. Results of model

(6) are reported in columns (1)-(3) of Table 10.

The table reveals one distinct evidence – the important role of common institutional investors

in aligning the CSR of their portfolio of customer and supplier firms. We find the two measures

of common institutional ownership and their interactions with the customer CSR Score to yield

positively and statistically significant effects on the next period’s supplier CSR Score. For example,

in column (1), the interaction term, CSRC Score×% of Shares HeldS , is 0.033 with robust standard

error of 0.015. Column (3) produces a qualitatively similar finding based on the number of common

institutional investors in both the supplier and customer firms. The coefficient of the interaction

term is 0.068 and is statistically significant at the 1% level. Thus, common institutional investors

do seek to mobilize investor voice towards positive social impact along the supply chain.

5.2.2. Board interlocks

Economically-linked firms can also be connected through shared directors, where directors serve

on boards of both the customer and supplier. One advantage of having shared directors is that

28

directors can act in concert to promote similar CSR practices at both corporations. We therefore

examine whether board connections through shared directors can, in part, influence CSR policies.

To evaluate this channel, we construct two measures of common directors. The first measure is

based on the log of the number of directors who serve on both the supplier’s and customer’s boards.

To cite an example of interlocking directorates, in year 2006, Rudy Provoost was an Executive Vice

President, division CEO, and board member of Philips, a Dutch technology company headquartered

in Amsterdam, and at the same time, he was the head of board of directors (Chairman) at Philips’s

supplier, LG Display Company. An alternative measure is the log number of positions held by

common board members. For example, in 2003, Garo Armen was CEO and Chairman of Agenus,

Inc., a Lexington, Massachusetts-based biotechnology company focused on immunotherapy, and

was also Board Chair and Acting CEO of Elan Corp, Agenus’s supplier located in Dublin, Ireland.

In this case, Armen held two positions at the supplier. Results are presented in Table 11. Consistent

with the findings of common ownership in Table 9, we also find that common directors serve as

another channel through which the customer firm is able to influence better CSR practice at its

supplier firms. The coefficients of both interaction terms (CSRC Score × # of Common Directors

and CSRC Score × # of Board Positions) in columns (1) and (2) are positive and statistically

significant at the 10% level.

Overall, the results provide corroborating evidence that the stakeholder bargaining position

and collaboration among common institutional owners and common directors are instrumental in

propagating CSR practices from customers to suppliers.

6. Economic Consequences of Customer Effects of CSR

In the preceding sections, we have shown that customers have a positive impact on suppliers’

CSR practices, suggesting that suppliers do respond to their customers and behave similarly in

socially responsible ways. However, a question that remains is whether there is any economic

benefit arising from customers pushing suppliers for greater social responsibilities.

Existing studies dispute whether the benefits of CSR outweigh its costs. Some studies find that

CSR initiatives can help firms build a social reputation (e.g., Fombrun, 2005), attract more produc-

29

tive employees (Burbano, 2016), exploit new markets for environmentally friendly products (Arora

and Gangopadhyay, 1995), and can be financially profitable through branding/reputation effects

on different stakeholders (Baron, 2001). Benabou and Tirole (2010) argue that CSR engagements

are beneficial to firms in the long run and help strengthen their market positions. Other studies,

however, show that adopting CSR policies is likely to increase costs and hurt firm performance,

as firms redefine their corporate social responsibilities under the pressure of various stakeholders.

CSR costs include major investment costs involving construction, equipment, or new environmen-

tal technologies and processes, permanent contributions such as scholarships, and other operating

costs of CSR implementation. Margolis, Elfenbein, and Walsh (2010) provide a meta-analysis of

the relationship between corporate social and financial performance, and document that the overall

effect is positive but small.

In this section, we ask whether customers are financially incentivized to impose better CSR

practices on their suppliers, and whether suppliers benefit from taking greater social responsibilities.

The economic implications of these collaborative CSR efforts between customers and suppliers are

closely related to the growing literature linking CSR to firm financial performance (e.g., Gillan

et al., 2010; Edmans, 2011; Deng, Kang, and Low, 2013; Servaes and Tamayo, 2013; Flammer,

2015a; Krueger, 2015; Lins, Servaes, and Tamayo, 2017). Most of these studies consider CSR

engagements as a firm’s own strategic choice, and investigate their direct and indirect effects on the

firm’s profitability and valuation. Some recent work also studies the role of various stakeholders,

such as institutional investors (Dimson, Karakas, and Li, 2015; Dyck et al., 2017) and competitors

(Cao, Liang, and Zhan, 2018) in driving CSR engagements and their value implications. Unlike

these studies, our analysis focuses on the economic consequences of the stakeholder (customer)

effect of CSR on suppliers and not of firms’ own CSR activities.

We postulate that collaborative CSR efforts between customers and suppliers would lead to

increased operating efficiency, sales growth, and firm value, probably through enhancing branding

and reputation effects, attracting more consumers, and generating greater sales. Improved CSR

standard along a supply chain may involve the increased focus of product responsibility and safety,

which in turn lower discretionary expenses, such as selling, general, and administrative expenses,18

18Kalwani and Narayandas (1995) show that maintaining long-term relationships with their customers decreasesdiscretionary expenses, such as selling, general, and administrative expenses, and hence, improves profitability.

30

through improved operational efficiency. Customers and suppliers have desires to promote a sus-

tainable relationship, through greater CSR efforts, if such efforts produce better future sales growth

and enhanced firm valuation for both the customer and supplier. Our analysis uses the ratio of

selling, general, and administrative expenses to total assets (SG&A), 3-year annualized future sales

growth, and market-to-book ratio as measures of firm performance (Performance) associated with

increased collaborative CSR efforts in a customer-supplier relationship. To evaluate the economic

benefits associated with such efforts, we run the following regression model.

Performance S or C(t+ 1) = a0 + a1CSRC Score(t− 1) × CSRS Score(t) + a2CSRC Score(t− 1)

+ a3CSRS Score(t) +

K∑k=1

bkXk(t) + FE(t) + ε(t+ 1).(7)

The variable of interest is the interaction of CSRC Score(t − 1) and CSRS Score(t) in (7). If the

CSR effect on suppliers benefits both the supplier and customer, we should expect the coefficient

of a1 to be negative for SG&A whereas positive for future sales growth and market-to-book ratio.

Estimates of (7) are shown in Table 12. The dependent variables are SG&A, sales growth, and

market-to-book ratio of the supplier in columns (1), (3), and (5), while those of its customer

counterparts are shown in columns (2), (4), and (6).

We find that the stakeholder effect of CSR generates favorable economic outcomes. It therefore

pays for customers to influence their suppliers to act socially responsibly, as such behavior has an

overall positive impact on the customer’s future performance. Customers enjoy not only improved

operational efficiency in terms of lower SG&A, but also greater future sales growth and, albeit

small,19 firm valuation. For example, the a1 coefficient is negative in column (2) but is positive in

columns (4) and (6). Similarly, it is also worthwhile for suppliers to strive for better CSR standard

that adheres to that of their socially responsible corporate customer – suppliers also experience

decreased SG&A and enhanced firm value following their adoption of improved CSR practices.

However, unlike their customers, suppliers do not experience any statistically significant increase

in 3-year annualized future sales growth, even though the sign of the coefficient is positive.

In summary, this section shows evidence of economic benefits associated with the customer

effect of CSR along the supply chain.

19In their meta analysis of several existing empirical studies, Margolis, Elfenbein, and Walsh (2010) conclude thatthe overall correlation between CSR and corporate firm performance is positive but small.

31

7. Conclusion

Our study represents the first to examine the role of a specific group of stakeholders, namely

corporate customers, as one important driver of CSR in supply chains worldwide and then to inves-

tigate the possible channels through which customers push suppliers to take greater responsibilities

for social and environmental issues. We employ two unique international databases on firm-level

networks of customers and suppliers around the world and on Thomson Reuter’s ASSET4 Environ-

mental (E), Social (S), and Corporate Governance (G) ratings of publicly-listed global firms from

50 countries for the 2003-2015 period. Using this large international sample of corporate customer-

supplier relationships, we find strong evidence of uni-directional customer effects of CSR practices

on suppliers, but not when either the supplier or the customer is from an emerging market. This

baseline result is robust to the inclusion of a multitude of firm-level control variables, the country’s

GDP per capita, social and environmental aspects of CSR, alternative CSR databases, as well as

different combinations of supplier-customer, country, industry, and year fixed effects. Further, using

several quasi-natural experiments and M&A acquisitions provides causal evidence that customers

improve suppliers’ subsequent CSR ratings.

We show two key channels that corporate customers are able to influence or pressure suppliers

to commit to better CSR standards. The first channel is through the stakeholder bargaining power,

and the other is through common ownership and interlocking directorates. Our analysis provides

supporting evidence that the two channels facilitate the customer effect of CSR on suppliers.

Finally, we find that the collaborative CSR effect delivers economic values to both suppliers

and customers. Customers have incentives to aim for better CSR at their suppliers as higher CSR

standard results in improved operational efficiency, sales growth and firm value through socially

and environmentally friendly production and through enhancing branding and reputation effects.

Similarly, suppliers also have the desire to engage in responsible business practices that adhere to

those of their customers, as such adherence contributes to improving both operational efficiency

and firm valuation.

If we view the evidence of our study at face value, several ideas emerge for the improvement

of social welfare and firm performance. As multinational corporations around the world are in-

32

creasingly interconnected via global supply chains and making significant impacts on billions of

consumers, their value creation process and social responsibilities have become the foremost issues

in public debate. Despite their importance, we still have limited knowledge on how their CSR

practices are influenced by each other and spilled over along supply chains. Therefore, a firm’s own

socially responsible practice might have a multiplier effect through economic links and generate big-

ger positive social payoffs. However, our finding of the uni-directional CSR effect from customers

to suppliers and only in some subsamples of countries suggest that such effects are bounded by a

firm’s relative position in the global network and its socio-cultural and institutional environment.

This potentially indicates that policies aiming at promoting socially responsible practices among

public companies cannot be universally applied.

Our findings also shed light on some fundamental issues in industrial organization and strategic

management, such as why some firms are incentivized to coerce others to adopt certain practices.

The answer might simply be that it makes economic sense because as corporate customers, they

can benefit from increased sales throughout the supply chain. It is therefore unsurprising that

these companies are also more responsive to product safety and human rights scandals, as these

events are closely related to consumer perception and future purchases of their products. These are

important strategic considerations for managers, especially pertaining to indirect costs and benefits

incurred by other upstream and downstream firms, when trading off their social investment and

other capital expenditures with limited corporate resources.

33

References

Aggarwal, R., Erel, I., Ferreira, M. and Matos, P., 2011. Does governance travel around the world? Evidenceof institutional investors. Journal of Financial Economics 100, 154-181.

Albuquerque, R., Koskinen, Y., and Zhang, C., 2018. Corporate social responsibility and firm risk: Theoryand empirical evidence. Management Science. Forthcoming.

Allen, J., and Phillips, G., 2000, Corporate equity ownership, strategic alliances, and product marketrelationships. Journal of Finance 55, 2791-2815.

Anton, M., Ederer, F., Gine, M., and Schmalz, M. C., 2016. Common ownership, competition, and topmanagement incentives. Available at SSRN 2802332.

Anton, M., and Polk, C., 2014. Connected stocks. Journal of Finance 69, 1099-1127.

Armour, H., and Teece, D., 1980. Vertical integration and technological innovation, Review of Economicsand Statistics 62, 470-474.

Arora, S., and Gangopadhyay, S., 1995. Toward a theoretical model of voluntary overcompliance. Journalof Economic Behavior & Organization 28, 3, 289-309.

Azar, J., Raina, S., and Schmalz, M.C., 2016. Ultimate ownership and bank competition. Available atSSRN 2710252.

Azar, J., Schmalz, M.C., and Tecu I., 2017. Anti-competitive effects of common ownership. Journal ofFinance, Forthcoming.

Balotti, R. and Hanks, J. 1999. Giving at the office: A reappraisal of charitable contributions by corpora-tions. Business Lawyer 54, 956-966.

Banerjee, S., Dasgupta, S., and Kim, Y., 2008. Buyer-supplier relationships and the stakeholder theory ofcapital structure. Journal of Finance 63, 2507-2552.

Baron, D., 2001. Private Politics, Corporate Social Responsibility, and Integrated Strategy. Journal ofEconomics and Management Strategy 10, 7-45.

Benabou, R., and Tirole, J., 2006. Incentives and prosocial behavior. American Economic Review 96,1652-1678.

Burbano, V.C., 2016. Social responsibility messages and worker wage requirements: Field experimentalevidence from online labor marketplaces. Organization Science 27, 1010-1028.

Cao, J., Liang, H., and Zhan, X., 2018. Peer effects of corporate social responsibility. Management Science,Forthcoming.

Cen, L., Doidge, C., and Schiller, C.M., 2016. Do countries matter for information diffusion in financialmarkets? Evidence from global supply-chain networks. Working Paper, University of Toronto.

Cen, L., Hertzel, M., and Schiller, C.M., 2017. Speed matters: Limited attention and supply-chain infor-mation diffusion. Working Paper, University of Toronto.

Cen, L., Maydew, E.L., Zhang, L., and Zuo, L., 2017. Customer-supplier relationships and corporate taxavoidance. Journal of Financial Economics 123, 377-394.

Chatterji, A., Durand, R., Levine, D., Touboul, S., 2016. Do ratings of firms converge? Implications formanagers, investors and strategy researchers. Strategic Management Journal 37, 1597-1614.

Chen, Y., Hung, M., and Wang, Y., 2017. The effect of mandatory CSR disclosure on firm profitabilityand social externalities: Evidence from China. Journal of Accounting and Economics, Forthcoming.

Cheng, I.H., Hong, H., and Shue, K., 2016. Do managers do good with other peoples’ money? Availableat SSRN 1962120.

34

Chu, Y., Tian, X., and Wang, Z., 2017. Corporate innovation along the supply chain. Management Science,Forthcoming.

Cronqvist, H., and Yu, F., 2017. Shaped by their daughters: Executives, female socialization, and corporatesocial responsibility. Journal of Financial Economics, Forthcoming.

De Bettignies, J., and Robinson, D.T., 2017. When is social responsibility socially desirable? Available atSSRN 3057289.

Deng, X., Kang, J.K., and Low, B.S., 2013, Corporate social responsibility and stakeholder value maxi-mization: Evidence from mergers, Journal of Financial Economics 110, 87-109.

Dhaliwal, D.S., Shenoy, J., and Williams, R., 2016. Common auditors and relationship-specific investmentin supplier-customer relationships. Working Paper, University of Arizona.

Dimson, E., Karakas, O., and Li, X., 2015. Active ownership. Review of Financial Studies 12, 3225-3268.

Dowell, G., Hart, S., and Yeung, B., 2000, Do corporate global environmental standards create or destroymarket value?, Management Science 46, 1059-1074.

Dyck, A., Lins, K.V., Roth, L., and Wagner, H.F., 2017. Do institutional investors drive corporate socialresponsibility? International evidence. Journal of Financial Economics, Forthcoming.

Dyer, J. H., and Singh, H., 1998. The relational view: Cooperative strategy and sources of interorganiza-tional competitive advantage. Academy of Management Review 23, 660-679.

Edmans, A., 2011, Does the stock market fully value intangibles? Employee satisfaction and equity prices,Journal of Financial Economics 101, 621-640.

Ferrell, A., Liang, H., and Renneboog, L., 2016. Socially responsible firms. Journal of Financial Economics122, 585-606.

Flammer, C., 2015a, Does corporate social responsibility lead to superior financial performance? A regres-sion discontinuity approach. Management Science 61, 2549-2568.

Flammer, C., 2015b, Does product market competition foster corporate social responsibility? Evidencefrom trade liberalization, Strategic Management Journal 36, 1469-1485.

Fombrun, C., and Shanley, M., 1990. What’s in a name? Reputation building and corporate strategy. TheAcademy of Management Journal 33, 233-258.

Freeman, K.M., 2017. The effects of common ownership on customer-supplier relationships. AFA 2018Paper.

Friedman, M., 1970, The social responsibility of business is to increase its profits, New York Times Magazine,122-126.

Gillan, S. L., Hartzell, J. C., Koch, A., and Starks, L. T., 2010. Firm’s environmental, social and governance(ESG) choices, performance and managerial motivation. Working paper, University of Texas at Austin.

Gofman, M., Segal, G., and Wu, Y., 2017. Production networks and stock returns: The role of creativedestruction. Available at SSRN 2981447.

Harford, J., Jenter, D., and Li, K., 2011. Institutional cross-holdings and their effect on acquisition deci-sions. Journal of Financial Economics 99, 27-39.

He, J., and Huang, J., 2017. Product market competition in a world of cross ownership: Evidence frominstitutional blockholdings. Review of Financial Studies 30, 2674-2718.

Hong, H.G., Kubik, J.D., and Scheinkman, J.A., 2012, Financial constraints on corporate goodness. Avail-able at SSRN 1734164.

Hsu, P., Liang, H., and Matos, P., 2017, Leviathan Inc. and corporate environmental engagement. WorkingPaper, Singapore Management University.

35

Ioannou, I., and Serafeim, G., 2012. What drives corporate social performance? The role of nation-levelinstitutions. Journal of International Business Studies 43, 834-864.

Joskow, P., 1987. Contract duration and relationship-specific investments: Empirical evidence from coalmarkets. American Economic Review 77, 168-185.

Kalwani, M. U., and Narayandas, N., 1995. Long-term manufacturer-supplier relationships: Do they payoff for suppliers? Journal of Marketing 59, 1-16.

Kitzmueller, M., and Shimshack, J., 2012. Economic perspectives on corporate social responsibility. Journalof Economic Literature 50, 51-84.

Krueger, P., 2015. Corporate goodness and shareholder wealth. Journal of Financial Economics 115,304-329.

Li, J., and Wu, D., 2017. Does corporate social responsibility benefit society? Working Paper. Universityof Michigan-Ann Arbor.

Liang, H. and Renneboog, L., 2017. On the foundations of corporate social responsibility, Journal ofFinance, 853-909.

Lins, K.V., Servaes, H., and Tamayo, A., 2017. Social capital, trust, and firm performance: The value ofcorporate social responsibility during the financial crisis, Journal of Finance 72, 1785-1824.

Margolis, J. D., Elfenbein, H. A., and Walsh, J. P., 2010. Does it pay to be good? A meta-analysisand redirection of research on the relationship between corporate social and financial performance.Working Paper, Harvard University.

Masulis, R.W., and Reza, S.W., 2015. Agency problems of corporate philanthropy. Review of FinancialStudies 28, 592-636.

Matten, D., and Moon, J., 2008. “Implicit and “explicit CSR: A conceptual framework for a comparativeunderstanding of corporate social responsibility. Academy of Management Review 33, 404-424.

Matvos, G. and Ostrovsky, M., 2008. Cross-ownership, returns, and voting in mergers. Journal of FinancialEconomics 89, 391-403.

Nelson, J., 2004. Leadership, accountability, and partnership: Critical trends and issues in corporate socialresponsibility, Report of the CSR Initiative Launch Event, Corporate Social Responsibility Initiative,Report No. 1. Cambridge, Mass.: Kennedy School of Government, Harvard University.

Schiller, C., 2018. Global supply-chain networks and corporate social responsibility. Available at SSRN3089311.

Schmalz, M.C., 2018. Common ownership concentration and corporate conduct. Annual Review of Finan-cial Economics, 1-29.

Servaes, H., and Tamayo, A., 2013. The impact of corporate social responsibility on firm value: The roleof customer awareness. Management Science 59, 1045-1061.

Williamson, O., 1985. The Economic Institutions of Capitalism. New York Free Press.

36

Figure 1A Snapshot of FactSet Revere Information on BMW and its Worldwide

Suppliers

87 suppliers

37

Tab

le1

Su

mm

ary

Sta

tist

ics

The

table

rep

ort

sth

enum

ber

of

cust

om

er-s

upplier

pair

obse

rvati

ons

(NO

bs)

and

sum

mary

stati

stic

sof

vari

ous

corp

ora

teso

cial

resp

onsi

bilit

y(C

SR

)sc

ore

s,firm

-lev

elch

ara

cter

isti

csand

GD

Pp

erca

pit

aem

plo

yed

as

contr

ol

vari

able

s,and

the

vari

able

sem

plo

yed

for

the

mec

hanis

ms

of

CSR

pro

pagati

on

effec

ts.

Thes

eva

riable

sare

rep

ort

edfo

rb

oth

supplier

sand

cust

om

ers.

The

CSR

vari

able

sare

rati

ngs

ass

oci

ate

dw

ith

the

com

posi

teC

SR

score

,en

vir

onm

enta

l(E

nv),

soci

al

(Soc)

,gov

ernance

(Gov

),pro

duct

resp

onsi

bilit

y,and

hum

an

rights

CSR

issu

es.

Fir

mch

ara

cter

isti

csin

clude

lever

age,

retu

rnon

ass

ets

(RO

A),

Tobin

’sQ

(Q),

log

of

tota

lass

ets

(TA

sset

s),

close

ly-h

eld

share

s(C

lose

Share

s),

sale

sgro

wth

,as

wel

las

the

countr

y’s

gro

ssdom

esti

cpro

duct

per

capit

a(G

DP

C).

The

channel

vari

able

sin

clude

R&

D,

pate

nts

,H

erfindahl-

Hir

schm

an

Index

(HH

I),

%of

share

shel

d,

num

ber

sof

com

mon

owner

sand

dir

ecto

rsin

both

cust

om

erand

supplier

firm

s,and

#of

com

mon

dir

ecto

rhold

ing

mult

iple

board

posi

tions

at

the

supplier

.A

llth

eva

riable

sare

defi

ned

inA

pp

endix

Table

A.3

.

Variablesassociate

dwith

Suppliers

Variablesassociate

dwith

Customers

Vari

able

NO

bs

Mea

nM

edia

nStd

evM

in25th

75th

Max

NO

bs

Mea

nM

edia

nStd

evM

in25th

75th

Max

CSR

Composite

andComponen

tScores

CSR

82,1

13

64.2

174.3

828.9

62.5

338.8

290.6

898.4

982,1

13

71.7

383.5

126.7

32.5

356.8

292.6

198.6

0E

nv

82,1

13

60.9

374.0

232.1

48.2

725.2

891.4

497.4

982,1

13

69.4

486.3

530.1

28.2

746.4

892.8

297.5

0Soc

82,1

13

60.1

668.5

130.3

63.5

431.5

188.2

998.8

082,1

13

69.3

081.0

127.2

83.5

452.1

890.9

698.8

8G

ov82,1

13

66.0

374.1

125.2

71.2

253.2

085.4

898.7

782,1

13

65.0

875.4

127.5

61.1

847.2

587.0

698.7

7P

roduct

Res

ponsi

bilit

y81,6

80

57.4

359.6

029.4

02.1

932.1

386.2

199.2

281,6

80

58.6

059.4

728.0

42.1

935.5

186.2

299.2

2H

um

an

Rig

hts

81,6

80

58.5

662.5

632.8

02.6

421.2

892.8

699.9

581,6

80

67.1

287.2

831.4

32.6

429.5

893.9

799.9

5

TheChannel

Variables

R&

D82,1

13

0.0

50.0

30.0

50.0

00.0

10.0

70.1

852,2

51

0.0

50.0

20.0

30.0

00.0

00.0

50.1

4P

ate

nts

2,6

33

243.3

812

567.3

11.0

02.0

054.0

02803

3,8

02

222.0

710.0

0533.3

91.0

02.0

081.0

02803

HH

I82,1

05

0.1

80.0

80.2

20.0

10.0

40.2

31.0

077,3

44

0.2

20.1

10.2

50.0

10.0

50.3

01.0

0%

of

Share

sH

eld

82,1

13

0.3

70.3

00.2

90.0

00.1

20.6

02.1

382,1

13

0.0

10.0

00.0

40.0

00.0

00.0

00.9

6#

of

Com

mon

Ow

ner

s82,1

13

212.9

7150

218.1

0.0

071

280

2280

82,1

13

212.9

7150

218.1

0.0

071

280

2280

#of

Com

mon

Dir

ecto

rs82,1

13

0.0

60.0

00.4

90.0

00.0

00.0

026

82,1

13

0.0

60.0

00.4

90.0

00.0

00.0

026

#of

Board

Posi

tions

82,1

13

0.0

70.0

00.5

60.0

00.0

00.0

030

ControlVariables

Lev

erage

82,1

13

0.2

50.2

30.1

70.0

00.1

30.3

60.6

078,8

24

0.2

50.2

30.1

50.0

00.1

40.3

60.5

8R

OA

82,1

13

0.0

80.0

70.0

7-0

.10

0.0

40.1

20.2

178,8

24

0.0

80.0

70.0

7-0

.05

0.0

40.1

20.2

2Q

82,1

13

1.7

91.5

10.8

70.8

61.1

52.1

34.2

778,8

24

1.6

21.3

60.7

20.8

91.0

81.9

03.5

8M

B82,0

64

2.9

42.3

02.1

70.5

41.3

73.7

48.7

778,7

47

2.6

92.0

22.0

50.5

11.2

33.4

18.2

8T

Ass

ets

82,1

13

22.9

722.8

11.5

914.8

721.8

924.0

228.8

678,8

24

24.0

924.1

21.6

115.6

322.9

825.2

928.9

6C

lose

Share

s82,1

13

0.1

60.0

60.2

00.0

00.0

10.2

31.0

078,8

24

0.1

70.0

70.2

20.0

00.0

00.2

81.0

0Sale

sGro

wth

82,1

13

0.0

80.0

60.1

4-0

.16

-0.0

10.1

40.5

178,8

24

0.0

60.0

50.1

2-0

.15

-0.0

20.1

20.4

0G

DP

C82,1

13

10.7

510.7

99.4

96.9

010.6

410.8

711.6

978,8

24

10.7

110.7

89.5

96.9

010.6

210.8

711.6

9

38

Tab

le2

Eff

ects

of

Corp

ora

teS

ocia

lR

esp

on

sib

ilit

y(C

SR

)P

rop

agati

on

alo

ng

Glo

bal

Su

pp

lyC

hain

s

Th

ista

ble

rep

orts

resu

lts

from

the

regr

essi

onof

sup

pli

erC

SR

(CS

RS

)S

core

on

cust

om

erC

SR

(CS

RC

)S

core

,as

foll

ows.

CS

RS(C

)S

core

(t+

1)

=a0

+a1C

SR

C(S

)S

core

(t)

+

K ∑ k=1

b kX

k(t

)+

FE

(t)

+ε(t

+1).

Th

ed

epen

den

tva

riab

leis

sup

pli

er(c

ust

omer

)C

SR

Sco

rean

dth

eke

yin

dep

end

ent

vari

ab

leis

cust

om

er(s

up

pli

er)

CS

RS

core

inco

lum

ns

(1)-

(4)

(col

um

ns

(5)-

(8))

.C

ontr

olva

riab

lesX

kin

clu

de

firm

-ch

ara

cter

isti

csof

the

sup

pli

erin

the

firs

tth

ree

colu

mn

san

dth

ose

of

cust

om

erin

the

last

thre

eco

lum

ns.

Inco

lum

ns

(4)

and

(8),

the

contr

ol

vari

ab

les

incl

ud

eb

oth

of

the

sup

pli

eran

dcu

stom

er.

Th

eyare

lever

age,

retu

rnon

ass

ets

(RO

A),

Tob

in’s

Q,

log

ofto

tal

asse

ts(T

Ass

ets)

,cl

ose

ly-h

eld

share

s(C

lose

Sh

are

s),

sale

sgro

wth

,as

wel

las

GD

PC

.A

llth

eva

riab

les

are

defi

ned

inA

pp

end

ixT

able

A.3

.N

Ob

sis

the

nu

mb

erof

firm

-yea

rcu

stom

er-s

up

pli

erp

air

ob

serv

ati

on

s.T

he

regre

ssio

ns

als

oin

clu

de

inte

rcep

tsand

com

bin

atio

ns

ofd

iffer

ent

cust

omer

-su

pp

lier

,in

du

stry

,co

untr

y,an

dye

ar

fixed

effec

ts(F

E),

an

dall

stan

dard

erro

rsre

port

edin

pare

nth

eses

are

clu

ster

edat

the

cust

omer

-su

pp

lier

-pai

rle

vel

.*,

**,

***

are

sign

ifica

nce

leve

lsd

enote

dat

the

10%

,5%

an

d1%

,le

vels

,re

spec

tive

ly.

Su

pp

lier

CS

RS

Sco

reC

ust

om

erC

SR

CS

core

Var

iab

le(1

)(2

)(3

)(4

)(5

)(6

)(7

)(8

)

CS

RC

0.04

4***

0.02

1***

0.0

12**

0.0

11*

(0.0

05)

(0.0

05)

(0.0

06)

(0.0

07)

CS

RS

0.0

04

-0.0

04

-0.0

08

0.0

09

(0.0

05)

(0.0

05)

(0.0

05)

(0.0

06)

Lev

erag

eS-1

8.38

4***

-11.

340*

**

-11.7

68***

-11.4

11***

1.1

33

(1.0

42)

(1.0

23)

(1.0

08)

(1.0

40)

(0.9

38)

RO

AS

38.6

64**

*36

.732

***

38.1

40***

38.5

47***

-1.4

80

(1.4

43)

(1.4

23)

(1.4

45)

(1.5

04)

(1.2

90)

QS

0.71

9***

1.24

5***

1.2

96***

1.3

72***

0.0

61

(0.1

51)

(0.1

58)

(0.1

57)

(0.1

62)

(0.1

50)

Tot

alA

sset

sS10

.097

***

10.9

89***

11.5

39***

11.5

34***

-0.7

77***

(0.1

05)

(0.1

05)

(0.1

07)

(0.1

09)

(0.1

25)

Clo

seS

har

esS

-11.

319*

**-1

3.49

4***

-13.3

16***

-13.3

71***

0.7

56

(0.6

91)

(0.6

89)

(0.7

18)

(0.7

33)

(0.6

50)

Sal

esG

row

thS

-7.2

47**

*-6

.879

***

-7.5

34***

-7.6

12***

0.1

62

(0.5

09)

(0.5

11)

(0.5

30)

(0.5

59)

(0.4

60)

GD

PC

S-1

.973

***

0.35

50.7

85

0.5

35

1.4

77

(0.3

40)

(0.4

24)

(1.0

60)

(1.1

61)

(1.1

52)

39

Lev

erag

eC1.7

91*

-7.3

36***

-7.7

28***

-8.0

60***

-8.4

96***

(0.9

65)

(1.0

41)

(1.1

08)

(1.0

72)

(1.1

20)

RO

AC

-1.1

33

47.1

20***

44.0

96***

44.8

34***

45.7

44***

(1.8

22)

(1.7

50)

(1.7

68)

(1.7

49)

(1.8

42)

QC

-0.3

28

-0.0

85

-0.3

03

-0.2

96

-0.2

55

(0.2

35)

(0.2

04)

(0.2

16)

(0.2

12)

(0.2

24)

Tot

alA

sset

sC0.0

48

7.6

79***

8.8

05***

8.9

23***

9.0

02***

(0.1

29)

(0.1

34)

(0.1

55)

(0.1

49)

(0.1

46)

Clo

seS

har

esC

0.5

81

-10.6

29***

-10.9

07***

-10.7

74***

-11.2

84***

(0.6

66)

(0.6

63)

(0.6

70)

(0.6

78)

(0.7

01)

Sal

esG

row

thC

-0.4

02

-8.1

38***

-7.6

98***

-7.8

30***

-8.3

81***

(0.6

00)

(0.5

47)

(0.5

51)

(0.5

56)

(0.5

95)

GD

PC

C-1

.627

0.5

89*

1.0

43***

0.8

60

0.4

34

(1.2

46)

(0.3

04)

(0.3

52)

(0.9

53)

(1.0

21)

NO

bs

55,6

9455,6

85

52,5

55

48,1

63

51,1

16

51,1

15

51,1

05

45,8

43

Yea

rF

EY

esY

esY

esY

esY

esY

esY

esY

esS

up

pli

erIn

du

stry

xN

oY

esN

oY

esC

ust

omer

Ind

ust

ryF

ES

up

pli

erC

ountr

yx

No

Yes

No

Yes

Cu

stom

erC

ountr

yF

ES

up

pli

erIn

du

stry

FE

Yes

Yes

Yes

Yes

Cu

stom

erIn

du

stry

FE

Yes

Yes

Yes

Yes

Su

pp

lier

Cou

ntr

yF

EY

esY

esY

esY

esC

ust

omer

Cou

ntr

yF

EY

esY

esY

esY

es

40

Table 3

Environmental, Social, and Governance Issues in Global Supply Chains

This table reports results from the regression of supplier (customer) CSR Component score on customer(supplier) component CSR score, as follows.

CSRS(C) ComScore(t+ 1) = a0 + a1CSRC(S) ComScore(t) +

K∑k=1

bkXk(t) + FE(t) + ε(t+ 1).

The dependent variable is supplier (customer) CSR ComScore and the key independent variable is customer(supplier) CSR ComScore in columns (1)-(4) (columns (5)-(8)). The CSR components are environmental(Env), social (Soc), and governance (Gov) issues. Unreported control variables Xk include firm-characteristicsof the customer (C) or supplier (S), and they are leverage, return on assets (ROA), Tobin’s Q, log of total assets(TAssets), closely-held shares (CloseShares), sales growth, as well as GDPC. All the variables are defined inAppendix Table A.3. NObs is the number of customer-supplier pair observations. The regressions also includeintercepts and combinations of different customer-supplier, industry, country, and year fixed effects (FE), androbust standard errors reported in parentheses are clustered at the customer-supplier-pair level. *, **, *** aresignificance levels denoted at the 10%, 5% and 1%, levels, respectively.

EnvS SocS GovS EnvC SocC GovC

Variable (1) (2) (3) (4) (5) (6)

EnvC Score 0.032***(0.006)

SocC Score 0.013**(0.005)

GovC Score 0.024***(0.005)

EnvS Score -0.005(0.005)

SocS Score -0.003(0.005)

GovS Score 0.023***(0.005)

NObs 55,685 55,685 55,685 51,115 51,115 51,115Supplier Controls Yes Yes Yes No No NoCustomer Controls No No No Yes Yes YesYear FE Yes Yes Yes Yes Yes YesSupplier Industry x Yes Yes Yes Yes Yes Yes

Customer Industry FESupplier Country x Yes Yes Yes Yes Yes Yes

Customer Country FE

41

Table 4

Alternative CSR Rating Databases

This table reports results that replicate those of Table 2 using two alternative databases, namely MSCIIntangible Value Assessment (IVA) and Sustainalytics, and the regression model is as follows.

CSRS(C) Score(t+ 1) = a0 + a1CSRC(S) Score(t) +

K∑k=1

bkXk(t) + FE(t) + ε(t+ 1).

The dependent variable is either the supplier or customer CSR score and the key independent variableis its corresponding economically-linked firm’s CSR score. Unreported control variables Xk include firm-characteristics of the customer (C) or supplier (S), and they are leverage, return on assets (ROA), Tobin’sQ, log of total assets (TAssets), closely-held shares (CloseShares), sales growth, as well as GDPC. Allthe variables are defined in Appendix Table A.3. NObs is the number of firm-year customer-supplier pairobservations. The regressions also include intercepts and combinations of different customer-supplier, indus-try, country, and year fixed effects (FE), and standard errors reported in parentheses are clustered at thecustomer-supplier-pair level in columns (1) and (3) and at the customer-year level in (2) and (4). *, **, ***are significance levels denoted at the 10%, 5% and 1%, levels, respectively.

CSRS CSRC CSRS CSRC

Variable (1) (2) (3) (4)

MSCI IVA CSRC Score 0.014**(0.005)

MSCI IVA CSRS Score 0.010(0.006)

Sustainalytics CSRC Score 0.020***(0.008)

Sustainalytics CSRS Score -0.002(0.008)

NObs 28,056 23,600 22,173 20,805Supplier Controls Yes No Yes NoCustomer Controls No Yes No YesYear FE Yes Yes Yes YesSupplier Industry x Yes Yes Yes Yes

Customer Industry FESupplier Country x Yes Yes Yes Yes

Customer Country FE

42

Tab

le5

Locati

on

sof

Cu

stom

ers

an

dS

up

pliers

an

dS

up

plier

CS

R

Th

ista

ble

rep

orts

resu

lts

from

the

regr

essi

on

of

sup

pli

erC

SR

score

(CS

RS

Sco

re)

on

cust

om

erC

SR

score

(CS

RC

Sco

re),

wh

ere

cust

om

ers

and

sup

pli

ers

may

be

from

the

sam

eor

diff

eren

tco

untr

ies

CS

RS

(t+

1)S

core

=a0

+a1C

SR

C(t

)S

core

+

K ∑ k=1

b kX

k(t

)+

FE

(t)

+ε(t

+1).

Th

ere

gres

sion

isap

pli

edto

sup

pli

eran

dcu

stom

erfi

rms

from

the

sam

eco

untr

y,d

iffer

ent

cou

ntr

ies,

or

non

-US

cou

ntr

ies,

or

sup

pli

ers

and

cust

omer

sfr

omem

ergi

ng

(EM

G)

ord

evel

op

edco

untr

ies

(DE

V).

Un

rep

ort

edco

ntr

ol

vari

ab

lesX

kin

clu

de

firm

-ch

ara

cter

isti

csof

the

cust

omer

(C)

orsu

pp

lier

(S),

and

they

are

leve

rage,

retu

rnon

ass

ets

(RO

A),

Tob

in’s

Q,

log

of

tota

lass

ets

(TA

sset

s),

close

ly-h

eld

share

s(C

lose

Sh

ares

),sa

les

grow

th,as

wel

las

the

cou

ntr

y’s

gro

ssd

om

esti

cp

rod

uct

per

cap

ita

(GD

PC

).A

llth

eva

riab

les

are

defi

ned

inA

pp

end

ixT

able

A.3

.N

Ob

sis

the

nu

mb

erof

cust

om

er-s

up

pli

erp

air

obse

rvati

on

s.T

he

regre

ssio

ns

als

oin

clu

de

inte

rcep

tsan

dco

mb

inati

on

sof

diff

eren

tcu

stom

er-s

up

pli

er,in

du

stry

,co

untr

y,an

dye

ar

fixed

effec

ts(F

E),

an

dro

bu

stst

an

dard

erro

rsre

port

edin

pare

nth

eses

are

clu

ster

edat

the

cust

omer

-su

pp

lier

-pai

rle

vel.

*,**

,***

are

sign

ifica

nce

level

sd

enote

dat

the

10%

,5%

an

d1%

,le

vel

s,re

spec

tive

ly.

Sam

eD

iffer

ent

Non

-US

Su

pp

lier

sfr

om

Cu

stom

ers

from

Cou

ntr

yC

ou

ntr

ies

Cou

ntr

ies

EM

GD

EV

EM

GD

EV

Var

iab

le(1

)(2

)(3

)(4

)(5

)(6

)(7

)

CS

RC

Sco

re0.

008

0.0

14**

0.0

23**

-0.0

23

0.0

12**

0.0

23

0.0

16***

(0.0

09)

(0.0

07)

(0.0

09)

(0.0

20)

(0.0

06)

(0.0

15)

(0.0

06)

NO

bs

22,6

7229,8

83

14,3

14

3,3

66

49,2

19

52,0

63

49,8

69

Con

trol

sY

esY

esY

esY

esY

esY

esY

esY

ear

FE

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Su

pp

lier

Ind

ust

ryx

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Cu

stom

erIn

du

stry

FE

Su

pp

lier

Cou

ntr

yx

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Cu

stom

erC

ountr

yF

E

43

Table 6

Linked and Delinked Customer-Supplier Relationships

This table reports results from the regression of supplier CSR score on customer CSR score,linked/delinked indicator, and their interaction as follows.

CSRS Score(t+ 1) = a0 + a1CSRC Score(t) × Post Link/Delink(t) + a2CSRC Score(t)

+a3Post Link/Delink(t) +

K∑k=1

bkXk(t) + FE(t) + ε(t+ 1).

The dependent variable is supplier CSR score (CSRS Score) and the key independent vari-ables are the customer CSR score (CSRC Score), Post Link/Delink variable, and their in-teraction. Post Link (Delink) is a binary variable that equals 1 if the customer and supplierfirst establishes (severes) the relationship and 0 otherwise. Unreported control variables Xk

include firm-characteristics of the customer (C) or supplier (S), and they are leverage, returnon assets (ROA), Tobin’s Q, log of total assets (TAssets), closely-held shares (CloseShares),sales growth, as well as the country’s gross domestic product per capita (GDPC). All thevariables are defined in Appendix Table A.3. NObs is the number of customer-supplier pairobservations. The regressions also include intercepts and combinations of different customer-supplier, industry, country, and year fixed effects (FE), and robust standard errors reportedin parentheses are clustered at the customer-supplier-pair level. *, **, *** are significancelevels denoted at the 10%, 5% and 1%, levels, respectively.

Variable (1) (2)

CSRC Score x Post Link 0.020***(0.004)

Post Link 0.209(0.275)

CSRC Score x Post Delink 0.001(0.004)

Post Delink 0.808***(0.296)

CSRC Score -0.007* 0.001(0.003) (0.003)

NObs 233,881 233,881Controls Yes YesYear FE Yes YesSupplier Industry x Yes Yes

Customer Industry FESupplier Country x Yes Yes

Customer Country FE

44

Table 7

Human Rights and Product Responsibility

This table reports results from the regression of supplier CSR component Score (ComScore) (i.e., humanrights or product responsibility measures of CSR) on its customer CSR counterpart, the year humanrights or product responsibility event occurred (Event), and the interaction between Event and customerCSR component as follows.

CSRS ComScore(t+ 1) = a0 + a1CSRC ComScore(t) × Event + a2CSRC ComScore(t)

+a3Event +

K∑k=1

bkXk(t) + FE(t) + ε(t+ 1).

The dependent variable is the supplier CSR component score (CSRS ComScore) and the key independentvariables are the customer CSR subcomponent score (CSRC ComScore) and Event variable. We evaluatetwo human rights scandals (the 2005 Nike’s workers abuse report and the 2013 building collapse inBangladesh) and two product responsibility scandals (the 2008 food safety scandal in China and the2013 Takata airbag and Toyota car recalls). Unreported control variables Xk include supplier’s leverage,return on assets (ROA), Tobin’s Q, log of total assets (TAssets), closely-held shares (CloseShares),sales growth, and GDPC. All the variables are defined in Appendix Table A.3. NObs is the numberof customer-supplier pair observations. The regressions also include intercepts and combinations ofdifferent customer-supplier, industry, country, and year fixed effects (FE), and robust standard errorsreported in parentheses are clustered at the customer-supplier-pair level. *, **, *** are significancelevels denoted at the 10%, 5% and 1%, levels, respectively.

Worker Workplace Food Safety Airbag & CarAbuse Safety Scandal Recalls

Variable (1) (2) (3) (4)

Human RightsC x 2005 0.343***(0.093)

Human RightsC x 2013 0.031**(0.012)

Human RightsC 0.025 0.019(0.019) (0.020)

Product ResponsibilityCx 2008 0.250*(0.145)

Product ResponsibilityCx 2013 0.268**(0.122)

Product ResponsibilityC -0.026 -0.082(0.057) (0.076)

NObs 3,940 3,940 753 423Controls Yes Yes Yes YesYear FE Yes Yes Yes YesSupplier Industry x Yes Yes Yes Yes

Customer Industry FESupplier Country x Yes Yes Yes Yes

Customer Country FE

45

Table 8

Stakeholder Effects of CSR in Target Supplier and Customer Firms

This table reports results from the regression of supplier CSR score on customer CSR score, binary indicator for post-M&A event of either a customer or a supplier being a target firm, and the interaction between post-M&A indicator (PostM&A) and customer CSR score as follows.

CSRS Score(t+ 1) = a0 + a1CSRC Score(t) × Post M&A S or C(t) + a2CSRC Score(t) + a3Post M&A S or C(t)

+

K∑k=1

bkXk(t) + FE(t) + ε(t+ 1).

The dependent variable is supplier CSR score (CSRS Score) and the key independent variables are customer CSR score(CSRC Score) and Post M&A indicator when a customer or a supplier is a target firm. The table also shows results whenevaluating a customer whose CSR is above the median rating (CSRC

High) in the sample of customer firms. Unreportedcontrol variables Xk include firm-characteristics of the customer (C) or supplier (S), and they are leverage, return onassets (ROA), Tobin’s Q, log of total assets (TAssets), closely-held shares (CloseShares), sales growth, as well as thecountry’s gross domestic product per capita (GDPC). All the variables are defined in Appendix Table A.3. NObs isthe number of customer-supplier pair observations. The regressions also include intercepts and combinations of differentcustomer-supplier, industry, country, and year fixed effects (FE), and robust standard errors reported in parentheses areclustered at the customer-supplier-pair level. *, **, *** are significance levels denoted at the 10%, 5% and 1%, levels,respectively.

Target=Supplier/Customer Target=Supplier Target=Customer

Variable (1) (2) (3) (4) (5) (6)

CSRC Score x Post M&A -0.051** -0.075** -0.028(0.023) (0.032) (0.032)

CSRC Score 0.013** 0.013** 0.012**(0.006) (0.006) (0.006)

CSRCHigh x Post M&A -1.595* -2.307** -0.038

(0.896) (1.153) (1.419)

CSRCHigh 0.382* 0.372* 0.336

(0.219) (0.218) (0.218)

Post M&A 4.492** 1.514* 7.894*** 3.381*** 1.603 -0.504(1.860) (0.830) (2.482) (1.044) (2.639) (1.308)

NObs 52,555 52,555 52,555 52,555 52,555 52,555Controls Yes Yes Yes Yes Yes YesYear FE Yes Yes Yes Yes Yes YesSupplier Industry x Yes Yes Yes Yes Yes Yes

Customer Industry FESupplier Country x Yes Yes Yes Yes Yes Yes

Customer Country FE

46

Table 9

The Effect of Bargaining Power in the Supply Chain and Supplier CSR

This table reports results from the regression of supplier CSR score (CSRS Score) on customer CSR score(CSRC Score), a measure of the bargaining power (Power) in the customer-supplier relationship, and theinteraction between Power and CSRC Score as follows.

CSRS Score(t+ 1) = a0 + a1CSRC Score(t) × Power(t) + a2CSRC Score(t) + a3Power(t)

+

K∑k=1

bkXk(t) + FE(t) + ε(t+ 1).

Our analysis examines the degree of bargaining power in a supply chain through the supplier’s level ofinnovation, measured by its R&D to total assets and the log of the number of patents, in columns (1)-(2),and the industry competitiveness of suppliers and customers, as measured by the Herfindahl-HirschmanIndex (HHI), in columns (3)-(4). The key independent variables are CSRC Score, the proxy for Power,and their interaction. Unreported control variables Xk include firm-characteristics of the customer (C)or supplier (S), and they are leverage, return on assets (ROA), Tobin’s Q, log of total assets (TAssets),closely-held shares (CloseShares), sales growth, as well as the country’s gross domestic product per capita(GDPC). All the variables are defined in Appendix Table A.3. NObs is the number of customer-supplierpair observations. The regressions also include intercepts and combinations of different customer-supplier,industry, country, and year fixed effects (FE), and robust standard errors reported in parentheses areclustered at the customer-supplier-pair level. *, **, *** are significance levels denoted at the 10%, 5%and 1%, levels, respectively.

Variable (1) (2) (3) (4)

CSRC Score x R&DS -0.323***(0.114)

R&DS 65.482***(9.280)

CSRC Score x PatentsS -0.013*(0.008)

PatentsS 0.795(0.737)

CSRC Score x HHIS -0.567***(0.191)

HHIS 32.811**(13.939)

CSRC Score x HHIC 0.048**(0.023)

HHIC -4.235**(2.004)

CSRC Score 0.037*** 0.042 0.034*** 0.010(0.009) (0.042) (0.007) (0.007)

NObs 35,493 1,364 55,682 52,552Controls Yes Yes Yes YesYear FE Yes Yes Yes YesSupplier Industry x Yes Yes Yes Yes

Customer Industry FESupplier Country x Yes Yes Yes Yes

Customer Country FE

47

Table 10

Common Institutional Ownership and Supplier CSR

This table reports results from the regression of supplier CSR score on customer CSR score, Ownership, andthe interaction between Ownership and customer CSR score as follows.

CSRS Score(t+ 1) = a0 + a1CSRC Score(t) × OwnershipS or C(t) + a2CSRC Score(t)

+a3OwnershipS or C(t) +

K∑k=1

bkXk(t) + FE(t) + ε(t+ 1).

The dependent variable is supplier CSR score (CSRS Score) and the key independent variables are thecustomer CSR score (CSRC Score) and Ownership variable. Ownership is measured using the percentageof shares held by common institutional investors in a supplier firm or customer firm, or the log number ofinstitutional investors holding both the supplier and customer firms (# of Common Owners). Unreportedcontrol variables Xk include firm-characteristics of the customer (C) or supplier (S), and they are leverage,return on assets (ROA), Tobin’s Q, log of total assets (TAssets), closely-held shares (CloseShares), salesgrowth, as well as the country’s gross domestic product per capita (GDPC). All the variables are definedin Appendix Table A.3. NObs is the number of customer-supplier pair observations. The regressions alsoinclude intercepts and combinations of different customer-supplier, industry, country, and year fixed effects(FE), and robust standard errors reported in parentheses are clustered at the customer-supplier-pair level.*, **, *** are significance levels denoted at the 10%, 5% and 1%, levels, respectively.

Variable (1) (2) (3)

CSRC Score x % of Shares HeldS 0.033**(0.015)

% of Shares HeldS 3.421***(1.244)

CSRC Score x % of Shares HeldC 0.035**(0.014)

% of Shares HeldC 4.160***(1.078)

CSRC Score x # of Common Owners 0.068***(0.024)

# of Common Owners 0.040**(0.018)

CSRC Score -0.010 -0.004 -0.026(0.008) (0.008) (0.018)

NObs 52,555 52,555 55,685Controls Yes Yes YesYear FE Yes Yes YesSupplier Industry x Yes Yes Yes

Customer Industry FESupplier Country x Yes Yes Yes

Customer Country FE

48

Table 11

Common Directors and Supplier CSR

This table reports results from the regression of supplier CSR score on customer CSR score, commondirectors (CDirectors), and the interaction between Directors and customer CSR score as follows.

CSRS Score(t+ 1) = a0 + a1CSRC Score(t) × CDirectors S or C(t) + a2CSRC Score(t)

+a3CDirectors S or C(t) +

K∑k=1

bkXk(t) + FE(t) + ε(t+ 1).

The dependent variable is supplier CSR score (CSRS Score) and the key independent variables arethe customer CSR score (CSRC Score), CDirectors variable, and their interaction. CDirectors ismeasured using the log number of directors who serve on the boards of both the supplier and thecustomer, or the log number of common directors holding multiple positions in the supplier firm.Unreported control variables Xk include firm-characteristics of the customer (C) or supplier (S), andthey are leverage, return on assets (ROA), Tobin’s Q, log of total assets (TAssets), closely-held shares(CloseShares), sales growth, as well as the country’s gross domestic product per capita (GDPC).All the variables are defined in Appendix Table A.3. NObs is the number of customer-supplier pairobservations. The regressions also include intercepts and combinations of different customer-supplier,industry, country, and year fixed effects (FE), and robust standard errors reported in parenthesesare clustered at the customer-supplier-pair level. *, **, *** are significance levels denoted at the10%, 5% and 1%, levels, respectively.

Variable (1) (2)

CSRC Score x # of Common Directors 0.053*(0.027)

# of Common Directors -4.858**(2.229)

CSRC Score x # of Board Positions 0.048*(0.025)

# of Board Positions -4.348**(2.022)

CSRC Score 0.019*** 0.019***(0.007) (0.007)

NObs 34,024 34,024Controls Yes YesYear FE Yes YesSupplier Industry x Yes Yes

Customer Industry FESupplier Country x Yes Yes

Customer Country FE

49

Tab

le12

Fir

mP

erf

orm

an

ce

an

dC

ust

om

er

Eff

ects

of

CS

R

Th

ista

ble

rep

orts

resu

lts

from

regr

essi

ng

firm

per

form

an

ce,

mea

sure

dby

SG

&A

,3-y

ear

an

nu

ali

zed

sale

sgro

wth

,an

dfi

rmva

luati

on

on

sup

pli

erC

SR

Sco

re,

cust

omer

CS

RS

core

,an

dth

ein

tera

ctio

nb

etw

een

cust

om

erC

SR

Sco

rean

dsu

pp

lier

CS

RS

core

as

foll

ows.

Per

form

ance

Sor

C(t

+1)

=a0

+a1C

SR

CS

core

(t−

1)×

CS

RS

Sco

re(t

)+a2C

SR

CS

core

(t−

1)

+a3C

SR

SS

core

(t)

+

K ∑ k=1

b kX

k(t

)+

FE

(t)

+ε(t

+1).

Th

ed

epen

den

tva

riab

leis

the

sup

pli

er’s

per

form

an

ce(P

erfo

rman

ce)

inco

lum

ns

(1),

(3),

an

d(5

),an

dof

cust

om

er’s

in(2

),(4

),an

d(6

),an

dke

yex

pla

nat

ory

vari

able

sin

clu

de

sup

pli

erC

SR

score

(CS

RS

Sco

re),

cust

om

erC

SR

score

(CS

RC

Sco

re)

an

dth

eir

inte

ract

ion

.P

erfo

rman

ceis

mea

sure

dby

SG

&A

,3-

year

annu

aliz

edsa

les

grow

th,

and

mark

et-t

o-b

ook

equ

ity

valu

e.U

nre

port

edco

ntr

ol

vari

ab

lesX

kin

clu

de

firm

-ch

ara

cter

isti

csof

the

cust

omer

(C)

orsu

pp

lier

(S),

and

they

are

leve

rage,

retu

rnon

ass

ets

(RO

A),

Tob

in’s

Q,

tota

lass

ets

(TA

sset

s),

close

ly-h

eld

share

s,sa

les

gro

wth

,as

wel

las

the

cou

ntr

y’s

gros

sd

omes

tic

pro

du

ctp

erca

pit

a(G

DP

C).

All

the

vari

ab

les

are

defi

ned

inA

pp

end

ixT

ab

leA

.3.

NO

bs

isth

enu

mb

erof

cust

omer

-su

pp

lier

pai

rob

serv

atio

ns.

Th

ere

gre

ssio

ns

als

oin

clu

de

inte

rcep

tsan

dco

mb

inati

on

sof

diff

eren

tcu

stom

er-s

up

pli

er,

ind

ust

ry,

cou

ntr

y,an

dye

arfi

xed

effec

ts(F

E),

and

rob

ust

stan

dard

erro

rsre

port

edin

pare

nth

eses

are

clu

ster

edat

the

cust

om

er-s

up

pli

er-p

air

leve

l.*,

**,

***

are

sign

ifica

nce

level

sd

enot

edat

the

10%

,5%

and

1%

,le

vel

s,re

spec

tive

ly.

SG

&A

3-Y

ear

An

nu

ali

zed

Sale

sG

row

thM

ark

et-

to-B

ook

Valu

e

Su

pp

lier

Cu

stom

er

Su

pp

lier

Cu

stom

er

Su

pp

lier

Cu

stom

er

Var

iab

le(1

)(2

)(3

)(4

)(5

)(6

)

CS

RC

*CS

RS

-0.0

003*

**-0

.0003***

0.0

21

0.0

29*

0.0

04**

0.0

003*

(0.0

000)

(0.0

000)

(0.1

49)

(0.0

17)

(0.0

01)

(0.0

001)

CS

RS

0.05

48**

*0.0

257***

0.0

06

0.0

25

0.2

70**

-0.0

65***

(0.0

0747

)(0

.00868)

(0.1

22)

(0.0

15)

(0.1

21)

(0.0

13)

CS

RC

0.02

35**

*0.0

341***

0.3

40***

-0.0

21*

-0.2

11*

-0.0

14

(0.0

0771

)(0

.00720)

(0.1

08)

(0.0

12)

(0.1

09)

(0.0

14)

NO

bs

31,2

3228,2

38

32,2

42

32,2

44

33,2

02

32,2

42

Su

pp

lier

Con

trol

sY

esN

oY

esN

oY

esN

oC

ust

omer

Con

trol

sN

oY

esN

oY

esN

oY

esY

ear

FE

Yes

Yes

Yes

Yes

Yes

Yes

Su

pp

lier

Ind

ust

ryx

Yes

Yes

Yes

Yes

Yes

Yes

Cu

stom

erIn

du

stry

FE

Su

pp

lier

Cou

ntr

yx

Yes

Yes

Yes

Yes

Yes

Yes

Cu

stom

erC

ountr

yF

E

50

Appendix Table A.1

Distribution of the numbers of customers and suppliers, together with their meanCSR scores by year.

This table shows numbers of suppliers and customers, as well as their average CSR (CSRS and CSRC ,respectively) scores, by year.

Number of AverageSuppliers Customers CSRS Score CSRC Score

Year (1) (2) (3) (4)

2003 5,484 5,388 58.21 64.11

2004 6,159 6,161 64.73 72.29

2005 6,181 6,588 67.30 75.55

2006 6,494 7,111 66.37 75.12

2007 6,638 7,355 67.51 74.49

2008 7,150 7,835 66.60 73.46

2009 9,207 9,173 65.30 72.67

2010 12,371 12,544 68.18 73.09

2011 15,273 16,131 68.05 74.03

2012 18,792 19,902 66.02 72.43

2013 23,097 23,493 65.92 71.94

2014 20,866 20,758 67.01 71.89

2015 23,556 20,749 64.35 73.00

51

Appendix Table A.2

Distribution of numbers of customers and suppliers, together with their mean CSRscores by country.

This table shows the numbers of suppliers and customers, as well as their average CSR (CSRS and CSRC ,respectively) scores, by year.

Number of AverageSuppliers Customers CSRS Score CSRC Score

Country (1) (2) (3) (4)

Australia 1,580 2,195 81.30 63.05Austria 115 186 69.63 62.97Belgium 178 193 76.61 72.43Bermuda 107 303 21.95 22.40Brazil 1,338 768 70.17 59.98Canada 2,450 3,195 75.20 57.79Chile 458 293 38.64 34.32China 874 469 50.11 41.04Colombia 61 18 80.45 74.22Czech Republic 19 4 45.02 49.64Denmark 246 238 75.24 72.71Egypt 6 2 4.86 5.46Finland 437 760 90.39 90.99France 3,857 4,500 88.55 83.90Germany 3,622 3,061 82.36 73.41Greece 104 45 77.83 45.13Hong Kong 871 639 60.75 62.45Hungary 35 12 50.62 56.90India 713 398 76.58 74.54Indonesia 348 244 58.05 58.63Ireland 215 261 62.43 64.56Israel 274 185 38.98 56.75Italy 811 503 82.80 88.58Japan 5,786 3,997 70.73 67.26Kuwait 10 2 75.13 24.99Luxembourg 346 219 80.14 66.48Malaysia 174 225 62.56 46.07Mexico 298 288 40.02 43.98Morocco 3 5 26.38 26.38Netherlands 1,665 1,224 88.33 79.34New Zealand 135 113 59.06 26.50Norway 435 287 89.01 82.95Peru 43 19 22.07 35.90Philippines 101 61 49.10 47.32Poland 172 267 46.75 26.23Portugal 162 39 86.15 87.05Qatar 13 5 7.12 9.45Russia 554 434 51.06 51.25Saudi Arabia 62 52 62.64 73.71Singapore 329 678 65.80 58.92South Africa 591 651 82.01 67.96South Korea 2,235 1,903 68.35 64.44Spain 799 630 90.64 87.32Sweden 693 1,052 86.36 80.74Switzerland 956 1,300 86.53 64.85Thailand 197 170 81.07 83.93Turkey 163 84 56.83 49.62United Arab Emirates 4 3 20.00 15.38United Kingdom 5,632 6,192 87.58 78.67United States of America 38,546 43,741 71.13 59.12

52

Ap

pen

dix

Tab

leA

.3

Vari

ab

leD

efi

nit

ion

an

dD

ata

Sou

rce

Vari

ab

leD

efi

nit

ion

an

dD

ata

Sou

rce

CS

RC

om

pon

ent

Sco

res

CS

RS

core

An

equ

alw

eigh

ted

CS

Rra

tin

gsc

ore

(AS

SE

T4)

Env

Asc

ore

asso

ciat

edw

ith

the

envir

on

men

tal

pilla

rof

CS

RR

ati

ng

(AS

SE

T4)

Soc

Asc

ore

asso

ciat

edw

ith

the

soci

al

resp

on

sib

lep

illa

rof

CS

RR

ati

ng

(AS

SE

T4)

Gov

Asc

ore

asso

ciat

edw

ith

the

corp

ora

tegov

ern

an

cep

illa

rof

CS

RR

ati

ng

(AS

SE

T4)

Hu

man

Rig

hts

Asc

ore

asso

ciat

edw

ith

the

hu

man

rights

pilla

rof

CS

RR

ati

ng

(AS

SE

T4)

Pro

du

ctR

esp

onsi

bil

ity

Asc

ore

asso

ciat

edw

ith

the

pro

du

ctre

spon

sib

ilit

yp

illa

rof

CS

RR

ati

ng

(AS

SE

T4)

Mec

han

ism

Vari

abl

esR

&D

Res

earc

han

dD

evel

op

men

tE

xp

ense

(Worl

dsc

op

eit

em01201)/

Tota

lass

ets

(Worl

dsc

op

eit

em02999)

Pat

ents

Nat

ura

llo

gari

thm

valu

eof

nu

mb

erof

succ

essf

ul

pate

nt

ap

pli

cati

on

sfi

led

plu

son

ein

aye

ar

(PA

TS

TA

T)

HH

IH

erfi

nd

ahl-

Hir

sch

man

Ind

exm

easu

red

by

the

sum

mati

on

of

squ

are

dm

ark

etsh

are

(base

on

sale

s)of

each

firm

wit

hin

the

sam

ein

du

stry

)(D

ata

stre

am

Worl

dsc

op

e)%

ofS

har

esH

eld

Su

mof

the

max

imu

m%

of

own

ersh

iph

eld

by

all

com

mon

inst

ituti

on

al

own

ers

inth

esu

pp

lier

(or

cust

om

er)

ina

giv

enyea

r(F

actS

etG

lob

alO

wn

ersh

ipD

ata

)#

ofC

omm

onO

wn

ers

#of

com

mon

own

ers

wh

oow

nsh

are

sof

both

the

sup

pli

eran

dcu

stom

erin

agiv

enye

ar

(Fact

Set

Glo

bal

Ow

ner

ship

Data

)#

ofC

omm

onD

irec

tors

#of

com

mon

dir

ecto

rsw

ho

serv

eon

the

board

sof

both

the

sup

pli

eran

dcu

stom

erin

agiv

enye

ar

(Board

Ex)

#of

Boa

rdP

osit

ion

s#

ofb

oard

pos

itio

ns

hel

dby

com

mon

dir

ecto

rsat

the

sup

pli

erin

agiv

enye

ar

(Board

Ex)

Con

trol

Vari

abl

esL

ever

age

Book

Val

ue

ofD

ebt

(Worl

dS

cop

eit

em03255)

/T

ota

lass

ets

(Worl

dsc

op

eit

em02999)

RO

AE

arn

ings

bef

ore

Inte

rest

an

dT

axes

(Worl

dsc

op

eit

em18191)/

Tota

lass

ets

(Worl

dsc

op

eit

em02999)

QM

arket

valu

eof

com

mon

equ

ity

(Worl

dsc

op

eIt

em08001)

+T

ota

lass

ets

(Worl

dsc

op

eIt

em02999)

-B

ook

valu

eof

com

mon

equ

ity

(Wor

ldsc

ope

item

03501)

an

dth

end

ivid

edby

the

net

valu

eby

Tota

lass

ets

(Worl

dsc

op

eit

em02999)

TA

sset

sN

atura

llo

gari

thm

valu

eof

tota

lass

ets

(Worl

dsc

op

eit

em02999)

plu

son

eC

lose

Sh

ares

Nu

mb

erof

Clo

sely

Hel

dS

hare

s/C

om

mon

Sh

are

sO

uts

tan

din

g(W

orl

dsc

op

eit

em08021)

Sal

esG

row

thO

ne

year

net

sale

sgro

wth

(Worl

dsc

op

eit

em08631)

GD

PC

Gro

ssd

omes

tic

pro

du

ctin

US

Dd

ivid

edby

dom

esti

cp

op

ula

tion

(Worl

dB

an

kIn

dic

ato

rs)

Per

form

an

ceV

ari

abl

esS

G&

AS

elli

ng,

gen

eral

and

ad

min

istr

ati

veex

pen

ses

(Worl

dsc

op

eit

em08631)

div

ided

by

tota

lass

ets

(Worl

dsc

op

eit

em02999)

3-Y

ear

An

nu

aliz

edS

ales

Gro

wth

3-ye

arn

etsa

les

grow

th(W

orl

dsc

op

eit

em08633)

Mar

ket-

toB

ook

Rat

ioM

arke

tva

lue

ofco

mm

on

equit

y(W

orl

dsc

op

eIt

em08001)

div

ided

by

Book

valu

eof

com

mon

equ

ity

(Worl

dsc

op

eit

em03501)

53

ONLINE APPENDIX

To accompany

Socially Responsible Corporate Customers

54

Online Appendix Table OA.1

Major versus Non-Major Customers and Supplier Corporate Social Responsibility

The table shows the regression of a supplier’s CSRS score on its customer firm’s CSRC

score, as follows.

CSRS Score(t+ 1) = a0 + a1CSRC Score(t) +

K∑k=1

bkXk(t) + FE(t) + ε(t+ 1).

The dependent variable is the supplier CSR score and the key independent variable is itsmajor or non-major customer CSR score. Unreported control variables Xk include thesupplier’s leverage, return on assets (ROA), Tobin’s Q, log of total assets (TAssets), closely-held shares (CloseShares), sales growth, and GDPC. NObs is the number of customer-supplier pair observations. The regressions also include intercepts and combinations ofdifferent customer-supplier, industry, country, and year fixed effects (FE), and robuststandard errors reported in parentheses are clustered at the customer-supplier-pair level.*, **, *** are significance levels denoted at the 10%, 5% and 1%, levels, respectively.

Variable (1) (2)

Major CSRC Score 0.020***(0.005)

Non-Major CSRC Score 0.017(0.020)

NObs 51,866 3,819Controls Yes YesYear FE Yes YesSupplier Industry x Yes Yes

Customer Industry FESupplier Country x Yes Yes

Customer Country FE

55

Online Appendix Table OA.2

CSR Gap and Supplier CSR

This table reports results from the regression of supplier CSR score (CSRS Score) on customer CSR score(CSRC Score), CSR Gap and the interaction between CSR Gap and CSRC Score as follows.

CSRS Score(t+ 1) = a0 + a1CSRC Score(t) × CSR Gap(t) + a2CSRC Score(t)

+a3CSR Gap(t) +

K∑k=1

bkXk(t) + FE(t) + ε(t+ 1).

The key independent variables are CSRC Score and the interaction between CSRC Score and CSR Gap. Incolumns (1)-(2), CSR Gap is measured by the difference between CSRC and CSRS Scores, and in columns(3)-(4), the CSR gap is a binary indicator which equals 1 if CSRC >CSRS (I(CSRC >CSRS)), and 0otherwise. Unreported control variables Xk include firm-characteristics of the customer (C) or supplier(S), and they are leverage, return on assets (ROA), Tobin’s Q, log of total assets (TAssets), closely-heldshares (CloseShares), sales growth, as well as the country’s gross domestic product per capita (GDPC).All the variables are defined in Appendix Table A.3. NObs is the number of customer-supplier pairobservations. The regressions also include intercepts and combinations of different customer-supplier,industry, country, and year fixed effects (FE), and robust standard errors reported in parentheses areclustered at the customer-supplier-pair level. *, **, *** are significance levels denoted at the 10%, 5%and 1%, levels, respectively.

Variable (1) (2)

CSRC Score x (CSRC−CSRS) 0.016***(0.001)

CSRC−CSRS -0.807***(0.458)

CSRC Score x I(CSRC >CSRS) 0.121***(0.916)

CSRC Score 0.804*** 0.056***(0.0038) (0.006)

NObs 48,283 48,283Controls Yes YesYear FE Yes YesSupplier Industry x Yes Yes

Customer Industry FESupplier Country x Yes Yes

Customer Country FE

56