The Determinants of ESG Ratings - Rater Ownership Matters

The Determinants of ESG Ratings: Rater Ownership Matters*
Dragon Yongjun Tang, Jiali Yan, Chelsea Yaqiong Yao
June 6, 2022
Abstract
Environmental, social, and governance (ESG) ratings are widely used in practice but
lack evidence of their underpinning. We find that firms sharing the same major shareholders
with the rater (“sister firms”) receive higher ESG ratings. We make causal inference for the
ownership effect by exploiting an acquisition event that created sister firms exogeneously.
Sister firms receive higher ratings when the common owners have larger stakes in the ESG rater.
Notwithstanding their initial higher ratings, sister firms have poorer future ESG outcomes.
These findings cast doubt on the quality of ESG ratings and caution practitioners and regulators.
Keywords: ESG, Rating agencies, Conflicts of interest, Ownership structure

JEL classification: G32, L32, M14

*
Corresponding author: Dragon Yongjun Tang, University of Hong Kong, Pokfulam Road, Hong Kong; (+852)
22194321; [email protected]; Jiali Yan, Accounting and Finance Group, University of Liverpool Management
School, Liverpool, L69 7ZH, UK; (+44) 1517942000; [email protected]; Yaqiong Yao, Department of
Accounting and Finance, Lancaster University Management School, Lancaster, LA1 4YX, UK; (+44)
1524510731; [email protected]. For thoughtful comments and helpful discussions, we thank Kalok
Chan, Si Cheng, Sudipto Dasgupta, Elroy Dimson, Caroline Flammer, John Griffin, Vasso Ioannidou, Lei Gao,
Bruce Grundy, Dirk Jenter, Julian Kölbel, Bo Li, Spencer Martin, Zacharias Saunter, Christoph Schiller, Martin
Schmalz, Tao Shu, Laura Starks, Neal Stoughton, Sheridan Titman, Brooke Wang, Jesse Wang, Yue Wu, Fei Xie,
Huiting Xu, Tong Yao, Aaron Yoon, Elaine Zhang, Alex Zhou, Qifei Zhu, and seminar and conference
participants at the University of Iowa, 2019 SFS Cavalcade Asia Pacific, 2020 FMA, 2020 NFA, 2021 EFA, 2021
Paris December Finance Meeting and 2022 Finance Down Under Conference.

The Determinants of ESG Ratings: Rater Ownership Matters
Abstract
Environmental, social, and governance (ESG) ratings are widely used in practice but
lack evidence of their underpinning. We find that firms sharing the same major shareholders
with the rater (“sister firms”) receive higher ESG ratings. We make causal inference for the
ownership effect by exploiting an acquisition event that created sister firms exogeneously.
Sister firms receive higher ratings when the common owners have larger stakes in the ESG rater.
Notwithstanding their initial higher ratings, sister firms have poorer future ESG outcomes.
These findings cast doubt on the quality of ESG ratings and caution practitioners and regulators.
Keywords: ESG, Rating agencies, Conflicts of interest, Ownership structure

JEL classification: G32, L32, M14

1. Introduction
ESG ratings, which measure the environmental, social, and governance (ESG)
performance of corporate issuers, are in the fervent embrace of investors. We use Google search
trends to create Figure 1, which reveals the surge of interest in ESG ratings in recent years and
the stark contrast with the stable public interest in credit ratings. A recent large-scale survey of
fund managers reports that third-party ESG ratings are as important as shareholder proposals,
formal valuation models, and hedging as a tool for institutional investors to manage climate
risk (Krüger, Sautner, and Starks, 2020). Trillions of dollars are invested based on ESG ratings,
and the trend has accelerated after the outbreak of COVID-19. Moreover, some important
conclusions from empirical studies critically hinge on ESG ratings.1 Despite their widespread
use, little is known about the determinants and objectivity of ESG ratings. In this study, we
take the first step to fill this gap in the literature.
[Insert Figure 1 here]
Understanding the ESG rating quality is critical to the further development of

sustainable investing because ESG rating uncertainty can discourage investment in projects
beneficial to society (Avramov, Cheng, Lioui, and Tarelli, 2021). Given the concerns on the
transparency and diversity of methodologies adopted by rating agencies, it is necessary to
examine whether ESG ratings truly reflect corporate ESG quality. For instance, Facebook,
Volkswagen, and Wirecard all received good ratings from major ESG raters before their
negative ESG incidents were uncovered. 2 Government and watchdogs are pushing the
standardization of ESG metrics to curb greenwashing. The chairman of the U.S. Securities and
Exchanges Commission (SEC) and the head of ESG at the UK’s Financial Conduct Authority
have publicly questioned the quality of ESG ratings and criticized their precision, and the
European Commission is looking into ways to improve their oversight in this area.3 Moreover,

1
Gibson, Glossner, Krüger, Matos, and Steffen (2022) provide an overview of sustainable investing around the
world. This investment trend accelerated during the COVID-19 pandemic. Recent examples relating asset prices
to ESG ratings include Cao, Titman, Zhan, and Zhang (2022), Engle, Giglio, Kelly, Lee, and Stroebel (2020),
Matos (2020), Pedersen, Fitzgibbons, and Pomorski (2021), and so on.
2
See another example, “MSCI, the largest ESG rating company, doesn’t even try to measure the impact of a
corporation on the world. It’s all about whether the world might mess with the bottom line.” Cam Simpson, Akshat
Rathi, and Saijel Kishan, Bloomberg, December 2021. https://www.bloomberg.com/graphics/2021-what-is-esg-
investing-msci-ratings-focus-on-corporate-bottom-line/
3
See “Regulators take aim at ESG ratings in fight against greenwashing”, Financial Times, May 28, 2022;
“SEC chair warns of risks tied to ESG ratings,” Financial Times, May 28, 2020; Question 20, Section 1.3,
“Consultation on the Renewed Sustainable Finance Strategy,” European Commission,
https://ec.europa.eu/info/sites/info/files/business_economy_euro/banking_and_finance/documents/2020-
sustainable-finance-strategy-consultation-document_en.pdf
1

investors and issuers care about the quality and outcomes of ESG ratings. Hartzmark and
Sussman (2019) show that mutual fund investors buy or sell funds based on ESG ratings of the
fund portfolios, suggesting that ESG ratings affect capital allocation. Companies even sue
rating agencies for allegedly poorly manufactured ratings, raising substantial concerns among
practitioners about their accuracy.4
We empirically examine the determinants of ESG ratings, with a focus on potential

conflicts of interest. Our analysis is guided by incentive-based theoretical predictions.
Theoretical models on credit ratings predict that, notwithstanding their concerns on reputation,
rating agencies can purposefully provide inaccurate—either inflated or deflated—ratings (see,
e.g., Bolton, Freixas, and Shapiro, 2012; Goldstein and Huang, 2020). There is a large literature
on credit ratings. In contrast, ESG ratings are a relatively new phenomenon and have been less
scrutinized than credit ratings. There are also differences between ESG ratings and credit
ratings. For instance, ESG performance lacks objective measures or external benchmarks.
Therefore, it is necessary to examine ESG ratings separately.
We compile a large dataset to uncover the determinants and integrity of ESG ratings.
We first show that in recent years, ESG ratings are more explicable by firm characteristics. For
example, large firms receive higher ESG ratings in the post-2010 period, and firm size is the
most statistically significant explanatory variable in multivariate regressions of ESG ratings.
More importantly, we demonstrate that companies owned by the same institutional investors
as their ESG rater (“sister firms”) receive higher ESG ratings. The finding of higher ratings for
sister firms is robust to the control for firm characteristics such as firm size. It is also robust to
alternative measures of sister firms, control groups, and estimation approaches.
We recognize the potential endogeneity in firm ownership and take a meticulous

approach to making causal inferences. Our identification of the common-owner effect on ESG
ratings is primarily from exogenous changes in rater ownership, which occur during corporate
mergers and acquisitions. Prior to 2009, Kinder, Lydenberg, Domini & Co. (KLD), the most
widely used ESG data source for academic studies and among the most popular for
practitioners, was a privately held company primarily owned by its founders and employees.
Its acquisition merged it with a publicly listed company with major shareholders identified by

4
In March 2020, Isra Vision, a German manufacturer of camera technology, sued ISS ESG, a well-established
ESG rating agency, for assigning it the lowest ESG rating, a D-. Consequently, the Munich District Court banned
ISS ESG from rating Isra Vision. This is the first known lawsuit based on poor ESG ratings. Nauman, Billy, 2020
March 4, “Heavy flows into ESG funds raise questions over ratings,” Financial Times.
2

mandatory SEC disclosures. We show that the newly created sister firms receive higher ESG
ratings after the merger. Moreover, ESG ratings increase only for newly added sisters, not
unrelated firms. The results from such a pseudo-natural experiment of acquisition lend support
to a causal interpretation of our main finding.
We show that ESG rating favoritism for sister firms is more pronounced when the
common institutional investor is dedicated (i.e., with a longer holding horizon), has a more
active management style, and owns a large equity stake in the rater’s holding company.
Therefore, the effect of rater ownership on ESG rating depends on owner characteristics. Such
findings are consistent with the theoretical prediction that active large shareholders extract
private benefits from firms rather than remain concentrated on their monitoring role (e.g.,
Goldman and Wang, 2021).
We categorize ESG ratings by the financial materiality of specific ESG issues according
to the classifications set out by the Sustainability Accounting Standards Boards (SASB). We
find that ESG rating inflation induced by ownership connection is driven by the immaterial
part of ESG ratings. Drilling down on the specific constructs of ESG ratings, we show how
rating inflation differs depending on whether the component is a strength or a concern. The
rating agency tends not to inflate the material part of ESG concerns, indicating it engages in
strategic planning when choosing the specific ESG item to inflate. As markets tend to pay more
attention to ESG concerns than strengths (Krüger, 2015), inflating the strengths would not set
off as many alarms. We also benchmark the KLD ratings against ratings from Refinitiv
(previously part of Thomson Reuters) and find qualitatively the same results: KLD gives higher
ratings to its sister firms relative to Refinitiv. Moreover, we uncover little evidence of a
convergence in ESG ratings not subject to ownership influence. In other words, the other rater,
Refinitiv, does not increase its rating for KLD’s sister firms in subsequent years.
As an out-of-sample test for rating inflation, we use future ESG incidents to evaluate
the accuracy of initial ESG ratings. RepRisk collects negative news about firms’ ESG activities
in real time (Li and Wu, 2020). The coverage is widespread with both local and nonlocal, minor
and major news. We find that, all else being equal, more negative news items are reported about
sister firms than non-sister firms. This ESG underperformance by sister firms is significant
after controlling for the initial ESG rating, especially the concern aspects. Therefore, the
inflated ratings of sister firms are revealed by their ex-post poor ESG performance.
3

The findings in this study have implications for increasingly mainstream ESG rating-
based investing. Low-quality ESG information can foster greenwashing (Wu, Zhang, and Xie,
2020). This paper is among the first to scrutinize the integrity of ESG ratings and identify
problems related to conflicts of interest. Our study shows that the large shareholder effect on
credit ratings documented by Kedia, Rajgopal, and Zhou (2017) is also prevalent in ESG
ratings. However, ESG ratings are different from credit ratings as ESG ratings are not paid by
issuers, so the “issuer pay” concern that plagued credit ratings is not the driver for conflicts of
interest for ESG ratings. Nevertheless, once institutional investors gain influence over ESG
rating agencies through their blockholdings, their portfolio companies can receive favorable
ratings, via suasion from the investor, catering by the rater, or other channels.5
This study of conflicts of interest inherent in ESG rating agencies is related to prior
work on other information intermediaries, such as corporate governance advisors. Li (2018)
finds that the voting recommendation advisor, Institutional Shareholder Service (ISS), tends to
make voting recommendations favorable for management. Such a bias is driven by ISS’s
revenue from consultation fees paid by the firms. Daines, Gow, and Larcker (2010) question
the quality of commercial governance ratings. Cohen, Gurun, and Nguyen (2021) reveal an
ESG rating deficiency in the energy sector, in which the firms producing most green patents
receive lower ESG ratings. The empirical findings on conflicts of interest in this paper are
consistent with theoretical predictions from Mehran and Stulz (2007) and Goldstein and Huang
(2020), which address rating inflation and the verifiability of ratings.
This study highlights a new determinant of ESG ratings: rater subjectivity, which
played a role in CDO credit ratings leading up to the 2008 crisis (see, e.g., Griffin and Tang,
2012). It is widely documented in both academic studies and public commentaries that ESG
ratings for the same firm often differ widely across rating agencies. Berg, Koelbel, and Rigobon
(2022), Christensen, Serafeim, and Sikochi (2022), and Gibson, Krüger, and Schmidt (2021)
reveal evidence of rating disagreements and show that the correlations between different ESG
ratings are generally low. Berg, Fabisik, and Sautner (2021) find inconsistent records of ESG
ratings from Refinitiv over time. Our study provides an underpinning for such a rater effect
through the ownership of the rater. The rating disagreements are neither random nor innocuous.
Different rating agencies are controlled by different owners, who then exert their own influence

5
There are studies on the impact of common ownership on product market competition, such as Azar, Schmalz,
and Tecu (2018). However, in our setting, the ESG rating firm and its sister firms are in different industries, and
there is little competition between them.
4

on ESG ratings. These findings suggest ESG rating agencies should be regulated like other
information intermediaries.
The rest of the paper is organized as follows. We introduce the ESG rating industry and
formulate our testable hypotheses in Section 2. Section 3 describes our data and empirical
setting. The effect of sisterhood via common ownership on ESG ratings is presented in Section
4. Section 5 discusses the channels and mechanisms for the sister firm effect. Section 6
concludes.
2. Institutional Background on ESG Ratings and Hypotheses Development
In this section, we first introduce the institutional background of ESG ratings, especially
market practices. We then discuss relevant literature and form testable hypotheses for our
empirical analyses.
2.1. ESG ratings
Sustainable Finance. Sustainable finance originated in the early 1990s to foster social and
environmental developments. However, such investments need specific metrics. Hence, social
and governance ratings were created for practitioners. The phrase “ESG” was first coined in
2004 to reference policies and actions related to environmental, social, and governance issues.
The United Nations (UN) has been a strong promoter of ESG market development. Across the
globe, large institutional investors integrate ESG factors into their portfolio analysis, equity
screening, or quantitative research. Amundi, the largest asset manager in Europe in terms of
total assets under management (AUM), pledged 100% ESG integration for their active funds
and 100% integration of ESG in voting by 2021. 6 A considerable number of banks in 37
countries have also adopted the Equator Principle, committing to consider borrowers’
environmental and social policies during their lending process.7 To gain access to external
financing at lower costs, firms have adopted incentives to improve their ESG profiles. Hong
and Kacperczyk (2009) find that irresponsible firms are eschewed by institutional investors
and analysts, thus facing a higher equity cost. Kim, Wan, Wang, and Yang (2019) show that
institutional investors play an important role in shaping corporate ESG performance. One
influential example is BlackRock, which since 2018 has sent annual “Dear CEO” letters to its

6
See Responsible Investment Amundi’s Ambition 2021.
7
https://equator-principles.com.
5

portfolio companies to remind them about ESG issues (Pawliczek, Skinner, and Wellman,
2021).
ESG Rating. An ESG rating is an assessment of a firm’s performance in its approach to

environmental, social, and governance practices. Institutional investors and regulators refer to
ESG ratings when evaluating a firm’s ESG profile. ESG ratings share similarities with other
rating systems, but they have their own unique features and prominent roles. However, no
universal reporting standards exist in this rather nascent industry, which is the primary obstacle
hobbling the practical use of ESG information, even while the ESG rating industry has
experienced rapid expansion and consolidation in recent years. The first major ESG rating
agency was KLD, founded in 1988 in Boston as a privately owned research firm. It has been
the most widely used ESG data vendor since the mid-1990s. In 2007, 33 out of 50 institutional
investors worldwide were using its research to integrate ESG factors into their investment
decisions.8 Its ratings are the primary data source for academic studies. KLD was acquired by
a public firm in 2010. Refinitiv/Asset4 is another major ESG rating provider with a data history
dating back to 2002. Refinitiv was a division of Thomson Reuters and is now owned by
Blackstone and the London Stock Exchange. Other examples include Bloomberg acquiring
New Energy Finance, Morningstar acquiring Sustainalytics, Standard and Poor’s acquiring
TruCost and RobecoSAM ESG Ratings, and Moody’s acquiring Vigeo Eiris. Bloomberg has
been providing ESG ratings from third-party data sources since 2012. It launched its own ESG
scores in August 2020 to improve the transparency of ESG data and ratings so investors can
make informed decisions.
Rater Ownership. Large investment managers, including several mutual and hedge funds,
own a big stake in the rating agency. Major shareholders of public listed companies could be
identified from regulatory fillings. Some owners hold concentrated portfolios and appear to be
mindful of ESG issues. Their investment strategies allow them to micromanage their investees
through means such as sitting on the board of portfolio companies. For example, ValueAct
Capital usually holds a dozen public companies. It also makes information about its holdings
public, such as its investment in Microsoft in 2013. It appears that most, if not all, major
institutional investors of the parent firm of the rater are conscious of the ESG ratings of their
portfolio companies. Dyck, Lins, Roth, and Wagner (2019) show that institutional investors,
motivated by financial returns, push for stronger firm-level ESG (they focus on E&S) globally.

8
http://web.archive.org/web/20070524034938/http://www.kld.com:80/about/index.html.
6

It is possible they also encourage rating agencies to grant the companies they invest in higher
ESG ratings. Investors can push for a certain ESG profile to justify their investments in those
companies.
Regulations on ESG Ratings. Despite the increasingly important role of ESG ratings in capital
markets, little research has been conducted on the ESG rating industry. This is concerning
given the number of reasons the industry should be regulated. First, ESG assessments are prone
to subjectivity, interpretation, and hypocrisy. As SEC Commissioner Hester M. Peirce
highlighted in her speech to the American Enterprise Institute on June 18, 2019:
[T]he ambiguity and breadth of ESG allow ESG experts great latitude to improve their
own judgments, which may be rooted in nothing at all other than their own preferences.
Not surprisingly then, there are many different scorecards and standards out there, each
of which embodies the maker’s judgments about any issues it chooses to classify as
ESG.9
The above quote leads to the second major concern regarding ESG ratings—the lack of
transparency in how information is assessed. The issue of limited transparency in the ESG
rating processes ranges from indicators and algorithms to the qualitative proprietary assessment
techniques applied. The third issue involves the completion of the unmanageable amount of
work many of the major ESG raters undertake, evaluating thousands of firms using hundreds
of indicators. Working with such a tremendous amount of data might compromise the quality
of the ratings. These problems all cause many to be wary of the trustworthiness of ESG ratings.
ESG rating agencies are part of the ecosystem of sustainable investing, which is one
element in the overarching framework of the Sustainable Development Goals (SDGs)
announced by the United Nations in 2015. ESG is an important element of the SASB, UNPRI,
and the UN Global Compact. Therefore, it is crucial to understand the determinants of ESG
ratings and whether they are prone to conflicts of interest, of which investors need to be
informed. However, ESG ratings and rating agencies are unregulated. Although they fulfill
similar functions as other information intermediaries, such as credit rating agencies, financial
analyst recommendations, and even auditors, ESG rating agencies are not subject to securities
law in any market. More research is needed to uncover important findings in practice.

9
Peirce, Hester M., 2019 June 18, Scarlet Letters: Remarks before the American Enterprise Institute,
https://www.sec.gov/news/speech/speech-peirce-061819.
7

2.2. Testable hypotheses
Rating agencies, as information intermediaries, have an incentive to provide accurate

ratings in order to establish a strong foundation for a long-lasting business. Bolton, Freixas,
and Shapiro (2012) provide a reputation-based model for credit rating agencies. In their model,
credit rating agencies are disciplined by their reputational concerns. Oftentimes, rating agencies
issue unbiased ratings given their information set. However, rating inflation does occur under
certain conditions. Goldstein and Huang (2020) further show that rating inflation can be an
equilibrium outcome when credit ratings have feedback effects and affect corporate financing
and investments. Critical to Goldstein and Huang’s (2020) model is the verifiability of credit
rating. Inflation occurs when ratings are only partially verifiable.
ESG ratings resemble credit ratings in some ways. However, the differences are notable,
perhaps none quite so much as that ESG ratings are harder to verify. The rater has room to
exert its own subjective views. Such subjectivity and opaqueness are a hotbed for conflicts of
interest (see, e.g., Mehran and Stulz, 2007). Potential conflicts of interest can originate from
customers and shareholders of the rating agencies, among others. Conflicts of interest arising
from customers are easily dismissed because ESG agencies are not paid by the companies being
rated. On the other hand, a major shareholder of a rating agency can influence ESG ratings in
ways that will further enhance the value of their portfolio.
Kedia, Rajgopal, and Zhou (2017) provide empirical evidence of credit rating inflation
for related companies. They find that Moody’s gives higher ratings relative to S&P to
companies held by the same institutional investors (e.g., Berkshire Hathaway and Davis
Selected Advisers). Further, they suggest that rater bias toward the economic interests of their
significant shareholders is more severe when opaque and complex issues are involved, such as
structured finance products. Their findings are consistent with the view that large shareholders
exert influence on firms, such as advising on takeovers and managerial appointments, in order
to reap benefits for themselves (Edmans, 2014).10 Above discussions lead us to conjecture the
following relationship between ESG ratings and ownership:

10
See Edmans (2014) for a detailed review of theoretical and empirical studies regarding the mechanisms through
which large shareholders can influence corporate governance. See McCahery, Sautner, and Starks (2016) for a
comprehensive survey of corporate governance channels that institutional investors prefer to adopt.
8

Hypothesis 1: ESG raters assign higher ESG ratings to their sister firms, which are
owned by the same blockholders than non-connected firms.
Some argue that ESG ratings reports and related information are read and analyzed by
machines and thus rating assessments may not be prone to subjectivity and human judgment.
Unlike credit ratings, which are regulated and tend to be governed by specific triggers such as
a firm breaching a certain level of leverage, ESG ratings are less regulated and monitored and
lack clear metrics, which makes rating inflation more plausible. If credit ratings are subject to
conflicts of interest, then ESG ratings are likely to be influenced by subjectivity as well.
If larger shareholders influence the ESG rating process, it is natural to expect the ESG
rater should treat them more favorably than other large shareholders. The ESG rater is likely
to cater to key institutional shareholders because their “voice” or “exit” can have serious
consequences on or pose significant threats to the ESG rater itself. In a similar vein, the ESG
rater is more inclined to accommodate major sister firms in which their large shareholders have
pronounced stakes. This leads to key sister firms obtaining even higher ESG ratings than other
sister firms. In other words, the closeness between an ESG rater and its large shareholders, and
the connectedness between large shareholders and sister firms, tend to increase rating inflation.
Many ESG raters acquire an existing ESG rating business as a means of entering the industry.
The sister firms of an ESG rater may then enjoy the benefits of being associated with such an
agency:
Hypothesis 2: The effect of sister firms on ESG rating is more pronounced when the
institution owners can and do exert a stronger influence on the rater.
3. Data and Summary Statistics
In this section, we outline the three main datasets used in the paper: ESG rating data,
Compustat corporate fundamentals data, and 13F institutional holding data. We then describe
our model specifications.
3.1. ESG rating data
ESG rating data is provided by the KLD database. In the existing literature, ESG ratings
are most commonly used to evaluate a firm’s ESG performance.11 The KLD ESG rating data

11
See, e.g., Hong and Kostovetsky (2012), Hong, Kubik, and Scheinkman (2012), DiGiuli and Kostovetsky
(2014), Cheng, Hong, and Shue (2016), Khan, Serafeim, and Yoon (2016), and Chen, Dong and Lin (2020).
9

is one of the longest ESG data time series available. The dataset covers S&P 500 firms from
1991 to 2000. Its coverage extends to Russell 1000 components and Russell 3000 component
firms through 2001 and 2003, respectively. It measures ESG strength and ESG concern
indicators for seven ESG categories: environment, community, employee relations, diversity,
product, humanity, and governance. If a firm meets (or fails to meet) the evaluation criteria for
an indicator, it receives one (zero) for that indicator. Each year, we add one point for each
strength (concern) to create an aggregate strength (concern) for a given firm. We then take the
difference between the strengths and the concerns across the categories to calculate the ESG
ratings. Higher ratings indicate more strengths or fewer concerns, and vice versa.
Founded as a private company in 1988 and acquired in 2010 by a public firm, KLD
experienced a change in ownership structure from a privately owned firm to a publicly traded
firm. Our analysis focuses on ESG ratings post-acquisition, from 2010 to 2015. Table 1
presents the summary statistics of the 11,145 firm-year observations in the sample. In Panel A,
the mean of the ESG strengths is 1.61, while the mean of the ESG concerns is 1.47. The average
of the ESG ratings is 0.13, and the median of the ESG ratings is 0, with a standard deviation of
2.74.
[Insert Table 1 here]
3.2. Identifying shareholders and sister firms
We obtain ownership structure information from the Thomson Reuters Institutional

Holdings 13F database (formerly known as CDA/Spectrum), which contains ownership
information on all institutional investors, operating under discretionary management, who
manage portfolios of at least $100 million. Common stock positions larger than 10,000 shares
or worth $200,000 or more must be disclosed, and institutional managers file 13F reports of
their stock ownership quarterly with the SEC. An institutional investor is defined as a large
shareholder if it owns at least 5% of shares of the parent firm of the rater in a given year.12 In
our sample period, investments by large shareholders can exceed 10%, the investment horizon
ranges from two to six years, and the management style of the owners varies from passive to

12
The ESG rating data in a given year is released in the second quarter of the subsequent year. For example, the
rating data for 2010 was published in the second quarter of 2011. The ESG ratings for 2010 are matched with the
institutional holding data from the third quarter of 2010 to the second quarter of 2011. Given that the ESG database
was acquired in June 2010, this study analyzes ESG rating data from 2010 to 2015, which are matched with the
institutional holding data from the third quarter of 2010 to the second quarter of 2016.
10

active. In the section below, we discuss how these factors influence the ESG ratings of
institutional investors’ holding firms.
We define an investee firm as a sister if it comprises at least 0.25% of a large

shareholder’s investment portfolio in a given year. 13 Panel B of Table 1 displays various
comparisons between sister firms and other firms. The average ESG rating of sister firms is
2.70 and -0.07 for other firms is -0.07, creating a difference of 2.77 at the 1% significance level.
Similarly, the median ESG rating of sister firms is 2.00, whereas it is 0.00 for others. The
difference test in medians is statistically significant at the 1% level.
Does the significant difference in ESG ratings between sisters and non-sisters hold
across the sample of firms? Figure 2 depicts the ESG ratings of sister and non-sister firms,
respectively, and shows that, in general, the frequencies of positive (negative) ESG ratings of
sister firms are higher (lower) than those of non-sisters. This distribution plot confirms that
ESG ratings of sister firms are higher than those of non-sisters across the sample of firms.
Further, we turn to the Kolmogorov-Smirnov test to compare the two distributions. We find
that the null hypothesis that the two distributions of ESG ratings of sisters and non-sisters
would be identical was rejected, with a Kolmogorov-Smirnov value of 0.3748 (p = 0.0000).
Nevertheless, significant differences may be driven by a number of different factors at the firm
and institution levels. To address this possibility, we perform regression tests to control various
factors.
3.3. Empirical model specification
The control variables used are the following firm-level variables that may potentially
account for variations in ESG ratings as documented in prior literature: total assets, book-to-
market ratio, firm age, sales per employee, ROA, prior-year return, capital expenditure,
research and development (R&D), R&D missing dummy, dividends, cash holdings, leverage,
advertising intensity, institutional ownership, analyst coverage, and blue state dummy.14 The

13
We adopt 0.25% as the cutoff because it is the 75th percentile holding weight in the Thomson Reuters 13F
universe during the sample period. For robustness, we use alternative definitions of Sister and find that the results
are qualitatively similar, as shown in the next section.
14
See Di Giuli and Kostovetsky (2014) on total assets, book-to-market ratio, and firm age; Dimson, Karakas, and
Li (2015) on sales per employee, ROA, and stock returns; Jensen and Meckling (1976) on capital expenditure,
R&D, and R&D missing; Hong, Kubik, and Scheinkman (2012) on dividends, cash holdings, and leverage;
Servaes and Tamayo (2013) on advertising intensity; Dyck, Lins, Roth, and Wagner (2019), Chen, Dong, and Lin
11

stock price information is obtained from the Center for Research in Security Prices (CRSP),
accounting variables from the Compustat Fundamentals Annual database, and analyst
information from the IBES detail file. The Appendix provides detailed variable definitions.
We begin by exploring the determinants of ESG ratings pre- and post-acquisition of the
ESG data business to test whether explanatory factors are time variant. We regress the ESG
ratings on all control variables defined in the Appendix, and include industry times year-fixed
effects. In Table 2, column (1) reports the results for the entire sample period of 2003 to 2015;
column (2) presents the results for the pre-acquisition period of 2003 to 2009, and column (3)
shows the results for the post-acquisition period of 2010 to 2015. The results reveal the
estimated coefficient on firm size (i.e., total assets) is insignificant pre-acquisition, but positive
and significant post-acquisition. This implies that big firms outperform small firms in terms of
ESG ratings post-acquisition of an ESG data business, which is not the case pre-acquisition. In
other words, this observation indicates that the size effect on ESG ratings exists only post-
acquisition.
Other explanatory variables, including firm age, financial constraint measures (i.e.,
cash holdings, dividends, and leverage), advertising intensity, analyst coverage, and blue state
dummy, explain the consistency of ESG ratings over time. The coefficients on cash holdings
and dividends are positive and significant, whereas the coefficient on leverage is negative and
significant, in line with the financial constraint hypothesis that states firms do good when they
do well (Hong, Kubik, and Scheinkman, 2012). The loading on advertising intensity is positive
and significant, consistent with the view that advertising spending strengthens the relation
between CSR and firm value (Servaes and Tamayo, 2013). The negative coefficient on
institutional ownership echoes agency motives behind ESG spending (Cheng, Hong, and Shue,
2016). The loadings on the blue state dummy are positive and significant, which is in line with
the finding that firms have higher ESG ratings when they are headquartered in Democratic
rather than Republican-leaning states (Di Giuli and Kostovetsky, 2014).

(2020), and Gibson, Glossner, Krüeger, Matos, and Steffen (2022) on total institutional ownership; Hong and
Kacperczyk (2009) on analyst coverage; Hong and Kostovetsky (2012) and Di Giuli and Kostovetsky (2014) on
political values.
12

Panel B of Table 1 presents the univariate analysis of firm characteristics of sister and
non-sister firms. Compared with non-sister firms, sister firms are larger, more profitable, and
mature; have more growth opportunities and higher institutional ownership; spend more on
advertising; are followed by more analysts; and are more likely to be headquartered in blue
states. In the univariate analysis, the difference tests on these firm characteristics are all
statistically significant. For example, the average of the logarithm of total assets of sister firms
is 9.28 and of non-sister firms is 7.13, with a difference of 2.15 at the 1% significance level.
For the blue-state dummy, the average percentage of sister firms headquartered in Democratic-
leaning states is 0.68, which is statistically significant compared to the average of non-sister
firms, at 0.60.
4. Sister Firms and ESG Ratings
The transition from a privately owned firm to a publicly listed firm suggests shareholder
interests have become the important driver of corporate actions, consequently changing ESG
ratings post-acquisition. In other words, large shareholders may extract private benefits. Most
large shareholders are investment firms and as such are not assigned an ESG rating.15 However,
through acquisition, the rater “adopts” many sister firms connected to these large shareholders.
Will sister firms receive higher ESG ratings based on institutional shareholders’ interests? Or
do machine involvement or reputational concerns prevent shareholders from exerting such
influence? In this section, we will bring those questions to the data.
4.1. ESG rating and sister firms
To answer these questions, we begin by examining the relationship between ESG

ratings and sister firms, using the following equation:
ESG ratingit  0  1Sisterit  2 X it  eit , (1)
where the dummy variable, Sister, equals one if a firm is a large investee owned by a large
shareholder of the rater’s parent firm and zero otherwise; Xit is a vector of firm-level

15
Despite most large shareholders of the rater not being assigned ESG ratings, in our sample, we identify one
firm, one of the top three large shareholders, a public firm on the S&P 500, which has ESG ratings available. This
firm received a nine-point boost in its ESG rating one year after the rater was acquired, compared with one year
before.
13

characteristics including total assets, book-to-market ratio, firm age, sales per employee, ROA,
prior-year return, capital expenditure, R&D, R&D missing dummy, dividends, cash holding,
leverage, advertising intensity, institutional ownership, analyst coverage, and blue state dummy.
In Table 3, column (1), we include the Sister dummy variable without any control
variables, and the regression is akin to the univariate analysis. The Sister coefficient, 2.775
with a t-statistic of 10.38, is positive and statistically significant, indicating a positive relation
between ESG ratings and sister firm status. In column (2), we include total assets as the only
independent variable, consistent with the size effect on ESG ratings previously discussed. The
estimated coefficient on total assets, 0.799 with a t-statistic of 24.33, is positive and significant
at the 1% level. In column (3), we include the Sister dummy and total assets as the two
independent variables. The coefficient on the Sister dummy is 1.199 with a t-statistic of 4.97,
indicating Sister can explain ESG ratings even when firm size is controlled. However, omitted
firm characteristics could affect this finding. In column (4), we include all firm-level control
variables and a blue state dummy, and control for industry and year-fixed effects to examine
whether sister status remains a key determinant of ESG ratings. The coefficient on the Sister
dummy, 0.795 with a t-statistic of 3.78, is positive and significant at the 1% level, showing that
adding firm-level controls and industry and year-fixed effects does not qualitatively affect the
results.
Our findings reveal sister firms are assigned higher ESG ratings, supporting our
hypothesis that these firms garner favorable treatment compared with non-sister firms. This is
consistent with the “partial verifiability constraint” view of Goldstein and Huang (2020), as the
validity of ESG ratings is difficult to verify. Consequently, the ambiguity and breadth of ESG
ratings leave ESG raters with the scope and discretion to treat their favorites favorably.
4.2. Robustness tests
In this subsection, we examine the robustness of our main findings by constructing

various alternative measures and using an alternative estimation method.
4.2.1. Large Shareholders. We propose an alternative definition of “large shareholders.”

We define an institutional investor as a large shareholder of the rater if it owns at least 4% of
shares of the rater, rather than 5% as defined in the previous subsection. An institutional
14

investor is required to file Schedule 13D/13Gforms with the SEC when acquiring more than
5% of a voting class of equity securities of a publicly traded company. 16 To avoid public
disclosure, the institutional investor may intentionally manage its holding positions to remain
below 5%, but will still be able to influence the management team due to its significant stakes.
The results are reported in Table 4, column (1). While all control variables in our baseline test
are included in the new test and the remaining tables, we only report the Sister dummy (and
any newly added variables) coefficients for the sake of brevity. Using the alternative cutoff to
define a large shareholder leads to similar findings. The coefficient on the Sister dummy, 0.712
with a t-statistic of 3.94, is positive and significant at the 1% level, which is similar to the
baseline result, 0.795 with a t-statistic of 3.78, as seen in Table 3, column (4).
4.2.2. Sister Firm Definition. We explore whether our results are robust to alternative
definitions of “sister firms.” The monitoring efforts allocated by institutional investors to an
investee firm depend on the relative importance of the investee firm’s stock in their portfolios
(Fich, Harford, and Tran, 2015). In other words, we expect that the larger stakes shareholders
have in investee firms, the more preferential treatment the investee firms receive from the rater.
In our previous analysis, we define a sister firm as one that accounts for at least 0.25% of one
of the rater’s large shareholder’s portfolios. Using a stricter cutoff, we define a sister firm as
one that accounts for at least 5% of one of the rater’s large shareholder’s portfolios (Fich,
Harford, and Tran, 2015). Table 4, column (2) reveals that the coefficient on the Sister dummy
(1.773) is twice that of our baseline model result of 0.795 (Table 3, column (4)).
In a similar spirit, we would expect to see sister firms’ effects on rating inflation
increase as institutional ownership cutoffs rise. Figure 3 plots the coefficients on the Sister
dummy estimated from our baseline model. We define a sister firm as one that accounts for 1%
to 5% of one of the rater’s large shareholder’s portfolios. The coefficients on the Sister dummy
increase monotonically as the cutoff rises from 1% to 5%. The results not only confirm that
our findings are robust to various alternative definitions of sister firms, but also show that the
bigger the stake that a large shareholder invests in a sister firm, the higher the ESG ratings
assigned to the sister firm.

16
https://www.sec.gov/fast-answers/answerssched13htm.html.
15

4.2.3. Continuous Measure of Connection. We use an alternative continuous measure

to quantify the stakes large shareholders invest in sister firms. The continuous measure is the
size of ownership, defined as the average of the large shareholders’ portfolio weights of a sister
firm in a given year. The results are reported in Table 4, column (3). The estimated coefficient
on the continuous measure, 0.272 with a t-statistic of 3.14, is positive and significant. Using
alternative continuous measures provides further evidence that sister firms are more likely to
receive favorable treatment when large shareholders retain larger stakes in such firms.17
4.2.4. Matched Sample. We implement propensity score matching to address a concern

about nonlinearity. The treatment group includes sister firms, while the untreated group is
comprised of the remaining firms that were evaluated by the same rater but are not its sister
firms. To control observable differences in attributes between the treatment and control groups,
we use the propensity score matching procedure, a technique using one-to-one nearest-
neighbor matching with replacement, which can reduce the bias of regression estimators. This
approach is suitable in our setting because we have more candidates for potential matches in
the untreated group than in the treated group, which increases the possibility of identifying
appropriate matches between sister and non-sister firms.
We begin the matching with a logit regression where the dependent variable equals one
if a firm is a sister firm and zero otherwise. Because independent variables need to influence a
firm’s ESG rating and institutional investors’ shareholding choices simultaneously, we use firm
characteristics in our baseline regression to match a sister firm to another firm with attributes
statistically closest to it. To alleviate the concern of imperfect matching, we also control the
same covariates in the regression analysis. Table 4, column (4) reports the results. The
coefficient on Sister, 0.666 with a t-statistic of 2.47, remains positive and statistically
significant. The matching analyses provide further supportive evidence of our main findings.

17
We also test whether ESG ratings increase as the length of time large shareholders invest in sister firms increases.
The length of the relationship between a large shareholder and a sister firm is defined as the number of quarters
that large shareholders hold those sister firms. We find that rating inflation is positively related to the duration of
ownership.
16

4.3. Identification
Despite the various precautions taken in the previous subsection, our findings may still
be spurious due to potential endogeneity problems. Omitted variables may drive the association
between sister firms and favorable ESG ratings. First, it is possible that large shareholders may
hold both the rater and rater’s sister firms due to their common unobservable characteristics.
Second, our findings may be confounded by the fact that the rater and its large shareholders
are capable of identifying firms with better ESG performance. In other words, specific firm
characteristics might entice large shareholders to purchase major stakes in those firms, but
those characteristics might also lead the rater to assign high ratings. To identify the causal
relationship between sister firms and ESG rating inflation, we use the acquisition of the ESG
business as an exogenous shock to examine ESG ratings of sister firms pre- and post-
acquisition of the ESG business in 2010. If higher ESG ratings of sister firms are due to private
benefits from corporate control, rather than unobservable common criteria, then the higher ESG
ratings assigned to sister firms should not exist before the takeover of the ESG business.
We begin by computing the average of sister and non-sister firms' ESG ratings post-
acquisition of the ESG business. We then define “pseudo-sisters” as firms held by the same set
of large shareholders prior to their classification as large shareholders post-acquisition, and
define “pseudo-non-sisters” as the rest of the firms pre-acquisition. We calculate the average
ESG ratings on pseudo-(non-)sisters before the acquisition. We plot the average of the ratings
for (non)sisters and those for pseudo-(non)sisters in Figure 4. Two thought-provoking
observations arise. First, the average of the ESG ratings on pseudo-sisters is negative pre-
acquisition, while the average of the ESG ratings of sisters is positive and rises to above one
post-acquisition. Second, the average of the ESG ratings of pseudo-non-sisters is negative pre-
acquisition, and the average of non-sisters is negative post-acquisition. The visual inspection
highlights the large differences in ESG ratings for pseudo-sisters pre-acquisition and sisters
post-acquisition, despite few differences in ESG ratings between pseudo-non-sisters and non-
sisters.
17

We then turn to formal tests, specifically, regression-based analyses. We extend our
sample period to cover 2003 to 2015, including six years before and after acquisition.18 We
augment our baseline regression by adding another dummy variable, Pseudo-Sister, which is
equal to one if a firm is a large investee firm consisting of the same large shareholders prior to
their being classified as large shareholders post-acquisition, and zero if not. If the rater loosened
its ESG rating standards only after the change in ownership, the coefficient of the Sister dummy
should be positive and significant, while the coefficient of the Pseudo-Sister dummy should
not. If the higher ratings assigned to connected firms are due to common criteria, then both the
coefficient of the Sister and Pseudo-Sister dummy should be positive and significant. As seen
in Table 5, column (1), the coefficient of the Sister dummy (1.791 with a t-statistic of 8.28) is
positive, whereas the coefficient of the Pseudo-Sister dummy (-0.329 with a t-statistic of -2.62)
is negative. This finding suggests that once connected with the rater, sister firms receive higher
ratings.19 Overall, we find little evidence to suggest that common criteria are the source of the
favorable ratings awarded to the rater’s sister firms, as higher ratings exist only after the
connection is established.
Our analyses above raise two critical questions. First, when a firm aligns with a rater
and becomes a sister firm, does this help upgrade a firm’s ESG rating? Second, when a sister
firm disassociates from its rater, does it lead to a downgrade in a firm’s ESG rating? To answer
the two questions, we create four dummy variables: Become Sister (Become Pseudo-Sister)
that takes the value of one if a firm becomes a sister firm to the rater after (before) 2010, and
zero otherwise, and Become Outsider (Become Pseudo-Outsider) that takes the value of one if
a firm is no longer a large investee in the rater after (before) 2010, and zero otherwise.
As shown in Table 5, column (2), the coefficient on the Become Sister dummy, 1.473
with a t-statistic of 8.25, is positive and significant, while the coefficient on Become Pseudo-
Sister, 0.162 with a t-statistic of 0.83, is insignificant. The results answer the first question:
becoming sister firms with the rater has little impact on ESG ratings pre-acquisition, but after

18
A potential concern is that our findings may be contaminated by confounding events that occurred during this
long sample period. To address this issue, we re-estimate our results using a one-year (three-year) window before
and after acquisition. We find this robustness check does not qualitatively change the results.
19
To rule out possible confounding effects due to macroeconomic or industry-level factors, we repeat the exercise,
replacing the dependent variable, ESG ratings, with relative ESG ratings as defined in Section 5.4. Using relative
ESG ratings does not qualitatively affect our results.
18

the acquisition it incurs favorable ESG ratings. As seen in Table 5, column (3), the coefficient
on the Become Outsider dummy (0.020 with a t-statistic of 0.14) is insignificant, suggesting
that if a firm is no longer a sister post-acquisition, no downgrade occurs, which answers the
second question. This finding may reflect the concept of the “once and forever club” in which
both the rater and their sister firms have incentives to maintain their relationship once it has
been established.
5. Channels and Mechanisms
The section above demonstrates that the rater tends to give preferential treatment to its
sister firms by assigning them higher ESG ratings. In this section, we explore channels and
mechanisms that drive such rating inflation and subsequent performance.
5.1. Owner style and influence
We begin by investigating how relationships with large shareholders affect the extent
of rating inflation. First, we examine whether rating inflation is related to the closeness between
the rater and its large shareholders. Does the duration that large shareholders hold the rater
matter? Chen, Harford, and Li (2007) suggest that short- and long-term large investors
approach monitoring activities differently. Compared with short-term investors, long-term
investors focus more on monitoring activities and less on short-term trading profits. McCahery,
Sautner, and Starks (2016) show that long-term institutional investors intervene in the activities
of their portfolio firms more intensively, through “voice” or direct intervention, than short-term
investors do. In this subsection, we examine differences in ESG ratings of sister firms between
short-term and long-term large shareholders of the rater.
We define an institutional investor as a dedicated (transient) large shareholder of the

rater if it holds at least 5% of the rater’s ownership for more than (equal to or less than) two
years, following Chen, Harford, and Li (2007). In Table 6, column (1), the dummy variable
Sister is equal to one if an investee firm is owned by a dedicated large shareholder of the rater,
and zero otherwise. As reported in column (1), when sister firms are held by dedicated large
shareholders, the coefficient on the Sister dummy, 0.842 (t-statistic of 3.56), is positive and
significant at the 1% level. In column (2), the dummy variable Sister is equal to one if an
investee firm is owned by a transient large shareholder of the rater, and zero otherwise. As
shown in column (2), when sister firms are owned by transient large shareholders, the
19

coefficient on the Sister dummy is insignificant. Our results indicate that ESG rating inflation
is limited to sister firms held by dedicated shareholders rather than transient ones.
Does the size of the stake in the rater held by large shareholders make a difference? We
rank large shareholders based on their average ownership of the rater in a given year during
our sample period. We define a dummy variable Top (Bottom) to indicate if a sister firm is held
by large shareholders that rank in the top (bottom) five of investors to indicate their average
quarterly stake in the rater. In Table 6, column (3), the coefficient on the Top dummy, 1.098
with a t-statistic of 3.88, is positive and significant. The results show that when large
shareholders hold more stakes in the rater, sister firms are more likely to receive preferential
treatment.
We then investigate whether the management style of large institutions matters. If large
shareholders are passive investors, they might not care about the ESG ratings of their holding
firms, and therefore sister firms may not receive preferential treatments from the rater. To
explore this possibility, we repeat our baseline test, but divide large shareholders into two
categories, active and passive, and reexamine the status of sister firms within each category. In
Table 6, column (5), Sister is a dummy variable that takes a value of one if a sister firm is held
by a large shareholder who is an active player in the fund industry, and zero otherwise; in
column (6), passive players are examined. As shown in column (5), when sister firms are held
by active institutional investors, the coefficient of the Sister dummy, 0.728 with a t-statistic of
3.71, is positive and significant. Column (6) reveals that when sister firms are held by passive
institutional investors, the coefficient of the Sister dummy is insignificant. These results are
consistent with our expectation that active institutional investors care more about the ESG
ratings of their holding firms.
Our analyses show that ESG rating inflation for sister firms is more pronounced when the
common owner is a long-term investor, holds a large ownership stake, and possesses a more
active management style. The results suggest that common owner characteristics may be
underlying forces behind our finding of ESG rating favoritism for sister firms.
20

5.2. Materiality of ESG ratings
We consider the implications of financial materiality in rating inflation. ESG issues can
be financially material and immaterial from an investor’s viewpoint. Unlike material ESG
issues, which may be taken into consideration by shareholders and monitored by regulators,
immaterial ESG issues may receive less attention, leaving raters with the scope and discretion
to give sister firms favorable ratings. In other words, the cost of inflating immaterial issues may
be lower. Meanwhile, investors may not be able to distinguish between material and immaterial
ESG issues because materiality classifications have been available only recently (Khan,
Serafeim, and Yoon, 2016). That is to say, the benefit of inflating immaterial issues may be
similar to that of inflating material issues. Given the cost-benefit balances, accordingly, we
would expect that rating inflation that favors sister firms may be mainly limited to immaterial
ESG issues.
To distinguish between material and immaterial ESG indicators, we follow SASB

standards. To alleviate the problems of the ambiguity of ESG ratings that investors and
regulators face, SASB, founded in 2011, aims to provide a clear set of standards for reporting
the sustainability information that matters most to their investors. Because the SASB standards
are designed from an investor viewpoint, the standard-setting process centers on financial
materiality. The SASB standards focus on sector-specific sustainability factors likely to affect
the financial conditions or operating performances of companies materially. They provide
guidance for 11 sectors encompassing 77 industries. 20 For example, greenhouse house
emissions are a material issue for the extractives & minerals processing sector but are
immaterial for the finance sector, while data security is a material issue for the consumer goods
sector, but not for the transportation sector.
We hand-map the firm-level ESG rating indicators with the SASB sector-specific
standards to classify the ESG indicators into financially material and immaterial categories.
We follow Khan, Serafeim, and Yoon (2016) to classify material or immaterial within each
sector accordingly. For example, Tesla, an American electric vehicle and clean energy
company, is in the transportation sector. In 2011, Tesla obtained one strength under the ESG
category of “climate change: carbon emissions,” which is hand-mapped to the SASB topic of

20
The 11 sectors are consumer goods, extractive and mineral processing, financials, food and beverage, health
care, infrastructure, renewable resources and alternative energy, resource transformation, services, technology and
communications, transportation, and services. https://www.sasb.org/standards-overview/materiality-map/.
21

“greenhouse emissions” as a material issue in the transportation sector. Meanwhile, Tesla
received one concern under each of the ESG categories of “reporting quality” and “board
diversity.” However, these are not classified as material issues in the transportation sector under
SASB standards. After hand-mapping SASB material topics to ESG indicators across sectors,
we then construct the material (immaterial) ESG ratings for each firm as the material
(immaterial) strengths minus the material (immaterial) concerns.
We repeat the baseline model, replacing the dependent variables with the material and
immaterial ESG ratings rather than the overall ESG ratings. The results are reported in Table
7. As shown in columns (1) and (2), when the material and immaterial ESG ratings are
dependent variables, the coefficient on the Sister dummy is 0.157 (t-statistic of 1.63) and 0.638
(t-statistic of 3.58), respectively. Consistent with our expectations, the results indicate that
preferential treatment of sister firms is mainly present in immaterial ESG issues, which may
not be monitored by regulators and receive less attention.
Next, we repeat the baseline test, but assess the implications of materiality on ESG
strengths and concerns separately. Columns (3) and (4) report the results for ESG strengths.
The coefficients of the Sister dummy are 0.213 (t-statistic of 2.66) and 1.222 (t-statistic of 6.69)
for material and immaterial ratings, respectively, again indicating raters’ looser standards for
sister firms regarding immaterial issues as opposed to material issues. When Immaterial
Concern Ratings is the dependent variable, as shown in column (6), the coefficient of Sister is
0.584 with a t-statistic of 6.02, less than half the magnitude of the immaterial strength ratings.
5.3. Benchmarking against alternative ESG ratings
Excluding the previously discussed factors, we consider whether any unobserved or

omitted factors could account for higher ratings awarded to sister firms. For example, certain
industrial or macroeconomic events may influence a firm’s ESG performance and affect its
rating. We employ a difference-in-difference method to examine relative ESG ratings rather
than absolute ESG ratings. The first difference examined is between the KLD rating and an
alternative rating. If a firm’s ESG performance is affected by an industry shock, it should be
reflected in both the KLD ratings and alternative ESG ratings, but not in relative ratings. The
second difference compares the ratings of sister firms with those of non-sister firms.
22

We begin by benchmarking the KLD ratings against the Refinitiv ESG ratings.
Compared to KLD, Refinitiv ESG data have a relatively short and small coverage,
encompassing S&P 500 firms since 2003 and extending to Russell 1000 components and
Russell 3000 components in 2011 and 2017, respectively. Refinitiv ESG ratings are the second
most popular ESG rating data used in ESG literature. The agency provides two ESG measures:
an ESG score and an ESG combined score. The Refinitiv ESG score includes only the positive
aspects of ESG, while the combined score adds negative factors. We use the Refinitiv ESG
combined score as it provides a more comprehensive evaluation of a firm’s sustainable impact
and behavior and is also more comparable to the KLD ESG rating.
We begin by grouping all firms into 20 quantiles based on the KLD ESG ratings and
Refinitiv ESG ratings, using the distribution of raw ratings in each year to create comparable
scales. We then apply ESG rankings—from 1 to 20—in the analysis. We construct the
dependent variable, Benchmarked ESG—the differences between KLD ESG rankings and
Refinitiv ESG rankings—using the following model:
Benchmarked ESGit  0  1Sisterit  2 X it  eit , (2)
where the specification in the model (2) resembles the specification in the model (1), except
the dependent variable is changed to Benchmarked ESG.
The regression estimation results are reported in Table 8. As shown in column (1), the
estimated coefficient of the Sister dummy is 1.226 with a t-statistic of 2.88, indicating that
compared to Refinitiv, KLD assigns higher ESG ratings to its sister firms. These results help
alleviate the concern that our findings may be driven by unobserved and omitted factors related
to ESG ratings. Our findings echo the well-documented rating divergence found in both
academic studies and public commentaries—ESG ratings for the same firm can vary
significantly across rating agencies.21 Berg, Koelbel, and Rigobon (2022) attribute the rating
divergence to the rater effect or the rater’s subjective view of the firm. We show that firms that
are connected to the rater tend to receive favorable treatment relative to the alternative rater,
which indicates the ownership structure of the rater can create the rater effect.

21
See, e.g., Berg, Koelbel, and Rigobon (2022), Christensen, Serafeim and Sikochi (2022), and Gibson and Krüger,
and Schmidt (2021).
23

Having demonstrated the significant differences between KLD ESG ratings and
alternative ESG ratings of sister firms, we next examine whether KLD and its competitors
adjust their ratings and whether the ratings assigned by two raters may converge in subsequent
years. If rating inflation reflects the information-driven timeliness of KLD’s rating decisions,
rather than ownership-driven leniency, then the ratings by two raters should eventually
converge. To explore this possibility, we estimate the response of future Refinitiv ESG ratings
to current KLD ratings assigned to sister firms. In columns (2) and (3), the dependent variable
is Future Refinitiv ESG rankings; that is, Refinitiv ESG rankings in Year t + 1 and t + 2. The
independent variable includes the Sister dummy, ESG Rankings, and the interaction between
them (Sister × ESG). If alternative ratings of sister firms increase and converge with those of
KLD, we would expect that the coefficient on the interaction term is positive.
The results are reported in Table 8, columns (2) and (3). In column (2), the dependent
variable is Refinitiv ESG Rankings in Year t + 1, and the coefficient on the interaction term
Sister × ESG is -0.012 with a t-statistic of -2.15. In column (3), we obtain similar results when
the dependent variable is Refinitiv ESG Rankings in Year t + 2. This shows that after KLD
assigns higher ratings to its sister firms, the Refinitiv ESG ratings of those firms do not increase
in subsequent years. The results provide little evidence that KLD’s high ratings of its sister
firms are driven by the information timeliness of KLD’s rating decisions, given the low ratings
assigned by its competitor. Further, the negative coefficients of the interaction terms in columns
(2) and (3) indicate the sister firms’ ratings will continue to diverge in subsequent years.
5.4. Incentive alignment
Why do institutional investors care about the ESG ratings of their portfolio companies?
Hartzmark and Sussman (2019) argue that good ESG metrics enable funds to attract greater
inflow from investors. Asset owners might prefer investments with higher ESG ratings (Bauer,
Ruof, and Smeets, 2021). To investigate incentive alignment between raters and large
shareholders, we examine two possibilities. First, the rater may have more incentive to provide
favorable treatment to sister firms held by multiple shareholders than sister firms held by single
shareholders. Second, a large shareholder of the rater that is powerful and influential due to
AUM may create more incentive to assign higher ratings to the large shareholder’s holding
firms. These large shareholders can derive greater benefits by obtaining easing ratings for
connected firms in their portfolios.
24

To examine the first possibility, we repeat the baseline analysis but distinguish whether
a large investee is held by one or multiple large shareholders. We report the results in Table 9.
In column (1), the Sister dummy takes the value of one if it represents at least 0.25% of the
portfolio of a large shareholder of the rater and is held by multiple large shareholders in a year,
and zero otherwise. In column (2), the Sister dummy takes the value of one if it represents at
least 0.25% of the portfolio of a large shareholder of the rater and is held by a single large
shareholder in a year, and zero otherwise. The coefficient on the Sister dummy in column (1),
1.221 with a t-statistic of 2.96, is more than double the corresponding coefficient on the Sister
dummy in column (2), 0.458 with a t-statistic of 2.36. This finding indicates that the more
shareholders hold sister firms, the higher ratings sister firms receive.
To explore the second possibility, we redo the baseline analysis but differentiate large
shareholders by their AUM. In Table 9, column (3), the Sister dummy takes the value of one if
a firm represents at least 0.25% of the rater’s large shareholder’s portfolio and its AUM is
above the median of other large shareholders in a year, and zero otherwise. In column (4), the
Sister dummy is defined similarly, but the AUM of the large shareholder of the rater is below
the median AUM of other large shareholders in a year. The coefficient on the Sister dummy in
column (3) is 1.038 with a t-statistic of 3.42; however, the corresponding coefficient in column
(4) is insignificant. The results suggest sister firms held by larger shareholders tend to improve
by more than one rating notch, and this relative easing in ESG ratings is both economically
meaningful and statistically significant, while the sister firms held by smaller shareholders may
not receive higher ratings. In other words, major large shareholders, compared to relatively
small shareholders, may enjoy more benefits from easing ESG ratings.
5.5. Large shareholders’ incentives
In this subsection, we investigate how large shareholders’ incentives relate to ESG

rating inflation. When large shareholders have more incentive to push for high ESG ratings of
their portfolio firms, ESG rating inflation might be more severe. We first examine the
differential responses of large shareholders to ESG proposals targeted at their significant
holding firms. Prior studies document that when ESG shareholder proposals are put up for
votes, it signals that the proposal sponsors (e.g., asset managers) and management are unable
to reach an agreement (Flammer, 2015; Krüger, Sautner, and Starks, 2020). When important
25

investee firms of the ESG rater’s shareholders are faced with such pressure (e.g., a large number
of CSR proposals), the shareholders’ attention may then focus on the investees’ ESG profile.
Shareholders could either engage with the targeted firm directly to create substantial
improvements or push for higher ESG ratings to restore the company’s image. Second, we use
the FTSE4Good US Select Index rebalance events as shocks to large investee firms’ ESG
reputation, which in turn may incentivize shareholders to seek high ESG ratings for their
investee firms. FTSE-Russell measures the ESG performance of firms using over 300
indicators. U.S. firms are added to the FTSE4Good US Select Index when they meet the criteria
and removed when they fail to do so. Since its inception, the Index has been viewed as a salient
ESG benchmark in the financial market (Shin, 2021; Starks, Venkat, and Zhu, 2020). As such,
we expect that when firms are removed, their shareholders may have a greater incentive to
repair firms’ ESG reputation by pursuing high ESG ratings.
We obtain information on CSR shareholder proposals from Institutional Shareholder

Services and rerun the baseline regression by differentiating large shareholders on the basis of
the average number of CSR proposals faced by their important investee firms at annual general
meetings. In Table 10, column (1), the dummy variable Sister is an indicator that equals one if
a firm represents at least 0.25% of the shareholders’ portfolio, and the average number of CSR
proposals faced by a large shareholders’ important investee firms is above the median of all
shareholders, and zero otherwise. In column (2), the dummy variable Sister is an indicator that
equals one if a firm represents at least 0.25% of the shareholders’ portfolio, and the average
number of CSR proposals faced by a large shareholders’ important investee firms is below the
median of all shareholders, and zero otherwise. The coefficient on the Sister dummy in column
(1) is 0.951 with a t-statistic of 9.07, and in column (2) it is 0.325 with a t-statistic of 2.29. The
difference test of the coefficients on the two Sister dummies is significant, with a probability
of 0.0181. The results suggest that when shareholders face greater public activism on CSR
issues involving their important investee firms, sister firms may receive higher ESG ratings.
We re-estimate the baseline regression but differentiate large shareholders based on the
average number of FTSE4Good US Select Index exclusion events experienced by their
important investee firms. In column (3), the indicator variable, Sister, is equal to one if a firm
represents at least 0.25% of the rater’s large shareholder’s portfolio, and the average number
of index exclusion events occurring at a large shareholders’ important investee firms is above
26

the median of all large shareholders, and zero otherwise. In column (4), the indicator variable,
Sister, is equal to one if a firm represents at least 0.25% of the rater’s large shareholder’s
portfolio, and the average number of index exclusion events occurring at a large shareholders’
important investee firms is below the median of all large shareholders, and zero otherwise. The
coefficient on the Sister dummy in column (3) is 1.709 with a t-statistic of 8.22, and the
coefficient on the Sister dummy in column (4) is 0.735 with a t-statistic of 3.89. The difference
test of the coefficients on the Sister dummies in two columns is significant, with a probability
of 0.0596. This indicates that large investors may be more incentivized to respond to reputation
shocks to their significant investee firms by pressing for higher ESG ratings to repair the
damage.
5.6. Future ESG performance
Finally, we examine whether higher ratings assigned to sister firms can be justified by
their better ESG performances. In other words, the higher ESG ratings received by sister firms
may merely reflect that their good ESG performances are being rewarded with high ratings.
We use ESG incidents as a proxy for ESG performance to examine how ESG ratings are related
to ESG performance. Unlike ESG ratings, ESG incidents are less likely to be under direct
managerial control. Although it may be plausible for firms to self-promote positive news
through their public relations departments, it is considerably harder to bury negative ESG news
coverage as media outlets compete for audiences.
We retrieve negative ESG news from RepRisk, a comprehensive database that has
compiled information on ESG and business conduct risks since 2007. The RepRisk dataset
covers negative ESG news that can be harmful to a firm’s reputation. Its data analysis is
conducted in two steps. First, using machine learning algorithms based on 8 million labeled
documents, RepRisk systematically screens over 500,000 documents for news items on ESG
incidents from more than 90,000 public sources and stakeholders in 20 major business
languages on a daily basis. Second, using a rule-based methodology, trained analysts conduct
two-level analyses of each ESG incident to verify and then classify them into 30 predefined
ESG incident categories (95 ESG factors), such as environmental degradation, overuse, and
wasting of resources, and impacts on communities. This dataset was previously used by
researchers such as Hummel and Schlick (2016), Kolbel, Busch, and Jancso (2017), and Li and
Wu (2020).
27

We obtain firm-level incident counts from RepRisk and merge them with our sample.
We begin by plotting the number of ESG incidents to make comparisons between sister and
non-sister firms. Figure 5 depicts the average count of ESG incidents pertaining to pseudo-
sisters and pseudo-non-sisters for each quarter pre-acquisition of the ESG business from 2007
to 2009. The pseudo-sisters endured slightly more ESG incidents than pseudo-non-sisters pre-
acquisition. Figure 5 also shows the mean of ESG incidents involving sister and non-sister
firms for each quarter post-acquisition from 2010 to 2015. If ESG ratings reflect true ESG
performance, then we should expect fewer ESG incident counts for sister firms compared to
non-sister firms in order to justify higher ratings for sister firms. What we find is the opposite.
The average count of ESG incidents attributed to sister firms is noticeably higher than the
average for non-sister firms post-acquisition. The significant discrepancy between ESG ratings
and ESG performance implies that higher ESG ratings for sister firms are likely not justified
by better ESG performance. Similarly, Li and Wu (2020) also use RepRisk data to show that
for the sake of shareholders’ best interests, public firms, as opposed to private firms, show little
real effort in ESG performance.22
The RepRisk data provide an external measure of actual ESG performance, which
allows us to answer the following question: Do inflated ratings for sister firms have predictive
power over subsequent ESG incidents? If large shareholders care about sister firms’ ESG
ratings rather than their actual ESG performance, then we expect to uncover a positive relation
between sister firms and future ESG incidents rather than a significant reduction in the sister
firms’ number of ESG incidents.
We begin by plotting the average number of future ESG incidents in the subsequent
year against the range of ESG concerns for sister firms (black bars) and non-sisters (gray bars).
Figure 6 reveals that for each range of concern, the number of future ESG incidents faced by
sister firms is generally higher than that faced by non-sister firms. For instance, for the level of
ESG concerns ranging from 9 to 11, sister firms encounter more than 20 ESG incidents in the
subsequent year, compared with fewer than 10 ESG incidents involving non-sister firms. The

22
Li and Wu (2020) suggest that the shareholder-stakeholder conflicts of interest are the important driver of
decoupled ESG action. Having compared the ESG performances of public firms and private firms, they find that
private firms tend to improve ESG performance as evidenced by a decrease in ESG incident levels, but this is not
the case for public firms.
28

results indicate that the lenient ratings bestowed on sister firms may not be justified. We then
turn to formal regression-based tests.
We employ a count data model to examine whether a relationship exists between future
ESG incidents and rating inflation. The primary independent variables of interest are the Sister
dummy and ESG Rating; the dependent variable is the number of future ESG incidents
predicted for the following year. We assume the number of ESG incidents follows a Poisson
distribution—the rate of distribution is determined by the status of being a sister firm, firm
characteristics, and industry year-fixed effects. Table 11, columns (1) and (2) report the count
data model estimation results. As shown in column (1), the coefficient on the Sister dummy is
positively related to the number of ESG incidents in the following year. In column (2), we add
lagged one-year ESG concerns and the identical set of control variables included in other tables
to control for past ESG performance and firm-level characteristics. We still find that the Sister
dummy significantly positively predicts future ESG incidents.
For robustness, we use a negative binomial distribution for the number of ESG incidents
and re-estimate the count data model. Table 11, Column (3) does not include past ESG ratings
or various control variables, while column (4) does. Using alternative distribution generates
qualitatively similar results. All findings indicate that rating inflation is associated with an
increased number of future negative ESG incidents.
6. Conclusions
ESG ratings have become increasingly important to investment and capital allocation.
However, little is known about the determinants of ESG ratings, especially whether such
ratings reflect the true quality of a firm’s ESG approaches. In this study, we find that rater
ownership plays a significant role in the determination of ESG ratings. Specifically, the rater
tends to give higher ESG ratings to sister firms owned by the same large shareholders. Our
findings are robust to different definitions of sister firms, alternative estimation methods, and
benchmarking against ratings from alternative raters. To identify the causal effect, we use the
exogenous change of ownership of the rater and find that sister firm results are present only
29

after the rater and sister firms are connected through the same owners. We show that the sister
firm rater effect is more pronounced when the large institutional investor is dedicated (e.g.,
with a longer holding horizon), has a more active management style, and holds a significant
ownership stake in the rater. When investigating whether rating inflation granted to sister firms
is justified, we find that sister firms instead are predicted to have more future negative ESG
incidents.
To the best of our knowledge, this study is the first to provide evidence that conflicts
of interest in market practices can undermine the integrity of ESG ratings. Across the world,
trillions of dollars are being invested in accordance with ESG ratings. Rating inflation that
benefits a specific group of firms can have significant real and negative consequences on the
capital markets. Our work highlights the specific conflicts of interest that affect the validity of
the ESG ratings, which should be considered by investors in various decision- and policy-
making situations. We support clearer and more transparent ESG metrics, which could make
rating inflation more detectable. The European Commission recently proposed the Corporate
Sustainability Reporting Directive in an attempt to produce more consistent and comparable
information. It remains to be seen to what extent the implementation of the included policies
help improve ESG rating quality.
30

References
Avramov, Doron, Si Cheng, Abraham Lioui, and Andrea Tarelli, 2021, Sustainable investing
with ESG rating uncertainty, Journal of Financial Economics, forthcoming.
Azar, José, Martin C. Schmalz, and Isabel Tecu, 2018, Anticompetitive effects of common
ownership, Journal of Finance 73, 1513–1565.
Berg, Florian, Kornelia Fabisik, and Zacharias Sautner, 2021, Rewriting history II: The
(un)predictable past of ESG ratings, Working Paper, MIT Sloan School of Management
and Frankfurt School of Finance and Management.
Berg, Florian, Julian F. Koelbel, and Roberto Rigobon, 2022, Aggregate confusion: The
divergence of ESG ratings, Review of Finance, forthcoming.
Bauer, Rob, Tobias Ruof, and Paul Smeets, 2021, Get real! Individuals prefer more sustainable
investments, Review of Financial Studies 34, 3976-4043.
Bolton, Patrick, Xavier Freixas, and Joel Shapiro, 2012, The credit ratings game, Journal of
Finance 67, 86–111.
Cao, Jie, Sheridan Titman, Xintong Zhan, and Weiming Zhang, 2022, ESG preference,
institutional trading, and stock return patterns, Journal of Financial and Quantitative
Analysis, forthcoming.
Chen, Tao, Hui Dong, and Chen Lin, 2020, Institutional shareholders and corporate social
responsibility, Journal of Financial Economics 135, 483–504.
Chen, Xia, Jarrad Harford, and Kai Li, 2007, Monitoring: Which institutions matter? Journal
of Financial Economics 86, 279–305.
Cheng, Ing-Haw, Harrison Hong, and Kelly Shue, 2016, Do managers do good with other
people’s money? Working Paper, Dartmouth College.
Christensen, Dane, George Serafeim, and Anywhere Sikochi, 2022, Why is corporate virtue in
the eye of the beholder? The case of ESG ratings, The Accounting Review 97, 147-175.
Cohen, Lauren, Umit Gurun, and Quoc Nguyen, 2021, The ESG-innovation disconnect:
Evidence from green patenting, Working Paper, Harvard Business School.
Daines, Robert M., Ian D. Gow, and David F. Larcker, 2010, Rating the ratings: How good are
commercial governance ratings? Journal of Financial Economics 98, 439–461.
Di Giuli, Alberta, and Leonard Kostovetsky, 2014, Are red or blue companies more likely to
go green? Politics and corporate social responsibility, Journal of Financial Economics 111,
158–180.
Dimson, Elroy, Oğuzhan Karakaş, and Xi Li, 2015, Active ownership, Review of Financial
Studies 28, 3225–3268.
Dyck, Alexander, Karl V. Lins, Lukas Roth, and Hannes F. Wagner, 2019, Do institutional
investors drive corporate social responsibility? International evidence, Journal of
Financial Economics 131, 693–714.
31

Edmans, Alex, 2014, Blockholders and corporate governance, Annual Review of Financial
Economics 6, 23–50.
Engle, Robert F., Stefano Giglio, Bryan Kelly, Heebum Lee, and Johannes Stroebel, 2020,
Hedging climate change news, Review of Financial Studies 33, 1184–1216.
Fich, Eliezer M., Jarrad Harford, and Anh L. Tran, 2015, Motivated monitors: The importance
of institutional investors’ portfolio weights, Journal of Financial Economics 118, 21–48.
Flammer, Caroline, 2015. Does corporate social responsibility lead to superior financial
performance? A regression discontinuity approach, Management Science 61, 2549–2568.
Gibson, Rajna, Simon Glossner, Philipp Krüeger, Pedro Matos, and Tom Steffen, 2022, Do
responsible investors invest responsibly?, Working Paper, Swiss Finance Institute.
Gibson, Rajna, Philipp Krüeger, and Peter S. Schmidt, 2021, ESG rating disagreement and
stock returns, Financial Analysts Journal 77(4), 104-127.
Goldman, Eitan, and Wenyu Wang, 2021, Weak governance by informed active shareholders,
Review of Financial Studies 34, 661–699.
Goldstein, Itay, and Chong Huang, 2020, Credit rating inflation and firms’ investments.
Journal of Finance 75, 2929–2972.
Griffin, John, and Dragon Yongjun Tang, 2012, Did subjectivity play a role in CDO credit
ratings? Journal of Finance 67, 1293–1328.
Hartzmark, Samuel M., and Abigail B. Sussman, 2019, Do investors value sustainability? A
natural experiment examining ranking and fund flows, Journal of Finance 74, 2789–2837.
Hong, Harrison, and Marcin Kacperczyk, 2009, The price of sin: The effects of social norms
on markets, Journal of Financial Economics 93, 15–36.
Hong, Harrison, and Leonard Kostovetsky, 2012, Red and blue investing: Values and finance,
Journal of Financial Economics 103, 1–19.
Hong, Harrison, Jeffrey D. Kubik, and Jose A. Scheinkman, 2012, Financial constraints on
corporate goodness, Working Paper, Columbia University.
Jensen, Michael C., and William H. Meckling, 1976, Theory of the firm: Managerial behavior,
agency costs and ownership structure, Journal of Financial Economics 3, 305–360.
Kedia, Simi, Shivaram Rajgopal, and Xing (Alex) Zhou, 2017, Large shareholders and credit
ratings, Journal of Financial Economics 124, 632–653.
Khan, Mozaffar, George Serafeim, and Aaron Yoon, 2016, Corporate sustainability: First
evidence on materiality, The Accounting Review 91, 697–1724.
Kim, Incheol, Hong Wan, Bin Wang, and Tina Yang, 2019, Institutional investors and
corporate environmental, social and governance policies: Evidence from toxics release
data, Management Science 65, 4451–4949.
32

Kolbel, Julian, Timo Busch, and Leonhardt Jancso, 2017, How media coverage of corporate
social irresponsibility increases financial risk, Strategic Management Journal 38, 2266–
2284.
Krüger, Philipp, 2015, Corporate goodness and shareholder wealth, Journal of Financial
Economics 115, 304–329.
Krüger, Philipp, Zacharias Sautner, and Laura T. Starks, 2020, The importance of climate risks
for institutional investors, Review of Financial Studies 33, 1067–1111.
Li, Tao, 2018, Outsourcing corporate governance: Conflicts of interest within the proxy
advisory industry, Management Science 64, 2951–2971.
Li, Jun, and Di Wu, 2020, Do corporate social responsibility engagements lead to real
environmental, social, and governance impact? Management Science 66, 2291–2799.
Matos, Pedro, 2020, ESG and responsible institutional investing around the world: A critical
review, CFA Institute Research Foundation.
McCahery, Joseph A., Zacharias Sautner, and Laura T. Starks, 2016, Behind the scenes: The
corporate governance preferences of institutional investors, Journal of Finance 71, 2905–
2932.
Mehran, Hamid, and René M. Stulz, 2007, The economics of conflicts of interest in financial
institutions, Journal of Financial Economics 85, 267–296.
Pawliczek, Andrea, A. Nicole Skinner, and Laura Wellman, 2021, A new take on voice: The
influence of BlackRock’s “Dear CEO” letters, Review of Accounting Studies, 1088–1136.
Pedersen, Lasse Heje, Shaun Fitzgibbons, and Lukasz Pomorski, 2021, Responsible investment:
The ESG-efficient frontier, Journal of Financial Economics 142, 572-597.
Servaes, Henri, and Ane Tamayo, 2013, The impact of corporate social responsibility on firm
value: The role of customer awareness, Management Science 59, 1045–1061.
Shin, David Dongheon, 2021, Corporate ESG profiles, matching, and the cost of bank loans,
Working Paper, Indiana University.
Starks, Laura T., Parth Venkat, and Qifei Zhu, 2020, Corporate ESG profiles and investor
horizons, Working Paper, University of Texas at Austin.
Wu, Yue, Kaifu Zhang, and Jinhong Xie, 2020, Bad greenwashing, good greenwashing:
Corporate social responsibility and information transparency, Management Science 66,
3095–3112.
33

Appendix: Variable Definitions
Variable Definition
ESG Data (Source: KLD ESG, Refinitiv ESG, RepRisk, and SASB)
ESG rating The sum of ESG ratings across environment, social, and
governance categories
ESG strengths The total number of strength scores under the environment,
social, and governance categories
ESG concerns The total number of concern scores under the environment,
social, and governance categories
ESG ranking 20-quantile ranks converted from the original ESG rating
Refinitiv ESG ranking 20-quantile ranks converted from the Refinitiv ESG combined
rating
ESG incident counts The number of ESG incidents annually per firm
Severity The level of the harshness of an ESG incident on a firm
Material Financially material ESG issues as classified by SASB
Immaterial Non-Financially material ESG issues as classified SASB
Fundamental Data (Source: Compustat and CRSP)
Advertising intensity The ratio of advertising expenditure to total sales

Blue state dummy Equal to 1 if a firm’s headquarter is located in a state where
voters predominantly support the Democratic Party
Book-to-market The book value of equity divided by the market value of equity
Capital expenditure A firm’s capital expenditure scaled by total sales
Cash holdings Cash balances scaled by total assets
Dividends Cash dividends over book assets
Firm age The number of years since the firm was first covered by the
Compustat fundamental annual file
Leverage Total debt divided by total assets
Total assets The natural logarithm of total assets
Log SG&A expenses The natural logarithm of SG&A expenses
Log (SG&A-advertising) The natural logarithm of SG&A expenses minus advertising
expenses
Net income The fiscal period income or loss disclosed by a firm
R&D R&D expenditures scaled by total assets
R&D dummy Equal to 1 if R&D expenditure is missing, and 0 otherwise
ROA Ratio of income before extraordinary items to total assets
Prior-Year return Past one-year returns
Sales per employee A firm’s total sales scaled by its number of employees
SG&A A firm’s selling, general, and administrative expenses
SG&A/revenues SG&A expenses divided by revenue
34

Appendix: Variable Definitions⸺Continued
Variable Definition
Institutional-level and Analyst Data (Source: Thomson Reuters 13F, PRI and IBES)
Dedicated Institutional investors with holding periods equal to or longer

than two years
Transient Institutional investors with holding periods fewer than two years
Top Institutional investors that rank in the top five in terms of
average ownership of the parent firm of the rater
Bottom Institutional investors that rank in the bottom five in terms of
average ownership of the parent firm of the rater
Active A large shareholder who is an active player in the fund industry
Passive A large shareholder who is a passive player in the fund industry
Sister Equal to 1 if a firm is a large investee firm owned by the rater’s
large shareholder, and 0 otherwise
Closer Sister Equal to 1 if a firm represents at least 5% of the rater’s large
shareholder’s portfolio in a year, and 0 otherwise
Pseudo-Sister Equal to 1 for large investee firms of large shareholders before a
rater was acquired and started to provide ESG ratings, and 0
otherwise
Become Sister Equal to 1 if a firm becomes a sister firm in a year after 2009,
and 0 otherwise
Become Pseudo-Sister Equal to 1 if a firm becomes a sister firm in a year before 2009,
and 0 otherwise
Become Outsider Equal to 1 if a firm is no longer a sister firm in a year after 2009,
and 0 otherwise
Become Pseudo-Outsider Equal to 1 if a firm is no longer a sister firm in a year before
2009, and 0 otherwise
Institutional ownership The average of a firm’s total institutional ownership across a
year
PRI Signatories Equal to 1 if a rater’s stockholder is a signatory of the Principles
for Responsible Investment (PRI), and 0 otherwise
PRI Tenure The number of years since a stockholder joined PRI
Analyst Coverage The natural logarithm of (1+ the number of financial analysts
following the firm)

35

Google search volume index
0
20
40
60
80
100
120
2012-01
2012-04
2012-07
2012-10
2013-01
2013-04
2013-07
2013-10
2014-01

2014-04
2014-07
2014-10
2015-01
2015-04
ESG rating
2015-07
2015-10
2016-01
36
2016-04
2016-07
and “Credit rating” from January 2012 to April 2021.

2016-10
2017-01
2017-04
2017-07
2017-10
2018-01
2018-04
2018-07
2018-10
2019-01
2019-04
Credit rating
2019-07
2019-10
2020-01
2020-04
2020-07
2020-10
2021-01
2021-04
monthly times series trends of the Google Search Volume Index of the phrases “ESG rating”
Figure 1. Google search volume index: ESG rating vs. credit rating. This figure depicts the

30%
25%
Non-sister firms
20%
Frequency
15%
10% Sister firms
5%
0%
-8 -7 -6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19
ESG ratings

Figure 2. The ESG rating histogram: Sister vs. non-sister firms. This figure shows the
distribution of the ESG ratings for sister and non-sister firms after the ESG rater was acquired.
The sample period is 2010–2015. A sister firm is defined as a firm that is a large investee firm
held by a rater’s large shareholder. The solid black line represents sister firms, while the dotted
gray line represents non-sister firms. The horizontal axis indicates the numerical values of the
ESG ratings. The Kolmogorov-Smirnov test statistic is 0.3748, which is statistically significant
at the 1% level.
37

3.5
3.0
Coefficient estimates 2.5
2.0
1.5
1.0
0.5
0.0
0 1 2 3 4 5
Ownership cutoffs (%) for sister firms
Figure 3. Coefficient estimates from different definitions of sister firms. This figure reports
the regression coefficients on the Sister dummy estimated from our baseline model. A sister
firm is defined as a firm that accounts for 1% to 5% (as shown in the horizontal axis) of the
portfolio of one of the rater’s large shareholders. We plot the coefficients on the Sister dummy,
which are the estimates representing the differences in trends in ratings between sister firms
and other firms. The error bars use standard errors clustered at the firm level. All specifications
include industry times year-fixed effects.
38

2.0
1.5
1.0
ESG ratings
0.5
0.0
-0.5
-1.0
-1.5
Year-2 Year-1 Year 0 Year+1 Year+2
Pseudo sisters Pseudo nonsisters Sister firms Non-sister firms
Figure 4. ESG ratings of sister vs. non-sister firms. This figure displays the annual average
of the ESG rating for sister firms and the average of the ESG rating for other firms after the
ESG rater was acquired. A sister firm is defined as a large investee firm held by the ESG rater’s
large shareholder. The figure also reports the annual average of the ESG rating for pseudo-
sisters and the annual average of the ESG ratings for other firms before the rater was acquired.
Pseudo-sisters are firms held by the same large shareholders pre-acquisition.
39

6
5
The number of ESG incidents
0
Mar-07 Mar-08 Mar-09 Mar-10 Mar-11 Mar-12 Mar-13 Mar-14 Mar-15
Pseudo sisters Pseudo nonsisters Sister firms Non-sister firms
Figure 5. ESG incident counts: Sister vs. non-sister firms. This figure depicts the number of
ESG incidents made by sister and non-sister firms in 2010–2015 and the number of ESG
incidents by pseudo-sister and pseudo-non-sister firms in 2007–2009.
40

60
Sister firms
Non-sister firms
50
Number of future ESG incidents
40
30
20
10
0
[0-2] [3-5] [6-8] [9-11] [12-14] [15-18]
ESG concerns
Figure 6. ESG concern ratings and ESG incident count: Sister vs. non-sister firms. This
figure depicts the average number of ESG incidents by prior-year ESG concern rating ranges
in 2010–2015. A sister firm is defined as a large investee firm held by the rater’s large
shareholder. The black bar represents sister firms, and the gray bar represents non-sister firms.
41

Table 1. Summary Statistics
The summary statistics for the main ESG variables used in this study. Panel A presents the summary
statistics of ESG ratings. The ESG rating is the sum of ratings across the environment, social, and
governance categories, which is the number of ESG strengths minus ESG concerns. Panel B presents
the results of a univariate analysis of the ESG ratings and firm characteristics of sister and non-sister
firms. A firm is defined as a sister firm if it represents at least 0.25% of the large shareholder’s portfolio
of the rater in a given year, while other firms represent all the other firms in the sample. The last
column displays p-values from the mean difference tests of the two groups. The sample ranges from
2010 to 2015. ***, **, and * denote significance at the 1%, 5%, and 10% levels, respectively.
Panel A: ESG Measures
Characteristics N Mean Median Std Min Max

ESG strengths 11,145 1.61 0 2.79 0 22
ESG concerns 11,145 1.47 1 1.58 0 16
ESG rating 11,145 0.13 0 2.74 -8 19
Panel B: Sister vs. Non-Sister Firms
t-stat of the
Sister Non-Sister
difference
ESG rating 2.70 -0.07 2.77***
Total assets 9.28 7.13 2.15***
Book-to-Market 0.31 0.54 -0.23***
Firm age (in years) 3.33 2.98 0.35***
Sales per employee 0.57 0.45 0.12***
ROA 0.08 0.01 0.07***
Prior-Year return 0.26 0.24 0.02
Capital expenditure 0.10 0.12 -0.02*
R&D 0.06 0.15 -0.09***
R&D missing dummy 0.26 0.38 -0.12***
Dividends 0.02 0.01 0.01***
Cash holdings 0.16 0.19 -0.03***
Leverage 0.22 0.21 0.01**
Advertising intensity 0.02 0.01 0.01***
Institutional ownership 0.70 0.67 0.03***
Analyst coverage 3.19 2.29 0.90***
Blue state dummy 0.68 0.60 0.08***

42

Table 2. Determinants of ESG Ratings
The table reports the coefficients for linear regression models estimating determinants of ESG ratings.
The dependent variable is the ESG rating. The control variables are defined in the same way as in the
appendix. Standard errors are clustered at the firm level. t-statistics are in parentheses. The sample
period is 2003–2015, 2003–2009, and 2010–2015 in columns (1), (2), and (3), respectively. ***, **,
and * denote statistical significance at the 1%, 5%, and 10% levels, respectively.
Dependent Variable: ESG Ratings
2003–2015 2003–2009 2010–2015
(1) (2) (3)
Total assets 0.350*** -0.018 0.834***
(8.62) (-0.40) (16.32)
Book-to-Market -0.040 -0.018 -0.070
(-0.96) (-0.47) (-1.22)
Firm age 0.175*** 0.118** 0.288***
(3.79) (2.14) (5.16)
Sales per employee -0.164* -0.197* -0.132
(-1.86) (-1.81) (-1.32)
ROA 0.371*** 0.664*** -0.186
(2.76) (4.27) (-0.96)
Prior-Year return -0.066*** -0.099*** -0.094***
(-2.89) (-3.40) (-2.74)
Capital expenditure 0.309*** 0.517*** -0.001
(3.75) (4.71) (-0.01)
R&D -0.002 -0.016 0.011
(-0.06) (-0.39) (0.27)
R&D missing dummy -0.219*** -0.131 -0.338***
(-2.73) (-1.33) (-3.53)
Dividends 10.020*** 8.413*** 10.870***
(6.94) (4.69) (6.60)
Cash holdings 0.762*** 0.316* 1.277***
(4.51) (1.70) (5.56)
Leverage -0.544*** -0.580*** -0.799***
(-3.05) (-2.66) (-3.73)
Advertising intensity 6.122*** 6.486*** 5.765***
(4.60) (4.21) (3.58)
Institutional ownership -0.723*** -0.497*** -0.654***
(-4.98) (-2.93) (-3.75)
Analyst coverage 0.204*** 0.125** 0.244***
(4.20) (2.58) (3.17)
Blue state dummy 0.338*** 0.389*** 0.316***
(4.96) (4.90) (3.88)
Industry Year FE Yes Yes Yes
Adjusted R2 0.207 0.096 0.385
Observations 25,814 14,691 11,123
43

Table 3. Sister Firms and ESG Ratings
The table reports the coefficients for linear regression models estimating the association between ESG
rating and sister firms. The dependent variable is the ESG rating. The Sister dummy is equal to one if a
firm represents at least 0.25% of the portfolio of a large shareholder of the rater in a year, and zero
otherwise. An institutional investor is defined as a large shareholder if it owns at least 5% of the shares
of the rater. Standard errors are clustered at the firm level. t-statistics are in parentheses. The sample
period is 2010–2015. ***, **, and * denote statistical significance at the 1%, 5%, and 10% levels,
respectively.
(1) (2) (3) (4)
Sister 2.775*** 1.199*** 0.795***
(10.38) (4.97) (3.78)
Total assets 0.799*** 0.732*** 0.783***
(24.33) (24.83) (15.66)
Book-to-Market -0.054
(-1.01)
Firm age 0.289***
(5.25)
Sales per employee -0.131
(-1.32)
ROA -0.217
(-1.14)
Prior-Year return -0.101***
(-2.91)
Capital expenditure -0.002
(-0.03)
R&D 0.009
(0.23)
R&D missing dummy -0.302***
(-3.19)
Dividends 10.070***
(6.33)
Cash holdings 1.217***
(5.44)
Leverage -0.700***
(-3.32)
Advertising intensity 5.624***
(3.56)
Institutional ownership -0.629***
(-3.66)
Analyst coverage 0.222***
(2.89)
Blue state dummy 0.294***
(3.62)
Industry Year FE No No No Yes
Adjusted R2 0.070 0.225 0.237 0.390
Observations 11,145 11,145 11,145 11,123

44

Table 4. Sister Firms and ESG Ratings⸺Robustness Checks
This table reports the results of the robustness tests between ESG rating and sister firms. The dependent
variable is the ESG rating. In columns (1) –(4), the independent variable Sister is a dummy variable that
equals one if a firm is a large investee firm held by a large shareholder of the rater, and zero otherwise.
In column (1), we adopt an alternative definition of large shareholders. An institutional investor is
defined as a large shareholder if it owns at least 4% of rater shares. In column (2), we use an alternative
definition of Sister, which is equal to one if a firm represents at least 5% of the portfolio of a large
shareholder of the rater in a year and zero otherwise. In column (3), we define the variable Sister as the
average portfolio holding of large shareholders for each of their investees over a year. In column (4),
we adopt the propensity score matching with the logit model to find control firms for sister firms.
Propensity score matching is performed one year before the observation year with replacement. Each
treatment firm (i.e., a sister firm) is matched to a control firm using the nearest neighbor. Standard errors
are clustered at the firm level. t-statistics are in parentheses. The sample period is 2010–2015. ***, **,
Alternative Alternative Continuous Propensity
Large Sister Measure of Score Matched
Shareholders Definitions Connection Sample
(1) (2) (3) (4)
Sister 0.712*** 1.773** 0.272*** 0.666**
(3.94) (2.36) (3.14) (2.47)
Control variables Yes Yes Yes Yes
Industry Year FE Yes Yes Yes Yes
Adjusted R2 0.389 0.386 0.387 0.407
Observations 11,123 11,123 11,123 1,541
45

Table 5. Change in Sister Firm Status
The dependent variable is the ESG rating. In column (1), the dummy variable, Pseudo-Sister, is equal
to one if a firm is a large investee firm of the rater’s large shareholders before the rater was acquired.
In column (2), Become Sister (Become Pseudo-Sister) is equal to one if a firm becomes a sister firm
in a year after (before) 2009, and zero otherwise. In column (3), Become Outsider (Become Pseudo-
Outsider) is equal to one if a firm is no longer a sister firm in a year after (before) 2009, and zero
otherwise. The sample period is 2003–2015. Standard errors are clustered at the firm level. The
corresponding t-statistics are reported in parentheses. ***, **, and * denote statistical significance at
the 1%, 5%, and 10% levels, respectively.
(1) (2) (3)
Sister 1.791***
(8.28)
Pseudo-Sister -0.329**
(-2.62)
Become Sister 1.473***
(8.25)
Become Pseudo-Sister 0.162
(0.83)
Become Outsider 0.020
(0.14)
Become Pseudo-Outsider -0.269
(-1.58)
Control variables Yes Yes Yes
Adjusted R2 0.235 0.246 0.218
Observations 23,735 23,735 23,735

46

Table 6. ESG Ratings and Relationships
This table reports the results of regressions of ESG ratings on different measures of the relation among
the rater, its large shareholders, and its sister firms. The dependent variable is the ESG rating. The Sister
dummy is equal to one if a firm is a large investee firm held by a large shareholder of the rater, and zero
otherwise. An institutional investor is defined as a large shareholder if it owns at least 5% of the shares
of the rater. In columns (1)–(2), Sister is equal to one if a sister firm is owned by the rater's dedicated
and transient large shareholders, respectively, and zero otherwise. Dedicated and transient shareholders
are investors with holding periods over the rater for longer than and less than two years, respectively. In
columns (3)–(4), Sister takes the value of one if a sister firm is held by the rater’s large shareholders,
which rank in the top five and bottom five in terms of average ownership on the rater, respectively. In
columns (5)–(6), Sister is equal to one if a sister firm is held by a large shareholder who is an active and
passive player in the fund industry, respectively, and zero otherwise. Standard errors are clustered at the
firm level. t-statistics are in parentheses. The sample period is 2010–2015. ***, **, and * denote
statistical significance at the 1%, 5%, and 10% levels, respectively.
Investment Horizon Ownership Size Management Style
Dedicated Transient Large Small Active Passive
(1) (2) (3) (4) (5) (6)
Sister 0.842*** 0.273 1.098*** 0.347 0.728*** 0.669
(3.56) (1.02) (3.88) (1.18) (3.71) (0.68)
Control variables Yes Yes Yes Yes Yes Yes
Industry Year FE Yes Yes Yes Yes Yes Yes
Adjusted R2 0.389 0.385 0.390 0.385 0.383 0.385
Observations 11,123 11,123 11,123 11,123 11,123 11,123
47

Table 7. Sister Firms and ESG Rating by ESG Materiality
This table shows the results of analyzing the effect of sister-firm status on material and immaterial ESG.
The firm-level ESG rating indicators are mapped using the Sustainability Accounting Standards Board
(SASB) sector-specific values to classify the ESG indicators into financially material and immaterial
categories. The SASB standards define sector-specific sustainability likely to materially affect the
financial conditions or operating performances of companies. In columns (1)–(2), (3)–(4), and (5)–(6),
ESG, ESG strengths, and ESG concerns ratings are classified into material and immaterial types
according to the SASB Materiality Map. The sample period is 2010–2015. ∗∗∗, ∗∗, and ∗ indicate
statistical significance at 1%, 5%, and 10%, respectively.
ESG Ratings ESG Strengths ESG Concerns
Material Immaterial Material Immaterial Material Immaterial
(1) (2) (3) (4) (5) (6)
Sister 0.157 0.638*** 0.213*** 1.222*** 0.056 0.584***
(1.63) (3.58) (2.66) (6.69) (0.91) (6.02)
Control variables Yes Yes Yes Yes Yes Yes
Industry Year FE Yes Yes Yes Yes Yes Yes
Adjusted R2 0.257 0.362 0.327 0.454 0.381 0.442
Observations 11,123 11,123 11,123 11,123 11,123 11,123
48

Table 8. Sister Firms and Benchmarked ESG Ratings
This table presents the results of employing alternative ESG ratings provided by Refinitiv as benchmark
ratings. In column (1), the dependent variable, Benchmarked ESG, is defined as KLD ESG rankings
minus Refinitiv ESG rankings. In columns (2)–(3), dependent variables are alternative ESG rankings
provided by Refinitiv in the subsequent first and second years, respectively. Standard errors are
clustered at the firm level. t-statistics are in parentheses. The sample period is 2010–2015. ***, **, and
* indicate statistical significance at the 1%, 5%, and 10% levels, respectively.
Benchmarked ESG Alternative ESG Alternative ESG
Year t Year t + 1 Year t + 2
(1) (2) (3)
Sister 1.226*** 0.117 0.101
(2.88) (1.26) (1.05)
ESG 0.031*** 0.030***
(10.28) (9.71)
Sister × ESG -0.012** -0.012*
(2.15) (1.96)
Control variables Yes Yes Yes
Adjusted R2 0.144 0.168 0.165
Observations 3,882 3,882 3,882
49

Table 9. Incentive Alignment with Owner

This table reports how the relation between ESG rating and sister firms varies with the number of large
shareholders holding a sister firm as well as with the total net assets of large shareholders. An
institutional investor is defined as a large shareholder if it owns at least 5% of the shares of the rater.
The dependent variable is the ESG rating. In column (1), the Sister dummy is equal to one if a firm
represents at least 0.25% of the rater’s large shareholder’s portfolio and is held by multiple large
shareholders in a year, and zero otherwise; in column (2), Sister is equal to one if a firm represents at
least 0.25% of the rater’s large shareholder’s portfolio and is held by a single large shareholder in a year,
and zero otherwise. In column (3), the Sister is equal to one if a firm represents at least 0.25% of the
rater’s large shareholder’s portfolio and the asset under management (AUM) of the large shareholder is
above the median compared to all large shareholders in a year, and zero otherwise. In column (4), Sister
is equal to one if a firm represents at least 0.25% of the rater’s large shareholder’s portfolio and the
AUM of the large shareholder is below the median compared to all large shareholders in a year, and
zero otherwise. Standard errors are clustered at the firm level. The sample period is 2010–2015. ***, **,
Dependent Variable: ESG ratings
Number of AUM of
Large Shareholders Large Shareholders
Multiple Single High Low
(1) (2) (3) (4)
Sister 1.221*** 0.458** 1.038*** 0.259
(2.96) (2.36) (3.42) (1.16)
Adjusted R2 0.388 0.386 0.388 0.385
Observations 11,123 11,123 11,123 11,123

50

Table 10. Large Shareholders’ Incentives
This table reports the effect of large shareholders' incentives on ESG ratings. An institutional investor
is defined as a large shareholder if it owns at least 5% of the shares of the rater. The dependent variable
is the ESG rating. In column (1), the Sister dummy is equal to one if a firm represents at least 0.25% of
the rater’s large shareholder’s portfolio, and the average number of CSR proposals presented by a large
shareholders’ important investee firms is above the median compared to all large shareholders, and zero
otherwise. In column (2), Sister is equal to one if a firm represents at least 0.25% of the rater’s large
shareholder’s portfolio, and the average number of CSR proposals presented by a large shareholders’
important investee firms is below the median compared to all large shareholders, and zero otherwise. In
column (3), Sister is equal to one if a firm represents at least 0.25% of the rater’s large shareholder’s
portfolio, and a large shareholders’ important investee firms encounter an above-average number of
FTSE4 Good Index exclusion events, and zero otherwise. In column (4), Sister is equal to one if a firm
represents at least 0.25% of the rater’s large shareholder’s portfolio and a large shareholders’ important
investee firms encounter a below-average number of FTSE4 Good Index exclusion events, and zero
otherwise. Standard errors are clustered at the firm level. The sample period is 2010–2015. ***, **, and
* denote statistical significance at the 1%, 5%, and 10% levels, respectively.
Dependent Variable: ESG ratings
Number of Number of
CSR Proposals Index Exclusion Events
High Low High Low
(1) (2) (3) (4)
Sister 0.951*** 0.326** 1.709*** 0.735**
(9.07) (2.29) (8.22) (3.89)
Adjusted R2 0.388 0.384 0.388 0.385
Observations 11,123 11,123 11,123 11,123
51

Table 11. Sister Firms and Future ESG Incidents

This table displays the count data model estimation results of the relationship between sister firms and
future ESG risk incidents. The dependent variable is the count of ESG incidents for a firm in a year as
captured by RepRisk. Poisson regressions and negative binomial regressions are estimated in columns
(1)–(2), and (3)–(4), respectively. Standard errors are clustered at the firm level. The sample period is
2010–2015. ∗∗∗, ∗∗, and ∗ indicate statistical significance at the 1%, 5%, and 10% levels, respectively.
Dependent Variable: ESG Incident Counts
Poisson Model Negative Binomial Model
(1) (2) (3) (4)
Sister 2.227*** 0.189** 2.639*** 0.206***
(15.62) (2.12) (14.40) (2.80)
ESG Concerns 0.178*** 0.191***
(12.02) (12.16)
Control variables No Yes No Yes
Log pseudo likelihood -15384.24 -6900.07 -8377.10 -6122.72
Observations 10,255 9,921 10,225 9,921

52

The Determinants of ESG Ratings - Rater Ownership Matters

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

The Determinants of ESG Ratings - Rater Ownership Matters

Uploaded by

Copyright:

Available Formats

The Determinants of ESG Ratings: Rater Ownership Matters*

Dragon Yongjun Tang, Jiali Yan, Chelsea Yaqiong Yao

Keywords: ESG, Rating agencies, Conflicts of interest, Ownership structure

Keywords: ESG, Rating agencies, Conflicts of interest, Ownership structure

[Insert Figure 1 here]

Understanding the ESG rating quality is critical to the further development of

We empirically examine the determinants of ESG ratings, with a focus on potential

We recognize the potential endogeneity in firm ownership and take a meticulous

2. Institutional Background on ESG Ratings and Hypotheses Development

2.1. ESG ratings

ESG Rating. An ESG rating is an assessment of a firm’s performance in its approach to

Rating agencies, as information intermediaries, have an incentive to provide accurate

3. Data and Summary Statistics

3.1. ESG rating data

[Insert Table 1 here]

3.2. Identifying shareholders and sister firms

We obtain ownership structure information from the Thomson Reuters Institutional

We define an investee firm as a sister if it comprises at least 0.25% of a large

[Insert Figure 2 here]

3.3. Empirical model specification

[Insert Table 2 here]

4. Sister Firms and ESG Ratings

4.1. ESG rating and sister firms

To answer these questions, we begin by examining the relationship between ESG

[Insert Table 3 here]

4.2. Robustness tests

In this subsection, we examine the robustness of our main findings by constructing

4.2.1. Large Shareholders. We propose an alternative definition of “large shareholders.”

[Insert Table 4 here]

4.2.3. Continuous Measure of Connection. We use an alternative continuous measure

4.2.4. Matched Sample. We implement propensity score matching to address a concern

[Insert Figure 4 here]

[Insert Table 5 here]

5. Channels and Mechanisms

5.1. Owner style and influence

We define an institutional investor as a dedicated (transient) large shareholder of the

[Insert Table 6 here]

To distinguish between material and immaterial ESG indicators, we follow SASB

[Insert Table 7 here]

5.3. Benchmarking against alternative ESG ratings

Excluding the previously discussed factors, we consider whether any unobserved or

[Insert Table 8 here]

5.4. Incentive alignment

[Insert Table 9 here]

5.5. Large shareholders’ incentives

In this subsection, we investigate how large shareholders’ incentives relate to ESG

We obtain information on CSR shareholder proposals from Institutional Shareholder

[Insert Table 10 here]

5.6. Future ESG performance

[Insert Figure 5 here]

[Insert Figure 6 here]

[Insert Table 11 here]

Fundamental Data (Source: Compustat and CRSP)

Advertising intensity The ratio of advertising expenditure to total sales

Dedicated Institutional investors with holding periods equal to or longer

Google search volume index

and “Credit rating” from January 2012 to April 2021.

10% Sister firms

Coefficient estimates 2.5

Pseudo sisters Pseudo nonsisters Sister firms Non-sister firms

Panel A: ESG Measures

Characteristics N Mean Median Std Min Max