Pacific-Basin Finance Journal 78 (2023) 101952

Contents lists available at ScienceDirect

Pacific-Basin Finance Journal

journal homepage: www.elsevier.com/locate/pacfin

The green fog: Environmental rating disagreement and

corporate greenwashing
Xinwen Hu a, Renhai Hua b, *, Qingfu Liu c, d, Chuanjie Wang c, *
a
School of Management and Economics, The Chinese University of Hong Kong, Shenzhen, China
b
School of Finance, Nanjing University of Finance and Economics, Nanjing, China
c
School of Economics, Fudan University, Shanghai, China
d
Yanqi Lake Beijing Institute of Mathematical Sciences and Applications, Beijing, China

A R T I C L E I N F O A B S T R A C T

JEL classification: We investigate whether the environmental rating disagreement (ERD) would influence future
D22 corporate greenwashing behavior. We find that ERD would increase the future probability of
Q56 corporate greenwashing, mainly through mechanisms of agency costs and corporate information
G24
opacity. The relationship between ERD and corporate greenwashing is stronger in heavily-
Keywords: polluted firms, and would be significantly affected by internal incentives. In the presence of
Environmental rating disagreement
ERD, corporate greenwashing of firms with higher internal supervision intensity and external
Corporate greenwashing
investor attention exhibits a smaller or even insignificant increase. Overall, our findings indicate
Agency costs
Firm opacity that greenwashing could be curbed through the unification of environmental rating standards, the
improvement of internal supervision improvement, and the expansion of external monitoring.

“If a company scores highly on one ESG ranking but not another, what conclusions should an investor draw?”1
Financial Times, March 19, 2022.

1. Introduction

A flourishing industry of ESG rating consultants has sprung up, as issues of climate change and sustainability move up the investor
agenda (Li et al., 2022). It is widely acknowledged that rating agencies should serve as gatekeepers to capital markets (Xia, 2014), since
the proper allocation of resources in an economy requires institutions that provide information intermediation (Healy and Palepu,
2001; Serafeim and Yoon, 2022). However, contrary to the original intention of ESG ratings, ESG ratings from different providers
disagree substantially (Chatterji et al., 2016; Berg et al., 2022), and even create chaotic signals. As the ESG objective is becoming a
primary focus in asset management, an emerging stream of literature has focused on the fact that differences or uncertainties in ESG
ratings significantly affect stock returns, volatility, and external financing possibilities (Avramov et al., 2022; Christensen et al., 2022).
Regardless of the rating agency, credit ratings for a particular issuer are broadly similar. This is not the case for ESG ratings (Dimson

* Corresponding authors.
E-mail addresses: [email protected] (X. Hu), [email protected] (R. Hua), [email protected] (Q. Liu), [email protected]
(C. Wang).
1
This paragraph is taken from the article entitled Boom in ESG ratings leaves trail of confusion published on March 19, 2022. See more details from
https://www.ft.com/content/c34fe314-838b-4b00-ae25-9a4f0d93f822.

https://doi.org/10.1016/j.pacfin.2023.101952
Received 19 October 2022; Received in revised form 7 January 2023; Accepted 26 January 2023
Available online 30 January 2023
0927-538X/© 2023 Elsevier B.V. All rights reserved.
X. Hu et al. Pacific-Basin Finance Journal 78 (2023) 101952

et al., 2020). Typically, in the dataset constructed by Berg et al. (2022), the correlations between ESG ratings range from 0.38 to 0.70.
Further, Berg et al. (2022) systematically sorted out the drivers of rating disagreement: scope divergence, measurement divergence,
and weight divergence. Consequently, the rating confusion, like a fog, has obscured investors’ vision, especially on how one can judge
their quality. As a guide to investors, some literature has studied the performance of relevant portfolios, and shown a positive cor­
relation between ESG rating divergence and stock returns (Gibson Brandon et al., 2021). In addition, the disagreement can also in­
crease the CAPM alpha and effective beta (Avramov et al., 2022). However, to date, little attention has been devoted to the signal from
rating disagreement conveyed to companies or managers, that is, how firms and their managers take actions or respond to the rating
disagreement.
Against this backdrop, this paper aims to fill this gap by analyzing how the rating disagreement impacts corporate greenwashing
behavior. Flammer (2021) defined greenwashing as making unsubstantiated or misleading claims about the company’s environmental
commitment. With the growing popularity of corporate social responsibility, critics pointed out that firms tend to use catchy termi­
nologies like ESG as a marketing gimmick, often labeled as greenwashing (Wu et al., 2020). Recent evidence suggests that green­
washing may undermine favorable perceptions and a firm’s profitability, and result in ethical harm (Nyilasy et al., 2014; Du, 2015;
Szabo and Webster, 2021). While several studies have shown the proliferation of greenwashing is driven by various external, orga­
nizational, and individual issues (Guo et al., 2017), one question remains to be addressed, i.e., whether the disagreement of the third
party’s rating will stimulate corporate greenwashing.
To complete the identification task, we restrict our analysis of rating disagreement to the environmental dimension, i.e., envi­
ronmental rating disagreement (hereinafter referred to as ERD). The reason is straightforward. The ESG rating includes environmental,
social, and governance dimensions, while the latter two seem to have little to do with greenwashing. That is, companies with potential
motives for greenwashing have a stronger perception of ERD than ESG rating disagreement.
Further, it is central to our paper on how we construct the measurements or proxies of ERD and corporate greenwashing. Following
Avramov et al. (2022), Christensen et al. (2022), and Serafeim and Yoon (2022), we can easily define ERD as the standard deviation of
environmental ratings or the average of pairwise rating uncertainty across rating agencies. The measurement construction for
corporate greenwashing remains a barrier for our empirical setting. Although one set of studies has deeply analyzed greenwashing, for
example, Parguel et al. (2011), no consensus has been reached on the greenwashing measurement. Inspired by Loughran et al. (2009),
we construct the greenwashing proxy variable in our paper by comparing the inconsistency between oral green publicity and actual
green performance. As a robustness check, following Yu et al. (2020) and Zhang (2022a), we also consider an alternative proxy for
corporate greenwashing, comparing the environmental disclosure score and the environmental real-performance score.
We approach our problem using Chinese common stocks listed on the main board of the Shanghai and Shenzhen Stock Exchange
from 2015 to 2020. We collect environmental ratings from five major rating agencies: Hexun, Bloomberg, Huazheng, SynTao Green
Finance, and the Wind ESG database. With these firm-level data, our empirical analysis starts with examining how the ERD affects
corporate greenwashing. To better mitigate the endogeneity concern and reflect the corporate greenwashing behavior lag, we consider
incorporating the independent and control variables with a one-period lag. In this setup, corporate greenwashing exhibits a significant
positive relation to ERD. Firms and their managers thus tend to conduct greenwashing behavior when they observe environmental
disagreement across different agencies. The results are robust to a battery of robustness tests, including altering the dependent variable
and the model setting.
The second question relates to the mechanisms or channels through which ERD affects corporate greenwashing. To this end, our
first mechanism is placed on the firm’s opacity. Following Hutton et al. (2009), we construct the proxy for corporate information
opacity, and find that ERD would increase firm-level opacity, thus stimulating the incentives of corporate greenwashing. Next, we
measure the firm’s agency costs following Singh and Davidson (2003). The regression results confirm that ERD could push up agency
costs, finally resulting in corporate greenwashing.
To understand whether firm’s internal and external characteristics will magnify or weaken the main effect, our third analysis is on a
sequence of heterogeneity tests. First, we find that the positive effect is more pronounced in heavily-polluted subsamples than in less-
polluted ones. Second, we demonstrate empirically how the internal incentives affect the positive relationship between ERD and
corporate greenwashing, and the results indicate that firms with higher-paid executives and lower financial constraints, are more
“energetic” in conducting greenwashing behavior. Third, the role of internal supervision is also included in the further analysis.
Notably, if firms have more independent directors or a higher proportion of foreign shares, the positive effect of ERD on corporate
greenwashing will be reduced. Lastly, we also find that external investor attention would suppress the motives of corporate green­
washing, but investors’ positive sentiment would have the opposite influence.
This paper contributes to several strands of the literature. First, we contribute to the literature examining rating disagreement’s
consequences. Prior work has focused on the capital market reaction to rating disagreement, especially ESG rating disagreement, and
provides evidence from investors’ perspectives (Gibson Brandon et al., 2021; Christensen et al., 2022; Serafeim and Yoon, 2022). We
add to this stream of literature by verifying the impact of rating disagreement on corporate behaviors. To the best of our knowledge, we
are the first to analyze the impact of environmental rating disagreement on corporate greenwashing, which is an important concern
about the firm’s greenness.
Second, we contribute to the growing literature on measures to constrain corporate greenwashing. Delmas and Burbano (2011)
argued that mitigating greenwashing is particularly challenging in a context of limited and uncertain regulation. A few prior studies
show that sustainability ratings might deter greenwashing (Parguel et al., 2011), and firm-level governance factors effectively
attenuate firms’ greenwashing behavior (Yu et al., 2020). Our contribution is to comprehensively propose how to prevent green­
washing from internal and external perspectives, that is, reducing the greenwashing motives of the firm’s executives, increasing the
internal supervision power, and strengthening the supervision of external public attention.

2
X. Hu et al. Pacific-Basin Finance Journal 78 (2023) 101952

Third, our work enriches academic discussions in constructing greenwashing measurements. Due to the lack of consensus on
quantifying greenwashing, a stream of literature measures corporate greenwashing relying on the ESG ratings, for example, Yu et al.
(2020) and Zhang (2022a). Combining the textual analysis technology and the widely-accepted definition of greenwashing (Loughran
et al., 2009; Flammer, 2021), this paper proposes a novel measurement, i.e., by comparing the inconsistency between the oral green
publicity and the actual green performance. In terms of greenness, if a firm speaks well and does poorly, it will be marked as
greenwashing.
The remainder of this paper is organized as follows. Section 2 presents the literature review and our hypotheses. Section 3 describes
the data and the main variables used. Section 4 presents the research design and main results. Section 5 provides the results of further
analysis. The conclusion follows in Section 6.

2. Literature review and hypotheses development

In this section, we review the related literature and propose our main hypotheses, including the relationship between environ­
mental rating disagreement and corporate greenwashing, the mechanisms, and several further analyses.

2.1. ERD and corporate greenwashing

The growing salience of ESG is not unique to companies and is also prevalent in the asset management industry. For example, the
assets under management (AUM) of the United Nations Principles for Responsible Investment (PRI) signatories ballooned to 121.3
trillion dollars in 2021. However, even in the fund industry, Liang et al. (2022) argued that managers may engage in greenwashing.
Shifting our focus onto the implication of ERD for firms and their managers, if environmental ratings appropriately reflect a
management’s efforts to limit negative environmental events and to promote positive environmental events, different rating agencies
can also form ratings with little difference. However, when firms and managers observe notably divergent environmental ratings, they
may regard it as a signal that the rating agencies are confused. Meanwhile, ERD appears to introduce uncertainty regarding the
greenness of a firm (Christensen et al., 2022). This may serve as market friction that stimulates the firm’s motive to conduct green­
washing, since a monopolistic rating agency can credibly threaten to punish issuers (Fulghieri et al., 2014), but divergent ones cannot.
Our first hypothesis then is:
H1. There is a positive relationship between environmental rating disagreement and corporate greenwashing.

2.2. The mechanisms

Morgan (2002) referred to opacity (or non-transparency) as a situation in which risks are hard to observe for an outsider. Numerous
empirical studies have argued that rating splits could exacerbate or even serve as a proxy for the opacity of banks (Bannier et al., 2010),
corporate bonds (Güntay and Hackbarth, 2010; Hauck and Neyer, 2014), and loans (Drucker and Puri, 2009). Kothari et al. (2009), and
Jin and Myers (2006) further argued that the lack of transparency enables managers to hide bad news from investors for extended
periods. Thus, these arguments lead to our second hypothesis:
H2. ERD could increase corporate opacity, then stimulate greenwashing behavior.
Next, we make predictions on the role of agency costs. Previous literature has shown that transparency is the paramount
consideration in models of agency costs (Edelen et al., 2012). Bhattacharya et al. (2003) also showed that better accounting infor­
mation helps investors distinguish between good and bad managers, decreasing agency costs. In such a case, since rating disagreement
decays the information transparency, increasing the firm’s agency costs, firms will likely choose greenwashing strategy to improve its
ostensible greenness and corresponding reputation. This leads us to set forth the following hypothesis:
H3. ERD could increase agency costs, then stimulate greenwashing behavior.

2.3. The effects of internal and external characteristics

We expect that several firm characteristics will moderate the relationship between ERD and corporate greenwashing.
First, from prior literature examining the difference between heavily- and lowly-polluted firms, we know that the former suffers
from high pollution abatement costs and the costs of meeting the environmental regulation requirements (Zhang, 2022b). Meanwhile,
firms operating in a deregulated environment are more likely to greenwash (Kim and Lyon, 2015). Thus, heavily-polluted firms would
be more devoted to conducting greenwashing. To test the above statement, we set forth our fourth hypothesis as follows:
H4. The positive relationship between ERD and corporate greenwashing will be stronger in the heavily-polluted firms.
Second, the theory has indicated that a firm’s internal incentives would affect corporate behavior. In our analysis, we incorporate
two typical elements that would impact the internal incentives, i.e., executive salary and the firm’s financial status. First, although
manager incentives are a powerful tool that can mitigate the agency problems inherent in the separation of ownership and control,
they can also lead to negative behaviors of managers, such as financial fraud (Johnson et al., 2009). On the one hand, we anticipate that
managers with a higher salary would likely whitewash the firm’s green performance, to maintain their compensation and reputation,
since the disagreement shows that it is challenging to link CEO compensation to environmental rating performance (Berg et al., 2022).

3
X. Hu et al. Pacific-Basin Finance Journal 78 (2023) 101952

On the other hand, the internal incentive of greenwashing is remarkably restrained by financial sources. According to studies that can
collectively be termed the resource constraints literature, resource constraints alter how resources are garnered and expended, forcing
managers to improve allocative efficiency, instead of greenwashing (Baker and Nelson, 2005; George, 2005). These arguments lead to
our fifth hypothesis:
H5. The positive relationship between ERD and corporate greenwashing will be weaker in the financially-constrained firms, but
stronger with higher executive salaries.
Third, the theory has also indicated that a firm’s internal supervision would affect corporate behavior. This paper mainly focuses on
two critical sources of supervisory power: independent directors and foreign investors. Recent literature suggests that regulation,
especially government regulation, has reduced corporate greenwashing behavior (Delmas and Burbano, 2011; Sun and Zhang, 2019).
Under such a circumstance, we also expect that the strength of internal supervision will have similar effects. For one thing, several
pieces of literature have found that foreign ownership plays a role in organizational processes and changes in corporate governance
patterns (Desender et al., 2016). Such an effect is related to higher commitment and longer-term involvement of foreign capital (Wei
et al., 2005), leading to increased firm value (Douma et al., 2006). For another thing, independent directors have also been verified to
be an effective monitoring power (Byrd and Hickman, 1992), and firms with more independent directors are associated with better
operating performance and higher firm value (Masulis and Mobbs, 2014). Under this scenario, our sixth hypothesis is:
H6. The positive relationship between ERD and corporate greenwashing will be weaker in the firms with more independent directors
and a higher shareholding ratio of foreign investors.
Finally, we consider the influence of investors’ attention and sentiment on the corporate greenwashing. There is a rich literature in
accounting and finance that examines the impact of investor attention, for instance, on IPO performance (Bajo et al., 2016), managers’
disclosure choices (Abramova et al., 2020) and corporate green innovation (He et al., 2022b). Thus, the general finding is that investor
attention forces managers to review their own behavior, which may weaken the positive impact of ERD on corporate greenwashing.
Apart from investor attention, another stream of literature suggests that investor sentiment has a significantly impact on managerial
decisions (Bergman and Roychowdhury, 2008; Brown et al., 2012). Amin et al. (2021) further found that high investor sentiment has
been linked with opportunistic managerial behavior. Together, these arguments lead to our last hypothesis:
H7. The positive relationship between ERD and corporate greenwashing will be weaker with more external investor attention and
negative investor sentiment.

3. Data

3.1. Data sources

Our sample consists of all common stocks listed on the main board of the Shanghai and Shenzhen Stock Exchange. We collect
environmental rating data from five vendors, including Bloomberg, Hexun, Huazheng, SynTao Green Finance, and the Wind ESG
database. These data providers represent the major players in the Chinese ESG rating market, and their ratings are widely used by
practitioners as well as in a growing number of academic studies (Chang et al., 2021; Avramov et al., 2022; Chen et al., 2023; He et al.,
2022a; Zhou et al., 2022). To construct the corporate greenwashing measurement, we obtain text data of the firm’s annual reports from
the Wingo Platform, and environmental-related punishment data from CSMAR and CNRDS databases. Annual financial statement data
come from the Wind database.
The entire sample period ranges from 2015 to 2020, and for each firm-year observation, we require environmental ratings from at
least three data vendors. We have also excluded firms in the financial industry with special capital structures. Since our panel data are
unbalanced, our total sample for the main analysis includes around 10,800 unique firm-year observations.

3.2. Main variables

3.2.1. Corporate greenwashing

Greenwashing is the practice of making unsubstantiated or misleading claims about the company’s environmental commitment
(Flammer, 2021). To quantify the corporate greenwashing in our paper, we need to measure the firm’s oral green publicity (Oral) and
actual environmental performance (Actual), respectively. First, following Loughran et al. (2009) and Hoberg and Lewis (2017), we
have constructed a greenness- or environment-related term set, including the term “Green”, “Environmental Protection”, “Low-Carbon”
and “Environment”. Then, for each firm-year observation, we analyze the frequency of these words in the MD&A section of the annual
report.2 The dummy variable Oral equals one if the frequency of the observation is greater than the median of the same industry in the

2
In our robustness check, we also incorporate the sentence frequency, including these items, to construct the variable Oral.

4
X. Hu et al. Pacific-Basin Finance Journal 78 (2023) 101952

same period, and vice versa. Next, the actual environmental performance Actual equals one if firm-year observation is subject to
environmental punishment, and vice versa.3 Taken together, the proxy variable for corporate greenwashing (DGW) is calculated as
follows:
{
1, if Orali,t = 1 and Actuali,t = 1
DGW i,t = (1)
0, o.w.

The connotation of the indicator variable equal to one is very intuitive; that is, the firm has carried out too much greenness
publicity, but the actual environmental performance is abysmal the contrary. We also follow Zhang (2022a) for the robustness test to
measure a firm’s peer-relative greenwashing score, as defined in Eq. (2).
( ) ( )
ERi,t − ER¯dis ERi,t − ER¯per
GW i,t = − (2)
σdis σ per

The first term denotes the normalized measure of a firm’s position relative to its peers in the distribution of the disclosure score of
environmental rating (ER), and the second term is a normalized measure of a firm’s position relative to its peers in the distribution of its
real-performance score of environmental rating. Specifically, ER¯dis and ER¯per represent the average value of environmental disclosure
and performance scores, respectively. σ dis and σper represent the standard deviations of environmental disclosure and performance
scores, respectively. Practically, the Bloomberg environmental rating is regarded as the ER disclosure score, while the Wind envi­
ronmental rating is regarded as the ER real-performance score.

3.2.2. Environmental rating disagreement

We focus on the environmental rating from each data provider, and construct two types of ERD measurement. Since ratings from
the different providers are organized in different structures, following previous studies (Christensen et al., 2022; Serafeim and Yoon,
2022), we standardize all ratings to a range of 0 to 1, making them comparable across various sources. Under such a circumstance, we
define the first proxy variable for environmental rating disagreement ERD_I as the standard deviation of these environmental ratings.
Further, we construct our second proxy variable for environmental rating disagreement ERD_II following Avramov et al. (2022).
Specifically, let ERi, t, A and ERi, t, B denote the environmental rating for stock i in year t from raters A and B, respectively. The pairwise
⃒ ⃒/√̅̅̅
rating disagreement is calculated as ⃒ERi,t,A − ERi,t,B ⃒ 2, and then we compute the firm-level environmental rating disagreement as
the average pairwise rating disagreement across all rater pairs.

3.2.3. Control variables

Following the previous literature, for instance, Kim and Lyon (2015), we incorporate several control variables in our regression
model. The variable ln(Age) is the log of the firm’s age. Size is the log of the firm’s total assets. Board and Dep are the log of board size
and the proportion of independent directors on the board, respectively. Dual is a dummy variable indicating whether the Chairman and
CEO are the same people. Lev is the leverage ratio measured by total liabilities divided by total assets. ROA and ROE are defined as the
return on assets and return on equity, respectively. SOE is also a dummy variable indicating the firm’s ownership status. All of the
variable definitions are listed in Table 1.

Table 1
Variable definitions.
Variable Definition

DGW A dummy variable of a firm’s greenwashing that takes the value of 1, if in the “Management Discussion and Analysis” Section of the annual report, the
frequency of words related to environmental protection is higher than the median of firms in the same industry in the same period, but the firm is
punished for environmental issues in the same year, and vice versa.
ERD_I The standard deviation of the environmental rating scores from three rating agencies.
ERD_II The average pairwise environmental rating uncertainty across all rater pairs.
ln(Age) The natural logarithm of firm’s age.
Size The natural logarithm of firm’s total assets.
Board The natural logarithm of board size.
Dep The ratio of the independent directors on the board.
Dual An indicator variable that equals 1 if the Chairman and CEO are the same people, and vice versa.
Lev The leverage ratio, equal to total liabilities divided by total assets.
ROA Return on assets.
ROE Return on equity.
SOE An indicator variable that equals 1 if the firm is state-owned, and vice versa.

3
One of the responsibilities of China’s Ministry of Ecology and Environment, as well as environmental protection departments at all levels of
government, is to supervise and enforce the law and punish those who break the rules. Therefore, when listed companies are punished for violating
environmental laws, both the government and the company will disclose the punishment details. In this case, we determine that the company
suffered environmental punishment that year.

5
X. Hu et al. Pacific-Basin Finance Journal 78 (2023) 101952

Table 2
Summary statistics.
Variable Obs. Mean S.D. P10 Median P90

DGW 19,207 0.028 0.166 0.000 0.000 0.000

ERD_I 10,973 0.095 0.079 0.000 0.077 0.207
ERD_II 10,973 0.212 0.088 0.059 0.236 0.302
ln(Age) 19,206 2.969 0.313 2.565 2.996 3.332
Size 19,206 22.407 1.394 20.723 22.265 24.272
Board 18,963 2.124 0.200 1.946 2.197 2.398
Dep 18,963 0.374 0.053 0.333 0.333 0.429
Dual 18,866 0.261 0.439 0.000 0.000 1.000
Lev 19,206 0.450 0.205 0.181 0.441 0.728
ROA 19,147 0.058 0.078 − 0.003 0.054 0.146
ROE 19,110 0.066 0.171 − 0.022 0.076 0.205
SOE 19,207 0.359 0.480 0.000 0.000 1.000

Table 2 presents the summary statistics of the main variables used in this paper. The mean value of DGW is about 2.8%, implying
that 2.8% of the total observations are labeled as the corporate greenwashing sample. The mean value of ERD_I and ERD_II are 0.095
and 0.212, respectively. As for other firm-level characteristics, an average firm has the ln(Age) of 2.969, Size of 22.407, Board of 2.124,
Dep of 0.374, Lev of 0.450, ROA of 0.058, ROE of 0.066, and 35.9% of the observations is state-owned.

4. Research design and Main results

4.1. Research design

In our main regression, we need to verify whether ERD predicts future corporate greenwashing. Since the dependent variable is a
dummy variable, we can choose the linear-probability model (LPM), the logit model, or the probit model. Jiménez et al. (2012) argued
that LPM has two advantages over the logit or probit model. On the one hand, LPM could avoid sample selection problems. On the
other hand, for the interaction terms, the main focus of the analysis, the estimated coefficients are directly interpretable, and the
standard errors require no corrections. According to Dagher and Sun (2016), when N → ∞ and T is fixed, the LPM has an important
√̅̅̅̅
advantage over probit and logit models, since the estimates are generally inconsistent in probit and logit, but are N consistent using
LPM. Anderson et al. (2017) also reported that they prefer the linear probability model as nonlinear models with fixed effects suffer
from the incidental parameter problem, which can bias asymptotic standard errors downwards.
Thus, to test our first hypothesis H1, we estimate the probability of being in the greenwashing group using the linear-probability
model (LPM). Following the previous literature (Brown et al., 2007; Spalt, 2013; Choi and Laschever, 2018; Tsang et al., 2022), We
have also used a logit or probit model in robustness test, and obtained very similar qualitative and quantitative results. Our main
reduced-form specification can be written as follows:
DGW i,t+1 = β0 + β1 ERDi,t + γX + μi + vt + εi,t (3)

The DGW is the dummy variable of corporate greenwashing, ERD includes the ERD_I and ERD_II defined in Table 1. The control
variables vector X contains ln(Age), Size, Board, Dep, Lev, ROA, ROE and SOE introduced in Section 3.2.3. μi and vt represent the firm
fixed effects and year fixed effects, respectively. Standard errors are robust to heteroscedasticity and clustered at the firm level.
Further, we should test the mechanisms proposed in H2 and H3. Also employing the panel dataset, we estimate the following two
models:
Opacityi,t+1 = β0 + β1 ERDi,t + γX + μi + vt + εi,t (4)

Agency Costsi,t+1 = β0 + β1 ERDi,t + γX + μi + vt + εi,t (5)

The basic model setup is in line with Eq. (3), including the independent variables, control variables, fixed effects, and standard
errors. When examining H2 and H3, the dependent variables will be proxy variables for the firm’s opacity and agency costs,
respectively. First, we follow Hutton et al. (2009) to construct the proxy for corporate opacity, i.e., the three-year moving sum of the
absolute value of annual discretionary accruals:
∑
2
( )
Opacityi,t+1 = Abs DisAcci,t− j (6)
j=0

DisAcci, t denotes the value of firm i’s annual discretionary accruals at year t. To distinguish between normal and discretionary accruals,
Hutton et al. (2009) employed the modified Jones model (Dechow et al., 1995). Specifically, we estimate the following cross-sectional
regression equation using the firms in each industry for each fiscal year. Then, discretionary annual accruals are then calculated using
the parameter estimates from Eq. (7):

6
X. Hu et al. Pacific-Basin Finance Journal 78 (2023) 101952

TAi,t 1 ΔRevi,t PPEi,t

= α1 + α2 + α3 + εi,t (7)
Asseti,t− 1 Asseti,t− 1 Asseti,t− 1 Asseti,t− 1
( )
TAi,t 1 ΔRevi,t − ΔReci,t PPEi,t
DisAcci,t = − α
̂1 +α
̂2 +α
̂3 (8)
Asseti,t− 1 Asseti,t− 1 Asseti,t− 1 Asseti,t− 1

TAi, t denotes total accruals for firm i during year t, Asseti, t denotes total assets for firm i at the end of year t, ΔRevi, t denotes change in
sales, and PPEi, t denotes fixed assets, ΔReci, t denotes the change in accounts receivable. Hutton et al. (2009) proposed that, the
straightforward idea behind this measure is that firms with consistently large absolute values of discretionary accruals are more likely
to be managing reported earnings, thus revealing less firm-specific information to investors. Therefore, the larger Opacityi, t is, the
higher the opacity of corporate information.
Next, following Singh and Davidson (2003), we also construct the firm’s agency costs measure, i.e., the ratio of annual sales to total
assets. This measure of asset utilization indicates the amount of sales and ultimately cash flow generated for a given level of assets.
When the ratio decreases, it indicates that management is using assets in non-cash flow generating and probably value-destroying
ventures. This is similar to the variables used by Ang et al. (2000) and An et al. (2022).

4.2. Main results

We present the main results in Table 3, and the dependent variables are the variable DGW in all specifications. In columns (1) and
(3), we control for firm and year fixed effects and include no control variables. In columns (2) and (4), we additionally include control
variables. The coefficient estimates of our interest are the ̂
β 1 on the ERD_I or ERD_II. The results show that, consistent with our H1,
environmental rating disagreement from different raters would increase the probability of future corporate greenwashing, across all
specifications. Specifically, a firm with a standard deviation higher average ERD_I and ERD_II than the base group would exhibit a
7.52% and 11.38% higher probability of future corporate greenwashing at the 5% significance level.

Table 3
Main results.
Variable (1) (2) (3) (4)

ERD_I 0.0723** 0.0752**

(0.0343) (0.0349)
ERD_II 0.1104** 0.1138**
(0.0498) (0.0502)
ln(Age) − 0.0175 − 0.0155
(0.0581) (0.0578)
Size − 0.0023 − 0.0029
(0.0073) (0.0073)
Board − 0.0581** − 0.0577**
(0.0290) (0.0289)
Dep − 0.0786 − 0.0790
(0.0763) (0.0764)
Dual − 0.0077 − 0.0076
(0.0074) (0.0074)
Lev 0.0483* 0.0498*
(0.0268) (0.0268)
ROA − 0.0700 − 0.0680
(0.0625) (0.0625)
ROE 0.0364 0.0359
(0.0284) (0.0284)
SOE 0.0070 0.0066
(0.0119) (0.0119)
Constant 0.0197*** 0.2565 0.0032 0.2459
(0.0032) (0.2430) (0.0105) (0.2407)
Firm FEs YES YES YES YES
Year FEs YES YES YES YES
Obs. 10,895 10,800 10,895 10,800
Adj. R2 0.2185 0.2170 0.2188 0.2172

Notes: This table reports the main results, and the dependent variables are the variable DGW in all specifications. In columns (1) and (3), we control
for firm and year fixed effects and include no control variables. In columns (2) and (4), we additionally include control variables. All variables are
defined in Table 1. The standard errors (in parentheses) are clustered at the firm level. *, **, and *** denote significance at the 10%, 5%, and 1%
levels, respectively.

7
X. Hu et al. Pacific-Basin Finance Journal 78 (2023) 101952

4.3. Mechanism analyses

To further examine what drives the impact of higher ERD on corporate greenwashing, we ask whether the mechanisms are firm
opacity or agency costs, i.e., testing H2 and H3. To investigate these hypotheses, we run the regressions in Eq. (4) and Eq. (5), and
Table 4 reports the corresponding results. In columns (1) and (2), the dependent variable is the measure of corporate information
opacity. In columns (3) and (4), we incorporate the proxy for agency costs as the dependent variable. In all columns, we include
controls variables, the year and firm fixed effects as in Table 3.
The first two columns show that the corporate information opacity is higher when the proxies for environmental rating
disagreement are larger. The findings are quite straightforward, i.e., ERD erodes firm transparency significantly, with coefficient
estimates approximately equal to 2.3%, which also supports our hypothesis H2.
Columns (3) and (4) report the results of Eq. (5). When the proxy for agency costs is used as the dependent variable, the coefficients
of ERD are positive and significant at a level of at least 5%. The coefficient estimates on ERD_I and ERD_II are − 10.41% and − 17.51%
with control variables, implying that the ERD decreases the utilization ratio of total assets. Firms with lower asset utilization ratios are
making non-optimal investment decisions or using funds to purchase unproductive (non-revenue-generating) assets, and creating
agency costs for shareholders (Henry, 2010). To summarize, the results in the two columns provide evidence that supports the view
that an increase in environmental rating disagreement leads to managers’ agency costs increase, confirming our hypothesis H3.

Table 4
Mechanism analysis.
Var. Opacity Agency Costs

(1) (2) (3) (4)

ERD_I 0.0239** − 0.1041**

(0.0120) (0.0459)
ERD_II 0.0234* − 0.1751***
(0.0141) (0.0596)
ln(Age) 0.0813** 0.0822*** 0.1172 0.1152
(0.0316) (0.0317) (0.1157) (0.1156)
Size 0.0072 0.0072 − 0.1318*** − 0.1308***
(0.0048) (0.0048) (0.0141) (0.0141)
Board 0.0052 0.0054 − 0.0665* − 0.0670*
(0.0124) (0.0124) (0.0348) (0.0349)
Dep − 0.0139 − 0.0132 0.0001 0.0015
(0.0276) (0.0275) (0.0949) (0.0950)
Dual − 0.0013 − 0.0013 − 0.0047 − 0.0048
(0.0041) (0.0041) (0.0095) (0.0095)
Lev 0.0538* 0.0540* 0.1684*** 0.1661***
(0.0313) (0.0314) (0.0454) (0.0455)
ROA 0.2630 0.2635 0.2717*** 0.2683***
(0.2034) (0.2036) (0.0805) (0.0804)
ROE − 0.0789 − 0.0792 0.0297 0.0306
(0.0492) (0.0493) (0.0328) (0.0328)
SOE − 0.0120** − 0.0120** 0.0090 0.0095
(0.0059) (0.0059) (0.0259) (0.0257)
Constant − 0.3799** − 0.3856** 3.3516*** 3.3634***
(0.1570) (0.1572) (0.4698) (0.4691)
Firm FEs YES YES YES YES
Year FEs YES YES YES YES
Obs. 8029 8029 10,800 10,800
Adj. R2 0.4170 0.4170 0.8788 0.8789

Notes: This table reports the results of mechanism tests. Columns (1) and (2) incorporate the firm’s opacity proxy as the dependent variable. Columns
(3) and (4) incorporate the proxy for agency costs as the dependent variable, which is the utilization ratio of total assets, equal to the ratio of annual
sales to total assets. A lower utilization ratio of total assets implies a higher level of agency costs. In all columns, we control for firm and year fixed
effects and include all control variables. All variables are defined in Table 1. The standard errors (in parentheses) are clustered at the firm level. *, **,
and *** denote significance at the 10%, 5%, and 1% levels, respectively.

8
X. Hu et al. Pacific-Basin Finance Journal 78 (2023) 101952

Table 5
Robustness tests – altering the dependent variable.
(1) (2) (3) (4)

Panel A: Altering the calculation basis of the dependent variable DGW

ERD_I 0.0628** 0.0654**

(0.0320) (0.0327)
ERD_II 0.1003** 0.1033**
(0.0451) (0.0457)
Obs. 10,895 10,800 10,895 10,800
Adj R2 0.2267 0.2254 0.2269 0.2256

Panel B: Introducing the greenwashing measurement following Zhang (2022a)

GW GW

ERD_I 2.4149*** 2.4187***

(0.7802) (0.7930)
ERD_II 2.6580** 2.7229**
(1.2665) (1.2859)
Obs. 2006 1980 2006 1980
Adj R2 0.7265 0.7286 0.7225 0.7247

Controls NO YES NO YES

Constant YES YES YES YES
Firm FEs YES YES YES YES
Year FEs YES YES YES YES

Notes: This table reports the results of robustness tests. Panel A reconstructs the DGW with greenness- or environmental-related sentence frequency
instead of word frequency. Panel B incorporates the variable GW following Zhang (2022a). In columns (1) and (3), we control for firm and year fixed
effects and include no control variables. In columns (2) and (4), we additionally include control variables. All variables are defined in Table 1. The
standard errors (in parentheses) are clustered at the firm level. *, **, and *** denote significance at the 10%, 5%, and 1% levels, respectively.

Table 6
Robustness tests – altering the model setting.
Panel A: Incorporating environmental disclosure score and average rating score

DGW DGW

ERD_I 0.0723** 0.2216***

(0.0343) (0.0766)
ERD_II 0.1104** 0.2145***
(0.0498) (0.0694)
Firm FEs YES YES YES YES
Obs. 10,895 5981 10,895 5981
Adj R2 0.2185 0.2593 0.2188 0.2597

Panel B: Altering the regression model

Logit (DGW) Probit (DGW)

ERD_I 0.0884*** 0.0920***

(0.0327) (0.0283)
ERD_II 0.0716*** 0.0743***
(0.0324) (0.0299)
Firm FEs NO NO NO NO
Obs. 10,874 10,874 10,874 10,874
Pseudo R2 0.0721 0.0673 0.0720 0.0670

Panel C: Incorporating industry-year fixed effects

DGW GW

ERD_I 0.0866** 2.6594***

(0.0360) (0.7809)
ERD_II 0.1040** 3.1213**
(0.0520) (1.2440)
Firm FEs YES YES YES YES
Obs. 10,800 10,800 1980 1980
Adj R2 0.2150 0.2150 0.7722 0.7676
Controls YES YES YES YES
Constant YES YES YES YES
Year FEs YES YES YES YES

Notes: This table reports the results of robustness tests. Panel A further incorporates the environmental disclosure score and average rating score as
control variables following Christensen et al. (2022). Panel B incorporates the logit and probit models to repeat the main regression. Panel C
additionally introduces industry-year fixed effects. All variables are defined in Table 1. The standard errors (in parentheses) are clustered at the firm
level. *, **, and *** denote significance at the 10%, 5%, and 1% levels, respectively.

9
X. Hu et al. Pacific-Basin Finance Journal 78 (2023) 101952

Table 7
Further analysis – heavily-polluted industry.
(1) (2) (3) (4)

Panel A: The first classification of heavily-polluted industry

Polluted Non-Polluted Polluted Non-Polluted

ERD_I 0.2210** 0.0378
(0.1001) (0.0338)
ERD_II 0.2243* 0.0698
(0.1288) (0.0468)
Obs. 3286 7078 3286 7078
Adj R2 0.2589 0.1484 0.2585 0.1487

Panel B: The second classification of heavily-polluted industry

Polluted Non-Polluted Polluted Non-Polluted

ERD_I 0.2039** 0.0370
(0.0892) (0.0342)
ERD_II 0.2228* 0.0615
(0.1191) (0.0456)
Obs. 3629 6735 3629 6735
Adj R2 0.2461 0.1682 0.2459 0.1684

Controls YES YES YES YES

Constant YES YES YES YES
Firm FEs YES YES YES YES
Year FEs YES YES YES YES

Notes: This table reports the results of subsample regression divided by the heavily-polluted firms. Panel A and Panel B incorporate two mainstream
standards to split the sample. In all columns, we control for firm and year fixed effects and include control variables. All variables are defined in
Table 1. The standard errors (in parentheses) are clustered at the firm level. *, **, and *** denote significance at the 10%, 5%, and 1% levels,
respectively.

4.4. Robustness tests

In this section, we perform a series of robustness tests to guarantee that our main results hold with alternative settings.
In Table 3, we rely on the frequency of greenness- or environmental-related words to construct the proxy for corporate green­
washing. The related words can, however, be concentrated in several sentences rather than scattered in many parts of the text. As a
result, the Oral indicator might be a noisy proxy for the firm’s oral green publicity. To address this concern, we introduce the related-
sentences frequency to construct the Oral indicator and, consequently, the GW variable. In Panel A of Table 5, we repeat our analyses in
Table 3 using this new GW variable. The results continue to show ERD’s positive and statistically significant impact on corporate
greenwashing.
As stated in Section 3.2.1, we also incorporate the corporate greenwashing developed by Zhang (2022a), and present the results in
Panel B of Table 5. The coefficient estimates on ERD_I and ERD_II are still positive at the 1% and 5% significance level, respectively,
implying that the main results will not be overturned by the measurement of GW.
Next, Christensen et al. (2022) reported that greater ESG disclosure actually leads to greater ESG rating disagreement. Thus, to
alleviate the concern that environmental disclosure instead of the ERD drives the current results, we incorporate the variable E_dis­
closurei,t and E_avgi,t as the additional control variables. Following Christensen et al. (2022), E_disclosurei,t indicates the i-th firm’s
environmental disclosure score for the ESG report pertaining to year t’s performance, while E_avgi,t implies the average environmental
rating the firm i received for year t’s environmental performance.4 After incorporating the two variables, as shown in Panel A of
Table 6, the coefficient estimates are still significantly positive. Therefore, the main results are not influenced by the greater envi­
ronmental disclosure.
Since our dependent variable GW is an indicator variable, we further utilize the logit and probit model to conduct the empirical
analysis. We report the results in Panel B of Table 6. Following Anderson et al. (2018) and Agrawal et al. (2022), we do not include firm
or industry fixed effects in the logit or probit model to avoid the incidental parameters problem that arises in the context of nonlinear
panel models. The regression estimates capture the marginal effects of ERD on the probability of future corporate greenwashing, which
is still significantly positive at the 1% level. This suggests that the model chosen is unlikely to be driving our results.
In Panel C of Table 6, we further control for unobservable factors that change over time at the industry level, that is, incorporating
industry-year fixed effects (Bernstein et al., 2017). We continue to find a positive and statistically significant impact of the ERD on
corporate greenwashing, which also indicates the robustness of our main results.

4
Christensen et al. (2022) utilized ESG disclosure scores from Bloomberg to proxy for the extent of firms’ ESG disclosure practices. Thus, we also
utilize the environmental disclosure scores from Bloomberg to proxy for the extent of firms’ environmental disclosure practices.

10
X. Hu et al. Pacific-Basin Finance Journal 78 (2023) 101952

Table 8
Further analysis – internal incentive.
(1) (2) (3) (4)

Panel A: The salary of top three executives

High Low High Low

ERD_I 0.1003** 0.0362
(0.0410) (0.0691)
ERD_II 0.1018* 0.1410
(0.0593) (0.0972)
Obs. 5593 4659 5593 4659
Adj R2 0.2600 0.1709 0.2597 0.1719

Panel B: The level of financial constraints

Constrained Not-Constrained Constrained Not-Constrained

ERD_I 0.0368 0.1229**
(0.0472) (0.0552)
ERD_II 0.0060 0.1965***
(0.0839) (0.0710)
Obs. 5636 4677 5636 4677
Adj R2 0.2101 0.2605 0.2100 0.2615

Controls YES YES YES YES

Constant YES YES YES YES
Firm FEs YES YES YES YES
Year FEs YES YES YES YES

Notes: This table reports the results of subsample regression divided by the internal incentives. Panel A presents the impact of the salary of the top
three executives. Panel B shows the impact of the financial constraint level. In all columns, we control for firm and year fixed effects and include
control variables. All variables are defined in Table 1. The standard errors (in parentheses) are clustered at the firm level. *, **, and *** denote
significance at the 10%, 5%, and 1% levels, respectively.

5. Further research

5.1. Industry heterogeneity

In Table 7, we examine whether the impact of ERD on corporate greenwashing varies between heavily-polluted and lowly-polluted
firms. There are many ways to divide heavy pollution industries, and this paper selects two mainstream standards. The first is based on
the Classified Management Directory of Environmental Protection Verification Industry of Listed Companies,5 and the second is based on the
Guidelines for Environmental Information Disclosure of Listed Companies.6 The two methods are roughly the same, but also slightly
different. Against this backdrop, Panel A and B report the two sets of results based on the two different standards, which have no
notable impact on the results. In columns (1) and (3), we present the regression results of heavily-polluted subsamples, while we show
the results of lowly-polluted ones in columns (2) and (4).
Consistent with our hypothesis H4, whether from the perspective of significance or coefficient estimates magnitude, the impact of
ERD on corporate greenwashing is greater and significant in the subsamples of heavily-polluted enterprises. It indicates that they have
more motives to whitewash their public image of greenness.

5.2. The effect of internal incentives

To verify hypothesis H5, we incorporate two proxy variables. One is the salary of the top three executives, and the other is the WW
index constructed by Whited and Wu (2006). The data of the WW index are obtained from the CSMAR database.7 According to the
median of these two variables, this paper divides the full sample into two subsamples. In detail, if the salary of the top three executives
is higher than the median in the current year, the sample will be included in the “High” salary subsamples. Otherwise, it will be
included in the “Low” salary subsamples. Similarly, since a higher WW index indicates a higher level of financial constraints, the firm-
year observation with the WW index higher than the median will be included in the “Constrained” subsamples. Otherwise, it will be
included in the “Not-Constrained “subsamples.
The evidence presented in Panel A of Table 8 supports our hypothesis that the higher the executives’ salary, the higher the incentive
they will have to engage in greenwashing, and thus look forward to maintaining the company’s rating performance and their

5
The policy document is from http://www.gov.cn/gzdt/2008-07/07/content_1038083.htm.
6
The policy document is from https://www.chinaacc.com/upload/html/2010/9/15/sunxia520820109159215316576.pdf.
7
From the description of the CSMAR database, the WW index is constructed following Whited and Wu (2006) as − 0.091 × CF-0.062 × DivPos +
0.021 × Lev-0.044 × Size+0.102 × ISG-0.035 × SG, where CF denotes the ratio of cash flow to total assets, DivPos denotes the dummy variable
indicating the cash dividend, Lev denotes the ratio of long-term liabilities to assets, ISG is the average sales growth rate of the industry, SG denotes
sales revenue growth rate.

11
X. Hu et al. Pacific-Basin Finance Journal 78 (2023) 101952

Table 9
Further analysis – internal supervision.
(1) (2) (3) (4)

Panel A: The shareholding ratio of foreign investors

High Low High Low

ERD_I 0.0596 0.1081**
(0.0577) (0.0521)
ERD_II 0.1124 0.1513**
(0.0857) (0.0711)
Obs. 4142 6099 4142 6099
Adj R2 0.2176 0.2626 0.2178 0.2630

Panel B: The ratio of independent directors on the board

High Low High Low

ERD_I 0.0587 0.0692
(0.0499) (0.0579)
ERD_II 0.0419 0.1556*
(0.0666) (0.0817)
Obs. 4642 5484 4642 5484
Adj R2 0.2840 0.2063 0.2838 0.2071

Controls YES YES YES YES

Constant YES YES YES YES
Firm FEs YES YES YES YES
Year FEs YES YES YES YES

Notes: This table reports the results of subsample regression divided by the internal supervision. Panel A presents the impact of the shareholding ratio
of foreign investors. Panel B shows the impact of the proportion of independent directors. In all columns, we control for firm and year fixed effects and
include control variables. All variables are defined in Table 1. The standard errors (in parentheses) are clustered at the firm level. *, **, and *** denote
significance at the 10%, 5%, and 1% levels, respectively.

Table 10
Further analysis– external attention.
(1) (2) (3) (4)

Panel A: Number of stock posts

High Low High Low

ERD_I 0.0383 0.1537**
(0.0532) (0.0602)
ERD_II 0.0824 0.1755**
(0.0763) (0.0816)
Obs. 5129 4655 5129 4655
Adj R2 0.2458 0.1690 0.2461 0.1685

Panel B: Number of posts reading

High Low High Low

ERD_I 0.0454 0.1387**
(0.0504) (0.0612)
ERD_II 0.0701 0.1895**
(0.0725) (0.0878)
Obs. 5194 4650 5194 4650
Adj R2 0.2479 0.1700 0.2480 0.1703

Panel C: Investor sentiment status

Negative Positive Negative Positive

ERD_I − 0.0696 0.1469***
(0.0602) (0.0488)
ERD_II − 0.1160 0.1945***
(0.0989) (0.0581)
Obs. 4654 5107 4654 5107
Adj R2 0.2137 0.2431 0.2140 0.2436

Controls YES YES YES YES

Constant YES YES YES YES
Firm FEs YES YES YES YES
Year FEs YES YES YES YES

Notes: This table reports the results of subsample regression divided by the external attention. Panels A, B, and C show the impact of stock posts, stock
posts reading and investor sentiment. In all columns, we control for firm and year fixed effects and include control variables. All variables are defined
in Table 1. The standard errors (in parentheses) are clustered at the firm level. *, **, and *** denote significance at the 10%, 5%, and 1% levels,
respectively.

12
X. Hu et al. Pacific-Basin Finance Journal 78 (2023) 101952

reputation. In terms of results, this is directly reflected in the positive impact of ERD on corporate greenwashing at the level of at least
10% among the subsamples with higher executives’ salaries. In contrast, there is no significant effect in the pair subsamples.
After reviewing the influence of financial constraints shown in Panel B, the findings are in accordance with our anticipation that the
internal finance source would restrain corporate greenwashing behavior. Thus, comparing the results between the two subsamples, the
coefficient estimates on ERD are smaller and insignificant within the financially-constrained firms. Taken together, these results have
supported our hypothesis H5 concerning the firm’s internal incentives. That is, the positive relationship between ERD and corporate
greenwashing will be weaker in the financially-constrained firms, but stronger with higher executive salaries.

5.3. The effect of internal supervision

In this section, we incorporate the shareholding ratio of foreign investors and the proportion of independent directors on the board
to measure internal supervision intensity. As the above treatment, we divide the sample into two subsamples according to the median
of the two variables, respectively.
As presented in Table 9, we can easily find that, the positive relationship between ERD and corporate greenwashing will be weaker
in the firms with higher internal supervision, i.e., the higher shareholding ratio of foreign investors and a higher ratio of independent
directors. To some extent, the magnitude and the significance level of the key independent variables support our hypotheses H6.
Specifically, when the sample is split according to the median of two variables, the effect is more evident in the subsamples with
weaker internal supervision.

5.4. The effect of external attention

To measure the external attention, we choose the posts from Eastmoney Stock Post Bar as the research object. We collect the
number of posts, readings, and sentiment index related to each listed firm each year. The sentiment index is equal to the number of
negative words divided by the total number of positive and negative words. The data comes from the CSAMR database.
Results in Panel A and Panel B of Table 10 represent the effect of investor attention, in which we also try to divide the entire sample
into two subsamples, i.e., “High” and “Low”, according to the median of the number of posts or the number of readings, respectively.
The results show that higher external attention would suppress the motives of corporate greenwashing, since the coefficient estimates
of ERD on corporate greenwashing are significantly positive in the “Low” subsamples.
We further use the median of the calculated sentiment index to divide the “Positive” subsamples and the “Negative” subsamples.
Results in Panel C support that higher investor sentiment has been linked with opportunistic managerial behavior, inducing more
corporate greenwashing. To sum up, these findings can support our hypothesis H7. The positive relationship between ERD and
corporate greenwashing will be weaker with more external investor attention and negative investor sentiment.

6. Conclusions

In this paper, we focus on a relatively covert relation between environmental rating disagreement and corporate greenwashing.
Environmental rating disagreement is becoming increasingly evident, confusing investors and the market like a fog. Under such a
circumstance, our paper highlights that firms and their managers are motivated to wrap the corporate greenwashing in a disguise of
rating divergence.
We focus on empirically investigating whether and how ERD impacts corporate greenwashing. Our findings can be summarized as
follows. First, we find that environmental rating disagreement increases the future probability of corporate greenwashing, and such
impact could be explained by the perspective of agency costs and firm opacity. Second, we find that several internal and external
characteristics would change the effect of ERD on corporate greenwashing. On the one hand, internal incentives would restrain or
stimulate corporate greenwashing, while internal supervision would decrease the probability of corporate greenwashing facing the
same ERD level. On the other hand, external investor attention would suppress the motives of corporate greenwashing, but investors’
positive sentiment would have the opposite influence. Our main results remain robust across a battery of robustness tests.
Our findings suggest that environmental rating disagreement would not only confuse investors, but also send greenwashing signals
to firms or managers, since they will rarely be punished by the raters even if a firm does so. Prior empirical studies mainly examine the
impact of rating disagreement on the capital market, but less attention has been devoted to the negative influence on the firm’s
behavior. We provide a novel perspective that can contribute to curbing the greenwashing phenomenon, for instance, enhancing
internal supervision and external investor attention.

Acknowledges

This work was supported by the National Key R&D Program of China (2021YFC3340703) and the National Natural Science
Foundation of China (71973063, 71991471, 71871066 & 72121002).

13
X. Hu et al. Pacific-Basin Finance Journal 78 (2023) 101952

References

Abramova, I., Core, J.E., Sutherland, A., 2020. Institutional investor attention and firm disclosure. Account. Rev. 95 (6), 1–21.
Agrawal, A., Gonzalez-Uribe, J., Martínez-Correa, J., 2022. Measuring the ex-ante incentive effects of creditor control rights during bankruptcy reorganization.
J. Financ. Econ. 143 (1), 381–408.
Amin, K., Eshleman, J.D., Guo, P., 2021. Investor sentiment, misstatements, and auditor behavior. Contemp. Account. Res. 38 (1), 483–517.
An, Y., Jin, H., Liu, Q., Zheng, K., 2022. Media attention and agency costs: evidence from listed companies in China. J. Int. Money Financ. 124, 102609.
Anderson, R.W., Johnson, N.D., Koyama, M., 2017. Jewish persecutions and weather shocks: 1100–1800. Econ. J. 127 (602), 924–958.
Anderson, R.W., Bustamante, M.C., Guibaud, S., Zervos, M., 2018. Agency, firm growth, and managerial turnover. J. Financ. 73 (1), 419–464.
Ang, J.S., Cole, R.A., Lin, J.W., 2000. Agency costs and ownership structure. J. Financ. 55 (1), 81–106.
Avramov, D., Cheng, S., Lioui, A., Tarelli, A., 2022. Sustainable investing with ESG rating uncertainty. J. Financ. Econ. 145 (2), 642–664.
Bajo, E., Chemmanur, T.J., Simonyan, K., Tehranian, H., 2016. Underwriter networks, investor attention, and initial public offerings. J. Financ. Econ. 122 (2),
376–408.
Baker, T., Nelson, R.E., 2005. Creating something from nothing: resource construction through entrepreneurial bricolage. Adm. Sci. Q. 50 (3), 329–366.
Bannier, C.E., Behr, P., Güttler, A., 2010. Rating opaque borrowers: why are unsolicited ratings lower? Rev. Fin. 14 (2), 263–294.
Berg, F., Koelbel, J.F., Rigobon, R., 2022. Aggregate confusion: the divergence of ESG ratings. Rev. Fin. 26 (6), 1315–1344.
Bergman, N.K., Roychowdhury, S., 2008. Investor sentiment and corporate disclosure. J. Account. Res. 46 (5), 1057–1083.
Bernstein, S., Lerner, J., Sorensen, M., Str_mberg, P., 2017. Private equity and industry performance. Manag. Sci. 63 (4), 1198–1213.
Bhattacharya, U., Daouk, H., Welker, M., 2003. The world price of earnings opacity. Account. Rev. 78 (3), 641–678.
Brown, J.R., Liang, N., Weisbenner, S., 2007. Executive financial incentives and payout policy: firm responses to the 2003 dividend tax cut. J. Financ. 62 (4),
1935–1965.
Brown, N.C., Christensen, T.E., Elliott, W.B., Mergenthaler, R.D., 2012. Investor sentiment and pro forma earnings disclosures. J. Account. Res. 50 (1), 1–40.
Byrd, J.W., Hickman, K.A., 1992. Do outside directors monitor managers?: evidence from tender offer bids. J. Financ. Econ. 32 (2), 195–221.
Chang, K., Cheng, X., Wang, Y., Liu, Q., Hu, J., 2021. The impacts of ESG performance and digital finance on corporate financing efficiency in China. Appl. Econ. Lett.
1–8.
Chatterji, A.K., Durand, R., Levine, D.I., Touboul, S., 2016. Do ratings of firms converge? Implications for managers, investors and strategy researchers. Strateg.
Manag. J. 37 (8), 1597–1614.
Chen, Y., Li, T., Zeng, Q., Zhu, B., 2023. Effect of ESG performance on the cost of equity capital: evidence from China. Int. Rev. Econ. Financ. 83, 348–364.
Choi, H.S., Laschever, R.A., 2018. The credit card debt puzzle and noncognitive ability. Rev. Fin. 22 (6), 2109–2137.
Christensen, D.M., Serafeim, G., Sikochi, A., 2022. Why is corporate virtue in the eye of the beholder? The case of ESG ratings. Account. Rev. 97 (1), 147–175.
Dagher, J., Sun, Y., 2016. Borrower protection and the supply of credit: evidence from foreclosure laws. J. Financ. Econ. 121 (1), 195–209.
Dechow, P.M., Sloan, R.G., Sweeney, A.P., 1995. Detecting earnings management. Account. Rev. 193–225.
Delmas, M.A., Burbano, V.C., 2011. The drivers of greenwashing. Calif. Manag. Rev. 54 (1), 64–87.
Desender, K.A., Aguilera, R.V., Lópezpuertas_Lamy, M., & Crespi, R., 2016. A clash of governance logics: foreign ownership and board monitoring. Strateg. Manag. J.
37 (2), 349–369.
Dimson, E., Marsh, P., Staunton, M., 2020. Divergent ESG ratings. J. Portfolio Manag. 47 (1), 75–87.
Douma, S., George, R., Kabir, R., 2006. Foreign and domestic ownership, business groups, and firm performance: evidence from a large emerging market. Strateg.
Manag. J. 27 (7), 637–657.
Drucker, S., Puri, M., 2009. On loan sales, loan contracting, and lending relationships. Rev. Financ. Stud. 22 (7), 2835–2872.
Du, X., 2015. How the market values greenwashing? Evidence from China. J. Bus. Ethics 128 (3), 547–574.
Edelen, R.M., Evans, R.B., Kadlec, G.B., 2012. Disclosure and agency conflict: evidence from mutual fund commission bundling. J. Financ. Econ. 103 (2), 308–326.
Flammer, C., 2021. Corporate green bonds. J. Financ. Econ. 142 (2), 499–516.
Fulghieri, P., Strobl, G., Xia, H., 2014. The economics of solicited and unsolicited credit ratings. Rev. Financ. Stud. 27 (2), 484–518.
George, G., 2005. Slack resources and the performance of privately held firms. Acad. Manag. J. 48 (4), 661–676.
Gibson Brandon, R., Krueger, P., Schmidt, P.S., 2021. ESG rating disagreement and stock returns. Financ. Anal. J. 77 (4), 104–127.
Güntay, L., Hackbarth, D., 2010. Corporate bond credit spreads and forecast dispersion. J. Bank. Financ. 34 (10), 2328–2345.
Guo, R., Tao, L., Li, C.B., Wang, T., 2017. A path analysis of greenwashing in a trust crisis among Chinese energy companies: the role of brand legitimacy and brand
loyalty. J. Bus. Ethics 140 (3), 523–536.
Hauck, A., Neyer, U., 2014. Disagreement between rating agencies and bond opacity: a theoretical perspective. Econ. Lett. 123 (1), 82–85.
He, F., Du, H., Yu, B., 2022a. Corporate ESG performance and manager misconduct: evidence from China. Int. Rev. Financ. Anal. 82, 102201.
He, F., Yan, Y., Hao, J., Wu, J., 2022b. Retail investor attention and corporate green innovation: evidence from China. Energy Econ. 106308.
Healy, P.M., Palepu, K.G., 2001. Information asymmetry, corporate disclosure, and the capital markets: a review of the empirical disclosure literature. J. Account.
Econ. 31 (1–3), 405–440.
Henry, D., 2010. Agency costs, ownership structure and corporate governance compliance: a private contracting perspective. Pac. Basin Financ. J. 18 (1), 24–46.
Hoberg, G., Lewis, C., 2017. Do fraudulent firms produce abnormal disclosure? J. Corp. Finan. 43, 58–85.
Hutton, A.P., Marcus, A.J., Tehranian, H., 2009. Opaque financial reports, R2, and crash risk. J. Financ. Econ. 94 (1), 67–86.
Jiménez, G., Ongena, S., Peydró, J.L., Saurina, J., 2012. Credit supply and monetary policy: identifying the bank balance-sheet channel with loan applications. Am.
Econ. Rev. 102 (5), 2301–2326.
Jin, L., Myers, S.C., 2006. R2 around the world: new theory and new tests. J. Financ. Econ. 79 (2), 257–292.
Johnson, S.A., Ryan, H.E., Tian, Y.S., 2009. Managerial incentives and corporate fraud: the sources of incentives matter. Rev. Fin. 13 (1), 115–145.
Kim, E.H., Lyon, T.P., 2015. Greenwash vs. brownwash: exaggeration and undue modesty in corporate sustainability disclosure. Organ. Sci. 26 (3), 705–723.
Kothari, S.P., Shu, S., Wysocki, P.D., 2009. Do managers withhold bad news? J. Account. Res. 47 (1), 241–276.
Li, C., Wu, M., Chen, X., Huang, W., 2022. Environmental, social and governance performance, corporate transparency, and credit rating: some evidence from Chinese
A-share listed companies. Pac. Basin Financ. J. 74, 101806.
Liang, H., Sun, L., TEO, S. W. M., 2022. Responsible hedge funds. Rev. Fin. 26 (6), 1585–1633.
Loughran, T., McDonald, B., Yun, H., 2009. A wolf in sheep’s clothing: the use of ethics-related terms in 10-K reports. J. Bus. Ethics 89 (1), 39–49.
Masulis, R.W., Mobbs, S., 2014. Independent director incentives: where do talented directors spend their limited time and energy? J. Financ. Econ. 111 (2), 406–429.
Morgan, D.P., 2002. Rating banks: risk and uncertainty in an opaque industry. Am. Econ. Rev. 92 (4), 874–888.
Nyilasy, G., Gangadharbatla, H., Paladino, A., 2014. Perceived greenwashing: the interactive effects of green advertising and corporate environmental performance on
consumer reactions. J. Bus. Ethics 125 (4), 693–707.
Parguel, B., Benot-Moreau, F., Larceneux, F., 2011. How sustainability ratings might deter ‘greenwashing’: a closer look at ethical corporate communication. J. Bus.
Ethics 102 (1), 15–28.
Serafeim, G., Yoon, A., 2022. Stock price reactions to ESG news: the role of ESG ratings and disagreement. Rev. Acc. Stud. 1–31.
Singh, M., Davidson III, W.N., 2003. Agency costs, ownership structure and corporate governance mechanisms. J. Bank. Financ. 27 (5), 793–816.
Spalt, O.G., 2013. Probability weighting and employee stock options. J. Financ. Quant. Anal. 48 (4), 1085–1118.
Sun, Z., Zhang, W., 2019. Do government regulations prevent greenwashing? An evolutionary game analysis of heterogeneous enterprises. J. Clean. Prod. 231,
1489–1502.
Szabo, S., Webster, J., 2021. Perceived greenwashing: the effects of green marketing on environmental and product perceptions. J. Bus. Ethics 171 (4), 719–739.
Tsang, A., Yang, N., Zheng, L., 2022. Cross-listings, antitakeover defenses, and the insulation hypothesis. J. Financ. Econ. 145 (1), 259–276.

14
X. Hu et al. Pacific-Basin Finance Journal 78 (2023) 101952

Wei, Z., Xie, F., Zhang, S., 2005. Ownership structure and firm value in China’s privatized firms: 1991-2001. J. Financ. Quant. Anal. 40 (1), 87–108.
Whited, T.M., Wu, G., 2006. Financial constraints risk. Rev. Financ. Stud. 19 (2), 531–559.
Wu, Y., Zhang, K., Xie, J., 2020. Bad greenwashing, good greenwashing: corporate social responsibility and information transparency. Manag. Sci. 66 (7), 3095–3112.
Xia, H., 2014. Can investor-paid credit rating agencies improve the information quality of issuer-paid rating agencies? J. Financ. Econ. 111 (2), 450–468.
Yu, E.P.Y., Van Luu, B., Chen, C.H., 2020. Greenwashing in environmental, social and governance disclosures. Res. Int. Bus. Financ. 52, 101192.
Zhang, D., 2022a. Green financial system regulation shock and greenwashing behaviors: evidence from Chinese firms. Energy Econ. 106064.
Zhang, D., 2022b. Environmental regulation and firm product quality improvement: how does the greenwashing response? Int. Rev. Financ. Anal. 80, 102058.
Zhou, G., Liu, L., Luo, S., 2022. Sustainable development, ESG performance and company market value: mediating effect of financial performance. Bus. Strateg.
Environ. 31 (7), 3371–3387.

15