The effect of environmental credit rating on audit fees: A quasi-natural experiment from China

Abstract

Environmental credit rating (ECR) is a new policy that deeply integrates the construction of ecological civilisation and the social credit system in China; however, there is a paucity of research on the response of external auditors to the ECR. This study takes the environmental credit evaluation policy implementation as a quasi-natural experiment, using China's A-share listed companies in heavily polluting industries from 2008 to 2019 as samples. We construct a heterogeneous timing difference-in-differences model to empirically explore the impact of ECR on audit fees. The results show that the ECR significantly reduces companies' audit fees. Importantly, our analysis suggests that the ECR improves environmental information transparency and enhances sustainable operation ability, thereby reducing audit fees. Further analysis shows that the negative correlation between the ECR and audit fees is more obvious in non-state enterprises, in poor legal environments and low levels of trust. Our study provides scientific evidence for the economic consequences of the environmental credit evaluation policy and enriches the literature on the factors affecting audit fees. It has revelatory significance for China and other developing countries to implement and improve the environmental credit evaluation policies and better guide enterprises to fulfil their environmental responsibilities.

1. Introduction

Human beings' long-term unreasonable development and utilisation activities have increasingly tightened the constraints of the ecological environment [1], which has become a significant practical problem worldwide that must be solved urgently [2]. China has made remarkable economic achievements but faces increasing environmental and ecological destruction issues [[3], [4], [5]]. Fortunately, since the 18th CPC National Congress, the Chinese government has formulated a series of policies to continuously strengthen environmental regulation, such as green credit policy [[6], [7], [8]], emission trading system [[9], [10], [11]], accountability audit of natural resource [12,13], China's new environmental protection law (EPL) [14,15], and green tax reforms [16,17], which played an essential role in promoting the green transformation. However, the progress of the construction of ecological civilisation further exposes the environmental risks of enterprises. In 2006, China issued a series of audit standards. Among them, the Auditing Standards for Chinese Certified Public Accountants No. 1631: Consideration of environmental matters in the audit of financial reports suggested that the auditor should focus on the environmental matters of clients. It aims to reduce the risk of audit failure caused by environmental risk; however, only a small amount of literature has researched the reaction of external auditors to environmental events or risks [[18], [19], [20], [21], [22], [23], [24]], and some conclusions have not yet reached a consensus. Therefore, mitigating the environmental risks to reduce audit fees is worth in-depth exploration.

At the end of 2013, the former Ministry of Environmental Protection of China and three other departments of the State Council of China jointly issued the Measures for Enterprise Environmental Credit Evaluation (Trial) (hereafter, evaluation measures), which indicated that the Chinese government would implement the environmental credit evaluation policy. The environmental credit rating (ECR) is primarily used to evaluate and publicise the credit rating of enterprises' environmental behaviours to provide a decision-making reference for environmental stakeholders [25,26]. ECR is a new measure that deeply integrates the construction of ecological civilisation and social credit system and is suitable for establishing and improving the long-term mechanism of ‘trustworthy incentive and dishonesty punishment’. Therefore, in theory, this policy can induce the compliance constraint effect and the incentive and guidance effect for enterprises to fulfil their environmental responsibilities, which is helpful to auditors to reduce the perception of clients' environmental risks; however, to our knowledge, there is a paucity of research on the response of external auditors to the ECR. Therefore, we attempt to answer whether and how the ECR affects audit fees, filling this important research gap. Specifically, we use China's listed enterprises in heavily polluting industries from 2008 to 2019 as a sample, taking the implementation of the environmental credit evaluation policy as a quasi-natural experiment and constructing a heterogeneous timing difference-in-differences (DID) model. This study examines the impact of ECR on audit fees and its mediating mechanism, analysing whether the influence relationship will vary due to the heterogeneous property rights, legal environment and regional trust.

The potential contributions of this study are outlined as follows. First, this study directly explores the relationship between ECR and audit fees and uses a heterogeneous timing DID model to alleviate endogeneity. This approach enriches the literature on the impact factors of audit fees and expands the research scope of ECR. Second, this study provides the first empirical evidence that the ECR can reduce audit fees and its potential mechanism in China, which confirms that auditors have paid attention to corporate environmental behaviour. Third, this study discusses the impact of property rights, legal environment and regional trust on the ECR reduction in audit fees, which will offer constructive significance for improving environmental credit evaluation policies and reducing auditors’ audit risks.

The remainder of this paper is organised as follows. Section 2 is the background of ECR and the literature review. Section 3 presents the theoretical basis and research hypothesis. Section 4 is the study design. Section 5 presents the empirical results. Section 6 provides research on the heterogeneity analysis. Section 7 is the discussion, followed by the conclusion, recommendations and limitations in Section 8.

2. Institutional background and literature review

2.1. Institutional background

In 1989, the Norsk Hydro A.S. of Norway released the first environmental report in the world. The Coalition for Environmentally Responsible Economies in the U.S. launched the ‘Berdis principles’, the international investment community that introduced the concept of corporate environmental behaviour. Environmental risk has increasingly become important for agencies in carrying out credit ratings [[27], [28], [29], [30]]. Therefore, in 2003, Citigroup and some other banking organisations jointly launched the ‘equator principles’, which officially started to bring the ECR into the view of governments and financial institutions and gradually became an essential tool for ‘green allocation’ and ecological governance [26]. Although the ECR originated and imitated the credit rating system in the Western financial market, it has gradually developed towards an independent evaluation system. In the relatively mature carbon credit evaluation market, carbon rating agencies such as Standard & Poor's and Moody's further simplify the decisions of financiers, leading to capital support increasingly depending on the corporate carbon credit rating [31]. Large commercial banks and investment institutions are the main forces that promote the development of ECR in Western countries.

In contrast, the development of ECR in China has strong government promotion and can be divided into three stages. The first stage is ‘initial germination’ from 2000 to 2005. In 2000, Zhenjiang City, Jiangsu Province, piloted a World Banks project called the Research on Environmental Behaviour Information Disclosure of Industrial Enterprises, which officially evaluated the environmental behaviour of enterprises in China for the first time. The second stage is ‘partial exploration’ from 2006 to 2013. In 2006, the former State Environmental Protection Administration of China tried to expand enterprises' environmental behaviour assessment; however, the pilot was ineffective. The third stage is ‘orderly progress’ after 2014. The evaluation measures were issued at the end of 2013, and the ECR was officially implemented in 2014. It has attracted increasing attention in China and received positive responses from all over the country. According to the Ministry of Ecology and Environment of China, by the end of 2020, more than 20 provinces will have implemented the ECR system for more than 30,000 enterprises, and resulting in some positive effects.

2.2. Research on the ECR

Not much research on the ECR focus on bank credit and developed Western countries. Li (2011) found that more and more financial institutions had begun to use the ECR as an important reference for making credit decisions [32], which effectively reduced their environmental risk assessment costs [33]. Weber (2012) used Canada as an example to further explore how financial institutions integrated environmental credit risk into credit risk management [34]. Serrano-Cinca et al. (2016) found that although many micro-finance institutions tried to consider loans' social and environmental impact, they lacked support from data and evaluation systems [35]. China's banks also increasingly favoured those environmentally friendly enterprises under the influence of green credit policies [36].

Furthermore, some studies examine the environmental effectiveness of ECR based on empirical data. Yan et al. (2023) identified the effect of ECR on enterprises' environmental choice behaviour, finding that although environmental credit constraints could motivate enterprises to choose more active environmental behaviour and promote green innovation, some enterprises might also adopt evasive strategies due to increasingly production costs and debt [25]. Zuo and Wu (2022) found that the ECR promoted heavily polluting enterprises’ green innovation through reputation mechanism and financing mechanisms [26]. Di et al. (2023) analysed the pollution-reducing effect of the ECR system and found that the environmental credit constraints helped reduce emission intensity, because they could motivate innovation and optimise allocation structure [37].

2.3. Influencing factors of audit fees

The classic audit theory argues that audit fees mainly comprise the cost incurred during audit procedures and the compensation for potential audit risks [38]. Traditional literature focused on the financial risks [39] and examined the impact of client size [40,41], corporate governance [[42], [43], [44]], earning management [[45], [46], [47]], earning performance [48,49] and liability leverage [50,51] on audit fees. However, as the environmental and sustainability issues become increasingly severe, the environmental risks have gained increasing attention from auditors. Some researchers have explored the impact of environmental risks and their management level on audit fees. Yao et al. (2023) found that auditors raise audit fees after clients violate environmental regulations [18]. Xin et al. (2022) held a similar view on the environmental administrative penalties [19]. Tan et al. (2023) explored the effect of carbon emissions and found that a firm with a higher carbon emission level would be charged a higher audit fee due to the potential carbon risk [20]. Liu et al. (2021) found that auditors would charge higher audit fees due to China's new EPL, which further increased their perceived environmental risks [21]. Yao et al. (2020) found that The Measures for the Disclosure of Environmental Information in 2008 led auditors to pay more attention to environmental information during decision-making on audit pricing [22]. Furthermore, Qian et al. (2022) examined the impact of the green credit policy by using a DID model, and the results showed that auditors could perceive the environmental risks imposed by this policy, thereby increasing audit fees [24].

Many studies agree that audit fees are compensation for audit risks, including environmental risks, and they have become increasingly important; however, to our knowledge, there is little direct empirical evidence on whether and how audit fees are affected by the ECR.

3. Theoretical basis and research hypotheses

3.1. Theoretical basis

The environmental governance effect of environmental credit evaluation policy can be summarised as follows. On the one hand, it is the ‘compliance constraint effect’ based on the legitimacy theory [52,53]. Specifically, implementing an environmental credit evaluation policy can aggravate pressure on the environmental legitimacy of enterprises from three aspects. First, the principle of ‘consultation and joint evaluation’ will expand the sources of information acquisition [26] and reduce the cost of information acquisition. Second, the ‘result sharing’ principle will enable the ECR rating result to spread and offer more information support for classified treatment. Third is the principle of ‘lassified regulation’, Suppose an enterprise has a poor ECR rating result. In that case, it will face a series of punishments by the government (including but not limited to checking the applications for environmental administrative licence, increasing the environmental supervision, and suspending the environmental subsidies) and diversified quasi-regulatory pressures, such as stricter credit risk assessment from banks, negative environmental reports from media and more widespread public attention [25,37].

On the other hand, it is the ‘incentive and guidance effect’ based on the resource-based theory and the reputation theory [[54], [55], [56]]. Specifically, implementing an environmental credit evaluation policy strengthens the punishment for environmental dishonesty and attaches importance to the incentive of environmental trustworthiness [26]. The first is the green resource incentives from the government, such as environmental subsidies, green procurement lists, approval of environmental projects and additional emission indicators for the environmental trustworthiness enterprises. The second is the commercial incentives from the market, such as the support for credit granting and a more favourable environmental pollution liability insurance rate for the environmental trustworthiness enterprises. The third is the reputation incentive. As a special credit evaluation system provided by the government, the signal from official recognition will directly affect the social reputation of enterprises.

3.2. Hypotheses

Under the modern risk-oriented audit model, the risk perception of auditors is an important criterion for them to charge audit fees [57,58]. In recent years, environmental risks have gotten more attention from external auditors [59]. Due to the higher environmental risks, auditors must carry out more audit procedures and bear the higher risk of audit failure [60]. The compliance constraint effect and the incentive and guidance effect of the environmental credit evaluation policy will improve environmental information disclosure and achieve sustainable growth of enterprises [25,26,37], which can help weaken the negative impact of clients’ environmental risks on the risk perception of auditors. Therefore, we conjecture that the ECR can mitigate the environmental risks, thereby reducing audit fees from the following two aspects.

On the one hand, the ECR can reduce the auditor's perception of environmental risks by improving the environmental information transparency of enterprises based on the compliance constraint effect, thus charging lower audit fees. Specifically, the ECR provides alternative workloads for auditors, enabling them to obtain environmental information with higher endorsement value by the government environmental departments at a lower audit time and cost. Song et al. (2023) found that better environmental performance could reduce auditors' workloads and risk perception [61]. Second, the high-quality disclosure of environmental information is an important strategy to respond to the environmental stakeholders, which helps maintain the legitimacy of enterprises. Yao et al. (2020) found that managers maintained legitimacy by disclosing more environmental information [22]. Third, environmental information disclosure alleviates information asymmetry [62,63], which can add an incremental value to auditing financial reports. Cormier et al. (2011) indicated that environmental information differs from financial information: auditors would consider the inconsistencies between financial data and environmental information [64]. Fourth, the transparency of environmental information inhibits agency costs [65], which can compress the space for executives to carry out environmental moral hazards and adverse selection [66].

On the other hand, the ECR can reduce the auditor's perception of environmental risk by enhancing enterprises' sustainable operation ability based on the incentive and guidance effect, thus charging lower audit fees. The poor environmental performance of enterprises will bring more uncertainty to their cash flow and sustainable operation ability [67,68]. Moreover, public boycotts, government fines and even being closed will dim enterprises' development prospects and the security of capital and earnings stability also encounter more challenges [21]; however, the ECR attaches importance to the incentive for environmental trustworthiness [26]. First, the government will provide tax relief, environmental subsidies and government procurement to enhance their competitiveness according to the results of ECR [26]. Second, the government encourages the banks to fully implement a green credit policy, which is conducive to improving availability, reducing costs and alleviating the constraints of financing [25]. Zhu and Wu (2022) found that the ECR improves the green innovation of heavily polluting enterprises due to the better financing mechanism, thus enhancing sustainable development ability [26]. Third, better ECR results can help improve enterprises' reputation and social image. Du et al. (2018) and Kong et al. (2014) found that the reputation incentive could help enterprises attract consumers, customers, and shareholders, and then build a more stable trust and cooperation relationship [67,69], which is good for increasing the stability of financial performance.

In summary, after implementing the environmental credit evaluation policy, the ECR forces audit clients to improve environmental information transparency and enhance their sustainable operation ability, reducing auditor's perception of the environmental risk and thereby charging lower audit fees. Based on the above logic, we propose the following hypotheses:

H1

Since the environmental credit evaluation policy is implemented, the ECR can significantly reduce audit fees.

H1a

The ECR reduces audit fees by improving environmental information transparency.

H1b

The ECR reduces audit fees by enhancing the sustainability of operations.

4. Research design

4.1. Sample selection and data sources

Heavily polluting enterprises are the key evaluation objects in the environmental credit evaluation policy; therefore, we choose China's A-share listed companies in heavily polluting industries as the research object. However, the emergence of the COVID-19 pandemic severely affected the work of auditors, such as audit plans, revision of identified risks, assessment of accounting estimates and policies and any other difficulties encountered during the audit engagement due to physical distancing [70]. To ensure the validity of the results, we do not involve the samples during the COVID-19. We refer to Bongiovanni and Fiandrino (2024) in defining the timing of the COVID-19 spread [71]. As the overall crisis period went from 02 January 2020 (i.e. the first discovery of cases of pneumonia and the closure of the Huanan Seafood Wholesale Market), we cover the 2008–2019 period as the initial sample.

Furthermore, the initial sample is screened as follows. (1) We eliminated the companies of ST, *ST, suspension of listed and delisting. (2) We deleted the companies with missing research variable data, (3) excluded the companies transformed into non-polluting industries and (4) omitted newly listed companies. Finally, we obtained 6707 valid observations. All the continuous variables in our research are winsorised at the 1% and 99% levels to avoid the influence of extreme observations.

The ECR data are collected and matched manually through the websites of local government environmental protection departments in China. Environmental information is collected from enterprises’ annual reports and social responsibility, sustainable development and environmental reports. All other data are obtained from the CSMAR, Wind and CNRNS databases.

4.2. Model and variables

Referring to Zuo and Wu (2022) [26], we construct the following heterogeneous timing DID model to estimate the impact of ECR on audit fees in heavily polluting enterprises.

FEE_ipt = α₀ + α₁ECR_ipt + Σ_jα_jControls_ipt + u_t + δ_p + ε_ipt (1)

where i, p, t and j represent the listed company, region, time and the number of control variables, respectively. u_t and δ_p represent the fixed effect of the year (Year-fe) and region (Province-fe), respectively, and ε_ipt is the error term.

Fee is the dependent variable, which indicates a company's audit fees. Following Li et al. (2020), we measure it by the natural logarithm of the audit fees incurred by listed companies [72].

The ECR is the core explanatory variable. Specifically, according to the implementation time of the environmental credit evaluation policy of the province where a listed company is located, we construct a dummy variable for the heterogeneous timing DID model. If the province where a company is located implemented the environmental credit assessment policy in period t from 2008 to 2019, the value is 1 for both the current year and subsequent years; otherwise, it is 0. In particular, if a province has issued the ECR policy in the first half of the year (before 30 June), it is deemed that the policy was implemented that year; otherwise, it was implemented the following year.

Controls include the following variables. Following the variable selection criteria in the previous literature, we choose the following control variables: enterprise size (Size) [49], asset–liability ratio (Lev) [26], return on assets (Roa) [73], property rights (Soe) [40], growth capability (Growth) [26], company value (TQ) 40], equity concentration (Largest) [25], operational complexity (Complexity) [51], board size (Board) [26], proportion of independent directors (Indep) [40], executive shareholding (Executives) [73], CEO duality (Dual) [40], audit opinion type (Opinion) [73], auditor change (Chang) [51], auditor type (Big4) [47] and audit effort (Workload) [51]. Table 1 defines all the variables.

Table 1.

A brief description of the variables.

Variable	Description
FEE	The natural logarithm of the audit fees
ECR	If the province where a company is located implements the environmental credit assessment policy, the value is 1 for the current year and after years; otherwise, it is 0.
Size	The logarithm of total assets
Lev	The total liabilities divided by the total assets
Roa	Net profit divided by average total assets
Soe	1 if a corporation is state-owned, 0 otherwise
Growth	(Current operating income − previous operating income)/previous operating income
TQ	The market value of the firm divided by the book value of total assets
Largest	Percentage of shares held by the largest shareholder
Complexity	(Total accounts receivable + total inventory)/total assets
Board	The logarithm of the number of directors on the board
Indep	The proportion of independent directors
Executives	The total of shares held by directors, supervisors and senior managers
Dual	A dummy variable that is equal to 1 if the chairperson is also the general manager and 0 otherwise
Opinion	1 if it is ‘non-standard without reservation’; otherwise, 0
Chang	1 if replacement occurs in the current period; otherwise, 0
Big4	1 if the company hires the Big 4 accounting firms; otherwise, 0
Workload	The difference between the audit report date of the current year and the financial statement date

To verify how the ECR affects audit fees, drawing on Baron and Kenny (1986) [74], combined with Formula (1), we construct (2), (3) for analysis.

Mediator_ipt = α₀ + α₁ECR_ipt + Σ_jα_jControls_ipt + u_t + δ_p + ε_ipt (2)

FEE_ipt = α₀ + α₁ECR_ipt + α₂Mediator_ipt + Σ_jα_jControls_ipt + u_t + δ_p + ε_ipt. (3)

Mediator refers to the mediating variables of this study, which are environmental information transparency and sustainable operation ability, respectively. EIT represents environmental information transparency, measured by the quality of environmental information disclosure. Referring to Clarkson et al. (2008) and Wiseman (1982) [75,76], and following the Index and Scoring Method for Enterprise Environmental Credit Evaluation (Trial), we construct the evaluation indicators for environmental information disclosure, as presented in Table 2. Next, according to the core of an indicator, a different weight is assigned. Finally, based on the unified scoring rules, we score the above evaluation indicators individually and use the method of comprehensive determination of weight to multiply the score of all indicators by the sum of weight ratios to obtain the final score (the possible maximum score is 100); this value is used as the quality of environmental information disclosure. The higher the final score, the better the EIT. Sustainable operation ability (SOA) represents the sustainable operation ability. Referring to Bryan et al. (2018) [77], we use the earnings volatility to measure it. The smaller the value, the higher the SOA.

Table 2.

The specific evaluation indicators for the quality of environmental information disclosure.

Serial number	Project details
	Indicator category	Detailed indicators	Weight
1	Environmental liabilities	Discharge of wastewater	5%
2		Exhaust emissions	5%
3		Soot or dust emissions	5%
4		Production of industrial solid waste	4%
5	Environmental costs	Energy consumption per 10,000 RMB GDP	3%
6		Total water consumption	3%
7		Total standard coal	3%
8		Pay pollution discharge fee	2%
9	Environmental management	Environmental protection agency, full-time and part-time staffing	4%
10		Environmental protection education, training and publicity	1%
11		Certification of cleaner production	3%
12		Environmental risk response plan and drill	4%
13		Implementation of the ‘three simultaneous’ system	3%
14		Certification of ISO environmental management system	2%
15		Declaration of pollution discharge permit	4%
16		Environmental auditing	1%
17	Pollution prevention	Energy saving, environmental protection technology R&D and process innovation	3%
18		Investment in green environmental	3%
19		Operation of pollution control facilities	4%
20		Standardised setting of sewage outlets	3%
21		Enterprise self-environmental monitoring	2%
22	Governance Performance	Wastewater discharge compliance rate	4%
23		Comprehensive utilisation rate of industrial solid waste	4%
24		Decline rate of comprehensive energy (standard coal)	4%
25		Discharge reduction of wastewater	4%
26		Emission reduction of exhaust gas	4%
27		Soot or dust emission reduction	4%
28		Punishment for environmental violations and occurrence of major environmental problems	5%
29		Compliance of noise emission in the plant area	3%
30		Environmental protection honours and commendations	1%
Total			100%

5. Empirical results

5.1. Descriptive statistics analysis

The descriptive statistics of variables are presented in Table 3. The average value of FEE is 13.570, with a minimum value of 12.430, a maximum value of 15.680, and a standard deviation of 0.620, indicating that audit fees vary relatively significantly among different enterprises. The average value of ECR is 0.470, indicating that approximately 47% of the observations are affected by implementing the environmental credit evaluation policy. Furthermore, the control variables are similar to the findings of previous related studies.

Table 3.

Descriptive statistical results.

Variable	N	Average	Median	Minimum	Maximum	Std.
FEE	6707	13.570	13.460	12.430	15.680	0.620
ECR	6707	0.470	0	0	1	0.499
Size	6707	22.180	22.020	19.480	25.910	1.262
Lev	6707	0.434	0.430	0.051	0.975	0.216
Roa	6707	0.044	0.038	−0.194	0.252	0.065
Soe	6707	0.444	0	0	1	0.497
Growth	6707	0.220	0.081	−0.719	5.999	0.773
TQ	6707	2.099	1.639	0.874	8.890	1.396
Largest	6707	0.351	0.332	0.092	0.750	0.1478
Complexity	6707	0.205	0.187	0.011	0.591	0.126
Board	6707	2.166	2.197	1.609	2.708	0.195
Indep	6707	0.369	0.333	0.308	0.571	0.050
Executives	6707	0.098	0	0	0.660	0.178
Dual	6707	0.229	0	0	1	0.420
Opinion	6707	0.042	0	0	1	0.201
Change	6707	0.108	0	0	1	0.310
Big4	6707	0.050	0	0	1	0.218
Workload	6707	92.440	94	28	118	20.220

5.2. Basic regression analysis

The regression results for the effect of ECR on audit fees are shown in Table 4. In Column (1), the coefficient of ECR is negative and significant at the 5% level without any control variables. Column (2) shows the result after adding the control variables, and the coefficient of ECR coefficient is −0.047 with a significance at the 1% level, indicating that after implementing the environmental credit evaluation policy, ECR further reduced companies’ audit fees. The above results support hypothesis H1.

Table 4.

Basic regression results.

Variable	(1)	(2)
	FEE	FEE
ECR	−0.057** (−2.314)	−0.047*** (−2.672)
Size		0.373*** (54.518)
Lev		−0.005 (−0.153)
Roa		−0.548*** (−6.050)
Soe		−0.000 (−0.034)
Growth		0.006 (0.816)
TQ		0.037*** (7.850)
Largest		−0.001*** (−3.835)
Complexity		0.251*** (6.130)
Board		0.046 (1.315)
Indep		0.003 (0.025)
Executives		0.244*** (8.757)
Dual		−0.002 (−0.190)
Opinion		0.060** (2.296)
Change		−0.019 (−1.194)
Big4		0.585*** (17.305)
Workload		0.001*** (4.814)
Constant	13.55*** (149.927)	4.948*** (28.465)
Year-fe	Yes	Yes
Province-fe	Yes	Yes
N	6707	6707
R2	0.357	0.562

Note: Values in parentheses are t-statistics. ***, ** and * denote significance at the 1%, 5% and 10% levels, respectively.

5.3. Mechanism analysis

The baseline regression demonstrated that ECR further reduces audit fees in heavily polluting enterprises. Next, we will verify its underlying mechanism: EIT and SOA.

Columns (1) and (2) in Table 5 show the results of the mediation effect of EIT. The coefficient for environmental credit evaluation policy support in Column (1) is significantly positive at the 10% level, indicating that ECR promotes EIT. The results in Column (2) show that the coefficient of ECR is −0.047 with a significance at the 1% level, and the EIT support coefficient is significantly positive at the 5% level. The above results demonstrate that the ECR reduce audit fees by promoting the transparency of environmental information, thereby verifying hypothesis H1a.

Table 5.

Test of mediation effects.

Variable	(1)	(2)	(3)	(4)
	EIT	FEE	SOA	FEE
ECR	0.403* (1.875)	−0.047*** (−2.653)	−0.003** (−2.333)	−0.046*** (−2.609)
EIT		0.002** (2.266)
SOA				−0.421*** (−3.791)
Size	1.785*** (23.546)	0.367*** (55.260)	−0.006*** (−7.488)	0.366*** (57.104)
Lev	−1.371*** (−3.733)	−0.005 (−0.150)	0.024*** (4.961)	−0.014 (−0.453)
Roa	0.067 (0.061)	−0.488*** (−5.652)	−0.030 (−1.504)	−0.472*** (−5.514)
Soe	0.035 (0.266)	−0.002 (−0.157)	0.003*** (2.697)	−0.003 (−0.313)
Growth	−0.284*** (−3.947)	0.003 (0.480)	0.002 (1.467)	0.003 (0.443)
TQ	−0.015 (−0.300)	0.036*** (8.211)	0.007*** (9.478)	0.032*** (7.362)
Largest	0.027*** (6.077)	−0.001*** (−3.544)	0.000*** (2.722)	−0.001*** (−3.851)
Complexity	1.849*** (3.757)	0.248*** (6.423)	−0.041*** (−8.008)	0.265*** (6.803)
Board	2.026*** (5.190)	0.004 (1.090)	−0.005 (−1.559)	0.004 (1.045)
Indep	−0.118 (−0.085)	−0.029 (−0.273)	−0.005 (−0.416)	−0.026 (−0.245)
Executives	−0.623 (−1.516)	0.228*** (8.356)	−0.003 (−1.020)	0.231*** (8.506)
Dual	−0.257* (−1.697)	−0.004 (−0.338)	0.000 (0.186)	−0.003 (−0.302)
Opinion	0.357 (1.024)	0.063** (2.563)	0.049*** (9.516)	0.038 (1.529)
Change	−0.656*** (−3.444)	−0.017 (−1.124)	0.007*** (3.396)	−0.019 (−1.268)
Big4	1.328*** (4.197)	0.546*** (17.622)	0.004** (1.969)	0.541*** (17.503)
Workload	−0.013*** (−4.235)	0.001*** (5.097)	−0.000* (−1.783)	0.001*** (5.341)
Constant	−35.670*** (−18.546)	5.112*** (33.679)	0.172*** (9.580)	5.103*** (34.232)
Year-fe	Yes	Yes	Yes	Yes
Province-fe	Yes	Yes	Yes	Yes
N	6707	6707	6707	6707
R2	0.247	0.562	0.071	0.564

Columns (3) and (4) of Table 5 present the results of the mediation effect of SOA. Column (3) displays the results of Formula (2), with the dependent variable being the SOA of enterprises. The coefficient of the environmental credit evaluation policy is significantly negative at the 5% level, which indicates that the ECR enhances the SOA. Column (4) reports the results of audit fees on the basic independent variable (ECR) and mediator (SOA). The coefficients of both ECR and SOA on audit fees are significantly negative at the 1% level, demonstrating that enhancing the sustainable operation ability is a channel for the environmental credit evaluation policy to affect audit fees of heavily polluting enterprises. Therefore, the hypothesis H1b is also verified.

5.4. Robustness checks

5.4.1. Parallel trend test

Constructing a DID model to estimate the impact of ECR on audit fees must satisfy the parallel trend test. Referring to Dyreng et al. (2016) [78], we revise Formula (1). Specifically, we re-constructed the interaction items in the 2 years before (Pre_2, Pre_1), the current year (Current) and 2 years after (Post_1, Post_2) of the environmental credit evaluation policy. Column (1) of Table 6 results show that the coefficients failed to pass the significance test in the prior two periods (Pre_2, Pre_1); however, the coefficients are significantly negative in the current period (Pre_2, Pre_1) and after two periods (Pre_2, Pre_1). Therefore, this study conforms to the parallel trend hypothesis, making the conclusion reliable.

Table 6.

Robustness test results.

Variable	(1)	(2)	(3)	(4)	(5)
	FEE	FEE	FEE	FEE	FEE
ECR		−0.051** (−2.196)	−0.018 (−0.871)	−0.051*** (−2.825)	−0.43** (−2.474)
Pre_2	0.004 (0.882)
Pre_1	−0.024 (−1.392)
Current	−0.038*** (−2.983)
Post_1	−0.055*** (−2.725)
Post_2	−0.069*** (−2.902)
Controls	Yes	Yes	Yes	Yes	Yes
Year-fe	Yes	Yes	Yes	Yes	Yes
Province-fe	Yes	Yes	Yes	Yes	Yes
Firm-fe	NO	NO	NO	NO	Yes
N	6707	4974	6707	6130	6707
R2	0.521	0.550	0.503	0.552	0.567

5.4.2. PSM-DID method

We use the propensity score matching-DID (PSM-DID) method to mitigate the impact of sample feature differences on results. The results in Column (2) of Table 6 present that the coefficient of ECR maintains a significant negative result at the 5% level, and the research conclusion remains unchanged.

5.4.3. Placebo test

We conducted a placebo test to verify that the impact of ECR on audit fees is not affected by other policies or random factors. Specifically, we reset the heterogeneous timing DID time point variable of the ECR. That is, the implementation time of the environmental credit evaluation policy is moved back to 2012, and then the placebo test is conducted again using the policy implementation year that has been virtualised. The results in Column (3) of Table 6 show that the coefficient of ECR is non-significant, indicating that the company's audit fees have no significant change without the exogenous impact of the environmental credit evaluation policy, which further supports the basic regression conclusion.

5.4.4. Deleting the disputed samples

A few regions in China, such as Jiangsu Province, conducted environmental impact assessments on the environmental behaviour of some key polluting enterprises before 2014 [26]. There are many differences from the ECR, which may also confuse the effect of the environmental credit evaluation policy and affect the reliability of the conclusions. Therefore, we eliminate the samples from Jiangsu Province and then re-perform a regression. The results in Column (4) of Table 6 show that the conclusion of H1 is robust.

5.4.5. Firm fixed effect test

To alleviate the interference of the differences in firm-level factors on the research results, we control the regressions’ firm fixed-effect (Firm-fe) and retest. The results in Column (5) of Table 6 show that the conclusion continues to be supported.

6. Further research on heterogeneity analysis

Although the basic regression has provided empirical evidence to support the use of ECR to reduce audit fees, the impact of government macro-policy on enterprise behaviour is often greatly influenced by enterprise micro-system [79]. Among them, property rights are regarded as the most basic micro-system of Chinese enterprises [3]. Existing literature has shown that the legal environment will directly impact enterprises' environmental governance strategies and auditors' responses [80,81]. Furthermore, apart from the legal environment, auditors’ behaviour may also depend on the characteristics of the informal institutional environment, such as the regional trust with transition economies [82]. Therefore, we further explore the effectiveness of ECR on audit fees across heterogeneous property rights, legal environments and regional trust to offer a more convincing result.

6.1. Property rights

As the vanguard of ecological civilisation construction, state-owned enterprises (SOEs) play the role of green image spokesperson of government [83], and their willingness to support environmental regulation policies is usually stronger. Furthermore, SOEs tend to obtain more high-quality resources, such as bank credits and government subsidies and derive technological advantages due to their political ties with the government [84]. Therefore, the effect of environmental governance and green transformation is more evident for (SOEs) [3,85]. As a result, the environmental risks of (SOEs) may be relatively lower, weakening the ECR's impact on audit fees. Based on the above logic, we argue that the negative relationship between the ECR and audit fees may be more evident in non-state enterprises.

To verify the above inference, we divide our sample into SOE and non-SOE according to the property rights. Moreover, we construct an interaction item of the group dummy variable (Soe) and ECR (Soe*ECR), which we then test. The related regression results are presented in Columns (1), (2) and (3) of Table 7. The coefficient of ECR is significantly negative at the 1% level in Column (2), but the result of Column (1) is insignificant, and the Permutation test is significant at the 5% level. Furthermore, in Column (3), the coefficient of Soe*ECR remains significantly positive at the 1% level. The above test results demonstrates that the negative effect of ECR on audit fees is more effective in non-SOEs than in SOEs. The possible reasons for this finding are that compared with SOEs, non-SOEs usually not only face more fierce market competition and deep-rooted credit discrimination [86,87], leading them to ignore ecological and environmental problems [88,89]. However, the environmental credit evaluation policy can press them to promote environmental information disclosure and achieve sustainable development, reducing auditors’ perception of environmental risks in non-SOEs and decreasing audit fees.

Table 7.

Further analysis results.

Variable	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)	(9)
	SOEs	non-SOE	Full sample	G-Law	P-Law	Full sample	H-Trust	L-Trust	Full sample
ECR	−0.034 (−1.362)	−0.075*** (−3.011)	−0.006** (−2.323)	−0.038 (−0.984)	−0.069*** (−3.126)	−0.061*** (−3.227)	−0.006 (−1.161)	−0.065*** (−3.151)	−0.055*** (−2.968)
Soe*ECR			0.092*** (5.052)
Law						−0.060*** (−3.820)
Law*ECR						0.052*** (3.002)
Trust									0.031 (0.316)
Trust*ECR									0.027** (2.330)
Size	0.377*** (39.223)	0.353*** (40.852)	0.352*** (26.374)	0.374*** (50.012)	0.345*** (28.592)	0.357*** (26.820)	0.375*** (49.457)	0.333*** (28.633)	0.358*** (26.744)
Lev	−0.033 (−0.734)	0.034 (0.795)	0.036 (0.781)	−0.027 (−0.739)	0.082 (1.515)	0.034 (0.731)	0.004 (0.097)	0.021 (0.409)	0.036 (0.770)
Roa	−0.669*** (−5.330)	−0.339*** (−2.804)	−0.177** (−2.198)	−0.575*** (−5.501)	−0.322** (−2.199)	−0.171** (−2.098)	−0.678*** (−6.469)	−0.145 (−0.990)	−0.173** (−2.112)
Soe			−0.043* (−1.787)	0.000 (0.017)	−0.004 (−0.240)	−0.083 (−1.593)	−0.001 (−0.082)	−0.002 (−0.142)	−0.083 (−1.585)
Growth	−0.003 (−0.324)	0.011 (1.057)	0.012 (1.345)	−0.002 (−0.237)	0.014 (1.303)	0.012 (1.312)	−0.007 (−0.750)	0.028*** (2.648)	0.012 (1.313)
TQ	0.046*** (6.909)	0.029*** (4.892)	0.013*** (3.045)	0.030*** (5.590)	0.050*** (6.580)	0.015*** (3.497)	0.030*** (5.608)	0.045*** (6.579)	0.015*** (3.509)
Largest	−0.001** (−2.287)	−0.001*** (−2.746)	0.117 (1.497)	−0.001*** (−3.015)	−0.001* (−1.726)	0.104 (1.328)	−0.002*** (−3.724)	−0.000 (−0.378)	0.104 (1.325)
Complexity	0.328*** (5.596)	0.169*** (3.218)	0.100** (2.218)	0.305*** (6.364)	0.129** (1.971)	0.113** (2.383)	0.250*** (5.138)	0.203*** (3.225)	0.110** (2.340)
Board	0.033 (0.662)	0.0428 (0.965)	0.045 (0.993)	0.040 (1.008)	0.027 (0.455)	0.042 (0.928)	0.083** (2.022)	−0.029 (−0.539)	0.044 (0.963)
Indep	−0.089 (−0.543)	0.026 (0.174)	−0.042 (−0.394)	0.043 (0.327)	−0.205 (−1.005)	−0.051 (−0.468)	0.217 (1.641)	−0.476** (−2.518)	−0.050 (−0.457)
Executives	0.315*** (5.301)	0.205*** (6.581)	0.048*** (4.931)	0.207*** (6.662)	0.346*** (5.902)	0.033*** (4.641)	0.211*** (6.619)	0.301*** (5.609)	0.034*** (4.662)
Dual	0.013 (0.706)	−0.012 (−0.864)	−0.002 (−0.198)	−0.001 (−0.055)	−0.005 (−0.218)	−0.003 (−0.257)	0.004 (0.278)	−0.011 (−0.552)	−0.003 (−0.268)
Opinion	0.061* (1.750)	0.068* (1.895)	0.056*** (2.637)	0.024 (0.792)	0.083** (2.034)	0.058*** (2.734)	−0.013 (−0.415)	0.146*** (3.785)	0.058*** (2.765)
Change	−0.019 (−0.864)	−0.012 (−0.576)	−0.009 (−1.224)	−0.019 (−0.920)	−0.004 (−0.172)	−0.008 (−1.068)	−0.002 (−0.103)	−0.019 (−0.795)	−0.008 (−1.076)
Big4	0.612*** (15.484)	0.475*** (10.170)	0.232*** (3.832)	0.518*** (15.678)	0.608*** (7.072)	0.234*** (3.811)	0.495*** (14.783)	0.715*** (9.403)	0.230*** (3.730)
Workload	0.001** (2.371)	0.001*** (4.558)	0.044*** (3.324)	0.001*** (3.658)	0.002*** (3.945)	0.048*** (3.619)	0.001*** (4.156)	0.001*** (3.512)	0.048*** (3.601)
Constant	4.986*** (20.811)	5.361*** (23.853)	5.309*** (15.827)	5.021*** (25.219)	5.533*** (18.909)	5.270*** (15.823)	4.797*** (23.655)	6.037*** (23.585)	5.184*** (15.818)
Year-fe	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Province-fe	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
N	2976	3731	6707	4551	2156	6707	4370	2337	6707
R2	0.603	0.511	0.566	0.651	0.546	0.563	0.553	0.565	0.562
Permutation test	Bdifference = −0.041 P = 0.043			Bdifference = −0.031 P = 0.011			Bdifference = −0.059 P = 0.024

6.2. Legal environment

Generally, in the areas with poor law levels, the intensity of environmental enforcement will be weaker, and the cost of environmental violations may be lower, thus shielding and even contributing to the environmental opportunistic behaviour of enterprises [90]. However, a good legal environment increases the pressure on enterprises' environmental legitimacy, urging them to improve their environmental governance capacity and create a favourable basis for auditors to conduct audit activities. As a result, their audit risks are relatively stable and controllable in areas with a good legal environment, which will weaken the ECR's impact on audit fees. Based on the above logic, we argue that the negative correlation between the ECR and audit fees may be more transparent in poor legal environments.

To verify the above inference, according to the median value of each region in the sub-score of ‘market intermediary organization development and legal system environment’ in the Fan Gang marketisation index, we introduce a variable of the legal environment (Law) and divide the full samples of this study into two groups as follows: the good legal environment (G-Law) and the poor legal environment (P-Law). Moreover, we construct an interaction item of the Law and ECR (Law*ECR), which we then test. The related regression results are listed in Columns (4), (5) and (6) of Table 7. It can be seen that the coefficient of ECR is significantly negative at the 1% level in Column (5), but the result of Column (4) is not significant, and the Permutation test is significant at the 5% level. Furthermore, in Column (6), the coefficient of Law*ECR remains significantly positive at the 1% level. Based on the above test results, the negative effect of ECR on audit fees is more effective in enterprises with a poor legal environment than in those with a good legal environment. One possible reason for this finding is that compared with those located in a good legal environment, auditors significantly reduce their sense of trust for the role of environmental enforcement of local government and the effect of environmental risk management in areas with a poor legal environment. This situation will lead the auditors to rely more on the support provided by the environmental credit evaluation policies to carry out audit activities.

6.3. Regional trust environment

On the one hand, regarding the integrity of executives, the self-implementation of the social paradigm and the cost of dishonesty will lead executives with high levels of regional trust to pay more attention to integrity, which can restrict their illegal and immoral behaviour [91]. On the other hand, in terms of access audit evidence, the complex structure of social networks and emphasis on the integrity and reputation not only make the high regional trust to expand auditors' access to information beyond the client (including but not limited to former auditors, customers, suppliers, and other stakeholders), but also can reduce the cost of information acquisition [92]. As a result, their audit risks are relatively stable and controllable in the areas with a high level of regional trust, which is likely to weaken the ECR's impact on audit fees. Based on the above logic, we conjecture that the negative correlation between the ECR and audit fees may be more robust in those with a lower level of regional trust.

To verify the above inference, referring to Deng et al. (2022) and Deng and Yu (2021) [93,94], we introduce a variable of regional trust (Trust) and measure it based on the survey data of relevant issues from the Chinese General Social Survey. According to the median value of Trust, we divide this study's total samples into two groups: the higher regional trust (H-Trust) and the lower regional trust (L-Trust). Moreover, we construct an interaction item of the Trust and ECR (Trust*ECR), which we then test. The related regression results are shown in Columns (7), (8) and (9) of Table 7. It can be seen that the coefficient of ECR is significantly negative at the 1% level in Column (8), but the result of Column (7) is not significant, and the Permutation test is significant at the 5% level. Furthermore, in Column (9), the coefficient of Trust*ECR is significantly positive at the 5% level. The above test results confirm that the negative effect of ECR on audit fees is more effective in enterprises with a lower regional trust than those with a higher regional trust. One possible reason for this finding is that compared with a higher regional trust, auditors significantly reduce their sense of trust for enterprises' executives and have low access to obtain audit evidence in enterprises with a lower regional trust. This situation will lead the auditors to rely more on the support provided by the environmental credit evaluation policies to carry out audit activities.

7. Discussion

Compared with previous studies, we delve into an interesting and often overlooked subject: the effect of ECR on audit fees and its mechanisms. The main differences between our research and previous studies are the following two aspects.

First, related conclusions have not yet reached a consensus regarding the relationship between environmental regulation policies and audit fees. For example, Liu et al. (2021) argued that China's new EPL further enhanced audit fees for heavily polluting enterprises [21]; this relationship was positive. However, Du et al. (2018) found that auditors applauded environmentally friendly enterprises, thereby charging lower audit fees [67]; the relationship should be negative. A possible reason for the mixed conclusions was that previous studies focused on only the command-and-control environmental regulation [18,19,21] or the market-based incentive environmental regulation [20,24], which were defective and might be regarded as a good opportunity by some enterprises but treated as a significant burden by some other enterprises. Therefore, it is worthy of in-depth exploration. Furthermore, to our knowledge, there is a paucity of research on the effect of ECR on audit fees, and the environmental credit evaluation policy has both natures for the constraint and incentive [26]. We examine the impact of ECR on audit fees using the implementation of an environmental credit evaluation policy as a quasi-natural experiment for the first time. Therefore, our study can help fill this gap.

Second, in terms of research interpretation, some researchers have explored the reaction of auditors to the environmental events or environmental risks; however, the existing research on how environmental regulation affects audit fees can still be broadly categorised into the ‘audit cost effect’ and the ‘risk compensation effect’ from the traditional perspective [20,24]. Nonetheless, related environmental channels are largely neglected. In other words, few studies connect audit fees to related environmental channels [21]. Although Liu et al. (2021) examined the effect of environmental information disclosure [21], the research finding might be challenging to motivate enterprises to disclose environmental information. Compared with previous studies, we identify two distinct channels through which the ECR influences audit fees: promoting the transparency of environmental information and enhancing sustainable operations. Therefore, our research findings demonstrate the role of the ECR in reducing auditors' perception level of environmental risks for enterprises and can further incentivise enterprises to disclose environmental information, which is another feature of this study.

8. Conclusion, recommendations and limitations

8.1. Conclusion

Taking the implementation of environmental credit evaluation policy as a quasi-natural experiment, based on the environmental governance effect of the environmental credit evaluation policy, and using the heterogeneous timing DID model method, this study investigates the impact of ECR on audit fees. The results show that the ECR can promote EIT and enhance SOA, reducing audit fees. The empirical results remain significant and credible when subjected to robust testing, such as parallel trend test, PSM-DID, placebo test, deletion of the disputed samples and the firm-fixed effect test. Further research shows that the negative correlation between the ECR and audit fees is more evident in non-state enterprises, enterprises with a low legal environment and those with lower regional trust.

8.2. Policy recommendations

First, the government should accelerate the implementation of environmental credit evaluation policies. On the one hand, it is necessary to cover this policy nationwide as soon as possible. On the other hand, it should expand the scope of industries and enterprises included in the ECR policy. Second, enterprises should attach more importance to the environmental credit evaluation policy, which can enhance their SOA and help auditors provide higher-quality audit services. Third, the government should attach importance to constructing the rule of law in the ecological environment and improve regional trust, creating a better external environment for auditors to conduct audit activities.

8.3. Limitations

First, our research is conducted within the distinctive context of China. It is necessary to consider the impact of ECR on audit fees extended to other developing countries in the future. Second, there is still a lack of a uniform environmental information disclosure system in China, leading to more green-washing, which may restrict the rationality of EIT. Therefore, it is necessary to consider other more reasonable ways to measure it in future. Third, our data are limited to listed companies. If data from non-listed companies are available and can obtain the environmental credit score of each enterprise in the future, the impact of ECR on audit fees can be revealed more intuitively and robustly. Furthermore, although we have adopted a heterogeneous timing DID model and conducted some robustness tests, the empirical results may nevertheless be affected by selection bias. Therefore, we should deploy more advanced econometric techniques to solve it in the future.

Funding

This study has been funded by the National Science Foundation of China (Grant number 71402082) and the National Social Science Foundation of China (Grant number 21AGL013).

Data availability statement

Data can be accessible through corresponding author.

Ethics declarations

Not applicable.

Additional information

No additional information is available for this paper.

CRediT authorship contribution statement

Jianghan Wang: Writing – review & editing, Writing – original draft. Haiyan Zhong: Validation, Supervision, Resources, Funding acquisition. Minxin Li: Methodology, Formal analysis

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.