Does signing auditors’ communist party membership shape audit quality? Evidence from China

Abstract

In this article, we investigate the effect of signing auditors’ communist party membership on audit quality. Relying on a sample of Chinese listed firms from the period 2001 to 2019, we find that firms with signing auditors who have communist party membership conduct less earnings management, indicating that signing auditors with party membership provide high level of audit quality. Moreover, the above relationship is more pronounced in small audit firms. We also find that firms who are audited by signing auditors with party status have lower likelihood of financial statements and loss avoidance. Further analyses suggest that signing auditors with party status can earn audit fee premium. The positive relationship between signing auditors with party membership and audit quality is more pronounced in non-specialists auditors and high client importance.

Keywords:

• Signing auditors
• communist party membership
• large auditors
• audit quality

1. Introduction

Recently, there is an increasing appreciation about how signing auditors’ personal traits influence audit quality (Chen et al., 2017; Guan et al., 2016; Gul et al., 2013; Robert Knechel et al., 2015; Taylor, 2011; Wang et al., 2015). These archival studies have shown the evidence that audit quality varies within the audit firm and much of this variation can be explained by the characteristics of signing auditors who are in charge of audit engagements (Chen et al., 2020; Chi & Chin, 2011; Chin & Chi, 2009; Chu et al., 2021; Hardies et al., 2016; Hou et al., 2020; Ittonen et al., 2013, 2015; Kallunki et al., 2019; Lee et al., 2019). In this article, we extend this line of literature by exploring the effect of signing auditors’ political status on audit quality.

A growing strand of accounting and economic literature, based on the social sciences theories (e.g. behavioral consistency theory, upper echelon theory, and social identity theory), analyzes how the CEOs’ or managers’ political status influences micro-corporate decisions and behaviours. These studies mainly focus on the U.S. capital markets and document the empirical evidence that CEOs’ or managers’ political preference is closely associated with corporate innovation (Han, 2019), corporate social responsibility (Chin et al., 2013; Di Giuli & Kostovetsky, 2014; Unsal et al., 2016), investment decisions (Elnahas & Kim, 2017), audit-client contracting (Bhandari et al., 2020), financial conservatism (Hutton et al., 2014), credit rating (Bhandari & Golden, 2021), and tax avoidance (Christensen et al., 2015; Francis et al., 2016). However, there is virtually limited empirical evidence on the relationship between signing auditors’ political status and audit quality. Therefore, we attempt to bridge the void by empirically examining the impact of signing auditors’ political identity on the quality of audits. Specially, we focus on the Chinese background and test how signing auditors’ communist party membership shapes audit quality.

As an important individual characteristic, Communist Party of China (CPC) member is an important status for Chinese individuals. The selection process for becoming a CPC member is strict as the total number of party members is limited (Dickson & Rublee, 2000; Ma & Iwasaki, 2021). Hence, those who have membership of CPC are usually regarded as social elite (Cheng & White, 1990; Dickson, 2007). For instance, CPC members are more likely to be highly educated and highly skilled (Dickson & Rublee, 2000; Ma & Iwasaki, 2021). Individuals with CPC status can establish wide social network (Zhang & Anderson, 2014; Nikolov et al., 2020), are imprinted with communist sociology (Marquis & Qiao, 2020), are endowed with more social responsibility (Li et al., 2020), and are bound by party discipline and rules (Gore, 2016; Heilmann, 2005). Hence, communist party membership is an important characteristic that cannot be ignored in the study of individuals’ behaviours. Following this chain of logic, limited but growing literature has validated that firms managed by CEOs or managers with communist party membership exhibit varying corporate behaviours (Dong et al., 2016; Li et al., 2020; Yan & Xu, 2022; Zhou et al., 2021). Signing auditors who are in charge of audit work spend significant time and effort to perform a variety of task including client risk assessment, audit plan making, clients communication, and audit opinion issuance (Gul et al., 2013; Taylor, 2011). Their difference in risk appetite, ethical standards, and ability could inevitably lead to variation in audit quality (Chen et al., 2017; Hou et al., 2020; Wang et al., 2015). Accordingly, CPC membership has important influence on signing auditors’ behaviours and thereby would exert significant impact on the audit outcomes.

To dig into our research question, this article uses a sample of Chinese listed firms between the year 2001 and 2019 to conduct empirical tests. At present, only a few countries or regions (e.g. Australia, China, Sweden, Taiwan, the United Kingdom, and the United States) require to disclose signing auditors (Chen et al., 2016; Guan et al., 2016). The Chinese audit report is required to disclose names of the signing auditors and signing auditors characteristics can be obtained from the China Institute of Certified Public Accountants (CICPA), which makes us available to conduct empirical research. Our empirical results show that signing auditors with communist party membership are negatively related with earnings management and thus indicate that signing auditors who are members of CPC provide high level of audit quality. This findings support our conjecture that CPC membership is an important factor influencing individuals’ behaviours. Moreover, we find that the positive relationship between signing auditors’ CPC membership and audit quality is more pronounced in small audit firms, suggesting that audit firm size mitigates the positive effect of signing auditors’ CPC membership on audit quality. Furthermore, we conduct a variety of robustness tests including alternative proxies for audit quality and signing auditors’ CPC membership and results remain unchanged, which bolsters the credibility of our hypotheses. This article documents that firms who select signing auditors with CPC membership have lower likelihood to financial misstatements and loss avoidance. The results are still valid after conducting a series of tests to tackle with potential endogeneity issue. In addition, further analyses illustrate that signing auditors with party status can earn audit fee premium. The positive relationship between signing auditors with CPC membership and audit quality is more pronounced in non-specialists auditors and high client importance (both at signing auditors level and at audit firms level).

This article may make contributions to the literature along several dimensions. First, this article enriches the growing literature about the effect of signing auditors’ personal traits on audit quality. Recent relevant studies have devoted to explore how the signing auditors’ demographics characteristics such as gender, professional experience, education level, industry expertise influence the quality of audits undertaking (Chen et al., 2020; Chi & Chin, 2011; Chin & Chi, 2009; Chu et al., 2021; Hardies et al., 2016; Hou et al., 2020; Ittonen et al., 2013, 2015; Lee et al., 2019). However, these existing research rarely provides an analysis of the association between signing auditors’ political status and audit outcomes. In this article, we try to address this void by empirically examining the role of signing auditors’ communist party status in audit quality. In this regard, our empirical findings shed new light on the important influence of signing auditors’ characteristics on audit quality and add to the literature on the influencing factors of audit quality.

Second, this article contributes to research about the economic consequence of Communist Party status. As an important personal characteristics of individuals, the status of CPC membership is closely associated with individuals’ behaviours (McLaughlin, 2017; Nikolov et al., 2020; Zhang & Anderson, 2014). A strand of literature has provided empirical evidence that CEOs or managers who have party membership have important effect on corporate social behaviours, firm’s international expansion, bank loans obtainability, overinvestment (Dong et al., 2016; Li et al., 2020; Marquis & Qiao, 2020; Yan & Xu, 2022; Zhou et al., 2021). Yet, there remains silence on the effect of party membership on audit behaviours. Based on a sample of Chinese listed firms, this study documents a positive correlation between the signing auditors’ communist party status and audit quality. The results highlight the important role of CPC membership on audit outcomes and thereby provide new empirical evidence for the economic consequences of individuals’ party membership.

The article is organised as follows. The next section summarises the background literature and develops the hypotheses. Section 3 describes the sample, data, and research design. Section 4 reports the descriptive statistics, the univariate analysis and multivariate results. In the final section, we summarise our main results and conclude the article.

2. Literature review and hypotheses development

In recent years, auditing research has begun to push the analysis unit down to investigate whether the signing auditors affect audit quality considering the decisive role of signing auditors in audit outcomes (Aobdia et al., 2015; Chen et al., 2016; Robert Knechel et al., 2015; Wang et al., 2015). Some studies use the fixed effects method to empirically examine the role of signing auditors in explain audit quality variation. For example, Gul et al. (2013) use a sample of Chinese listed firms to investigate whether signing auditors matter to audit quality and empirical results document that signing auditors can serve as an important influencing factor of audit quality incremental to audit firms. Similarly, Cameran et al. (2018) provide the evidence that signing auditors have more explanation on audit quality than audit firm and branch office. Further, a growing number of studies try to open the ‘black box’ by exploring how the signing auditors’ personal attributes shape audit quality. Chin and Chi (2009) suggest that firms who select signing auditors with industry expertise have lower likelihood of financial restatements. Similarly, Chi and Chin (2011) find that signing auditor with industry expertise is negatively associated with earnings management and thus indicates that signing auditors who are industry expertise are capable to provide high quality audits. Ittonen et al. (2013) and Lee et al. (2019) argue that female auditors provide high-quality auditing in contrast male auditors. Ittonen et al. (2015) show that signing auditors who have more clients deliver high level of audit quality because they are inclined to keep high level of independence. In contrast, Chen et al. (2020) show that signing auditors’ workload is negative associated with audit quality. Guan et al. (2016) document that school ties between signing auditors and client executives negatively influence quality of audits. Chen et al. (2017) show that signing auditors’ foreign working experience is associated with high level of audit quality because signing auditors with foreign experience tend to maintain high independence and scepticism. Kallunki et al. (2019) use Sweden listed firms and find that signing auditors’ IQ scores are closely related with audit quality. Hou et al. (2020) examine the relationship between signing auditors’ foreign education experience and audit quality and document that foreign education experience enables signing auditors to conduct high quality of audits. Chu et al. (2021) document the evidence that signing auditors with high education level provide high audit quality. However, there is generally lack of research about the effect of signing auditors’ political status on audit quality. In this article, we fill this gap to address how signing auditors’ communist party membership influences audit quality.

As an important characteristic of individuals, communist party status plays a significant role in explaining individuals’ behaviours in China. As the total number of party members is limited, individuals who want to be a party membership must undergo a rigid selection process (Dickson & Rublee, 2000; Ma & Iwasaki, 2021). Hence, party membership can represent one’s ability or other characteristics that can not be directly observed (Cheng & White, 1990; Dickson, 2007). The economic consequences of party status are widely concerned and studied. It is shown in the existing literature that the party member status facilitates individuals to enhance social capital and obtain wage premium (Dickson, 2014; Li et al., 2007; McLaughlin, 2017; Nikolov et al., 2020; Zhang & Anderson, 2014). In addition, some scholars have begun to pay attention to the important role of managers’ party membership in corporate decision-making and behaviour. For instance, Dong et al. (2016) and Cheng (2022) find that the presence of the Communist Party of China guides firms to engage in more employment protection. Li et al. (2020) suggest that managers’ CPC member can improve corporate investment efficiency. Yan and Xu (2022) document that private entrepreneurs with party status are inclined to participant more in environmental protection. Accordingly, we conjecture that signing auditors with party status exhibit different audit styles and eventually lead to varying audit quality.

Prior literature has shown that party membership increase individuals’ social capital (Ma & Iwasaki, 2021; Zhang & Anderson, 2014). That is, party status facilitates individuals to access wide social network that could yield valuable resources and connections (Nikolov et al., 2020). As Talavera et al. (2012) find that, managers’ CPC membership with their social capital advantage help firms obtain bank loans. Hence, we argue that signing auditors with party membership have more expansive social network. Social capital enhances signing auditors’ professional competence to perform the audit task and in turn lead to high level of audit quality (Cohen et al., 2008; Hitt et al., 2001). Specially, wide social networks grant signing auditors to acquire valuable information including market conditions, industry trends and regulations and thus help signing auditors to improve professional competence (Bianchi et al., 2020; Horton et al., 2012). Consequently, signing auditors with party status are capable to serve the external monitoring role.

Moreover, the values and ideologies contained in the CPC have an important effect on individuals’ behaviours (Yan & Xu, 2022). The Party Constitution clearly states that party members should have the qualities of ‘altruistic spirit of serving the people’, ‘awareness of discipline and law-abiding rules’, ‘practical and realistic work style’, and so on. With the long-term and regular participation of learning and education activities, these spirits and values are deeply rooted in the thinking mode and behaviour patterns of party members (Marquis & Qiao, 2020; Wang et al., 2019). Specific to audit quality, low-quality audit is contrary to safeguard the interests of stakeholders, which is obviously contrary to the values of ‘altruism’, ‘seeking truth from facts’, and ‘prudence’. Under the influence of the communist ideology, signing auditors with party status have more motivations to keep independence and curb managers’ earnings manipulation behaviours.

In addition, party membership can represent one’s ability or excellent qualities that cannot be directly observed (Dickson, 2014). Generally speaking, party members can be regarded as elite class and are also excellent representative of ‘pioneer model’ (Ma & Iwasaki, 2021). This status not only brings reputation and economic benefits to individuals, but also endows individuals with more social responsibility and attention (Li et al., 2020). Individuals with party membership are encouraged to ‘take the lead’ and serve as examples to the rest of the population (Dickson & Rublee, 2000). As a result, signing auditors who are CPC members are more eager to maintain independence and prudence and ensure high-quality audit services to maintain their reputation and fill the social responsibility. Otherwise, these signing auditors would face high cost of reputation loss, which would seriously ruins their market competitiveness of signing auditors.

Finally, party members should not only abide by basic laws and regulations, but also abide by party discipline and party rules (Gore, 2016; Heilmann, 2005). For example, according to Chinese Communist Party Disciplinary Regulations, ‘if a party member violates the national laws and regulations, the rules and regulations of enterprises, institutions or other social organisations, in addition to being punished by the national laws, the party organisation should also impose party disciplinary sanctions in accordance with the regulations after verifying the seriousness of the circumstances’. Hence, this internal supervision mechanism greatly increases the cost of audit failure as signing auditors with party status are subject to audit standards and party rules. Therefore, under the restraint and deterrence of party discipline and party rules, these signing auditors tend to be more cautious in the audit process and are more inclined to curb clients firms’ accounting information manipulation, which is conducive to provide higher quality audit services.

Based on the above discussions, we hypothesise that:

Hypothesis 1. Ceteris paribus, signing auditors’ communist party membership is positively associated with audit quality.

Previous literature shows that auditor size is positively related with audit quality (Lim et al., 2016; Svanberg & Öhman, 2016). As a response, we further test whether the positive association between signing auditors’ communist party membership and audit quality is influenced by auditor size.

In contrast with small auditors, large auditors with larger number of clients have strong incentives to identify accounting fraud and withstand client pressure (DeAngelo, 1981). This is because no single client is important to a large auditor and the auditors have more to lose if they misreport. Hence, large auditors have strong incentives to maintain audit independence and objectivity, which motivates signing auditors to better play the monitoring role. Moreover, with the high reputation, large audit firms have more motivations to protect their reputation by providing high audit quality to avoid their reputations damage and the cost litigation (Francis & Wilson, 1988; Lim et al., 2016). As such, it is less likely for signing auditors who work in large audit firms to collude with clients firms. In addition, large audit firms usually have more resources to support training and better quality control systems (Svanberg & Öhman, 2016), which is helpful to enhance signing auditors’ professional competence and hinder signing auditors from making unethical decisions. Based on the above discussions, we argue that larger audit firms help signing auditors improve their professional competence and independence and thus weaken the advantages of signing auditors with party membership, which makes the positive impact of signing auditors’ party status on audit quality less prominent. Accordingly, we propose our Hypothesis 2.

Hypothesis 2. Ceteris paribus, the positive effect of signing auditors’ communist party membership on audit quality is more pronounced in small audit firms.

3. Research design

3.1. Sample selection and data

Our initial sample consists of Chinese listed firms on Shanghai and Shenzhen Exchanges between the year 2001 and 2019. Referring to the relevant literature, we conduct the following procedures. (1) We delete firm-year observations that are belong to banking, insurance, and other financial industries (1,231); (2) We eliminate firms whose listed age is below one year(3,080); (3) we exclude observations without sufficient data to calculate discretionary accruals (4,060); (4) we delete observations with missing data on signing auditors (3,467); (5) we exclude observations with missing control variables (822). Finally, we obtain a sample of 29,450. The detailed sample selection is listed in Appendix A.

The data we employed in this study is compiled from multiple sources: First, data on signing auditors’ party membership and other characteristics are manually obtained from the CICPA (http://cmis.cicpa.org.cn); Second, other data we used is collected (calculated) from the China Stock Market and Accounting Research database (CSMAR).

3.2. Empirical model

We apply the following model to examine our hypotheses. Referring to relevant literature (Chen et al., 2016; Guan et al., 2016; Hou et al., 2020; Xiong et al., 2020), we incorporate audit quality, signing auditors’ CPC membership, auditor-specific characteristics, firm-level controls, and other potential variables into our model (1). (Model 1) $$\begin{array}{c}\left|DA\right|={\alpha }_0+{\alpha }_{1}AR\_\mathit{CPC}+{\alpha }_{2}EDU+{\alpha }_{3}F\mathit{EMALE}+{\alpha }_{4}AR\_\mathit{EXPERT}\\ +{\alpha }_{5}AF\_\mathit{EXPERT}+{\alpha }_{6}BIG10+{\alpha }_{7}S\mathit{IZE}+{\alpha }_{8}LEV+{\alpha }_{9}BTM\\ +{\alpha }_{10}ROE+{\alpha }_{11}T\mathit{URNOVER}+{\alpha }_{12}INV+{\alpha }_{13}L\mathit{ISTAGE}+{\alpha }_{14}S\mathit{TATE}\\ +\sum AND+\sum \mathit{YEAR}+\sum \mathit{IND}+\epsilon \end{array}$$(Model 1)

In model (1), our outcome variable is audit quality. We adopt earnings management as the surrogate measurement of audit quality (Guan et al., 2016; Gul et al., 2013) and specially the following modified Jones model is employed to calculate discretionary accruals (Dechow et al., 1995). (Model 2) $$\frac{{\mathit{ACC}}_{j,t}}{{TA}_{j,t-1}}={\mathrm{\beta }}_1\mathrm{ }\frac{1}{{TA}_{j,t-1}}\mathrm{ }+\mathrm{ }{\mathrm{\beta }}_2\frac{{\Delta \mathit{REV}}_{j,t}}{{TA}_{j,t-1}}\mathrm{ }+\mathrm{ }{\mathrm{\beta }}_3\mathrm{ }\frac{{\mathit{PPE}}_{j,t}}{{TA}_{j,t-1}}\mathrm{ }+{\mathrm{\epsilon }}_{\mathrm{j},\mathrm{t}}$$(Model 2) (Model 3) $$\frac{{\mathit{NACC}}_{j,t}}{{TA}_{j,t-1}}={\mathrm{\beta }}_1\mathrm{ }\frac{1}{{TA}_{j,t-1}}\mathrm{ }+\mathrm{ }{\mathrm{\beta }}_2\mathrm{ }\frac{{(\Delta \mathit{REV}}_{j,t-}{\Delta \mathit{REC}}_{j,t})}{{TA}_{j,t-1}}\mathrm{ }+\mathrm{ }{\mathrm{\beta }}_3 \frac{{\mathit{PPE}}_{j,t}}{{TA}_{j,t-1}}{\mathrm{\epsilon }}_{\mathrm{j},\mathrm{t}}$$(Model 3) (Model 4) $${\mathrm{ }DA}_{j,t}\mathrm{ }={\mathit{ACC}}_{j,t}\mathrm{ }-\mathrm{ }{\mathit{NACC}}_{j,t}$$(Model 4)

In Model (2), j and t denote for the firm and the year, respectively. TA_j,t-1 denotes total assets in year t-1. ACC_j,t denotes the total accruals which is calculated as net income minus operating cash flows. ΔREV_j,t equals the change in net incomes from year t-1 to year t. PPE_j,t is the gross value of the property, plant and equipment at the end of year t. ACC, ΔREV, PPE, TA are all from financial statements. In model (3), ΔREC_j,t is the change in accounting receivables from year t-1 to year t. We base the Model (3) using the estimated coefficients β₁, β₂, β₃ from Model (2) to calculate non-discretionary accruals (NACC_j,t). Finally, we calculate the discretionary accruals (DA) using the model (4) in which ACC is obtained from the financial statements and NACC is obtained from the model (3). The absolute value of discretionary accruals is adopted to capture the extent of earnings management (|DA|). The high values of |DA| indicate high level of earnings management and low level of audit quality.

Our main test variable, AR_CPC, captures signing auditors’ party status. The party membership of signing auditors is defined as ‘equalling to 1 if signing auditors have communist party status and 0 otherwise’. There are at least two signing auditors (three signing auditors in some case) in the Chinese audit reports. Hence, variables of signing auditor characteristics in this study are measured as the average value of two (or more) signing auditors.

To better examine the incremental effects of the signing auditors’ CPC membership on audit quality, we include a number of control variables in model (1). First, we refer to the relevant audit quality research and incorporate a set of signing auditors’ characteristics including education level, gender, industry expertise and audit firms characteristics including industry expertise and size (Chen et al., 2017; Chi & Chin, 2011; Hou et al., 2020; Xiong et al., 2020). Second, to address the impacts of firm characteristics on audit quality, firm-level characteristics are controlled for in model (1) (Chen et al., 2017; Chu et al., 2021) and firm size (SIZE), firms leverage (LEV), book to market value (BTM), return on equity (ROE), assets turnover (TURNOVER), inventory ratio (INV), listing age (LISTAGE), nature of ownership (STATE) are controlled in empirical model. Finally, a set of audit firm, year and industry dummies (the CSRC classification) are added to control for the fixed effect of audit firms, calendar years, and industries.

The Appendix B provides detailed definitions of variables used in our empirical model. In order to attenuate the undue effect of outliers, we winsorize all the continuous variables at the 1st (99th) percentiles of their annual distributions throughout this study.

4. Empirical results

4.1. Descriptive results

In Table 1, all the variables adopted in our main tests are descripted. The mean value of |DA| is 0.070, with a maximum value of 0.489. AR_CPC has a mean value of 0.238, indicating that signing auditors with CPC membership account for 23.8% in Chinse capital market. With regard to the control variables, on average and approximately, 13% of signing auditors have graduate degree or above (EDU). 31.1% of signing auditors are female. 6.7% of signing auditors are with industry expertise (AR_EXPERT) and 20.9% of audit firms are industry expertise (AF_EXPERT). 48% of Chinese listed firms in our sample are audited by large auditors (BIG10). In terms of firm-level control variables, the mean value of firm size is 21.986 (SIZE). The average ratio of long-term debts account for 7.4% of total assets (LEV). The mean value of book to market value is 0.570 and the return on equity (ROE) is 0.044. TURNOVER has a mean value of 0.662. The average ratio of inventory to total assets is 0.158 (INV). The listing age of listed firms is 10.367. 49.6% of Chinese listed firms in our sample are state-owned firms.

Table 1. Descriptive statistics.

Variables	N	MEAN	SD	MIN	p25	p50	p75	MAX
|DA|	29450	0.070	0.081	0.001	0.020	0.045	0.088	0.489
AR_CPC	29450	0.238	0.306	0.000	0.000	0.000	0.500	1.000
EDU	29450	0.130	0.241	0.000	0.000	0.000	0.000	1.000
FEMALE	29450	0.311	0.328	0.000	0.000	0.500	0.500	1.000
AR_EXPERT	29450	0.067	0.250	0.000	0.000	0.000	0.000	1.000
AF_ EXPERT	29450	0.209	0.407	0.000	0.000	0.000	0.000	1.000
BIG10	29450	0.480	0.500	0.000	0.000	0.000	1.000	1.000
SIZE	29450	21.986	1.307	19.165	21.072	21.842	22.731	25.909
LEV	29450	0.074	0.100	0.000	0.000	0.029	0.113	0.446
BTM	29450	0.570	0.260	0.086	0.364	0.550	0.763	1.155
ROE	29450	0.044	0.200	−1.253	0.023	0.063	0.111	0.523
TURNOVER	29450	0.662	0.477	0.047	0.345	0.548	0.832	2.650
INV	29450	0.158	0.146	0.000	0.062	0.120	0.200	0.725
LISTAGE	29450	10.367	6.203	1.000	5.000	9.000	15.000	25.000
STATE	29450	0.496	0.500	0.000	0.000	0.000	1.000	1.000

Note: Table 1 reports the results of descriptive statistics, including the values of mean, median, standard deviation, minimum, maximum, 25 percentile, 75 percentile and maximum.

Source: Authors.

4.2. Pearson correlation test

Table 2 shows the Pearson’s correlations matrix between the variables used in our regression analyses. It can be seen that AR_CPC is negatively correlated with earnings management (|DA|) and thus provides preliminary support to our hypothesis 1. EDU and FEMALE are significantly negatively associated with |DA|, which indicates that signing auditors with high education level and female auditors provide high audit quality. AF_ EXPERT and BIG10 are negatively associated with earnings management, indicating that industry expertise auditors and large auditors are capable to provide high audit quality. LEV, BTM, ROE, and STATE (INV, LISTAGE) are positively (negatively) associated with audit quality. These results highlight that it is important to include these control variables into our regressions. The results of Table 2 also indicate our empirical analyses suffer no serious collinearity problem as absolute values of most correlation coefficients are relatively low.

Table 2. Pearson correlation test.

Variables		（1）	（2）	（3）	（4）	（5）	（6）	（7）	（8）
|DA|	（1）	1.000
AR_CPC	（2）	−0.014**	1.000
EDU	（3）	−0.012**	0.140***	1.000
FEMALE	（4）	−0.012**	−0.001	−0.010*	1.000
AR_EXPERT	（5）	0.031***	0.002	0.004	−0.010*	1.000
AF_ EXPERT	（6）	−0.024***	−0.026***	−0.026***	0.003	0.095***	1.000
BIG10	（7）	−0.054***	−0.043***	−0.028***	0.025***	−0.033***	0.301***	1.000
SIZE	（8）	−0.121***	−0.042***	−0.005	0.021***	−0.087***	0.040***	0.243***	1.000
LEV	（9）	−0.014**	−0.020***	0.014**	0.003	0.021***	−0.013**	0.038***	0.414***
BTM	（10）	−0.134***	−0.006	0.022***	0.000	−0.042***	−0.046***	0.019***	0.556***
ROE	（11）	−0.117***	0.022***	0.010*	0.020***	−0.017***	0.025***	0.045***	0.115***
TURNOVER	（12）	0.009	0.012**	0.022***	0.013**	−0.050***	−0.025***	0.027***	0.080***
INV	（13）	0.127***	−0.001	0.007	0.003	−0.038***	−0.036***	−0.015***	0.090***
LISTAGE	（14）	0.046***	−0.033***	−0.020***	0.013**	0.013**	0.041***	0.067***	0.286***
STATE	（15）	−0.067***	0.007	0.069***	0.023***	0.021***	−0.061***	−0.063***	0.190***
Variables		（9）	（10）	（11）	（12）	（13）	（14）	（15）
LEV	（9）	1.000
BTM	（10）	0.337***	1.000
ROE	（11）	−0.009	−0.067***	1.000
TURNOVER	（12）	−0.215***	0.010*	0.158***	1.000
INV	（13）	0.041***	0.111***	0.034***	0.020***	1.000
LISTAGE	（14）	0.156***	0.180***	−0.029***	0.013**	0.099***	1.000
STATE	（15）	0.179***	0.239***	−0.001	0.079***	0.005	0.241***	1.000

Note: Table 2 reports the results of Pearson correlation analysis. *p < 10%; **p < 5%; ***p < 1% (two-tailed).

Source: Authors.

4.3. Main results

4.3.1. Signing auditors’ communist party membership and audit quality

The empirical results of the relationship between signing auditors’ communist party membership and audit quality are tabulated in Table 3. The coefficient on AR_CPC is significantly negative at the 1% level (-0.004 with t = -2.88), suggesting the level of earnings management is mitigated when signing auditors have party status. The above results validate that signing auditors with communist party membership are more capable to provide high level of audit quality undertaking.

Table 3. Signing auditors’ communist party membership and audit quality.

Variables	|DA|
	Coefficient	T value
AR_CPC	−0.004***	−2.88
EDU	−0.002	−1.19
FEMALE	−0.003**	−2.11
AR_EXPERT	−0.001	−0.26
AF_ EXPERT	0.001	0.44
BIG10	−0.004**	−2.28
SIZE	−0.002***	−3.02
LEV	0.039***	6.00
BTM	−0.054***	−19.81
ROE	−0.054***	−12.78
TURNOVER	0.010***	7.53
INV	0.041***	8.36
LISTAGE	0.001***	11.94
STATE	−0.013***	−11.41
INTERCEPT	0.131***	9.90
YEAR	YES
IND	YES
AND	YES
Observations	29450
Adj R^2	0.099
F	23.344

Note: Table 3 reports the OLS regression results of the model (1). *p < 10%; **p < 5%; ***p < 1% (two-tailed). All reported t statistics are based on standard errors adjusted for Huber-White (White, 1980).

Source: Authors.

With regard to signs and significances of control variables, the significantly negative coefficient on FEMALE suggests that female auditors have high audit quality. BIG10 is negatively associated with |DA|, meaning that large auditors provide high quality of audits. SIZE, BTM, and ROE have negative and significant coefficients which indicate that firms with large size, high book to market value, and well financial performance are related with low level of earnings management. Firm leverage (LEV), assets turnover (TURNOVER), and inventory ratio (INV) are positively correlated with earnings management. The significantly negative coefficient on STATE suggests that state-owned firms are less likely to engage in earnings management.

4.3.2. Signing auditors’ communist party membership, audit firm size and audit quality

In Table 4, we divide our full sample into two subsamples based on the audit firm size: the large auditor subsample (BIG10 = 1) and the small auditor subsample (BIG10 = 0). As shown in columns (1)-(2) of Table 4, the coefficient on AR_CPC is significantly negative (-0.005 with t = -2.03) (the small auditor subsample). In contrast, coefficient on AR_CPC is insignificant in in columns (3)-(4) of Table 4 (the large auditor subsample). These empirical results shown in Table 4 mean that the positive effect of AR_CPC on audit quality is more pronounced in the small auditor subsample. The chow test for the difference between two subsamples shown in the last line of Table 4 is significant at the 1% level, indicating the rationality of subsample tests based on audit firm size.

Table 4. Signing auditors’ communist party membership, audit firm size and audit quality.

variables	(1)		(2)
	BIG10 = 0		BIG10 = 1
	Coefficient	T value	Coefficient	T value
AR_CPC	−0.005**	−2.03	−0.003	−1.37
EDU	−0.003	−1.09	−0.002	−0.78
FEMALE	−0.005**	−2.19	−0.001	−0.71
AR_EXPERT	0.000	0.01	−0.002	−0.53
AF_ EXPERT	−0.002	−0.56	0.002	1.01
SIZE	−0.003***	−3.12	−0.001	−1.45
LEV	0.036***	3.79	0.041***	4.64
BTM	−0.068***	−16.72	−0.040***	−10.90
ROE	−0.048***	−8.92	−0.060***	−9.08
TURNOVER	0.007***	3.52	0.014***	7.58
INV	0.042***	6.24	0.042***	5.71
LISTAGE	0.002***	10.10	0.001***	6.30
STATE	−0.014***	−8.59	−0.012***	−7.50
INTERCEPT	0.171***	8.47	0.114***	6.03
YEAR	YES		YES
IND	YES		YES
AND	YES		YES
Observations	15303		14147
Adj R^2	0.107		0.087
F	22.732		22.023
Chow test for two subsamples	123.07***(0.000)

Note: Table 4 conducts the subsample tests based on the audit firm size. * p < 10%; ** p < 5%; *** p < 1% (two-tailed). All reported t statistics are based on standard errors adjusted for Huber-White (White, 1980).

Source: Authors.

4.4. Robustness tests

4.4.1. Robustness tests using other measurement of earnings management

Alternatively, we further adopt the augmented Jones model of Ball and Shivakumar (2006) considering cash flow changes to estimate the discretional accruals (|DA_DCF|) and re-run the model (1). As shown in Table 5, when using |DA_DCF| as dependent variables, AR_CPC has positive effect on audit quality and this effect is more obvious in large audit firms, thus validating Hypotheses 1 and 2 again.

Table 5. Robustness test using other measurement of earnings management.

Variables	|DA_DCF|
			BIG10 = 0		BIG10 = 1
	(1)		(2)		(3)
	Coefficient	T value	Coefficient	T value	Coefficient	T value
AR_CPC	−0.003***	−2.81	−0.003*	−1.95	−0.002	−1.01
EDU	−0.003*	−1.79	−0.003	−1.20	−0.002	−1.18
FEMALE	−0.001	−1.01	−0.002	−1.02	−0.001	−0.72
AR_EXPERT	−0.002	−0.92	0.000	0.06	−0.005**	−2.05
AF_ EXPERT	−0.002	−1.19	−0.004	−1.34	−0.001	−0.40
BIG10	−0.001	−1.18
SIZE	−0.000	−0.87	−0.001*	−1.94	0.000	0.34
LEV	0.028***	5.52	0.025***	3.52	0.031***	4.37
BTM	−0.046***	−21.35	−0.053***	−16.45	−0.038***	−13.14
ROE	−0.057***	−15.85	−0.051***	−11.02	−0.065***	−11.34
TURNOVER	0.007***	6.12	0.004***	2.90	0.010***	6.48
INV	0.021***	5.90	0.023***	4.58	0.021***	3.99
LISTAGE	0.001***	12.59	0.001***	10.65	0.001***	6.53
STATE	−0.010***	−10.78	−0.011***	−8.81	−0.009***	−6.60
INTERCEPT	0.082***	7.82	0.111***	6.70	0.072***	4.58
YEAR	YES		YES		YES
IND	YES		YES		YES
AND	YES		YES		YES
Observations	29450		15303		14147
Adj R^2	0.092		0.099		0.086
F	33.548		20.465		21.174
Chow test for two subsamples			153.34***(0.000)

Note: Table 5 reports the robustness tests using alternative measurement of earnings management. * p < 10%; ** p < 5%; *** p < 1% (two-tailed). All reported t statistics are based on standard errors adjusted for Huber-White (White, 1980).

Source: Authors.

4.4.2. Robustness tests using to financial misstatements to measure audit quality

To evaluate the robustness of our results, in Table 6, financial misstatements are adopted as another surrogate to measure audit quality (Chen et al., 2020; Guan et al., 2016). We define financial misstatements as ‘equalling to one if firms restate the income in future years and zero otherwise’ (MIS_DUM). Results in Table 6 show that firms who choose signing auditors with communist membership are less likely to conduct financial misstatements. Moreover, the positive relationship between AR_CPC and MIS_DUM is more pronounced in small audit firms. Collectively, these results lend further support to our findings.

Table 6. Robustness test using financial misstatements.

Variables	MIS_DUM
			BIG10 = 0		BIG10 = 1
	(1)		(2)		(3)
	Coefficient	Z value	Coefficient	Z value	Coefficient	Z value
AR_CPC	−0.172*	−1.93	−0.186*	−1.77	−0.026	−0.15
EDU	−0.131	−1.21	−0.047	−0.38	−0.389*	−1.67
FEMALE	−0.224***	−2.61	−0.062	−0.60	−0.594***	−3.70
AR_EXPERT	0.062	0.55	0.167	1.19	−0.349	−1.44
AF_ EXPERT	−0.227**	−2.37	−0.069	−0.47	−0.070	−0.45
BIG10	−0.265***	−3.54
SIZE	−0.027	−0.77	−0.030	−0.67	−0.027	−0.45
LEV	0.739**	2.40	0.686*	1.91	0.942	1.47
BTM	−0.037	−0.23	−0.169	−0.86	0.274	1.02
ROE	−0.544***	−4.74	−0.326**	−2.32	−1.016***	−5.41
TURNOVER	−0.317***	−4.04	−0.308***	−3.23	−0.314**	−2.27
INV	−0.545**	−2.42	−0.355	−1.28	−1.121***	−2.77
LISTAGE	0.030***	4.98	0.030***	3.81	0.029***	2.95
STATE	−0.087	−1.44	−0.110	−1.53	−0.138	−1.19
INTERCEPT	−1.884**	−2.47	−1.990**	−2.19	−3.513**	−2.39
YEAR	YES		YES		YES
IND	YES		YES		YES
AND	YES		YES		YES
Observations	29450		15303		14147
Pseudo R2	0.098		0.086		0.095
Chi2	1199.690		635.387		429.665
Chow test for two subsamples			207.17***(0.000)

Note: Table 6 reports the robustness tests using alternative measurement of audit quality (Financial misstatements). * p < 10%; ** p < 5%; *** p < 1% (two-tailed). All reported t statistics are based on standard errors adjusted for Huber-White (White, 1980).

Source: Authors.

4.4.3. Robustness tests using loss avoidance to measure audit quality

Prior literature has suggested that firms have motivations to systematically manage earnings to avoid losses (Brown & Caylor, 2005; Burgstahler & Dichev, 1997). In particularly, Chinese listed firms may have strong loss avoidance incentives for regulatory concerns such as initial public offerings, delisting thresholds, relisting applications, stock option executions, and seasoned equity offerings (Chen et al., 2001; Jiang & Wang, 2008). According to the relevant literature (Chen et al., 2020; Gul et al., 2013), we proxy loss avoidance with a small-profit dummy variable (SP), which is measured as ‘equalling to one if a firm reports ROA between 0 and 2% in a given year and zero otherwise’. In Table 7, we perform robustness tests to investigate whether our findings in Tables 3 and 4 are still stand for. The empirical results shown in Table 7 are consistent with our hypothesis, and thereby provide further support to our results.

Table 7. Robustness test using loss avoidance.

Variables	SP
			BIG10 = 0		BIG10 = 1
	(1)		(2)		(3)
	Coefficient	Z value	Coefficient	Z value	Coefficient	Z value
AR_CPC	−0.110**	−2.26	−0.161**	−2.42	−0.002	−0.03
EDU	−0.105*	−1.69	−0.037	−0.44	−0.142	−1.49
FEMALE	−0.135***	−3.04	−0.164***	−2.71	−0.130*	−1.95
AR_EXPERT	0.044	0.66	0.218**	2.40	−0.091	−0.85
AF_ EXPERT	0.021	0.44	−0.061	−0.55	0.103	1.48
BIG10	−0.082**	−2.06
SIZE	−0.450***	−22.50	−0.545***	−18.53	−0.394***	−14.17
LEV	1.135***	6.69	0.784***	3.39	1.771***	6.83
BTM	3.432***	37.68	3.587***	27.51	3.328***	25.64
ROE	0.059	1.15	0.361***	5.67	−0.355***	−3.97
TURNOVER	−0.455***	−10.96	−0.463***	−8.25	−0.401***	−6.46
INV	1.001***	8.44	0.817***	5.12	1.316***	7.03
LISTAGE	0.025***	9.10	0.033***	8.09	0.021***	5.32
STATE	0.185***	5.49	0.206***	4.61	0.155***	2.91
INTERCEPT	6.844***	16.60	8.420***	14.00	5.631***	9.07
YEAR	YES		YES		YES
IND	YES		YES		YES
AND	YES		YES		YES
Observations	29450		15303		14147
Pseudo R2	0.086		0.086		0.094
Chi2	2539.874		1313.880		1323.081
Chow test for two subsamples			236.18***(0.000)

Note: Table 7 reports the robustness tests using alternative measurement of audit quality (Loss avoidance). * p < 10%; ** p < 5%; *** p < 1% (two-tailed). All reported t statistics are based on standard errors adjusted for Huber-White (White, 1980).

Source: Authors.

4.4.4. Robustness tests using other measurement of signing auditors’ communist party membership

In the main tests, we use the average value of signing auditors’ party status to run the regression. As a robustness test, we calculate the minimum and maximum value of the signing auditors’ party membership to re-run the model (1) (AR_RC_1, AR_RC_2). Results in Tables 8 and 9 are similar to our main Tables 3 and 4, indicating that our findings are still valid after using other proxies for signing auditors’ communist party membership.

Table 8. Robustness test using other measures of signing auditors’ communist party membership.

Variables	|DA|
			BIG10 = 0		BIG10 = 1
	(1)		(2)		(3)
	Coefficient	T value	Coefficient	T value	Coefficient	T value
AR_CPC_1	−0.005**	−2.56	−0.005*	−1.92	−0.003	−1.27
EDU	−0.003	−1.34	−0.003	−1.09	−0.002	−0.83
FEMALE	−0.003**	−2.09	−0.005**	−2.42	−0.001	−0.72
AR_EXPERT	−0.001	−0.24	0.001	0.17	−0.002	−0.52
AF_ EXPERT	0.001	0.39	−0.000	−0.12	0.002	0.99
BIG10	−0.003**	−2.22
SIZE	−0.002***	−3.13	−0.003***	−3.08	−0.001	−1.46
LEV	0.039***	5.98	0.037***	3.91	0.041***	4.63
BTM	−0.054***	−19.97	−0.068***	−16.73	−0.040***	−10.89
ROE	−0.054***	−12.81	−0.048***	−8.91	−0.060***	−9.10
TURNOVER	0.010***	7.53	0.007***	3.60	0.014***	7.60
INV	0.041***	8.35	0.042***	6.19	0.042***	5.72
LISTAGE	0.001***	11.92	0.002***	10.18	0.001***	6.28
STATE	−0.013***	−11.38	−0.014***	−8.84	−0.012***	−7.49
INTERCEPT	0.130***	10.22	0.165***	8.02	0.114***	6.01
YEAR	YES		YES		YES
IND	YES		YES		YES
AND	YES		YES		YES
Observations	29450		15303		14147
Adj R^2	0.099		0.107		0.087
F	37.710		25.033		22.018
Chow test for two subsamples			123.20***(0.000)

Note: Table 8 reports the robustness tests using alternative measurement of signing auditors’ communist party membership. * p < 10%; ** p < 5%; *** p < 1% (two-tailed). All reported t statistics are based on standard errors adjusted for Huber-White (White, 1980).

Source: Authors.

Table 9. Robustness test using other measures of signing auditors’ communist party membership.

Variables	|DA|
			BIG10 = 0		BIG10 = 1
	(1)		(2)		(3)
	Coefficient	T value	Coefficient	T value	Coefficient	T value
AR_CPC_2	−0.002**	−2.22	−0.002*	−1.72	−0.001	−0.83
EDU	−0.003	−1.28	−0.003	−0.98	−0.002	−0.86
FEMALE	−0.003**	−2.10	−0.005**	−2.41	−0.001	−0.77
AR_EXPERT	−0.001	−0.24	0.000	0.13	−0.002	−0.50
AF_ EXPERT	0.001	0.37	−0.000	−0.10	0.002	0.97
BIG10	−0.004**	−2.29
SIZE	−0.002***	−3.80	−0.004***	−3.68	−0.001*	−1.82
LEV	0.042***	6.21	0.040***	4.09	0.044***	4.78
BTM	−0.055***	−19.51	−0.068***	−16.31	−0.041***	−10.60
ROE	−0.048***	−10.63	−0.042***	−7.08	−0.057***	−8.00
TURNOVER	0.010***	7.10	0.007***	3.32	0.014***	7.35
INV	0.041***	8.07	0.042***	5.97	0.042***	5.57
LISTAGE	0.001***	12.19	0.002***	10.39	0.001***	6.41
STATE	−0.013***	−11.34	−0.015***	−8.86	−0.012***	−7.39
INTERCEPT	0.140***	10.78	0.181***	8.42	0.121***	6.22
YEAR	YES		YES		YES
IND	YES		YES		YES
AND	YES		YES		YES
Observations	29450		15303		14147
Adj R^2	0.094		0.102		0.083
F	35.793		23.839		20.971
Chow test for two subsamples			123.36***(0.000)

Note: Table 9 reports the robustness tests using alternative measurement of signing auditors’ communist party membership. * p < 10%; ** p < 5%; *** p < 1% (two-tailed). All reported t statistics are based on standard errors adjusted for Huber-White (White, 1980).

Source: Authors.

4.4.5. Endogeneity tests using the two-stage least squares (2SLS) approach

Considering our results may be affected by the potential endogeneity issue, we next adopt two-stage least squares (2SLS) approach to tackle with the potential endogeneity problem. We identify GMQ—the natural logarithm of the number of old revolutionary base areas in the province where firms are located, and other control variables in the first stage regression. As expected, GMQ is positively associated with AR_CPC. Then, we use the fitted value of AR_CPC obtained from the first-stage regression (AR_CPC*) as the independent variable in the second-stage regression. Result in Columns (3)-(8) of Table 10 are qualitatively similar to those in Tables 3 and 4 and validate that our main findings are still stand after controlling the potential effect of the endogeneity issue.

Table 10. Endogeneity tests using the 2SLS approach.

Variables	The First stage		The Second stage
					BIG10 = 0		BIG10 = 1
	Coefficient	T value	Coefficient	T value	Coefficient	T value	Coefficient	T value
GMQ	0.001**	2.32
AR_CPC*			−0.423***	−3.85	−0.487***	−3.01	−0.182	−1.22
EDU	0.170***	21.49	0.069***	3.68	0.079***	2.88	0.029	1.12
FEMALE	−0.005	−0.91	−0.005***	−3.49	−0.007***	−3.28	−0.003	−1.34
AR_EXPERT	−0.020***	−2.60	−0.009***	−2.82	−0.009*	−1.93	−0.006	−1.43
AF_ EXPERT	−0.004	−0.60	−0.001	−0.42	−0.002	−0.53	0.002	0.67
BIG10	−0.032***	−5.53	−0.017***	−4.44
SIZE	0.005**	2.46	0.000	0.45	−0.000	−0.38	−0.000	−0.16
LEV	−0.038*	−1.84	0.023***	3.02	0.019*	1.72	0.034***	3.32
BTM	−0.013	−1.26	−0.059***	−19.21	−0.073***	−16.13	−0.042***	−10.06
ROE	0.031***	3.67	−0.040***	−7.37	−0.033***	−4.40	−0.054***	−6.62
TURNOVER	−0.003	−0.64	0.009***	6.66	0.006***	3.02	0.014***	7.35
INV	0.016	1.13	0.048***	9.06	0.050***	6.87	0.046***	5.88
LISTAGE	0.001*	1.85	0.001***	11.83	0.002***	10.11	0.001***	5.55
STATE	−0.018***	−4.41	−0.021***	−8.88	−0.023***	−6.81	−0.015***	−4.73
INTERCEPT	0.266***	5.64	0.245***	7.43	0.294***	6.30	0.158***	3.82
YEAR	YES		YES		YES		YES
IND	YES		YES		YES		YES
AND	YES		YES		YES		YES
Observations	29450		29450		15303		14147
Adj R^2	0.073		0.099		0.108		0.089
F	28.578		36.192		22.434		21.887

Note: Table 10 reports the endogeneity tests using 2SLS approach. * p < 10%; ** p < 5%; *** p < 1% (two-tailed). All reported t statistics are based on standard errors adjusted for Huber-White (White, 1980).

Source: Authors.

4.4.6. Endogeneity tests using the heckman two-stage method and propensity score matching approach

Moreover, we adopt Heckman two-stage method to tackle with the issue of survivorship bias. We define the AR_CPC_DUM as ‘equalling to one if any of signing auditors are communist party membership and zero otherwise’. The results shown in Table 11, taken together, confirm that after controlling the potential self-selection bias, our results are still valid.

Table 11. Endogeneity tests using the Heckman two-stage method.

Variables	The First stage		The Second stage
	AR_CPC_DUM				BIG10 = 0		BIG10 = 1
	Coefficient	Z value	Coefficient	T value	Coefficient	T value	Coefficient	T value
GMQ	0.007***	3.06
AR_CPC			−0.009**	−2.56	−0.010*	−1.95	−0.006	−1.23
EDU	−0.009	−0.39	−0.003**	−2.13	−0.005**	−2.22	−0.001	−0.71
FEMALE	0.654***	20.44	−0.001	−0.70	−0.002	−0.73	−0.001	−0.50
AR_EXPERT	−0.105***	−2.97	−0.001	−0.30	−0.000	−0.01	−0.002	−0.54
AF_ EXPERT	0.013	0.46	0.001	0.43	−0.002	−0.58	0.002	1.01
BIG10	−0.071***	−2.72	−0.004**	−2.38
SIZE	0.013	1.37	−0.002***	−2.97	−0.003***	−3.07	−0.001	−1.44
LEV	−0.027	−0.28	0.039***	5.97	0.035***	3.76	0.041***	4.63
BTM	−0.011	−0.25	−0.054***	−19.82	−0.068***	−16.74	−0.040***	−10.90
ROE	0.117***	2.95	−0.053***	−12.75	−0.048***	−8.89	−0.060***	−9.07
TURNOVER	−0.004	−0.20	0.010***	7.52	0.007***	3.49	0.014***	7.59
INV	0.066	1.04	0.041***	8.38	0.042***	6.25	0.042***	5.73
LISTAGE	0.003**	2.15	0.001***	11.96	0.002***	10.12	0.001***	6.31
STATE	−0.092***	−5.10	−0.013***	−11.43	−0.014***	−8.61	−0.012***	−7.50
IMR			0.002	1.50	0.002	1.18	0.001	0.74
INTERCEPT	−0.007	−0.03	0.132***	9.98	0.172***	8.54	0.115***	6.07
YEAR	YES		YES		YES		YES
IND	YES		YES		YES		YES
AND	YES		YES		YES		YES
Observations	29450		29450		15303		14147
Pseudo R^2 / Adj R^2	0.054		0.099		0.107		0.087
LR_Chi^2 / F	2000.223		35.754		22.480		21.692

Note: Table 11 reports the endogeneity tests using Heckman two-stage method. * p < 10%; ** p < 5%; *** p < 1% (two-tailed). All reported t statistics are based on standard errors adjusted for Huber-White (White, 1980).

Source: Authors.

Further, we use the propensity score matching approach. Specially, we match firms that select signing auditors with communist party status (the treated sample) to firms that select signing auditors without communist party status following one-to-one non-repeated matching principle and using ±5% as the calliper of the propensity score. We finally obtain a sample of 21,574 firm-year observations including 10,787 treated firms and 10,787matching firms after conducting the propensity score matching process. The results, show in Table 12, are similar to the results in main table, which provides further support to our hypothesis.

Table 12. Endogeneity tests using the using propensity score matching approach.

Variables	The First stage		The Second stage
	AR_CPC_DUM				BIG10 = 0		BIG10 = 1
	Coefficient	Z value	Coefficient	T value	Coefficient	T value	Coefficient	T value
GMQ	0.007***	3.06
AR_CPC			−0.004**	−2.42	−0.005*	−1.84	−0.002	−0.78
EDU	−0.009	−0.39	−0.004***	−2.67	−0.007***	−2.88	−0.002	−0.98
FEMALE	0.654***	20.44	−0.001	−0.44	−0.002	−0.59	−0.001	−0.23
AR_EXPERT	−0.105***	−2.97	−0.001	−0.47	0.003	0.82	−0.006*	−1.69
AF_ EXPERT	0.013	0.46	0.002	0.79	−0.004	−0.93	0.003	1.28
BIG10	−0.071***	−2.72	−0.003*	−1.79
SIZE	0.013	1.37	−0.002***	−3.00	−0.003***	−2.91	−0.002*	−1.75
LEV	−0.027	−0.28	0.046***	6.12	0.050***	4.41	0.041***	4.08
BTM	−0.011	−0.25	−0.055***	−17.73	−0.071***	−15.45	−0.040***	−9.34
ROE	0.117***	2.95	−0.049***	−9.52	−0.046***	−7.01	−0.050***	−6.23
TURNOVER	−0.004	−0.20	0.010***	6.82	0.007***	3.34	0.014***	6.63
INV	0.066	1.04	0.045***	7.90	0.043***	5.37	0.050***	6.02
LISTAGE	0.003**	2.15	0.001***	10.05	0.001***	7.81	0.001***	6.03
STATE	−0.092***	−5.10	−0.012***	−9.01	−0.013***	−6.70	−0.011***	−6.20
INTERCEPT	−0.007	−0.03	0.137***	9.08	0.170***	7.30	0.121***	5.66
YEAR	YES		YES		YES		YES
IND	YES		YES		YES		YES
AND	YES		YES		YES		YES
Observations	29450		21574		10776		10798
Pseudo R^2 / Adj R^2	0.054		0.100		0.111		0.087
LR_Chi^2 / F	2000.223		27.025		16.956		17.281

Note: Table 12 reports the endogeneity tests using PSM method. * p < 10%; ** p < 5%; *** p < 1% (two-tailed). All reported t statistics are based on standard errors adjusted for Huber-White (White, 1980).

Source: Authors.

4.4.7. Endogeneity tests controlling the firm fixed effects and signing auditors fixed effects

In addition, we employ firm fixed effects model to address the potential effect of unobservable client firms’ characteristics that may influence audit quality (Hou et al., 2020). The results, listed in columns (1)-(2) of Table 13, show that coefficient on AR_CPC is significantly negative. We also control for signing auditors fixed effects. In China, each audit report is signed by two or three signing auditors and thus we conduct our test using observations at the signing auditors-client-year level. Consequently, the sample is changed to 57,814. The columns (3)-(4) of Table 13 listed show that after controlling for the signing auditors fixed effects, our results still stand. Taken together, after controlling the potential effect of unobservable factors, our results are still valid.

Table 13. Robustness test using controlling firms fixed effects and signing auditors fixed effects.

Variables	|DA|
	Coefficient	T value	Coefficient	T value
AR_CPC	−0.004*	−1.96	−0.009*	−1.79
EDU	0.000	0.18	0.010	1.31
FEMALE	−0.001	−0.50	0.008	1.36
AR_EXPERT	−0.000	−0.01	−0.003	−1.39
AF_ EXPERT	0.002	1.14	0.002	0.98
BIG10	−0.004**	−2.28	−0.002	−1.35
SIZE	−0.002*	−1.94	−0.005***	−6.57
LEV	0.039***	5.19	0.058***	8.29
BTM	−0.060***	−15.74	−0.053***	−18.99
ROE	−0.041***	−15.66	−0.015***	−5.11
TURNOVER	0.016***	8.62	0.007***	5.52
INV	0.024***	4.05	0.055***	11.75
LISTAGE	0.008*	1.81	0.002***	14.68
STATE	−0.009***	−3.85	−0.016***	−12.62
INTERCEPT	0.136***	6.14	0.189***	13.06
YEAR	YES		YES
IND			YES
FIRM	YES
AND	YES		YES
AR			YES
Observations	29450		57814
Adj R^2	0.044		0.095
F	16.183		24.692

Note: Table 13 reports the robustness tests controlling firms fixed effects and signing auditors fixed effects. * p < 10%; ** p < 5%; *** p < 1% (two-tailed). All reported t statistics are based on standard errors adjusted for Huber-White (White, 1980).

Source: Authors.

4.4.8. Placebo test

Next, a placebo test is conducted to further tackle with the problems of unobservable factors. Theoretically, if our empirical results are caused by some omitted variables, these results are still stand after randomly disrupting. Hence, we re-define the independent variable by disorganising the signing auditors’ CPC membership randomly (AR_RC_CPC). Then, we re-run our model (1) by using the AR_RC_CPC as our independent variable. We conduct 100 times of randomisation and then draw the distribution diagram of t-value of the estimated coefficient on AR_RC_CPC. We can see from the Figure 1 that most t- values are around 0 and not significant and thereby indicates that our findings are unlikely to be driven by the non-observable factors.

Figure 1. Placebo tests.

Source: Authors.

4.4.9. Further tests: signing auditors’ communist party membership and audit fees

This article further examines the relationship between the party membership of signing auditors and audit quality. On one hand, in order to maintain their reputation and avoid audit failure, signing auditors with party status would be more diligent in the audit process, which demands great engagement effort. On the other hand, signing auditors with party membership are more competitive in the audit market because they provide higher audit service quality. Therefore, this article predicts that signing auditors who have party membership can obtain high audit fees. We use the logarithmic transformation of the audit fees (LNFEE) as the outcome variables. In addition, we refer to prior literature and calculate abnormal audit fees (ABLNFEE) (Choi et al., 2010; Krishnan et al., 2005). The Table 14 show the empirical results that party status enables signing auditors to obtain high audit fees, which provides further supports to our hypotheses.

Table 14. Further tests: signing auditors’ communist party membership and audit fees.

variables	LNFEE		ABLNFEE
	Coefficient	T value	Coefficient	T value
AR_CPC	0.024***	3.05	0.028***	3.51
EDU	0.003	0.28	0.005	0.43
FEMALE	0.028***	3.79	0.033***	4.53
AR_EXPERT	−0.011	−0.96	−0.019*	−1.69
AF_ EXPERT	0.015	1.50	0.014	1.46
BIG10	0.021**	2.49	0.020**	2.39
SIZE	0.391***	109.23	−0.028***	−7.79
LEV	−0.078**	−2.36	0.274***	8.39
BTM	−0.274***	−17.59	−0.255***	−16.43
ROE	−0.207***	−14.80	−0.081***	−5.90
TURNOVER	0.088***	14.07	0.089***	14.54
INV	−0.069***	−3.47	−0.082***	−4.20
LISTAGE	0.009***	18.22	0.002***	3.46
STATE	−0.080***	−13.98	0.000	0.03
INTERCEPT	4.994***	67.94	0.619***	8.48
YEAR	YES		YES
IND	YES		YES
AND	YES		YES
Observations	28349		28349
Adj R^2	0.707		0.169
F	654.065		45.558

Note: Table 14 reports the effect of signing auditors’ communist party membership on audit fees. * p < 10%; ** p < 5%; *** p < 1% (two-tailed). All reported t statistics are based on standard errors adjusted for Huber-White (White, 1980).

Source: Authors.

4.4.10. Further tests: considering the effect of industry expertise and client importance

Auditors’ industry expertise is an important influencing factors of audit quality and thus we examine whether auditors’ industry expertise serves as a mitigating role. Prior literature shows the evidence that auditors with industry expertise are capable to provide high audit quality because these signing auditors have high level of professional competence and audit independence (Chi & Chin, 2011; Chin & Chi, 2009). Hence, we argue that auditors’ industry expertise attenuate the positive effect of signing auditors’ communist party membership on audit quality. The results are shown in Table 15 suggest that the positive relationship between signing auditors’ party status and audit quality is more pronounced in non-industry expertise subsample, which validates our conjecture.

Table 15. Further tests considering the effect of auditors’ industry expertise.

Variables	AR_EXPERT = 0		AR_EXPERT = 1		AF_EXPERT = 0		AF_EXPERT = 1
	Coefficient	T value	Coefficient	T value	Coefficient	T value	Coefficient	T value
AR_CPC	−0.004***	−2.92	−0.000	−0.04	−0.005***	−3.04	−0.000	−0.01
EDU	−0.003	−1.41	−0.001	−0.06	−0.005**	−2.28	0.008*	1.86
FEMALE	−0.003**	−2.18	0.002	0.29	−0.004**	−2.36	−0.000	−0.03
AR_EXPERT					0.000	0.03	−0.003	−0.66
AF_ EXPERT	0.001	0.70	−0.006	−1.20
BIG10	−0.003**	−2.01	−0.008	−1.39	−0.004**	−2.37	−0.003	−0.70
SIZE	−0.001**	−2.13	−0.011***	−3.77	−0.002***	−2.95	−0.001	−0.73
LEV	0.039***	5.84	0.044	1.64	0.029***	4.18	0.077***	4.89
BTM	−0.055***	−19.53	−0.043***	−3.36	−0.055***	−17.82	−0.051***	−8.49
ROE	−0.054***	−12.39	−0.045***	−3.17	−0.053***	−11.68	−0.058***	−5.12
TURNOVER	0.010***	7.38	0.011	1.63	0.010***	6.27	0.014***	4.73
INV	0.041***	8.10	0.043*	1.93	0.038***	6.95	0.053***	4.61
LISTAGE	0.001***	10.92	0.002***	4.50	0.001***	10.70	0.001***	5.28
STATE	−0.013***	−11.14	−0.012**	−2.37	−0.013***	−9.68	−0.015***	−6.18
INTERCEPT	0.117***	8.60	0.337***	5.71	0.135***	9.01	0.124***	4.18
YEAR	YES		YES		YES		YES
IND	YES		YES		YES		YES
AND	YES		YES		YES		YES
Observations	27476		1974		23288		6162
Adj R^2	0.099		0.105		0.101		0.092
F	34.396		3.974		30.196		8.962

Note: Table 15 reports the subsample tests based on the auditors’ industry specialists. * p < 10%; ** p < 5%; *** p < 1% (two-tailed). All reported t statistics are based on standard errors adjusted for Huber-White (White, 1980).

Source: Authors.

Furthermore, relevant studies suggest that when facing important client firms, auditors have the motivation to compromise with mangers in order to retain clients firms (Chen et al., 2010). It can be seen that the economic dependence between auditors and client firms would reduce the audit quality (Chi et al., 2012). According to the previous analysis, this article argues that signing auditors with party membership are more likely to adhere to professional ethics and maintain independence and prudence in the audit process. Then, in the face of important clients, signing auditors with party status have less incentives to lose their independence for sake of their own economic interests. Therefore, this article tests the impact of client importance on the relationship between signing auditors with party membership and audit quality. We define the client importance as ‘the natural logarithm of the assets of a particular client scaled by the sum of the natural logarithm of the assets of all clients of signing auditors’ (CI_AR) and ‘the natural logarithm of the assets of a particular client scaled by the sum of the natural logarithm of the assets of all clients of audit firms’ (CI_AF). Then, we partition the full sample into the high CI_IA (CI_AF) subsample and the low CI_IA (CI_AF) subsample. As shown in Table 16, the positive association between AR_CPC and audit quality is more pronounced in high clients importance subsample, further supporting our hypotheses.

Table 16. Further tests considering the effect of client importance.

Variables	CIDUM = 1		CIDUM = 0		CFDUM = 1		CFDUM = 0
	Coefficient	T value	Coefficient	T value	Coefficient	T value	Coefficient	T value
AR_CPC	−0.005**	−2.18	−0.002	−1.06	−0.004*	−1.79	−0.003	−1.61
EDU	−0.005**	−2.06	0.001	0.50	−0.005*	−1.83	−0.000	−0.13
FEMALE	−0.004**	−2.25	−0.001	−0.64	−0.003	−1.44	−0.003	−1.42
AR_EXPERT	−0.001	−0.27	−0.000	−0.04	−0.001	−0.17	−0.002	−0.68
AF_ EXPERT	−0.001	−0.30	0.001	0.54	−0.000	−0.01	0.001	0.62
BIG10	−0.003	−1.30	−0.006***	−3.06	−0.002	−0.80	−0.004	−1.47
SIZE	−0.002***	−2.79	−0.002	−1.63	−0.002**	−2.33	−0.002**	−2.52
LEV	0.039***	4.60	0.038***	3.91	0.029***	3.29	0.053***	5.73
BTM	−0.049***	−12.95	−0.058***	−14.56	−0.055 ***	−14.03	−0.053***	−13.73
ROE	−0.059***	−10.48	−0.049***	−7.69	−0.057***	−9.83	−0.049***	−8.13
TURNOVER	0.009***	5.03	0.012***	5.72	0.007***	3.81	0.014***	7.39
INV	0.043***	6.41	0.041***	5.52	0.038***	5.49	0.042***	5.71
LISTAGE	0.001***	8.36	0.001***	8.36	0.001***	9.67	0.001***	7.38
STATE	−0.014***	−8.78	−0.013***	−7.84	−0.012***	−7.18	−0.016***	−9.41
INTERCEPT	0.133***	7.65	0.126***	6.20	0.120***	6.61	0.148***	7.60
YEAR	YES		YES		YES		YES
IND	YES		YES		YES		YES
AND	YES		YES		YES		YES
Observations	14636		14814		14636		14814
Adj R^2	0.112		0.088		0.102		0.101
F	23.601		18.347		20.536		23.742

Note: Table 16 reports the subsample tests based on the auditors’ client importance. * p < 10%; ** p < 5%; *** p < 1% (two-tailed). All reported t statistics are based on standard errors adjusted for Huber-White (White, 1980).

Source: Authors.

5. Conclusions

Audit research has begun to explore the audit quality variation at signing auditors level. In this article, we examine whether signing auditors’ communist party status influences audit quality using a sample of Chinese listed firms from the year 2001 to 2019. Consistent with theoretical analysis, the empirical results show that signing auditors who have communist party membership are significantly negatively associated with audit quality, suggesting that signing auditors with communist party membership provide high quality of audits. Moreover, the relationship between signing auditors with party status and audit quality is more pronounced in small audit firms. We also find that firms who select signing auditors with CPC membership have lower likelihood of financial misstatements and loss avoidance. In addition, the results are still valid after adopting a series of tests to tackle with potential endogeneity issue. Further analyses illustrate that signing auditors with party status can earn audit fee premium. Auditors’ industry expertise and client importance mitigate the positive relationship between signing auditors with CPC membership and audit quality. This study enriches the existing literature about the economic consequence of individuals’ party status and provides new evidence about the influencing factors of audit quality.

In addition to the theoretical contributions listed in the ‘1. Introduction’ part, this article may also engender some policy implications. Our study analyses whether and how signing auditors with party status have an influence on audit quality and finds that party membership serves as an important factor that influences signing auditors’ behaviours and then audit quality. In fact, the CICPA has promulgated a series of policies to strengthen the party construction in the audit industry and encourage signing auditors to join the CPC. Hence, audit firms can take signing auditors’ CPC membership into consideration when evaluating the professional competence of signing auditors. Meanwhile, this observable attributes can be an important consideration when audit firms recruit and train signing auditors. Moreover, firms, as the demand of audit service, should not only consider the audit firm characteristics, but also take into account characteristics of signing auditors when choosing external auditors. Besides, it is important for investors to consider the signing auditors characteristics when making decisions. Furthermore, our findings suggest that audit firm size mitigate the positive association between signing auditors’ party status and audit quality. Hence, since the audit firm size has an important role in the audit market and thereby the relevant departments should encourage the audit firms to become bigger and stronger.

Disclosure statement

No potential conflict of interest was reported by the authors

Additional information

Funding

This study is supported by National Natural Science Foundation of China (the approval numbers: NSFC-72102119) and Department of Education of Guizhou Province Project ([2021]123)

Appendix A. Sample selection

 

Table

The initial sample		42,110
Eliminate firms pertaining to the banking, insurance, and other financial industries		(1,231)
Eliminate firms whose listed age below one year		(3,080)
Eliminate observations without sufficient data to calculate discretionary accruals		(4,060)
Eliminate observations with missing data on signing auditors		(3,467)
Eliminate observations with missing control variables		(822)
Available firm-year observations		29,450
Unique firms		3,299

Source: Authors.

Appendix B. Variable definition

 

Table

Variables	Definitions
|DA|	The absolute value of discretionary accruals based on the modified Jones model;
AR_CPC	Signing auditors’ party status, equalling 1 if signing auditors have communist party status and 0 otherwise; We take the average value of signing auditors’ party status;
EDU	Signing auditors’ educational level, equalling 1 if signing auditors have obtained a graduate degree or above and 0 otherwise; We take the average value of signing auditors’ educational level;
FEMALE	Signing auditors’ gender, equalling 1 if one or more signing auditors are women and 0 otherwise; We take the average value of signing auditors’ gender;
AR_EXPERT	An indicator variable for signing auditors industry specialisation, equalling 1 if any of signing auditors have the largest market share based on the number of clients in each two-digit CSRC industry;
AF_ EXPERT	An indicator variable for audit firm industry specialisation, equalling 1 if the audit firm has the largest market share based on the number of clients in each two-digit CSRC industry;
BIG10	An indicator variable, equalling 1 if the audit firm is one of the top 10 auditors or their affiliated firms the CICPA annual ranking and 0 otherwise;
SIZE	Firm size, measured as the natural logarithm of total assets;
LEV	Financial leverage, measured as the ratio of long-term liabilities to total assets;
BTM	Book to market value；
ROE	Return on equity, measured as the net profit scaled by total equity;
TURNOVER	Total sales divided by average total assets;
INV	The inventory deflated by total assets;
LISTAGE	The number of years since a firm’s IPO;
STATE	A dummy variable, equalling 1 if a firm’s ultimate owner is a local or central government or state-owned enterprise and 0 otherwise.

Source: Authors.