State Control,Related-Party Transactions and Audit Reporting: Evidence From Key Audit Matters

Abstract

We examine whether political forces in Chinese State-Owned Enterprises (SOEs) influence audit reporting, specifically the disclosure of Key Audit Matters (KAMs). We test two competing predictions that offer alternative explanations for the relation between SOEs and KAMs disclosure. Using a sample of Chinese listed firms and controlling for related determinants of KAMs reporting, we document that, compared to non-SOEs, SOEs have fewer abnormal KAMs (relative to their industry peers). SOEs are also more likely to avoid the disclosure of expected KAMs, especially in the subject areas of inventory, revenue and related party transactions (RPTs). Taken together, results suggest that SOEs have strong political motives and power to obscure transparency and withhold potentially costly news. In line with this conjecture, we show that the aforementioned effects are more pronounced when SOEs have more concentrated state ownership, operate in industries of strategic importance to the state, or are involved in tunneling RPTs. Supplementary analysis indicates that SOEs have less extensive KAMs disclosures and auditor responses, which moreover are less risk-oriented. Overall, our study provides new evidence on how state control and related institutional factors affect audit practices.

Keywords

key audit matters SOEs political control related party transactions G39 M41 M42

Introduction

We study how political forces, namely state control, shape audit reporting, namely the disclosures of Key Audit Matters (KAMs) in expanded audit reports. KAMs disclosure is a landmark change in audit regulation worldwide. Auditors are required to move beyond the traditional pass/fail model and to communicate information, known as KAMs, about areas of high-risk of material misstatement, areas involving significant auditor and management judgment (including highly uncertain estimates) as well as information about significant events or transactions that occurred during the period (Minutti-Meza, 2021, p. 556).

The consequences of the adoption of KAMs reporting in multiple jurisdictions constitute an active area of research currently yielding mixed evidence (see review by Minutti-Meza, 2021). Thus, understanding why auditors report the KAMs that they do as well as documenting the institutional factors that influence KAMs reporting is inherently interesting and potentially important. Accordingly, in the current study, we examine the determination of KAMs reporting and focus on the role of state control. To our knowledge, prior literature does not provide insights on this issue of expanded audit reporting. Yet, state ownership plays an increasingly significant role in today’s global economy. In 2021, there were 100 State-Owned Enterprises (SOEs) in Fortune Global 500 (FG500), accounting for approximately 22% of the total revenues of all FG500 corporations (FORTUNE, 2021). Moreover, state ownership often concentrates in strategically important industries on which large parts of the broader economy depend (OECD, 2014).

We address the following questions in this paper. Does the disclosure of KAMs in audit reports vary between SOEs and non-SOEs? Does the level of state control influence KAMs disclosure by SOEs? Do related party transactions (RPTs), and in particular tunneling, influence KAMs reporting by SOEs? Our investigation is motivated by prior research that highlights the role of political institutions in shaping financial reporting and auditing (e.g., Bushman & Piotroski, 2006; Hope et al., 2020; Wang et al., 2008). It is also motivated by an emerging literature examining the new generation of expanded audit reports (Minutti-Meza, 2021).

China offers a natural setting for a study of the above questions for the following two reasons. First, even after four decades of market-oriented reforms, many of China’s institutions remain relatively underdeveloped whilst its capital market is still under strong government influence (Lennox & Wu, 2022). In 2019, there were still more than 130,000 SOEs in China, with total assets of $28,182 billion and total revenues of $8274 billion (NPC, 2020). Second, contrary to the U.K. and the U.S., China has several local institutional features that relate to state dominance and that are particularly relevant to our research questions. These are high ownership concentration, pervasive political interference, an opaque and potentially biased corporate information environment, a relatively less independent audit market where auditors have modest regulatory, litigation and reputational incentives, and frequent RPTs (Lennox & Wu, 2022; Piotroski et al., 2015; Piotroski & Wong, 2012; Wong, 2016). Thus, we exploit China’s unique setting to understand how state control and other related institutional factors shape the disclosure of KAMs.

Conceptually, there are two contradictory predictions about the impact of state control on KAMs disclosure. By their very nature, SOEs are distinctive because the state is a political actor. SOEs therefore have different objectives and incentives (Wong, 2016). Theory and prior evidence suggest that, relative to non-SOEs, Chinese SOEs are more likely to engage in earnings manipulation and aggressive accounting practices (e.g., Aharony et al., 2000; Chen et al., 2010). To the extent that increased earnings management and reduced conservative accounting suggest higher financial reporting risks, we expect more extensive KAMs disclosures in SOEs’ than in non-SOEs’ audit reports. However, there is a counter-argument. Relative to non-SOEs, SOEs are more inclined to suppress the flow of bad news in order to avoid financial, and ultimately political costs (Piotroski et al., 2015). KAMs disclosure is a type of risk disclosure regarding the reliability of the financial statements and the operations of the firm (Chang et al., 2024; Klevak et al., 2023), with potentially adverse financial consequences for companies, such as higher cost of debt (e.g., Porumb et al., 2021) and negative investor sentiment (e.g., Christensen et al., 2014). Also, Chinese SOEs, and in particular the government, has the power to impose costs on auditors, such as influencing their licensing (e.g., Chen et al., 2000), whereas auditors in China have relatively weak incentives for delivering high-quality audits (Lennox & Wu, 2022; Piotroski & Wong, 2012). Thus, given this context, to the extent that more KAMs-related information conveys potentially costly news with adverse political externalities, we expect fewer KAMs disclosures in SOEs’ than in non-SOEs’ audit reports. We evaluate which of these incentives and powers dominate by examining whether KAMs disclosure varies systematically between SOEs and non-SOEs.

Based on a sample of Chinese listed firms during 2017–2020 with hand-collected KAMs data, we report a series of results that are new to the literature. In our baseline analysis, we employ two outcome variables for KAMs disclosure, namely the abnormal number of KAMs relative to industry peers and expected KAMs estimated using the prediction models developed by Burke et al. (2023). Regarding the first measure, we document that, after controlling for other determinants of KAMs disclosure, Chinese SOEs disclose 4.3% fewer abnormal KAMs than non-SOEs. As for expected KAMs, we find that Chinese SOEs display a 7.2% higher probability of avoiding the disclosure of KAMs in subject areas where they are predicted, especially in areas related to inventory, revenue and RPTs. Together, these results are consistent with the conjecture that SOEs have strong political motives and power to obfuscate transparency and withhold potentially unfavorable news. We find that our baseline findings are not driven by SOEs’ greater government support or lower firm risks. Also, our results are robust to several tests addressing endogeneity concerns.

To corroborate our baseline inferences, we perform a series of cross-sectional analyses. We first focus on SOEs’ ability to insert political influence and posit that SOEs under stronger state control have a higher ability to exercise political pressure (Jin et al., 2022). Consistent with this, we show that the documented negative (positive) association between SOEs and the number of abnormal KAMs (the likelihood of missed KAMs) is meaningfully stronger for SOEs with higher state ownership concentration and SOEs in strategic or public welfare industries that are subject to tighter state scrutiny. For example, for SOEs with strategic importance to the state, the abnormal number of KAMs drops by an incremental 4.4%. For SOEs with high ownership concentration, the likelihood of avoiding the disclosure of expected KAMs increases by an additional 4.2%. These results reinforce our primary finding on the role of SOE political forces in shaping KAMs disclosure.

Next, we perform cross-sectional analyses based on SOEs’ incentives to insert political influence. We expect SOEs with tunneling RPTs (i.e., providing loans or guarantees to related parties) to have stronger incentives to exercise political control and suppress KAMs disclosure because tunneling RPTs are opportunistic wealth expropriation activities that benefit the state but damage other stakeholders (Jiang et al., 2010). Results show that SOEs engaged in tunneling transactions disclose even fewer KAMs related to RPTs. Specifically, SOEs with tunneling RPTs display an additional 2.8% reduction in the number of abnormal KAMs and an additional 4.2% increase in the likelihood of not reporting expected KAMs. Overall, rendering audit reports less informative to provide cover for potential wealth expropriation activities for the benefit of the state is consistent with SOEs having strong motives to suppress potentially costly information. The documented cross-sectional results also reveal that the average effect of state ownership on KAMs masks significant heterogeneity.

Lastly, in additional analysis, we investigate further whether the extent of KAMs disclosures varies between SOEs and non-SOEs by employing a range of variables related to the textual characteristics of KAMs descriptions and auditor responses to identified KAMs. This analysis reveals that, relative to non-SOEs, SOE KAMs descriptions (auditor responses) consist of 4.9% (7.0%) less words; SOE KAMs cite 4.2% fewer financial statement footnotes; and SOE auditors communicate less by 3.3% about their procedures when addressing identified KAMs as well as use 8.5% fewer risk-related words when responding to KAMs. Collectively, our findings point to the fact that state ownership is an important factor with negative consequences for KAMs reporting.

Our study contributes to the auditing literature, and in particular the emerging literature of expanded audit reporting, in two main ways. First, a recent wave of studies examines the informativeness and usefulness of KAMs disclosure. So far, these studies have produced mixed evidence.¹ On one hand, several archival studies using data from Western economies find little or no evidence that KAMs are incrementally informative to investors (e.g., Bedard et al., 2018; Burke et al., 2023; Gutierrez et al., 2018). On the other hand, based again on data from Western jurisdictions other studies document that KAMs have information content for both equity and debt market participants. For example, the switch to expanded audit reporting is found to be associated with higher earnings response coefficients (ERCs) (Reid et al., 2019) as well as with changes in firms’ loan contracting terms, especially for borrowers with a poor information environment (Porumb et al., 2021).

Using UK data, Lennox et al. (2023) investigate potential explanations for why investors do not find KAMs disclosures to be informative and show that investors are already informed about the vast majority of KAMS as these are available in other disclosures (e.g., conference calls) made by companies, prior to the release of expanded audit reports. However, Klevak et al. (2023) document a significantly negative association between short-term market returns and the extent of Critical Audit Matters (CAMs) disclosures in the US, even after controlling for other information discussed in the earnings conference call. Moreover, Lennox et al. (2023, p. 523) acknowledge that the UK’s information environment could limit the generalizability of their results to other countries. This may be particularly the case for China, where the broader corporate information environment is of lower quality and less transparent (e.g., Lennox & Wu, 2022). Thus, the “leakage” of KAMs information via other channels is less likely in settings like China. Also, even if some KAMs-related information is available before the release of the annual reports (e.g., via conference calls), auditors’ responses to identified KAMs, disclosed in expanded audit reports, is ex ante unknown and thus potentially ex-post informative. In line with this argument, Goh et al. (2023) show that auditors’ responses are associated with a stronger market reaction and a higher ERC.

Focusing on China, contrary to Liao et al. (2022) who do not find evidence of an association between the introduction of KAMs and the information value of the audit report, Goh et al. (2023) show that trading volume and ERCs increase significantly following the adoption of expanded audit reporting, especially for non-SOEs than SOEs. Similarly, Chang et al. (2024) explore the heterogeneity of KAMs reporting in Taiwan and document that more extensive client/risk-specific KAMs disclosures are associated with ERCs, suggesting that such disclosures are useful and informative.

Against this background of no compelling evidence regarding the consequences of KAMs reporting, our study represents an important step forward by attempting to understand its determinants and the forces that may shape it. In doing so, we document that the political forces underlying SOEs may impair KAMs reporting. This may help us understand the conditions under which expanded audit reports may indeed matter.

Second, a few studies examine the determinants of KAMs disclosure whilst Burke et al. (2023) develop prediction models for expected KAMs. These studies identify several firm-level determining factors, including operational complexity, past reporting issues, profitability, and the magnitude of accounts requiring high degrees of judgment (Burke et al., 2023; Sierra-García et al., 2019); as well as auditor-level factors, such as audit partner gender (Abdelfattah et al., 2020). We add to this line of research by documenting that state control and the related political institutions are important determinants of KAMs reporting. In doing so, we exploit the unique setting of China, where the state maintains strong control of the capital market and SOEs account for a significant portion of the market, firms’ information environment is less transparent, equity ownership is highly concentrated, audit market is less independent and auditors have weak incentives for delivering high-quality audits, and RPTs are pervasive.

Our paper also extends prior research on how political forces shape financial reporting and auditing. Several papers document that state ownership and political connections undermine information transparency and earnings quality (e.g., Bushman & Piotroski, 2006; Chen et al., 2020; Fan & Wong, 2002; Hope et al., 2020). In addition, prior research documents the effects of state ownership and political affiliations on audit pricing and auditor choice (e.g., Guedhami et al., 2009; Guedhami et al., 2014; Gul, 2006). In China, specifically, prior work suggests that SOEs have strong incentives to obscure information and pursue control over auditors. For example, Wang et al. (2008) document the propensity of SOEs to hire local smaller auditors, who are potentially more conducive to rendering financial statements less informative. Similarly, Chan et al. (2012) find that SOEs are more likely to secure favorable audit opinions. We extend research on the link between state control and auditing practices by documenting its adverse consequences for KAMs disclosure in expanded audit reports.

Our study is topical in light of the recent and continuous adoption of expanded audit reports across several jurisdictions worldwide. We provide users of audited financial statements and regulators with new insights into the determinants of information covered in expanded audit reporting. As discussed, prior work suggests that KAMs disclosure provides limited value to investors in well-developed economies, possibly because they already know about the risks before auditors disclose them (Lennox et al., 2023). Our findings suggest that, in less-developed economies, the informativeness and overall usefulness of KAMs disclosure may be compromised by strong political forces in the capital market.

Institutional Background: The Audit Profession and KAMs Disclosure in China

China’s auditing profession was established in 1980. It is overseen by the Chinese Institute of Certified Public Accountants (CICPA), under joint regulation by the China Securities Regulatory Commission (CSRC) and the Ministry of Finance (MOF). Until the early 1990s, most audit firms were affiliated with governments. However, the opening of stock exchanges in Shanghai and Shenzhen in 1990 and 1991, respectively, stimulated the development of the audit market. Due to investor demands for higher audit quality, the CICPA and the MOF launched a disaffiliation program in 1996 to separate audit firms from the government (Gul et al., 2009; Yang et al., 2001). Chinese audit firms now operate under relatively competitive market forces (Lennox & Wu, 2022).

The CICPA and the MOF are responsible for establishing auditing standards in China, including Auditing Standard No. 1504 for Certified Public Accountants of China - Communication on Key Audit Matters in Audit Reports, issued on 23 December 2016 (CICPA, 2016; MOF, 2016). This standard mandates KAMs disclosure in Chinese public firms’ expanded audit reports. Firms listed on the stock exchanges in both Mainland and Hong Kong China (i.e., A + H share firms) are required to adopt the standard after January 1, 2017. Firms listed only on stock exchanges in Mainland China (i.e., A share firms) are required to adopt the standard after January 1, 2018.

According to Auditing Standard No. 1504, KAMs are defined as those matters that, in the auditor’s professional judgment, are of the most significance in the audit of the financial statements of the current period (CICIPA, 2016, Article 7). In determining KAMs, the auditor shall consider areas of high risk of material misstatement, significant auditor judgments related to major managerial discretion estimates, and current material transactions or events (CICPA, 2016, Article 9). When responding to identified KAMs auditors should describe the audit procedures implemented as well as the audit resources and efforts expended to address the identified KAMs and related risks (BICPA, 2018). Auditor responses provide detailed insights about auditor professional judgment, thus enhancing the transparency of audit work and potentially the informational content of audit reports (Yan & Zhuang, 2021).

The primary objective of the new audit disclosures is to convey useful information to investors (Lennox et al., 2023). Moreover, the language used in KAMs often signals the potentially high financial reporting uncertainty and/or operating risks of a firm. For example, the 2019 auditor’s report of Hisense Visual Technology disclosed an RPT-related KAM as follows: “Due to the large number of related parties and the diverse types of related party transactions involved, there is a risk that Hisense Vision does not adequately disclose the related party relationships and their transactions in the notes to the financial statements. Therefore, we have identified the related party relationships and their transactions as a key audit matter.” Appendix OA1 (Online Appendix) provides more examples of KAMs descriptions and related auditor responses. KAMs should be communicated in a separate section within the audit report. If some matters result in a qualified audit opinion, those matters should not be communicated as KAMs, albeit the auditors may still consider other matters to be reported as KAMs (CICPA, 2016, Article 5).

Informal discussions with auditors in China (ranging from associates to senior managers) suggest that the identification of KAMs is initiated by senior auditors (i.e., partners and senior managers) at the planning stage. They provide a list of matters that may be considered as KAMs based on their professional judgment. Next, the audit team conducts audit procedures and collects evidence to address the most important matters on the list. During this process and prior to the disclosure of KAMs and related auditor responses in the report, auditors should discuss these issues with their clients.

Prior Literature and Hypothesis Development

In SOEs, the controlling shareholder is the government, which pursues multiple objectives other than value maximization, such as maintaining social stability and political power (Jin et al., 2022; Lin et al., 2020). Consequently, SOEs are not fully accountable for their performance and often depend on state support in case of financial difficulties, resulting in soft budget constraints and incentive problems. Indeed, Chairmen and CEOs of listed SOEs in China are appointed and terminated by the local/central government. Many of these key corporate executives are quasi-government officials (e.g., they have vice-minister ranking), enjoy fringe benefits of government officials and have strong aspirations to climb up the government hierarchy and pursue a political career (Wong, 2016). To improve their political performance and advance their careers, they are motivated to actively implement the state’s principles, policies and resolutions in SOEs (Jin et al., 2022). Moreover, executive compensation in SOEs is highly regulated and ineffective in incentivizing managers (Lennox & Wu, 2022; Lin et al., 2020). In a system like this, SOE managers are urged to solve social problems, such as unemployment and social instability, and ultimately to enhance their political capital (Lin et al., 2020; Wong, 2016). In sum, it is mainly politics rather than financial incentives that drive SOEs.

Driven by political aspirations, managers of SOEs have strong incentives to engage in earnings management, especially given the opaque pyramid structure of SOE corporate groups (Piotroski & Wong, 2012; Wong, 2016). Such incentives may, to some extent, be attributed to China’s bright-line profitability rules for share issuance and delisting (Jian & Wong, 2010). Specifically, central and local governments as the ultimate controlling owners of SOEs have strong incentives to help them maintain their listing status (Bai et al., 2005) and qualify for raising more funds, as such firms are better able to help relieve unemployment problems and enhance the infrastructure development of the ministries where the firms belong or of the regions where the firms operate (Li & Zhou, 2005). In line with these arguments, prior research documents higher accruals-based earnings manipulation (Aharony et al., 2000) and abnormal sales to related parties for SOEs than non-SOEs (Jian & Wong, 2010). Additionally, because of the soft budget constraint, lenders tend to be less concerned by SOEs’ than non-SOEs’ downside risk, resulting in weaker demand for conservative accounting for SOEs. In line with this, Chen et al. (2010) document less timely loss recognition by SOEs than non-SOEs. Taken together, to the extent that more earnings manipulation and aggressive accounting practices suggest less reliable and more uncertain accounting estimates and thus, higher financial reporting risks, we expect SOEs to have more KAMs disclosures than non-SOEs.

However, governments, politicians and SOE managers have incentives to avoid the disclosure of negative news about their activities because such transparency highlights inefficiencies and economic disparities, attracts unwanted scrutiny over policy failures and corruption, and ultimately, damages authority and career prospects (e.g., Piotroski et al., 2015; Rajan & Zingales, 2003). Likewise, theories of authoritarian politics argue that governments have to demonstrate strengths or at least reveal fewer weaknesses to maintain power (Tullock, 1987). In line with this reasoning, Piotroski et al. (2015) show that SOEs are more likely to suppress the flow of negative information to reduce the political costs of releasing bad news.

As mentioned, KAMs disclosure is potentially a type of unfavorable risk disclosure regarding the reliability of the financial statements and the operations of the company, and thus may have negative consequences for companies. Consistent with this view, Chang et al. (2024) report that more extensive KAMs disclosures related to the client’s risk assessment reduce ERCs, indicating that the market perceives such information as a warning of heightened risk. Similarly, Klevak et al. (2023) find that the extensiveness of CAMs reporting is positively associated with stock price volatility and analyst forecasts’ dispersion, implying again that KAMs signal increased perceived uncertainty. Moreover, Porumb et al. (2021) find that the number of KAMs is positively associated with loan spreads, suggesting that KAMs are indicative of risks priced by lenders whereas Burke et al. (2023) find that KAMs disclosed and not anticipated by the market induce negative cumulative abnormal returns (CAR). In addition to the aforementioned empirical evidence, experimental research (Christensen et al., 2014; Dennis et al., 2019; Rapley et al., 2021) also finds that KAMs raise investors’ attention to companies’ risks and restrain their willingness to invest. Kachelmeier et al. (2020) find that experimental participants have less confidence in accounts that are identified in the audit report as CAMs. Collectively, these studies indicate that KAMs contain potentially costly news. Thus, given the propensity of SOEs to suppress the flow of negative information, we expect them to disclose fewer KAMs than non-SOEs.

As discussed, auditors are responsible for initiating the identification of potential KAMs, collecting relevant evidence, communicating with the client, and ultimately determining which KAMs to disclose. Some might argue that many KAMs disclosures are boilerplate, suggesting that auditors can easily enhance their KAMs reporting without necessarily exercising additional effort or adversely affecting the auditor-client relationship. However, recent research using data from Taiwan documents that only 8% of a KAM’s content is identified as generic language, implying that KAMs may not necessarily be boilerplate (Chang et al., 2024). Further, regarding the incentives of auditors, we observe the following. First, China’s reliance on public enforcement to incentivize auditors has fundamental limitations because regulators face capacity constraints and are susceptible to political control (Lennox & Wu, 2022; Piotroski & Wong, 2012). Second, although auditor litigation threat has increased recently, it still remains low; civil lawsuits are relatively rare in China mainly because plaintiffs face significant obstacles to winning cases against auditors (Lennox & Wu, 2022). Third, Chinese auditors can suffer significant losses from a reputational impairment, albeit reputational incentives are very much dependent on the effectiveness of regulatory oversight (Lennox & Wu, 2022).

More specifically, in the case of SOEs, the government may threaten auditors by asking its SOEs not to use the auditors’ services or by influencing the administration of qualifying exams, the licensing of audit firms, and the regulation of their day-to-day operations (Chen et al., 2000; Piotroski & Wong, 2012). These are potentially costly outcomes for SOE auditors. Taken together, the preceding discussion suggests a less independent audit market in China, where the state has the power to influence auditing practices and auditors have modest regulatory, litigation and reputational incentives. In line with these arguments, prior research provides evidence that SOEs are more inclined to hire local smaller auditors, driven to some extent by collusion incentives (Wang et al., 2008). Similarly, SOEs are more likely to secure favorable audit opinions, especially if they are politically-connected (Chan et al., 2006, 2012; He et al., 2017). Moreover, Lennox et al. (2016) report more downward than upward audit adjustments of accruals but only for non-SOEs.²

In sum, the role of SOEs in determining the disclosure of KAMs in expanded audit reports remains an empirical question that hinges on which incentives and powers dominate. We, largely for expositional convenience, state our hypothesis as follows:

Hypothesis 1

Compared to non-SOEs, SOEs have fewer and less extensive KAMs disclosures in the audit reports.

Research Design

Empirical Specification

To test our hypothesis on the relation between SOEs and KAMs disclosure, we employ the following specification for firm $i$ in year $t$ :

{K A M d i s c l o s u r e d e p v a r}_{i, t} = α_{0} + α_{1} {s o e}_{i, t} + α_{2} {s i z e}_{i, t} + α_{3} {l e v}_{i, t} + α_{4} {r o a}_{i, t} + α_{5} {l o s s}_{i, t} + α_{6} {r e v_g r o w}_{i, t} + α_{7} {c u r r e n t}_{i, t} + α_{8} {i n_a r}_{i, t} + α_{9} {a b s d a c c}_{i, t} + α_{10} {o c f}_{i, t} + α_{11} {s t d o c f}_{i, t} + α_{12} {a h}_{i, t} + α_{13} {b m}_{i, t} + α_{14} {i s s u e}_{i, t} + α_{15} {m a}_{i, t} + α_{16} {r e s t r u c t}_{i, t} + α_{17} {i m p a i r l o s s}_{i, t} + α_{18} {p p e_t o a}_{i, t} + α_{19} {l n a g e}_{i, t} + α_{20} {p d}_{i, t} + α_{21} {a u d i t o r c h a n g e}_{i, t} + α_{22} {a u d i t o r t e n u r e}_{i, t} + α_{23} {a o q u a l i f i e d}_{i, t} + α_{24} {r e s t a t e}_{i, t} + α_{25} {w e a k}_{i, t} + α_{26} {l n a r_w c}_{i, t} + α_{27} {d u a l}_{i, t} + α_{28} {l a r g e s t h o l d e r}_{i, t} + α_{29} {i n d e d i r e c t o r}_{i, t} + α_{30} {b o d m g t s t h o l d e r}_{i, t} + α_{31} {i i s t h o l d e r}_{i, t} + I n d u s t r y F E + Y e a r F E + A u d i t F i r m F E + ε_{i, t}

(1)

In equation (1) we use two dependent variables to perform our baseline analysis, namely $a b n_k a m_n u m b e r$ and $m i s s e d_k a m$ , which aim to capture abnormal and expected KAMs reporting, respectively. We define $a b n_k a m_n u m b e r$ as the difference between the total number of KAMs disclosed in the firm’s auditor report in year $t$ and the average number of KAMs of all firms in the same industry in year $t$ . This measure provides an industry-year-adjusted benchmark for KAMs disclosure and proxies for KAMs reporting that deviates from industry norms. With this dependent variable, we perform OLS regression.³

To construct missed_kam, we follow a two-step process. First, we manually classify all KAMs in our sample into seven subject areas that are commonly seen in Chinese firms’ audit reports, namely property, plant, and equipment (PPE); goodwill; inventory; receivables; revenues; RPTs; and other.⁴ This classification is consistent with the most commonly reported KAMs in China as shown by Zeng et al. (2021). For each firm-year, we generate an indicator variable for each subject area that equals to 1 if the firm has at least one KAM in this subject area, and 0 otherwise. This gives us, for each firm-year, seven subject area indicator variables, namely $k a m_p p e$ , $k a m_g o o d w i l l$ , $k a m_i n v e n t o r y$ , $k a m_r e c e i v a b l e s$ , $k a m_r e v e n u e$ , $k a m_r p t$ , $and k a m_o t h e r .$ Then, following Burke et al. (2023), we predict the probability of a firm having a KAM in each subject area in a year using the following model:

{k a m s u b j e c t a r e a d e p v a r}_{i, t, s} = α_{0} + α_{1} {s u b j e c t a r e a r e l a t e d a c c o u n t}_{i, t} + Σ C o n t r o l s + I n d u s t r y F E + Y e a r F E + A u d i t F i r m F E + ε_{i, t}

(2)

Equation (2) is estimated for firm $i$ in year $t$ for each KAM subject area. Specifically, we logit regress the indicator variable for each KAM subject area on the control variables of equation (1) as well as the subject area’s related financial statement account, assuming a higher likelihood of KAMs in certain subject areas when the related financial statement account is material.⁵

Next, as in Burke et al. (2023), we construct a firm-year aggregate variable for missed KAMs (i.e., $m i s s e d_k a m$ ). It is an indicator variable equal to 1 if the auditor’s report does not disclose a KAM in at least one of the seven subject areas, where equation (2) predicts the disclosure probability to be in the top quintile of the sample, and 0 otherwise. This $m i s s e d_k a m$ indicator variable is our second dependent variable in equation (1). It captures the non-disclosure of KAMs in subject areas where firm fundamentals and other firm-specific factors predict KAMs disclosure. When $m i s s e d_k a m$ is the dependent variable, we estimate equation (1) using logit regression. In supplementary analyses, we employ additional KAMs-related outcome variables (see Section “Supplementary Analysis”).

Our independent variable of interest is $s o e$ , which is an indicator equal to 1 if the firm is a state-owned enterprise (i.e., SOE), and 0 otherwise (Bortolotti & Faccio, 2009; Piotroski et al., 2015; Wei, 2021). When the dependent variable is $a b n_k a m_n u m b e r$ , the coefficient on $s o e$ is expected to be negative if SOEs have fewer KAMs relative to their industry peers, than non-SOEs. When the dependent variable is $m i s s e d_k a m$ , the coefficient on $s o e$ is predicted to be positive if SOEs are more likely to evade disclosing KAMs in subject areas that are expected.

We include an array of control variables following prior research on the determinants of KAMs (Burke et al., 2023; Sierra-García et al., 2019). Specifically, we control for firm characteristics related to performance, complexity and risk, including size ( $s i z e$ ), leverage ( $l e v$ ), return on assets ( $r o a$ ), profitability ( $l o s s$ ), revenue growth ( $r e v_g r o w$ ), current ratio ( $c u r r e n t$ ), inventory and accounts receivables ( $i n_a r$ ), discretionary accruals ( $a b s d a c c$ ), operating cash flows and its volatility ( $o c f$ and $s t d o c f$ ), dual-listing status in the mainland and Hong Kong China ( $a h$ ), book-to-market ratio ( $b m$ ), stock issuance ( $i s s u e$ ), M&A and restructuring activities ( $m a$ and $r e s t r u c t$ ), impairment loss ( $i m p a i r l o s s$ ), tangibility of operating assets ( $p p e_t o a$ ), firm age ( $l n a g e$ ), the probability of default ( $p d$ ), prior restatements ( $r e s t a t e$ ), and internal control weaknesses ( $w e a k$ ). We also control for auditor characteristics related to audit reporting, namely whether there is an auditor change ( $a u d i t o r c h a n g e$ ), whether the auditor issues a qualified opinion ( $a o q u a l i f i e d$ ), and auditor tenure ( $a u d i t o r t e n u r e$ ). Moreover, we control for the length of annual reports ( $l n a r_w c$ ) because longer reports can be more complex and potentially contain more risks. Finally, we include a set of governance variables, namely whether the Chairman and the CEO are the same person ( $d u a l$ ), the percentage of equity held by the largest shareholder ( $l a r g e s t h o l d e r$ ), the proportion of independent directors ( $i n d e d i r e c t o r$ ), the percentage of equity held by directors and executives ( $b o d m g t s t h o l d e r$ ), and the percentage of equity held by institutional investors ( $i i s t h o l d e r$ ). Appendix A defines all variables.

In equation (1), we include industry fixed effects and year fixed effects to control for unobservable industry-invariant and time-invariant factors, respectively. We also include audit firm fixed effects to account for unobservable audit firm characteristics, such as auditor knowledge and expertise. We cluster the standard errors at the firm level and winsorize all continuous variables at the 1 and 99 percentiles.

Sample Construction

Our sample consists of firms listed on the Shanghai or Shenzhen Stock Exchange that are under the mandatory KAMs disclosure requirement for the fiscal years 2017–2020. We obtain our data from the following sources: KAMs are hand-collected from audit reports available in company annual reports; firm and auditor characteristics variables from China Stock Market & Accounting Research (CSMAR); and data on the probability of default from the Credit Risk Initiative (CRI) of the National University of Singapore (NUS).⁶ Table 1 Panel A presents the sample construction procedure. We identify an initial sample of audit reports consisting of 15,078 firm-year observations over 2017–2020. After excluding observations with a disclaimer of audit opinion or with missing data from CSMAR or NUS CRI, financial firms, and fixed effect singletons, we obtain a final sample of 12,143 firm-year observations relating to 3508 unique firms.

Table 1.

Sample Construction and Distribution.

Panel A: Sample Construction
	Unique Issuers	Observations
Fiscal years 2017–2020 audit reports of firms listed on Shanghai or Shenzhen Stock Exchanges	4298	15,078
After excluding
Observations with a disclaimer of audit opinion	4285	14,938
Observations with missing CSMAR data	3617	12,608
Observations with missing NUS default data	3537	12,237
Financial firms	3509	12,146
Fixed effect singletons	3508	12,143
Final sample	3508	12,143

Panel B: Sample Distribution by Year and Ownership Type
	Observations
Fiscal Year	SOE	Non-SOE	Total
2017	859	1876	2735
2018	870	2177	3047
2019	921	2200	3121
2020	931	2309	3240
Total	3581	8562	12,143

Panel C: Sample Distribution by Industry and Ownership Type
	Observations
Industry	SOE	Non-SOE	Total
A: Agriculture, forestry and fishing	68	81	149
B: Mining	144	104	248
C1: Manufacturing – food and textile	204	581	785
C2: Manufacturing – chemical and products	508	1843	2351
C3: Manufacturing – metal and equipment	1115	3558	4673
C4: Manufacturing – instruments	28	190	218
D: Utility	284	97	381
E: Construction	133	195	328
F: Wholesale and retail trade	232	299	531
G: Transportation, warehousing and postal services	219	75	294
I: Communications, Software and information technology	178	871	1049
K: Real estate	192	196	388
L: Rental and leasing, and business services	73	133	206
M: Scientific research and technical services	46	110	156
N: Water conservancy, environment and public infrastructure management	64	144	208
R: Culture, Sports and entertainment	93	85	178
Total	3581	8,562	12,143

Notes. This table presents the sample. Panel A presents the sample construction process. Panel B presents the distribution of the sample by year and ownership type. Panel C presents the distribution of the sample by industry and ownership type.

Table 1 Panel B shows the distribution of the sample across fiscal years and by ownership type. We have slightly more observations in later years than in early years without any specific year dominating the sample. Table 1 Panel C reports the sample distribution across industry sectors and by ownership type. Our sample covers sixteen industries whilst more than half of the observations are in the manufacturing sectors.⁷

Baseline Analysis

Descriptive Statistics

Table 2 provides the descriptive statistics for all variables in equations (1) and (2). Panel A refers to the full sample. On average, our sample firms have two (zero) KAMs (abnormal KAMs) whereas the number of KAMs ranges from 1 to 4 (see Figure 1). Regarding the subject areas, Figure 2 reveals that revenue and receivables are the most frequent KAM subject areas, accounting for 36.3% and 18.4% of total KAMs, respectively; whereas PPE and RPTs are the least frequent KAM subject areas (each accounting for 2.8% of total KAMs). The distributions are comparable to those documented by prior literature using data from China (e.g., Zeng et al., 2021). In line with recent studies using comparable sample periods (e.g., Li & Sun, 2023; Zhang & Zhang, 2023), SOEs account for 29.5% of our sample. Untabulated correlations do not suggest any unusual behavior or multicollinearity issues.

Table 2.

Descriptive Statistics.

Panel A: Full sample (N = 12,143)
Variable		Mean	SD	P25	Median	P75
abn_kam_number		0.000	0.638	−0.149	−0.055	0.056
kam_number		2.059	0.644	2.000	2.000	2.000
kam_ppe		0.057	0.232	0.000	0.000	0.000
kam_goodwill		0.287	0.452	0.000	0.000	1.000
kam_inven		0.217	0.412	0.000	0.000	0.000
kam_rec		0.381	0.486	0.000	0.000	1.000
kam_rev		0.751	0.433	1.000	1.000	1.000
kam_rpt		0.057	0.232	0.000	0.000	0.000
kam_other		0.321	0.467	0.000	0.000	1.000
soe		0.295	0.456	0.000	0.000	1.000
size		22.255	1.271	21.356	22.098	22.973
lev		0.423	0.202	0.264	0.414	0.566
roa		0.030	0.087	0.013	0.037	0.067
loss		0.119	0.324	0.000	0.000	0.000
rev_grow		0.078	0.198	−0.013	0.048	0.134
current_ratio		2.333	2.039	1.182	1.675	2.665
in_ar		0.266	0.157	0.146	0.252	0.364
abs_dacc		0.055	0.056	0.017	0.038	0.071
ocf		0.050	0.068	0.011	0.050	0.089
std_ocf		0.043	0.037	0.018	0.032	0.055
ah		0.017	0.129	0.000	0.000	0.000
bm		0.508	0.348	0.263	0.420	0.653
stock_issue		0.120	0.325	0.000	0.000	0.000
ma		0.458	0.498	0.000	0.000	1.000
restructure		0.737	0.440	0.000	1.000	1.000
impair_loss		0.524	0.499	0.000	1.000	1.000
ppe_toa		0.296	0.201	0.137	0.267	0.425
lnage		2.977	0.277	2.833	2.996	3.178
pd		0.008	0.009	0.002	0.005	0.010
auditor_change		0.111	0.315	0.000	0.000	0.000
audit_tenure		1.727	0.895	1.099	1.946	2.398
ao_qualified		0.023	0.151	0.000	0.000	0.000
restate		0.137	0.344	0.000	0.000	0.000
weak		0.021	0.145	0.000	0.000	0.000
dual		0.311	0.463	0.000	0.000	1.000
lnar_wc		11.710	0.408	11.409	11.582	12.035
largestholder		0.327	0.142	0.217	0.303	0.417
indedirector		0.378	0.053	0.333	0.364	0.429
bodmgtstholder		0.147	0.196	0.000	0.024	0.279
iistholder		0.405	0.246	0.185	0.416	0.604
goodwill		0.039	0.080	0.000	0.001	0.036
inventory		0.133	0.120	0.054	0.106	0.170
receivables		0.133	0.107	0.048	0.112	0.192
revenue		0.591	0.384	0.343	0.509	0.728
total_rpt_amt		0.135	0.249	0.004	0.028	0.157
Panel B: SOE versus non-SOE subsamples (SOE N = 3581; non-SOE N = 8562)
Variable	SOE Mean	Non-SOE Mean	Mean difference p-Value	SOE Median	Non-SOE Median	Median Difference p-Value
abn_kam_number	−0.044	0.018	0.000	−0.075	−0.038	0.000
kam_number	2.005	2.082	0.000	2.000	2.000	0.000
kam_ppe	0.093	0.042	0.000	0.000	0.000	0.000
kam_goodwill	0.185	0.329	0.000	0.000	0.000	0.000
kam_inven	0.234	0.210	0.003	0.000	0.000	0.003
kam_rec	0.321	0.407	0.000	0.000	0.000	0.000
kam_rev	0.683	0.779	0.000	1.000	1.000	0.000
kam_rpt	0.087	0.045	0.000	0.000	0.000	0.000
kam_other	0.405	0.285	0.000	0.000	0.000	0.000
size	22.921	21.977	0.000	22.845	21.859	0.000
lev	0.493	0.394	0.000	0.499	0.384	0.000
roa	0.027	0.031	0.041	0.028	0.041	0.000
loss	0.105	0.126	0.001	0.000	0.000	0.001
rev_grow	0.065	0.084	0.000	0.037	0.055	0.000
current_ratio	1.840	2.540	0.000	1.389	1.800	0.000
in_ar	0.250	0.273	0.000	0.222	0.259	0.000
abs_dacc	0.049	0.057	0.000	0.034	0.039	0.000
ocf	0.048	0.051	0.007	0.049	0.050	0.007
std_ocf	0.041	0.044	0.000	0.030	0.033	0.000
ah	0.042	0.007	0.000	0.000	0.000	0.000
bm	0.676	0.437	0.000	0.595	0.379	0.000
stock_issue	0.121	0.119	0.858	0.000	0.000	0.858
ma	0.447	0.463	0.109	0.000	0.000	0.109
restructure	0.648	0.774	0.000	1.000	1.000	0.000
impair_loss	0.525	0.524	0.912	1.000	1.000	0.912
ppe_toa	0.343	0.277	0.000	0.300	0.259	0.000
lnage	3.083	2.933	0.000	3.091	2.944	0.000
pd	0.007	0.008	0.000	0.004	0.005	0.000
auditor_change	0.130	0.103	0.000	0.000	0.000	0.000
audit_tenure	1.670	1.751	0.000	1.792	1.946	0.000
ao_qualified	0.009	0.029	0.000	0.000	0.000	0.000
restate	0.144	0.134	0.171	0.000	0.000	0.171
weak	0.012	0.025	0.000	0.000	0.000	0.000
dual	0.103	0.397	0.000	0.000	0.000	0.000
lnar_wc	11.712	11.710	0.751	11.593	11.578	0.789
largestholder	0.371	0.308	0.000	0.347	0.292	0.000
indedirector	0.372	0.380	0.000	0.333	0.364	0.000
bodmgtstholder	0.012	0.204	0.000	0.000	0.139	0.000
iistholder	0.560	0.340	0.000	0.562	0.312	0.000
goodwill	0.016	0.049	0.000	0.000	0.003	0.000
inventory	0.144	0.128	0.000	0.108	0.105	0.839
receivables	0.105	0.144	0.000	0.071	0.127	0.000
revenue	0.611	0.583	0.000	0.514	0.508	0.896
total_rpt_amt	0.211	0.104	0.000	0.095	0.014	0.000

Notes. This table presents descriptive statistics of all variables used in equations (1) and (2). Panel A presents descriptive statistics of all variables for the full sample. Panel B presents the comparison of variable mean and median values between the SOEs and non-SOEs subsamples.

Panel B of Table 2 compares variables between SOEs and non-SOEs. On average, SOEs have significantly fewer abnormal KAMs than non-SOEs. Specifically, SOEs report fewer KAMs than their industry peers (mean = −0.044), whereas non-SOEs disclose more KAMs relative to their competitors (mean = 0.018). Moreover, SOEs report considerably more KAMs in certain areas than non-SOEs, such as PPE (mean = 0.093 vs. 0.042) and RPTs (mean = 0.087 vs. 0.045); and fewer KAMs in other areas, such as goodwill (mean = 0.185 vs. 0.329). Turning to control variables, it is apparent that SOEs are systematically different from non-SOEs, consistent with expectations. For example, SOEs are larger than non-SOEs (mean = 22.921 vs. 21.977), older (mean = 3.083 vs. 2.933), more leveraged (mean = 0.493 vs. 0.394), less profitable (mean = 0.027 vs. 0.031), and more capital-intensive (mean = 0.343 vs. 0.277); whilst they report losses less frequently (mean = 0.105 vs. 0.126), have higher ownership concentration (mean = 0.371 vs. 0.308), lower proportion of independent directors (mean = 0.372 vs. 0.380), less director and executive shareholdings (mean = 0.012 vs. 0.204), and higher institutional ownership (mean = 0.560 vs. 0.340). These differences are at least partially attributable to industry idiosyncrasies, which we control for in our empirical tests. We also perform additional tests to mitigate endogeneity concerns that our findings could be driven by differences between SOEs and non-SOEs (see Section “Endogeneity and Other Concerns”).

The Relation Between SOEs and KAMs Disclosure

Table 3 Panel A reports the results from estimating equation (1) on the full sample with

a b n_k a m_n u m b e r

as the dependent variable. In Column (1), we include firm-level controls. The coefficient on

s o e

is −0.082 and significant at the 1% level. Column (2) further controls for auditor characteristics. The coefficient on

s o e

is −0.089 and remains highly significant. In Column (3), we include the corporate governance variables; the coefficient on

s o e

is −0.089 and still significant at the 1% level. In terms of economic significance, results in Column (3) suggest that SOEs display 4.3% fewer KAMs relative to their industry peers.⁸

Table 3.

The Association Between SOEs and the (Abnormal) Number of KAMs.

Panel A: The association between SOEs and the abnormal number of KAMs
	(1) abn_kam_number	(2) abn_kam_number	(3) abn_kam_number
soe	−0.082***	−0.089***	−0.089***
soe	(−3.91)	(−4.25)	(−3.94)
size	0.065***	0.067***	0.071***
size	(5.75)	(5.96)	(6.11)
lev	0.009	0.005	0.008
lev	(0.12)	(0.07)	(0.11)
roa	−0.526***	−0.645***	−0.593***
roa	(−4.00)	(−5.01)	(−4.57)
loss	0.041	0.045	0.043
loss	(1.42)	(1.56)	(1.52)
rev_grow	0.005	−0.005	−0.000
rev_grow	(0.16)	(-0.14)	(-0.01)
current	−0.012**	−0.012**	−0.011**
current	(−2.38)	(−2.38)	(−2.17)
in_ar	0.149**	0.129*	0.137**
in_ar	(2.13)	(1.86)	(1.98)
absdacc	0.012	0.045	0.058
absdacc	(0.09)	(0.36)	(0.46)
ocf	−0.149	−0.136	−0.106
ocf	(−1.36)	(−1.25)	(−0.98)
stdocf	−0.432**	−0.419**	−0.439**
stdocf	(−2.15)	(−2.11)	(−2.21)
ah	0.110	0.119	0.125
ah	(1.28)	(1.38)	(1.45)
bm	0.008	0.003	−0.007
bm	(0.24)	(0.09)	(-0.19)
stock_issue	0.026	0.021	0.021
stock_issue	(1.31)	(1.03)	(1.03)
ma	0.007	0.004	0.005
ma	(0.55)	(0.34)	(0.38)
restructure	0.081***	0.080***	0.072***
restructure	(5.79)	(5.73)	(5.07)
impairloss	−0.064***	−0.064***	−0.062***
impairloss	(−3.71)	(−3.74)	(−3.62)
ppe_toa	−0.360***	−0.372***	−0.362***
ppe_toa	(−6.29)	(−6.54)	(−6.36)
lnage	−0.017	−0.015	−0.027
lnage	(−0.54)	(−0.48)	(−0.85)
pd	2.592**	3.331**	3.001**
pd	(2.04)	(2.64)	(2.38)
lnar_wc	0.300***	0.300***	0.300***
lnar_wc	(6.74)	(6.75)	(6.71)
auditorchange		0.050**	0.043
auditorchange		(1.98)	(1.72)
auditortenure		−0.018	−0.022*
auditortenure		(−1.47)	(−1.80)
aoqualified		−0.404***	−0.403***
aoqualified		(−7.21)	(−7.25)
restate		0.032*	0.029
restate		(1.81)	(1.61)
weak		0.076	0.077
weak		(1.42)	(1.45)
dual			0.022
dual			(1.29)
largestholder			−0.096
largestholder			(−1.30)
indedirector			−0.060
indedirector			(−0.41)
bodmgtstholder			−0.146**
bodmgtstholder			(−2.25)
iistholder			−0.085
iistholder			(-1.46)
intercept	−4.826***	−4.835***	−4.757***
intercept	(−9.78)	(−9.79)	(−9.42)
N	12,143	12,143	12,143
Adjusted $R^{2}$	0.12	0.13	0.13

Panel B: The association between SOEs and the number of KAMs
	(1) kam_number	(2) kam_number	(3) kam_number
soe	−0.083***	−0.090***	−0.090***
soe	(−3.94)	(−4.28)	(−3.96)
Firm characteristic controls	Yes	Yes	Yes
Audit firm characteristic controls	No	Yes	Yes
Governance controls	No	No	Yes
N	12,143	12,143	12,143
Adjusted $R^{2}$	0.14	0.14	0.14

Notes. This table presents the results from estimating equation (1) based on the full sample. Panels A and B present the results using abn_kam_number and kam_number as the dependent variable respectively. All models include industry, year, and audit firm fixed effects. All variables are defined in Appendix A. All continuous variables are winsorized at the 1 and 99 percentiles. t-statistics reported in parentheses are based on standard errors clustered by firm. *, **, and *** stand for significance at 10%, 5%, and 1% levels, respectively.

Regarding control variables, results show that KAMs disclosure can vary significantly across firms. For example, firms’ audit reports disclose more abnormal KAMs if they are bigger, engage in restructuring, and have a higher probability of default, or a lengthier annual report. On the other hand, firms have fewer abnormal KAMs if they are more profitable and capital-intensive, as well as if they report an impairment loss or receive a qualified audit opinion.⁹ As discussed, our focus is on abnormal/industry-adjusted KAMs reporting. But for completeness, we also examine the determinants of the unadjusted total number of KAMs based on equation (1). We report this analysis in Panel B. As shown, the coefficient of soe is significantly negative across all columns, indicating that SOEs disclose fewer total KAMs.¹⁰

Next, Table 4 reports the results from estimating missed KAMs. As explained in Section “Research Design”, to construct

m i s s e d_k a m

, we first estimate equation (2) where we use logit regressions to predict the disclosure probability of each KAM subject area. Panel A reports the estimation results. The full set of control variables and fixed effects described earlier are included in all models, but to conserve space, we do not report coefficient estimates for these variables.¹¹ In line with Burke et al. (2023), across all columns, the coefficient on the KAM subject area’s related account variable is positively correlated with the disclosure of a KAM in that subject area, indicating that auditors are more likely to report KAMs in a specific subject area when they observe a high amount in the related financial statement account.¹² Similar to Burke et al. (2023), our prediction models display high areas under the Receiver Operating Characteristic (ROC) curve, ranging from 0.729 to 0.938, indicating high predictive power.

Table 4.

The Association Between SOEs and Missed KAMs.

Panel A: Prediction models for missed_kam
	(1) kam_ppe	(2) kam_goodwill	(3) kam_inventory	(4) kam_receivables	(5) kam_revenue	(6) kam_rpt	(7) kam_other
ppe	5.697*** (13.40)
goodwill		38.813*** (16.23)
inventory			8.238*** (17.15)
receivables				9.053*** (18.93)
revenue					0.030 (0.27)
total_rpt_amt						1.371*** (8.40)
Controls	Yes	Yes	Yes	Yes	Yes	Yes	Yes
N	11,815	12,143	12,143	12,138	12,138	12,032	12,143
Pseudo $R^{2}$	0.26	0.47	0.20	0.18	0.12	0.14	0.12
ROC curve	0.862	0.938	0.799	0.782	0.736	0.784	0.729

Panel B: Descriptive statistics
Variable	SOE Mean	Non-SOE Mean	Mean Difference p-Value	SOE Median	Non-SOE Median	Median Difference p-Value
missed_kam	0.604	0.389	0.000	1.000	0.000	0.000
missed_kam_ppe	0.223	0.124	0.000	0.000	0.000	0.000
missed_kam_goodwill	0.011	0.024	0.000	0.000	0.000	0.000
missed_kam_inventory	0.105	0.088	0.004	0.000	0.000	0.004
missed_kam_receivables	0.050	0.055	0.221	0.000	0.000	0.222
missed_kam_revenue	0.013	0.015	0.388	0.000	0.000	0.388
missed_kam_rpt	0.301	0.109	0.000	0.000	0.000	0.000
missed_kam_other	0.122	0.061	0.000	0.000	0.000	0.000

Panel C: Regression results
	(1) missed_kam	(2) missed_kam_ppe	(3) missed_kam_goodwill	(4) missed_kam_inventory	(5) missed_kam_receivables	(6) missed_kam_revenue	(7) missed_kam_rpt	(8) missed_kam_other
soe	0.373***	−0.123	−0.003	0.309**	0.235	0.827***	0.278***	−0.155
soe	(4.92)	(-0.87)	(-0.01)	(2.12)	(1.23)	(2.93)	(2.71)	(-1.13)
Controls	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
N	12,143	11,815	11,604	10,959	11,004	10,070	11,894	12,005
Pseudo $R^{2}$	0.17	0.48	0.16	0.20	0.20	0.21	0.50	0.30
ROC curve	0.774	0.950	0.835	0.837	0.848	0.883	0.943	0.912

Notes. This table presents the results from estimating missed KAM decisions. Panel A reports the results from the prediction models of missed_kam, following Equation (2). Panel B displays the descriptive statistics of missed_kam. Panel C presents the results from estimating Equation (1) based on the full sample with missed_kam being the dependent variable. All models include industry, year, and audit firm fixed effects. All variables are defined in Appendix A. All continuous variables are winsorized at the 1 and 99 percentiles. z-scores reported in parentheses are based on standard errors clustered by firm. *, **, and *** stand for significance at 10%, 5%, and 1% levels, respectively.

We use the regression results in Panel A to construct our firm-year aggregate missed_kam variable. Moreover, we also construct a separate missed KAM indicator variable for each one of the seven subject areas, in order to examine which subject areas are more likely to be affected by state control. Panel B of Table 4 compares the mean and median of these variables between SOEs and non-SOEs. On average, 60.4% (38.9%) of SOEs (non-SOEs) have at least one missed KAM, and the difference is statistically significant. In terms of specific KAM subject areas, SOEs have more missed KAMs related to PPE, inventory, and RPTs than non-SOEs. However, these statistics do not control for the determining factors of missed KAMs, so we investigate these issues further in our regression analysis.

Panel C of Table 4 reports the logit regression results of equation (1) when the dependent variable is $m i s s e d_k a m$ . In column (1), the coefficient on $s o e$ is 0.373 and significant at the 1% level, indicating that SOEs have 45% ( $e^{0.373} - 1 = 0.45$ ) more odds of missing at least one expected KAM than non-SOEs. The marginal effect suggests that SOEs are, on average, 7.2% more likely to avoid the disclosure of expected KAMs in at least one expected subject area than non-SOEs.^13,¹⁴ Columns (2)–(8) examine the association between SOEs and missed KAMs in each subject area. The coefficient on $s o e$ is positive and significant in Columns (4), (6) and (7), indicating that the most commonly missed KAM subject areas for SOEs are inventory, revenue, and RPTs, respectively.¹⁵ Given that 52% of SOEs in our sample are manufacturing firms, these SOEs have potentially strong incentives to suppress expected KAMs related to inventory, a material asset for the manufacturing sector. SOEs may also have incentives to withhold expected KAMs related to revenue because it is an important financial measure that is often used by government authorities to evaluate SOE performance (Du et al., 2012). Finally, SOEs have incentives to suppress expected KAMs on RPTs given that such transactions are widely used by SOEs to manipulate performance and expropriate wealth (e.g., Jiang et al., 2010).

Taken together, Tables 3 and 4 suggest that SOEs have fewer KAMs disclosures, consistent with our prediction that the political forces behind state ownership influence audit reporting.

Cross-Sectional Analysis

State Control and RPTs

So far, we have documented that SOEs on average have fewer abnormal KAMs and are more likely to evade the disclosure of expected KAMs, than non-SOEs. These findings are likely to be explained by SOEs’ ability and incentives to exert political influence in order to obscure disclosures and withhold bad news. To shed more light on the above inference, we perform cross-sectional analyses. Specifically, we estimate our baseline equation (1) conditioning on the firm’s level of state control (SOEs’ ability), and firms’ engagement in tunneling RPTs (SOEs’ incentives).

Regarding state control, we examine the differential effect of being an SOE on KAMs disclosure, depending on the strategic importance of the industry and the level of state ownership concentration. SOEs in strategic or public welfare industries (such as national safety, national economy, and utilities) are subject to greater state monitoring and experience more stringent state control (Jin et al., 2022). Similarly, increased equity holdings by the state suggest tighter political control. Accordingly, we expect even fewer abnormal KAMs and a higher probability of missed KAMs for SOEs with greater state ownership concentration or in strategic/public welfare industries. We generate an indicator variable strg_pubwelfare equal to 1 if the firm is in a strategic industry or a public welfare industry, and 0 otherwise. We classify firms into the respective industries following Wei et al. (2017) and Jiang (2021). We proxy for state ownership concentration with the largest shareholder equity stake variable (Gul et al., 2010), which we transform into an indicator based on its median value for the SOE subsample ( $h i g h_l a r g e s t h o l d e r$ ).¹⁶

Regarding RPTs, we examine the differential effect of being an SOE on the disclosure of RPT-related KAMs, conditioning on whether the firm engages in tunneling transactions. Tunneling RPTs refer to the various ways in which the controlling shareholder siphons resources and extracts wealth from the firm at the expense of minority shareholders and creditors; lending from the firm to related parties is the major tunneling mechanism (Milhaupt & Pargendler, 2018). Sun & Li, 2024 report that most loans from listed firms to their controlling shareholders and their affiliates are typically long-term loans with no guarantees or collateral, and are frequently written off as bad debt. Consistent with the wealth-expropriating and value-decreasing role of tunneling, prior research documents that such RPTs are associated with negative market reactions (Cheung et al., 2009). Similarly, Chinese firms with high levels of related party loans and guarantees experience negative economic consequences, including sharp declines in profitability, a higher likelihood of receiving a “probation” status as well as of entering financial distress (Jiang et al., 2010). Moreover, tunneling RPTs also have earnings management implications. For example, Liu and Lu (2007) and Jian and Wong (2010) find that the pattern of earnings management observed among Chinese firms is consistent with their desire to facilitate and sustain long-term tunneling.

Collectively, these findings suggest greater business and financial reporting risks for companies engaging in tunneling. Accordingly, prior research documents that auditors are more likely to provide modified audit opinions to firms with high levels of intercorporate loans (Jiang et al., 2010) or to firms with related lending but not related borrowing (Fang et al., 2018).¹⁷ Therefore, one would expect more KAMs in the subject area of RPTs for firms involved in tunneling. However, Cheung et al. (2009) document that RPTs representing tunneling are accompanied by significantly less information disclosure (e.g., financial advisor “fairness” reports). This evidence suggests strong incentives for firms engaged in tunneling transactions, and in particular SOEs given their political motivations, to obscure transparency and obfuscate the disclosure about wealth expropriation for the benefit of the controlling shareholder. Thus, one would expect to see fewer RPT-related KAMs for SOEs with tunneling RPTs.

Following prior studies (Fang et al., 2018; Jia et al., 2013), we construct two tunneling indicator variables: rp_lend (rp_progrt) equal to 1 if the firm lends (provides guarantees) to its controlling shareholder or other firms controlled by the controlling shareholder, and 0 otherwise.¹⁸

Descriptive Statistics

Table 5 reports the descriptive statistics of the partitioning variables for the cross-sectional analyses. Panel A shows that 20.6% of our observations are in a strategic or public welfare industry. Moreover, 12.2% (5.5%) of our observations lend (provide guarantees) to its controlling shareholder or other firms controlled by the controlling shareholder. Panel B compares the SOE and the non-SOE subsamples. In line with expectations, there are more SOEs in strategic or public welfare industries (mean = 0.346 vs. 0.147). Also, SOEs have higher ownership concentration (mean = 0.499 vs. 0.352) and are more involved in tunneling RPTs (mean = 0.258 vs. 0.066 for rp_lend, and mean = 0.080 vs. 0.044 for rp_progrt).

Table 5.

Descriptive Statistics of Partitioning Variables in Cross-Sectional Tests.

Panel A: Full sample (N = 12,143)
Variable	Mean	SD	P25	Median	P75
strg_pubwelfare	0.206	0.404	0.000	0.000	0.000
high_largestholder	0.395	0.489	0.000	0.000	1.000
rp_lend	0.122	0.328	0.000	0.000	0.000
rp_progrt	0.055	0.228	0.000	0.000	0.000

Panel B: SOE versus non-SOE subsamples comparison (SOE N = 3581; non-SOE N = 8562)
Variable	SOE Mean	Non-SOE Mean	Mean Difference p-Value	SOE Median	Non-SOE Median	Median Difference p-Value
strg_pubwelfare	0.346	0.147	0.000	0.000	0.000	0.000
high_largestholder	0.499	0.352	0.000	0.000	0.000	0.000
rp_lend	0.258	0.066	0.000	0.000	0.000	0.000
rp_progrt	0.080	0.044	0.000	0.000	0.000	0.000

Notes. This table presents the descriptive statistics of the partitioning variables used in the cross-sectional tests. Panel A presents the descriptive statistics for the full sample. Panel B presents the comparison of variable mean and median values between the SOEs and non-SOEs subsamples. All variables are defined in Appendix A.

Cross-Sectional Results

Table 6 presents the cross-sectional results on state control (Panel A) and tunneling (Panel B). The full set of control variables and fixed effects are included, but not reported for brevity.

Table 6.

Cross-Sectional Tests.

Panel A: State control
	(1) abn_kam_number	(2) abn_kam_number	(3) missed_kam	(4) missed_kam
soe	−0.068***	−0.061**	0.311***	0.286***
soe	(−2.72)	(−2.24)	(3.72)	(3.18)
strg_pubwelfare	−0.032		0.066
strg_pubwelfare	(−0.98)		(0.62)
soe*strg_pubwelfare	−0.091**		0.283*
soe*strg_pubwelfare	(−2.10)		(1.91)
high_largestholder		0.023		0.010
high_largestholder		(0.82)		(0.09)
soe*high_largestholder		−0.071**		0.219*
soe*high_largestholder		(−1.97)		(1.74)
Controls	Yes	Yes	Yes	Yes
N	12,143	12,143	12,143	12,143
Adjusted $R^{2}$	0.13	0.13	-	-
Pseudo $R^{2}$	-	-	0.17	0.17
p-value
soe + soe* strg_pubwelfare	<0.001		<0.001
soe + soe* high_largestholder		<0.001		<0.001

Panel B: Tunneling
	(1) abn_kam_number_rpt	(2) abn_kam_number_rpt	(3) missed_kam_rpt	(4) missed_kam_rpt
soe	0.005	0.014	0.096	0.255**
soe	(0.55)	(1.44)	(0.88)	(2.44)
rp_lend	0.024*		0.413**
rp_lend	(1.80)		(1.97)
soe*rp_lend	0.011		0.593**
soe*rp_lend	(0.56)		(2.34)
rp_progrt		0.059**		0.823***
rp_progrt		(2.56)		(3.21)
soe*rp_progrt		−0.057*		0.120
soe*rp_progrt		(-1.94)		(0.35)
Controls	YES	YES	YES	YES
N	12,143	12,143	11,894	11,894
Adjusted $R^{2}$	0.04	0.04	-	-
Pseudo R²	-	-	0.51	0.50
p-value
soe + soe*rp_lend	0.683		0.006
soe + soe*rp_progrt		0.067		0.008

Notes. This table presents the results from estimating equation (1) conditioning on the level of state control (Panel A) and on tunneling transactions (Panel B). In Panel A, Columns 1 and 2 present the results using abn_kam_number as the dependent variable. Columns 3 and 4 present the results using missed_kam as the dependent variable. In Panel B, Columns 1 and 2 present the results using abn_kam_rpt_number as the dependent variable. Columns 3 and 4 present the results using missed_kam_rpt as the dependent variable. All models include industry, year, and audit firm fixed effects. All variables are defined in Appendix A. All continuous variables are winsorized at the 1 and 99 percentiles. t-statistics or z-scores reported in parentheses are based on standard errors clustered by firm. *, **, and *** stand for significance at 10%, 5%, and 1% levels, respectively.

In Panel A Columns (1) and (2), where the dependent variable is $a b n_k a m_n u m b e r$ , the coefficient on $s o e$ is always negative and highly significant, whereas the coefficients on the interaction terms $s o e * s t r g_p u b w e l f a r e$ and $s o e * h i g h_l a r g e s t h o l d e r$ are also significantly negative (−0.091 and −0.071, respectively, at the 5% level). To put in context, results indicate, for example, that the number of abnormal KAMs drops by an incremental 4.4% for SOEs operating in a strategically important industry (see Column 1). In Columns (3) and (4), where the dependent variable is $m i s s e d_k a m$ , the coefficient on $s o e$ is always positive and highly significant, whilst the interaction terms $s o e * s t r g_p u b w e l f a r e$ and $s o e * h i g h_l a r g e s t h o l d e r$ are also positive and significant (0.283 and 0.219, respectively, at the 10% level). In terms of magnitude, these findings suggest, for example, that SOEs with high ownership concentration are more likely by an additional 4.2% to avoid the disclosure of expected KAMs (see Column 4). Overall, we report that SOEs in industries experiencing greater state scrutiny as well as those with highly concentrated state ownership disclose even fewer KAMs.

In Panel B, we test the differential effect of SOEs on KAMs disclosure conditioning on the engagement with tunneling RPTs. We employ two dependent variables related to KAMs in the subject area of RPTs, namely $a b n_k a m_n u m b e r_r p t$ (Columns 1 and 2) and $m i s s e d_k a m_r p t$ (Columns 3 and 4).¹⁹ As reported, $s o e$ is insignificant in Columns (1) and (2), whereas the interaction term $s o e * r p_p r o g r t$ in Column (2) is significantly negative (−0.057 at the 10% level), indicating an additional decrease of 2.8% in abnormal RPT-related KAMs for SOEs providing guarantees. In Columns (3) and (4), $s o e$ is positive, albeit significant only in the latter Column. More importantly, the interaction term $s o e * r p_l e n d$ (Column 3) is also positive (0.593) and highly significant, suggesting a 4.2% incremental increase in the likelihood of missing predicted KAMs for SOEs providing loans. Collectively, we find that SOEs involved with tunneling transactions report even fewer RPT-related KAMs.^20,²¹ This is consistent with SOEs involved in tunneling having stronger incentives to render audit reports less informative so as to provide cover for potential wealth expropriation activities for the benefit of the state. Overall, the results in Table 6 support our predictions on the adverse consequences of tightened state control behind SOEs for KAMs disclosure in audit reports.

Supplementary Analysis

In this section, we perform supplementary analysis using textual features of KAMs, as outcome variables, to provide more detailed and granular evidence on how state ownership affects KAMs reporting.²² The results of this analysis are presented in Table 7.

Table 7.

Supplementary Tests.

Panel A: Descriptive statistics for full sample (N = 12,143)
Variable	Mean	SD	P25	Median	P75
description_wordcount	378.811	188.697	242.000	348.000	482.000
response_wordcount	601.915	242.856	435.000	574.000	734.000
description_riskwordcount	2.008	1.445	1.000	2.000	3.000
response_riskwordcount	1.468	1.548	0.000	1.000	2.000
footnote_count	2.885	1.824	2.000	3.000	4.000
low_kam_effort	0.250	0.433	0.000	0.000	0.000

Variable	SOE Mean	Non-SOE Mean	Mean Difference p-Value	SOE Median	Non-SOE Median	Median Difference p-Value
Panel B: Descriptive statistics for SOE versus non-SOE subsamples (SOE N = 3581; non-SOE N = 8562)
description_wordcount	369.083	382.880	0.000	337.000	352.000	0.000
response_wordcount	569.497	615.473	0.000	546.000	588.000	0.000
description_riskwordcount	1.959	2.028	0.017	2.000	2.000	0.061
response_riskwordcount	1.234	1.565	0.000	1.000	1.000	0.000
footnote_count	2.871	2.890	0.589	2.000	3.000	0.049
low_kam_effort	0.288	0.234	0.000	0.000	0.000	0.000

	(1)	(2)	(3)	(4)	(5)	(6)
Panel C: Textual features of KAMs
	description_wordcount	response_wordcount	description_riskwordcount	response_riskwordcount	footnote_count	low_kam_effort
soe	−18.640***	−42.197***	−0.073	−0.124**	−0.122*	0.195**
soe	(-2.84)	(-4.85)	(-1.27)	(-2.17)	(-1.84)	(2.23)
Controls	Yes	Yes	Yes	Yes	Yes	Yes
N	12,143	12,143	12,143	12,143	12,143	12,138
Adjusted R²	0.21	0.20	0.06	0.17	0.14	-
Pseudo R²	-	-	-	-	-	0.08

Notes. This table presents the results from supplementary analysis based on textual features of KAMs. Panel A presents the descriptive statistics for the full sample. Panel B presents the comparison of variable mean and median values between the SOEs and non-SOEs subsamples. Panel C reports the regression results. All models include industry, year, and audit firm fixed effects. All variables (including the dependent variables) are defined in Appendix A. All continuous variables are winsorized at the 1 and 99 percentiles. t-statistics or z-scores reported in parentheses are based on standard errors clustered by firm. *, **, and *** stand for significance at 10%, 5%, and 1% levels, respectively.

In line with concurrent studies (e.g., Goh et al., 2023; Klevak et al., 2023), we employ a range of variables measuring the overall as well as risk-focused extent of KAMs descriptions and related audit responses. Specifically, we employ the following textual feature variables: (a) the total word count of all KAMs descriptions (description_wordcount), (b) the total word count of all auditor responses to KAMs (response_wordcount), (c) the number of risk words in all KAMs descriptions (description_riskwordcount), (d) the number of risk words in all auditor responses to KAMs (response_riskwordcount), (e) the number of financial statement footnotes referenced in KAMs (footnote_count), and (f) the auditor’s effort in responding to KAMs (low_kam_effort).

Regarding the risk-related variables, we construct a risk wordlist following prior literature in the Chinese setting (He et al., 2022b; Liu et al., 2022; Wang & Wang, 2023; Wang & Zeng, 2019).²³ Also, low_kam_effort is an indicator variable that takes the value of 1 if kam_effort is in the bottom quartile, and 0 otherwise. As in Goh et al. (2023), kam_effort is measured as the number of words in each KAM audit response divided by the number of words in each KAM description at the firm-year level. The intuition behind this proxy is that when auditors tend to communicate less (in terms of the number of words) about their procedures relative to the description of the risks in the identified KAM, it signifies the extent of the audit effort expended to address the aforementioned risks.

Panels A and B of Table 7, respectively, describe the above variables for the full sample and separately for SOEs and non-SOEs. As shown in Panel B, KAMs reporting by SOEs involves significantly fewer total or risk-related words, both for KAMs descriptions and auditor responses. For example, the mean of description_wordcount is 369.083 and 382.880 for SOEs and non-SOEs, respectively; whereas the equivalent values of response_riskwordcount are 1.234 and 1.565. Similarly, SOEs’ KAMs reference fewer footnotes than those of non-SOEs (based on the median values of footnote_count). Also, auditors of SOEs are more likely to exert less effort in addressing specific KAMs than non-SOEs’ auditors (the mean value of low_kam_effort is 0.288 and 0.234 for SOEs and non-SOEs, respectively).

Panel C of Table 7 reports the results of our empirical analysis. We begin with description_wordcount and response_wordcount, which proxy for the total amount of information provided in KAMs. We expect SOEs to have shorter KAMs descriptions and auditor responses than non-SOEs if the former are more likely to constrain KAMs information. In line with our prediction, Columns (1) and (2) in Panel C report that the coefficient of soe is negative (−18.640 and −42.197, respectively) and highly significant in both cases.

Then, we focus on the extent to which KAMs explicitly describe risks. This is important because KAMs are, to some extent, designed to provide information on high-risk areas of a company’s financial reporting. Again, we expect SOEs’ KAMs descriptions and related auditor responses to discuss risk issues less frequently than non-SOEs if the former have incentives and the ability to withhold potentially negative news. In Panel C, Column (3), when we use description_riskwordcount as the outcome variable, the coefficient on soe is negative (−0.073) but insignificant. In Column (4), the outcome variable is response_riskwordcount and the coefficient on soe is significantly negative (−0.124 at the 5% level). This is consistent with our prediction.

Next, we analyze the extent to which KAMs explicitly refer to financial statement footnotes. As reported in Column (5), when the outcome variable is footnote_count, the coefficient on soe is negative (−0.122) and significant (t-statistic = −1.84), indicating that SOEs’ KAMs refer to fewer financial statement footnotes than in the case of non-SOEs, consistent with our expectation. Lastly, we examine the auditor’s effort in responding to KAMs, namely low_kam_effort. Accordingly, we expect a higher likelihood for SOE auditors to expend less audit effort in addressing KAMs, if SOEs tend to obfuscate KAMs reporting. Consistent with our prediction, in Panel C, Column (6), the coefficient on soe is positive (0.195) and highly significant (t-statistic = 2.23).²⁴

To put in context, overall, our supplementary analysis reveals that, relative to non-SOEs, SOE KAMs descriptions (auditor responses) consist of 4.9% (7.0%) fewer words; SOE KAMs cite 4.2% fewer financial statement footnotes; and SOE auditors communicate less by 3.3% about their procedures when addressing identified KAMs as well as use 8.5% fewer risk-related words when responding to KAMs. This is all consistent with state ownership having negative consequences for KAMs reporting.

Endogeneity and Other Concerns

Entropy Balancing and Heckman Two-Stage Correction

As reported in Table 2 Panel B, there are several fundamental differences between SOEs and non-SOEs, raising endogeneity concerns about our findings. In this section, we perform two tests to mitigate such concerns, namely entropy balancing and Heckman two-stage correction. In Table 8, we re-estimate baseline analyses reported in Tables 3 and 4 Panel C with the full set of control variables and fixed effects (not tabulated).

Table 8.

Endogeneity Tests.

	Entropy Balancing		Heckman Two-Stage Correction
	(1)	(2)	(3)	(4)	(5)
	abn_kam_number	missed_kam	soe	abn_kam_number	missed_kam
soe	−0.148***	0.307**		−0.081***	0.364***
soe	(−3.62)	(2.25)		(−3.44)	(4.61)
reg_ind			0.450***
reg_ind			(6.56)
Inverse mills ratio				−0.092	0.099
Inverse mills ratio				(−1.42)	(0.39)
Controls	Yes	Yes	Yes	Yes	Yes
1st stage	-	-	Yes	No	No
2nd stage	-	-	No	Yes	Yes
N	12,143	12,143	12,143	12,143	12,143
Pseudo $R^{2}$	-	0.22	0.43	-	0.17
Adjusted $R^{2}$	0.27	-	-	0.13	-

Notes. This table presents the results from estimating equation (1) based on entropy balancing and Heckman two-stage correction. Columns 1 and 2 present regression results based on entropy balancing. Columns 3 to 5 present the regression results based on Heckman two-stage correction. The first stage of entropy balancing (Heckman two-stage correction) includes industry and year fixed effects (year fixed effects). All other models include industry, year, and audit firm fixed effects. All variables are defined in Appendix A. All continuous variables are winsorized at the 1 and 99 percentiles. t-statistics or z-scores reported in parentheses are based on standard errors clustered by firm. *, **, and *** stand for significance at 10%, 5%, and 1% levels, respectively.

Entropy balancing is a generalization of conventional propensity score weighting, designed to ensure balance in the distributions of covariates across the treatment and control groups (Hainmueller, 2012). Entropy balancing has advantages over propensity score weighting or matching because it does not depend on the correct specification of a propensity score model or researchers’ subjective calibration choices. Specifically, we balance the SOE and non-SOE groups on the first, second and third moment of all control variables used in equation (1), as well as on industry- and year-fixed effects. A covariate balance between SOEs and non-SOEs is achieved for all control variables in terms of mean, median or skewness (untabulated). Estimations of equation (1) using the entropy balanced sample are presented in Columns (1) and (2). The coefficient on $s o e$ remains negative when $a b n_k a m_n u m b e r$ is the dependent variable, and positive with $m i s s e d_k a m$ as the dependent variable. In both cases, the coefficients are significant.²⁵

Next, to mitigate the potential selection bias of SOEs, we apply the Heckman (1979) two-stage correction method. This method is used widely to deal with selection bias (e.g., Bradshaw et al., 2019; Lennox et al., 2012). In the first stage, we follow prior literature (Bradshaw et al., 2019) and use regulated industries ( $r e g_i n d$ ), where the state retains tight control, to predict the probability of a firm being an SOE (Central Committee of the Communist Party of China, 2003; State Council of the People’s Republic of China, 2020). Accordingly, $r e g_i n d$ is an indicator variable equal to 1 for firms in the following sectors: mining, railroads, trucking, airlines, telecommunications, energy supply, media and technology (Bradshaw et al., 2019; Chang et al., 2014).²⁶ In the first stage, we regress $s o e$ on $r e g_i n d$ and all control variables in equation (1) and present results in Column (3). As predicted, the coefficient on $r e g_i n d$ is positive and highly significant. We then compute the inverse Mills ratio based on the first stage results and include it in the second stage regressions. As shown in Columns (4) and (5), the coefficient on $s o e$ is significantly negative when the dependent variable is $a b n_k a m_n u m b e r$ , and significantly positive when the dependent variable is $m i s s e d_k a m$ , after adjusting for the selection bias. Overall, our primary findings remain qualitatively unchanged. That is, SOEs have fewer abnormal KAMs and are more likely to evade expected KAMs than non-SOEs. Taken together, the results of the above analyses provide additional reassurance against endogeneity concerns.

Ruling out the Alternative Explanation

It may be argued that our baseline results in Tables 3 and 4 are driven by the fact that SOEs are less risky than non-SOEs. That is, SOEs do not need to disclose as many KAMs as non-SOEs because SOEs are supported and protected by the government, enjoy soft budget constraints, and are less threatened by financial difficulties. To rule out this alternative explanation, we re-estimate equation (1) after partitioning our sample on the level of government support (proxied by the amount of government subsidies that a firm receives) and firm risk (proxied by leverage and probability to default). Results (untabulated) show that SOEs report fewer abnormal KAMs and are more likely to have missed KAMs irrespective of their level of government support or risk, ruling out the above alternative explanation.

Conclusions

During the past few years, regulators around the world have been mandating expanded audit reports, in which auditors are required to communicate new information on KAMs. Theory and empirical evidence suggest that political institutions play an important role in shaping financial reporting and auditing. Yet, research has been silent on how state control affects the disclosure of KAMs. We address this topic by exploiting idiosyncratic features of the Chinese economy and examining if the strong political forces behind SOEs affect KAMs reporting.

We document that, compared to non-SOEs, SOEs have fewer KAMs than their peers, in the audit reports. SOEs are also more likely to avoid the disclosure of expected KAMs, especially in the subject areas of inventory, revenue and RPTs. Cross-sectional analyses reveal that the effects of SOEs on KAMs disclosure are more prominent for SOEs subject to amplified state control via higher state ownership concentration or for SOEs in politically strategically important industries. The effects are also stronger when SOEs have greater political incentives for less transparency resulting from tunneling RPTs. Additional analysis shows that being an SOE has adverse consequences for the content and textual characteristics of KAMs descriptions and auditor responses.

Taken together, our findings indicate that the communication of new information on KAMs can be compromised in the case of state-owned companies, especially when they have strong politically motivated incentives and sufficient political power. Overall, our study represents a first attempt at understanding the role of state control and related institutional factors in determining KAMs disclosure in expanded audit reports. Naturally, our study is subject to limitations, the most important one being the SOE status is not assigned randomly and is highly stable over time. While we employ various techniques to mitigate endogeneity concerns, we note that none of them is perfect. Nonetheless, the results of the various analyses corroborate with each other and provide consistent support for our findings. Future research may extend our findings by employing a longer time-series to examine the consequences of suppressing KAMs disclosure on SOE managers’ political capital and their subsequent career path in the political hierarchy.

Supplemental Material

Supplemental Material - State Control, Related-Party Transactions and Audit Reporting: Evidence From Key Audit Matters

Supplemental Material for State Control, Related-Party Transactions and Audit Reporting: Evidence From Key Audit Matters by Annita Florou, Xiaoxi Wu, Shuai Yuan, and Vincent (Qiru) Zhang in Journal of Accounting, Auditing & Finance

Footnotes

Acknowledgements

We thank Pietro Bianchi, Mark DeFond, April Klein, Clive Lennox, Katherine Schipper, Regina Wittenberg-Moerman, T.J. Wong, Donghui Wu, Zhifang Zhang, participants at the European Accounting Association 2022 Annual Congress, and workshop participants at Bocconi University, Free University of Bozen-Bolzano, INSEAD, Queen’s University Belfast, and CUNEF university for their helpful and constructive comments. We thank Dapeng Zhu for her contribution in data collection. We also thank Zhenda Guan, Xuanchun Huang, Meiru Jiang, Tianqing Li, Yizhou Liu, Junyi Luo, Wanzhi Mao, Huilin Wang, Xu Xia, Xier Xu, Chen Yang, and Yifan Zhao for their assistance in data collection. Hand-collected data is available upon request.

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: Shuai Yuan would like to thank the support of the Ministry of Education of China under Humanities & Social Sciences Programme with project code 24YJC790222.

ORCID iDs

Annita Florou

Shuai Yuan

Supplemental Material

Supplemental material for this article is available online.

Notes

Variable Definitions

This appendix defines variables used in our empirical analysis.

Variable	Definition
Dependent variables
abn_kam_number	The difference between the total number of KAMs disclosed in the firm’s auditor report in the current year and the average number of KAMs of all firms in the same industry and in the same year.
abn_kam_number_rpt	The difference between the total number of KAMs related to related party transactions disclosed in the firm’s auditor report in the current year and the average number of KAMs related to related party transactions of all firms in the same industry and in the same year.
kam_number	The total number of KAMs disclosed in the firm’s auditor report in the current year.
missed_kam	Indicator variable. It is 1 if the auditor does not disclose KAMs in at least one of the seven subject areas where the models predict disclosure (predicted probability in the top quintile) in the current year, and 0 otherwise.
kam_ppe	Indicator variable. It is 1 if a KAM related to property, plant and equipment is disclosed in the auditor’s report in the current year, and 0 otherwise.
kam_goodwill	Indicator variable. It is 1 if a KAM related to goodwill is disclosed in the auditor’s report in the current year, and 0 otherwise.
kam_inventory	Indicator variable. It is 1 if a KAM related to inventory is disclosed in the auditor’s report in the current year, and 0 otherwise.
kam_receivables	Indicator variable. It is 1 if a KAM related to receivables is disclosed in the auditor’s report in the current year, and 0 otherwise.
kam_revenue	Indicator variable. It is 1 if a KAM related to revenue is disclosed in the auditor’s report in the current year, and 0 otherwise.
kam_rpt	Indicator variable. It is 1 if a KAM related to related party transactions is disclosed in the auditor’s report in the current year, and 0 otherwise.
kam_other	Indicator variable. It is 1 if a KAM related to other subject areas is disclosed in the auditor’s report in the current year, and 0 otherwise.
missed_kam_ppe	Indicator variable. It is 1 if the auditor does not disclose KAMs related to property, plant and equipment where the models predict disclosure (predicted probability in the top quintile) in the current year, and 0 otherwise.
missed_kam_goodwill	Indicator variable. It is 1 if the auditor does not disclose KAMs related to goodwill where the models predict disclosure (predicted probability in the top quintile) in the current year, and 0 otherwise.
missed_kam_inventory	Indicator variable. It is 1 if the auditor does not disclose KAMs related to inventory where the models predict disclosure (predicted probability in the top quintile) in the current year, and 0 otherwise.
missed_kam_receivables	Indicator variable. It is 1 if the auditor does not disclose KAMs related to receivables where the models predict disclosure (predicted probability in the top quintile) in the current year, and 0 otherwise.
missed_kam_revenue	Indicator variable. It is 1 if the auditor does not disclose KAMs related to revenue where the models predict disclosure (predicted probability in the top quintile) in the current year, and 0 otherwise.
missed_kam_rpt	Indicator variable. It is 1 if the auditor does not disclose KAMs related to related party transactions where the models predict disclosure (predicted probability in the top quintile) in the current year, and 0 otherwise.
missed_kam_other	Indicator variable. It is 1 if the auditor does not disclose KAMs related to other subject areas where the models predict disclosure (predicted probability in the top quintile) in the current year, and 0 otherwise.
description_wordcount	The total word count of all KAMs descriptions in the firm’s auditor report in the current year.
response_wordcount	The total word count of all auditor responses to KAMs in the firm’s auditor report in the current year.
description_riskwordcount	The risk word count of all KAMs descriptions in the firm’s auditor report in the current year.
response_riskwordcount	The risk word count of all auditor responses to KAMs in the firm’s auditor report in the current year.
footnote_count	The number of financial statements footnotes referenced in KAMs in the firm’s auditor report in the current year.
low_kam_effort	Indicator variable. It is 1 if kam_effort is in the bottom quartile, and 0 otherwise. Following Goh et al. (2023), kam_effort is defined as the average of the following ratio: the number of words in each KAM auditor response divided by the number of words in each KAM description.
Independent variable of interest
soe	Indicator variable. It is 1 if a firm is controlled ultimately by the government or government agencies, and 0 otherwise.
Control variables
size	The natural logarithm of total assets in the current year.
lev	The value of total debts in the current year divided by total assets in the current year.
roa	Return on assets, defined as net profit in the current year divided by total assets in the current year.
loss	Indicator variable. It is 1 if net profit in the current year is less than zero, and 0 otherwise.
rev_grow	Sales in the current year minus sales in the last year divided by sales in the last year.
current	Total current assets in the current year divided by total current liabilities in the current year.
in_ar	Inventory in the current year plus receivables in the current year divided by total assets in the current year.
absdacc	Absolute value of Kothari et al. (2005) performance matched discretionary accruals in the current year.
ocf	Cash from operations in the current year divided by total assets in the current year.
std_ocf	The standard deviation of ocf for the current and previous two years.
ah	Indicator variable. It is 1 if the firm is dual-listed in the mainland and Hong Kong in the current year, and 0 otherwise.
bm	Book to market ratio, defined as total assets in the current year divided by the market value of the firm in the current year.
stock_issue	Indicator variable. It is 1 if the company engages in stock issuance in the current year, and 0 otherwise.
ma	Indicator variable. It is 1 if the company engages in M&A in the current year, and 0 otherwise.
restructure	Indicator variable. It is 1 if the company engages in restructuring in the current year, and 0 otherwise.
impairloss	Indicator variable. It is 1 if the company reports any type of impairment loss in the current year, and 0 otherwise.
ppe_toa	PPE in the current year divided by total assets in the current year.
lnage	The natural logarithm of the number of years that the firm has been founded.
pd	Probability to default. NUS PD risk measure for 12-month forecast horizon.
auditorchange	Indicator variable. It is 1 if the auditor is newly appointed in the current year, and 0 otherwise.
auditortenure	The natural logarithm of the number of years of the audit firm-client relationship.
aoqualified	Indicator variable. It is 1 if the auditor issues a qualified opinion in the current year, and 0 otherwise.
restate	Indicator variable. It is 1 if the firm reports a restatement in current year, and 0 otherwise.
weak	Indicator variable. It is 1 if the firm has a material weakness in current year, and 0 otherwise.
lnar_wc	The natural logarithm of annual report word count in the current year.
dual	Indicator variable. It is 1 if the Chairman and the CEO of the firm are the same person in the current year, and 0 otherwise.
largestholder	The percentage of equity held by the largest shareholder in the current year.
indedirector	The percentage of independent directors on the board in the current year.
bodmgtstholder	The percentage of equity held by the board of directors, the supervisory board, and executives in the current year.
iistholder	The percentage of equity held by institutional investors in the current year.
goodwill	Goodwill in the current year divided by total assets in the current year.
inventory	Inventory in the current year divided by total assets in the current year.
receivables	Accounts receivable in the current year divided by total assets in the current year.
revenue	Revenue in the current year divided by total assets in the current year.
total_rpt_amt	Total amount of related party transactions in the current year divided by total assets in the current year.
Cross-sectional analyses variables
high_largestholder	Indicator variable. It is 1 if the firm’s largest shareholder’s shareholding in the current year is higher than or equal to the median of the SOE subsample, and 0 otherwise.
strg_pubwelfare	Indicator variable. Following Wei et al. (2017) and Jiang (2021), it is 1 if the firm is in a strategic or public welfare industry (e.g., utilities), and 0 otherwise.
rp_lend	Indicator variable. It is 1 if the firm reports any intercorporate loans from the firm to related parties in the current year, and 0 otherwise.
rp_progrt	Indicator variable. It is 1 if the firm reports any guarantees provided by the firm to related parties in current year, and 0 otherwise.
Endogeneity analyses variables
reg_ind	Indicator variable. It is1 if the firm is from the following industry: Mining, railroads, trucking, airlines, telecommunications, energy supply, media and technology; and 0 otherwise.

Author Biographies

Annita Florou is Professor of Accounting at Bocconi University. She is also the Academic Advisor to the Institute of Chartered Accountants in England and Wales (ICAEW); Editor of Journal of Business, Finance and Accounting; and an Editorial Board Member of Accounting and Business Research, British Accounting Review, Contemporary Accounting Research, Corporate Governance: An International Review and Journal of Accounting and Public Policy. Her recent research interests cover auditor reporting and regulation as well as sustainability reporting and assurance.

Xiaoxi Wu is an Assistant Professor of Accounting at Bocconi University. Her research interests include corporate disclosure, managerial financial and non-financial communication, and audit reporting. She is also interested in the application of inter-disciplinary theories and methods.

Shuai Yuan is an Assistant Professor of Accounting at University of Nottingham Ningbo, China. His research interests include auditor regulation, corporate governance and corporate finance. He also engages in inter-disciplinary research.

Vincent (Qiru) Zhang is an Assistant Professor of Accounting at Santa Clara University. His research investigates how accounting information helps investors make decisions in the municipal bond market. His papers also examine the impact of regulatory changes and accounting/auditing standard changes.

References

Abdelfattah

Elmahgoub

Elamer

A. A.

(2020). Female audit partners and extended audit reporting: UK evidence. Journal of Business Ethics, 174(1), 177–197. https://doi.org/10.1007/s10551-020-04607-0

Aharony

Lee

C. W. J.

Wong

T. J.

(2000). Financial packaging of IPO firms in China. Journal of Accounting Research, 38(1), 103–126. https://doi.org/10.2307/2672924

Bai

C.-E.

Liu

Song

F. M.

(2005). Bad news is good news: Propping and tunnelling evidence from China. University of Hong Kong, Working paper.

Bédard

Gonthier-Besacier

Schatt

(2018). Consequences of expanded audit reports: Evidence from the justifications of assessments in France. Auditing: A Journal of Practice & Theory, 38(3), 23–45. https://doi.org/10.2308/ajpt-52339

Beijing Institute of Certified Public Accountants . (2018). Expert tips from the Professional Technical Committee of the Beijing Institute of Certified Public Accountants [2018] No. 9- Issues to be concerned about in determining and communicating key audit matters. https://bicpa.org.cn/p1/tzgg/20181101/45397.html

Bortolotti

Faccio

(2009). Government control of privatized firms. Review of Financial Studies, 22(8), 2907–2939. https://doi.org/10.1093/rfs/hhn077

Bradshaw

Liao

M. S.

(2019). Agency costs and tax planning when the government is a major shareholder. Journal of Accounting and Economics, 67(2–3), 255–277. https://doi.org/10.1016/j.jacceco.2018.10.002

Burke

Hoitash

Xiao

(2023). The disclosure and consequences of U.S. critical audit matters. The Accounting Review, 98(2), 59–95. https://doi.org/10.2308/tar-2021-0013

Bushman

R. M.

Piotroski

J. D.

(2006). Financial reporting incentives for conservative accounting: The influence of legal and political institutions. Journal of Accounting and Economics, 42(1–2), 107–148. https://doi.org/10.1016/j.jacceco.2005.10.005

10.

Central Committee of the Communist Party of China . (2003). Decision of the Central Committee of the Communist Party of China on some issues concerning the improvement of the socialist market economy.

11.

Chan

K. H.

Lin

K. Z.

P. L. L.

(2006). A political–economic analysis of auditor reporting and auditor switches. Review of Accounting Studies, 11(1), 21–48. https://doi.org/10.1007/s11142-006-6394-z

12.

Chan

K. H.

Lin

K. Z.

Wang

R. R.

(2012). Government ownership, accounting-based regulations, and the pursuit of favorable audit opinions: Evidence from China. Auditing: A Journal of Practice & Theory, 31(4), 47–64. https://doi.org/10.2308/ajpt-50227

13.

Chang

Chen

Liao

(2014). What are the reliably important determinants of capital structure in China? Pacific-Basin Finance Journal, 30, 87–113. https://doi.org/10.1016/j.pacfin.2014.06.001

14.

Chang

Y. T.

Chi

Stone

D. N.

(2024). Is client-specific information useful to investors? Evidence from key audit matter reports. Journal of Accounting, Auditing and Finance, 39(3), 786–806. https://doi.org/10.1177/0148558x221091804

15.

Chen

C. J.

Zhao

Kong

C. U. o. H.

(2000). An emerging market’s reaction to initial modified audit opinions: Evidence from the Shanghai Stock Exchange. Contemporary Accounting Research, 17(3), 429–455. https://doi.org/10.1092/gcjp-5599-quwb-g86d

16.

Chen

J. Z.

Lobo

G. J.

Wang

(2010). Association between borrower and lender state ownership and accounting conservatism. Journal of Accounting Research, 48(5), 973–1014. https://doi.org/10.1111/j.1475-679x.2010.00385.x

17.

Chen

Cheng

Hao

Liu

(2020). GDP growth incentives and earnings management: Evidence from China. Review of Accounting Studies, 25(3), 1002–1039. https://doi.org/10.1007/s11142-020-09547-8

18.

Cheung

Y. L.

Jing

Rau

P. R.

Stouraitis

(2009). Tunneling and propping up: An analysis of related party transactions by Chinese listed companies. Pacific-Basin Finance Journal, 17(3), 372–393. https://doi.org/10.1016/j.pacfin.2008.10.001

19.

China Securities Regulatory Commission (CSRC) . (2017). Circular of the China Securities Regulatory Commission and the State-owned Assets Supervision and Administration Commission of the State Council on Several Issues Concerning the Standardization of Capital Exchanges between Listed Companies and Related Parties and the External Guarantees of Listed Companies. (In Chinese). https://www.csrc.gov.cn/tibet/c105486/c1304233/content.shtml

20.

Chinese Institute of Certified Public Accountants (CICPA) . (2016). Auditing Standard No. 1504 for Certified Public Accountants of China - Communication on Key Audit Matters in Audit Reports. (In Chinese). https://www.cicpa.org.cn/xxfb/news/201612/W020210118390501351623.pdf

21.

Christensen

B. E.

Glover

S. M.

Wolfe

C. J.

(2014). Do critical audit matter paragraphs in the audit report change nonprofessional investors’ decision to invest? Auditing: A Journal of Practice & Theory, 33(4), 71–93. https://doi.org/10.2308/ajpt-50793

22.

Dennis

S. A.

Griffin

J. B.

Zehms

K. M.

(2019). The value relevance of managers’ and auditors’ disclosures about material measurement uncertainty. The Accounting Review, 94(4), 215–243. https://doi.org/10.2308/accr-52272

23.

Tang

Young

S. M.

(2012). Influence activities and favoritism in subjective performance evaluation: Evidence from Chinese state-owned enterprises. The Accounting Review, 87(5), 1555–1588. https://doi.org/10.2308/accr-50196

24.

Fan

J. P. H.

Wong

T. J.

(2002). Corporate ownership structure and the informativeness of accounting earnings in East Asia. Journal of Accounting and Economics, 33(3), 401–425. https://doi.org/10.1016/s0165-4101(02)00047-2

25.

Fang

Lobo

G. J.

Zhang

Zhao

(2018). Auditing related party transactions: Evidence from audit opinions and restatements. Auditing: A Journal of Practice & Theory, 37(2), 73–106. https://doi.org/10.2308/ajpt-51768

26.

FORTUNE . (2021). Global 500. https://fortune.com/ranking/global500/2021/

27.

Goh

B. W.

Wang

(2023). Informativeness of the expanded audit report: Evidence from China. Auditing: A Journal of Practice & Theory. forthcoming.

28.

Guedhami

Pittman

J. A.

Saffar

(2009). Auditor choice in privatized firms: Empirical evidence on the role of state and foreign owners. Journal of Accounting and Economics, 48(2–3), 151–171. https://doi.org/10.1016/j.jacceco.2009.08.003

29.

Guedhami

Pittman

J. A.

Saffar

(2014). Auditor choice in politically connected firms. Journal of Accounting Research, 52(1), 107–162. https://doi.org/10.1111/1475-679x.12032

30.

Gul

F. A.

(2006). Auditors’ response to political connections and cronyism in Malaysia. Journal of Accounting Research, 44(5), 931–963. https://doi.org/10.1111/j.1475-679x.2006.00220.x

31.

Gul

F. A.

Kim

J. B.

Qiu

A. A.

(2010). Ownership concentration, foreign shareholding, audit quality, and stock price synchronicity: Evidence from China. Journal of Financial Economics, 95(3), 425–442. https://doi.org/10.1016/j.jfineco.2009.11.005

32.

Gul

F. A.

Sami

Zhou

(2009). Auditor disaffiliation program in China and auditor independence. Auditing: A Journal of Practice & Theory, 28(1), 29–51. https://doi.org/10.2308/aud.2009.28.1.29

33.

Gutierrez

Minutti-Meza

Tatum

K. W.

Vulcheva

(2018). Consequences of adopting an expanded auditor’s report in the United Kingdom. Review of Accounting Studies, 23(4), 1543–1587. https://doi.org/10.1007/s11142-018-9464-0

34.

Habib

Jiang

Zhou

(2015). Related party transactions and audit fees: Evidence from China. Journal of International Accounting Research, 14(1), 59–83. https://doi.org/10.2308/jiar-51020

35.

Hainmueller

(2012). Entropy balancing for causal effects: A multivariate reweighting method to produce balanced samples in observational studies. Political Analysis, 20(1), 25–46. https://doi.org/10.1093/pan/mpr025

36.

Pan

Tian

G. G.

(2017). Political connections, audit opinions, and auditor choice: Evidence from the ouster of government officers. Auditing: A Journal of Practice & Theory, 36(3), 91–114. https://doi.org/10.2308/ajpt-51668

37.

Pan

Tian

G. G.

Cai

(2022a). How does reciprocal rent-seeking between politicians and auditors influence audit quality? Evidence from China. Accounting Horizons, 36(3), 103–126. https://doi.org/10.2308/horizons-19-118

38.

Y. X.

Q. C.

Ding

C. C.

(2022b). Research on corporate multinational investment decisions from the perspective of uncertainty perception (an textual analysis based on listed firms’ annual reports). World Economy Studies, 7(59–75), 136.

39.

Heckman

J. J.

(1979). Sample selection bias as a specification error. Econometrica, 47(1), 153–161. https://doi.org/10.2307/1912352

40.

Hope

O. K.

Yue

Zhong

(2020). China’s anti‐corruption campaign and financial reporting quality. Contemporary Accounting Research, 37(2), 1015–1043. https://doi.org/10.1111/1911-3846.12557

41.

Jia

Shi

Wang

(2013). Coinsurance within business groups: Evidence from related party transactions in an emerging market. Management Science, 59(10), 2295–2313. https://doi.org/10.1287/mnsc.1120.1703

42.

Jian

Wong

T. J.

(2010). Propping through related party transactions. Review of Accounting Studies, 15(1), 70–105. https://doi.org/10.1007/s11142-008-9081-4

43.

Jiang

Lee

C. M.

Yue

(2010). Tunneling through intercorporate loans: The China experience. Journal of Financial Economics, 98(1), 1–20. https://doi.org/10.1016/j.jfineco.2010.05.002

44.

Jiang

X. H.

(2021). Classification reform of mixed ownership in state-owned enterprises and capital allocation efficiency. Contemporary Finance & Economics, 0(7), 127–137. https://www.cnki.com.cn/Article/CJFDTotal-DDCJ202107012.htm

45.

Jin

Xin

Adhikari

(2022). Political governance in China’s state-owned enterprises. China Journal of Accounting Research, 15(2), Article 100236. https://doi.org/10.1016/j.cjar.2022.100236

46.

Kachelmeier

S. J.

Rimkus

Schmidt

J. J.

Valentine

(2020). The forewarning effect of critical audit matter disclosures involving measurement uncertainty. Contemporary Accounting Research, 37(4), 2186–2212. https://doi.org/10.1111/1911-3846.12583

47.

Klevak

Livnat

Pei

D. S.

Suslava

(2023). Critical audit matters: Possible market misinterpretation. Auditing: A Journal of Practice & Theory, 42(3), 45–70. https://doi.org/10.2308/ajpt-2020-113

48.

Kothari

S. P.

Leone

A. J.

Wasley

C. E.

(2005). Performance matched discretionary accrual measures. Journal of Accounting and Economics, 39(1), 163–197. https://doi.org/10.1016/j.jacceco.2004.11.002

49.

Lennox

J. S.

(2022). A review of China-related accounting research in the past 25 years. Journal of Accounting and Economics, 74(2–3), Article 101539. https://doi.org/10.1016/j.jacceco.2022.101539

50.

Lennox

Zhang

(2016). The effect of audit adjustments on earnings quality: Evidence from China. Journal of Accounting and Economics, 61(2–3), 545–562. https://doi.org/10.1016/j.jacceco.2015.08.003

51.

Lennox

C. S.

Francis

J. R.

Wang

(2012). Selection models in accounting research. The Accounting Review, 87(2), 589–616. https://doi.org/10.2308/accr-10195

52.

Lennox

C. S.

Schmidt

J. J.

Thompson

A. M.

(2023). Why are expanded audit reports not informative to investors? Evidence from the United Kingdom. Review of Accounting Studies, 28(2), 497–532. https://doi.org/10.1007/s11142-021-09650-4

53.

Zhou

L.-A.

(2005). Political turnover and economic performance: The incentive role of personnel control in China. Journal of Public Economics, 89(9–10), 1743–1762. https://doi.org/10.1016/j.jpubeco.2004.06.009

54.

Z. Q.

Sun

(2023). Disclosure of key audit matters, media attention, and risk of stock price collapse. Communication of Finance and Accounting, 909(1), 27–31. https://qikan.cqvip.com/Qikan/Article/Detail?id=7108716765

55.

Liao

Minutti-Meza

Zhang

Zou

(2022). Consequences of the adoption of the expanded auditor’s report: Evidence from Hong Kong. University of Miami Business School Research Paper, No. 3392449.

56.

Lin

K. J.

Zhang

Zheng

(2020). State-owned enterprises in China: A review of 40 years of research and practice. China Journal of Accounting Research, 13(1), 31–55. https://doi.org/10.1016/j.cjar.2019.12.001

57.

Liu

Wang

J. Q.

Qiu

M. Q.

(2022). Research on the impact of risk information disclosure of customers’ annual report on cash holding decisions of enterprises. Chinese Journal of Management, 19(12), 1863–1873. https://www.cnki.com.cn/Article/CJFDTotal-GLXB202212015.htm

58.

Liu

Z. J.

(2007). Corporate governance and earnings management in the Chinese listed companies: A tunneling perspective. Journal of Corporate Finance, 13(5), 881–906. https://doi.org/10.1016/j.jcorpfin.2007.07.003

59.

Milhaupt

C. J.

Pargendler

(2018). Related party transactions in state-owned enterprises: Tunneling, propping, and policy channeling. In The Law and Finance of Related Party Transactions. Cambridge University Press.

60.

Ministry of Finance of the People’s Republic of China (MOF) . (2016). Auditing Standard No. 1504 for Certified Public Accountants of China - Communication on Key Audit Matters in Audit Reports. https://www.mof.gov.cn/zhengwuxinxi/zhengcefabu/201612/W020161229380915020654.pdf

61.

Minutti-Meza

(2021). The art of conversation: The expanded audit report. Accounting and Business Research, 51(5), 548–581. https://doi.org/10.1080/00014788.2021.1932264

62.

NPC . (2020). The state council’s decision on the supervision of enterprises in the state owned assets supervision system for the year 2019 special report on the management of state-owned assets—At the 22nd meeting of the Standing Committee of the 13th National People’s Congress on October 15, 2020 (In Chinese).

63.

Organisation for Economic Co-operation and Development (OECD) . (2014). The size and sectoral distribution of SOEs in OECD and partner countries. OECD Publishing.

64.

Piotroski

J. D.

Wong

T. J.

(2012). Institutions and information environment of Chinese listed firms. In Capitalizing China (pp. 201–246). University of Chicago Press.

65.

Piotroski

J. D.

Wong

T. J.

Zhang

(2015). Political incentives to suppress negative information: Evidence from Chinese listed firms. Journal of Accounting Research, 53(2), 405–459. https://doi.org/10.1111/1475-679x.12071

66.

Porumb

V. A.

Zengin‐Karaibrahimoglu

Lobo

G. J.

Hooghiemstra

De Waard

(2021). Expanded auditor’s report disclosures and loan contracting. Contemporary Accounting Research, 38(4), 3214–3253. https://doi.org/10.1111/1911-3846.12697

67.

Rajan

R. G.

Zingales

(2003). The great reversals: The politics of financial development in the twentieth century. Journal of Financial Economics, 69(1), 5–50. https://doi.org/10.1016/s0304-405x(03)00125-9

68.

Rapley

E. T.

Robertson

J. C.

Smith

J. L.

(2021). The effects of disclosing critical audit matters and auditor tenure on nonprofessional investors’ judgments. Journal of Accounting and Public Policy, 40(5), Article 106847. https://doi.org/10.1016/j.jaccpubpol.2021.106847

69.

Reid

L. C.

Carcello

J. V.

Neal

T. L.

Francis

J. R.

(2019). Impact of auditor report changes on financial reporting quality and audit costs: Evidence from the United Kingdom. Contemporary Accounting Research, 36(3), 1501–1539. https://doi.org/10.1111/1911-3846.12486

70.

Sierra-García

Gambetta

García-Benau

M. A.

Orta-Pérez

(2019). Understanding the determinants of the magnitude of entity-level risk and account-level risk key audit matters: The case of the United Kingdom. The British Accounting Review, 51(3), 227–240. https://doi.org/10.1016/j.bar.2019.02.004

71.

State Council of the People's Republic of China . (2020). Opinion on speeding up the improvement of the socialist market economy in the new era.

72.

Sun

D. J.

J. Q.

(2024). Analysis of the problem of capital occupation by major shareholders of the company - based on the perspective of investor protection. Times of Economy & Trade, 572(9), 132–134. http://www.sdjmqk.com.cn/qikan_detail.asp?id=8576

73.

Tullock

(1987). Autocracy. Kluwer Academic Publishers.

74.

Wang

Q. K.

(2023). The contagion of firm’s perception of economic policy uncertainty: Based on the textual analysis of annual reports and interlocking board networks. China Journal of Econometrics, 3(2), 548–569. https://doi.org/10.12012/CJoE2022-0085

75.

Wang

Wong

T. J.

Xia

(2008). State ownership, the institutional environment, and auditor choice: Evidence from China. Journal of Accounting and Economics, 46(1), 112–134. https://doi.org/10.1016/j.jacceco.2008.04.001

76.

Wang

X. Y.

Zeng

(2019). Risk disclosure in annual reports and bank loan interest rate. Journal of Financial Research, 463(1), 54–71. http://www.jryj.org.cn/EN/Y2019/V463/I1/54

77.

Wei

(2021). State ownership and target setting: Evidence from publicly listed companies in China. Contemporary Accounting Research, 38(3), 1925–1960. https://doi.org/10.1111/1911-3846.12665

78.

Wei

Cai

Liu

(2017). Research on governance of Chinese state-owned listed companies based on classification. Journal of Sun Yat-sen University (Social Science Edition), 57(4), 175–192. in Chinese. https://www.cnki.com.cn/Article/CJFDTotal-ZSDS201704019.htm

79.

Wong

T. J.

(2016). Corporate governance research on listed firms in China: Institutions, governance and accountability. Foundations and Trends® in Accounting, 9(4), 259–326. https://doi.org/10.1561/1400000039

80.

Yan

M. M.

Zhuang

F. P.

(2021). Analysis and evaluation of the current situation of key audit matters disclosure: Based on the experience data of Shanghai and Shenzhen Stock Exchanges in the first year of comprehensive implementation of the new audit report standards. Accounting and Finance Communication, 111–116+121.

81.

Yang

Tang

Kilgore

Yi Hong

(2001). Auditor-government associations and auditor independence in China. The British Accounting Review, 33(2), 175–189. https://doi.org/10.1006/bare.2001.0162

82.

Yang

(2013). Do political connections add value to audit firms? Evidence from IPO audits in China. Contemporary Accounting Research, 30(3), 891–921. https://doi.org/10.1111/j.1911-3846.2012.01177.x

83.

Zeng

Zhang

J. H.

Zhang

(2021). Key audit matters reports in China: Their descriptions and implications of audit quality. Accounting Horizons, 35(2), 167–192. https://doi.org/10.2308/horizons-19-189

84.

Zhang

D. L.

(2023). Does auditor’s perception of fraud affect the disclosure of key audit matters? Audit and Economic Research, 38(1), 31–39. https://www.cnki.com.cn/Article/CJFDTotal-SJYJ202301004.htm

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.31 MB