![Sample our Economics, Finance,Business & Industry journals, sign in here to start your access, latest two full volumes FREE to you for 14 days]({"type":"image" , "src" : "/sda/5931/TOC-Subject-Sample-2021-Economics-Finance-Business-Industry.jpg"})
Abstract
In December 2010, Hong Kong regulators allowed 12 mainland Chinese registered audit firms to audit companies incorporated in mainland China and listed in Hong Kong (H-shares). In this study, we examine whether access to the Hong Kong audit market improves the quality of audits conducted by these Chinese audit firms for clients listed in the mainland markets (A-shares). Using data from 2008 to 2016, we find that mainland auditors with H-share clients provide higher quality audits, as measured by more modified audit opinions, higher audit fees and less earnings management, than auditors without H-share clients. This effect is more pronounced when the auditors are non-Big 4 firms and the clients are listed only in mainland China. Further analysis shows that A-share investors react positively to the initial announcement that mainland auditors are permitted to conduct H-share audits. Overall, our findings suggest that H-share audits have a positive spillover effect on the quality of A-share audits.
Acknowledgement
We thank the guest editors of the ABR special issue and two reviewers for their many helpful comments and suggestions.
Disclosure statement
No potential conflict of interest was reported by the author(s).
Notes
1 We use the terms ‘auditor’ and ‘firm’ interchangeably for the audit firm and the terms ‘company’ and ‘client’ interchangeably for the listed company.
2 Initially, 12 audit firms were included on the approved list, i.e. PwC, Deloitte, EY, KPMG, Lixin, Dahua, Tianjian, ShineWing, Crowe Horwath, Jingdu Tianhua, Zhongrui Yuehua and Daxin. Jingdu Tianhua changed its name to Zhitong in 2012. In 2013, Zhongrui Yuehua and Crowe Horwath merged and the company name was changed to Ruihua. Since then, 11 firms have qualified for H-share audits.
3 Under the dual-audit regime, one AH company is required to have two auditors; the A-share financial statements using CAS are audited by a mainland auditor, and the H-share statements using HKAS or IAS are audited by a Hong Kong auditor.
4 Ke et al. (Citation2015) provide a detailed comparison of the institutions affecting audit quality in Hong Kong and mainland China. They explicitly attribute the spillover effect to the fact that Hong Kong’s stronger institutions induce higher quality audits from Hong Kong auditors.
5 Wang and Xin (Citation2011) show that 69.44% of AH companies hire the two auditors from the same Big 4 firm.
6 Tian et al. (Citation2017) and Zhang et al. (Citation2019) also examine the effect of cancelling the mandatory dual-audit system on the quality of A-share audits. They generally find that A-share audit quality is higher for AH companies that keep dual audits and lower for AH companies that forego dual audits, indicating that the spillover effect arising from dual audits continues in the post-policy change period.
7 The CICPA announced a comprehensive strategy to accelerate the development of China’s CPA industry in 2007, and the MOF formally endorsed it in 2009 (‘Document 56’ MOF Citation2009).
8 In additional and sensitivity analyses, we use a constant sample of auditors to control for the quality effect of audit firm mergers and include an indicator variable, LLP, in our models to control for the potential confounding effect of LLP transformation. We do not consider the quality effect of joining international accounting membership networks, as past research finds no evidence that such membership affects audit quality (Mao et al. Citation2017). We include auditor indicators in our models.
9 Since 2010, almost all of the H-share financial statements audited by mainland auditors have been reviewed by Hong Kong regulators, although they are unable to inspect these mainland auditors. Thus far, no cases of material non-compliance with the accounting and auditing requirements have been identified by the reviewers (FRC Citation2010–Citation2017).
10 One example of this is the positive spillover effect from Hong Kong auditors to mainland auditors under the dual-audit system for AH companies (Ke et al. Citation2015, Wang and Xin Citation2011, Lin et al. Citation2014, Tian et al. Citation2017, Zhang et al. Citation2019).
11 Due to the lack of an independent regulator for the auditors of listed companies, Hong Kong was removed from a list of jurisdictions deemed to have regulatory equivalency with the European Commission (EC) in 2013 (Gillis Citation2018). To obtain EC equivalence, Hong Kong amended its law in 2019, granting the FRC the powers of inspection, investigation and sanction.
12 Gillis (Citation2011) summarises the development of this scheme and analyses the short-term and long-term effects of this scheme on the auditing profession in Hong Kong and mainland China. Two recent empirical studies (Liu and Lin Citation2019, Fan et al. Citation2020) examine the influence of this scheme on H-share companies’ audit fees and audit quality. In contrast, our study examines the impact on A-share companies’ audit quality following mainland auditors’ access to H-share audits.
13 All are AH companies. Shandong Molong Petroleum Machinery (002490.SZ and 0568.HK) hired Deloitte, Dongjiang Environmental (002672.SZ and 0895.HK) hired ShineWing, Tsingtao Brewery (600600.SH and 0168.HK) hired PwC and Chongqing Iron and Steel (601005.SH and 1053.HK) hired KPMG. Previously, their H-share auditors were the Hong Kong affiliates of Deloitte, ShineWing, PwC and KPMG, respectively.
14 For example, following the investigation of Ruihua by mainland regulators in July 2019, the number of A-share companies audited by Ruihua plummeted from 246 in 2018 to 23 in 2019 and all of its six H-share clients announced the termination of their contracts with Ruihua (according to the CSMAR database and HKEX website).
15 We simply include all the non-treatment client-years in the test. To better control for potential endogeneity, we also select the control group using conventional matching by auditor size and year and by client industry, year and size. The results are reported in Section 4.2.2.
16 Zhitong is eligible for H-share audits but had no H-share clients in the 2010–2016 period. Our inferences do not change when we exclude the client-years audited by Zhitong from our sample.
17 As a sensitivity check, following Krishnan et al. (Citation2017), we also include ACCESS in the regression models. Our inferences remain unchanged.
18 In an untabulated analysis, we adopt a typical difference-in-differences design by including the variables POST, ACCESS and POST*ACCESS in the test, where POST is an indicator variable for A-share client-years in 2010 and onward. The interaction variable POST*ACCESS represents the incremental quality effect of the approved auditors in the post-2010 period when they have H-share audit qualifications. The result shows a significant increase in audit quality post-2010 but not a significant interaction between POST and ACCESS, indicating that gaining H-share audit qualification itself has no incremental influence on A-share audit quality. Thus, we define POSTACCESS in our test as being engaged by H-share clients, i.e. gaining actual access to H-share clients.
19 Audit fees are generally regarded as a weak measure of audit quality, particularly if multiple reforms take place during the sample period that could influence fees. In additional analyses, we use restatements, going-concern opinions and sanctions against auditors to strengthen the measures of audit quality.
20 As a sensitivity check, we also re-calculate the signed discretionary accruals using the modified Jones model (Dechow et al. Citation1995, Jones et al. Citation2008) and re-estimate Equation (1) for various samples. Our inferences remain unchanged.
21 To avoid potential misinterpretations of the coefficient of the interaction term in non-linear models (Ai and Norton Citation2003), following Lamoreaux (Citation2016) and Fung et al. (Citation2017), we use a linear probability model to estimate an auditor’s propensity to issue MAOs. The problem of predicted values falling outside [0,1] is not common in our sample. For example, the predicted value of MAO in the estimation of falls within [-0.09, 0.75]. We also run a logistic regression for the MAO and find the coefficient on POSTACCESS to be positive and significant.
22 In an untabulated analysis, we also estimate the MDD and PMATCHP models using income increasing discretionary accruals (i.e. MDD > 0, PMATCHP > 0) and find that the coefficient on POSTACCESS is negatively associated with PMATCHP, significant at the 10% level, but insignificantly associated with MDD. Similar results are also obtained for the matched samples.
23 For brevity, we present only the coefficients on POSTACCESS and other variables of interest in the tables that follow.
24 In an untabulated analysis, following the literature (Ke et al. Citation2015, Lamoreaux Citation2016, Fung et al. Citation2017, Krishnan et al. Citation2017), we also select the control group using the propensity score of hiring an auditor with H-share qualifications. We find similar results but with insignificant POSTACCESS for the MAO models.
25 Zheng et al. (Citation2011) report ratios of 72% in 2008, 67% in 2009 and 63% in 2010. The ratios for two audits by the Hong Kong and mainland affiliates of the same non-Big 4 firm are only 12%, 15% and 17%, respectively.
26 Year indicators are not included due to multilinearity problems.
27 We compute the daily abnormal returns (AR) using the market-adjusted model, ARit = Rit – (αi + βi * Rmt), where Rit is the return for company i on day t, Rmt is the return on the value-weighted market returns on day t, and α and β are, respectively, the intercept and beta coefficients for company i, estimated over 60 trading days before the event window starts. We then accumulate the daily AR to obtain the CAR using short event windows that range from 2 days before to 2 days after the event date. Patell’s (Citation1976) t-statistic is used to examine whether the average standardised CARs are significantly different from zero.
28 One of the potential benefits to H-share companies of forgoing dual audits is a reduction in audit costs, which is supposed to be good news for H-share companies. However, the negative market reaction suggests that the investors’ concerns about the quality of mainland firms’ audits overshadow the benefits of the cost reduction. Liu and Lin (Citation2019) find the audit fee saving after switches to mainland auditors to be relatively trivial.
29 In sensitivity tests, we consider an alternative design to control for auditors’ expectation of access to H-share audits.
30 We also use 2010 and 2011 as the first access years and our results remain unchanged.
31 The data are collected from the annual ranking list of CICPA’s Top 100 accounting firms.
32 For SameAuditor and SOE, we use ratios, i.e. the number of clients with SameAuditor = 1 (SOE = 1) divided by the total number of clients for a given auditor each year. We obtain 103 auditor-year observations, 53 of which have POSTACCESS = 1.
33 As a sensitivity test, the predicted values from Equation (2) are used as a proxy for auditors’ choice of actual access, Prob(POSTACCESS). We then re-estimate Equation (1) on the full sample with Prob(POSTACCESS) included as an additional independent variable. Our results remain unchanged.
34 Gillis (Citation2011) posits that mainland regulators closely monitor the qualified H-share auditors to avoid scandals that would have long-term implications for the mainland audit profession.
35 The data are collected from https://pcaobus.org/Registration/Firms/Pages/RegisteredFirms.aspx#k=, accessed 18 March 2020. None of the firms have disciplinary proceedings or inspection reports. Among the 12 approved auditors with H-share qualifications, only ShineWing and Daxin have no PCAOB status.
36 We add DUAL, TOP1SHR and an indicator variable, LMAO, in the regression of the restatement and sanction models. LMAO equals one if the client received an MAO in the previous year, and zero otherwise. Standard errors are not adjusted for clustering on the client companies.
37 For example, in June 2012, Jingdu Tianhua (eligible for H-share audits) merged with Tianjian Zhengxin (ineligible for H-share audits) and the company name was changed to Zhitong. In May 2013, Zhongrui Yuehua and Crowe Horwath (both eligible for H-share audits) merged and the company name was changed to Ruihua. In 2008, Zhongshen Yatai formed as the combination of two firms that were both ineligible for H-share audits. Thus, we exclude the client-years audited by the above firms.
38 For the auditors that began H-share audits in 2010, the sample period is five years, from 2008 to 2012, covering the two years preceding and two years following the access year. Similarly, for the firms that began H-share audits in 2011 (2012), the sample period is seven (nine) years from 2008 to 2014 (2008 to 2016), covering the three (four) years preceding and following the access year. No restrictions are applied to the remaining auditors.