Abstract
CHINA JOURNAL OF ACCOUNTING STUDIES 2019, VOL. 7, NO. 2, 170–183 https://doi.org/10.1080/21697213.2019.1676037 ARTICLE Do critical audit matters signal higher quality of audited financial information? Evidence from asset impairment a a b Xi Wu , Yujiang Fan and Yulong Yang a b School of Accountancy, Central University of Finance and Economics, Beijing, China; College of Management and Economics, Tianjin University, Tianjin, China ABSTRACT KEYWORDS Audit report; critical auditing Chinese auditing standards mandate the disclosure of critical audit matter; risk-oriented matters (CAMs) in audit reports for all listed companies since 2017. auditing; asset impairment Risk-oriented auditing requires auditors to assess material misstate- ment risks and provide reasonable assurance on financial state- ments that should reflect the firm’s underlying economics, regardless whether a CAM is disclosed. However, given a material misstatement risk, if some auditors effectively identify it as a CAM while others do not, the financial information with a CAM would exhibit higher quality than that without a CAM, leading to a positive association between CAM disclosure and the audited information quality. Using asset impairment-related CAMs, we show the relation between asset impairment and worsened economics is notably stronger for companies with an impairment-related CAM than those without any. Further, this association is more pronounced in smaller audit firms. Our findings reveal inadequate implementation of risk-oriented auditing, particularly in audit firms with greater resource constraints. 1. Introduction How should the information users interpret the critical audit matters (CAMs) disclosed under the new auditing standard? This becomes a problem because information users usually focus on the quality of audited accounting information rather than the material matters identified by the auditors throughout the audit process, like CAMs. Risk-oriented auditing principles require auditors to provide reasonable assurance regarding the poten- tial risks behind disclosed CAMs. In other words, does it make a substantive difference to information users to publicly disclose CAMs originated from the audit process? Fiscal year 2017 was the first year that the new auditing standard was mandated for all Chinese A-share listed firms. Although A + H share firms had piloted the new auditing standard in 2016, the test sample is rather limited. Up to the end of April 2018, more than 3,400 audit reports provide rich data for large-sample analysis. According to our manual work, CONTACT Yulong Yang yulongyang@tju.edu.cn College of Management and Economics, Tianjin University, Tianjin, China Paper accepted by Kangtao Ye. This article has been republished with minor changes. These changes do not impact the academic content of the article. © 2019 Accounting Society of China CHINA JOURNAL OF ACCOUNTING STUDIES 171 78% of A-share firms mentioned at least one asset impairment-related CAM, and this category of CAMs accounts for 45% of all 7,217 CAMs disclosed by A-share firms in fiscal year 2017. Hence, focusing on asset impairment-related CAMs, this paper examines whether the disclosure of such CAMs in audited reports signals higher-quality audited financial infor- mation. The quality of information with respect to asset impairment lies in the extent to which such impairment reflects an underlying deterioration in the economic condition of the firm (Francis, Hanna, & Vincent, 1996; Lobo, Paugam, Zhang, & Casta, 2017; Riedl, 2004): the stronger the association between asset impairment and underlying degraded economic conditions, the more the accounting information reflects the economic funda- mentals of the firm, that is, the higher the information quality of asset impairment. Regardless of whether asset impairment-related CAMs are disclosed, there should be no significant difference in the quality of audited information regarding asset impairment if all audit firms conduct their audits under risk-oriented auditing principles, which require auditors to identify all material misstatement risks and provide reasonable assurance regarding those risks when the audits are completed. In such a case, there should be no association between asset impairment-related CAM disclosure and the quality of the audited asset impairment information. However, if a subset of audit firms implement risk- oriented auditing principles inadequately by allocating more audit resources to asset impairment matters identified as CAMs but do not provide sufficient, appropriate audit evidence for the non-CAM asset impairments, it possibly leads to CAM asset impairment information better representing worsened underlying economics, while non-CAM asset impairment information could suffer from potential misstatement, such as an inadequate impairment estimate, in the completed audit. To summarise, if risk-oriented auditing is only partially implemented among auditors, the disclosure of asset impairment-related CAMs will be positively correlated to the information quality of audited asset impairment. Empirical evidence supports this prediction. Disclosure of asset impairment-related CAMs is significantly positively correlated with the information quality of audited asset impair- ment, indicating insufficient implementation of risk-oriented auditing principles. The rela- tionship is more pronounced for smaller audit firms that face greater resource constraints. Implications for the way information users interpret CAMs are twofold. First, the infor- mation quality of non-CAM asset impairment matters is relatively low compared with asset impairment matters identified as CAMs. Second, the information quality of non-CAM asset impairment matters is lower when the incumbent auditors are relatively small. By focusing on these issues, this study extends the existing literature on the new auditing reporting standards and deepens our understanding of the informational implications of CAMs. The remainder of this study is organised as follows: Section II provides background information and reviews existing research about the new auditing standard. Section III develops our hypotheses and Section IV describes the research design. Section V reports our main results as well as supplementary analysis. Section VI concludes. 2. Institutional background 2.1. New auditing standard The content and form of audit reports have always been controversial, even though prior studies find that traditional audit opinions do have information value (Church, Davis, & 172 X. WU ET AL. McCracken, 2008;Tang, 2015; Zhang, Cai, & Liu, 2016). Oneofthe focal pointsofthe controversy is the highly formatted ‘Clean/Modified’ opinion audit report standard. Given that the vast majority of firms receive a clean opinion, the limited amount of information incorporated in these audit reports cannot satisfy the growing need for more information (IAASB, 2013). Accordingly, some developed countries (e.g. the U.K.) adopted, and others (e.g. the U.S.) proposed adopting a new auditing standard in 2013, while the International Auditing and Assurance Standards Board (IAASB) issued a revised auditing standard (ISA701) in January 2015. Auditors are now required to disclose CAMs in audit reports under the new auditing standard. Specifically, a matter should be identified as a CAM if it involves material misstatement or fraud, significant judgement and estimation, or it involves significant value. In 2016, China revised its auditing standard (Auditing Standard No. 1504 for Certified Public Accountants of China – Communication on Critical Audit Matters in Audit Reports), requiring A + H share listed firms to disclose CAMs in fiscal year 2016, and requiring all A-share listed firms to implement the new auditing standard in fiscal year 2017. We manually collected audited reports on these firms and identified 7,217 CAMs disclosed by 3,449 firms in fiscal year 2017 (through 30 April 2018). Each firm disclosed an average of two CAMs approximately in this period (the maximum was six). Among the 7,217 disclosed CAMs, CAMs related to asset impairment account for the largest number (3,229 CAMs, 44.8%), while income-related CAMs rank second (2,331 CAMs, 32.3%). Apart from these two main categories, 371 M&A-related CAMs accounted for 5.1% of the full sample, and no other type of CAM exceeded 2%. 2.2. Prior literature Research on new auditing reporting models, especially related to CAMs, has emerged both internationally and in China in recent years, but most of the research is experimental (Backof, Bowlin, & Goodson, 2017; Brasel, Doxey, Grenier, & Reffett, 2016; Brown, Majors, & Peecher, 2015; Christensen, Glover, & Wolfe, 2014; Doxey, 2014; Gimbar, Hansen, & Ozlanski, 2016; Kachelmeier, Schmidt, & Valentine, 2017; Sirois, Bédard, & Bera, 2014; Zhang et al., 2016; Zhang & Han, 2014; Zhang, He, & Han, 2015). However, extant experimental studies fail to reach a consensus regarding many topics. For instance, some studies find that disclosure of CAMs reduces the willingness of participants to invest in the firms (Christensen et al., 2014), while other studies argue that disclosure of CAMs adds to the auditing quality perceived by information users (Doxey, 2014; Zhang & Han, 2014). A number of studies suggest that disclosure of CAMs leads information users to undervalue the liability that auditors bear for material mis- statement (Brasel et al., 2016; Brown et al., 2015; Kachelmeier et al., 2017; Zhang et al., 2015). Meanwhile, another group of studies finds that disclosure of CAMs (surprisingly) increases a jury’s perception of audit negligence (Backof et al., 2017; Gimbar et al., 2016). Based on data from multiple regimes following the implementation of new auditing standards, a series of studies examine the consequences of disclosure of CAMs in the U.K.’s The Auditing Standard No. 1504 for Certified Public Accountants of China mentions three aspects that determine when auditors must identify CAMs: a high assessment of risk of material misstatement or identified special risk, significant audit judgement relating to areas pertinent to significant management judgement (including accounting estimations with high estimation uncertainty), and the effect significant transactions or events pose on the audit during the period. Please refer to the review by Bédard, Coram, Espahbodi, and Mock (2016). CHINA JOURNAL OF ACCOUNTING STUDIES 173 capital markets (Gutierrez, Minutti-Meza, Tatum, & Vulcheva, 2018;Lennox,Schmidt,& Thompson, 2018; Reid, Carcello, Li, & Neal, 2018) and in China’s capital markets (Wang & Li, 2019;Wang,Xu,Wang,&Yu, 2018). Gutierrez et al. (2018) find no evidence of a stock market reaction, and no significant changes in audit fees or audit quality accompanying the disclosure of CAMs. Reid et al. (2018) argue that disclosure of CAMs leads to higher-quality financial reports and insignificant changes in audit fees. Lennox et al. (2018) point out that the more CAMs a firm discloses, the smaller the firm’searningsresponsecoefficient (ERC). However, this pattern pre-exists the disclosure of CAMs by one year, suggesting that information users had already captured the potential risk information embedded in the number of CAMs. Using the data from Chinese A + H share firms for fiscal year 2016, Wang et al. (2018) provide evidence showing that the magnitude of change in cumulative abnormal returns (CARs) for firms adopting the new auditing standard is larger than for the samples that disclose traditional audit reports. In addition, Wang and Li (2019) find that implementing the new auditing standard suppress the stock price synchronicity of the audited firms. In general, these studies of Chinese capital market primarily reveal the value of information communicated by the new audit reports. The above-mentioned studies mainly focus on the economic consequences of imple- menting the new auditing standards, or the overall characteristics of the CAMs (e.g. the number of CAMs disclosed), but research targeting a specific type of CAM is scarce. This study aims to deepen the understanding of the information value of CAMs by analysing the relation between specific CAMs and the quality of audited accounting information. As described earlier, asset impairment is the most common type of disclosed CAM. Such matters signal the potential loss of economic value for the firm’s assets but involve manage- ment estimation and the professional judgement of auditors due to the significant uncer- tainty involved. The essence of asset impairment is the deterioration of a firms’ economic foundation. Therefore, if management makes a reasonable estimation of asset impairment, the provision for such asset impairment should effectively reflects the worsened economic condition of the firm (Francis et al., 1996; Lobo et al., 2017; Riedl, 2004). 3. Hypothesis development 3.1. Hypothesis H1 According to risk-oriented auditing principles, auditors are supposed to provide reasonable assurance regarding material misstatement risks identified through the audit process, if such risks exist in the CAMs. In other words, auditors should have resolved their concerns about the risks of material misstatement during the audit by responding, at both overall and assertion levels, to the CAMs that contain material misstatement risk as identified in the audit planning and performance stages. We illustrate the fundamental logic of this concept through Figure 1. Figure 1 depicts the relationships among the quality of unaudited accounting informa- tion, whether to identify a specific accounting issue as a CAM, and the quality of audited accounting information. We define the audited quality of asset impairment matters (IQ ) as the extent to which asset impairments reflect the worsened economic condition AUD of the firm. First, we assume the auditor faces three typical clients: firm A, firm B and firm C. In firm A, the unaudited information quality of asset impairment matters is high, as management has already fully booked asset impairments. In the risk assessment stage, 174 X. WU ET AL. Figure 1. Unaudited information quality, CAMs identification and audited information quality. the auditor perceives that the extent to which asset impairment has been booked is in accordance with the actual worsened economic condition of the firm, and therefore does not identify asset impairment matters as CAMs during the audit process. In other words, the auditor recognises that management’s judgement of the asset impairment of the firm is appropriate; therefore, the asset impairment information fully reflects the worsened economic condition of firm A and is recognised at the end of the audit. In firm B, the unaudited information quality of asset impairment matters is low, as management has failed to record the impairments adequately. In the risk assessment stage, the auditor perceives that asset impairment insufficiently provides for the worsened economic condition of the firm and thus identifies asset impairment matters as CAMs during the audit process. After applying adequate audit measures, the auditor requires the firm to increase the provision for asset impairment. Firm B accepts the adjustment and books sufficient asset impairment at the end of audit. Consequently, the audited asset impairment information fully reflects the worsened economic condition of the firm. In firm C, the unaudited information quality of asset impairment matters is low as management has failed to record the impairments adequately. However, the auditor fails to implement risk-oriented auditing principles rigorously and neither identifies asset impair- ment matters as CAMs nor requires the firm to increase provisions for asset impairment. As a result, the asset impairment information fails to fully reflect the worsened economic condition of firm C, as the asset impairment shown at the end of audit is insufficient. If there are only two types of firms in the market, namely firm type A (firms with high accounting information quality prior to auditing) and firm type B (firms with low accounting information quality prior to auditing, but whose auditors adequately implement risk-oriented auditing principles), then the information quality of audited asset impairment is high for both, regardless of which party (management or the auditor) identifies the asset impairment-related CAMs. This suggests there is no correlation between disclosure of asset impairment-related CAMs and the information quality of audited asset impairment (IQ ). AUD Even if there were only two types of firms (firm A and firm B) in the market and there is a significant correlation between CAM and information quality of unaudited asset impairment (IQ ), that is to say, IQ is lower when the audit report pre pre mentions a corresponding CAM and vice versa, it could be difficult for the researchers to test for such a pattern because the unaudited asset impairment information is unobservable. CHINA JOURNAL OF ACCOUNTING STUDIES 175 But if firms of type C (whose auditors fail to adequately implement risk-oriented auditing principles) do exist, a comparison between firms of type B and of type C suggests that the information quality of audited asset impairment is higher for type B firms (those that identify asset impairment matters as CAMs) than for type C firms (those that do not identify asset impairment matters as CAMs). Accordingly, the disclosure of asset impairment-related CAMs is positively correlated with the information quality of audited financial statements that include asset impairment. In addition, the identification of CAMs may be driven by asset impairment matters involving significant transactions or events that do not involve management judgement and the related material misstatement risks, according to Auditing Standard No. 1504 for Certified Public Accountants of China. In this circumstance, the decision to identify asset impairment matters, as CAMs have nothing to do with the intrinsic material misstatement risks that asset impairment matters carry, and therefore have no systematic correlation with the information quality of audited asset impairment. Accordingly, we propose our first hypothesis in the form of an alternative hypothesis: H1. Audited asset impairment better reflects the worsened economic condition for firms with asset impairment-related CAMs, compared to firms that do not identify asset impair- ment matters as CAMs. 3.2. Hypothesis H2 The null hypothesis in H1 (no correlation between the disclosure of impairment-related CAMs and the information quality of audited asset impairment) is more likely to hold for audit firms that have abundant audit resources. This is because audit firms with abundant audit resources can use their ample supplies of human resources, expertise, technical standards and quality control systems to identify areas containing material misstatement risks as fully and precisely as possible. In contrast, resource-limited audit firms are more likely to omit or fail to identify areas containing material misstatement risk, for they face constraints with respect to human resources, expertise, technology, quality control sys- tems and independence. Accordingly, we propose the second hypothesis: H2. The positive association between the disclosure of asset impairment-related CAMs and the information quality of audited asset impairment is more pronounced in audit firms with abundant audit resources than in audit firms with greater resource constraints. An implicit assumption of hypothesis H2 is that audit firms with greater resource con- straints do not misjudge the CAMs they identify. In other words, for those CAMs that do contain material misstatement risks, the auditors provide reasonable assurance about them Among the reasons for CAM identification that were disclosed, we identify 148 observations where the identification of CAMs is solely driven by asset impairment matters involving significant transactions or events (set CAM_MATACCOUNT = 1, otherwise 0) but not by involving management judgement or material misstatement risks. The other 2208 observations attribute their CAM identification to asset impairment matters involving significant subjective judgement or material misstatement risks (set CAM_MATJUDGRISK = 1, otherwise 0). Untabulated tests suggest that the regression result of Model (1) in Table 2 is driven by LOWPERF×CAM_MATJUDGRISK (t-stat. = 4.39) instead of driven by LOWPERF×CAM_MATACCOUNT (t-stat. = 0.93). 176 X. WU ET AL. during the implementation stage of the audit. However, violation of this implicit assumption indicates a weaker association between the disclosure of asset impairment-related CAMs and the information quality of audited asset impairment, making it difficult to identify the pattern posited in H2. 4. Research design 4.1. Empirical model According to prior studies, a firm tends to book a larger impairment of assets when its economic performance has deteriorated. Following Lobo et al. (2017), we test H1 by estimating the following model. IMPAIR ¼ α þ α LOWPERF þ α LOWPERF CAM AI þ α CAM AI 0 1 2 3 (1) þ Controls þ ε In the model, the dependent variable IMPAIR equals impairment loss (if any) divided by total asset at the beginning of the year. Following Lobo et al. (2017), LOWPERF is the indicator of deteriorating firm economics. LOWPERF equals 1 if any of the following three indicators of economic impairment are met: market-to-book ratio less than one, ROA in the lowest quartile of the firms in a given industry and year, and operating cash flows divided by total assets in the lowest quartile of the same industry/year cohort, and zero otherwise. Under the rationale that asset impairment reflects deteriorating economic performance, we expect α to be positive. CAM_AI equals 1 if the auditor discloses at least one CAM in the audit report, and 0 otherwise. Our interest lies in the coefficient of LOWPERF×CAM_AI. According to H1, audited financial information is supposed to reflect the firm’s underlying economics regardless of whether a CAM is disclosed, in which case we anticipate α to be insignif- icant. However, if the disclosure of CAMs indicates higher informational quality in terms of asset impairment, one should observe a significant and positive α . We control the following variables in Model (1) in accordance with prior studies (Francis et al., 1996; Lobo et al., 2017; Riedl, 2004): the natural log of year-end total assets (SIZE), total liability to total assets ratio (LEV), the industry median growth of operating income (IND_SALESCHG), Big 10 audit firm indicators according to the annual ranking by CICPA (BIG10), as well as industry and year fixed effects. We estimate Model (1) using the Tobit regression because of the non-negativity of the dependent variable, IMPAIR. To test H2, we need to differentiate between audit firms with relatively abundant audit resources and those that face more resource constraints. As larger audit firms have ample supply of human resources, expertise, technical standards and quality control systems (DeAngelo, 1981), we divide our sample into two subsamples deter- mined by BIG10 (= 0 vs. = 1) and re-estimate the equations for each of the subsamples. We expect the coefficient α to be larger for firms audited by more resource-constrained auditors (BIG10 =0)thanfor firms whose auditors face fewer resource constraints (BIG10 =1). We multiply the original ratio by 100 for better presentation of coefficients in the regression analysis. The results remain robust when adopting OLS regression. CHINA JOURNAL OF ACCOUNTING STUDIES 177 4.2. Data and sample We obtain all financial data from WIND database, while data on CAMs are manually collected from the 2017 audit reports. We collected CAM data from 3,449 annual reports of A-share listed firms through 30 April 2018. We deleted 116 firms that received modified audit opinions to guarantee comparability in the assurance level of financial reports. In addition, we deleted 76 firms from financial sectors and 26 firms due to missing values. We have 3,231 firm observations in the final sample. 4.3. Descriptive statistics Table 1 reports the descriptive statistics for the variables in Model (1). According to Table 1, the mean (median) of the dependent variable IMPAIR is 0.784 (0.468), suggesting that the mean (median) percentage of asset impairment loss relative to beginning total assets is 0.784% (0.468%). The incidence of LOWPERF is 17.3%, and 72.9% of firms disclosed at least one CAM related to asset impairment in 2017. Meanwhile, 58% of firms were audited by top 10 audit firms in China for that year. 5. Empirical results 5.1. Regression results for hypothesis H1 Table 2 reports the Tobit regression results of Model (1). The first column of Table 2 shows the basic relationship between LOWPERF and IMPAIR. The significant positive coefficient of LOWPERF (t-stat. = 9.77) indicates that the asset impairment booking reflects the wor- sened economic condition of the firm. It is consistent with the concept that accounting information represents the underlying economics of the firm. The second column in Table 2 examines the information quality of CAMs. The coeffi- cient of LOWPERF×CAM_AI is 0.429, and is significantly different from zero at the 1% level (t-stat. = 4.41). The result implies that the ability of impairment accounting (IMPAIR) to Table 1. Descriptive Statistics (N = 3231). Variable Mean S.D. Min Median Max IMPAIR 0.784 0.876 0 0.468 3.428 LOWPERF 0.172 0.378 0 0 1 CAM_AI 0.729 0.444 0 1 1 SIZE 22.173 1.223 19.830 22.063 24.755 LEV 0.402 0.192 0.116 0.387 0.803 IND_SALESCHG 0.183 0.058 0.050 0.185 0.338 BIG10 0.580 0.494 0 1 1 By 30 April 2018, 65 of the total 3,514 listed firms had not disclosed any CAM information in their audit reports, including five whose stocks are suspended, two that had delayed release of their annual reports, 20 newly listed firms that were exempt from CAM disclosure and 38 that failed to obey the new auditing standards. Auditing Standard No. 1504 for Certified Public Accountants of China regulates that auditors shall not communicate in the section of CAM on the issues that have caused a qualified audit opinion (Article 12). This means that CAM information would be not comparable between firms receiving clean and modified audit opinions. Robustness tests suggest that the findings hold if we include financial firms and firms receiving modified opinions in the regression sample. 178 X. WU ET AL. Table 2. Tobit Regression of Model (1). Coefficient Coefficient Dependent variable: IMPAIR (t-stat.) (t-stat.) LOWPERF 0.505 0.137 (9.77)*** (1.71)* LOWPERF×CAM_AI 0.429 (4.41)*** CAM_AI 0.339 (10.36)*** SIZE −0.075 −0.086 (−4.53)*** (−5.34)*** LEV 0.390 0.395 (3.37)*** (3.49)*** IND_SALESCHG −2.883 −1.938 (−1.75)* (−1.20) BIG10 −0.015 −0.027 (−0.46) (−0.86) Constant 2.399 2.333 (5.75)*** (5.71)*** Industry fixed effects Control Control N 3,231 3,231 Pseudo R 0.05 0.07 ***, *denote statistically significant at 1% and 10% levels, respectively (two-tailed). reflect lower current performance (LOWPERF) is stronger when asset impairment matters are identified as CAMs; hence H1 holds. We notice that when LOWPERF×CAM_AI is added to the model, the coefficient of LOWPERF drops from 0.505 to 0.137, and becomes marginal (p < 10%; t-stat. = 1.71). This decline in both magnitude and significance suggests that the ability of asset impair- ment information (IMPAIR)toreflect a worsened economic condition (LOWPERF) is notably weaker when no asset impairment-related CAMs are disclosed in the audit report (CAM_AI = 0). Overall, the results in Table 2 support hypothesis H1. In terms of control variables, the coefficient of CAM_AI is statistically different from zero at the 1% level (t-stat. = 10.36), indicating that even for those firms whose economic condition shows no sign of deterioration (LOWPERF = 0), firms with an asset impairment- related CAM book significantly more asset impairments compared to firms without any asset impairment-related CAMs. It reveals the fact that certain CAMs are due to material transactions (rather than misreporting risks), which is consistent with the definition of a CAM according to accounting regulations. In addition, it shows that smaller firms with higher financial leverage book larger assets impairments, and the negative coefficient of IND_SALESCHG suggests that firms in more prosperous industries are less likely to record assets impairments. 5.2. Regression results for hypothesis H2 Table 3 reports the regression results of Model (1) for subgroups. The coefficient of LOWPERF×CAM_AI is 0.245 (p < 5%; t-stat. = 2.00) for firms audited by a non-big 10 auditor, compared to 0.646 (p < 1%; t-stat. = 4.22) for firms audited by a Big 10 auditor. The comparison test shows that the difference between these two coefficients is signifi- cant (F-stat. = 4.21; p-value = 0.04), which supports H2. CHINA JOURNAL OF ACCOUNTING STUDIES 179 Table 3. Tobit Regression of Model (1) in Subgroups. BIG10 =1 BIG10 =0 Dependent variable: IMPAIR Coefficient Coefficient (t-stat.) (t-stat.) LOWPERF 0.284 −0.030 (2.87)*** (−0.24) LOWPERF×CAM_AI 0.245 0.646 (2.00)** (4.22)*** CAM_AI 0.358 0.331 (8.67)*** (6.05)*** Controls YES YES N 1,873 1,358 Pseudo R 0.07 0.07 ***, **denote statistically significant at 1% and 5% levels, respectively (two-tailed). Moreover, Column 2 in Table 3 shows that among firms audited by smaller audit firms (BIG10 = 0), the ability of asset impairment to reflect a deterioration in the firm’s underlying economic condition is rather weak when none of the CAMs is related to asset impairment (CAM_AI =0), as thecoefficient of LOWPERF is not significantly different from zero (t-stat. = −0.24). In contrast, Column 1 in Table 3 suggests that for firms audited by larger audit firms (BIG10 =1), thecoefficient of LOWPERF is highly positive (p < 1%; t-stat. = 2.87), suggesting that in these firms, the relationship between asset impairment and deteriorating economic conditions is present and strong, even if a firm does not disclose any asset impairment-related CAMs (CAM_AI =0). Taken together, the results reported in Table 3 suggest that audited asset impairment is more closely linked to a decline in underlying economics, provided that an asset impairment-related CAM is disclosed when the auditor is subject to resource constraints compared to audits where the auditor faces fewer resource constraints. 5.3. The ranking order of disclosed CAMs Does the relative ranking order of an individual CAM matter relative to other CAMs a firm discloses in a single audit report? A possible interpretation is that the greater attention a CAM gains from the auditor, the more likely it is be assigned a higher ranking. On the other hand, the ranking order of CAMs may not signal any difference in the attention given by the auditor. To answer this question empirically, we construct Model (2) as follows: IMPAIR ¼ β þ β LOWPERF þ β LOWPERF CAM AI1 þ β CAM AI1 0 1 2 3 (2) þ β LOWPERF CAM AI2 þ β CAM AI2 þ Controls þ ε 4 5 In Model (2), CAM_AI1 equals 1 if an asset impairment matter is identified as a CAM and its standardised ranking order among the CAMs in the audit report is not less than 0.5, and 0 otherwise; CAM_AI2 equals 1 if the asset impairment matter is identified as a CAM and its standardised ranking order in the audit report is less than 0.5, and 0 otherwise. The standardised ranking order of a CAM = 1-(n-1)/(N-1), where n = 1 . . . N, denoting the natural order of CAMs in the audit report, and N is 180 X. WU ET AL. the total numbers of CAMs the firm discloses. Other model specifications remain consistent with Model (1). In Model (2), we focus primarily on the coefficients of the variables LOWPERF×CAM_AI1 and LOWPERF×CAM_AI2. If a higher ranking order of asset impairment-related CAMs implies higher informational quality, we would expect β > β , but if the difference in ranking order 2 4 does not represent a difference in information quality, we would expect β = β . 2 4 Table 4 shows the regression results for Model (2). The first column shows the result for the full sample, while the second and third columns show subgroup results based on audit firm size. The first column of Table 4 suggests that the coefficients for both LOWPERF×CAM_AI1 and LOWPERF×CAM_AI2 are significantly positive at the 1% level (t-stat. = 4.17 and t-stat. = 3.18, respectively), and the difference test within the group accepts the null hypothesis (F-stat. = 0.62). Similarly, results shown in the other two columns suggest there is no significant difference between the coefficients of LOWPERF×CAM_AI1 and LOWPERF×CAM_AI2 (F-stat. = 0.06 and F-stat. = 1.01, respectively), regardless of the size of the audit firm. To summarise, the results reported in Table 4 indicate there is no significant difference in the information quality across CAMs, regardless of the order in which the auditor ranks them. 5.4. Measuring worsened economic condition using continuous variable In addition to employing a dummy variable, LOWPERF, to indicate a worsened economic condition for a given firm, we use continuous measures for robustness checks. First, we treat an auditor’s explicit mention of a ‘significant going concern uncertainty’ in the audit report as an external assessment of the firms degraded economic condition. Following Table 4. CAMs with Different Ranking Orders. Full sample BIG10 =1 BIG10 =0 Dependent variable: IMPAIR Coefficient Coefficient Coefficient (t-stat.) (t-stat.) (t-stat.) LOWPERF 0.138 0.285 −0.027 (1.72)* (2.87)*** (−0.21) LOWPERF×CAM_AI1 0.462 0.254 0.726 (4.17)*** (1.86)* (4.06)*** CAM_AI1 0.368 0.374 0.372 (9.06)*** (7.46)*** (5.42)*** LOWPERF×CAM_AI2 0.371 0.217 0.548 (3.18)*** (1.40) (3.14)*** CAM_AI2 0.316 0.345 0.295 (8.74)*** (7.51)*** (4.92)*** Controls YES YES YES N 3,231 1,873 1,358 Pseudo R 0.07 0.07 0.07 Difference tests within the subgroups(H : LOWPERF× CAM_AI1 = LOWPERF×CAM_AI2) F-stat. 0.62 0.06 1.01 (p-value) (0.43) (0.81) (0.31) ***, *denote statistically significant at 1% and 10% levels, respectively (two-tailed). When multiple asset impairment-related CAMs exist in a single audit report, the CAM with higher ranking order shall prevails. CHINA JOURNAL OF ACCOUNTING STUDIES 181 previous studies (e.g. Mo, Rui, & Wu, 2015), we estimate a going concern opinion model based on a sample spanning the period 2007 to 2017, where the dependent variable GCO equals 1 if the firm’s auditors explicitly mentioned ‘significant going concern uncertainty’ in the audit report, and 0 otherwise. Second, we predict the probability of receiving a going concern opinion (PROBGCO) for sample firms in 2017. Since our primary model already limits the sample to firms that received clean opinions, the larger the value of PROBGCO, the worse the economic condition as assessed by external professionals. Third, we replace LOWPERF with PROBGCO in the asset impairment model. The regression results reported in Column 1 of Table 5 suggest that the coefficient of PROBGCO is significantly positive without adding the interaction term to the model. The positive coefficient means that the worse the economic condition as assessed by the external auditors, the larger the asset impairment the firm books, adding to the validity of PROBGCO as a proxy for a firm’s degraded economic condition. Results in the second column show that the coefficient of PROBGCO×CAM_AI is significantly greater than zero, suggesting the positive association between worsened economic condition and asset impairment is more pronounced in firms with asset impairment-related CAMs, which is consistent with the findings when LOWPERF is used as the proxy. 5.5. Eliminating potential interference of negative earnings management Normally, a deteriorating economic condition for a firm means a higher risk of asset impairment. However, asset impairment may not reflect the firm’s underlying economics if management engages in manipulation to lower earnings. To eliminate such potential interference on our conclusion, we follow previous studies (e.g. Francis et al., 1996; Riedl, 2004) to identify firm-years with higher levels of negative earnings management, and remove them from our sample. First, we estimate a prediction model where the dependent variable, EARCHG, equals the earnings growth from the previous year scaled by beginning assets of the year. Explanatory variables include firm size, financial leverage, current ratio, revenue growth, ROA, market-adjusted stock return and a dummy variable of loss in the previous year. We run an OLS regression based on sample of firms from 2007–2017, and estimate the residual for firms in 2017. We then narrow the dataset down to observations that include Table 5. Tobit Regression of Asset Impairment Model: Using Continuous Variable PROBGCO to Proxy Worsened Economic Condition. Coefficient Coefficient Dependent variable: IMPAIR (t-stat.) (t-stat.) PROBGCO 3.454 −0.438 (3.17)*** (−0.54) PROBGCO×CAM_AI 6.742 (4.31)*** CAM_AI 0.347 (10.00)*** Controls Control Control N 3,231 3,231 Pseudo R 0.04 0.06 ***denotes statistically significant at 1% level (two-tailed). 182 X. WU ET AL. a negative EARCHG (earnings decline) and a negative residual according to the prediction model, and rank the negative residuals in ascending (increasingly negative) order. Finally, we remove the observations with residuals lower than certain thresholds (25% quartile, 50% quartile and 75% quartile, respectively) and rerun the regressions in our main tests. Untabulated results show that our findings remain robust (the coefficients of LOWPERF×CAM_AI for thresholds at 25%, 50% and 75% quartiles are 0.279, 0.229 and 0.254, significant at 1%, 1% and 5%, respectively). 6. Conclusions Based on the first wave of CAMs disclosure in audit reports of year 2017, this study examines the intrinsic relationship between CAMs, as informationonauditingprocess,and thequality of audited outcomes. We focus on asset impairment, the most commonly disclosed category of CAMs by Chinese listed firms in 2017 and investigate whether asset impairment-related CAMs signal higher audit quality in terms of asset impairment. We find that the capability that audited asset impairment reflects the worsened economic condition is notably stronger when assetimpairmentmatters areidentified as CAMs. Supplementary analyses suggest that this pattern is more pronounced in smaller audit firms with greater resource constraints. It implies that reasonable inference behind the phenomenon is that auditors fail to fully identify all material misstatements of their clients, leading to higher information quality of asset impair- ment as CAMs and lower information quality of non-CAM asset impairment. For financial information users, our study contributes to the interpretation of CAMs under the new auditing standard. Theoretically, information users would not anticipate a close relationship between CAMs (audit process) and post-audit earnings quality, if the audit was conducted properly. However, if a close relationship between CAMs and audited earnings quality is detected, we suggest information users be vigilant regarding the quality of non-CAM accounting information, for there may exist deficiencies in the implementation of risk-oriented auditing principles. There are limitations to this study. We examine only the most common CAMs, namely those related to asset impairment, so future research could focus on other types of CAMs. With the surge in CAM disclosures, the auditors’ behaviour in the auditing practice may be influenced by various information users. Such topics are worthy of investigation in the future. Disclosure statement No potential conflict of interest was reported by the authors. References Backof, A., Bowlin, K., & Goodson, B.M. (2017). The impact of proposed changes to the content of the audit report on Jurors’ assessments of auditor negligence (Working paper). Bédard, J., Coram, P., Espahbodi, R., & Mock, T.J. (2016). Does recent academic research support changes to audit reporting standards? Accounting Horizons, 30(2), 255–275. Brasel, K., Doxey, M.M., Grenier, J.H., & Reffett, A. (2016). Risk disclosure preceding negative out- comes: The effects of reporting critical audit matters on judgments of auditor liability. The Accounting Review, 91(5), 1345–1362. CHINA JOURNAL OF ACCOUNTING STUDIES 183 Brown, T., Majors, T., & Peecher, M. (2015). The influence of evaluator expertise, a judgment rule, and critical audit matters on assessments of auditor legal liability (Working paper). 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. Church, B.K., Davis, S.M., & McCracken, S.A. (2008). The auditor’s reporting model: A literature overview and research synthesis. Accounting Horizons, 22(1), 69–90. DeAngelo, L.E. (1981). Auditor size and audit quality. Journal of Accounting and Economics, 3(3), 183–199. Doxey, M. (2014). The effects of auditor disclosures regarding management estimates on financial statement users’ perceptions and investments (Working paper). Francis, J., Hanna, J.D., & Vincent, L. (1996). Causes and effects of discretionary asset write-offs. Journal of Accounting Research, 34, 117–134. Gimbar, C., Hansen, B., & Ozlanski, M.E. (2016). The effects of critical audit matter paragraphs and accounting standard precision on auditor liability. The Accounting Review, 91(6), 1629–1646. Gutierrez, E., Minutti-Meza, M., Tatum, K.W., & Vulcheva, M. (2018). Consequences of adopting an expanded auditor’s report in the United Kingdom. Review of Accounting Studies, 23(4), 1543–1587. International Auditing and Assurance Standards Board (IAASB). (2013). Reporting on audited financial statements: Proposed new and revised International Standards on Auditing (ISAs). Kachelmeier, S.J., Schmidt, J.J., & Valentine, K. (2017). The disclaimer effect of disclosing critical audit matters in the auditor’s report (Working paper). Lennox, C.S., Schmidt, J.J., & Thompson, A. (2018). Does an expanded model of audit reporting improve the information environment for investors? Evidence from the UK (Working paper). Lobo, G.J., Paugam, L., Zhang, D., & Casta, J.F. (2017). The effect of joint auditor pair composition on audit quality: Evidence from impairment tests. Contemporary Accounting Research, 34(1), 118–153. Mo, P.L.L., Rui, O.M., & Wu, X. (2015). Auditors’ going concern reporting in the pre- and post- bankruptcy law eras: Chinese affiliates of big 4 versus local auditors. The International Journal of Accounting, 50(1), 1–30. Reid, L.C., Carcello, J.V., Li, C., & Neal, T.L. (2018). Impact of auditor report changes on financial reporting quality and audit costs: Evidence from the United Kingdom. Contemporary Accounting Research, 36(3), 1501–1539. Riedl, E.J. (2004). An examination of long-lived asset impairments. The Accounting Review, 79(3), 823–852. Sirois, L., Bédard, J., & Bera, P. (2014). The informational value of emphasis of matter paragraphs and auditor commentaries: Evidence from an eye-tracking study (Working paper). Tang, J.H. (2015). The evaluation and analysis on IAASB’s audit report reform. Auditing Research, 1, 60–66 (In Chinese). Wang, M.Z., & Li, D. (2019). New audit reporting and stock price synchronicity. Accounting Research, 1,86–92 (In Chinese). Wang, Y.Y., Xu, R., Wang, C.L., & Yu, L.S. (2018). Can key audit matters enhance the communication value of the audit report? Accounting Research, 6, 86–93 (In Chinese) Zhang, J.X., Cai, Y.D., & Liu, W.H. (2016). The effect of improvement of audit report and manager auditor relationship on managers’ communication willingness——An experimental evidence. Auditing Research, 3,77–83 (In Chinese). Zhang, J.X., & Han, D.M. (2014). The improvement of standard auditor’s report and investors’ perceptions of relevance and usefulness, their investment decisions——An experimental evi- dence. Auditing Research, 3,51–59 (In Chinese). Zhang,J.X.,He,C.,&Han,D.M.(2015). Proposed amendments to the standard audit report and investor perception of auditor′s responsibility——An experimental evidence. Auditing Research, 3,56–63 (In Chinese).
Journal
China Journal of Accounting Studies
– Taylor & Francis
Published: Apr 3, 2019
Keywords: Audit report; critical auditing matter; risk-oriented auditing; asset impairment
