Audit Partner Disciplinary Actions and Financial Restatements

Background: The Audit Partner Disciplinary System in Taiwan

Audit partners in Taiwan are regulated by the government, headed by the Financial Supervisory Commission (FSC), which is equivalent to the SEC in the US 2 Currently two agencies have the authority to invoke sanctions against audit partners. One is the Certified Public Accountant (CPA) Discipline Committee empowered by the Taiwanese Certified Public Accountant Act. The other is the FSC empowered by the Taiwanese Securities Exchange Act.

According to the Taiwanese Certified Public Accountant Act, the CPA Discipline Committee has four types of disciplinary actions, including warning, reprimand, suspension between two months and two years, 3 and disbarment. If the audit partners disagree with a disciplinary action, they can request a re-examination of the case by the CPA Disciplinary Rehearing Committee. If they are still dissatisfied with the results of the re-examination, they can initiate an administrative litigation. When a final decision is reached the resolution is posted on the government gazette and website. Both the CPA Discipline Committee and the CPA Disciplinary Rehearing Committee comprise 15 members appointed by the FSC. These members include: (1) representatives of the Taiwanese CPA associations; (2) scholars or highly regarded members with law or accounting backgrounds; and (3) officials of related administrative agencies. Part-time employees of both committees come from the FSC and are responsible for keeping the meeting minutes and managing other in-house affairs. Committee members and their staff are not compensated for their services.

To ensure a timely process, the Taiwanese Securities Exchange Act empowers the FSC to investigate and adjudicate disciplinary charges against the audit partners of public companies regardless of the processes of the CPA Discipline Committee. The types of adjudication of the FSC include warning, suspension from providing attestation services to public companies for less than two years, and revocation of permission to provide attestation services to public companies. A disciplinary action is announced immediately after a decision by the FSC. If the audit partners are dissatisfied with the result, they can initiate an administrative remedy.

The efficacy of the two agencies differs. The Taiwanese Securities Exchange Act only applies to public companies. Therefore, audit partners under the FSC suspension or revocation are not allowed to render attestation services to public companies but can still render them to non-public companies. However, if audit partners are suspended or disbarred by the Taiwanese Certified Public Accountant Act, they are prohibited from providing any public accounting services during the sanction period.

Literature Review

Studies on auditor sanctions can be classified into four streams. The first stream is about market responses to sanctions. Firth (1990) reports a drop in stock prices of other clients of disciplined auditors. Moreland (1995) finds that the earnings response coefficients of other clients of sanctioned auditors decrease significantly. Further, the more serious the sanction, the greater the magnitude of the decrease. Using the accounting scandal of Germany ComROAD AG and its auditor KPMG, Weber et al. (2008) document that KPMG's clients sustain negative abnormal returns of 3% around the period surrounding the scandal. These returns are more negative for companies that are likely to have higher demands for audit quality.

The second stream of the literature addresses the causes of disciplinary actions. Using Taiwanese sanction data during 1988–1998, Yu et al. (2000) list the common causes of disciplinary actions, including the lack of conclusions in working papers, insufficient and incompetent audit evidence, missing management representation letters or letters with erroneous dates, and lack of evidence pertaining to the physical count of cash. Firth et al. (2005), using Chinese sanction data for the period 1996–2002, report that auditors suffer more sanctions in cases in which they fail to detect and report material mis-statements than in cases involving inadequate disclosure. With regard to material mis-statement/fraud cases, revenue-related material mis-statements bring about more auditor sanctions than asset-related mis-statements.

The third stream of the literature investigates the consequences of disciplinary actions on auditors. Both Davis and Simon (1992) and Firth (1990) report an audit fee reduction for disciplined auditors. Disciplined auditors lose incumbent clients as well (Firth, 1990; Wilson and Grimlund, 1990; Moreland, 1997; Weber et al., 2008; Skinner and Srinivasan, 2012). The fourth stream of the literature concerns the effects of sanctions on auditor behaviour. Bannister and Wiest (2001) find that auditors subject to SEC disciplinary actions require clients to recognize more income-decreasing accruals during the investigation periods but have no such requirement after the event.

Aside from adjudicated sanctions, peer reviews and the PCAOB's inspections are also considered as auditor monitoring and discipline mechanisms. Using data from an insurance company, Casterella et al. (2009) define audit failure as the auditors' claims of malpractice because of their negligence. They find that peer review findings are useful in predicting audit failure, and are also associated with other firm-specific indicators of potentially weak quality control or risky practices within audit firms. As a result, they conclude that peer reviews provide effective signals regarding audit firm quality. Lennox and Pittman (2010) investigate the information value of the PCAOB inspection reports. They find that audit clients do not perceive the PCAOB's inspection reports to be valuable in signalling audit quality, probably because the PCAOB fails to disclose in its reports certain information, specifically, the quality control weaknesses and overall ratings of audit firms.

Hypothesis Development

An audit partner is responsible for planning an audit, supervising the audit team, making decisions on significant matters in the audit process, issuing an audit report, and communicating with the client's management. Studies on auditing judgment and decision making suggest that audit quality is affected by individual auditor attributes, such as knowledge, ability, risk profile, and cognitive style (Nelson and Tan, 2005). These personal attributes are considered fixed at a given point in time (Tan and Kao, 1999). Apart from personal attributes, engagement or task factors, such as complexity, may affect an auditor's performance (Bonner, 1994; Chang et al., 1997; Tan and Kao, 1999; Tan et al., 2002).

When an audit partner is sanctioned for his or her negligence involving a specific engagement, one of the following two possibilities may hold. The first possibility is that a disciplinary action results from an isolated case of low audit quality, possibly due to engagement-specific complexity. In this case, the audit quality of the disciplined audit partner for his/her other clients is expected to be the same as that of non-sanctioned audit partners. The second possibility is that a disciplinary action systematically reflects the average audit quality of the sanctioned audit partner. The disciplined audit partner delivers inferior quality audit work due to personal attributes, for example, insufficient knowledge, poor problem-solving abilities, or aggressive risk-taking attitude. In this case, we expect the audit quality of the sanctioned audit partner for his/her other clients to be significantly lower than that of non-sanctioned audit partners.

Krishnan (2005) finds that the clients of Arthur Andersen's Houston office, which audited Enron, exhibit less timely reporting of bad news compared to a sample of Houston-based clients audited by other Big 6 audit firms, as well as clients of Andersen's Atlanta office, in the same year as the Enron audit failure. Francis and Michas (2013) document that earnings quality is lower for clients in office-years with client misreporting compared to a control sample of office-years with no restatements. Both studies indicate that audit failure can reveal the poor quality of concurrent audits of auditors involved in a failed audit. Accordingly, we present the following hypothesis:

The drivers behind the auditors' pursuit of high quality audits may include protection of reputation or fear of litigation (Khurana and Raman, 2004). Sanctions tarnish reputations and lead to either a loss of clients or a reduction in audit fees (Firth, 1990; Wilson and Grimlund, 1990; Davis and Simon, 1992; Moreland, 1997; Weber et al., 2008; Skinner and Srinivasan, 2012). Furthermore, if disciplined auditors are perceived to be lower quality auditors, investors or regulatory agencies will scrutinize their audit reports more discerningly and critically. For example, Huang et al. (1998) report that, in Taiwan, the regulatory agencies examine not only the files of the investigated client but also the files of all other clients when an audit partner is subject to a sanction investigation. These examinations increase the likelihood of discovering defects in the financial statements of their clients and increase the risk of litigation or enforcement actions.

In summary, reputation loss and litigation/regulation risk provide disciplined auditors with incentives to improve their audit quality in the post-sanction period. Moreover, Bannister and Wiest (2001) argue that disciplined auditors have strong incentives to improve their audit quality to rebuild their reputations and retain existing clients or attract new ones without requiring excessive fee discounting. Therefore, this study suggests the following hypothesis:

Legally, successor auditors are not liable for defects in the financial statements that were audited by predecessor auditors. However, if previous mis-statements remain uncorrected or if inappropriate accounting methods persist, the successors are held liable (Cahan and Zhang, 2006; Romanus et al., 2008). Using the audit failure of Enron, prior studies show that successor auditors limit their litigation risk from ex-Andersen clients by decreasing discretionary accruals (Cahan and Zhang, 2006), issuing more going-concern audit opinions (Krishnan et al., 2007) or raising audit fees (Kealey et al., 2007).

Following Cahan and Zhang (2006), we posit that the change in auditors because of sanctions increases the litigation risk of successor auditors in at least two ways: (1) if the audit quality of predecessor disciplined auditors is lower, the probability of financial mis-statements by their clients is higher; and (2) if the predecessor disciplined auditors are perceived as lower quality auditors, investors or authorities will scrutinize the financial statements associated with the disciplined auditors, thus increasing the probability of discovering mis-statements. Successor auditors may be subject to litigation when they have taken over the engagement and do not detect and correct mis-stated beginning balances (Lys and Watts, 1994). In response to the increase in litigation risk from the new clients accepted from predecessor disciplined auditors, we posit that successor non-disciplined auditors have incentives to increase their diligence to ensure audit quality, resulting in lower probabilities of financial restatements of the new clients. This study articulates the above statements by the following hypothesis:

Empirical Model

This study establishes the following empirical model to test our hypotheses:

(1)

where,

RESTATE

= 1 if a company restates its annual financial statements, and 0 otherwise;

PRE

= 1 if a company is audited by a disciplined audit partner in the pre-sanction period, and 0 otherwise;

POST

= 1 if a company is audited by a disciplined audit partner in the post-sanction period, and 0 otherwise;

SUCC

= 1 if a company is audited by a successor non-disciplined audit partner in the post-sanction period, and 0 otherwise; and

Controls

= vector of the control variables.

This study uses γ₁ to capture the likelihood of financial restatements by the clients of disciplined audit partners before sanctions. According to H1, audit partner disciplinary actions signal lower quality audit partners and thus γ₁ is expected to be positive. γ₂ assesses the likelihood of financial restatements by the clients of disciplined audit partners after sanctions. Based on H2, as disciplined audit partners have learned from past lessons and improve subsequent audit quality, we expect that γ₂ is less than γ₁, namely, γ₁ – γ₂ > 0. We employ γ₃ to evaluate the likelihood of financial restatements by the clients of successor non-disciplined audit partners after sanctions. According to H3, when accepting clients from disciplined audit partners, the successors increase their diligence to limit litigation or regulation risk. We thus expect that γ₃ is less than γ₁, namely, γ₁ – γ₃ > 0.

Variable Measurements

Research variables: PRE, POST, and SUCC

Audit clients have three choices after audit partner sanctions. One is to retain the incumbent disciplined audit partner. Another is to replace the disciplined audit partner with another disciplined partner. The third is to replace the disciplined audit partner by a non-disciplined audit partner. On the other hand, if audit partners are not sanctioned, audit clients also have three choices. One is to retain the non-disciplined audit partner. The second is to hire another non-disciplined audit partner. The third is to replace the incumbent non-disciplined audit partner by an audit partner who has been disciplined before. Figure 1 depicts the relation between sanctions and audit partner choices. Cases 1 and 2 involve non-disciplined audit partners only. Audit partner B in period t–1 belongs to the pre-sanction period and is used to construct the research variable PRE. In contrast, we utilize disciplined audit partners in period t (post-sanction period) to establish the research variable POST. When POST is sub-divided, Case 3 refers to a situation in which a company changes its non-disciplined audit partner to a disciplined one. This case is called POST_NEW, and it represents the new clients of the disciplined audit partner. Case 4 refers to a company that retains its disciplined audit partner. This case is called POST_KEEP, and it denotes the existing clients of the disciplined audit partner. Case 5 indicates a situation in which a company engages another disciplined audit partner after the sanction of the incumbent audit partner. This case is called POST_DOUBLE, and it represents another situation in which a disciplined audit partner solicits new clients. 6 Finally, Case 6 is the situation in which a company changes its disciplined audit partner to a non-disciplined one. It is used to construct research variable SUCC.

If companies are audited by disciplined audit partners in the pre-sanction period (Cases 4, 5, and 6 in period t–1), then PRE is equal to 1, and 0 otherwise. If companies are audited by disciplined audit partners in the post-sanction period (including POST_KEEP, POST_NEW, and POST_DOUBLE), then POST is equal to 1, and 0 otherwise. 7 If companies change to a non-disciplined audit partner after their audit partner is sanctioned, SUCC equals 1, and 0 otherwise. If the sample companies have audit partners who have not been sanctioned at least during our sample period, then PRE, POST, and SUCC equal zero. These companies will serve as a control group.

Sample

Our sample data come from two channels of audit partner sanction law, namely, the Taiwanese Certified Public Accountant Act and the Taiwanese Securities Exchange Act. Sanction data on the former are obtained from the National Central Library Gazette, and sanction data on the latter are gleaned from the FSC Website. Other data come from the Taiwan Economic Journal (TEJ).

The research period of this study is 2000–2006. However, we need data in period t–1 to construct some variables and data in periods t + 1 to t + 3 to check whether financial statements are restated. Thus, data are collected for the period from 1999 to 2009. The initial sample comprises 8,023 listed firm-year observations. We eliminate companies for which the audit partners were disciplined (i.e., 17 companies with 49 observations) because they are already identified as companies with an audit deficiency and will thereby result in a bias toward H1 if they are included in the sample. To ease the comparability of financial data, we exclude 107 non-calendar year observations. We also delete 1,027 observations with missing financial data and eight with missing director data. 10 As stated earlier, this study utilizes mis-statement-type financial restatements to proxy for low audit quality. If restatements result from a qualified audit opinion, then the restatements will not indicate low audit quality and will therefore be excluded. The total number of observations with a qualified audit opinion is 66. Table 1 displays the sample selection process. The final sample includes 6,766 firm-year observations, comprising 1,411 companies and 82 financial restatements. The 82 financial restatements represent 59 companies and 63 different causes of restatements. 11

Table 2 lists the sample distribution, Panel A by year and Panel B by industry. The number of observations for PRE = 1 is 724, which represents the clients of 33 disciplined audit partners. In contrast, the number of observations for POST = 1 is 346, which denotes the clients of 22 disciplined audit partners. 12 The number of observations for SUCC = 1 is 449, which stands for the clients of 122 successor non-disciplined audit partners. In the yearly distribution, the earlier years have more pre-sanction samples. However, the total of sanction-related samples in the column subtotal indicates an immaterial difference in annual sample distributions. Industrial sample distribution reports two industries that incur more frequent audit partner sanctions and financial restatements. These industries are information electronic and textile, the two leading industries in Taiwan. In total, no unusual yearly or industrial clustering exists in this study.

Descriptive Statistics

Table 3 displays the descriptive statistics for the control variables. To reduce any adverse effects from outliers, the top and bottom 1% of the corresponding distributions of all continuous variables are winsorized. Mean GROWTH is 11.47%. FIN has a mean value of 0.06, median value of 0, and maximum value of 0.55. This indicates that only a few companies conducted fundraising. If they do raise funds, however, their fundraising could be as high as 55% of the beginning total assets. A mean LEV of 0.43 indicates that 43% of the capital comes from debt financing and 57% from equity financing. Mean LOSS of 0.25 denotes that 25% of the sample companies incur net losses. Average SIZE is 15.04 and its untransformed amount indicates average total assets of NT$10,678 million. Average LNAGE is 2.95 and its untransformed figure reveals that the mean number of years since company establishment is 22. The shortest age is 3 years and the longest is 61. Mean BIG4 of 0.82 indicates that the Big 4 auditors occupy over 80% of the listed company audit market. On average, DUAL is 0.29, indicating that 29% of the chairmen of the board of directors are also the CEOs of listed companies. Finally, BDSIZE has a mean value of 10.51, a minimum value of 3, and a maximum value of 32.

Table 4 displays the results of the univariate test to compare the differences in ratios of financial restatements between disciplined vs. non-disciplined audit partners, and pre-sanction vs. post-sanction periods (2 × 2). The proportion of restating financial statements audited by disciplined audit partners in the pre-sanction period (PRE) is 0.02 and that of non-disciplined audit partners (CONTROL) is 0.01. The ratio difference reaches a statistically significant level (t = 2.10 with p-value = 0.04, z = 2.75 with p-value = 0.01). For disciplined audit partners, the proportion of financial restatements in the pre-sanction period (PRE) is 0.02 and reduces to 0.01 in the post-sanction period (POST). The difference in ratios reaches a marginal significance level (t = 1.48 with one-tailed p-value = 0.07). The proportion of restating financial statements audited by successor non-disciplined audit partners in the post-sanction period (SUCC) is 0.00. The difference in the restatement ratios between PRE and SUCC reaches a statistical significance level (t = 2.95 with p-value = 0.00, z = 2.51 with p-value = 0.01). Finally, the difference in the restatement ratios between POST and CONTROL is insignificant but the restatement ratio of SUCC is significantly less than that of CONTROL (t = –3.18 with p-value = 0.00). Combining the results in Table 4, the audit quality of disciplined audit partners is lower in the pre-sanction period and improves in the post-sanction period. For companies that switch from a disciplined to a non-disciplined audit partner, the audit quality of the new partners is higher than that of the predecessors in the pre-sanction period.

Table 5 presents both Pearson and Spearman correlation coefficients with a maximum of 0.47. As a check on multicollinearity, we calculate variance inflation factors (VIFs). The maximum VIF is 3.38 (less than 10), which indicates that no serious multicollinearity exists in the regression model and hence the collinearity problem is negligible (Kennedy, 1992).

Regression Results

Column 1 of Table 6 displays the logit regression results of financial restatements. We estimate the z-statistics using robust standard error and covariance. We have a LR-statistic of 209.77 (p-value = 0.00), pseudo R² of 24%, and p-value of Hosmer and Lemeshow's (1989) test of fitness of 0.94, denoting a good fit of our regression model. 1

The coefficient on PRE is positive and significant and H1 is supported. The coefficient on POST is not significant and the coefficient on SUCC is marginally significant and negative (p-value = 0.09). Consistent with our expectations, both coefficients on POST and on SUCC are significantly less than that of PRE (p-value of PRE – POST is 0.09 and p-value of PRE – SUCC is 0.00), which supports H2 and H3, respectively. -6 In summary, the audit quality of disciplined audit partners is lower and the probability of financial restatements of their clients is higher in the pre-sanction period. After sanctions, disciplined audit partners significantly improve their audit quality and make it comparable with that of non-disciplined partners. Successor non-disciplined audit partners provide higher audit quality to the new clients accepted from disciplined audit partners to limit the litigation/regulation risk; therefore, the probability of financial restatements of the new clients is lower in the post-sanction period.

The preceding analysis on SUCC in Column 1 does not distinguish between changes in audit firms (where both audit firms and audit partners are different from the previous ones) or audit partner changes (where audit firms are kept but audit partners are changed). Accordingly, Column 2 of Table 6 further divides SUCC into audit firm and audit partner changes. If a company changes to a new audit firm after its audit partner sanction, SUCC_F is equal to 1, and 0 otherwise. If a company changes to a new audit partner of the same audit firm after its audit partner sanction, SUCC_P is set as 1, and 0 otherwise. The results in Column 2 of Table 6 reveal a marginally significantly negative coefficient on SUCC_P that is significantly less than that of PRE, while both SUCC_F and the related Wald tests are insignificant. Audit quality in the early stages of audit engagements is commonly argued to be lower because of auditors' lack of familiarity with their new clients (Gul et al., 2009). When companies change both disciplined audit partners and audit firms, the new audit partners' unfamiliarity with clients offsets part of the deterrent effects in the disciplinary system. As a result, the effects of SUCC mainly come from SUCC_P.

Moreover, the significant difference in audit quality between disciplined audit partners in the pre-sanction period and successor non-disciplined audit partners may be a result of lower audit quality of PRE and not necessarily of the higher audit quality of SUCC. In order to exclude this explanation, we add two research variables to Column 3 of Table 6: AFC and ROTATE. If a company changes its audit firm for reasons other than sanction, AFC equals 1, and 0 otherwise. Accordingly, both AFC and SUCC_F have the same effect of audit firms being unfamiliar with new clients. However, as SUCC_F has an additional deterrent effect of sanctions on AFC, we predict the coefficient on SUCC_F to be smaller than that on AFC. If a company retains its audit firm but changes its audit partner for reasons other than sanction, ROTATE equals 1, and 0 otherwise. Both SUCC_P and ROTATE denote audit partners that although belonging to the same audit firm lack familiarity with new clients. However, as SUCC_P has an additional deterrent effect of sanctions on ROTATE, we expect the coefficient on SUCC_P to be smaller than that on ROTATE.

Column 3 of Table 6 indicates that the significance levels of SUCC_P and SUCC_F do not change substantially. The coefficient on ROTATE is significantly positive. This indicates that the audit quality of the first year of audit partner rotation is poorer, consistent with prior literature (e.g., Chi et al., 2009). The Wald test of linear combination of coefficients displays no significant difference between SUCC_F and AFC (p-value = 0.88). However, the coefficient on SUCC_P is significantly less than that on ROTATE (p-value = 0.01). This finding indicates that audit partners who accept new clients previously audited by disciplined audit partners are more diligent in reducing the probabilities of financial restatements of their new clients than audit partners whose new clients are previously not audited by disciplined audit partners.

The results of the control variables reveal significantly positive coefficients for LEV, LOSS, BDSIZE, and RES_1, supporting the positive financial restatements effects caused by debt covenant incentives, financial distress, inefficient monitoring of the board of directors, and prior period restatements.

Chow Tests

Table 6 assumes that slope coefficients of the control variables are constant in the pre- and post-sanction periods. Only the changes in intercepts are affected by auditor disciplinary actions. Table 7 relaxes the assumption of constant slope coefficients and employs the Chow Test to test the differences in the coefficients of sanction-related variables. Column 1 of Table 7 performs the logit regression analysis of financial restatements for disciplined audit partners in the pre-sanction period with a sample size of the sum of firm-years audited by disciplined audit partners in the pre-sanction period (N = 724), and firm-years audited by non-disciplined audit partners (N = 5,247). Column 2 is the logit regression results for disciplined audit partners in the post-sanction period with a sample size of the sum of firm-years audited by disciplined audit partners in the post-sanction period (N = 346), and firm-years audited by non-disciplined audit partners (N = 5,247). Finally, Column 3 displays the logit regression results for successor non-disciplined audit partners with a sample size of the sum of firm-years switched to successor non-disciplined audit partners (N = 449), and firm-years audited by non-disciplined audit partners (N = 5,247). The results of Chow Tests in Table 7 do not substantially differ from those in Column 1 of Table 6.

Pre-sanction Analysis of Audit Quality Level

Audit Quality Enhancement by Successor Non-disciplined Audit Partners

In this section we compare the difference in audit quality between disciplined audit partners in the pre-sanction period (N = 724) and successor non-disciplined audit partners (N = 449). We use the non-disciplined audit partners (N = 5,247) as a control group and obtain a final sample size of 6,420 in this section. Related empirical results are presented in Table 10.

Based on whether the successor non-disciplined audit partners are from the Big 4 or Non-Big 4 auditors, Column 1 of Table 10 divides SUCC into two dummy variables, SUCC_UB4 and SUCC_DNB4. When successor non-disciplined audit partners come from the Big 4 auditors, either from Big 4 to Big 4 or from Non-Big 4 to Big 4, SUCC_UB4 equals 1, and 0 otherwise. The same rule applies to SUCC_DNB4. The empirical results show that the coefficient on SUCC_UB4 is marginally significantly negative and is significantly less than that of PRE. However, the coefficient on SUCC_DNB4 and the difference in coefficients between PRE and SUCC_DNB4 are not significant. This indicates that the audit quality of the successor non-disciplined audit partners from the Big 4 auditors is higher than that of disciplined audit partners in the pre-sanction period. Column 2 lists the results on the type of sanctions. If a company switches to a successor non-disciplined audit partner due to a serious sanction (e.g., suspension and revocation) imposed on a disciplined audit partner, SUCC_STYPE equals 1, and 0 otherwise. By contrast, if a company switches to a successor non-disciplined audit partner because of a minor sanction (e.g., warning and reprimand) imposed on a disciplined audit partner, SUCC_MTYPE equals 1, and 0 otherwise. The empirical results indicate that the coefficient on SUCC_STYPE is marginally significantly negative and is significantly less than that on PRE, while the coefficient on SUCC_MTYPE and the related Wald test results are not significant. This means that the more severe the disciplinary action, the higher the deterrent effect of the sanction on successor non-disciplined audit partners.

In Taiwan, two audit partners are required to sign the audit report. Based on whether the successor non-disciplined audit partner is the other signing audit partner or not, Column 3 of Table 10 divides SUCC into two dummy variables, namely, SUCC_NEWP and SUCC_KEEPOP. If a company replaces the two signing audit partners because one or both are being disciplined with two new non-disciplined audit partners, SUCC_NEWP equals 1, and 0 otherwise. -24 If a company replaces one disciplined audit partner (with a new non-disciplined audit partner) but retains the other non-disciplined incumbent audit partner, SUCC_KEEPOP equals 1, and 0 otherwise. The empirical results indicate that the coefficients on SUCC_NEWP and SUCC_KEEPOP are not significantly negative but are significantly less than that on PRE. This indicates that Taiwanese audit partner disciplinary actions improve the audit quality of successor non-disciplined audit partners regardless of whether the successor is the other signing audit partner or not.