| 英文摘要 |
The purpose of this paper is to examine the influence of auditor industry specialization on earnings quality in the banking industry. A growing body of research investigates the importance of audit firm industry expertise and most of the findings show that an industry specialist offers a higher level of assurance than does a nonspecialist (Craswell, Francis, and Taylor 1995; Gramling and Stone 2001). Prior research also suggests greater concentration in regulated industries and different patterns of growth for industry experts in regulated vs. nonregulated industries (Hogan and Jeter 1999). However, no existing paper directly examines or tests the effectiveness of auditor industry specialization on one of the key regulated industries － the banking industry. Herein, we develop a testable hypothesis and use Taiwanese banks as a sample to test our conjecture, and, provide an insight for future research on this topic. The provision for bad debts could expose the evaluation on the quality of loans (an asset for banks). Liu and Ryan (1995) suggested that when banks expect future cash flows to increase, they tend to manipulate discretionary bad debts in order to smooth out earnings. Meanwhile, Liu, Ryan and Wahlen (1997) argued that in order to comply with the capital adequacy requirements established by regulators, 'at risk' bank managers tend to report loan loss provision aggressively. Wahlen (1994) indicated that as the commercial bank loan portfolios are 10 to 15 times greater than bank equity, the discretionary component of unexpected loan loss provisions are positively correlative with stock returns. In addition, the changes in bad debts are also an important signal to the market. Therefore, bank managers have incentives to use a discretionary portion of the provisions to manage earnings. Krishnan (2003) suggested that specialist auditors mitigate accruals-based earnings management more than nonspecialist auditors. Also, audit specialists are argued to be suppliers of higher quality audits because they have superior knowledge and ability in identifying 'industry-specific' problems and issues' (O'Reilly and Reisch 2002). We expected and found that earnings quality of banks audited by industry specialists is higher than by nonspecialists. Different from most of the past studies, this paper focused on a specific regulated industry - the banking industry, to test whether the discretionary part of provision for bad debts, a proxy of earnings quality, are higher for the banks audited by an industry specialists. 2. Hypothesis DeAngelo (1981) defines audit quality as the probability of detecting and reporting major financial statement errors, which will depend on the auditor's proficiency and independence. The first probability is a testimony of the professionalism of auditors. However, Bell, Knechel, Payne and Willingham (1998) found that in the sample of misrepresentations detected by auditors, only 3.6% of them meant to be misleading; most of the misrepresentations were caused by random errors. In other words, only a very small portion of misrepresented financial statements detected by auditors were the result of fraudulent intentions. This fact highlights the importance of audit quality, and auditor industry specialization is part of the audit quality. Gramling and Stone (2001) show that audit firms develop industry in order to accomplish multiple objectives. For example, audit industry expertise helps firms increase the demand for audit or nonaudit services within the focal industry. Audit industry expertise also potentially improves firm efficiency through economies of scale resulting from concentrating resource and technology investments in a focal industry. In addition, audit firm industry expertise also helps differentiate audit market products and creates barriers to entry from competitors. Finally, within the market for audit services, industry expertise is likely to affect market performance, including audit fees and audit quality. Liu (1999) argued that bank managers can manage earnings via accrual items on the income statements (such as the provision of bad debts) and the timing of realizing profits and losses (such as the disposal of long-term assets). Window dressing of financial statement is particularly obvious in the fourth quarter, when the fourth quarter ratios of interim income manipulation items are large and more variable than those of the other three quarters, illustrating that the Taiwanese banking industry is actively engaged in earnings management. Bank managers may manipulate earnings under political pressures, or to smooth out earnings or to meet the target specified in the reward scheme for managers. Accordingly, Liu (1999) argued that in good years, managers may recognize more expenses or losses as a reserve to manipulate earnings at bad times. Meanwhile, a commercial bank manager has incentives to reduce regulatory costs imposed when the bank's capital adequacy ratio falls below its regulatory minimum. On the other hand, to avoid investors from questioning their dividend policies, banks are likely to keep capital adequacy ratios at a moderate level. A number of literature indicated that commercial bank loan loss provisions have significant influence on stock returns (Liu and Ryan 1995; Wahlen 1994; and Grammatikos and Saunders 1990). The changes in bad debt provisions or the recognition of bad debt expenses are signals to the market, but the direction of their effects are not consistent. There is growing literature that links industry specialization with financial reporting quality or earnings quality (Balsam, Krishnan and Yang 2003; Gramling and Stone 2001). Earnings quality is a concept that is not observable, and therefore a variety of proxies are used in the literature. However, loan loss provisions are a relatively large accrual for commercial banks and therefore, have a significant impact on banks' earnings and regulatory capital (Ahmed, Takeda and Thomas 1999). Also, Liu, Ryan and Wahlen (1997) indicated that commercial bank loan loss provisions are largely discretionary. This linkage leads our argument that a banking industry specialist auditor should be able to control the absolute value of dis cretionary part of bad debts, a proxy of earnings quality, generated by the bank's accounting system. Hence, this paper established the following hypothesis: Hypothesis: The absolute value of the discretionary part of the provision of banks audited by specialist auditors is lower than that of the banks audited by nonspecialist auditors. 3. Research Methods This paper referred to the audit industry specialization as the test variable to explore the correlation between the industry specialization and the absolute value of the discretionary part of the provisions. The model is as follows: AbsDBADi,t= 0+ 1SPi,t+ 2BISi,t+ 3EVARi,t+ 4EBTPi,t+ 5SIZEi,t+ 6|CFO|i,t + 7STOCKi,t+ 8FIRMAGEi,t+ 9Y2001+ 10Y2002 + 11Y2003+ i,t (1) where, AbsDBAD: The absolute value of the provision for bad debts; SP: The variable of interest, SP, is defined in six ways. They were LEADER, the sum of the sales revenues of audit firm i / sales revenues of all banks; DOMINANCE, square root of total assets audited by audit firm i / square root of assets for all banks; MOSTCL, dummy variable of 1 for the audit firm with the largest number of banking clients and 0 for otherwise; OTHERIND, number of banking clients audited by audit firm i / number of all clients audited by audit firm i; SHARECL, market share of the audit firm calculated by the number of banking clients; NCLIENTS, number of clients of the audit firm; BIS: Capital adequacy ratios; EVAR: Standard deviation of earnings. EBTP: Pretax earnings before bad debts deflated by total assets at the beginning of the period; SIZE: Bank size, defined as total loans divided by total assets; |CFO|: The absolute value of operating cash flows deflated by total assets at the beginning of the period; STOCK: Percentage of shares owned by the insiders; FIRMAGE: Audit-firm tenure (years); Y2001: Dummy variable, 1 for fiscal year 2001, 0 otherwise; Y2002: Dummy variable, 1 for fiscal year 2002, 0 otherwise; and Y2003: Dummy variable, 1 for fiscal year 2003, 0 otherwise. The coefficient of 1 in Equation (1) is used to test our hypothesis, and expected to be positive. That is, there is a positive association between measures of earnings quality and auditor industry specialization. Six different indicators based on sales revenue, total assets, and the number of clients, were used to proxy the audit industry specialists. Also, several variables have been identified that are correlated with the absolute value of the provision for bad debts. BIS defined as ratio of actual regulatory capital. The capital management hypothesis (Ahmed, Takeda and Thomas 1999) posits that managers of banks with low regulatory capital have incentives to increase discretionary provision for bad debts. This suggests that the coefficient of BIS will be negative. EVAR is measured by the standard deviation of earnings over the sample period. Variability in the earnings series increases the bank's probability of bankruptcy as well as its perceived risk. In turn, the coefficient of EVAR is expected to be positive. EBTP is measured as earnings before tax and provision deflated by total assets at the beginning of the period. This variable is used to control for the potential effects on the absolute value of the discretionary provision for bad debts of motivation related to income smoothing. The coefficient of EBTP is expected to be negative. SIZE is measured as total loans divided by total assets. Kanagaretnam, Lobo and Yang (2005) suggest that using total loans outstanding as a proxy for size is better than other alternative proxies such as market value of equity and total assets because the room for managerial discretion over discretionary provision for bad debts mainly depends upon the magnitude of outstanding loans. We expected that the coefficient of SIZE is positive. The absolute value of operating cash flows (|CFO|) is used to control for the effects of future cash flows on loan loss provisions. Whalen (1994) and Liu and Ryan (1995) show that provision for loan losses are positive correlative with future cash flows and stock returns. STOCK is measured as the percentage of shares held by the insiders. Managers have stronger incentives to manipulative earnings while they have more shares owned in their company. That is, the coefficient of STOCK is expected to be positive. We also include audit-firm tenure (FIRMAGE) to control for the effect of audit quality on earnings quality. However, we did not expect the sign of the coefficient of FIRMAGE. Y2001, Y2002, and Y2003 are year dummies, and used to control for the time effects. Most of past studies followed the estimation equation developed by McNichols and Wilson (1988) for the selection of bad debt estimation models. This paper also adopted the same model to estimate the expected bad debts. The equation is as follows: BADi,t= 0+ 1*LLRi,t-1+ 2*WOFFi,t+ 3*WOFFi,t+1+ i,t (2) BADi,t: The provision of bad debts of company i during period t, deflated by period t net sales; LLRi, t-1: The beginning balance in the allowance for bad debts period t, deflated by period t net sales; WOFFi, t: Write-offs for period t, deflated by period t net sales; WOFFi,t+1: Write-offs for period t+1, deflated by period t net sales; and : Error term. This model is based on a balance sheet perspective. Under the null hypothesis that management does not manipulate earnings, the provision depends on the allowance for bad debts, the current period's write-offs, and the management's expectation of future write-offs condi tional on their current information. Discretionary provision for bad debts (DBAD) is the residual terms in equation (2) and the absolute value of DBAD is denoted as AbsDBAD. 4. Research Findings This paper investigated the correlation between auditor industry specialization and earnings quality of banks by referring to discretionary bad debts as an indicator. It analyzed whether the banks audited by the industry specialists from 2001 to 2004 had higher earnings quality. A total of six indicators based on sales revenue, total assets, and the number of clients, were established to evaluate the audit industry specialists. Discretionary or unexpected part of bad debts is estimated using the model proposed by McNichols and Wilson (1988). Data for this study are sourced from the Taiwan Economic Journal database. Banks which have merged from 2001 to 2004 and show no available data are deleted from the sample pool, with the final number of banks at 24 and with a total of 95 observations for this study. Table 1 is the sample descriptive statistics, where the absolute value of discretionary bad debts (AbsDBAD) is 0.0609. Data for industry specialist shows that all measured variables indicate that industry specialists have a strong foothold in terms of their market share. The list of audit firms in the banking industry from 2001 to 2004 is shown in Table 2. Table 2 indicates that from 2001 to 2004, Deloitte has the most number of audited clients and has the highest market share versus the other Big 3 and other audit firms in the market. Tables 3 to 7 are the results for LEADER, DOMINANCE, NCLIENT, SHARECL, and OTHERIND which show the univariate results between the Big 4(5) and other audit firms, results indicate that each industry specialist variable for all six audit firms are different and significant (p-value<0.0001). The mean differences between groups show that Deloitte has a higher and significant mean versus the other audit firms. Table 8 is the AbsDBAD univariate analysis results between the Bi g 4(5) and other audit firms, results indicate that there exist significant differences between the Big 4(5) and other audit firms. From the univariate analysis results (Table 3 to table 8), Deloitte appears to be the industry specialist leader in the Taiwan's banking industry. Table 9 shows the Pearson correlation test results between group (2x2) variables, with the between group variable correlations ranging from -0.715 to -0.001; further, all industry specialist variables and AbsDBAD show negative yet statistically insignificant correlation. The empirical findings in Table 10, as expected, we find that all six indicators of auditor industry specialist are negatively correlative with the absolute value of the discretionary part of bad debts. Only the findings in DOMINANCE (relative market shares measured with total assets) indicator fail to reach conventional statistical significance level. Findings in the SHARECL (market share of the audit firm calculated by the number of banking clients) and NCLIENTS (number of clients of the audit firm) indicators are significantly at 5% significant level. Overall, these findings moderately supported our conjecture that the discretionary bad debts of the banks audited by the industry specialists were lower than that of the banks audited by nonspecialsts. The adjusted R2 was mostly between 11% and 15%, an indication of the explanatory power of the model as a whole. The use of proxies to measure auditor industry specialization is a limitation of this study. Future studies should focus on refinements of the industry specialists' measures, with a view to being able to incorporate the effect of individual auditors. |
本站僅提供期刊文獻檢索。
