Mobile Edition
Print
Get the Mag
Weekly UpdatesTheoretical Framework for the Effects of Audit Quality
The audit process, in general, has been shown to improve internal and organizational controls. Wallace (1980) asserts that the audit process deters fraud by creating the threat of discovery. Auditors' review of the internal control system improves the control environment and helps eliminate carelessness. Auditors not only find errors, but also recommend process improvements; auditors typically issue a report to management containing suggestions and recommendations for improving the accounting system and process.
A number of theoretical and empirical studies have argued that large or Big 5 auditors are recognized as brand name suppliers of audits and that these audits are perceived to be of higher quality (e.g., DeAngelo, 1981; Simunic and Stein, 1987; Schwartz, 1997; Solomon et al., 1999). In this study we rely on DeAngelo's (1981) seminal theory of auditor reputation that uses traditional agency theory to describe why larger auditors have incentives to provide high quality services. She argues that in order to maintain their investment in reputation capital, larger firms will provide higher quality services. Due to their size, larger auditors have the resources to invest more in personnel training and technology and, therefore, have greater skill in developing internal controls for clients and detecting breaches of the accounting system. Also, due to their size, larger auditors are less dependent on any given client for fee revenue and thus can be more resistant to client pressures in reporting accounting breaches, increasing the auditors credibility with non-management constituents. In summary, larger auditors have better accounting skills and are more independent, which ultimately creates brand name reputation.
Our study extends DeAngelo's (1981) theory to the regulator as a constituent of the bank who relies on audit quality. The role of the auditor in our model is based on the effect of auditor reputation on the regulator's perception (i.e., the credibility of the auditor's opinion) and the skill and ability of the auditor to improve management and organizational control. Adopting DeAngelo's (1981) theory for banking firms, a high quality auditor has a greater probability of improving accounting controls and detecting accounting breaches because of greater accounting skills, and has a greater probability of reporting a breach because the auditor is larger in size (i.e., size enhances independence). We also suggest that since management has significant influence in hiring the auditor, the choice of a large, high quality auditor also indicates management's willingness to undergo a rigorous review which sends a signal to all constituents that management is confident and competent in its financial management and repo rting.
Our prediction of the effect of audit quality on regulators is developed as follows. If regulatory examiners value the certification work of external auditors, they may rely more on audited financial reports, limiting their own work and improving their evaluation of management controls and financial results. Bank managers, realizing these effects, may influence the extent of the regulatory examination by selecting high quality auditors. Since previous research indicates high quality auditors improve the reliability of both financial statements and internal control systems, auditors may then influence the examiners' perception of the bank's financial condition and the quality of its management. Thus, we expect to see a positive, significant relation between auditor quality and regulatory evaluation results. Our hypothesis, stated in alternative form, is:
[H.sub.A]: Regulators' evaluations are positively influenced by managers' choice of high quality auditors.
RESEARCH DESIGN
Model
To test our hypothesis, we would ideally examine how CAMEL ratings are associated with proxies for audit quality. CAMEL ratings are a direct output measure of the regulator's onsite examination and are based on objective and subjective measures of bank financial condition. Auditor choice is not explicitly defined as an input to the CAMEL in the regulatory process, but could be incorporated subjectively through the regulator's personal judgment. However, CAMEL ratings are not publicly available. The results of bank examinations are intended for bank directors and management only. The privacy of these reports is enforced in the banking industry.
Instead we use ratings provided by the Sheshunoff Service for the year 1996. The Sheshunoff ratings provide an objective composite measurement of historical bank performance based on four of the five CAMEL factors: Capital Adequacy, Asset Quality, Earnings, and Liquidity (Sheshunoff does not calculate a rating for Management). The Sheshunoff ratings are calculated using publicly available financial information obtained from the release of the preliminary reports of condition (analogous to a balance sheet) and reports of income (analogous to an income statement) from the Federal Reserve Call Reports. Thus, these ratings are mathematically derived and do not include a subjective evaluation of bank management. However, two factors suggest that the Sheshunoff rating can acceptably proxy for the overall CAMEL rating. First, the composite Sheshunoff rating focuses on the bank's health and potential for failure, rather than current performance. Both the measures selected for Capital Adequacy, Asset Quality, Earning s and Liquidity and the weighting of those measures are statistically determined to yield a measure of the probability of long-term success. Second, all raw scores are compared to a five-year industry average by placing each institution's raw score on the industry representative (normal) distribution to determine each bank's rating. This process ensures that the bank's rating reflects its true condition (i.e., healthy banks will not be penalized with a low rating simply because other institutions have higher numbers). These two computational steps help to enrich the raw data and make the Sheshunoff rating a more suitable proxy for the CAMEL rating.
Although the Federal Reserve was unwilling to release CAMEL ratings for our sample period, we did obtain CAMEL ratings for 58 of the largest banks from 1984 to 1986. These CAMEL ratings were obtained by Dr. Chris James while he was a visiting scholar at the Federal Reserve Bank of San Francisco and have been used in previous studies (James, 1988; Cargill, 1989). The Pearson correlation coefficient between the overall CAMEL rating and Sheshunoff ratings during the 1984 to 1986 time period is -0.752 (p-value [less than] .01), indicating that Sheshunoff ratings have a high, negative correlation with CAMEL ratings, which again suggests that the Sheshunoff ratings are a reasonable proxy for CAMEL ratings.
We would like to use Dr. James' sample of 58 banks from 1984 in our analysis, but there are a number of important limitations to the data. First, the sample is small (only 58 versus 252 banks in the current study), which reduces the power of our tests. Second, and most importantly, is the lack of variation in the auditor choice variable: all 58 banks from the 1984 sample used a Big 5 auditor (Big 8 at that time). This limitation prevents us from using the traditional Big 5/Non Big 5 dichotomy as a proxy for audit quality. Third, as the largest 58 banks in the U.S. at that time, this sample is not a random draw from the general population of U.S. banks. This would reduce the external validity of testing. Fourth, the structure of the auditing and banking industries has changed as both industries have consolidated significantly. To make our study contemporaneously relevant, we examine a current sample of banks and auditors. Since the algorithm to calculate Sheshunoff and CAMEL ratings has not changed over time, we expect the correlation between Sheshunoff and CAMEL ratings to be stable, suggesting that Sheshunoff ratings are a reliable proxy for CAMEL ratings.
Since the Sheshunoff ratings are an objective combination of the four weighted CAMEL factors, the use of an audit quality proxy as a direct input factor to the Sheshunoff rating is excluded. We argue, however, that audit quality indirectly affects the Sheshunoff rating by directly affecting the individual CAMEL factors. Informal discussions with regulators indicate that they are aware of the auditor's reputation and the extent of the auditor's work. While we believe that audit quality is a subjective factor in the CAMEL ratings, we assume that its affect is direct. Because the correlation between CAMEL and Sheshunoff ratings indicates an inverse relationship, we suggest the following system of linear equations:
RATING = [[alpha].sub.10] + [[alpha].sub.11]C + [[alpha].sub.12]A + [[alpha].sub.13]E + [[alpha].sub.14]L (1)
C = [[beta].sub.20] + [[beta].sub.21][X.sub.2] + [[beta].sub.22]AUDITOR + [[epsilon].sub.2] (2)
A = [[beta].sub.30] + [[beta].sub.31][X.sub.3] + [[beta].sub.32]AUDITOR + [[epsilon].sub.3] (3)
E = [[beta].sub.40] + [[beta].sub.41][X.sub.4] + [[beta].sub.42]AUDITOR + [[epsilon].sub.4] (4)
L = [[beta].sub.50] + [[beta].sub.51][X.sub.5] + [[beta].sub.52]AUDITOR + [[epsilon].sub.5] (5)
RATING is the Sheshunoff peer or national composite rating; C, for capital adequacy, is core capital as a percentage of assets; A, for asset quality, is adjusted nonperforming assets as a percentage of total assets; E, for earnings, is the return on average assets; L, for liquidity, is liquid assets as a percentage of total liabilities; [X.sub.i] is a vector of bank characteristics and performance measures such as profitability, size, risk, and operational complexity that affect the ith CAMEL factor (specifically, C, A, E, or L); and AUDITOR is a measure of audit quality. Equation (1) does not have a disturbance term since the relationship between the Sheshunoff rating and the CAMEL factors is exact or deterministic. Equations (2) through (5) include a stochastic disturbance term ([[epsilon].sub.i]-N(o,[o.sup.2][I.sub.T])) since these relationships are not deterministic and may be measured with error. Substituting equations (2), (3), (4), and (5) into equation (1) provides the following single equation:
FEED