278 /304
沙賓法
404
條及審計準則第
5
號是否會減少內部控制揭露錯誤？
278
5. Sensitivity and Additional Tests
5.1 Measurement Validity of Type I and Type II Errors
Because ICFR disclosure-quality cannot be easily observed or directly gauged, we
regard restatement companies as more likely to have material internal control weaknesses
and, therefore, to receive adverse ICFR-audit opinions. Although prior studies suggest that
misstatements are indicative of ineffective internal control systems (Kinney and
McDaniel, 1989; DeFond and Jiambalvo, 1991; McMullen et al., 1996; Eilifsen and
Messier, 2000; Leone, 2007; Rice and Weber, 2012), some extenuating or unique
situations in which a misstatement is not always parallel to ineffective internal controls
(PCAOB, 2004) may still exist. Likewise, non-restatement companies cannot be
guaranteed perfection in ICFR. That is, companies may conclude that their actual weak
internal control systems are ineffective, but their financial reports are accurately stated.
Therefore, those companies do not make Type I errors in ICFR disclosures, and our
measurement of the Type I error likelihood may be overestimated in this study.
One benefit of ICFR audits is to identify and remediate control deficiencies or
weaknesses, in turn leading to improvement of internal control systems. Schroeder and
Shepardson (2016) have shown that the implementation of SOX 404 is associated with
improved internal control system quality. Based on their results, we expect fewer non-
restatement companies that conclude their actually weak internal control systems as
ineffective after the implementation of SOX 404. If our measurement of the Type-I error
likelihood is significantly overestimated because of the inclusion of those non-restatement
companies, the overestimation is expected to lessen under the SOX 404 regime. However,
we found no evidence that the enactment of SOX 404 affected the Type-I error rate. Thus,
we believe that the possible lack of measurement validity does not drive our empirical
results.
To further settle the concern on the measurement validity in the study, we try an
alternative method to identify companies with ineffective ICFR. Ashbaugh-Skaife,
Collins, Kinney, and LaFond (2006) and Doyle et al. (2007a) have suggested that internal
control problems are associated with lower accrual quality, because companies with weak
internal controls fail to detect intentionally biased accruals resulting from earnings
management or unintentional accrual estimation errors more probably. As a result, low
accrual quality could be viewed as a strong indicator of material weaknesses in internal
control. We measured accrual-estimation error based on the method developed by Dechow