The Effect of Internal Control Certification Regulatory Changes on Real and Accrual-based Earnings Management

ABSTRACT

This study examines the effect of changes in internal control certification requirements (ICCR) on the earnings management choices of Australian firms in the period 2007–2015. The Australian setting is unique as the certification requirements change from voluntary in 2004–2007 to mandatory in 2008–2014, before being abolished in 2015. Consistent with the notion that real earnings management (REM) is less susceptible to detection, the results suggest that firms place greater reliance on REM than on accrual-based earnings management (ABEM) when having to comply with certification requirements. In particular, I find voluntary certifiers have lower REM and ABEM relative to first-time certifiers in the mandatory period between 2008 and 2014, and there is an increase in REM activities among first-time mandatory period certifiers. Moreover, firms that discontinue certification, after the abolition of the requirement in 2015, switch from REM to ABEM. This suggests that regulatory ICCR changes affect firms’ earnings management choices.

Acknowledgements

I thank the two anonymous reviewers and Professor M. Daniel Beneish for their insightful comments and constructive suggestions received during the review process. I acknowledge the valuable feedback provided by participants at the European Accounting Association Annual Congress, Valencia, 2017. I am grateful to Farshid Navissi and Carla Wilkin for their valuable suggestions to improve the paper. I am also grateful to Cameron Troung and Abu Zafar Shahriar at Monash University for their valuable comments. Errors and omissions are my responsibility.

Notes

1 A thorough search of post-2014 annual reports and corporate governance statements was conducted to determine whether firms continue to disclose certification post-abolition. Because since 2015 the certification disclosure is no longer part of the best practice recommendation (see Principle 7), only 197 firms from a sample of 921 firms continue to make the disclosure in their 2015 annual reports. Of the 243 firms that certified in 2014, when it was mandatory to disclose, only 21 firms provide certification disclosure during the abolition period (2015). A possible explanation for this is that these 21 firms improved the quality of their internal controls and, therefore, had the confidence to provide certification disclosures even after the abolition of ICCR.

2 The internal control documentation required by the Israeli security regulator is similar to SOX302 (Weiss, 2014). In the UK, the Combined Code and the Turnbull Guidance do not require certification by CEOs and CFOs of the quality of financial reports based on internal controls. Similarly, in Sweden there is no requirement for certification from CEOs and CFOs of internal controls and risk management.

3 In Germany, listed firms must establish a wide-ranging risk management system. However, they do not need to publish an explicit statement or certify the effectiveness of the internal controls and risk management system.

4 Based on an extensive search conducted on the ASX website and disciplinary announcements concerning regulatory action for noncompliance with the ASX guidelines on ICCR reporting, there is no evidence in Australia of regulatory action for falsely certifying internal controls. Further, there is no evidence of false certification and related regulatory action on the Australian Securities and Investment Commission (ASIC) website or in general search engines.

5 Firms that do not make any disclosure or comment in compliance with Recommendation 7.2 are considered to be non-certifying firms. This is the only way to segregate certifiers from non-certifiers, and it remains a limitation of my study of the Australian setting.

6 Measures of ABEM using discretionary accruals are criticized because they exhibit considerable variation in performance (e.g., Dechow & Dichev, 2002; McNichols, 2002). However, Jones, Krishnan, and Melendrez (2008) find that all 10 measures of discretionary accruals are significantly associated with a fraudulent event. To address the concerns of McNichols (2002) and Dechow and Dichev (2002), two approaches are adopted in the study based on measures that are widely used in the earnings management literature (e.g., Cohen & Zarowin, 2010; Cohen et al., 2008; Kothari et al., 2016; Zang, 2012).

7 A restructuring charge is not included in the model because the information on restructuring includes five different types of restructures, of which some are material and others are not, making it difficult to determine the impact of restructuring on earnings quality. Restructuring expense information relevant to the model is not available in the notes to the financial statements.

8 Hypotheses Hp1a, Hp1b, and Hp1c are tested using a sample based on the 2008–2014 financial years. Hypothesis Hp2a and Hp2b are tested using a 2015 financial year sample.

9 A thorough search of post-2014 annual reports and corporate governance statements is conducted to determine whether firms continue to disclose ICCR certification post-abolition. Since certification disclosure is no longer part of the best practice recommendation for Principle 7 beginning in 2015, only 197 firms from a sample of 921 firms continue to make the disclosure in their 2015 annual reports. Of the 243 firms that certified in 2014, when it was mandatory to disclose, only 21 firms provide certification disclosure during the abolition period (2015). These 21 firms are more likely to have improved the quality of their ICCR and, therefore, had the confidence to provide certification disclosures even after the abolition of ICCR.

10 The variance inflation factor (VIF) of MKTVAL and BSEGMENT is higher, at 2.480. However, the main results remain unchanged when one of these control variables is dropped.

11 To conserve space, results not reported.