“The impact of external auditor size on the relationship between audit committee effectiveness and earnings management”

This research aims to explore new evidence on the nature of the relationship between the effectiveness of audit committee and earnings management in one of the emerging economies, Jordan. In addition, it investigates how external auditor size might moderate this relationship. For this purpose, a panel data consisting of 64 industrial firms listed on Amman Stock Exchange (ASE) is used, covering the period between 2009 and 2014. An index consisting of four characteristics is developed to measure the effectiveness of audit committee, namely audit committee independence, size, meetings and financial expertise. Results show that audit committee effectiveness has a significant and negative impact on earnings management. Moreover, a positive interaction effect of external auditor size and audit committee effectiveness on earnings management is found, which is supportive of the substitute relationship between the external auditor size and effective audit committee in reducing earnings management. Policy makers and professional accounting bodies in Jordan might benefit from these results, as they show that legislative reforms can motivate firms to adopt good governance practices to mitigate earnings management. A considerable research has tested the influence of the individual characteristics of AC mechanism on managed earnings. This method has been criticized in recent studies for its inability to represent the effectiveness of AC. According to DeZoort al. the AC effectiveness framework could improve significantly if AC characteristics are considered together. Following the same rationale, the current study examines AC characteristics (in-dependence, size, frequent meetings and financial expertise) as a com-©


INTRODUCTION
Corporate governance (hereafter, CG) has developed a number of control mechanisms considering maximizing stakeholders' interests. These mechanisms are numerous, yet the audit committee (hereafter, AC) and external audit (hereafter, EA) remain the most appealing in enhancing the quality of financial reports (Cohen et al., 2004). Moreover, both of ACs and EAs have the ability to mitigate opportunities available to managements for earnings manipulation (Bradbury et al., 2006).
As regards the effectiveness of the AC mechanism, Mangena and Pike (2005) state that AC characteristics, for instance, independence, degree of financial expertise, size and the number and nature of meetings are a measurement of its effectiveness. A considerable research has tested the influence of the individual characteristics of AC mechanism on managed earnings. This method has been criticized in recent studies for its inability to represent the effectiveness of AC. According to DeZoort et al. (2002), the AC effectiveness framework could improve significantly if AC characteristics are considered together. Following the same rationale, the current study examines AC characteristics (independence, size, frequent meetings and financial expertise) as a com-posite measure to capture their combined influence on both the practical and the theoretical sides of earnings management in the emerging market of Jordan.
EA is thought of as another vital monitoring mechanism that helps in aligning the interests of shareholders and managers. This would consequently reduce the likelihood for management opportunistic behavior. For instance, Frankel et al. (2002) provide evidence of the adverse relationship between the monitoring quality of a high quality EA and earnings management. Yet, for this mechanism to act properly on this matter, EA must provide high quality audits. Several researchers state that most previous research has mainly proxied audit quality using auditor size (e.g. Francis, 2004;Ronen & Yaari, 2008). This auditor differentiation is based on the argument of DeAngelo (1981) who proposes that audit firm size is an appropriate proxy for audit quality, because no single client is important to large auditors, as they have superior reputation to lose should they behave opportunistically. Using this reasoning, this study uses auditor size to proxy for external audit quality.
Moreover, while the aforementioned CG mechanisms aim to ensure the quality of the process of financial reporting, the relationship between AC effectiveness and EA quality has been considered as controversial one. Existing literature emphasizes two possible approaches in an attempt to clarify how AC effectiveness and EA quality interact: ACs and EAs can be considered as either substitutes or complements for each other (Hay et al., 2008). The first perspective entails that these CG mechanisms substitute each other given that internal control mechanisms react to changes in external control and vice versa. Thus, a negative effect is anticipated between the AC and the EA. The second perspective acknowledges that internal governance mechanisms are required for the external mechanisms to act, resulting in a complementary relationship between these mechanisms. Hence, a positive relationship is expected between AC and EA (Adjaoud et al., 2007). Therefore, the current study attempts to extend prior literature through examining both approaches through the measurement of the moderating influence of EA quality on the relationship between AC effectiveness and earnings management.
Accordingly, the current study will be organized as follows: section 1 is devoted to previous literature and the theoretical framework and hypotheses, whereas section 2 deals with the methodological approach, section 3 is devoted to the discussion of results and finally last section presents the conclusions.

Audit committee effectiveness and earnings management
Monitoring and controlling the process of financial reporting are often considered the main duties of boards of directors. However, this role is often delegated to a subcommittee called the AC. The main role of the AC stems from the internal CG of firms, because it is about monitoring and establishing the financial reporting process in order to deliver appropriate and highly credible information for stakeholders (McMullen, 1996). According to the agency theory, the AC is an effective monitoring mechanism conditional on having independent directors, members with financial expertise, sufficient directors, and frequent meetings (Carcello et al., 2006). It should be noted that these studies examine relationships on an individual basis. The results from previous research that used the effectiveness of a single governance mechanism might be ambiguous by proving that the effect of some single mechanisms on corporate performance disappeared in the combined measure (Agrawal & Knoeber, 1996). More recently, the function of governance mechanisms can be better as a group compared with the stand-alone mechanism (Ward et al., 2009 uses an index for AC characteristics and investigates its effectiveness in deterring earnings management. Their results provide strong evidence that the CG quality negatively affects earnings management. Nevertheless, apart from being the only research found to have established an index for AC characteristics, its index only includes the existence of AC, number of meetings, and financial expertise, and thus excludes ACs independence and size from the index. This might be explained, at least partially, by the low level of non-financial information reported by firms listed on ASE. Since the effectiveness of AC is considered one of the most crucial elements in the mechanisms of internal CG, and depends on its characteristics, it could be argued that AC that have a higher score for its effectiveness have more ability to monitor the management, protect shareholders' interests and help to restrict earnings management practices. Hence, this study uses a composite measure for AC effectiveness as proposed by Jenkins (2002) and Kent et al. (2010) and seeks to explore the effective-ness of AC characteristics in deterring practices of earnings management in Jordanian firms. As such, this research provides a more comprehensive index for AC effectiveness and measures its effect on accruals earnings management.
On this basis, firms with high quality ACs in Jordan are expected to have enhanced high quality financial reporting systems. Thus, the following hypothesis is developed: H1: There is an adverse relationship between audit committee effectiveness and earnings management.

Impact of Big 4 audit firms on the relationship between AC effectiveness and EM
Early research conducted by Eichenseher and Shields (1985) and Menon and Williams (1994) documents significant effect of the audit firm size and the formation of AC. The researchers conclude that firms hiring big audit firms are more exposed to voluntarily form an AC. This interaction between ACs and EAs can probably provide external stakeholders with financial information of high quality (Mitchell et al., 2008). However, such interaction raises the issue of the substitution and complementary contrasting approaches to these two monitoring mechanisms. That is, previous research introduces two approaches to justify the relation between external audit quality and AC effectiveness: ACs and EAs are thought of as either substitutes or complements to each other.
On the one hand, the substitute approach demonstrates that the AC effectiveness is adversely linked to audit quality. This is since either ACs or EAs monitoring mechanisms is likely to be adequate in guaranteeing high quality external reporting.

Population and sample selection
The population of this study consists of industrial firms registered on ASE during the period 2009-2014. The sample includes all industrial firms that have available data related to CG and earnings man-agement. The total number of industrial firms is 81, 64 of which are used in the study. Of the 17 excluded firms, 11 firms are omitted, because they represent the mining industry 1 , and 6 firms are omitted due to data unavailability regarding CG information and/ or financial information for the test period.
As regards data collection, ASE website (www.ase. com.jo) provides all financial information about listed industrial firms that are needed for both the estimation of discretionary accruals and hypotheses testing. However, the annual reports of research sample do not provide sufficient information about CG variables. Due to data unavailability, this study follows Gabrielsen et al.'s (2002) approach to collect the needed nonfinancial data. That is, a questionnaire is specifically designed and distributed for the sole purpose of gathering data that are not available on ASE website.

Empirical models
To facilitate testing hypotheses of this study, two regression models are used. The first model is designed to examine the impact of AC effectiveness on EM. The second model is designed to measure the effect of Big 4 auditors on the relationship beetween AC effectiveness and EM through the inclusion of an interaction term between AC effectiveness and Big 4 audit firms.
Model (1) 1 Idris (2012) reports that the mining sector cause the normality of residuals assumption to be violated because of the heterogeneity of accounts, operations, and size between mining companies and the rest of industrial companies in Jordan.

Independent variables
_ ACE Score -AC effectiveness, a composite score measuring the effectiveness of audit committee ranging between 0 and 4 with 0 indicating the lowest effectiveness and 4 the highest effectiveness. The score is formed by aggregating the composite scores obtained from four AC constructs, independence, size, meetings, and financial expertise; ACIND -AC independence is coded "1" if the percentage of independent directors on the committee is above the sample median and "0" otherwise; ACSIZE -AC size is coded "1" if the number of audit committee is above the sample median, and "0" otherwise; ACMEET -AC meeting is coded "1" if the number of AC meetings is above the sample median and "0" otherwise; ACFIN -AC financial expertise is coded "1" if the proportion of financial experts on the committee is above the sample median and "0" otherwise;

Estimation of discretionary accruals
In line with the previous studies such as Klein (2002) and Alves (2012), the current study uses the magnitude of discretionary accruals (DA) as a proxy for the EM through using Kothari et al.'s (2005) model, because this model proved to be more robust in detecting earnings management in recent years. This model is as follows:   Table 2 illustrates that ACE_ Score has a negative significant relationship with earnings management supporting the first hypothesis of the research. This finding supports the agency theory prediction, which posits that an effective AC leads to improved financial reporting quality. It is also worth noting that this result is similar to that of Abbadi et al. (2016) in spite of the differences in the characteristics included in the indices of AC effectiveness. Additionally, the results suggest that ACs is more able to constrain earnings management when all effective characteristics are present simultaneously. This means that ACs that consist of a minimum of three members, consist of independent directors, include no less than one expert in finance, and meet no less than three times a year, mitigate earnings management more than ACs that do not possess all four characteristics. Therefore, such index seems to be much influential more measurement of effectiveness than any of the separate variables. As such, effective ACs in Jordan play a vital role in the improvement of cor- Note: Total number of observations for all variables is 384; DA (earnings management) -absolute value of discretionary accruals estimated by Kothari et al. (2005) model; ACE_Score (AC effectiveness) -score for effectiveness of the AC, which can range from 0-4; BIG4 (audit firm size) -equals "1" if the firm is audited by a BIG 4, and "0" otherwise; SIZE (firm size) -natural logarithm of total assets; LEV (leverage) -total liabilities scaled by total assets; BSIZE (board size) -total number of directors on the board.

Main empirical results
porate financial reporting through minimizing the level of opportunistic earnings management practices.

CONCLUSION
The key purpose of this study is to examine and explore the AC effectiveness in mitigating earnings management in the Jordanian setting. In addition, the study tests for the moderating effect of EA's size on the association between the score of AC effectiveness and earnings management. The results of this study found that earnings management is negatively associated with AC effectiveness as a composite measure. This is consistent with agency theory which posits that enhancing the effectiveness of the internal corporate mechanisms, such as AC, mitigates agency conflicts and hence should deter opportunistic earnings management practices.
Furthermore, this study provides evidence that AC effectiveness is significantly and negatively associated with earnings management. When controlling for the moderating effect of EA's size, the association between AC effectiveness and earnings management has weakened but remained significant. This means that EA is less effective in reducing earnings management when the AC is effective. Therefore, it could be argued that high audit quality is more effective in improving the quality of financial information in contexts with weak legal regulatory systems than in contexts with strong legal regulatory systems, because high audit quality might function as a substitute to internal CG mechanisms.
The first limitation of the study arises from the mere use of Jordanian data. The findings of the study are recommended to be cautiously generalized to contexts with diverse economic conditions and regulations. The study sample may represent another limitation, as it only considers listed industrial firms.