AUDIT COMMITTEE INDEPENDENCE AND A CONTRACTING PERSPECTIVE ON GOODwILL IMPAIRMENT: SINGAPOREAN EVIDENCE

. This study examines factors influencing and constraining the decision to recognize zero goodwill impairment in a sample of 52 Singaporean listed firms from 2010–2012. Using binary logistic regressions, the results reveal that firms that are approaching violation of their debt covenants have a higher likelihood of exercising the recognition choice, while a higher proportion of audit committee independence constrains this choice. The policy implication of this study is that to improve the quality of the financial statements, the relevant authorities need to monitor firms’ reporting incentives closely. This study contributes to the literature on IFRS by providing evidence that supports the applicability of the debt hypothesis in explaining the decision of Singaporean listed firms to recognize zero goodwill impairment. been presented at national and international level. Her research interests focus in the areas of financial reporting, accounting choice, corporate governance, goodwill impairment, ownership concentration and International Financial Reporting Standards (IFRS).

impairmentonly approach to goodwill because the board believed that comparability would be affected when there are alternative accounting methods to acquired goodwill (Fabi et al. 2014).
To date, most research on IFRS implementation tends to focus on listed firms in European countries (e.g., Murphy 2000, Jaafar and McLeay 2007, Yip and Young 2012, Cascino and Gassen 2015. This study attempts to address this gap in the literature by investigating IFRS implementation focus ing on goodwill impairment by listed firms in Singapore, a developed country that belong to the Association of South East Asian Nations (ASEAN). Specifically, this study ex amines factors influencing and constraining the decision to recognize zero goodwill impairment in a sample of 52 Singaporean listed firms from 2010-2012.
To examine factors influencing and constraining the decision to recognize zero goodwill impairment, this study selected firms that encountered booktomarket ratios above one for three consecutive years. With booktomarket

Introduction
The International Financial Reporting Standards (IFRS) have been dispersed worldwide; the standards are being applied in varying degrees by more than one hundred ju risdictions in six different continents (Daske et al. 2013, Danjou 2015. One of the driving forces for the worldwide implementation of the IFRS is the globalization of capital markets, which, in turn, creates increasing demand for transparent and comparable financial reporting by various stakeholders (Glaum et al. 2013).
A number of initiatives were proffered by the International Accounting Standards Board (IASB) to re duce alternative accounting methods with the hope of improving the comparability of financial statements. One example of these initiatives is the issuance of IFRS 3 Business Combinations and the revision of IAS 36 Impairment of Assets, which are related to acquired goodwill. In 2005, the IASB eliminated alternative accounting methods for acquired goodwill and required firms to implement an ratios persistently above one for three continuous years, these firms have indications that goodwill may be im paired, yet they reported zero goodwill impairment. This study selected Singapore as an institutional setting for the analysis of the decision to recognize zero goodwill impair ment because it is one of the few ASEAN countries that have fully implemented IFRS related to goodwill impair ment. More importantly, since ASEAN has become a vital economic entity (Saudagaran and Diga 1998) and is keen on integrating the region through the establishment of the ASEAN Economic Community (AEC) (ASEAN 2008), it is imperative to understand the potential drivers and con straints to IFRS implementation within this region. This study focuses on IFRS 3 Business Combinations, which is related to goodwill impairment 1 , due to the increasing focus from regulators, practitioners and academics (e.g., Beatty and Weber 2006, Godfrey and Koh 2009, Ramanna and Watts 2012, Abdul Majid 2015, 2017.
The remainder of this paper is structured into four sec tions. Section 1 highlights the prior literature and describes the development of the hypotheses. Section 2 outlines the research design. Section 3 presents and discusses the em pirical findings. Section 4 summarizes this study and pres ents its conclusion.

Prior literature and hypotheses development
The harmonization of accounting standards through an IFRS implementation has garnered increasing interest from regulators, academics, and practitioners, both in de veloped and developing countries (Baker andBarbu 2007, Yip andYoung 2012). One of the expectations of an IFRS implementation is to achieve the comparability of finan cial statements, which, in turn will facilitate international transactions and minimize exchange costs (Cairns et al. 2011, Phuong and Nguyen 2012, Yip and Young 2012. Nevertheless, an improvement in financial statements comparability does not depend entirely on the application of a uniform set of accounting standards (DeFond et al. 2011). Factors such as firms' reporting incentives, the level of discretion afforded by specific accounting standards and the strength of regulatory enforcement of the standards may impact the comparability of the financial statements (Ball et al. 2003, Daske et al. 2008. For example, Ball et al. (2003) report that firms' reporting incentives appear to play an important role in the application of accounting standards for listed firms in Singapore.
In the context of goodwill impairment, prior studies have examined the determinants of goodwill impairment losses in various countries, such as the US (e.g., Ramanna and Watts 2012), Australia (e.g., Godfrey and Koh 2009), the UK (e.g., AbuGhazaleh et al. 2011), andMalaysia (e.g., Abdul Majid 2015). These studies provide mixed findings regarding factors that influence the reporting of impairment losses. One group of studies reported that the impairment loss is influenced by managers' reporting incentives while another group of studies showed that the economic impair ment of goodwill drives such reporting behavior (Abdul Majid 2015: 199).
The above mixed results provide an opportunity for this study to examine the issue of goodwill impairment. In ana lyzing the factors that influence and constrain the decision to recognize zero goodwill impairment, this study focuses on the literature related to the contracting perspective of accounting choice, ownership concentration and corporate governance.

Contracting perspective
The contracting perspective was developed by Watts and Zimmerman (1986). This perspective extends the agency theory model of Jensen and Meckling (1976) by emphasi zing the role of accounting numbers in the contract (Watts and Zimmerman 1986). The contracting perspective su ggests that managers choose a particular accounting op tion to influence the accounting numbers employed in the contracts (Watts and Zimmerman 1990). This study explo res the application of the contracting perspective by testing two hypotheses, i.e., debt hypothesis and CEO reputation.

Debt hypothesis
From the contracting perspective, the motives for exerci sing an accounting choice is to influence firms' contractual arrangement with the debtholders (Fields et al. 2001). By exercising the accounting choice, these firms are intending to lessen the accountingbased restrictions stipulated in the debt covenants (Smith 1993), hence avoiding violation of the covenants (Dichev and Skinner 2002). In the context of goodwill impairment, prior studies investigating firms' decisions in reporting goodwill impairment postulate that firms that are approaching violation of their debt covenants have a lower likelihood of reporting goodwill impairment losses (Beatty andWeber 2006, Abdul Majid 2015).
In this study, similar to Abdul Majid (2015), leverage (LEVERAGE), which is measured as debt to total assets, is employed as a proxy for the closeness of firms to violating their debt covenants. Consistent with the debt hypothesis, this study tests the relation between leverage (LEVERAGE) and the decision to recognize zero goodwill impairment in an alternative form as follows: H 1a : Ceteris paribus, there is a significant positive as sociation between leverage (LEVERAGE) and the decision to recognize zero goodwill impairment.

CEO reputation
Prior studies argue that existing top managers may not have reported goodwill impairment losses (by reporting zero goodwill impairment) when their firms' booktomarket ratio was above one because of concern for their personal reputations (Beatty and Weber 2006, MastersStout et al. 2008, Ramanna and Watts 2012. The top managers who were directly involved in the creation of goodwill through business combinations may be reluctant to writeoff the goodwill as doing so would imply that they were unable to realize the expected synergies from the business combi nations (LapointeAntunes et al. 2008, MastersStout et al. 2008. Similar to prior studies (e.g., Beatty and Weber 2006, Ramanna and Watts 2012, Abdul Majid 2013, this study employs CEO tenure (CEOTENURE) as a proxy for whether the CEO was responsible for the existence of good will. In this study, following Beatty and Weber (2006: 271) and Abdul Majid (2013: 321), CEO tenure (CEOTENURE) is measured as the number of years that the CEO has held the position. This study tests the relation between CEO tenure (CEOTENURE) and the decision to recognize zero goodwill impairment in an alternative form as follows: H 1b : Ceteris paribus, there is a significant positive asso ciation between CEO tenure (CEOTENURE) and the decision to recognize zero goodwill impairment.

Ownership concentration
Prior studies argue that the ownership concentration influences the accounting method choice (e.g., Dhaliwal et al. 1982, Niehaus 1989. When shareholders hold a small proportion of the ownership interest, due to high monitoring costs, they may have fewer incentives to monitor the managers. Thus, greater discretion may be exercised by managers in firms with dispersed owners hip (Niehaus 1989). However, the exercise of managerial discretion may decrease as the ownership interests of the shareholders increase. This is because when shareholders hold a large proportion of the ownership interest, the be nefit of monitoring the management increases and tends to offset the costs (Niehaus 1989, Astami and Tower 2006, Abdul Majid 2013. In this study, ownership concentration (OWNCON) is measured as the ratio of the number of ordinary shares held by the five largest shareholders to the total number of ordi nary shares. This study tests the relation between ownership concentration (OWNCON) and the decision to recognize zero goodwill impairment in an alternative form as follows: H 2 : Ceteris paribus, there is a significant negative asso ciation between ownership concentration (OWNCON) and the decision to recognize zero goodwill impairment.

Constraints to the recognition choice: corporate governance mechanism
Prior studies suggest that effective governance mechanisms act as a constraint to managerial opportunism related to reporting goodwill impairment (LapointeAntunes et al. 2008, Abdul Majid 2015. Corporate governance mecha nisms are normally regarded to be effective when there are a high proportion of audit committee members who are in dependent (AUDITCOM) (LapointeAntunes et al. 2008).
In this study, similar to Abdul Majid (2015), audit committee independence (AUDITCOM) is measured as the proportion of independent nonexecutive directors on the audit committee. This study tests the relation between the audit committee independence (AUDITCOM) and the decision to recognize zero goodwill impairment in an al ternative form as follows: H 3 : Ceteris paribus, there is a significant negative as sociation between the proportion of audit committee mem bers who are independent (AUDITCOM) and the decision to recognize zero goodwill impairment.

Economic factors of impairment and control variables
To control for the economic factors of impairment, consis tent with prior studies (e.g., Riedl 2004, AbuGhazaleh et al. 2011, Abdul Majid 2015, this study incorporates the chan ge in operating cash flows (∆OCF). Similar to AbuGhazaleh et al. (2011) andAbdul Majid (2015: 212), the change in operating cash flows (∆OCF) is measured as the change in operating cash flows from the prior year to the current year, divided by total assets at the end of the prior year.
In addition, following Abdul Majid (2015: 212), this study controls for the effect of companyspecific factors by incorporating the size of firms (SIZE), the relative size of the goodwill balance (GWB), and the booktomarket ratio (BTM) into the regression model. Detailed definitions for these variables are presented in Section 2.1.

Research design
This study takes a positivist research approach in analy zing the factors that influence and constrain the decision to recognize zero goodwill impairment. The approach is considered appropriate as this study is concerned with testing theory.
In this study, the dependent variable is a binary variable, equal to one for firms that are regarded as choosing to rec ognize zero goodwill impairment, and zero otherwise. Due to the binary nature of the dependent variable, this study has applied a binary logistic regression, a multiple regres sion. The logistic regression includes a dependent variable, which is a categorical dichotomy and independent variables, which are categorical or continuous (Field 2005: 218, Abdul Majid 2013.

Logistic regression model
To examine factors influencing and constraining the decisi on to recognize zero goodwill impairment by Singaporean listed firms, this study employs the following logistic re gression model: where: LEVERAGE = Debt ratio, measured as total debts at the end of the prior year divided by total assets at the end of the prior year; CEOTENURE = CEO tenure, which is based on the number of years that the CEO has held the position; OWNCON = Ownership concentration refers to the number of ordinary shares held by the five largest shareholders, divided by the total number of issued and paid up ordinary shares; AUDITCOM = The proportion of independent nonexecutive directors on the audit commit tee; GWB = Relative size of the goodwill balance, which is measured as the opening goodwill balance in the current year divided by total assets at the end of the prior year; ∆OCF = Change in operating cash flows from the prior year to the current year, divided by total assets at the end of the prior year; SIZE = Natural logarithm of total assets at the end of the prior year; BTM = Booktomarket ratio, computed as the book value of equity divided by the market value of equity at the end of the current year. These variable definitions are adopted from Abdul Majid (2013: 319-321) and Abdul Majid (2015: 212). Following Abdul Majid (2013), the dependent vari able, GWIL (0,1) is the choice to recognize zero goodwill impairment for firms that have booktomarket ratios above one for three consecutive years (i.e., 2010-2012). Similar to Abdul Majid (2013: 259-260), these firms are referred to as a test group; they are tested against a con trol group of firms. The control group is a group of firms that experienced a similar market condition, in that their booktomarket ratio is above one for three consecutive years; however, they reported goodwill impairment losses at the end of the third year. To perform the analysis, the test group is coded as one, and the control group is coded as zero. Accordingly, the dependent variable is a dichoto mous variable, which is equal to one for the test group, and zero for the control group.

Sample selection
To obtain the total population of listed firms in Singapore that implemented IFRS 3 from 2010-2012 and that had a booktomarket ratio above one for three consecutive years, two criteria are imposed. First, the study selected all firms listed on Singapore stock exchanges that have a goodwill balance at the yearend from 2010-2012. A total of 800 firms are listed on the Singapore stock exchange in 2010 of which 241 firms have a goodwill balance. Second, the study selected firms that experienced booktomarket ratio above one for three consecutive years from 2010-2012. Firms that had a booktomarket ratio above one had market values that dropped below the book values of the net assets. According to IAS 36 Impairment of Assets, this incidence is one of the external indications that goodwill may be impai red. Of the 241 firms, 86 firms reported a booktomarket ratio above one for three consecutive years. Nevertheless, 34 firms are excluded due to incomplete data to run the regression analysis. Overall, based on these two selection criteria, there are 52 firms with complete data.
Data on goodwill impairment are handcollected and compared with the annual reports before transforming them into a binary variable. Similarly, data on CEO ten ure, ownership concentration and corporate governance mechanism are manually collected from annual reports. Financial data, such as ∆OCF, BTM and SIZE, are generated from Datastream.  Table 1 shows that on average, the five largest share holders (OWNCON) owned 24.881% of the Singaporean firms, which suggests that the ownership structure of these firms is concentrated. In addition, the length of CEO tenure (CEOTENURE) is high, at 13 years on average. Also, on average, goodwill (GWB) represent 5.6% of the total assets, suggesting that the amount of goodwill is small relative to the total assets. With regard to firms' governance structures, the table shows that 93.6% of the audit committee mem bers are independent (AUDITCOM). The high proportion of independent directors on the audit committee for these firms indicates a strong governance structure.

Descriptive statistics
To detect the issue of multicollinearity, the test for Pearson correlation coefficients is performed between the variables employed in the study. Table 2 shows that the highest pairwise correlation coefficient is 0.41. This cor relation is not considered a major concern because it is not excessively strong, i.e., not more than 50% (Vaus 2002). Thus, the results of the correlation coefficients suggest that multicollinearity is not an issue in this study. Table 3 presents the logistic regression results for the re gression model, which analyses the potential drivers and constraints to the decision to recognize zero goodwill im pairment by listed firms in Singapore. All of these listed firms had a booktomarket ratio above one for three con secutive years.

Multivariate analysis
The results, which are based on the full model (Model 1), indicate a Nagelkerke R square of 35.5% with a Chisquare of 12.515 (at pvalue < 10%). The marginally significant result of the Chisquares tests indicates the lack of fit for the full model. According to Menard (1995), the lack of fit of a regression model could be due to two reasons, i.e., a small number of observations and a large number of variables in the model.
The issue of the lack of fit of the model is addressed us ing two approaches. First, this study runs diagnostic tests by eliminating control variables that are least significant. In Model 2, this study removes the least significant control variable, i.e., ∆OCF; in Model 3, the three least significant control variables (∆OCF, SIZE, and BTM) are removed. After the diagnostic tests, the results of Model 2 show a Nagelkerke R square of 35.4% with a Chisquare of 12.475 (at p < 5%). The statistically significant results of the Chi squares test suggest that Model 2 fits significantly better than an empty model. The second approach in overcoming the issue of the lack of fit of the model is by performing a step wise regression (see Model 4). The results reveal that the re gression model improved further, it achieved a Nagelkerke R square of 21.5% and a Chisquare of 7.195 (at p < 1%).

Discussion of the findings
In H1a, this study tests the debt hypothesis. The result shows that in all of the models, the coefficient on the LEVERAGE is positive and statistically significant. The positive di rection for the LEVERAGE result provides support to the debt hypothesis, which is formulated in H1a. Consistent with Beatty and Weber (2006), this result suggests that the higher the LEVERAGE, the higher the likelihood of firms to exercise the choice in recognizing zero goodwill impair ment although they had a booktomarket ratio above one for three consecutive years.
In H1b, this study tests the association between the CEO tenure and the decision to recognize zero goodwill impair ment. The results in Model 1 to Model 3, show that the coefficient on CEOTENURE is nonsignificant. Unlike the findings of prior studies (e.g., Beatty and Weber 2006), this study could not provide sufficient evidence to support H1b.
In H2, the study tests the influence of ownership con centration (OWNCON, measured as a percentage of shares held by the five largest shareholders) on the decision to rec ognize zero goodwill impairment. Unlike the findings of prior studies (e.g., Astami and Tower 2006), this study finds that the coefficient on OWNCON is nonsignificant in all of the models.
In H3, this study tests factors constraining firms' de cision to recognize zero goodwill impairment, using the  proportion of independent directors on the audit committee as a test variable (AUDITCOM). The findings show that in all the models, the coefficient on AUDITCOM is negative and statistically significant. Similar to LapointeAntunes et al. (2008), this result supports H3. The result suggests that, as the proportion of independent directors on the au dit committee increases, there is a lower likelihood for the sample firms in Singapore to exercise the choice in recog nizing zero goodwill impairment. This study finds control variables, such as SIZE and BTM, to be nonsignificant.

Summary and conclusion
Overall, this study examines factors influencing and cons training the decision to recognize zero goodwill impair ment using a sample of 52 Singaporean listed firms from 2010-2012. Listed firms in this ASEAN country imple mented IFRS developed by the IASB and they had a book tomarket ratio above one for three consecutive years. The regression results reveal that the decision to recognize zero goodwill impairment by Singaporean listed firms is driven by the desire to avoid debt covenant violation, supporting the debt hypothesis. However, a high proportion of inde pendent directors on the audit committee is more likely to constrain this recognition choice.
In conclusion, the findings of this study contributes to the literature on IFRS by providing evidence to sup port the applicability of the debt hypothesis, which was developed in an advanced market with firms that have dis persed ownership, in explaining the decision to recognize zero goodwill impairment by Singaporean listed firms. The results of this study also contributes to the corporate governance literature by supporting the results of prior studies that found effective governance mechanisms acts as a constraint to managerial opportunism related to re porting goodwill impairment.
This study provides important implications for manag ers, policy makers and relevant authorities. To enhance the quality of the financial statements, especially for Singaporean listed companies, the relevant authorities need to closely monitor firms' reporting incentives and further strengthen corporate governance mechanisms.
This study is subjected to at least one limitation, es pecially regarding thoroughly testing the contracting perspective. Specifically, the lack of data on management compensation plans has hindered this study in testing the bonus plan hypothesis. Future research could conduct a survey questionnaire enquiring whether such manage ment compensation plans are applied by Singaporean listed firms. In addition, future research could extend this study by utilizing comparative study approaches among the ASEAN countries that have fully implemented IFRS.