Goodwill Impairment - Earnings Management during the New FRS 3 Transitions: Evidence from the Main Board of Bursa Malaysia

Transcription

1 Goodwill Impairment - Earnings Management during the New FRS 3 Transitions: Evidence from the Main Board of Bursa Malaysia NURUL HUSNA HARON Faculty of Accountancy, Accounting Research Institute, Universiti Teknologi MARA, Shah Alam 40450, Selangor, Malaysia Tel: nurul346@johor.uitm.edu.my RUHAYA ATAN Faculty of Accountancy, Accounting Research Institute, Universiti Teknologi MARA, Shah Alam 40450, Selangor, Malaysia Tel: ruhaya@salam.uitm.edu.my Abstract: - The purpose of this study is to examine earnings management behavior surrounding the change of the goodwill treatment upon the adoption of FRS 3 Business Combination that takes effect on 1 January 2006 in Malaysia. Under FRS 3, goodwill balance should be tested for impairment annually and more frequently if there are indications that goodwill is impaired. Amortization of goodwill is no longer allowed. The impairment test involves managerial discretion in which it could lead to earnings management. In this paper, discretionary current accrual (DCA) is the proxy for earnings management. The final sample consists of 180 companies listed on the main board of Bursa Malaysia in It is found that the DCA of goodwill companies are higher during the adoption year compared to the pre-adoption year. The goodwill balance after the impairment test due to the changes in the goodwill rule influenced the DCA level during the adoption year. This indicates that the management in the sample companies uses the provision in the FRS 3 to manage the goodwill figure by providing impairment after the new standard adoption. Keywords: Goodwill, Earnings Management, Discretionary Current Accruals (DCA), FRS 3 Business Combination, Malaysia 1.0 Introduction In this study, the earnings management situation would be looked into detail from the perspective of FRS 3 Business Combination. Under FRS 3, goodwill balance should be tested for impairment annually and more frequently if there are indications that goodwill is impaired. Amortization of goodwill is no longer allowed. The FRS is intended to shift the balance sheet reported values from the historical costs towards the fair values 1. The fair value in certain circumstances would involve management discretion. The allowed managerial discretion should be based on the appropriate assumptions and the conclusions reached are both reasonable and supported or otherwise it would lead to serious impacts on the financial statements. However, the inclusion of the managerial discretion would be a great opportunity for the management to perform the earnings management (Healy and Wahlen, 1999). From the definition of the recoverable amount 2 (in determining the goodwill impairment loss), managers have the opportunity to manipulate the value of the cost to sell and the value in use as it requires the management s estimation. Indeed, subjectivity is inherent in the two-step impairment test (as above) prescribed by the standard (Hayn and Hughes, 2005). The manipulation in fact is partially attributable to the Agency Theory (Fama 1980) where management (agent) and owners (principals) are motivated by self-interests leading to the earnings management. Agency theory suggests that managers will use unverifiable discretion opportunistically and the theory also predicts that managers will use the discretion in FRS 3 in circumstances where they have agency-based motives to do so (Watts, 2003 as cited in Jarva, 2008). Mohanram (2003) defines earnings management as the intentional 1 Fair values as defined in the standard is the amount for which an asset could be exchanged, or a liability settled, between knowledgeable, willing parties in an arm s length transaction. 2 Para 17 of FRS 136 defines recoverable amount as the higher of an asset's net selling price and value in use. Value in use is the present value of the future cash flows expected to be derived from an asset or cash generating asset. ISSN: ISBN:

2 misstatement of earnings leading to bottom line numbers that would have been different in the absence of any manipulation. The earnings management occurs when managers exercise their professional judgment in financial reporting to manipulate the earnings. The new goodwill guideline then might be used as an earnings management tool that could mislead the users. This is supported by Swanson (2007) who found that managers are exploiting their discretion in recognizing goodwill impairments to manage earnings. Hayn and Hughes (2005) added that the non-amortization provision of the new rule makes the potential for abuse even greater, despite the presence of the more stringent periodic review requirement. Sevin and Schroeder (2005) agreed on the same basis in which in their recent research, they found and suggested that the new goodwill accounting rule allowed companies to engage in earnings management. Further, they found that small companies experienced a significantly greater negative impact and were much more likely than large companies to take big bath 3 charges. Haman and Jubb (2008) who extends the research of Sevin and Schroeder (2005) again found that the managers do manage earnings using long-term accruals, particularly at the time of mandatory adoption of a new goodwill rule. Further, Ahmed and Guler (2007) found that the goodwill write-offs (impairment/amortization) and goodwill balance are more strongly associated with stock returns and stock prices respectively after the new goodwill rule adoption. One way of measuring the acquired company s performance is through the goodwill balance. The performance of the acquired entity is important to be assessed as to ensure how much the entity could contribute to the acquiring entity as a whole. However, if the goodwill balance has been manipulated through the earnings management, the available information does not provide adequate information for the users to predict future write offs of goodwill (Hayn and Hughes, 2005). It seems like the standard on this goodwill has not fully protected the users. Both the goodwill value and the level of earnings reported in the financial statements may be distorted if the rule is being used as a tool for Earnings Management (Swanson, 2007). The discretion requirements contain in the Standard might lead to inconsistencies among various companies RECENT ADVANCES in BUSINESS ADMINISTRATION upon the adoption. Some inconsistencies may remain due to the amount of subjective estimates and judgement involved in applying the Standards, but they should be occurring in a greatly reduced rate when compared to the situation before its adoption (Baker et al., 2000). The IFRS is introduced partly to harmonize the accounting standards across the globe. However, the pervasiveness of earnings management did not decline after the introduction of IFRS, and in fact it increased in France (Jeanjean and Stolowy, 2008). The earnings management arises mainly due to the allowed management discretion in the standard. In Malaysia, Financial Reporting Standard (FRS) 3 Business Combination (which is similar to IFRS 3 that replaces IAS 22) should be applied effectively from 1 January Therefore, Malaysian public companies would treat goodwill according to the new requirement of the standard. In Malaysia, no prior studies have been undertaken to examine the earnings management behavior surrounding the change of treatment of goodwill upon the adoption of the above standard. As such this paper aims to fulfill the following specific objectives: i- To investigate whether the DCA of goodwill companies are significantly higher during the adoption year as compared to preadoption year. ii- To determine the influence of goodwill balance after the impairment test on the DCA during the adoption year that could lead to the earnings management activities. The findings of this study is important because it adds to the literature and inform the standard setters on which standards and accruals are being used to manage earnings (Healy and Wahlen, 1999). As there are increasing presences of fair value accounting in FRS specifically in goodwill, it is important for researchers, standard setters and investors to understand how fair value accounting is implemented by managers as well as how it is interpreted by users of financial statements (Bens et al. 2007). In addition, Bernard and Skinner (1996) stated that the standard setters tend to act to reduce managers' ability to exercise discretion in the reporting process, apparently based on the assumption that managers exercise their accounting discretion opportunistically. 3 Big bath: When companies are way below their targets, they have an incentive to make things look even worse as it is highly unlikely that any amount of earnings management will get them over the target and the costs of being even worse are typically minimal (Mohanram 2003) 2.0 Research Methodology This study tests goodwill impairment write downs in a cross-sectional distributional analysis for the year 2006, ISSN: ISBN:

3 the first year following the effective date of the FRS 3 in Malaysia. The sample is taken from the financial information of publicly traded companies tracked in the Data Stream databases. The sample is focused on seven selected industries from the Main Board of the Bursa Malaysia. This study employed the DCA as the proxy for the earnings management. The computation of the DCA is based on the Haman and Jubb (2008) formulae. The selection begins by including all listed companies providing a first FRS annual report in Y-0. The FRS 3 stated that the effective date of adopting the standard is on or after 1 January Next, the selection excludes all companies from the financial services, utility and real estate industries. These industries are excluded because these companies have unique disclosures and/or are heavily regulated and/or have a different type of accrual accounting. This is consistent with Teoh et al. (1998) and Vafeas et al. (2003). The selection then excludes companies that do not meet the following criteria: (a) the sample companies should have been listed on the Bursa Malaysia for two years prior to the adoption year of the new goodwill rule (2004 for December year end (Dec y/e) companies and 2005 for non-dec y/e); (b) the data necessary to determine the level of discretionary accruals are available and RECENT ADVANCES in BUSINESS ADMINISTRATION (c) the sample companies should have a goodwill balance in the year prior to the adoption of the new goodwill rule (Y-1) (2005 for Dec y/e companies and 2006 for non-dec y/e) After excluding all these companies, the sample consists of 180 companies that report a goodwill balance in Y-1. From this number, 100 companies are having the Dec y/e and 80 companies are having the non-dec y/e. The pre adoption year is then designated accordingly. In this study, financial year 0 represents the adoption year of the new goodwill rule (FRS 3). This study observes and analyses data for each sampled companies for two financial years i.e. around the year of adoption (Y-0) and the pre-adoption year (Y-1). These signs of Y- 0 and Y-1 are essential to differentiate the data used. The Y-0 is taken as the first annual financial reporting period commencing on or after 1st January 2006 (the effective date of adopting the FRS 3). For 31 st December year end companies, this means the annual reports are as at 31st December 2006 (Dec y/e). For companies with later balance dates such as the end month of January to November, this means the date of the first annual financial reporting period ending in 2007 (non-dec y/e). SGW DCA SIZE MB LEV ROA TCA OCF BIG4 Diagram 1 Conceptual Framework 2.1 Research Framework The variables selected in this study consist of one dependent variable, one independent/ explanatory variable and eight control variables. The relationship among the variables is depicted in the Conceptual Framework in Diagram 1. The asset-scaled goodwill (SGW (Goodwill balance divided by the total assets)) is hypothesized to influence the management s discretionary current accruals (DCA) that act as the proxy of earnings management. To make sure that there is no distortion from other variables and to increase the reliability of the results, seven control variables are included in the multiple regression ISSN: ISBN:

4 [Chung and Kallapur (2003) as cited in Haman and Jubb, 2008]. They are natural logarithm of total assets (SIZE), growth prospects (MB), leverage (LEV), return on assets (ROA), total current accruals (TCA), operating cash flows (OCF) and a proxy for the quality of the external auditor (BIG4). The DCA would be looked at particularly on the impairment of assets (goodwill). Thus, DCA would act as the dependent variable. DCA would involve the managers accounting discretion in adopting the FRS 3 specifically on the goodwill. Thus, DCA acts as the proxy for the earnings management. The earnings management is measured from one year prior (Y-1) to the year of the new goodwill rule adoption (Y-0) (Haman and Jubb, 2008). For this purpose, a variation of the cross-sectional version of the modified Jones (1991) model is employed (Dechow et al., 1995). Dechow et al. (1995) found that the model performed best with cross-sectional data rather than longitudinal data as used by Jones (1991). Sevin and Schroeder (2005) and Haman and Jubb (2008) follow the same model with some modifications by adding some other variables. However, to eliminate the potential balance sheet impact of the adoption of FRS on or after 1 January 2006, the variables in the modified Jones (1991) model are deflated by assets at the end of the year instead of at the beginning of the year (Haman and Jubb, 2008). RECENT ADVANCES in BUSINESS ADMINISTRATION Based on Haman and Jubb (2008), the discretionary current accruals (DCA) are calculated by subtracting the nondiscretionary current accruals (NDCA) from the total current accruals (TCA). The TCA, or non-cash working capital accruals, are calculated as changes in the noncash current assets, excluding the changes in short-term investments (STI); minus the changes in current liabilities (CL) excluding changes in the short term debt (STD), which is derived from the current maturity of long term debt. As the DCA are not observable, total current accruals are first computed and the estimated non-discretionary component is then extracted (Soo, 1999). For the first equation, the TCA would be computed for both goodwill companies and non-goodwill companies. The first equation is depicted below:- TCA t = ( CA t Cash t STI t ) ( CL t STD t ) (1) where: TCA t = Total Current Accruals in year t Cash t = Cash in year t less Cash in year t-1 CA t = Current Assets in year t less Current Assets in year t-1 STI t = Short-term Investment in year t less Short-term Investment in year t-1 CL t = Current Liabilities in year t less Current Liabilities in year t-1 STD t = Short-term Debt in year t less Short-term Debt in year t Development of Hypotheses The standards provide a flexible choice for managers to self-assess their impairment value through the recoverable amount of the goodwill as it involves management discretionary. Thus, the new goodwill rule adoption might lead to changes in the provision of goodwill impairment that affect the overall DCA level. This DCA level would be examined in both years ie Y-0 and Y-1. The DCA level is expected to have significant difference if being compared for both years. Therefore this study predicts that the level of DCA of goodwill companies will be higher during the adoption year as compared to the pre-adoption year. Hence, Hypothesis 1 is developed as follows H1: The discretionary current accruals of goodwill companies are significantly higher during the adoption year as compared to the pre-adoption year. Next, after obtaining the TCA figure for every company in this study, Equation 2 would be computed for the control companies only. Companies that do not report goodwill balance at the end of years -2, -1 and 0 are classified as the control companies. In other words, nongoodwill companies are the control companies. There are 377 companies selected for this purpose. Employing the regression model in Equation 2 (as per the formulae below), the data would be regressed separately for the 7 industries. It would be performed industry by industry for each financial year, to capture its DCA s behavior (Haman and Jubb, 2008). The regression results would identify the coefficients (α 0 and α 1 ) of the cross-sectional variables for the control companies as they are captured in the same industry sector. DCA that acts as the dependent variable would be computed based on four equations (Equation 1 to Equation 4) by adopting the modified Jones model. Then, the DCA would complete the final regression model in the Equation 5. j = Control companies A j,t = Total Assets in year t REV j,t = Revenue in year t Revenue in year t-1 ISSN: ISBN:

5 TR j,t = Trade Receivables in year t Trade Receivables in year t-1 ε j,t = Error term in year t These coefficients would then be used in computing Equation 3 for the sampled companies. The coefficients reflect the non-discretionary current accruals (NDCA) of the sampled companies for the Y-1 and Y-0 respectively. Previous studies such as DeAngelo (1986) and Healy (1985) which also use some type of discretionary accruals measure, discuss the partitioning of total accruals into discretionary and nondiscretionary components (Jones, 1991). To obtain NDCA of the sampled companies, the α 0 and α 1 of control companies that are obtained from Equation 2 are applied to the α 0 and α 1 of the sampled companies (i.e., goodwill companies) in Equation 3 for Y-1 and Y-0 respectively. The NDCA would lead to the computation of the dependent variable of DCA, Where i = Sampled companies A i,t = Total Assets in year t REV i,t = Revenue in year t Revenue in year t-1 TR i,t = Trade Receivables in year t Trade Receivables in year t-1 DCA are then calculated for sampled companies of Y-0 and Y-1 by subtracting NDCA calculated as per Equation 4 from the total current accruals (TCA) scaled by assets respectively. RECENT ADVANCES in BUSINESS ADMINISTRATION management uses the accounting rules and manipulates earnings either by not recognizing impairment when it has occurred or by recognizing it only when it is advantageous to do so. Thus, it could lead to higher earnings or lower earnings. As discussed earlier, all companies that have goodwill balance during the adoption year are required by the FRS 3 to test the goodwill balance for impairment. The impairment test would then lead the sample companies to have impaired goodwill or non-impaired goodwill. Thus, the goodwill balance that might contain provision for impairment is expected to influence the overall DCA level. Once the DCA level is affected, it would trigger the earnings management activities as the DCA acts as the proxy for the earnings management. Using the big bath theory, Sevin and Schroeder (2005) found that US managers managed earnings downwardly after adoption of the new goodwill rule by taking big bath charges. In addition, Haman and Jubb (2008) found that discretionary accruals are negatively associated with the level of goodwill in the adoption years of the new goodwill rule. The next hypothesis is developed as follow:- H2: In the adoption year of the new goodwill rule, the goodwill balance after the impairment test will influence the discretionary current accrual. Thus, the following regression model would be used to test the hypothesis above. DCA it = β 0 + β 1 SGW it + β 2 SIZE it + β 3 MB it + β4lev it + β 5 ROA it + β 6 TCA it + β 7 OCF it + β 8 BIG4 it + ε it (5) This study then hypothesizes that the goodwill companies will use the goodwill impairment provided in the Standard to manage the company s earnings. In certain circumstances, the companies would comply with the Standard and might provide for the impairment when there is indicator to do so. However, the amount to be impaired is subject to the companies earning target. The flexibility could be seen either to increase the impairment losses and downward the company s earning or to decrease the impairment losses and upward the company s earning This flexibility could be modified by adjusting the figures of Recoverable Amount (higher of value in use and net selling price). Baker et al., (2000) stated that one view of write-offs of asset impairments is that β 1 SGW it is the Assets-scaled goodwill of a company i in year t. If the accruals reflect the financial and economic condition of the companies, the association between accruals and level of goodwill is positive (Haman and Jubb, 2008). Hence, the relationship between the SGW and DCA is expected to be negative. β 2 SIZE it is the natural logarithm of company i s current year assets measured at the end of financial year. Moses (1987) finds evidence that large firms have a bigger incentive to smooth earnings than small firm. In contrast, Siregar and Utama (2008) found evidence that large firms have less incentive to smooth earnings than small firms. Thus, the association between SIZE and DCA could have either positive or negative. β 3 MB it is the ratio of current year company i s market to book value of assets at the end of the financial year. High growth companies tend to manage earnings ISSN: ISBN:

6 upwardly to maintain share price (Krishnan, 2003). However, there is possibility that high growth companies do not manage earnings if their share price has already been high. Thus, the association between MB and DCA could have either positive or negative. β4 LEV it is the current year long-term debt of company i at the end of financial year scaled by assets at the end of the year. Becker et al. (1998) suggest that leverage can be a proxy for potential income-decreasing accruals management in companies suffering from financial distress. In contrast, the debt equity hypothesis (Watts & Zimmerman, 1986) suggests that high leverage works as an incentive for income-increasing earnings management. Thus, it could lead to both positive or negative sign. β 5 TCA it is the total current accruals of company i in the current year. Since DCA is derived from TCA minus NDCA, it is expected that TCA is associated positively with DCA. β 6 OCF it is the company i s current year operating cash flows. As the positive TA is derived from NPAT minus OCF and TCA is derived from NPAT plus depreciation minus OCF, it is expected that this variable to be negatively associated with DCA. β 7 ROA it is the company i s current year Return on Assets. ROA is derived from current year net income after tax scaled by assets at the end of the year. It is predicted that managers accounting discretion to manage earnings are driven by earnings growth leading to a positive sign. β 8 BIG4 it is the dummy variable set to 1 for company i that employs an external auditor from amongst the BIG4. It is expected that external auditors from the BIG4 may minimise earnings management. Therefore, BIG4 is expected to be negatively associated with income increasing DCA (positive sign), and positively associated with income decreasing DCA (negative sign). 2.3 Analysis of Data Paired Sample T-Test Analysis T-test analysis is used to examine the difference of the DCA level between adoption year data and pre-adoption year data. This study employs the multivariate test of linear regression by using the SPSS statistical package. The data would be regressed according to the hypotheses developed. RECENT ADVANCES in BUSINESS ADMINISTRATION 2.4 Findings And Discussion Descriptive Statistics The sample characteristics are summarized in Table 1. The median of CHGW and SCHGW in Y-0 both showed This indicates that most of the sampled companies did stop amortizing the goodwill balance during the adoption year as required by the Standard. The mean of CHGW is higher in Y-0 (11,169) as compared to Y-1 (4,189). This is possibly due to the merger and acquisition (M&A) that took place during the adoption year. Thus, it led to a higher goodwill balance for those companies. However, it could not be generalized that there was no goodwill earnings management as the quantum of the M&A for the companies involved could be high. This is supported by the decreased mean of SCHGW from (Y-1) to 0.005(Y-0) and decreased mean of GW from 56,501 (Y- 1) to 43,448 (Y-0). The mean increases for TTLA, NPAT, OCF, MB and LEV from Y-1 to Y-0. In contrast, the mean for ROA in Y-0 of the sampled companies is lower than Y-1. This suggests that, on average, the increase in NPAT and MB of the sample companies in Y-0 is not supported by the companies ROA Correlation Analysis The correlation matrix for the variables for Y-0 is presented in Table 2. Low correlations of below are reported for all variables as expected for both years except for three relationships. Therefore multicollinearity should not be a concern in the regression analysis. There are 3 correlations that are above 0.500; ((DCA and TCA ), (SIZE and OC ) and (MB and ROA ). To ensure that there is no multicollinearity issue from the exceptions above, Collinearity Statistics is run. The result is summarized in Table 3. It is found that VIF for all the variables are below 10 and the Tolerance are more than Thus, the exceptions above are accepted no multicollinearity issues noted. ISSN: ISBN:

7 Table 1: Sample Characteristics (N= 180) Y-1 Y-0 Data Mean Median Std. Deviation Mean Median Std. Deviation DCA GW 56,501 5, ,006 43,488 5, ,234 CHGW 4,189 (337) 115,866 11, ,069 TTLA 1,107, ,081 1,706,582 1,314, ,413 2,130,742 NPAT 17,779 10, ,206 53,828 17, ,399 OCF 48,437 17,717 98,446 61,267 16, ,402 MB SGW SCHGW (0.001) REV ROA LEV Goodwill (GW) in RM 000, Change in Goodwill (CHGW) in RM 000, Total Assets (TTLA) in RM 000, Net Income After Tax (NPAT) in RM 000, Operating Cash Flows (OCF) in RM 000, Market to Book (MB) in ratio, Assets-Scaled Goodwill (SGW) in percentage, Assets-Scaled Change in Goodwill (SCHGW) in percentage, Assets-Scaled Change in Net Revenue (REV) in percentage, Return on Assets (ROA) in percentage, Leverage (LEV) in percentage Table 2: The Correlation Matrix of Variables during the Adoption Year (Y-0) Pearson Correlation RECENT ADVANCES in BUSINESS ADMINISTRATION Variables DCA it SGW it SIZE it MB it LEV it ROA it TCA it OCF it BIG4 it DCA it (0.000)*** (0.037)** (0.005)*** (0.038)** (0.079)* (0.434) (0.000)*** (0.330) (0.474) SGW it (0.037)** (0.000)*** (0.014)** (0.414) (0.241) (0.156) (0.488) (0.399) (0.407) SIZE it (0.005)*** (0.014)** (0.000)*** (0.046)** (0.000)*** (0.138) (0.126) (0.000)*** (0.080)* MB it (0.038)** (0.414) (0.046)** (0.000)*** (0.293) (0.000)*** (0.045)** (0.000)*** (0.070)* LEV it (0.079)* (0.241) (0.000)*** (0.293) (0.000)*** (0.016)** (0.015)** (0.400) (0.181) ROA it (0.434) (0.156) (0.138) (0.000)*** (0.016)** (0.000)*** (0.213) (0.000)*** (0.165) TCA it (0.000)*** (0.488) (0.126) (0.045)** (0.015)** (0.213) (0.000)*** (0.166) (0.234) OCF it (0.330) (0.399) (0.000)*** (0.000)*** (0.400) (0.000)*** (0.166) (0.000)*** (0.147) BIG4 it (0.474) (0.407) (0.080)* (0.070)* (0.181) (0.165) (0.234) (0.147) (0.000)*** DCA it = Discretionary current accruals of company i in yeart, SGW it = Assets-scaled goodwill of company i in year t, SIZE it= the natural logarithm of company i s current year assets measured at the end of financial year, MB it= the ratio of current year company i s market to book value of assets at the end of financial year, LEV it= current year long-term debt of company i at the end of financial year scaled by assets at the end of the year, ROA it_1= company i s current year Return on Assets, TCA it = total current accruals of company i in the current year, OCFit = company i s current year operating cash flows, BIG4 it = dummy variable set to 1 for company i that employs an external auditor from amongst the BIG4. *** Significant at 1% level; ** Significant at 5% level; * Significant at 10% level ISSN: ISBN:

8 Table 3: Collinearity Statistics Collinearity Statistics Variables Tolerance VIF SGW it SIZE it MB it LEV it ROA it TCA it OCF it BIG4 it Multivariate Analysis The sample in this study consists of 180 goodwill companies. In order to make comparison between the two groups, the Y-0 DCA companies were matched against the Y-1 DCA companies. The process of distinguishing between those two groups is vital because of the first objective. The first objective of this study is to investigate whether the discretionary current accrual of goodwill companies are significantly higher during the adoption year as compared to pre-adoption year. Table 4 showed that there is significant difference between DCA of goodwill companies during the adoption year and pre-adoption year. It was indicated by the t-test result of and the p-value of In addition, the DCA mean of Y-0 is higher (-0.233) as compared to the DCA mean of Y-1 ( ). The DCA mean is higher in the adoption year as compared to the pre-adoption year. This situation would lead to higher total expenses and lower earnings to the companies. However, at this level it could not be generalized that the DCA level is higher because of the goodwill impairment during the adoption year. It will be tested further in the H2. From the t-test result, the Hypothesis 1 is thus supported. Equation 2 is regressed by using the control companies data. The coefficients obtained from the regression result would be applied in Equation 3 for the sample companies. The results obtained are shown in Table 5. The Coefficient figure is extracted from the Unstandardized Coefficients (Beta) rather than the Standardized Coefficients (Beta) as it reflects the actual data value. Table 4: T-Test Result between Y-0 and Y-1 DCA Changes t-statistics Y-1 Y * significant at less than 5% level P-value Sig. (2-tailed) * ISSN: ISBN:

9 Table 5: Summary of Coefficients data regressed from the Equation 2 Industries Y-1 Y-0 Unstandardized Adjusted Unstandardized Coefficients (Beta) R² Coefficients (Beta) Adjusted R² N 1/A jt ( REV jt - TR jt )/A jt 1/A jt ( REV jt - TR jt )/A jt Industial Construction Technology Cons. Product Plantation Trading Property A j,t = Total Assets for control companies in year t REV j,t = Revenue in year t Revenue in year t-1 TR j,t = Trade Receivables in year t Trade Receivables in year t-1 N = Samples The Equation 2 is regressed separately according to the industry. The Equation employed the data regarding the change in revenue, change in trade receivables and TCA. These data are deflated by the total assets respectively. The data are not simply comparable between different industries due to the difference in the nature of business. Thus, the regression is done separately and lead to different coefficients that could reflect the nature of individual industry. The samples are not enough (i.e. N < 30) for regression purposes for the Construction (27) and Technology (11) industries. It is the limitation of this study due to the nature of the sample selection. Also noted is that the adjusted R² showed negative signs for the Technology (Y-1: and Y-0: ) and Consumer Product (Y-0: ) industries. It is possibly due to the sample limitation too. However, as the adjusted R² is not being used for any equations, it would be ignored. This study would regress 2 sets of data using the same final model of Equation 5. First set using the Y-0 data, and second set using the Y-1 data. Then the results for Y-0 and Y-1 would be compared in order to determine the existence of the earnings management (Haman and Jubb, 2008). DCA acts as the dependent variable (proxy to earnings management) that is expected to be influenced by SGW (independent variable) and the results are controlled by seven control variables namely SIZE, MB, LEV, ROA, TCA, OCF and BIG4. This study employs a multivariate test of linear regression by using the SPSS statistical package. In H2, the goodwill is expected to influence the DCA level of the goodwill companies during the adoption year. ISSN: ISBN:

10 Table 6: Multivariate Results Dependent Predicted Direction Variables for Y-0 Y-1 Y-0 (Constant) (-0.058) (0.008) SGW it (0.124) (0.044**) SIZE it +/ (0.095*) (0.014**) MB it +/ (0.000***) (0.250) LEV it +/ (0.237) (0.441) ROA it (0.040**) (0.221) TCA it (0.000***) (0.000***) OCF it (0.155) (0.363) BIG4 it (0.000***) (0.4403) R² Adj R² F change p -value N Result Supported DCA it = Discretionary current accruals of company i in year t, SGW it = Assets-scaled goodwill of company i in year t, SIZE it = the natural logarithm of company i s current year assets measured at the end of financial year, MB it = the ratio of current year company i s market to book value of assets at the end of financial year, LEV it = current year long-term debt of company i at the end of financial year scaled by assets at the end of the year, ROA it = company i s current year Return on Assets, TCA it = total current accruals of company i in the current year, OCF it = company i s current year operating cash flows, BIG4 it = dummy variable set to 1 for company i that employs an external auditor from amongst the BIG4. *** Significant at 1% level; ** Significant at 5% level; * Significant at 10% level ISSN: ISBN:

11 By referring to Table 5, during the Y-0, The p-value of SGW is significant at less than 5% level. In addition, 2 out of 7 control variables significantly affect the DCA namely the SIZE and TCA at less than 5% and 1% level respectively. The variables explain 48.9% for the whole model based on the adjusted R². The model is significant with the p-values of and F-value of All of these variables showed a predicted relationship. During the pre-adoption year, all goodwill companies would amortize the goodwill balance for a predetermined period evenly (maximum of 20 years). Managerial discretion would only involve during the acquisition year in determining the amortization period. Subsequent years, the company would amortize the goodwill as per period being pre-determined during the acquisition year. The goodwill balance would keep on decreasing every year due to the amortization. However, it does not affect the DCA level as it does not require managers discretionary as compared to the impairment requirement in adoption year. Thus, the DCA during pre-adoption year on the goodwill perspective are properly accounted for. Thus, there is no way for a manager to manipulate the goodwill balance. Hence the relationship between the DCA and SGW (Y-0) is expected to be positive sign. In Y-1, the coefficient of SGW showed positive sign and its p-value is Hence, no earnings management on the goodwill DCA is detected during the pre-adoption year. In contrast, the p-value of SGW in Y-0 is and it is significant. As predicted, the coefficient of SGW showed a negative relationship with DCA. The lower the SGW would indicate more impairment has been recognized leading to higher DCA. As the DCA is the proxy for the earnings management, indirectly the goodwill balance during the adoption year has influenced and affected the earnings management activity. This shows that the management in the sampled companies uses the provision in the FRS 3 to manipulate the goodwill figure by providing impairment after the new standard adoption. The impairment could only be done if there is any indication internally or externally to show that the goodwill should be impaired. However, the standard contains loopholes as it allows for the management discretionary to flexibly manage the company s earnings that lead to this manipulation. The result might not indicate the true situation if the companies recognize lesser impairment amount as compared to what they are supposed to. Instead of recognizing say RM150,000 impairment value, they might only recognize say RM50,000 during the adoption year. The situation is particularly at the initial acquisition of goodwill in order to avoid subsequent charges to earnings through recognition of goodwill impairment (Ahmed and Guler, 2007). However, the compared results (between Y-0 and Y-1) have proved that the goodwill balance that contains the impairment is significantly influence the DCA of goodwill companies in Y-0. The situation leads to a higher DCA and thus, it supports Hypothesis 2. This finding is consistent with the finding by Swanson (2007) in which she found that managers are exploiting their discretion in recognizing goodwill impairments to manage earnings. Sevin and Schroeder (2005) also found and suggested that the new goodwill accounting rule allowed companies to engage in earnings management. In addition to SGW, the control variable of SIZE is found significant at less than 5% level. Size showed a positive relationship with the DCA. Size as mentioned earlier is the natural logarithm of current year s assets measured at the end of the financial year. The size indicated that the companies have high amount of total assets which are being used to generate the companies revenue. As it goes around, the revenue generated would be converted to total assets. The bigger the company size, the bigger is the DCA indicating that the companies have higher amount of goodwill impairment provided during the adoption year. These big companies tend to provide higher impairment possibly due to the cookies jar earnings management. Again, cookies jar means to reduce the earnings when companies are way above their targets so that expectations for the future are adjusted accordingly otherwise the future targets more difficult to attain. Reducing current period income, companies implicitly save some of these excess earnings for the future when they may be more valuable (Mohanram, 2003). This is ISSN: ISBN:

12 supported by the correlation results at Table 2 that showed positive relationship between SIZE & ROA and SIZE & MB. TCA as predicted showed a positive relationship with DCA and was significant at 1% level with the p-value of It is supported by the high correlation between the two variables in the correlation matrix. The management used the TCA as it involves the management discretion to perform the earnings management. From there, the companies objectives could be achieved by managing the companies earning. Schoeler (2005) provided opinion that it is usually seen as a very strong management incentive to avoid reporting earnings decreases or losses by controlling accounting accruals with discretion. Thus, it supported the positive relationship between the TCA and DCA. Schoeler (2005) found that average company manager control the accounting accruals with discretion to manage reported earnings changes where he can. It is managed through the adoption of the new goodwill rule by manipulating the loopholes in the Standard that require the management discretion. Thus, the findings in this study confirm the findings of Dechow et al. (1995), Burgstahler and Dichev (1997) and Hayn and Hughes (2005) that the reported earnings are managed. The other control variables are not significant to the DCA namely MB, LEV, ROA, OCF and BIG4. As summary, it is found that the DCA of goodwill companies is significantly higher during the adoption year as compared to the pre-adoption year. Moreover, the goodwill after the impairment test is found influence the DCA level. This indicates that the management in the sampled companies used the provision in the FRS 3 to manipulate the goodwill figure by providing impairment after the new standard adoption. The impairment could only be done if there is any indication whether internally or externally to show that the goodwill should be impaired/amortized. However, the standard contains loopholes that lead to this manipulation as it uses management discretionary to flexibly manage the company s earnings. 4.0 Conclusion The purpose of this study is to examine earnings management behavior surrounding the change of the goodwill treatment upon the adoption of FRS 3 that takes effect on 1 January 2006 in Malaysia. Under FRS 3, companies are required to stop amortizing the goodwill but to assess their goodwill balance for annual impairment test. The test involves the managerial discretion in which it could lead to earnings management which is proxied by Discretionary Current Accruals (DCA). It is found that the DCA of goodwill companies are higher during the adoption year compared to the pre-adoption year. The goodwill balance after the impairment test due to the changes in the goodwill rule influenced the DCA level during the adoption year. This indicates that the management in the sample companies uses the provision in the FRS 3 to manage the goodwill figure by providing impairment after the new standard adoption. In carrying out this study, a few limitations had been found. Firstly, the selection of goodwill companies is based on the companies that have goodwill balance in Y-1. In Y-1, the goodwill consisted of positive and negative goodwill. The disclosure in the financial statement requires the companies to separately disclose the positive and negative goodwill. However, for a number of companies, the amounts were not separated indicated by the notes below the amounts. Whereas in the adoption year (Y-0), any negative goodwill is recognised in the income statement immediately. Thus, it provides inconsistencies for comparison. Secondly, the SGW (assets-scaled goodwill balance) only concern the year end balance without specifying the goodwill movement. If a company has goodwill impairment less than the goodwill acquired during the year, the goodwill balance would show an increase only (as if there is no goodwill impairment at all). This situation could affect the whole regression result. Finally, the computation of the DCA is based on the balance sheet approach. (Refer to Equation 1 to Equation 4). Collins and Hribar (1999) argue that using the balance sheet approach to compute total current accruals is inferior in certain circumstances to a cashflows-statement-based approach. Specifically, they suggest that the balance sheet approach can lead to ISSN: ISBN:

13 serious estimation errors, particularly when the company has been involved in mergers and/or acquisitions. For future research, the computation of the DCA would be better reflected if this study includes the long term discretionary accrual and finally make up the total discretionary accruals. The total accruals should capture a larger portion of managers' manipulations (Jones, 1991). The future research might include merger and acquisition (M&A) as dummy variable into the final model. The inclusion of this M&A variable would reveal the true goodwill balance that is expected to be impaired. Additional suggestion is to include the long term discretionary accruals to make up the total discretionary accruals to capture larger portion of manager s discretionary behavior. The same study could also be expanded by comparing the earnings management using the various earnings management detection models. From there, the impact could be seen whether different detection models would produce similar outcome. It is suggested that the standard setters should be more aware of this kind of earnings management (the loopholes pertaining to the managerial discretion). If implemented effectively, the Standard should eliminate most of the periodic, large write offs and improve the usefulness of certain popular financial ratios used in the performance evaluation (Baker et al, 2000). References [1] Ahmed, A.S. and Guler, L., Evidence on the Effects of SFAS 142 on the Reliability of Goodwill Write-offs, Social Science Research Network, (2007) retrieved from on 30 August [2] Baker, Rue J.C. and Volkan, A.G., Impairment Write-offs: Truth of Manipulation?, International Public Accountant, Vol.45, No. 5 (2000) pp [3] Bens D.A., Heltzer W. and Segal, B., The Information Content of Goodwill Impairments and the Adoption of SFAS 142, Social Science Research Network, (2007) retrieved from on 30 August [4] Bernard, V.L. and Skinner, D.J., What Motivates Managers Choice of Discretionary Accruals?, Journal of Accounting & Economics, Vol. 22(1996) pp [5] Burgstahler, D. & Dichev, I., Earnings Management to avoid Earnings Decreases and Losses, Journal of Accounting and Economics, Vol. 24(1997) pp [6] Chung, H. and Kallapur, S., Client Importance, Non-Audit Services, and Abnormal Accruals, Accounting Review, Vol. 78 No.4, (2003) pp [7] DeAngelo, L. Accounting Numbers as Market Valuation Substitutes: A Study of Management Buyouts of Public Stockholders, The Accounting Review, Vol. 61 (1986) pp [8] Dechow, P., Sloan, R. & Sweeney, A. Detecting Earnings Management, The Accounting Review, Vol.70, No. 2, (1995) pp [9] Fama, E., Agency Problems and the Teory of the Firm, Journal of Political (1980) [10] Haman, J. and Jubb, C., Earnings Management around a Change of Goodwill Rule, Social Science Research Network, (2008) retrieved at on 30 August [11] Hayn, C. and Hughes, P., Leading Indicators of Goodwill Impairment, Workshop paper. (2005) [12] Healy, P.M. and Wahlen, J.M., A Review of the Earnings Management Literature and Its Implications for Standard Setting, Accounting Horizons, Vol.13, No.4, (1999) pp ISSN: ISBN:

14 [13] Healy, P., The Effect of Bonus Schemes on Accounting Decisions, Journal of Accounting & Economics, Vol.7, (1985) pp [14] Jarva, H., Do Firms Manage Fair Value Estimates? An Examination of SFAS 142 Goodwill Impairment, Social Science Research Network, (2008)retrieved at on 30 August [15] Jeanjean, T. and Stolowy H., Do Accounting Standard Matter? An Exploratory Analysis of Earnings Management Before and After IFRS Adoption, J.Account.Public Policy, Vol. 27, (2008), pp [16] Jones, J., Earnings Management during Import Relief Investigations, Journal of Accounting Research, Vol. 29, (1991) pp [17] Mohanram, P.S., How to Manage Earnings Management?, Accounting World, Institute of Chartered Financial Analyst of India. (2003) [18] Schoeler, F., Earnings Management to Avoid Earnings Decrease and Losses, Financial Reporting Research Group, (2005) pp [19] Sevin, S. and Schroeder, R. Earnings Management: Evidence from SFAS No. 142 Reporting, Managerial Auditing Journal, Vol.20, No. 1, (2005) pp [21] Swanson, N.J., The New Guideline for Goodwill Impairment Just another Tool for Earnings Management, ProQuest Information and Learning Company UMI Microform , (2007) pp [22] Teoh, S., Welch, I. and Wong, T.J., Earnings Management and the Long-Run Market Performance of Initial Public Offerings, Journal of Finance Vol. 53, No. 6 (1998)pp [23] Vafeas, N. Vlittis, A., Katranis, P. and Ockree, K., Earnings Management Around Share Repurchases: A Note, Abacus Vol. 39 No. 2 (2003) pp [24] Watts, R. L., Conservatism in Accounting Part I: Explanations and Implications, Accounting Horizons, Vol. 17, No. 3 (2003)pp Accounting standards Malaysia Accounting Standard Board (FRS 3) Business combination (2006) Malaysia Accounting Standard Board (MASB 21) Business Combination (2004) [20] Soo, B.S., Accrual Response to Mandated Accounting Principles: the Case of SFAS Nos. 2 and 34, Journal of Accounting and Public Policy, Vol. 18, (1999) pp ISSN: ISBN: