1 CORPORATE BOARD DYNAMICS AND CLASSIFICATION SHIFTNG OF EARNINGS MANAGEMENT IN EMERGING ECONOMIES   , Hope Ifeoma Orjinta Department of Accountancy, Faculty of Management Sciences, Chukwuemeka Odumegwu Ojukwu University, Igbariam Campus, Anambra State and   Emma i. Okoye Department of Accountancy, Faculty of Management Sciences Nnamdi Azikiwe University, Awka, Anambra State  
Abstract This study investigated whether corporate board dynamics of selected non-financial firms in Nigeria and Kenya constrains classification shifting of earnings management. Samples of 50 quoted non-financial firms were used for the period from 2010 to 2019. Our study used ex-post facto research design that cut across different sectors and different countries. The secondary sources of data were collected from annual reports of the quoted firms quoted in their respective stock exchange and four (4) specific objectives and hypotheses were tested and analyzed using panel regression techniques. The result revealed that corporate board dynamics such as female board gender has negative and significant effect in mitigating opaque manipulation practices such as classification shifting of quoted non-financial firms in Kenya and Nigeria which was statistically significant at 1% level of significance. While other corporate board dynamics such as independent board director, foreign board membership and board financial expertise were found to have negative and insignificant effect in curbing classification shifting of firms in both Nigeria and Kenya. Again it was discovered that adjusted R-squared which stood at 42.2% indicates that all the independent variables jointly explain about 42.2% of the system variation in classification shifting of our sampled companies while about 57.8% of the total variations were not explained, hence captured by the stochastic error term. The study therefore suggests that non-financial firms’ corporate board should be constituted by equal proportion of female to male directors as presence of women in the board help mitigates classification shifting. Keywords: Classification shifting, earnings management, corporate board dynamics, emerging economic, non-financial firms     1                                  INTRODUCTION Considering the number of accounting scandals in world known companies like Enron and WorldCom and resultant loss of investors’ fund.  For example, Enron and WorldCom intentionally reclassified some of its liabilities and losses which was part of core expenses to non-recurring expenses and move it to non-consolidated special purpose entities (Xudong, 2016). This reclassification shifting resulted to a decrease in unexpected non-recurring expenses of $1.01 billion in October 2001 and an equivalent increment in core earnings. Therefore, as a result of these accounting scandals, several specific rules have been adopted to maintain the credibility of annual reports and that of the organization in the eyes of external readers. These series of corporate scandals brought corporate board dynamics issues to the forefront of investors’ consideration. In emerging economies like Nigeria, as a result of these scandals aforementioned, attempt to address these concern has seen countries come up with different guidelines used in regulating firms across globe such as Code of Corporate Governance in Nigeria and Kenya. This is to enable the developing economies like Nigeria and Kenya to forge ahead. As a forward, efforts were made to institute corporate governance and one of the major components of corporate governance role that can determine the success of the company is the establishment of corporate board with diverse knowledge and experience. The corporate board of an organization constitutes of individuals from different ethnic groups, with different expertise and qualification, age and gender differences saddled with the obligations of checkmating the board of directors (Orjinta & Okoye, 2018). As a result of this, growing number of companies now disclose and explain their corporate governance generally and their corporate board dynamics specifically. Moreover, previous studies in earnings management paid more attention to discretionary accrual-based earnings management (DA) and real earnings management (REM) leaving the classification shifting which is the third dimension unattended to. This third dimension of earnings management is a relatively new research area and needed to be exploited. Only few study like Orjinta and Okoye (2018), Orjinta, Onuora and Agubata (2018), Athanasakou,Strong and Walker (2009),Barua,Lin and Sbaraglia(2010), Fan, Barua, Cready and Thomas (2010)and McVay (2006) have recently focused on this form of earnings management through classification shifting (CS) and they record inconsistent result. Not just that, most prior studies on earnings management were limited and has been found in advanced countries such as America and United Kingdom(Athanasakou, Strong & Walker2009;Peasnell,Pope& Young, 2005; Xie, Davidson & Dadalt, 2003;Zalata &Roberts, 2015). It is therefore evident from the above findings that the extant literature has created more need for further studies to substantiate the movement of the effect between corporate board dynamics and classification shifting. To the best of researchers’ knowledge, very few studies related to emerging economies like Nigeria and Kenya has been carried out and thus, this creates need for further studies. We also extended our study to cut across two African countries at the same time for a ten years period of time spanning 2010 to 2019 against what prior studies have done. This is the lacuna the current study intends to bridge therefore adding to the existing literature. Therefore, this paper is subdivided into five series together with this introduction. Section 2 covers the review of some related literature; section 3 concentrates on the methodology adopted while section 4 deals with discussion of results and analysis. Lastly in section 5, we draw the conclusion and proffer our recommendation for policy implementation. 2.1       CONCEPTUAL REVIEW Classification Shifting and Corporate Board Dynamics The use of classification shifting as a strategy to manage earnings was brought to light by the work of McVay (2006), when she developed empirical models designed to detect abnormal behaviour in core earnings and correlated unexpected core earnings with the incidence of special items. McVay (2006) investigated this third form of earnings management called classification shifting where she tested if management also alters core earnings since analysts and investors pay attention on core earnings, as opposed to bottom-line net income.   Corporate board is a sub-committee of corporate governance; therefore we focused on corporate board in this current study. They are instituted to checkmate the company’s activities and stop them from opportunistic behavior and at the same time help to mitigate classification shifting. When a firm is faced with a challenge which management has not faced before, the members of the board with expertise in that respective field can serve as a source of counsel to management. On this note, this current paper reviewed the following board dynamics (independent board directors, foreign board membership, board gender diversity and board financial expertise) as follows: Independent Board Directors: Independent directors are directors that have no personal or professional relationship with a company, other than being a board member (Ong & Djajadikerta, 2017). They are also often referred to as external directors. Independent director has the task of ensuring a balanced decision making, especially to protect minority stockholders and other relevant parties (Orjinta & Okoye, 2018). Therefore, outsider directors are appointed on the board to obtain independent monitoring mechanism over the board process thereby reducing agency conflicts and improve performance. Foreign Board Membership and Gender dynamics Foreign board diversity simply means the number of foreigners in the corporate board. That is people from other country that forms part of the management team. Gender dynamics simply means the percentage of female members in the corporate board. Oscar and Daniel (2013) argued that female board member improves board monitoring and hence prevent earnings management to a larger extent. This is because male counterpart are likely to view leadership as a series of transaction with subordinates, while female are more likely to have more interactive leadership style. Despite the fact that there are these arguments in favour of women directorship, in reality, their representation in the non-financial firms’ team of board of directors is very low, as some firms within the sector did not provide even a single seat for women. Board Financial Expertise: Board financial expertise is referred to as the board committee members who have the knowledge and experiences in accounting and finance or those who have membership in any accounting body or association (Hashim & Abdul Rahaman, 2011; Mohamad-Nor, Shafie & Wan-Hussin; 2010). Board financial expertise is captured as the totality of board members with accounting and financial backgrounds to the total number of board members (Yatim, Kent & Clarkson; 2006). The report of Rustam, Rashid & Zaman (2013) asserts that board members should be sound in financial matters. This is because board members that have experience in finance are more conversant with classification shifting strategies. It is only a board member that is grounded in accounting that can dictate when an item is misclassified or not.     2.2:      Theoretical Framework (Positive Accounting Theory) Our study was also anchored on the Positive Accounting Theory (PAT) propounded by Watts and Zimmerman in 1986 to explain accounting practice. Positive Accounting Theory takes the position of rationality of all individuals and believes that human beings act in self-interest to maximize their own utility, which corresponds with the agency theory perspective. Managers can either use their discretion to increase the wealth of all contracting parties, or to achieve own benefits at the cost of contracting parties. Such discretion possessed by the managers give them leeway into manipulating accounting figures to suit their self interest thereby engaging in earnings management. It was as result of this that we anchored our study on Positive accounting theory to show what drives managers into classification shifting of earnings management. 2.3:      EMPIRICAL LITERATURE AND HYPOTHESES DEVELOPMENT Independent Board Directors and Classification Shifting of Earnings management Literature on earnings management studies were of the opinion that that adding more independent director on the board enhances credibility and mitigates classification shifting. Take for instance, Zalata and Roberts (2016) discovered that companies with fewer independent directors are more likely to control fraud. On the contrary, Man and Wong (2013) further argued that firms with extensive earnings management are more likely to be controlled by independent directors. In the same vein, Abed, Al-Attar and Suwaidan (2012) documented a positive but insignificant relation between classification shifting and percentage of outsiders in the board. Nevertheless, considering the contradicting theoretical argument, this paper does not predict any sign for the proportion of independent directors but propose that there is a significant relation between the proportion of independent directors and earnings management (H01) Foreign Board Membership and Classification shifting of Earnings Management We found few studies in the literature relating to foreign board member diversity or national diversity to classification shifting of earnings management. One of the few studies was that of Enofe, Eyafekhe and Eniola (2017) who documented a negative effect between international diversity and earnings management. Based on the contradictory effect found on this variable and coupled with an apparent absence of studies from emerging economies of African countries perspective, the current study does not intend to propose any sign, rather the second hypothesis is drawn up in the null form as follows:there is no significant relation between foreign board membership and classification shifting of earnings management (H02) Gender Dynamicsand Classification shifting of Earnings management A lot of inconsistencies characterized the relationship that exists between board gender dynamics and classification shifting. A group of authors such as Orjinta and Okoye (2018), and Omoye and Eriki (2014) finds an inverse relationship between gender dynamics and earnings management. Similarly, Clikean, Geiger and Connell, (2001), Enofe, Iyafekhe, and Eniola (2017), and Zalata, Tauringana and Tingbani (2017), also reported similar findings and therefore argued that gender diversity mitigates the incidences of earnings management. Nevertheless, considering the contradicting theoretical argument, this paper does not wish to predict any sign for gender diversity in the board but propose that there is a significant relation between female representative and earnings management (H03). Board Financial Expertise and Classification Shifting of Earnings management Wasukan (2015) finds that the number of directors that are experienced in finance and accounting are indirectly associated with classification shifting, which supports the findings of Park and Shin (2004). In a study done by Cohen, Dey and Lys (2005), they finds that high expertise directors are more sensitive to curbing classification. In the same vein, Yang and Krishnan (2005) found that expertise has positive effect with earnings management. However, there are some inconsistencies that existed in the literature, for that reason, the current study does not intend to propose any sign, rather we hypothesize that there is a significant relation between expertise diversity and earnings management (H04).   Figure 2.1:  Conceptual Framework                 Source:  Researchers’ Idea (2020)   3.                     METHODOLOGY Research Design The study adopted expost-facto research design and a cross sectional data of non-financial firms quoted in the Nigeria and Nairobi Stock Exchange for the period of ten (10) years spanning 2010 to 2019. The population of non-financial firms quoted in selected emerging African countries was 116 non-financial firms in Nigeria and 43 non-financial firms in Kenya. Therefore, one hundred and fifty nine (159) companies constituted the population. Sixty-one (61) firms formed the sample size, which is selected using purposive sampling. It is worthy to note that sample size was arrived by adopting number estimation formula by Taro Yameni, (1967). Statistically, our sample size was chosen using Taro Yameni Formula stated as follows:                 n   =         N                                                                                         1 + N(e)2 Where  n= Sample size   N= Total population,      e = error term or significant level (10%) n=        159                                                                      1 + 159 (.10)2                                                             n =  159                                                             2.59                                                               n=  61.39                                                               n   = 61 firms In addition, the sample also excluded newly quoted companies that did not exist as at beginning of 2010 i.e. newly quoted and newly listed companies with missing data points were excluded, as this will result in missing data for the period being studied. Based on consideration of sampling, the size of sample in this study is sixty-one (61) firms but there are 11 companies that do not have the completeness of the data and they were filtered as follows: Sample Selection and Filtration                                                                                  61 firms Less:  Industry-years with number of observations < 10/Newly listed firms             4 Less:Removal of companies with unavailable data in the model                               5 Less: Companies that have been delisted                                                                    2 Final Sample  Size                                                                                                     50 Therefore, only 50 firms are with sufficient information and were finally selected to be sample of this study. Note that the 50 firms were selected based on complete availability of data. Modeling To vividly summarize what we did, we followed McVay (2006)expectation model and estimate unexpected core earnings and non-recurring items or unexpected operating expenses. We then associate unexpected core earnings with non-recurring expenses while variables such as return on asset (ROA) and firm size (FSIZE) were added to the models as control variables to control for performance. A positive relationship between unexpected core earnings and non-recurring items is an evidence of classification shifting from cost of goods sold (COGS) to operating expenses; thus evidence of classification shifting. The model to dictate the existence of classification shifting is specified as follows: UNEXPCE = β0 + β1 +NRECit+tβ2ROA1t+β3FSIZE1t+ℇ………………………………….(1)        While the main regression model to test whether corporate board dynamics collaborates or mitigates classification shifting is stated as follows: CSEMTit=β0+β1INBD1t+β2FBMP+β3BGED1t+β4BFXP1tβ5ROA1t+β6FSIZE1t+Ɛ1t…… (II) Where UNEXP-CE means Unexpected Core Earnings computed as difference between reported core earnings and expected core earnings where the expected core earning value is calculated using McVay (2006)model while NREC stands for non-recurring items measured as the difference between reported core earnings and bottom line net income scaled by sales. CSEMT stands for classification shifting of earnings management ascertained using equation 1 above. INBD means independent board director captured as the proportion of independent or non-executive directors on the board divided by the total number of directors on the board. FBMP stands for foreign board membership measured as the percentage of the number of foreigners in the board to total board of directors while BGED equals board gender dynasty proxy as the total number of female directors in the board. BFXP stands for board financial expertise measured as the proportion of financially literate board members to the total number of board members; ROA is the return on assets measured as net income divided by the average total assets and FSIZE means firm size measured as the natural log of total assets.   4.1                   RESULTS ESTIMATION AND DISCUSSION OF FINDINGS The study investigated the causal effect that exists between corporate board dynamics variables and classification shifting of earnings management of listed non financial firms between 2010 and 2019. The study carried out some preliminary tests like descriptive statistics and variance inflation factor (VIF) analysis. The descriptive statistics was used to analyze the data in order to ascertain the normality and nature of the data. To further check for the case of perfect correlation among variables, Variance inflation factor (VIF) was conducted to test for the presence of multi-colinearity. Finally, the study used panel regression analysis in obtaining functional causal effect between the regressors putting into consideration the fixed or random effect testing for interpretation of regression result. Descriptive Statistics The Table below shows the descriptive statistics of the selected non financial firms in both Nigeria and Kenya that make up our sample. Table 4.1:       Descriptive Statistics   CSEMT INBD FBMP BGED BFXP ROA FSIZE  Mean  1.415320  0.447300  0.128440  1.794000  3.298800  0.595560  46.51422  Median  0.745000  0.500000  0.100000  2.000000  3.000000  0.640000  44.31000  Maximum  141.7400  0.930000  1.020000  5.000000  7.000000  1.420000  317.1900  Minimum -40.52000 0.120000  0.000000  0.000000  0.000000 -0.430000  19.49000  Std. Dev.  12.22264  0.152822  0.130924  1.018646  0.864206  0.294281  21.52379  Skewness  3.709986  0.262054  3.076149 -0.012601 -0.534181 -0.566647  4.143789  Kurtosis  40.34722  4.363265  16.45631  2.448343  4.427786  3.186500  51.18642                  Jarque-Bera  30205.65  44.44130  4560.896  6.353342  66.24934  27.48206  49804.47  Probability  0.000000  0.000000  0.000000  0.041724  0.000000  0.000001  0.000000                  Sum  707.6600  223.6500  64.22000  897.0000  1649.400  297.7800  23257.11  Sum Sq. Dev.  74547.06  11.65385  8.553383  517.7820  372.6793  43.21414  231173.4                  Observations  500  500  500  500  500  500  500 Source: researcher summary of descriptive statistics result (2020) using E-view 10 This analysis was conducted to describe the overall distributional properties of the series and to discover any unusual patterns of observations if any. Thus, initial exploration of the data using simple descriptive tools was documented to summarize the nature of data generated for the study. It shows the mean values for each of the variables, their maximum values, minimum values, standard deviation and Jarque-Bera values which show the normality of the data. This section provides the descriptive statistics as per the objective of the study. Firstly, it was observed that over the period under review, the sampled firms have average positive classification shifting of 1.42% approximately. Within the period under review, the maximum and minimum values of classification shifting as a measure of earnings management were 141.74 and -40.52 respectively. The large difference between the maximum and minimum values of classification shifting indicates that the degree of misclassification differs greatly among the firms selected and over the period under review, this shows that the firms are not homogenous. The result of the independent directors in the board shows that on the average, quoted companies in Nigeria and Kenya have about 45% independent directors in their board as against 51% reported by Zalata and Roberts (2016) in United Kingdom, 43% reported by Peasnell et al. (2005). But supports 45% reported by Osma (2008) in the United States. However, some firms maintained only 12% independent director as their minimum number over the years while others have about maximum of 93% independent directors in the board. The mean foreign board membership was 0.128 suggesting that on average, the firms under study had about 12% foreigners on the board. The minimum and maximum foreign board membership was 0.000 and 1.02 respectively implying that some firms do not have any foreign member in their board at all while the firm with the highest number of foreign directors on the board had about I person or one foreign director. That is to show that firms with the maximum number of foreign directors on the board had about 1% of the total directors being foreigners. The standard deviation for foreign board membership was .130 demonstrating that out of the 50 listed non-financial firms studied, foreign board membership was spread around the mean with about 13%. The skewness for directors’ foreign membership was 3.076 implying that data about foreign board membership was positively skewed with most values bunched to the left. The kurtosis for foreign board membership was 16.45 which is more than 3 implying that it has leptokurtic distribution or having outliers but not strong enough to distort the generalization. Mean gender dynasty was 1.794 suggesting that on the average, the firms have at least one female member in the board while the minimum value of 0.000 implying that some firms do not even have one female board member in their total number of board of directors and the maximum female member in the board was 5 females.  This shows that some firms have a good number of female directors in their board up to a maximum of 5 female members in the board. The standard deviation for gender was 1.018 demonstrating that out of the 50 non-financial firms, female gender was spread around the mean with about 1.018 female board members. The skewness was -0.012 implying data about gender diversity was negatively skewed with most values bunched to the right. The value of kurtosis was 2.448 implying that the data about female gender was distributed with kurtosis less than 3 hence said to be platykurtic and having few outliers. Board financial expertise result showed that on the average, about 3 members of the board of quoted companies in selected African countries are financial expert, however, some firms over the years have maximum number of about 7 members who have financial experience/expertise while others have no board members with financial or accounting experience/expertise at all. Lastly, the Jarque -Bera (JB) and its probability which test for normality or existence of outlier shows that all the variables were normally distributed at 1% level of significance except female board gender dynasty that was normally distributed at 5% level. 4.2       Test of Multicollinearity or Variance Inflation Factor (VIF) Multicollinearity was tested by computing the Variance Inflation Factor (VIF) and its reciprocal or the tolerance to know whether the independent variables used are perfectly correlated. The result of the Variance Inflation Factor (VIF) is provided below in table 4.2. below: Table 4.2.       Variance Inflation Factor Result (Nigeria and Kenya) Variance Inflation Factors   Date: 11/25/20   Time: 02:30   Sample: 2010 2019   Included observations: 500                     Coefficient Uncentered Centered Variable Variance VIF VIF                 C  12.35616  8.371906  NA INBD  18.49923  3.521563  1.013767 FBMP  14.62910  1.170890  1.007375 BGED  0.372364  1.827692  1.015697 BFXP  0.368170  3.723717  1.009143 ROA  3.486162  1.848416  1.010617 FSIZE  0.000592  1.882103  1.014144                 Source: Researcher’s summary of VIF result (2020) As can be observed from the result of VIF in table 4.2 above, the mean value or the variance inflation factor (VIF) values of the independent variables coefficient is less than 10, therefore the effect of multi-collinearity is negligible. This implies that there was no multi-collinearity problem with the variables thus all the variables were maintained in the regression model. Therefore it can be concluded that there is no problem of multi-collinearity. It can also be seen from the table that all the variables had a variance inflation factor (VIF) of less than 10 as follows independent board director (1.014) approximately, foreign board membership (1.007) board gender dynasty (1.015), board financial expertise (1.009), return on assets (1.010) and finally, firm size (1.014). This implies that there was no multi-collinearity problem with the variables, thus all the variables were maintained in the regression model. This means that there are no variables with outlier, and none of the variables are highly correlated. Our finding also justifies the use of panel least square estimation techniques. Hence, any recommendations made to a very large extent would represent the characteristics of the true population of study and thus can be used to draw conclusion. 4.3:                  Regression Results The regression test below was conducted to check and ascertain whether non-financial firms in Nigeria and Kenya engage in classification shifting in a bid to smooth their earnings. It is worthy to note that when firms engage in classification shifting, the unexpected core earnings increases with non-recurring items, thereby giving rise to a positive and direct relationship between unexpected core earnings and non-recurring items. A positive relationship between unexpected core earnings and non-recurring item is an evidence of Classification shifting (CS) and it also suggests that firms shift recurring/core expenses to non-recurring items to inflate core earnings, thus evidence of classification shifting (CS). Therefore, the current study therefore supported the findings of Athanasakou et al. (2009) and considers a firm as engaging in classification shifting if it has both positive coefficient value between unexpected core earnings and non-recurring items. The probability value shows that the effect of non-recurring expenses on unexpected core earnings is statistically significant at 1% level.  This means that as non-recurring expenses is increasing, unexpected core earnings is also increasing, thus evidence of classification shifting among firms quoted in Nigeria and Kenya. Table 4.3.1     Regression of Unexpected Core Earnings on Non-Recurring Items (Eqn 1) Dependent Variable: UNEXPCE     Method: Panel Least Squares                         Variable Coefficient Std. Error t-Statistic Prob.                       C 0.845941 0.250283 3.379943 0.0008 NREC 30.78411 1.296397 23.74590 0.0000 FSIZE -0.181316 0.004314 -42.02699 0.0000 ROA 5.708417 0.203722 28.02060 0.0000                     R-squared 0.773618     Mean dependent var -6.935587 Adjusted R-squared 0.772708     S.D. dependent var 5.781251 S.E. of regression 2.756220     Akaike info criterion 4.870917 Sum squared resid 5667.175     Schwarz criterion 4.895557 Log likelihood -1822.594     Hannan-Quinn criter. 4.880411 F-statistic 849.7735     Durbin-Watson stat 1.133637 Prob(F-statistic) 0.000000                 Source: Reseachers summary of regression Result         Source: Researchers’ computation (2020) Now that we have confirmed the existence and evidence of classification shift among selected non-financial firms in emerging economies, we can now proceed with the main regression analysis to see if corporate board collaborates or mitigate classification shifting. Therefore, to examine the relationship between the dependent variable (classification shifting) and the independent variables (INBD, FBDP, BGED, BFXP) and to test the formulated hypothesis, we employed a panel regression analysis and regressed all the independent variables and control variables against classification shifting (earnings management). The summarized result of regression analysis is presented below. However, the study takes into cognizance the non homogeneity nature of the firms, hence the need for testing its effect on the data. This necessitated the use of Hausman effect test to ascertain which effect to explain. Hausman test is used to decide between fixed effect model or random effect model. When the Chi square (Prob) value is greater than 5%, you interpret random effect and said that random effect is more preferred to fixed effect but when it is less than 5%, you interpret fixed effect and said that fixed effect is more preferred to random effect. Below is the summary of the Hausman test result, Table 4.3.2 Hausman Test Correlated Random Effects - Hausman Test   Equation: Untitled