CEO characteristics and earnings management: Evidence from mergers and acquisitions

CEO characteristics and earnings management: Evidence from mergers and acquisitions Thai Quoc Nguyen 1 School of Business and Law University of East London E15 4LZ t.q.nguyen@uel.ac.uk Nguyet Nguyen Portsmouth Business School University of Portsmouth PO1 2UP nguyet.nguyen@port.ac.uk Tri Tri Nguyen School of Business and Law University of East London E15 4LZ t.t.nguyen@uel.ac.uk Chau Duong School of Business and Law University of East London E15 4LZ c.duong@uel.ac.uk 1 Corresponding author. Address: Stratford campus, School of Business and Law, University of East London, London E15 4LZ, UK. E-mail: t.q.nguyen@uel.ac.uk

Abstract Previous research suggests that share-financed acquirers inflate their earnings before merger and acquisition announcements. The existing literature also indicates that characteristics of chief executive officers (CEO) could affect earnings management. In this study, we extend prior studies by examining the relationships between CEO characteristics and accrual earnings management in share-financed acquirers before mergers and acquisitions. We find that CEOs with high financial expertise and high reputation are associated with lower abnormal accruals in stock acquirers. The correlations are statistically significant and consistently exist in the first year before deal announcements. The findings are robust as we measure abnormal accruals in different ways and employ different models. The evidence suggests that CEO characteristics have an impact on earnings management in the contexts of mergers and acquisitions and have some implications for practitioners. Keywords: accrual earnings management, chief executive officers, mergers and acquisitions

1. Introduction: There is a large volume of published studies showing that share-financed acquirers (stock acquirers) significantly inflate earnings before mergers and acquisitions (MA, hereafter) to reduce acquiring cost (Botsari and Meeks, 2008; Louis, 2004; Pungaliya and Vijh, 2009). Prior studies also provide evidence that characteristics of CEOs matter in the context of MA (Custódio and Metzger, 2013; Grinstein and Hribar, 2004; Malmendier and Tate, 2008; Walters et al., 2007). In this study, we hypothesise that CEO characteristics could impact earnings management before MA announcements. Characteristics investigated are financial expertise and reputation of CEOs. Firstly, there is evidence that managers with financial expertise are more likely to produce higher organisational outcomes. For example, Custódio and Metzger (2013) find that financial expert CEOs are often in a better position to negotiate in MA transactions so that the acquirers might pay less for the deals, resulting in higher post- MA returns. CEOs with financial expertise are also associated with better organisational outcomes such as more flexible financial policies, ability to access external funds in difficult credit situations (Custódio and Metzger, 2014). Also, Aier et al. (2005) find that financial expertise of chief financial officers (CFOs) is negatively associated with accounting restatements, which is a proxy for earnings management. Secondly, the existing literature suggests that reputation of managers affect corporate practices, but the evidence is mixed. On the one hand, Jian and Lee (2011) find that CEOs with high reputation is associated with favourable responses from the market to announcements of capital investment. Also, CEOs with high reputation can generate higher post-investment returns. Also, Francis et al. (2008) find that firms with high earnings management are more likely to hire CEOs with high reputation so that those

CEOs could help to reduce earnings management in subsequent periods. The evidence suggests that there is a negative relationship between the reputation of CEOs and earnings management. On the other hand, CEOs with high reputation are found to prioritise enhancing their reputation instead of improving the wealth of shareholders (Hirshleifer, 1993; Malmendier and Tate, 2009). In this paper, we specifically examine the influence of CEOs traits on earnings management before MA deals, a setting which existing evidence has suggested that the incentives for earnings management would be high for the acquirers CEOs with low financial expertise and low reputation. For measuring financial expertise, we use the working experience of CEOs as a chief financial officer in the past and financerelated qualification of CEOs (Aier et al., 2005; Nguyen et al., 2017). We measure reputation by using media coverage and firm financial performance during CEO tenure (Francis et al., 2008; Jian and Lee, 2011; Milbourn, 2003). Regarding proxies for earnings management, we estimate abnormal accruals by using models following Jones (1991) and Dechow et al. (1995). We use a sample of 2,832 firm-year observations of UK listed companies from 2007 to 2012, which includes 71 observations with share-financed MA deals. We find that, in the first year before deal announcements, CEOs with high expertise and high reputation are associated with lower earnings management in share and hybird-financed acquirers. The results are statistically significant and robust as we measure abnormal accruals in different ways and employ different models. The evidence is new and adds significantly to the growing literature investigating how CEOs traits influence corporate practices. The paper is the first research which examines the relationships between CEO characteristics and earnings management

in the contexts of MA. While previous research has only focused on earnings management of acquirers and targets before MA (Botsari and Meeks, 2008), we provide evidence that CEO characteristics affect earnings management in the first year before MA. Also, we provide further evidence to support the notion that CEO characteristics are important determinants financial policies around MA deals (Custódio and Metzger, 2013; Grinstein and Hribar, 2004; Malmendier and Tate, 2008; Walters et al., 2007). The findings of this study are useful for practitioners such as investors and auditors. Investors should be cautious when using information related to MA announcements of acquirers having powerful CEOs because earnings are more likely to be manipulated in the first year before MA. The reason is that inflated earnings can be reversed in subsequent periods, which in turn reduce abnormal returns from investments in stocks of acquirers. Similarly, when auditing financial statements, auditors could particularly pay attention to firms having powerful CEOs because the risks of earnings management are high in the first year before MA. The following part proceeds as follows: section 2 presents a literature review and hypothesis development, section 3 describe methodologies, section 4 discussed findings, and section 5 provides conclusions. 2. LITERATURE REVIEW AND HYPOTHESIS DEVELOPMENT 2.1. Earnings management in stock acquirers Mergers and acquisitions a key part of financial activities and corporate growth strategy. The research on acquirers earnings management or the effect of acquirers earnings management on shareholder wealth thus attracts attention from researchers.

Some research documents that stock acquirers manage earnings before mergers and acquisitions (Botsari and Meeks, 2008; Erickson and Wang, 1999; Louis, 2004). Stock acquirers have motivations to manage earnings to increase their stock price. If acquirers stock price is higher, the number of stock used for exchange in MA deals would be lower because the exchange ratio is inverse to the acquirers stock price. The motivation is explained well under the agency theory (Jensen and Meckling, 1976; Watts and Zimmerman, 1990) which suggests that acquirers manage earnings when the costs of undoing earnings management are lower than the costs of detecting earnings management (Erickson and Wang, 1999; Watts and Zimmerman, 1990). However, the risk of detection may not deter acquirers from involving in earnings management activities if earnings are managed within the general accepted accounting principles (GAAP). Thus, earnings management in acquirers is not easily detected nor prevented. 2.2. CEO characteristics and earnings management Although previous research documents that stock acquirers inflate their earnings management before MA, acquirers with different CEO characteristics could manipulate earnings differently. This section briefly discusses the relationship between CEO characteristics and earnings management. Previous research provides evidence that working experience and personalities of CEOs could affect corporate practices. Hambrick and Finkelstein (1987) argue that managers traits matter because managers rely on their own experience, values and personalities to interpret strategic circumstances. Miller et al. (1986) also provide a psychological explanation for the relationship between top executives characteristics and organisational outcomes.

There is also empirical evidence in line with the argument that CEO characteristics are determinants of organisational outcomes. Bertrand and Schoar (2003) find that management style of managers affects various corporate practices. Ge et al. (2011) also provide evidence that style of chief financial officers (CFO) affect corporate investment and financial policies. On the direct evidence on CEO characteristics, Custódio and Metzger (2014) find that financial expert CEOs are more likely to result in better organisational outcomes such as more flexible financial policies, ability to access external funds in difficult credit situations. In addition to the above evidence, the existing literature also suggests that executives characteristics influence accounting practices such as earnings management. Specifically, financial expertise is an important determinant of earnings management. Aier et al. (2005) examine whether CFO characteristics are related to accounting restatement, a measure of earnings management. They argue that financial expert CFOs could contribute much to a better accounting system. Thus they are more likely to result in good accounting outcomes. Consistent with their prediction, they find that firms which have financial expert CFOs are less likely to be involved in accounting restatement. The next characteristic influencing earnings management is reputation, but the evidence is mixed. One the one hand, Hirshleifer (1993) suggest that managers have motivations impact corporate investment to build up their reputation rather than the wealth of shareholders. Also, Malmendier and Tate (2009) evaluate the effect of CEOs achieving superstar status on firm performance. The researchers realise that compensation, status and reputation of executives in the United States are formed in a highly skewed distribution: a few number of superstars enjoy significant benefits of

the rewards. They find that, after award-winning, awarded CEOs underperform compared with their prior performance and compared with non-winning CEOs. The evidence also indicates that superstar CEOs extract higher compensation and firms earnings are managed upward after CEO winning awards. The evidence suggests that CEOs opportunistically inflate their firms earnings. On the other hand, Francis et al. (2008) also find that firms with high earnings management are more likely to hire CEOs with high reputation so that those CEOs could help to reduce earnings management in subsequent periods. The evidence suggests that there is a negative relationship between the reputation of CEOs and earnings management. In addition to the relevance of separate CEO characteristics on earnings management, Nguyen et al. (2017) offer an aggregate measure of various CEO characteristics, namely PSCORE, and find that PSCORE is positively correlated with different measures of accrual-based earnings management and real earnings management. They also find that PSCORE is associated with financial statement errors, measured by differences between actual distributions of first digits shown in financial statements and expected distribution following Benford s Law. 2.3. Do CEO characteristics matter for mergers and acquisitions? As discussed above, previous research documents that stock acquirers inflate their earnings management before MA (Botsari and Meeks, 2008; Nguyen and Duong, 2017). Prior evidence also suggests that CEO characteristics are important determinants of earnings management (Feng et al., 2011; Francis et al., 2008; Malmendier and Tate, 2009). Thus, it is possible that CEO characteristics may affect earnings management around MA deals.

There are reasons to believe that CEO traits are relevant in the contexts of MA. Grinstein and Hribar (2004) provide an explanation why CEO characteristics are important for MA deals. They find that powerful CEOs are more likely to impact decisions boards of directors to extract higher compensation and bonus related to the completion of MA deals. Walters et al. (2007) also explain that CEOs impact MA deals to extend their tenure in acquirers. Custódio and Metzger (2013) show that financial expert CEOs are often in a better position to negotiate in MA transactions so that the acquirers might pay less for the deals, resulting in higher post-ma returns. In contrast with Custódio and Metzger (2013), Malmendier and Tate (2008) find that firms which have over-confident CEOs are more likely to make value-destroying MA because CEOs over-estimate subsequent returns of MA deals. 2.4. Hypothesis As explained earlier, the share-financed acquirers have the incentive to increase their stock price prior to MA. That is because if acquirers manage earnings to inflate stock price, the number of acquirers share are used to pay for target firms would be lower (Baik et al., 2007; Botsari and Meeks, 2008; Gong et al., 2008; Louis, 2004). Therefore, the first hypothesis is stated as below: H1: Share-financed acquirers significantly inflate accrual-based earnings prior to a merger announcement day. Although the importance of CEO characteristics on MA deals, the existing literature does not provide any evidence on the relationship between CEO characteristics and earnings management around MA. As a response to the above gap in the literature,

we examine how characteristics of CEO affect earnings management before MA deals. The existing literature as shown that CEOs of share-financed acquiring firms would expectedly try to inflate earnings in the year preceding the deals to minimize the number of shares exchanged for the target, assuming they are working to maximize shareholders wealth (Botsari and Meeks, 2008; Erickson and Wang, 1999; Louis, 2004). It has also been established that financial expertise would generally help mangers to be able to deliver better organisational outcomes (Aier et al., 2005; Albring et al., 2014; Badolato et al., 2014; Burak Güner et al., 2008). In the context of MA deals, CEOs with financial expertise would typically have better estimate of the financial costs and benefits of the deals (Ge et al., 2011). Therefore, we expect CEOs with financial expertise could be able to deliver good MA outcomes without having to resort to costly earning management options. Following that line of intuition, our first hypothesis is as follows: H2: CEOs with financial expertise are associated with lower earnings management in the year preceding share-financed MA deals. In a similar vein, the extant literature has shown that CEOs with more established reputation are more likely to produce better organisational outcomes (Francis et al., 2008; Jian and Lee, 2011). Taking that evidence in the context of share-financed MA deals, we believe more reputable CEOs would be less likely to resort to costly earnings management, not only just because their reputation would imply a good deal is more likely to be reached, but also because engaging in earnings management would risk their reputation being damaged. We, therefore, form our second hypothesis as follows:

H3: More reputable CEOs are associated with lower earnings management in the year preceding share-financed MA deals. 3. DATA AND METHODOLOGIES 3.1. Sample selection We use a sample of UK listed companies from 2007 to 2012. The sample includes 1,814 firms which are listed on the London Stock Exchange as at 31 st December 2012. We download data from Datastream, including financial variables needed for all analyses in this paper, together with the International Securities Identification Number (ISIN), company name, and industry level-six code. We eliminate utility and financial firms. Financial statements not presented in sterling ( ) are removed. We also exclude observations of which the number of days in a fiscal year is more than 400 or less than 330 and observations. We then collect data on all MA deals with a UK acquirer for the period from 2007 to 2012 from Bloomberg, including announcement date, payment method, the ISIN of the acquirer. Deals which are not completed are eliminated. Finally, we keep only deals which are financed in full or partly by share. The MA data is linked with the data downloaded from Datastream via ISIN of the acquirers. For each company remained in the sample, we identify the CEO as the individual listed in Bloomberg as the Chief Executive Officer or equivalent in Bloomberg. If a firm has two CEOs in a firm-year, we exclude the CEO has less than 6 months in role. For Co- EO firm-years, we will exclude CEO has less time in role than another. Next, we hand collect data on characteristics of each CEO in the sample from Bloomberg, including

the CEO s financial certificate and/or qualification 2, if any. If data for a CEO is missing from Bloomberg, we obtain the data directly from the annual reports downloaded from Key Note. If we are unable to obtain CEO s financial certificate and/or qualification and compensation using the above procedures, we drop the observation from the sample. For media coverage of a CEO in a year, we conduct a search for business news related to the CEO from the LexisNexis database using CEO s name and company s name. Following Francis et al. (2008), we restrict our search to UK national newspapers within the three-year period prior to and including the considered fiscal year. We keep only observations which have all of the data needed for the main analyses. The process yields a sample of 7,197 firm-year observations (including 1,707 separate companies across 54 industries). The sample consists of 135 share-financed deals. All continuous variables are winsorized at the 1 st and 99 th percentiles to mitigate the influence of outliers. 3.2. CEO characteristics The CEO s characteristics investigated in this study include financial expertise, and reputation of CEOs. Firstly, following previous studies (Aier et al., 2005; Nguyen et al., 2017), we have two proxies for financial expertise. Finance and accounting certification and CFO working experience help CEOs gain financial expertise, which is relevant in the context of earnings management. The first proxy is finance and accounting certification (ccert), which is a dummy variable with the value of one if a 2 Qualification is a professional accounting certification issued by one of five current qualifying bodies: Chartered Certified Accountants (ACCA), Association of International Accountants (AIA), Institute of Chartered Accountants in England and Wales (ICAEW), Chartered Accountants Ireland (CAI) and Institute of Chartered Accountants of Scotland (ICAS).

CEO has a master of business administration or a chartered accountant certification accredited by the Financial Reporting Council (Financial Reporting Council, 2016) or equivalent, zero otherwise. The second proxy is CFO experience (cexp), which is a dummy variable with the value of one if a CEO has worked as a chief financial officer in the past, zero otherwise. Under this construction, CEOs have high expertise when ccert and cexp are equal one. Secondly, we use media coverage (cpress) as the first proxy for the reputation of CEOs (Francis et al., 2008; Jian and Lee, 2011; Milbourn, 2003; Nguyen et al., 2017). cpress is a dummy variable with the value of one if the number of news covering CEO name and his company in year t is one or higher than one, zero otherwise. We also use firm financial performance during CEO tenure (croa) which is average returns on assets during the last three years of CEO tenure as the second proxy for the reputation of CEOs. croa is a dummy variable with the value of one if the average ROA during the last three years is higher than median average ROA during the last three years, zero otherwise. Under this construction, CEOs have high reputation when cpress is equal one and croa is positive. 3.3. Accrual-based earnings management Accrual-based earnings management refers to the use of accounting choices to affect reported earnings (Healy and Wahlen, 1999). Under accounting standards, accruals help financial statements reflect true and fair financial performance and net assets because accrual-based accounting requires transactions to be recorded at the time of occurrence rather than at the time cash is received or paid. In accounting research, normal accruals are estimated based on firm characteristics, and differences between

actual and predicted values are abnormal accruals 3 (Dechow et al., 1995; Jones, 1991; Kothari et al., 2005; Van Caneghem, 2002). Abnormal accruals are used as a proxy for earnings management. Although there are various models to estimate accrual-based earnings management (Ball, 2013; Dechow et al., 2010; DeFond, 2010; Fields et al., 2001; Holthausen et al., 1995), Peasnell et al. (2000) argue that the Jones model and its modified versions (Dechow et al., 1995; Jones, 1991) remain among the most effective models to proxy for earnings management in the UK. Thus, in this study, we use the original model of Jones (1991) and the modified-jones model (Dechow et al., 1995) to estimate abnormal accruals. We firstly run the following regression for each industry in each year with at least six observations: TA i,t 1 = α + β A 1 ( ) + β i,t 1 A 2 ( REV i,t ) + β i,t 1 A 3 ( PPE i,t ) + ε i,t 1 A i,t (1) i,t 1 Where, TA i,t is the total accruals, which equal net income minus cash flow from operations, of firm i in year t. is the total assets of firm i at the end of year t 1, REV i,t is the change in revenues of firm i in from year t-1 to year t, and PPE i,t is the gross value of property, plant and equipment of firm i in year t. Using the coefficients α, β 1, β 2, and β 3 estimated from equation (1), we compute abnormal accruals under the original Jones model as follows: ATA_JM i,t = TA i,t 1 [α + β 1( ) + β 2( REV i,t ) + β 3 ( PPE i,t )] (2) 3 In this study, we use terms abnormal accruals and discretionary accruals interchangeably.

The second measure of abnormal accruals is estimated using the modified-jones model (Dechow et al., 1995). In this model, changes in revenues ( REV i,t ) are deducted by changes in receivables ( REC i,t ) in the second stage to mitigate the concern of earnings management through sales. In particular, we run regression (1) as above, but in the second stage we calculate abnormal accruals as follows: ATA_MJM i,t = TA i,t 1 [α + β 1( ) + β 2( REV i,t REC i,t ) + β 3 ( PPE i,t )] (3) Abnormal accruals as estimated above are widely used as proxies for earnings management. However, there are some criticisms that those models rely on total accruals, which include long-term components, such as depreciation, which are less likely to be managed by managers because of their high visibility to financial statement users (Peasnell et al., 2000). Thus, we also employ abnormal working capital accruals to mitigate the concerns. We calculate two measures of abnormal working capital accruals, one follows the original Jones model (AWCA_JM) and the other follows the modified Jones model (AWCA_MJM). The estimation procedures are the same as for ATA_JM and ATA_MJM, but we replace total accruals (TA) in equations (1), (2) and (3) by working capital accruals, calculated as net income before extraordinary items minus depreciation and amortisation expenses. 3.4. Main regressions 3.4.1. Main regressions To test hypothesis H1, we run regressions to investigate whether sharefinanced acquirers inflate accruals earnings prior to MA. Specifically, the regression is as follows:

AA i,t = α + β 1 (MA i,t ) + β 2 (amtb i,t 1 ) + β 3 (asize i,t 1 ) + β 4 (aroa i,t ) + β 5 (alev i,t ) + β 6 (SEO i,t ) + β 7 (anoa i,t ) + Year Fixed Effects + Industry Fixed Effects + ε i,t (4) AA i,t are replaced by abnormal total accruals estimated by Jones and modified Jones models under cash flow approach (ATA_JM i,t and ATA_MJM i,t ) and abnormal working capital estimated by Jones and modified Jones models under cash flow approach (AWCA_JM i,t and AWCA_MJM i,t ). asize i,t 1, amtb i,t 1, alev i,t, SEO i,t, anoa i,t and NEGZSCORE i,t are control variables to control the effects of firm size, growth opportunities, debt, equity issuance and the level of net operating assets. T will be replaced by t-1, t-2 and t-3 for investigating whether acquirers inflate accruals earnings in Year t-1, Year t-2 and Year t-3. MA i,t is an indicator variable which is set to one for a share-financed acquiring firms in year t, zero otherwise. To study the relationship between CEO characteristics and earnings management, we use the following set of regressions: A = α + β 1 (X i,t 1 ) + β 2 (X i,t 1 MA i,t ) + β 3 (MA i,t ) + Control Variables + Year Fixed Effects + Industry Fixed Effects + ε i,t (5) Where A can be ATA_JM i,t 1, ATA_MJM i,t 1, AWCA_JM i,t 1, and AWCA_MJM i,t 1 (used as substitutes); X i,t 1 is the CEO characteristics which can be cexp i,t 1, ccert i,t 1, cpress i,t 1, cro and cpayslice i,t 1 in Year t-1 the first year prior to a MA deal; MA i,t is equal to one if firm i announces a share-financed MA in year t, zero otherwise; 3.4.2. Control variables

Previous studies have revealed that there are many factors which are able to affect companies earnings management. In this section, we will introduce the factors that may influence company s accrual earnings management and how these factors are calculated to control the impact on the results of this study. Industry-year adjustment: The measures of accrual earnings management are calculated for each industry-year. Therefore, the control variables in this study will be adjusted for the corresponding mean of the industry-year, except for dummy variables. Industry-adjusted values of the actual value of each industry except for the respective industry average. The equation of industry-year adjusted are as following: ac i,t = C i,t C t,k (6) Where, ac i,t is industry-year adjusted control variable of firm i in year t; C i,t is the actual control variable of firm i in year t; C t,k is the mean of corresponding all firms in industry k in year t (k is Datastream level-six industry numbers). Incentive: Equity issue and stock overvaluation The first group of factors that influence earnings management is the company s incentive such as equity issue and stock overvaluation. In case of equity issue, previous research has shown that companies are more

likely to increase earnings before the issue of equity (Cohen and Zarowin, 2010; DuCharme et al., 2004; Iqbal et al., 2009; Iqbal and Strong, 2010; Rangan, 1998; Shivakumar, 2000; Teoh et al., 1998). In this study, we will control the equity issuance factor by using the equity issue dummy control variable SEO i,t 1. The SEO i,t 1 equals to 1 if the company s outstanding shares increase by at least 5% compared to last year, zero otherwise. We expect the coefficients of SEO is positive. Previous research has also shown that companies with stocks that are overvaluated the real value of the stock, having a pressure to increase their earnings (Jensen, 2005; Skinner and Sloan, 2002) Skinner and Sloan, 2002). The pressure from the fact that these companies try to maintain their high stock prices. In order to control the highly valued stock, we use industry-adjusted market to book ratio (amtb i,t 1 ), where MTB is defined as the ratio of market value divided by book value of equity. We expect the coefficient of MTB is positive. Firm s characteristics: The second group of factors that could influence firm s earnings management is the firm s characteristics. We firstly control for the firm size. Lang and Lundholm (1993) show that the large firms find difficult in managing earnings due to their high public visibility. Consistent with Lang and Lundholm (1993), Dechow and Dichev (2002) find that firm size is negatively correlated with earnings management. Therefore, we control for firm size by using industry-adjusted size (asize i,t 1 ), where SIZE equals to

natural log of the market value of equity at the end of fiscal year. We expect the coefficient of asize is negative. The second firm characteristic could affect to earnings management is financial leverage. Houmes and Skantz (2010) show that firms with a high level of leverage are more likely to inflate their earnings because the debts usually are tied to the firm s performance. In line with Houmes and Skantz (2010), Alsharairi (2012) find that the non-cash acquirers with high leverage is positively correlated to earnings management. To control the financial leverage, we use industry-adjusted leverage (alev i,t 1 ), where LEV is the total of long-term and shortern debts at the end of year t divided by total assets at the end of year t. According to the previous studies, the sign of the coefficient of alev is mixed. Thus, we do not expect the sign of the coefficient of alev. Lastly, previous research documents that the ability to use accruals in current period affect firms earnings management because accruals are reversed in the later periods under the accrual basis accounting. Barton and Simko (2002) provide an evidence that firms with a higher current net operating assets (NOA) have greater past earnings manipulations. The reason is that the effects of prior accounting choices is accumulated in the balance sheet. Therefore, we control for systematic variation in estimating accrual-based earnings management by using industry-adjusted net operating asset (ano ), where NOA equals total of book value of equity,

long-term and short-term debt, cash and equivalents, all divided by sales. We expect the coefficient of anoa is negative. 4. FINDINGS 4.1. Descriptive statistics and correlations Table 1 provides an overview of CEOs characteristics (Panel A), proxies of earnings management (Panel B), firms characteristics (Panel C) and control variables (Panel D). In Panel A, the descriptive statistics show that the sample has less CEOs with financial expertise than CEOs without financial expertise (medians of ccert and cexp are 0). The statistics also indicate that high reputable CEOs are less than low reputable CEOs (median of cpress is 0 and cpress is 0). In Panel B, the panel reports the descriptive statistics of proxies for earnings management. The results show that, on average, abnormal accruals are negative in all proxies. Finally, firm statistics (Panel C) and control variables (Panel D) are similar to those reported in prior 100research (Goh and Gupta, 2016; Nguyen et al., 2015; Nguyen et al., 2017). [INSERT TABLE 1 ABOUT HERE] Table 2 reports correlations among selected variables. The figures show that the correlations between measures of earnings management and control variables are statistically significant. Also, correlation coefficients among control variables are small or insignificant. [INSERT TABLE 2 ABOUT HERE] 4.2. Univariate tests

Table 3 reports mean differences in abnormal accruals for each measure of financial expertise of CEO characteristics of MA firms (135 observations). Panel A and B report the mean differences in earnings management between stock acquirers which have financial expert CEOs and stock acquirers which do not have financial expert CEOs. In Panel A, Abnormal accruals of acquirers of the full sample (135 observations) are significant at 10% level. The abnormal accruals of acquirers which have CEOs without financial experience (cexp = 0) are significantly positive, while abnormal accruals of acquirers which have CEOs with financial experience (cexp = 1) are insignificantly mixed. The results also show that acquirers which have CEOs without financial experience inflate earnings higher than those with have CEOs with financial experience do, but insignificant. In the Panel B, the abnormal accruals of acquirers which have CEOs without financial certificate (ccert = 0) are significantly positive, while abnormal accruals of acquirers which have CEOs with financial certificate (ccert= 1) are insignificantly negative. The results also show that acquirers which have CEOs without financial certificate significantly inflate earnings higher than those with have CEOs with financial certificate do at 5% and 10% level. The results suggest that CEOs without financial expertise inflate more earnings than CEOs with financial expertise do. Panel C and Panel D of Table 3 report the mean differences in earnings management between stock acquirers which have high reputation CEOs and stock acquirers which have low reputation CEOs. In Panel C, abnormal accruals of acquirers which have CEOs with low press coverage (cpress = 0) are significantly positive, while abnormal accruals of acquirers which have CEOs with high press coverage (cpress = 1) are insignificantly negative. The abnormal accruals of acquirers which have CEOs with low press coverage (cpress = 0) is also significantly higher than those of acquirers

which have CEOs with high press coverage (cpress = 1). In Panel D, abnormal accruals of acquirers which have CEOs with low performance (croa = 0) are significantly positive, while abnormal accruals of acquirers which have CEOs with high performance (croa = 1) are insignificantly negative. The abnormal accruals of acquirers which have CEOs with low performance (croa = 0) is also significantly higher than those of acquirers which have CEOs with high performance (croa = 1). The results suggest that CEOs with low reputation inflate more earnings than CEOs with high reputation do. [INSERT TABLE 3 BOUT HERE] 4.3. Main regression results 4.3.1. Earnings management prior to mergers and acquisitions Table 4 and 5 present the accruals earnings management of share-financed acquirers prior to a MA. The table 4 shows that the coefficients of MA i,t are positive (0.028 when dependent variable is ATA_JM i,t 1 ; and 0.031 when dependent variable is ATA_MJM i,t 1 ) and both are significant at the 5% level in Year t-1. The results mean that abnormal total accruals of share-financed acquirers are higher the rest of the sample by 2.8% for using Jones model and 3.1% for using modified Jones model. Thus, the evidence supports hypothesis H1. In Year t-2, the coefficients of MA i,t are positive (0.003 when dependent variable is ATA_JM i,t 2 ; and 0.007 when dependent variable is ATA_MJM i,t 2 ), but insignificant. In year t-3, the coefficients of MA i,t are positive (0.003 when dependent variable is ATA_JM i,t 3 ; and 0.003 when dependent variable is ATA_MJM i,t 3 ), but insignificant. These results

implicate that share-financed acquiring firms manage total accrual upward in the first year prior to MA after controlling a number of control variables. [INSERT TABLE 4 ABOUT HERE] Table 5 presents the coefficients of MA i,t are significant positive (0.023 when dependent variable is AWCA_JM i,t 1 and significant at the 10% level; and 0.025 when dependent variable is AWCA_MJM i,t 1 ) and significant at the 5% level in Year t-1, while the coefficients of MA i,t are mixed (-0.001 when dependent variable is AWCA_JM i,t 2 ; and 0.004 when dependent variable is AWCA_MJM i,t 2 ), but insignificant in Year t-2. In Year t-3, the coefficients of MA i,t are negative (-0.001 when dependent variable is AWCA_JM i,t 3 ; and - 0.002 when dependent variable is AWCA_MJM i,t 3 ), but insignificant. Thus, the evidence supports hypothesis H1. These results implicate that sharefinanced acquiring firms manage accrual upward in the first year prior to MA. The results are consistent in four alternative accruals earnings management proxies. [INSERT TABLE 5 ABOUT HERE] 4.3.2. Financial expertise of CEOs and earnings management in stock acquirers Table 6 presents the results of main regression (5), where cexp is used in the model for year t-1 (one year prior to MA deals) and where different measures of earnings management are used as substitutes in the model.

Column (2) and (3) of Table 6 report results of regressions where abnormal accruals are used (ATA_JM and ATA_MJM). The evidence shows that the coefficients of MA cexp are negative and statistically significant at 5% level (coefficients of ATA_JM and ATA_MJM are -0.065 and -0.067, respectively). Similarly, in Column (4) and (5) report findings of regressions where abnormal working capital accruals are used (AWCA_JM and AWCA_MJM), the coefficients of MA cexp are also negative and statistically significant at 10% level (coefficients of AWCA_JM and AWCA_MJM are -0.046 and - 0.048, respectively). In general, the results indicate the absence of financial expert CEOs (cexp = 0) in stock acquirers is associated with an increase in abnormal accruals in year t-1 (one year prior to MA deals) and this increase is statistically significant. In other words, the presence of financial expert CEOs (cexp = 1) is correlated with a reduction in abnormal accruals in year t-1. This evidence is consistent with the hypothesis 2, and is in the line with the literature that CEO characteristics matter in the context of MA (Custódio and Metzger, 2013; Grinstein and Hribar, 2004; Malmendier and Tate, 2008; Walters et al., 2007). [INSERT TABLE 6 ABOUT HERE] Table 7 presents the results of main regression (5), where ccert is used in the model, and abnormal total accruals (Column (2) and (3)) and abnormal working capital accruals (Column (4) and (5)) are used. We find that the coefficients of MA ccert are negative and statistically significant at 5% and 10% level (coefficients of ATA_JM and ATA_MJM are -0.059 and -0.056, respectively). Similarly, in Column (4) and (5) report findings of regressions where abnormal working capital accruals are used (AWCA_JM and AWCA_MJM), the coefficient of MA ccert is also negative and statistically significant at 10% level (coefficients of AWCA_JM and -0.046) and

insignificant (coefficients of AWCA_MJM -0.048). In general, the results indicate the absence of financial expert CEOs (ccert = 0) in stock acquirers is associated with an increase in abnormal accruals in year t-1 (one year prior to MA deals) and this increase is statistically significant. [INSERT TABLE 7 ABOUT HERE] 4.3.3. Reputation of CEOs and earnings management in stock acquirers Table 8 presents the results of main regression (5), where croa is used in the model. We find that the coefficients of MA croa are negative and statistically at 5% level when abnormal accrual are estimated under the original Jones model (coefficients of ATA_JM and AWCA_JM are -0.065 and -0.067, respectively). Also, the coefficients of MA croa are also negative and statistically significant at 10% level when abnormal accruals are estimated under the modified-jones model (coefficients of ATA_MJM and AWCA_MJM are -0.046 and -0.048, respectively). The results are in the line with our hypothesis H3 that CEOs with high reputation (croa=1) are less likely to be involved in earnings management in one year prior to MA deals. [INSERT TABLE 8 ABOUT HERE] Table 9 presents the results of main regression (5), where cpress is used in the model. We find that the coefficients of MA cpress are negative and statistically at 1% level when abnormal accrual are estimated under the original Jones model (coefficients of ATA_JM and AWCA_JM are -0.085 and -0.089, respectively). Also, the coefficients of MA croa are also negative and statistically significant at 5% level when abnormal accruals are estimated under the modified-jones model (coefficients of ATA_MJM are -0.082) and insignificant when abnormal accruals are estimated under the modified-

Jones model (coefficients of AWCA_MJM are -0.083). The results are in the line with our hypothesis H3 that CEOs with high reputation (cpress=1) are less likely to be involved in earnings management in one year prior to MA deals. [INSERT TABLE 9 ABOUT HERE] 5. CONCLUSIONS In this study, we hypothesize that CEO characteristics affect earnings management in stock acquirers prior to MA deals. Using a sample of 7,197 firm-year observations of UK listed companies from 2007 to 2012 (135 observations with share deals and 7,062 firm-year non-share deal observations), we find consistent evidence that, in one year prior to deal announcement, financial expertise and reputation of CEOs are associated with a reduction in abnormal accruals, which is used as a proxy of earnings management. The correlations are statistically significant. The findings are robust as we measure abnormal accruals in various ways. In general, the evidence contributes to the existing literature by providing new evidence on the relationship of CEO characteristics and earnings management in MA deals. The findings have some implications for practitioners.

Appendix: Variable definitions ATA_JM i,t = TA i,t [α + β 1( 1 ) + β 2( REV i,t ) + β 3 ( PPE i,t )], is abnormal total accruals estimated by the original Jones model of firm i in year t. α, β 1, β 2, and β 3 are the estimated coefficients from the following regression, which is run in each industry-year with at least six observations: TA i,t = α + β 1 ( 1 ) + β 2 ( REV i,t ) + β 3 ( PPE i,t ) + ε i,t, where TA i,t is total accruals (net income minus cash flows from operation), is total assets of firm i at the end of year t 1; REV i,t is the changes in sales from year t 1 to year t of firm i; and PPE i,t is gross plant, property and equipment of firm i at the end of year t. ATA_MJM i,t = TA i,t [α + β 1( 1 ) + β 2( REV i,t REC i,t ) + A β 3 ( PPE i,t )], is abnormal total i,t 1 accruals estimated by the modified-jones model of firm i in year t; REC i,t is changes in receivables; α, β 1, β 2 and β 3 are the estimated coefficients from the following regression, which is run in each industry-year with at least six observations: TA i,t = α + β 1 ( 1 ) + β 2 ( REV i,t ) + β 3 ( PPE i,t ) + ε i,t AWCA_JM i,t = WCA i,t [α + β 1( 1 ) + β 2( REV i,t ) + β 3 ( PPE i,t )], is abnormal working capital accruals estimated by the Jones model of firm i in year t. α, β 1, β 2 and β 3 are the estimated coefficients from the following regression, which is run in each industry-year with at least six observations: WCA i,t = α + β 1 ( 1 ) + β 2 ( REV i,t ) + β 3 ( PPE i,t ) + ε i,t, where WCA i,t is working capital accruals, which is equals to the difference between net income before extraordinary items (NI i,t ) as reported in the cash flow statement and operating cash flow which excludes depreciation and amortisation (CFO i,t - D&A i,t ).

AWCA_MJM i,t = WCA i,t [α + β 1( 1 ) + β 2( REV i,t REC i,t ) + A β 3 ( PPE i,t )], is abnormal i,t 1 working capital accruals estimated by the modified-jones model of firm i in year t. α, β 1, β 2 and β 3 are the estimated coefficients from the following regression, which is run in each industry-year with at least six observations: WCA i,t = α + β 1 ( 1 ) + β 2 ( REV i,t ) + β 3 ( PPE i,t ) + ε i,t. ccert is a dummy variable with the value of one if a CEO does not has a master of business administration or a chartered accountant certification accredited by the Financial Reporting Council (Financial Reporting Council, 2016) or equivalent, zero otherwise. cexp is a dummy variable with the value of one if a CEO has not worked as a chief financial officer in the past, zero otherwise. cpress is a dummy variable with the value of one if the number of news covering CEO name and his company in year t is higher than zero, zero otherwise. For collecting cpress (press coverage) of a CEO in a year, we conduct a search using CEO s name and company s name on business news from the LexisNexis database. Following Francis et al. (2008), We restrict our search to UK national newspapers within the three-year period prior to and including the considered fiscal year. croa is a dummy variable with the value of one if the average ROA during the last three years is higher than median average ROA during the last three years of the sample, zero otherwise.

asize i,t 1 is industry-adjusted size, which is the difference of the actual size and the mean size of the corresponding industry, where firm s size is defined as the market value of equity at the end of fiscal year. amtb i,t 1 is industry-adjusted MTB, which is the difference of the actual MTB and the mean MTB of the corresponding industry, where MTB is market-to-book ratio which is defined as market value of equity at the end of fiscal year divided by book value of equity. alev i,t 1 is industry-adjusted debt ratio, which is the difference of the actual Leverage and the mean Leverage of the corresponding industry, where Leverage is defined as total of long-term and short-term debts divided by total assets. ano is industry-adjusted net operating assets ratio, which is the difference of the actual NOA and the mean NOA of the corresponding industry, where NOA is defined as total of book value of equity, long-term and short-term debts, cash and equivalents, all divided by sales. SEO i,t equals one if the firm engaged in a seasoned equity offering in the fiscal year, and zero otherwise. A seasoned equity is identified when (1) number of common shares outstanding increases by more than 5%, and (2) proceeds from issuing stocks are positive.

Table 1: Summary descriptive statistics: STANDARD STATISTICS N MEAN DEVIATION Panel A: Summary statistics for CEO's characteristics MIN 25 TH PERCENTILE MEDIAN 75 TH PERCENTILE cexp i,t-1 7197 0.186 0.389 0 0 0 1 1 ccert i,t-1 7197 0.237 0.425 0 0 0 1 1 cpress i,t-1 7197 0.164 0.371 0 0 0 1 1 croa i,t-1 7197 0.020 0.138-0.655 0 0.125 0.365 0.647 Panel B: Earnings management proxies ATA_JM i,t-1 7197-0.010 0.133-0.527-0.059-0.005 0.049 0.405 AWCA_JM i,t-1 7197-0.017 0.146-0.567-0.076-0.010 0.051 0.422 ATA_MJM i,t-1 7197-0.008 0.104-0.374-0.051-0.006 0.036 0.359 AWCA_MJM i,t-1 7197-0.001 0.097-0.349-0.039 0.000 0.037 0.335 Panel C: Firm characteristics AT i,t-1 7197 4,028,305 13,903,644 525 14,378 77,100 812,006 97,812,000 IB i,t-1 7197 264,203 1,080,662-446,000-1,545 1,172 31,012 8,007,866 MACAP i,t-1 7197 5,449,167 21,149,732 771 13,549 71,680 786,637 165,126,390 SALE i,t-1 7197 3,069,966 10,147,448 0 5,921 55,900 713,356 66,216,000 Panel D: Control variables SEO i,t-1 7197 0.291 0.454 0 0 0 1 1 alev i,t-1 7197-0.029 0.251-1.592-0.130-0.051 0.079 0.780 amtb i,t-1 7197 0.024 6.324-28.196-1.601-0.334 1.248 29.577 anoa i,t-1 7197-6.434 38.527-161.192-4.020-0.612 0.035 197.612 asize i,t-1 7197 0.156 2.627-4.825-1.817-0.156 1.804 6.705 Note: The table reports statistics of CEO characteristics variables, earnings management proxies and control variables for the UK sample from 2007 and 2012. Definitions of variables are in the Appendix. MAX

Table 2. Pearson Correlations Variable (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13) (1) ATA_JM i,t-1 1.000 (2) AWCA_JM i,t-1 0.975 1.000 (3) ATA_MJM i,t-1 0.980 0.955 1.000 (4) AWCA_MJM i,t-1 0.955 0.981 0.974 1.000 (5) cexp i,t-1 0.033 0.034 0.036 0.039 1.000 (6) ccert i,t-1 0.023 0.024 0.026 0.028 0.483 1.000 (7) cpress i,t 0.000-0.004 0.003 0.000 0.367 0.463 1.000 (8) croa i,t-1 0.148 0.138 0.153 0.145 0.161 0.130 0.143 1.000 (9) SEO i,t-1 0.045 0.045 0.035 0.040 0.021-0.024 0.059 0.101 1.000 (10) alev i,t-1-0.032-0.033-0.032-0.034-0.050-0.037 0.009-0.163-0.017 1.000 (11) amtb i,t-1-0.018-0.025-0.012-0.020 0.007 0.016 0.033-0.044 0.039-0.003 1.000 (12) anoa i,t-1-0.034-0.025-0.033-0.024-0.021 0.004-0.002-0.045-0.032-0.126 0.004 1.000 (13) asize i,t-1-0.005-0.017-0.001-0.013-0.097-0.152 0.053 0.129-0.144 0.063 0.051-0.083 1.000 Note: This table reports pooled Pearson correlations for the entire sample of 7,197 firm-years over the period 2007-2012. The values reported in italic indicate the corresponding coefficients are significant at 10% level. Please see Appendix for variable descriptions.

Table 3: T statistics Panel A: Earnings management of CEOs' experience Total sample cexp = 1 cexp = 0 Different Mean (1) (2) (3) (4) (5) ATA_JM i,t-1 0.024 * -0.006 0.030 ** 0.037 AWCA_JM i,t-1 0.027 * 0.005 0.032 ** 0.027 ATA_MJM i,t-1 0.019-0.018 0.027 * 0.045 AWCA_MJM i,t-1 0.022 * -0.005 0.028 ** 0.033 Observations 135 21 114 Panel B: Earnings management of CEOs' certificate Total sample ccert = 1 ccert = 0 Different Mean ATA_JM i,t-1 0.024 * -0.017 0.039 *** 0.056 ** AWCA_JM i,t-1 0.027 * -0.013 0.042 *** 0.055 * ATA_MJM i,t-1 0.019-0.019 0.033 ** 0.052 ** AWCA_MJM i,t-1 0.022 * -0.012 0.035 ** 0.047 * Observations 135 33 102 Panel C: Earnings management of CEOs' press coverage Total sample cpress = 1 cpress = 0 Different Mean ATA_JM i,t-1 0.024 * -0.038 0.041 *** 0.079 *** AWCA_JM i,t-1 0.027 * -0.034 0.044 *** 0.079 ** ATA_MJM i,t-1 0.019-0.041 0.035 ** 0.076 *** AWCA_MJM i,t-1 0.022 * -0.035 0.038 ** 0.073 ** Observations 135 26 109 Panel D: Earnings management of CEOs' performance Total sample croa = 1 croa = 0 Different Mean ATA_JM i,t-1 0.024 * -0.024 0.029 ** 0.053 ** AWCA_JM i,t-1 0.027 * -0.013 0.032 ** 0.045 ** ATA_MJM i,t-1 0.019-0.048 0.026 * 0.075 * AWCA_MJM i,t-1 0.022 * -0.035 0.029 ** 0.064 ** Observations 135 12 109 Note: The table reports mean of earnings management proxies of acquirers with different CEOs characteristics for the 135 MA observations from 2007 and 2012. Columns 2 shows the mean of abnormal accruals of acquirers of total sample. Columns 3 to 5 in Panel A, B, C and D shows the mean of high cexp and low cexp, high ccert and low ccert, high cpress and low cpress and high croa and low croa of share financed acquiring firms in year t-1. Under difference, standard t- statistics from t-test for the differences in means. An ***, **, and * indicate 1%, 5%, and 10% level of significance respectively. Please see Appendix A for variable descriptions.