ACCOUNTING AND REAL EARNINGS MANAGEMENT IN FAMILY FIRMS

Transcription

1 ACCOUNTING AND REAL EARNINGS MANAGEMENT IN FAMILY FIRMS Ann-Kristin Achleitner Center for Entrepreneurial and Financial Studies (CEFS) Technische Universität München Nina Fichtl Center for Entrepreneurial and Financial Studies (CEFS) Technische Universität München Christoph Kaserer* Center for Entrepreneurial and Financial Studies (CEFS) Technische Universität München *Corresponding Author Center for Entrepreneurial and Financial Studies (CEFS) TUM School of Management Technische Universität München Arcisstr München Tel: + 49 (0) Fax: + 49 (0)

2 Accounting and Real Earnings Management in Family Firms ABSTRACT This paper starts from the presumption that firms follow multiple earnings management strategies simultaneously, i.e. they engage in accounting (AEM) and real earnings management (REM) at the same time. We hypothesize that the outcome of this trade-off is driven by earnings management motives influenced by the ownership structure of the firm and the institutional environment. In order to test this hypothesis, we use a large hand-collected panel data set of listed German firms over the period 1998 to Specifically, we test the hypothesis that firms owned by their founding families should engage more in AEM and less in REM activities as compared to their non-family counterparts. In a nutshell, this is because at the one side control motives are a strong driver for AEM in family firms, whereas at the other side, family shareholders tend to dislike REM because of its negative performance impact. Our results clearly support this hypothesis, even when we control for a potential selection bias. JEL Classification: G32, M4, M41, M48 Keywords: family firms, accounting earnings management, real earnings management, earnings quality, ownership structures, Germany.

3 1. Introduction In this paper we scrutinize the trade-off between accounting earnings management (AEM) and real earnings management (REM). It is agreed in the literature that AEM refers to the way in which the discretion granted by accounting standards is exploited, whereas REM affects the timing and structuring of business activities. Hitherto, the overwhelming part of the literature focuses on drivers of AEM (for surveys cf. e.g. Healy and Wahlen, 1999; Armstrong et al., 2010, Kothari et al., 2010). Some recent papers investigate drivers of REM (e.g. Gunny 2010; Roychowdhury, 2006), whereby only very few papers take an integrated view by looking at both earnings management mechanisms simultaneously (Cohen et al., 2008; Cohen and Zarowin 2010; Zang, 2012). However, such an integrated view is a prerequisite for understanding earnings manipulation, as these papers point out that firms engage in multiple earnings management strategies at the same time. Such an understanding is important for any discussion about accounting choices and enforcement. In this paper we make a contribution to this strand of literature by scrutinizing the impact of ownership structures on both AEM as well as REM. More specifically, we hypothesize that by taking the specific benefits of earnings manipulation as well as the incentive structures governing the firms into account, closely-held firms should be less inclined to use REM, but more inclined to use AEM as compared to widely-held firms. We test this hypothesis by analyzing earnings management in family-owned firms. Our hypothesis is based on two pillars. First, we follow Cohen and Zarowin (2010) and Zang (2012) in pointing out that the choice between AEM and REM is, among others, determined by the specific motivation that is pursued. Such motivation for earnings management can be related to (i) meeting capital market expectations, (ii) influencing 2

4 management compensation, (iii) avoiding the breach of debt covenants or (iv) influencing payout levels. While for motives (iii) and (iv) no direct preference for either AEM or REM arises (as long as the costs of these instruments are not taken into account), a management pursuing motivations (i) or (ii) might prefer REM over AEM. This is because as far as earnings smoothing is concerned, it could be argued that the capital market, e.g. financial analysts, is able to look behind the veil of financial accounting policy. For example, Graham et al. (2005) report that 80% of surveyed managers would decrease discretionary spending on R&D or maintenance in order to meet formerly published earnings targets. Moreover, to the extent that these REM actions have an impact on the stock price, REM is also an instrument to influence stock-based compensation in the short-term. Actually, both family and non-family firms are interested in earnings management. However, the underlying motives are quite different. While in widely-held firms earnings management is mostly driven by the desire to smooth earnings and to influence management compensation, the dominant motive in family-owned firms is to stay in control of the company, i.e. to avoid the breach of covenant clauses (Prencipe et al., 2008), and to pay smooth dividends (Schmid et al., 2010). Hence, from the benefit side familyowned firms are, at best, indifferent between AEM and REM, while non-family firms may have a clear preference for REM. Second, for getting the whole picture about the trade-off between AEM and REM also the costs of these two instruments have to be taken into account (Zang, 2012). While AEM is based on pure accounting choices and, therefore, has no cash flow consequences, this does not hold to be true for REM. By definition, REM implies a deviation from normal business practices with the goal to avoid the reporting of a loss or to smooth earnings. 3

5 Roychowdhury (2006) shows that this effect is obtained by boosting earnings through increased price discounts, overproduction in order to allocate less overhead costs to the cost of goods sold and aggressively reducing discretionary expenses such as R&D, advertising and SG&A expenses. Therefore, it is not surprising, that REM has a negative impact on future performance (Gunny, 2010). As a consequence, the propensity to use REM instead of AEM is related to the degree of interest alignment between shareholders and management. Therefore, from the cost side perspective we would expect closely-held firms, e.g. family firms, to prefer AEM over REM as in these firms a stronger alignment between the interest of shareholders and managers can be assumed. For firms with atomistic shareholder structures this alignment may be less effective so that the negative performance consequences of REM are given a lower weight. The results presented in this paper clearly support our hypothesis that family firms are less inclined to use REM, but more inclined to use AEM as compared to widely-held firms. More specifically, our findings can be summarized as follows. First, while all firms use AEM as well as REM simultaneously, there is strong evidence that family-owned firms use more AEM and less REM compared to their non-family counterparts. Second, this result is not driven by family management, i.e. the fact that members of the founding family hold positions in the board. This is what we would expect according to our theoretical reasoning, as the motives for earnings management are related to the position of the founding family as a blockholder in the company. Third, this reasoning is further supported by the fact that we also find non-family firms with significant insider ownership to rely 4

6 more heavily on AEM. It should be noted that the results are robust against a potential selfselection bias. The paper contributes to the literature along several dimensions. First, while accounting earnings management in family firms has been addressed by previous research (e.g. Wang, 2006; Ali et al,. 2007; Prencipe et al., 2008; Tong, 2007, Jaggi et al., 2009; Bar-Yosef and Prencipe, 2011), this paper to the best of our best knowledge studies the effects of family involvement on the trade-off between AEM and REM for the first time. Second, the paper corroborates former studies indicating that control considerations are an important motive for earnings management in family firms (Prencipe et al., 2008; Bar-Yosef and Prencipe, 2011). In this sense, it raises some doubts against other studies concluding that because of higher interest alignment earnings quality is superior in family firms (Wang, 2006). Without taking into account the trade-off between AEM and REM, it is quite difficult to get a complete picture about earnings management activities in family firms. Fourth, the paper offers a potential explanation why evidence on AEM in family firms hitherto is mixed. While evidence on US and UK firms commonly suggests that family and managerial ownership is negatively associated with accounting earnings management (Warfield et al., 1995; Peasnell et al., 2005; Wang, 2006; Ali et al., 2007; Tong, 2007), results on European and Asian firms indicate that family influence increases the level of accounting earnings management (Gabrielsen et al., 2002; Jara-Bertin and Lopez Iturriaga, 2008; Jaggi et al., 2009). If the trade-off between AEM and REM depends, among others, on ownership structures and the institutional framework, the optimal choice between AEM and REM is likely to differ across countries. 5

7 Third, beyond the family-firm literature, we also add some evidence to the literature on the interaction between AEM and REM, especially because we relate this trade-off to ownership structures. This issue has not yet been scrutinized in the literature (Cohen and Zarowin, 2010; Zang, 2012). The rest of the paper is organized as follows. Section 2 develops the theoretical framework and discusses related literature. Section 3 presents the empirical models, while section 4 reports the sample and descriptive statistics. Results of the analyses are presented in section 5. Section 6 concludes the paper and points out to avenues for future research. 2. Related Literature and Hypotheses Earnings management can be defined as a purposeful intervention in the external financial reporting process, with the intent of obtaining some private gain. (Schipper, 1989: 92). Thereby, earnings management comprises AEM and REM. While the former is the outcome of accounting choices, the latter results from cash flow choices including operating, financing and investment decisions (e.g. Schipper, 1989; Roychowdury, 2006; Dechow and Skinner, 2000). Typical examples for REM activities include the deferral of discretionary expenses such as marketing or R&D expenses (e.g. Baber et al., 1991; Dechow and Sloan, 1991; Holthausen et al., 1995; Bushee, 1998; Bens et al., 2003), the sale of profitable assets or timing of income recognition from the disposal of a long-lived asset (Bartov 1993; Herrmann et al., 2003), inventory management (e.g. Hunt et al., 1996; Thomas and Zhang, 2002) or the cutting of prices to boost sales in the current period and other means to accelerate sales (Jackson and Wilcox, 2000). It is agreed in the theoretical and empirical literature that AEM and REM are complements rather than substitutes (Ewert and Wagenhofer, 2005; Cohen and Zarowin, 2010; Zang, 6

8 2012). Therefore, there is a trade-off between the two instruments the outcome of which depends on the associated benefits and costs. This trade-off most likely depends on the ownership structure of a company. Actually, given that the family is concerned about the longevity of the firm, family firms are presumed to be less subject to managerial myopia and more inclined to invest into positive NPV projects rather than boost current earnings by engaging in risky projects (James, 1999). Hence, there is more interest alignment between the shareholders and the management in a family-owned firm as compared to widely-held firms. The long-term investment horizon of family shareholders reduces the capital market pressure to meet short-term earnings targets such as analyst or management forecasts (Graham et al., 2005). While the alignment of interests is considered to reduce the level of earnings management, family firms could exhibit higher levels of earnings management due to conflicts between family and minority shareholders. Previous studies argue that family firms may mask performance to conceal expropriation of minority shareholders (Leuz et al., 2003; Wang, 2006; Jara-Bertin and Lopez Iturriaga 2008; Jaggi et al., 2009). Moreover, long-term orientation does not only mitigate agency costs arising from short-termism but also results in the wish to maintain a controlling position in the firm and risk aversion. Family firms could have incentives to engage in earnings management in order to conceal true economic performance with the purpose of defending financial and non-financial private benefits of control from owning the company. Two important goals pursued by the family blockholder should be mentioned in this context. First, family firms could make use of accounting policy discretion to meet debtrelated earnings targets, i.e. debt covenants (Prencipe et al., 2008). Second, it should not be 7

9 forgotten that families often have an interest to smooth earnings in order to smooth dividend payouts (Kasanen et al., 1996; García Lara et al., 2005; Schmid et al., 2010). However, these family-specific goals can be pursued by pure AEM. There is not so much need to engage in REM, especially since the latter is associated with negative performance consequences (Gunny, 2010). Indeed, family firms as long-term investors may have lower incentives to sacrifice firm value to meet short-term earnings targets (Hutton, 2007; Salvato and Moores, 2010). As opposed to that, the management in a widely-held firm may engage in REM rather than AEM because the latter is more easily detected by the market. In fact, financial analysts at least to some extent may be able to look behind the veil of financial accounting policy (Graham et al., 2005). As managers are considered to have strong incentives to manage earnings for compensation and reputation purposes (e.g. (Healy, 1985; Holthausen et al., 1995; Bergstresser and Philippon 2006), they would use REM instead of AEM instruments. Because of these interacting effects between AEM and REM, it is not surprising that previous evidence on AEM in family firms is mixed. While evidence on US and UK firms commonly suggests that family and managerial ownership is negatively associated with accounting earnings management (Dhaliwal et al., 1982; Warfield et al., 1995; Peasnell et al., 2005; Wang, 2006; Ali et al., 2007; Tong, 2007), results on European and Asian firms indicate that insider ownership and family influence tend to increase the level of accounting earnings management (Gabrielsen et al. 2002; Jara-Bertin and Lopez Iturriaga, 2008; Jaggi et al., 2009). According to our main hypothesis these differences can be well explained by differences in the institutional framework and in the level of family ownership. The outcome of the trade- 8

10 off between AEM and REM is driven by two mechanisms. First, the level of family ownership is relevant as it determines the degree of interest alignment. As ownership structures in Asian and European countries tend to be more concentrated as compared to the US or the UK we expect, on the margin, AEM to be less attractive relative to REM in the latter countries as compared to the former ones. It should be noted that the level of family ownership in our sample is 36% as compared to an average of 10% in the sample of S&P 500 firms in Wang (2006). Second, investor protection rights should set a limit to the degree of AEM put in place by the management. In fact, as far as AEM is concerned, it is well known that it is confined by the level of investor protection (Ball et al., 2000, 2003; Leuz et al., 2003). Therefore, we can expect that this influences the level of REM. As a consequence, it could be that in countries with higher investor protection rights, the management of widely-held firms has less discretion in implementing AEM making REM relatively more attractive. As an additional aspect, Wang (2006) points out that there could be a non-linear relationship between family ownership and AEM. Incentives to manage earnings in family firms are likely to change with the level of ownership. Previous studies point out that incentives to defend private benefits of control are particularly pronounced at medium levels of managerial or family ownership (Morck et al., 1988; Himmelberg et al., 1999; Anderson and Reeb, 2003). We are able to observe a high heterogeneity regarding the level of family ownership in our sample of German listed firms and test if there is a non-linear relationship between family ownership and earnings management. Family members have different possibilities to influence earnings depending on whether they only act as shareholders or are involved in the board. Germany is characterized by a 9

11 two-tier board structure. Compared to unitary board systems, dual board systems separate distinctly between the management and the supervision of a company. We take this effect into account by analyzing the impact of family ownership and the impact of family management separately. According to our reasoning family ownership should be the driving factor for AEM. However, it may well be that control incentives are more pronounced in those firms where the family is also active in the management. Therefore, it will be interesting to see whether the existence of a family-ceo has an additional impact on AEM activities. To sum up, the main hypotheses tested in our paper are as follows: (i) By taking the specific benefits of earnings manipulation as well as differences in the incentive structures into account, family-owned firms should be less inclined to use REM, but more inclined to use AEM as compared to widely-held firms. This should at least hold to be true in a German corporate governance context with relatively weak investor protection rights. (ii) Earnings management activities in family firms should be driven by family ownership and not by family management. (iii) While we have no clear prediction as far as the impact of the ownership level on the degree of AEM is concerned, we expect REM to be the lower the larger the ownership stake of the family is. (iv) We expect family firms where the founder is appointed as the CEO to engage in more AEM because of the long-term control motive of the founder. 10

12 3. Research Design 3.1 Measurement of Family Involvement The family firm definition used in this study follows the founding family concept used in the literature (Anderson and Reeb, 2003). Accordingly, firms classify as family firms when the founder or a member of his/her family by either blood or marriage serve as directors in the management and/or supervisory board or act as blockholder, either individually or as a group (Villalonga and Amit, 2006). In order to account for the more concentrated ownership, studies related to the German capital market set the threshold for a family blockholding at 25% of voting rights (Andres, 2008; Ampenberger et al., 2012). In this study we follow this approach. It should also be noted that according to the German Stock Corporation Act, important corporate decisions can be blocked with 25% of the voting rights. According to this definition, we use the dichotomous variable Dummy_FF indicating whether a specific firm is a family or non-family firm. However, we also use other proxies measuring family involvement. Family ownership FF_OWN corresponds to the percentage of common shares controlled by the founding family, while FMB is a dummy variable which indicates the presence of the founding family in the management board. Moreover, in order to analyze the role of the CEO, we include the following indicator variables in the regression model: a dummy variable indicating whether the CEO is a founder of the company (F_CEO), a descendant of the founding family (D_CEO), or an outside hired CEO (H_CEO). The presence of a founding family member as a chairman in the supervisory board is indicated by the dummy variable FF_Chair. 11

13 3.2 Model Description Accounting Earnings Management Following Wang (2006), accounting earnings management is estimated based on the nonlinear discretionary accruals model by Ball and Shivakumar (2006). Discretionary accruals are estimated in each year and industry using the following piecewise linear regression: ACC t = α 0 + α 1 CF t + α 2 CF t-1 + α 3 CF t+1 + α 4 DCF t + α 5 DCF t CF t + ε t ACC t is total accruals at t, defined as net income before extraordinary items less operating cash flows at t scaled by average total assets at t, CF t is operating cash flows at t, scaled by average total assets at t, CF t-1 is operating cash flow at t-1, scaled by average total assets at t, CF t+1 is operating cash flow at t+1, scaled by average total assets at t, DCF t is one if the operating cash flow at t is negative, and zero otherwise, DCF t CF t serves as proxy for economic losses, ε t denotes the error term and is presumed to capture the proportion of unexpected or abnormal accruals. The model is estimated in each industry and year based on ICB-industry codes and requires at least 30 observations in each industry-year regression (Ball and Shivakumar, 2006). We use absolute, i.e. unsigned, residuals from the first equation as absolute discretionary accruals (ABS_ACC it ) as dependent variable in the following regression: ABS_ACC it = α 0 + α 1 FAM_PROXY it + α 2 NF_INSIDER it + α 3 HERF it + α 4 SIZE it + α 5 PERF it + α 6 LEV it + α 7 GROWTH it + α 8 AGE it + α 9 LOSS it + α 10 INT_ACC it + ε it Consistent with previous studies on earnings quality in family firms, we use absolute values for discretionary accruals as firms may either engage in income increasing or decreasing earnings management (e.g. Wang, 2006). 12

14 The experimental variable FAM_PROXY denotes the variables that proxy for family involvement as explained above. Following Wang (2006) and other prior literature, the following control variables are included in the model: firm size (SIZE) measured as natural logarithm of total assets, profitability (PERF) measured as net income scaled by average total assets, leverage (LEV) defined as debt to total assets, a dummy variable that indicates if the firm exhibits a loss in the observation period (LOSS), growth (GROWTH) measured as sales growth and age (AGE) measured as observation year less founding year. Our observation period covers the years 1998 to 2008 and is hence characterized by the internationalization of accounting standards in Germany. To control for differences in the accounting standard applied, we include a dummy variable INT_ACC in our analysis that equals one if international accounting standards (IFRS or US GAAP) are applied in the respective observation period and zero if financial statements are prepared according to German GAAP. We differentiate between effects of family ownership and ownership concentration by including the Herfindahl index (HERF) for ownership concentration. Furthermore, nonfamily insider ownership (NF_INSIDER) is added as a control variable to examine if effects are due to insider or family ownership (Wang, 2006) Real Earnings Management Following Roychowdhury (2006), we use discretionary cash flows to analyze effects of family ownership and management on real earnings management. Discretionary cash flows are presumed to result from sales manipulation. Managers may be tempted to temporarily increase sales. Increased sales from price discounts are presumed to vanish when the firm returns to its old prices. As a consequence, the cash inflow per sale 13

15 net of price discount is lower because margins decrease. Roychowdhury (2006) expects that sales management activities are reflected in lower current period cash flows from operations and higher production costs as compared to levels common for the respective sales level. The estimation model for discretionary cash flows is expressed by the following regression equation: CFO t /A t-1 =α 0 + α 1 (1/A t-1 ) + α 2 (S t /A t-1 ) + α 3 (ΔS t /A t-1 ) + ε t All variables used in the model are scaled by lagged total assets (A t-1 ). CFO t is cash flows from operating activities in t, S t is sales in t, ΔS t is change in sales from t-1 to t. The regression is run cross-sectionally for each industry and year with a minimum of 15 observations per industry-year based on ICB codes. The residuals are presumed to capture the amount of discretionary cash flows (Discr_CF). Higher levels of discretionary cash flows are interpreted as lower levels of real earnings management. Effects of family influence on discretionary cash flows are examined based on the following regression model: Discr_CF it = α 0 + α 1 FAM_PROXY it + α 2 NF_INSIDER it + α 3 HERF it + α 4 SIZE it + α 5 PERF i + α 6 LEV it + α 7 GROWTH it + α 8 AGE it + α 9 LOSS it + α 10 INT_ACC it + ε it Experimental and control variables in the models on real earnings management are as defined in our models on accounting earnings management. All models are calculated based on pooled OLS and between effects models. 14

16 4. Data and Descriptive Statistics 4.1 Sample Selection Our sample covers eleven years of panel data for the period from 1998 to We start our sample selection process by identifying all German corporations whose common stock is listed in the CDAX in the respective year. In this way, we get 7,642 firm-year observations. The CDAX corresponds to the market segment that comprises the EU regulated market of the German stock exchange. We eliminate observations of financial firms based on the ICB industry classification in Thomson Financial Datastream. Observations are only included if full information regarding the founder, ownership structures and/or board structures as well as the accounting standard applied in the respective year is available. Outliers are eliminated based on the top and bottom 1%- percentile. Based on this sampling procedure, we include 4,937 firm-year observations of non-financial firms (708 cross-sections) in our analysis (2,335 observations for family firms and 2,602 observations for non-family firms). Data on ownership and board structures is derived from Hoppenstedt Aktienführer which publishes annual data on ownership structures of listed German firms. This information is verified using several further databases including Bureau van Dijk s Amadeus database, Commerzbank s Wer gehört zu wem, the director dealings database of the Bundesanstalt für Finanzdienstleistungsaufsicht (BaFin) and web research. Data on the founder is derived from Hoover s Online Profile, Factiva, LexisNexis as well as web-based research. Accounting data comes from the Worldscope database. Data on accounting standards is verified by hand using annual reports of the respective observation year. 15

17 4.2 Descriptive Statistics of Family and Non-Family Firms Table 1 shows the distribution of family to non-family firms in the sample. Insert Table 1 around here Family firms constitute almost half of the firms in the sample (47.3%). The average percentage of common stock held by founding families is about 35.8%. Descriptive statistics suggest that more than half of the family firms are managed by a founder CEO (F_CEO), while descendants of the founder only serve as CEOs in around 8.2% of the firms (D_CEO). More than a third (35.7 %) of the family firms is managed by an outside hired CEO (H_CEO). In only 15.1% of the family firms, family members hold the position of the chairman in the supervisory board (FF_Chair). It might be interesting to note that in more than 80% of the family firms, founding families hold an ownership stake and are involved in the management or supervisory board. Thereof, family firms characterized by family ownership and management contribute to around 60% of the family firms in the sample, whereas family firms characterized by family ownership and board membership in the supervisory board only contribute to around 20% of family firms in the sample. Insert Table 2 around here Summary statistics for family and non-family firms are reported in Table 2. Average absolute discretionary accruals (ABS_ACC) are for family firms (median 0.053) and for non-family firms (median 0.035). This difference is highly significant giving us a first indication that German family-owned firms exhibit higher levels of AEM than their non-family counterparts. In comparison, average discretionary cash flows (DISCR_CF) are for family firms (median 0.000) and for non-family firms (median 0.007). 16

18 Even though this difference is only weakly significant, it nevertheless is in line with our hypothesis that non-family firms engage more actively in REM than family firms. It should be noted that the level of REM decreases in discretionary cash flows. Results emerging from Table 2 indicate that family firms tend to be significantly smaller in terms of total assets (SIZE). They exhibit lower return on assets (PERF) and leverage (LEV). Family firms appear to have higher sales growth (GROWTH) than non-family firms. Consistent with lower ownership concentration in family firms than in non-family firms (HERF), untabulated statistics indicate that large block holdings of non-family insiders, financial investors or other shareholder groups can be found more frequently in non-family than in family firms. This evidence indicates that founding families tend to be the dominant shareholder in listed family firms in Germany. Descriptive statistics on the industry distribution are displayed in Table 3 and show that family firms are represented in all industries among German listed firms but are particularly concentrated in the health care, the consumer services, the telecommunication and the technology industry. Insert Table 3 around here Pearson correlation coefficients between the variables used in the analysis are displayed in Table 4. Abnormal accruals (ABS_ACC) are significantly negatively correlated with discretionary cash flows (DISCR_CF). This relation can possibly be explained by the fact that firms engage in accounting and real earnings management at the same time, i.e. they use multiple earnings management strategies. This is consistent with the results in Roychowdhury (2006). Insert Table 4 around here 17

19 5. Results 5.1 Family Firms and Accounting Earnings Management Table 5 shows the results from the pooled ordinary least squares (OLS) regression on accounting earnings management with absolute abnormal discretionary accruals (ABS_ACC) as dependent variable. Results from the pooled OLS regression are consistent with results from between effects models which are therefore not tabulated. Insert Table 5 around here First of all, all model specifications indicate that family firms have significantly higher levels of discretionary accruals than non-family firms, regardless how family involvement is measured. Moreover, model 3 suggests that family ownership not family management is the main driver of AEM in family firms. The coefficient on family ownership (FF_OWN) is much more relevant than the coefficient on family board management (FMB) from a statistical as well as an economic point of view. This corroborates our hypothesis that the private interests of the founding family are driving the AEM activities. In line with this result, we find in model 4 that AEM is particularly prevalent in family firms where either the founder or an external manager is appointed as the CEO (F_CEO, H_CEO). As far as the founding CEO is concerned, this is not surprising as the motives for earnings management, especially to stay in control of the firm, may be strongly developed also in this case. Moreover, the founding CEO may also be driven by socio-emotional factors arising from being in control of the firm (Stockmans et al., 2010). What is probably more intriguing is the fact that also externally hired CEOs engage strongly in AEM, if they are hired by a family firm. However, our primary hypothesis also delivers an explanation for that. The externally hired CEO, similarly to the CEO in a 18

20 widely-held firm, has an incentive to influence his compensation by earnings management activities. However, as opposed to the widely-held firm, where REM may be a preferred instrument for doing that, in the family-owned firm there may be strong limits set to such REM activities because of the tight control exercised by the family. Therefore, the externally hired CEO in a family firm has a different trade-off rate between AEM and REM as the CEO in a widely-held firm leading him to engage more in AEM. As an additional explanation one should take into account that variable compensation in family firms tends to be rather accounting than stock based as compared to non-family firms (Cheng and Firth, 2006; Young and Tsai, 2008; Gomez-Mejia et al., 2003). Therefore, there is an additional incentive for the external manager to prefer AEM over REM. The coefficient on ownership concentration (HERF) is negative and significant in models 2 and 3. This result may indicate that earnings management on average decreases with increasing ownership concentration. One explanation might be that the existence of other large shareholders sets a limit to the AEM activities, which is in line with our reasoning that the degree of earnings management is confined by the degree of corporate control. To assess the robustness of our results on AEM, we use the model applied by DeFond and Jiambalvo (1994), Kasznik (1999) and Cohen and Zarowin (2010) among others to estimate discretionary accruals. Since this alternative measure produces consistent results with the Ball and Shivakumar (2006) model, results are not tabulated. 5.2 Family Firms and Real Earnings Management Consistent with the analysis on accounting earnings management, four models are examined to analyze effects of founding family involvement on real earnings management based on pooled OLS and between effects models. The results in Table 6 are from the 19

21 pooled OLS regressions. Results are largely consistent with results derived from the between effects models which are therefore not tabulated. Insert Table 6 around here According to the results presented in Table 6 we find some indication that REM is less pronounced in family firms. Regardless of the model specification, coefficients are positive indicating that there is a lower level of REM in family-owned firms. However, one has to admit that from a statistical point of view the evidence is rather weak, as only in three models the coefficients are statistically significant. Two of these three significant coefficients measure family ownership, which corroborates our reasoning that family ownership has a dampening impact on REM. The third coefficient relates to the appointment of an externally hired CEO. As already explained in the preceding section, we expect that such an externally hired CEO should have an incentive to engage in AEM rather than in REM. However, by splitting up the impact into family ownership and family management variables, we do not get any significant results. As an alternative measure for real earnings management, we use discretionary expenses. Results are consistent with those presented in Table 6 and indicate that REM is significantly lower in family firms. Due to limited space, results are not tabulated. 5.3 Additional Analyses Earnings Management and Self-Selection Bias Two objections against our results could be raised. First, one could argue that AEM and REM should be analyzed in an integrated framework as it has been presumed that firms are using both instruments simultaneously. Second, and more importantly, it may well be that firms self-select into the group of family or non-family firms. As has already been shown 20

22 in Table 3 the industry distribution of family firms is quite different from the distribution of non-family firms. For different industries, however, the earnings management incentives as well as constraints may be totally different. Therefore, it may well be that other firm characteristics are responsible for forming this selection effect, which are much harder to control for. We therefore repeat our analysis by taking into account these two objections. For that purpose we estimate a two-stage model using the Heckman (1979) method. In the first stage, we explain total earnings management (Total_EM), i.e. AEM and REM, by a model that takes economic variables into account that are expected to explain earnings management activities with the exception of ownership variables (Cohen and Zarowin, 2010). In the second stage we explain, conditional on the results in the first stage, the preference for either AEM or REM. Results are presented in Table 7. Insert Table 7 around here From a technical perspective, it should be noted that Total_EM is our measure whether a firm-year observation is classified as an earnings management firm-year observation or not. It is an indicator variable that equals one if either discretionary accruals are above or discretionary cash flows are below the median in the respective year, and zero otherwise. In the first stage, we get the expected result that earnings management activities are driven by profitability (negative impact) and the existence of a loss (positive impact). In the second stage, we see that these variables are still significant. However, on top of that also the ownership structure becomes relevant. In fact, it clearly emerges that AEM is higher in family firms, but also in insider-held firms, while REM is lower in these two types of firms. Moreover, the coefficients are highly significant in all models. This is also true for 21

23 the coefficient on the inverse Mills ratio strongly indicating that there might be a selection bias in the data. To sum up, even by correcting for a potential selection bias and by taking an integrated view on AEM as well as REM, we find results strongly corroborating our hypotheses that family firms engage more in AEM and less in REM as compared to their non-family counterparts Non-linear Effects of Family Ownership on Earnings Management We test for non-linearity between family ownership and earnings management in three ways. Following Wang (2006), we generate three indicator variables, one for family firms without family ownership but presence of family members on one of the boards (No_FAM), one for weak (W_FAM) and one for strong family ownership (S_FAM). The indicator variable on strong family ownership (S_FAM) equals one if the percentage of common stock owned by family members is greater than or equal to the median of family ownership in the respective year (about 30%), and zero otherwise. Accordingly, the dummy variable on weak family ownership (W_FAM) equals one if the percentage of common stock owned by the family is below the median value of family ownership in the respective year, and zero otherwise. In a second model, we add the square of family ownership to the models for accounting earnings management (FF_OWN and FF_OWN²). In a third model, we include dummy variables for family ownership according to the different control thresholds in German listed firms (FF_1 to FF_4): An ownership level between zero and 25% (blocking minority, FF_1), between 25% and 50% (simple majority, FF_2), between 50% and 75% (qualified majority, FF_3) and above the qualified majority of 75% (FF_4). Results are presented in table 8. Again, our results corroborate the hypotheses that AEM activities are driven by family ownership and not by family management. In fact, while the 22

24 coefficients on weak family ownership (W_FAM) and strong family ownership (S_FAM) are positive and statistically significant, we do not find any impact in those firms where families do not have significant ownership stakes but are present in the supervisory or management board (No_FAM). Insert Table 8 around here As far as different levels of ownership are concerned, our hypothesis does not make a clear prediction. Nevertheless, one might expect that the impact on AEM is the larger the larger the ownership stake of the family is. While the result in model 2 is not against this presumption, this is not true for the results of model 3 in Table 8. In fact, according to these results AEM activities do not seem to strongly depend on the ownership stake held by the founding family. This is at least true for ownership stakes of up to 75% of voting rights. Beyond this level the coefficient on family ownership is not significantly different from zero. This could be an indication that incentives to manage accounting earnings are particularly pronounced for medium levels of family ownership. However, it should be taken into account that this result may also be driven by data issues, as the number of family firms in which the family holds more than 75% of the voting rights becomes very small (4.5% of family firms). Insert Table 9 around here As far as REM is concerned, our hypothesis predicts that these activities should be less pronounced the larger the ownership stake of the family is. This is true, at least, if we assume that the cost of REM activities gets more strongly weighted the more this cost is borne by the family. The results presented in Table 9 slightly corroborate this hypothesis. In fact, REM is less pronounced in firms with strong family ownership (S_FAM), whereas 23

25 the coefficients on weak or no family influence are not significantly different from zero according to model 1. The results in model 2, however, are not in accordance with our hypothesis, while the results in model 3 only partially support it Presence of Other Large Shareholders As a final remark, it should be noted that results are robust when variables that can be used as proxy for the presence of other blockholders are included in the models. At first, we add the percentage held by outside blockholders, i.e. shareholders owning more than 5% of shares in the firm. Secondly, we control for other types of large shareholders that may affect earnings characteristics, i.e. banks, strategic shareholders, private equity and institutional investors. Untabulated results suggest that the presence of these types of outside blockholders is not systematically associated with AEM or REM and does not affect our results. 5. Summary and Conclusion When engaging in earnings management firms are faced with a trade-off between AEM and REM. While there are several papers investigating the two mechanisms on an isolated basis, only few studies have taken an integral view. This paper aimed to shed new light on the understanding of this trade-off. Specifically, we have argued that the outcome of this trade-off is driven by the motives driving earnings management, the ownership structure of the firm and the institutional environment. In order to test this hypothesis, we have used a large panel data set of listed German firms over the period 1998 to Our ownership variable of interest was related to the founding family concept. We hypothesized that firms owned by their founding families should engage more in AEM and less in REM activities. In a nutshell, this is 24

26 because at the one side they engage in earnings management not because of capital market pressures, but because of control motives (i.e. avoiding the breach of covenant clauses or influencing the dividends). At the other side, because of their large ownership stakes they dislike REM because of its negative performance impact. Our results corroborate this hypothesis. We find clear evidence that family firms engage more actively in AEM compared to their non-family counterparts, while they engage less in REM. This result is robust against a potential selection bias. In fact, by applying a twostage Heckman selection model we can clearly corroborate these results. Moreover, by splitting-up the family involvement into a family ownership and a family management component, it can be shown that the effect is driven by family ownership. However, with respect to the question whether ownership has a linear or non-linear impact on earnings management activities the findings are rather mixed. Our findings also shed some light on the question why evidence on AEM in family firms presented hitherto was mixed. We argue that the ownership structures as well as the institutional environment, especially the level of investor protection, influence the trade-off between AEM and REM. As this impact is different for family and non-family firms it may well be that the relative level of AEM changes across countries. 25

27 References Ali, A., Chen, T.-Y. and Radhakrishnan, S. (2007) Corporate disclosures by family firms, Journal of Accounting and Economics, 44(1-2), pp Ampenberger, M., Schmid, T., Achleitner, A.-K., Kaserer, C. (2012) Capital structure decisions in family firms: empirical evidence from a bank-based economy, Review of Managerial Sciences, Forthcoming. Anderson, R. C. and Reeb, D. M. (2003) Founding-Family Ownership and Firm Performance: Evidence from the S&P 500, The Journal of Finance, 58(3), pp Andres, C. (2008) Large shareholders and firm performance An empirical examination of founding-family ownership, Journal of Corporate Finance, 14 (4), pp Armstrong, C. S., Guay, W.R. and Weber, J.P. (2010) The Role of Information and Financial Reporting in Corporate Governance and Debt Contracting, Journal of Accounting and Economics, 50(2-3), pp Baber, W. R., Fairfield, P. M. and Haggard, J. A. (1991) The Effect of Concern about Reported Income on Discretionary Spending Decisions: The Case of Research and Development, The Accounting Review, 66(4), pp Ball, R., Kothari, S. P. and Robin, A. (2000) The effect of international institutional factors on properties of accounting earnings, Journal of Accounting and Economics, 29(1), pp Ball, R., Robin, A. and Wu, J. S. (2003) Incentives versus standards: properties of accounting income in four East Asian countries, Journal of Accounting and Economics, 36(1-3), pp Ball, R. and Shivakumar, L. (2006) The Role of Accruals in Asymmetrically Timely Gain and Loss Recognition, Journal of Accounting Research, 44(2), pp Bar-Yosef, S. and Prencipe, A. (2011) Earnings Management and Corporate Governance in Family-Controlled Companies, Journal of Accounting, Auditing and Finance, 26(2), pp

28 Bartov, E. (1993) The Timing of Asset Sales and Earnings Manipulation, The Accounting Review, 68(4), pp Bens, D. A., Nagar, V., Skinner, D. J. and Wong, M. H. F. (2003) Employee stock options, EPS dilution, and stock repurchases, Journal of Accounting and Economics, 36(1-3), pp Bergstresser, D. and Philippon, T. (2006) CEO incentives and earnings management, Journal of Financial Economics, 80(3), pp Bushee, B. (1998) The influence of institutional investors on myopic R&D investment behavior, The Acounting Review, 73(3), pp Cheng, S. and Firth, M. (2006) Family Ownership, Corporate Governance, and Top Executive Compensation, Managerial and Decision Economics, 27(7), pp Cohen, D. A., Dey, A. and Lys, T. Z. (2008) Real and accrual-based earnings management in the pre- and post-sarbanes-oxley periods, The Accounting Review, 83(3), pp Cohen, D.A. and Zarowin, P. (2010) Accrual-based and real earnings management activities around seasoned equity offerings, Journal of Accounting and Economics, 50(1), pp Dechow, P. M. and Skinner, D. J. (2000) Earnings Management: Reconciling the Views of Accounting Academics, Practitioners, and Regulators, Accounting Horizons, 14(2), pp Dechow, P. M. and Sloan, R. (1991) Executive incentives and the horizon problem: an empirical investigation, Journal of Accounting and Economics, 14(1), pp DeFond, M. L. and Jiambalvo, J. (1994) Debt covenant violation and manipulation of accruals, Journal of Accounting and Economics, 17(1-2), pp Dhaliwal, D. S., Salamon, G. L. and Dan Smith, E. (1982) The effect of owner versus management control on the choice of accounting methods, Journal of Accounting and Economics, 4(1), pp

29 Ewert, R. and Wagenhofer, A. (2005) Economic Effects of Tightening Accounting Standards to Restrict Earnings Management, The Accounting Review, 80(4), pp Gabrielsen, G., Gramlich, J. D. and Plenborg, T. (2002) Managerial Ownership, Information Content of Earnings, and Discretionary Accruals in a Non-US Setting, Journal of Business Finance & Accounting, 29(7/8), pp García Lara, J. M., García Osma, B. and Araceli, M. (2005) The Effect of Earnings Management on the Asymmetric Timeliness of Earnings, Journal of Business Finance & Accounting, 32(3/4), pp Gomez-Mejia, L. R., Larraza-Kintana, M. and Makri, M. (2003) The Determinants of Executive Compensation in Family-Controlled Public Corporations, The Academy of Management Journal, 46(2), pp Graham, J. R., Harvey, C. R. and Rajgopal, S. (2005) The economic implications of corporate financial reporting, Journal of Accounting and Economics, 40(1-3), pp Gunny, K. (2010) The Relation between Earnings Management Using Real Activities Manipulation and Future Performance: Evidence from Meeting Earnings Benchmarks', Contemporary Accounting Research, 27 (3), pp Healy, P. M. (1985) The effect of bonus schemes on accounting decisions, Journal of Accounting and Economics, 7(1-3), pp Healy, P. M. and Wahlen, J. M. (1999) A Review of the Earnings Management Literature and Its Implications for Standard Setting, Accounting Horizons, 13(4), pp Herrmann, D., Inoue, T. and Thomas, W. B. (2003) The Sale of Assets to Manage Earnings in Japan, Journal of Accounting Research, 41(1), pp Himmelberg, C. P., Hubbard, R. G. and Palia, D. (1999) Understanding the determinants of managerial ownership and the link between ownership and performance, Journal of Financial Economics, 53(3), pp Holthausen, R. W., Larcker, D. F. and Sloan, R. G. (1995) Annual bonus schemes and the manipulation of earnings, Journal of Accounting and Economics, 19(1), pp