Does founding family control affect earnings management?

Abstract Because of concentrated ownership stakes, board composition and longer-investment horizons, founding-family controlled firms provide an interesting setting for examining issues relating to governance and control. Anderson and Reeb (Citation2003a, Citationb, Citation2004), find that the founding-family controlled structure results in superior stock market and accounting performance and lower cost of debt compared to their nonfamily controlled counterparts. We add to their findings by examining the relationships between founding family control and earnings management. The unique characteristics of family controlled firms could insulate these firms from pressures to manage earnings. Our results support this notion, and find that family firms are significantly less likely to manage earnings. Notes 1 This is analogous to the literature on the board of directors. There is mixed evidence on the association between boards and firm performance. Thus, several studies focus on specific situations rather than on the overall firm performance. For examples, research has been conducted on the impact of boards on CEO turnover (Weisbach, Citation1988; Perry, Citation2000), on earnings management (Xie et al., 2002) and on executive compensation (Hallock, Citation1997; Core et al., Citation1999). 2 To the extent that shareholders make poor decisions based on managed accounting numbers, earnings management can be classified as an agency cost (Davidson et al., Citation2004). 3 This proportion of family-owned firms is remarkably consistent with Anderson and Reeb (Citation2003a,Citationb; Citation2004). They find that one-third of their sample is family owned. 4 Anderson and Reeb (Citation2003a,Citationb, Citation2004) make their classification of family firms available to the public. More detailed discussions about the classification can be found in their studies. 5 For instance, the Ablon family is perceived as controlling the Ogden Corporation as if they were majority owners but, in actuality, they hold merely 2% of the outstanding shares. On the contrary, at Nordstrom's, the family holds 24% of the equity to maintain control. 6 We employ the modified Jones (Citation1991) model because the model has been found to have 'the most power in detecting earnings management' (Dechow et al., Citation1995). In addition, Guay et al. (Citation1996) state that 'only the Jones and modified Jones models appear to have the potential to provide reliable estimates of discretionary accruals.' Moreover, Bartov et al. (Citation2001) find that only the Jones and modified Jones model are consistently able to detect earnings management. 7 This approach has been used in a large number of studies (Teoh et al., Citation1998a, Citationb; Davidson et al., Citation2003; Xie et al. Citation2003). 8 Other studies have used the same measure as a proxy for the combined effect of income-increasing and income-decreasing earnings management (Warfield et al., Citation1995; Becker et al., Citation1998; Bartov et al. Citation2001; Klein, Citation2002). 9 In a separate regression (data not shown), the raw DCA (not the absolute value) is used. The family firm dummy does not produce a significant coefficient. 10 As in Anderson and Reeb (Citation2003a,Citationb, Citation2004), the magnitude of the reduction in earnings management is computed as the coefficient of the family dummy divided by the average | DCA | for the sample = 0.017/0.047 = 0.36. 11 Although the R 2s appear to be low, this should not be particularly problematic for Lev (Citation1989) finds that a majority of earnings studies exhibit low R 2s. The R 2s in this study are comparable to those in other earnings studies. We also make sure that multi-collinearity is not a problem. We examine the variance inflation factors (VIF) and find that none of them are higher than 2. 12 The fixed-effect approach helps reduce the possibility that the documented relation is caused by unobservable omitted variables, i.e. the fact that the firm is family owned and the extent of earnings management may be driven by some unobservable variables that are not included in the model. The fixed-effects analysis has been found to resolve this potential problem.