Abstract
Does the presence of multiple and unrelated family controllers improve firm performance? Drawing on both agency and behavioral agency theories, we argue that multifamily firms outperform single-family firms since families in multifamily firms actively monitor owners’ socioemotional goals. Additionally, we suggest that a balanced distribution of control among the owning families facilitates the monitoring process. Finally, we argue that the focal relationship follows an inverted U-shaped pattern depending on the number of families controlling the firm. We test our hypotheses using a sample of Chilean publicly listed family firms. Our study extends current knowledge of the uniqueness of multifamily firms.
The financial performance benefits of family control remains an open question. Some studies have shown a positive effect of family control on firm performance because of the lower agency costs that family firms enjoy (Anderson & Reeb, 2003a, 2003b). However, other studies argue that family control can harm performance given the noneconomically motivated preferences of family owners, which is perceived as wealth expropriation by other investors (Gómez-Mejía, Larraza-Kintana, & Makri, 2003; Schulze, Lubatkin, & Dino, 2003a; Schulze, Lubatkin, Dino, & Buchholtz, 2001). On the other hand, a meta-analysis by O'Boyle, Pollack, and Rutherford (2012) found no relationship between family involvement and firm performance.
A common and potential problematic assumption in the majority of the family firm–firm performance literature is that family control is concentrated in the hands of a single and indivisible family (Bennedsen & Nielsen, 2010; Frantz & Instefjord, 2009; Faccio, Marchica, & Mura, 2011). However, the family ownership structure is more complex than previously assumed. Many family firms possess an ownership structure comprising stakes owned by at least two unrelated owning families that collectively enjoy and share ownership and strategic control over the firm (Jara-Bertin, López-Iturriaga, & López-de-Foronda, 2008; Pieper, Smith, Kudlats, & Astrachan, 2015). Ignoring this complexity may lead to a misunderstanding of the true advantages of family control. Following Pieper et al. (2015), we label family firms controlled by multiple and unrelated owning families as multifamily firms (MFFs), as opposed to family firms controlled by a single owning family (single-family firms; SFFs). In this study, we attempt to refine the extant literature in the area of family firm governance by addressing the following research question: Does the presence of multiple and unrelated family controllers improve firm performance? The study of performance differences between MFFs and SFFs may yield new insights into agency costs experienced in family firms.
We build on agency and behavioral agency theories (Morck, Wolfenzon, & Yeung, 2005; Wiseman & Gómez-Mejía, 1998) to suggest that MFFs outperform SFFs. We theorize that given the higher complexity of MFFs (Pagliarussi & Rapozo, 2011) and diversity of socioemotional interests among unrelated controlling families (Berrone, Cruz, & Gómez-Mejía, 2012), family owners have the ability and incentive to effectively monitor potential misbehavior of other controlling families (Attig, Guedhami, & Mishra, 2008; Gomes & Novaes, 2005). We then continue to theorize that the balance of control among the unrelated controlling families impacts the MFFs’ relative outperformance. Finally, we propose that the relative outperformance of MFFs improves with the increase in the number of owning families controlling the firm until a certain point after which any further increase in the number of families leads to deterioration in performance (inverted U-shaped relationship). We address our research question by examining the entire population of publicly listed family firms in the Chilean stock market.
This study contributes to the family firms and governance literature in two ways. First, while prior research extensively documents that family firms are a unique organizational form (Evert, Martin, McLeod, & Payne, 2016), there is still little knowledge about the performance outcomes of MFFs (Brigham & Payne, 2015). We extend the family firm heterogeneity literature (Chua, Chrisman, Steier, & Rau, 2012; Jaskiewicz & Dyer, 2017) by providing evidence that having multiple and unrelated owning families controlling the family firm fuels firm performance. Second, we extend the literature on control contestability (Boateng & Huang, 2017; Cai, Hillier, & Wang, 2016; Jara-Bertin et al., 2008) by opening the black box of family control. Whereas most of this literature studies the benefits of control contestability between the controller (e.g., family owners) and other large shareholders (e.g., institutional owners), by studying the phenomenon of MFF we focus on the control distribution among owners sharing a similar identity (i.e., families).
Theory and Hypotheses
Family Firm and Firm Performance
The family firm literature offers divergent predictions about the performance consequences of family involvement in the business. Agency theorists suggest that family ownership mitigates principal–agent conflicts (i.e., conflicts of interests between owners and managers). Family owners enjoy more incentives to monitor and discipline nonfamily managers (Anderson & Reeb, 2003a; 2003b) or manage the firm more efficiently (McConaughy, Walker, Henderson, & Mishra, 1998), mainly when family members are part of the top management team. However, other authors argue that family control exacerbates principal–principal agency conflicts (i.e., conflicts of interest between majority and minority owners) with a detrimental impact on family firm performance. They argue that socioemotional priorities (i.e., family-oriented goals) of family owners lead to entrenchment (Gómez-Mejía, Núñez-Nickel, & Gutierrez, 2001; Morck et al., 2005), suboptimal investment decisions (Gómez-Mejía, Makri, Kintana, & Larraza-Kintana, 2010), and different compensation practices (Gómez-Mejía et al., 2003). For example, family socioemotional concerns to maintain family control over the firm’s decisions lead family owners to name family members in top firm positions independently of their professional qualifications (Caselli & Gennaioli, 2013). 1
The lack of theoretical and empirical consensus on the performance consequences of family control (O'Boyle et al., 2012) suggests that such relationship is more complicated than previously thought. As a consequence, scholars have called for research that seeks to understand and take into account the heterogeneity within the group of family firms (Chua et al., 2012; Jaskiewicz & Dyer, 2017). The most studied sources of heterogeneity among family firms relate to differential voting power relative to other nonfamily blockholders (Attig et al., 2008; Fattoum-Guedri, Guedri, & Delmar, 2018), governance mechanisms (e.g., Chrisman, Chua, Le Breton-Miller, Miller, & Steier, 2018), the degree of family participation in firms’ leadership positions (e.g., Evert, Sears, Martin, & Payne, 2018; Nordqvist, Sharma, & Chirico, 2014), specific resources and capabilities that family owners bring to the family firm (e.g., Arregle, Duran, Hitt, & van Essen, 2017; Duran, Kammerlander, van Essen, & Zellweger, 2016), and particular family goals that family owners expect to achieve (e.g., Berrone et al., 2012; Chrisman, Sharma, Steier, & Chua, 2013). More recently, researchers have also begun to distinguish between MFFs and SFFs (e.g., Brigham & Payne, 2015; Pieper et al., 2015) as a new source of heterogeneity among family firms. However, there is still little knowledge about the performance consequences of multifamily businesses (Brigham & Payne, 2015).
MFFs and Firm Performance
It has been argued that MFFs are more complex organizations than SFFs (Pagliarussi & Rapozo, 2011). In SFFs, family leaders exert a strong influence and control over firm decisions through executive position and board participation (Schulze, Lubatkin, & Dino, 2003b). Such decisions can pursue family-centric motives relating to, for example, the long-term preservation of the firm in the family (Berrone et al., 2012). Socioemotional wealth preservation can lead to risk-averse strategic and organizational choices that can affect firm performance negatively (Gómez-Mejía, Cruz, Berrone, & De Castro, 2011; Schulze & Kellermanns, 2015). In MFFs, visions about the future of the firm, the perception of risk and uncertainty, and diverse family socioemotional interests must be continuously negotiated among the unrelated owning families (Brigham & Payne, 2015; Lumpkin & Brigham, 2011). Lack of agreement among unrelated family controllers may jeopardize the well-being of the family firm. Additionally, owning families may have the incentive to form a coalition and act as a single large blockholder, which exacerbates potential expropriation of minority shareholders (Attig et al., 2008; Jara-Bertin et al., 2008). Despite the difficulties involved, compared to SFFs, MFFs have unique characteristics that provide advantages concerning firm performance.
Although owning families in an MFF share a distinctive nature of achieving socioemotional goals, these do not always coincide. Family owners’ socioemotional concerns are numerous, and families’ values and desires evolve as the firm passes through generations (Berrone et al., 2012). Thus, MFFs face the challenge of responding to multiple (and sometimes inconsistent) family expectations among unrelated owning families. The necessity of preserving both families’ affective and economic endowment encourages mutual monitoring among controlling families and between families and managerial agents (Attig et al., 2008; Pieper et al., 2015). Since unrelated family controllers have in common the pursuit of dual objectives, namely, economic and socioemotional, owning families are in a good position to understand and assess the motivation underlying other families’ behavior and decision-making in MFFs. This provides owning families a relatively strong monitoring capacity to effectively oversee other families’ behavior and prevent potential deviation of firm resources and single-family entrenchment (Gomes & Novaes, 2005; Pagano & Roell, 1998). Also, it may encourage open communication and transparent decisions within the MFF (Pieper et al., 2015; Steier, Chrisman, & Chua, 2015). In comparison, nonfamily owners may lack deep understanding of the motives behind family decisions and actions, which reduce their ability to effectively execute their monitoring functions and thus mitigate principal–principal agency costs. Consequently, we expect that MFFs enjoy unique control advantages that allow them to generate superior performance. Hence, we hypothesize the following:
The literature on multiple blockholders (e.g., Andres, 2008; López-de-Foronda, López-Iturriaga, & Santamaría-Mariscal, 2007; Maury & Pajuste, 2005) argues that firm performance depends on whether smaller blockholders monitor or decide to form a controlling coalition with the largest shareholder. Monitoring is more likely to occur when firms are owned by different types of blockholders (e.g., family and institutional; Maury & Pajuste, 2005) or when these blockholders enjoy greater balance of control (i.e., a relatively similar distribution of controlling ownership among blockholders; Fattoum-Guedri et al., 2018; Laeven & Levine, 2008). On the other hand, if blockholders share the same identity (e.g., only families) or the distribution of controlling ownership is unequal, then it is more likely that a coalition will emerge (Laeven & Levine, 2008; Maury & Pajuste, 2005).
A distinct feature in our study is that existing family blockholders explicitly share the controlling ownership stake. Thus, unrelated owning families have already formed a coalition to take the firm control and, therefore, the remaining relevant question is whether they still have the incentives for mutual monitoring. As aforementioned, we suggest that unrelated owning families have in common the pursuit of both economic and socioemotional goals, but the weight that owning families give to each goal may differ (Bertrand, Johnson, Samphantharak, & Schoar, 2008; Eddleston & Kellermanns, 2007; Zellweger & Kammerlander, 2015). Thus, if unrelated families enjoy relatively equal controlling ownership, they have strong incentives to monitor each other to protect their particular interests. Additionally, equal control enhances the capacity of each family to intervene in particular family desires using diverse mechanisms, including private letters to family principals, public shareholder proposals, or even the threat of selling family equity (Edmans, 2014), which can be negatively perceived by external investors (Edmans & Manso, 2011). On the contrary, a higher concentration of shares in one controlling family reduces any bargaining power of other unrelated owning family principals to deter undesired initiatives, which motivates these families to collude with the largest shareholder and obtain control advantages (Jara-Bertin et al., 2008; Maury & Pajuste, 2005). Following this reasoning, we suggest that MFFs with an unbalanced distribution of controlling ownership among unrelated owning families may indeed negatively affect firm performance:
We argue that the beneficial effect of having multiple and unrelated owning families as controllers cannot be reaped above a certain number of families for several reasons. First, a higher number of owning families increases the costs of coordination among them, which reduces the effectiveness of mutual monitoring (López-de-Foronda et al., 2007). This situation causes competition between owning family controllers to tunnel resources out of the firm before other controllers do (Bertrand et al., 2008; Zellweger & Kammerlander, 2015). Second, too many unrelated families simultaneously involved in the firm can reduce the identification of family members with the firm. In other words, family owners experience a weak sense of belonging and personal meaning to the family business (Berrone et al., 2012), which might enhance agency issues in the form of moral hazard, particularly shirking, free riding, and the consumption of perks (Karra, Tracey, & Phillips, 2006; Maug, 1998). Third, family scholars suggest that family governance is associated with quick and efficient decision-making (Zahra, 2003), which provides a key advantage over other firms to capture and develop strategic opportunities. However, MFFs with an excessive number of controlling families may lead to corporate paralysis (Gomes & Novaes, 2005), in which complex decision rules and inefficient bureaucracy for the approbation of strategic project arise. Finally, an increasing number of family heirs exacerbate conflicts about the firm’s future leadership (De Massis, Sieger, Chua, & Vismara, 2016), which could damage the MFF’s performance (Anderson & Reeb, 2003a; Carney, 2005). Consequently, we suggest the following
Research Design
Sample and Data
To test our hypotheses, we built a dataset using the entire population of Chilean publicly listed firms from all market segments, except banks and other financial institutions. There are three reasons for this choice. First, Chile has been categorized as a family-based economy (Steier, 2009; Schneider, 2013), where the majority of Chilean publicly listed firms are controlled by families (Duran, Kostova, & van Essen, 2017). Second, the list of family-controlled firms in Chile includes MFFs and SFFs of all sizes, including family firms classified as global challengers given their rapid global expansion, such as Concha y Toro, Falabella, and LATAM (BCG, 2016). Third, Chilean laws mandate that publicly listed firms report the percentage of shares owned by the ultimate controllers. This allows us to distinguish between MFFs and SFFs clearly.
We relied on four sources to build the dataset: (a) the SVS (Superintendencia de Valores y Seguros; also called the Chilean SEC); (b) the Santiago stock exchange; (c) Economatica, a dataset that covers listed firms in Latin America (Duran et al., 2017); and (d) the Civil Registration and Identification of Chile (a public service that register births, marriages, and deaths in Chile). We relied on Economatica to collect firms’ financial and accounting data. We collected data from 2010 to 2015.
In 2010, the SVS obligated listed firms to disclose controllers’ identities and ownership stakes in firms’ annual reports. From these reports, we identified and classified the ultimate controllers into two categories: family-controlled (either MFFs or SFFs) and nonfamily-controlled firms (i.e., institution-owned, foreign-owned, and widely held firms). Following Duran et al. (2017), we classified a firm as family-controlled if multifamily members (related by blood or marriage)
have direct or indirect participation in the ownership of the firm and they have the power to (1) ensure the majority of votes in shareholder meetings and elect the majority of seats on board of directors or ensure the majority of votes in board meetings, and/or (2) decisively influence the management of the firm. (Duran et al., 2017, p. 479) 2
We captured family ties among individuals based on their official certificates of birth and marriage provided by the Civil Registration and Identification of Chile. These documents allowed us to identify the husband’s, wife’s, and parents’ names and national identification numbers of each individual. To capture sibling, cousin, and other family ties among individuals, we relied on the two-surname system employed in Chile. The person’s first surname concerns his/her father’s first surname. The person’s second surname relates to his/her mother’s first surname. This information, together with birth and marriage certificates, allowed us to trace the person’s family tree and then identify and distinguish owning families controlling the family firm. We found that 65% of the total population of publicly listed firms in Chile have a family or group of unrelated families as the ultimate controlling owner. After removing nonfamily-controlled firms and firms with insufficient financial or ownership information, we obtained a final usable sample of 432 firm-year observations related to 80 family-controlled firms (59 SFFs and 21 MFFs).
Dependent Variable
The dependent variable in our study is return on equity (ROE), which has been commonly used to assess the impact of family ownership on firm performance (e.g., King & Santor, 2008; Martínez, Stöhr, & Quiroga, 2007). For robustness, we also report our analyses using return on assets (ROA; Minichilli, Brogi, & Calabrò, 2016), Tobin’s Q (the ratio of market capitalization plus total debt to total assets; Anderson & Reeb, 2003b; Villalonga & Amit, 2006), and short-term sales growth (the annual growth rate of firm’s total sales; Chrisman, Chua, & Litz, 2004).
Independent Variables
MFF
We measured MFF using a dummy variable with a value of 1 if a family-controlled firm is controlled by a group of unrelated owning families (i.e., owning families that were not related by blood or marriage) and 0 otherwise (i.e., SFF). For robustness, we use MFF board as a more stringent measure of MFF. We measured MFF board as a dichotomous variable with a value of 1 if a family-controlled firm is classified as an MFF and at least two unrelated owning families have family members on the board, and 0 otherwise.
Contestability
To proxy for the balance of control among owning family controllers, we used contestability. This variable was measured as the ratio of the total ownership stake of the nonlargest owning family controller(s) to the ownership stake of the largest owning family controller. Contestability has been widely used to capture the monitoring role and relative power of the nonlargest shareholders (Attig et al., 2008; Boateng & Huang, 2017; Jara-Bertin et al., 2008). For the purpose of this study, the calculation of contestability did not include noncontrolling shareholders and treated related (by blood or marriage) family members as part of the same owning family. For robustness, we also employed the Herfindahl–Hirschman Index (HHI) measured as the sum of squares of the total control stakes of each owning family controller (Jara-Bertin et al., 2008; Maury & Pajuste, 2005) as an alternative of contestability. The greater the HHI value, the greater the lack of contestability of the power of the dominant owning family controller.
Number of families
This variable was computed as the number of unrelated owning families in control of the family firm. Similarly to Fattoum-Guedri et al. (2018), the variable number of families is right skewed: The number of families varies between 1 and 12, with an average of 4.496 families. In 42.0% of the total sample of MFFs, the number of families is greater than the average (ranging between 5 and 12 unrelated families); in 5.0% of the total sample of MFFs, the number of families is greater than seven families (ranging between 7 and 12 unrelated families).
Control Variables
We included eight control variables that are likely to influence firm performance: (a) firm size (natural log of assets); (b) asset growth (the annual growth rate of the firm’s total assets); (c) leverage (ratio of the firm’s long-term debt to total assets); (d) investment (ratio of capital expenditures to total assets); (e) cash ratio (ratio of cash plus cash equivalent to total assets); (f) controlling ownership (the cumulated ownership of the controlling group); (g) noncontrolling blockholders (the cumulated ownership of the noncontrolling shareholders who hold a 10% or greater ownership stake; Attig et al., 2008; Attig, El Ghoul, & Guedhami, 2009; Maury & Pajuste, 2005); and (h) generational stage (the generation that is in control of the management of the family firm; Eddleston, Kellermanns, Floyd, Crittenden, & Crittenden, 2013). Table 1 provides the description and sources of data for all variables in the study.
Definition of Variables.
EBIT, earnings before interest and taxes; ROA, return on assets; ROE, return on equity; SVS, Superintendencia de Valores y Seguros.
Empirical Model
Consistent with previous studies in corporate governance and controlling ownership (Bennedsen & Nielsen, 2010; Faccio et al., 2011; Schmid, 2013), we used a pooled ordinary least square (OLS) regression model to test our hypotheses. Since we suggest that MFFs’ performance depends on the continuous monitoring capacity of firms’ controllers, the independent variables were not lagged (Andres, 2008; Jara-Bertin et al., 2008; Maury & Pajuste, 2005). Additionally, family ownership in Chile is stable over time 3 , which limits the power of fixed-effect models to control for firms’ unobservable heterogeneity (Donelli, Larrain, Francisco Urzúa, & Urzúa, 2013; Li & Prabhala, 2007; Roberts & Whited, 2013). Thus, we followed Gormley and Matsa (2014) by estimating our model including two-digit North American Industry Classification System (NAICS) industry-fixed effects and year-fixed effects. Finally, the standard errors are corrected for clustering at the firm level.
Results
Descriptive Statistics and Correlation Analysis
Table 2 reports the means, standard deviation, and differences in means between MFFs and SFFs. The results indicate that overall MFFs enjoy higher performance than SFFs measured in terms of ROE, ROA, and Tobin’s Q; are slightly larger in terms of assets; and invest slightly more than SFFs. However, MFFs have a lower cash ratio than SFFs. These two types of family firms do not differ significantly in terms of sales growth, asset growth, leverage, level of controlling ownership, the total ownership stake in the hands of noncontrolling blockholders, and the generational stage in control of the firm management. For the case of MFFs, Table 2 shows that, on average, the unrelated owning families collectively control 63.2% of the MFFs’ total shares. Further analyses reveal that each unrelated owning family in MFFs owns, on average, 13.0% of the firm’ total shares. 4 For the case of SFF, on average, the single controlling family owns 65.0% of the firm’s shares. Additionally, Table 2 indicates that the average number of unrelated owning families in MFFs is 4.496 (ranging from 2 to 12 families).
Descriptive Statistics of Multifamily Firms (MFFs) and Single-Family Firms (SFFs).
See Table 1 for variable definitions.
a p < 0.10. *p < .05. **p < .01. ***p < .001.
Table 3 shows descriptive statistics and Pearson’s correlation coefficients between the variables. All financial-based variables were winsorized at the top and bottom 1% to mitigate the outliers’ effect. Correlations between the independent variables are low to moderately high, suggesting the absence of multicollinearity. We confirmed this observation by performing collinearity diagnostics in all models reported in Table 4. The results indicate that multicollinearity is not a problem in our analyses, with variance inflation factors (VIFs) of the examined variables well below 10 (Hair, Anderson, Tatham, & Black, 1998). 5
Descriptive Statistics and Pearson Correlation Matrix.
HHI, Herfindahl–Hirschman Index; MFF, multifamily firm; ROA, return on assets; ROE, return on equity;Tobin's Q, the sum of market capitalization and total debt, divided by total assets.
a p < 0.10. *p < .05. **p < .01. ***p < .001.
Results of Pooled Ordinary Least Square (OLS) Regressions.
Standard errors (in parentheses) are clustered at the firm level; 432 firm–year observations (80 firms). See Table 1 for variable definitions.
a p < 0.10. *p < .05. **p < .01. ***p < .001.
Regression Analysis
Table 4 presents the pooled OLS estimates of the association between MFF, contestability, number of families, and firm performance. Model 1 introduces the control variables. Model 2 tests Hypothesis 1. We find that the coefficient of the relationship between MFF and ROE is positive and statistically significant (β = .061, p < .010), suggesting that the presence of multiple and unrelated families as the firm’s controllers enhances its performance relative to firms controlled by a single family. The magnitude of this effect is not trivial. Based on Model 1, the presence of at least two controlling families corresponds to an increase of 30% of the standard deviation of ROE.
Model 3, Table 4, reports that the association between contestability and ROE is positive and statistically significant (β = .026, p < .001), thus supporting Hypothesis 2. In accordance with prior studies on the effect of noncontrolling blockholders on firm value (e.g., Attig et al., 2009; Cai et al., 2016; Jara-Bertin et al., 2008; Maury & Pajuste, 2005), we find that a higher voting power of the nonlargest family controller(s) relative to the voting power of the largest family controller has a positive effect on firm performance. In terms of economic significance, we find that an increase of the contestability within controlling families of 1.07 (one std. dev.) induces a 14% increase in the standard deviation of ROE.
Hypothesis 3 states an inverse U-shaped relation between the number of unrelated owning families controlling the family firm and firm performance. Model 4, Table 4, shows evidence of a concave relationship between these variables. Specifically, the linear coefficient of the number of controlling families is positive and statistically significant (β = .035, p < .010), while the quadratic coefficient of the number of controlling families is negative and statistically significant (β = −0.003, p < .010), thus suggesting that the positive effect of the number of controlling families on firm performance has a limit. Assuming that the optimal number of controlling families can be estimated based on the results obtained in Model 4, Table 4, the turning point for firm performance maximization is 5.83 families, a number that lies within our sample range. As suggested by Haans, Pieters, and He (2016), we tested the first half of the inverted U-shaped pattern separately (i.e., family firms with a number of families equal to or lower than 6). The unreported results show a positive and significant relationship between the number of families and ROE (β = .020, p < .001, N = 426), thus confirming that no turning point shift occurs below a number of six unrelated families. Unfortunately, we were not able to independently assess the second half of the inverted U-shaped relationship given the aforementioned right-skewed distribution of number of families. Figure 1 presents a predictive margin plot for the first half of the inverted U-shaped relationship. Here, we visually depict a positive and concave relationship between the number of families and firm performance. Since the range of number of unrelated controlling families fluctuates between one and six for 98.6% of the total sample (right-skewed distribution), we explored whether the results obtained in Model 4 of Table 4 are robust after excluding MFFs with a number of families greater than six (Haans et al., 2016). The unreported results show a positive but insignificant linear effect of number of families and ROE (β = .025, ns, N = 426) and a negative but insignificant nonlinear effect for the focal relationship (β = −0.001, ns, N = 426). Altogether, we did not find robust evidence of the predicted inverted U-shaped relationship between the number of unrelated controlling families and firm performance because of the nature of our sample.
First half of the inverted U-shaped relationship between the number of families and firm performance.
Finally, consistent with prior studies (e.g., Anderson & Reeb, 2003a; Maury, 2006; Villalonga & Amit, 2006), Table 4 shows that the following control variables are statistically significant throughout all models: firm size, asset growth, investment, and cash ratio. Overall, these results suggest that MFFs are a unique type of family firms, whose governance attributes allow them to mitigate agency issues and, consequently, overperform SFFs. Additionally, the results support arguments that the balance of control among the unrelated owning families matters for MFF performance. However, the results are inconclusive regarding the effect of having a large number of owning families controlling the MFF.
Robustness Checks
We performed several robustness checks of the results reported in Table 4. Table 5 offers a summary of the coefficient obtained and the interpretation of the results. First, we examined alternative measures of firm performance such as ROA, Tobin’s Q, and sales growth (Anderson & Reeb, 2003b; Isakov & Weisskopf, 2014; Maury, 2006; Memili, Eddleston, Kellermanns, Zellweger, & Barnett, 2010). The results indicate that (a) Hypothesis 1 is supported for ROA (β = .028, p < .050, N = 432), weakly supported for Tobin’s Q (β = .169, p < .100, N = 351), and rejected for sales growth (β = .004, ns, N = 432); (b) Hypothesis 2 is supported for both ROA (β = .015, p < .001, N = 432) and Tobin’s Q (β = .132, p < .001, N = 351), but this hypothesis is rejected for sales growth (β = −0.001, ns, N = 432); and (c) Hypothesis 3 is supported for the three measures of firm performance (β = −0.001, p < .050, N = 432 for ROA; β = −0.011, p < .050, N = 351 for Tobin’s Q; and β = −0.002, p < .010, N = 432 for sales growth). We explain the nonsignificant results for sales growth in testing Hypotheses 1 and 2 by considering the conservative behavior of family owners. Family firm literature suggests that family ownership, relative to other types of ownership, may invoke more conservative strategies such as more reliance on internal cash reserves and lower leverage (Belenzon, Patacconi, & Zarutskie, 2016) that lead to superior performance (e.g., ROE, ROA) but lower sales growth (Casillas, Moreno, & Barbero, 2010; Memili et al., 2010). In fact, Table 2 indicates that MFFs do not significantly differ from SFFs in asset growth and leverage, which may partially indicate why we obtained insignificant coefficients for Hypotheses 1 and 2. Finally, the weak positive but significant association between MFF and Tobin’s Q (Hypothesis 1) suggests that MFFs can be viewed with suspicion by investors given the risk of colluding among controlling families and acting against minority shareholders’ interests (Maury & Pajuste, 2005).
Summary of Robustness Tests.
MFF, multifamily firm; ROA, return on assets; ROE, return on equity; NS, not significant.
Second, it is reasonable to suggest that MFFs occur as a rational response by families with the aim to continue sharing the firm’s control. Consequently, we test for a potential self-selection bias by performing a Heckman’s two-stage test on our sample (cf. Miller, Le Breton-Miller, & Lester, 2011). The first stage consists of a probit model that captures the probability that a family firm is an MFF. We use the fraction of industry sales that comes from MFFs (measured as the sum of MFFs’ market share at the two-digit NAICS level) as an appropriate instrument to perform the model. This instrument is correlated with the probability that a firm in the industry is an MFF (Amit, Ding, Villalonga, & Zhang, 2015). The subsequent stage performs a linear regression on firm performance. Following Amit et al. (2015), we use industry-adjusted ROE (measured as the difference between the firm’s ROE and the median ROE of its main industry at the two-digit NAICS level) as the dependent variable, which is uncorrelated with the instrumental variable. The results reported in Table 5 show a positive and statistically significant association between MFF and firm performance (β = .272, p < .010, N = 432), thus supporting Hypothesis 1 and suggesting that endogeneity is not a concern in our study. Additionally, we add a 1-year lagged ROE as an independent variable to deal with potential reverse causality where firm performance influences family control. The results summarized in Table 5 (β = .058, p < .050, N = 365 for Hypothesis 1; β = .027, p < .010, N = 365 for Hypothesis 2; and β = −0.003, p < .050, N = 365 for Hypothesis 3) are consistent with those provided in Table 4, suggesting that reverse causality can be ruled out (Sciascia, Mazzola, & Kellermanns, 2014).
Finally, to test Hypotheses 1 and 2 we use MFF board and HHI as alternative measurements of MFF and contestability, respectively. Overall, we found a positive and significant association between MFF board and ROE (β = .063, p < .001, N = 432) and, as expected, a negative and significant relationship between HHI and ROE (β = −0.094, p < .010, N = 432), thus supporting both Hypotheses 1 and 2.
Post Hoc Analyses
We performed post hoc investigations to assess potential meaningful relationships that were not formally hypothesized but may contribute to a broader understanding about the phenomenon of MFFs (Hollenbeck & Wright, 2017). First, we extended our original sample by including nonfamily publicly traded firms (i.e., institutional-owned, foreign-owned, and widely held firms) and asked whether SFFs perform better or worse than nonfamily firms. The unreported results indicate a negative and significant association between SFF and ROE (β = −0.091, p < .010, N = 593), suggesting that SFFs’ leaders may favor socioemotional goals such as keeping family control of the business despite potential threats to firm profitability (Schulze & Kellermanns, 2015).
Second, we asked whether MFFs outperform all types of controlled corporations (i.e., SFFs, institutional-owned firms, and foreign-owned firms). We argue that if nonfamily-controlled firms possess ad hoc governance rules that mitigate controller-centric initiatives, then the relative outperformance of MFFs diminishes compared to nonfamily-controlled firms. Thus, by including the subgroup of nonfamily-controlled firms, we would expect a lower coefficient of the association between MFF and firm performance than the coefficient reported in Model 2 of Table 4. Interestingly, the unreported results indicate a slightly higher coefficient for the relationship between MFF and firm performance (β = .064, p < .001, N = 505).
Third, we asked whether MFFs overperform all types of publicly traded corporations including widely held firms. It is reasonable to argue that if the distribution of control among unrelated owning families in an MFF generates a free rider issue among controllers, then the outperformance of MFFs comes from the absence of an active controller (Edmans, 2014) rather than the alleviation of principal–principal problems as suggested in our arguments for Hypothesis 1. Thus, by including the subset of widely held firms in the sample, we would expect nonperformance differences between MFFs and other types of corporations. The unreported results indicate a positive and significant coefficient between MFF and firm performance (β = .067, p < .001, N = 593). Overall, we did not find evidence that the MFFs’ outperformance is caused by lack of active control over the firm due to the dispersion of controlling ownership among families. In sum, MFFs, on average, outperform the pool of other types of publicly listed firms including SFFs, nonfamily-controlled firms, and widely held firms.
Discussion
With this study, we attempt to respond to the call for further exploration into the outcome consequences between MFFs and SFFs as well as nonfamily firms (Brigham & Payne, 2015) and contribute theoretically and empirically to the multiple-family blockholder literature (Cacciotti & Ucbasaran, 2018; Pieper et al., 2015). By combining the uniqueness of family firms (socioemotional wealth) with the most general behavior of ownership (agency theory), we propose and find that MFFs benefit from the enhanced monitory capacity for constraining entrenchment stemming from the potential diversity of socioemotional goals among controlling families and thus improve performance. However, we theorize that this positive effect is affected by two governance contingencies: the balance of control among unrelated owning families and the number of unrelated families controlling the firm. In the first case, we show that an unequal distribution of controlling ownership among unrelated families disincentivizes mutual monitoring among families and even forms a coalition that may result in entrenchment. In the second case, due to data restrictions, we were not able to find robust evidence that the relationship between the number of families and firm performance follows an inverted U-shaped pattern. However, we did find that the positive effect of having multiple owning families controlling the MFF on firm performance has a limit. These results highlight the importance of exploring MFFs as unique organizational form since they provide important implications for research on family firms and governance.
Implications for Theory
The findings of our study yield essential insights for the literature on family firms. Family business scholars must break away from the assumption that family firms are controlled by a homogeneous and monolithic group of family owners sharing the same goals (Chua et al., 2012; Zellweger & Kammerlander, 2015). As argued earlier, family firms can be also controlled by multiple and unrelated owning families with potentially different socioemotional goals and family needs. The study of family firm heterogeneity and particularly a “further exploration into the multibusiness organizational form” (Brigham & Payne, 2015, p. 1344) may help address conflicting predictions of the agency and socioemotional perspectives. The literature remains inconclusive on whether family firms’ lower principal–agent costs outweigh the adverse effects of principal–principal conflicts (Villalonga & Amit, 2006) and the pursuit of socioemotional goals (Berrone et al., 2012) on firm performance, a question particularly relevant in less developed markets (see Duran, van Essen, Heugens, Kostova, & Peng, 2018). By employing a family blockholder perspective we show that MFFs are better suited, relative to SFFs, to mitigating principal–principal issues and blocking socioemotional interests (e.g., family altruism) that go against the interests of other family and nonfamily parties, thus enhancing profitability.
Our study also contributes to the corporate governance literature, specifically as it relates to controlling contestability and shareholders’ coalitions. First, we show that firm controllers are divisible principals, susceptible to different and competing incentives that create potential conflicts among them. Second, we theorize and show that control coalitions among unrelated investors can be more valuable for firm performance than a single control structure due to the mutual monitoring incentives and balance of power. Finally, and related to the prior point, this work argues that the balance of control among unrelated owning families in MFFs can generate similar positive effects on firm performance in the presence of other largest shareholders. Our results suggest that control contest models are also valid within controlling families, but contrary to previous studies (Bertrand et al., 2008; Jara-Bertin et al., 2008; Maury & Pajuste, 2005), we propose and find that shared control structures among unrelated owning families can fuel firm performance and that contestability can also exist within shared control structures.
Implications for Practice
Our study suggests implications for both family firm owners and policymakers. It has often been argued that one of the main goals that family owners pursue is ensuring family firms’ survival and success for future generations (Berrone et al., 2012). To accomplish this, family firms must avoid the risk of family conflicts or of hiring incompetent family successors as business leaders. We argue in this study that mutual monitoring among family controllers is one of the benefits of MFFs. Thus, adding unrelated families to the firm’s blockholders’ group may help to mitigate the likelihood that family conflicts, derived from multiple socioemotional goals that family members pursue, negatively affect the firm’s competitiveness. However, family owners should take into account that MFFs with a relatively equal distribution of control among owning families may be more effective than MFFs with unbalanced control in blocking less profit-driven socioemotional desires of other families. On the other hand, an uneven distribution of controlling ownership among families increases minority family owners’ incentives to collude with the largest family to gain control advantages (Maury & Pajuste, 2005).
Furthermore, our investigation shows evidence that MFFs reduce principal–principal costs, which should be of particular interest to policymakers. It is argued that a high degree of ownership concentration in the hands of a family creates incentives to expropriate minority owners (Morck & Yeung, 2003). Policymakers may address this issue, for example, by encouraging multiple blockholder structures in firms (Edmans & Manso, 2011). In countries where entrepreneurs trust more in family partnership (Duran et al., 2018; Fukuyama, 1995), and family capitalism prevails (Steier, 2009), we suggest that policymakers create conditions to increase the prevalence of MFFs.
Limitations and Future Research
Our study presents several limitations. First, the findings must be interpreted with caution since our sample (i.e., Chile) does not represent all contexts in which family firms operate, and the context determines the behavior and outcome of family firms (Duran et al., 2018). Future work should replicate and extend these results in other countries. Second, the performance effects of family ownership can be explained through multiple theoretical perspectives. We draw our hypotheses from both agency and behavioral agency theories, but there are other theories that may also contribute to understanding the phenomenon of MFFs. We encourage researchers to consider alternative lenses to better understand MFFs. For example, questions such as to why some families are more willing than others to join other families and thus form an MFF can be approached from a social capital perspective (Carney, 2005). Third, this study relied on family firms that are publicly listed. Thus, our findings may not generalize to privately held family firms. Future research should explore whether the presence of multiple and unrelated family owners in privately held family firms is beneficial for firm performance. Fourth, as mentioned earlier, we were not able to provide strong evidence to support our Hypothesis 3 due to data constraints. In our dataset, the distribution of the number of owning families is skewed toward the right. This may suggest that either potential family investors find it unattractive to belong to the controlling group with a large number of owning families or family controllers do not want to incorporate new families in the controlling alliance since coordination issues may negatively affect firm performance (Fattoum-Guedri et al., 2018). We encourage future studies to further explore these ideas. Fifth, to mitigate multicollinearity concerns, we tested Hypotheses 2 and 3 in separate models; thus, the results should be interpreted cautiously. However, future research using data from other countries could seek to test these hypotheses in one empirical model. Finally, MFFs in Chile include owning families with less than 5% of the firm’s total shares outstanding. Although this percentage is below the common cutoff control threshold (cf. Lins, 2003; Peng & Jiang, 2010), these owning families still may enjoy influence over governance and strategic direction in Chilean corporations. Future research might examine the generalizability of our operationalization of MFF in other contexts (Bruton, Ahlstrom, & Obloj, 2007).
Our study merely scratches the surface of the MFF phenomenon. Strategy scholars could explore, for example, both strategic and investment differences between MFFs and SFFs, which could consequently affect firm performance. We suggest that MFFs impede a greater orientation toward socioemotional goals in family owners. Thus, MFFs are less constrained in pursuing risky strategic choices. Additionally, future research could explore differences between families in MFFs whose wealth is diversified in many firms and families whose wealth is concentrated in an MFF. The former can be more aggressive in pursuing risky strategies and less committed to monitoring than the latter.
Future research can also inform succession decisions in MFFs. We conjecture that MFFs have strong incentives to generate a competitive succession process including both professional and family candidates. Pieper et al. (2015) suggest that MFFs are more willing to engage professional nonfamily managers to help families address the firm’s complexity. Furthermore, we suggest that in the case of MFFs with a strong socioemotional concern of exerting family control over a company’s strategic decisions, they might enjoy, relative to SFFs, a more competitive internal labor market with candidates from several families who compete for leadership positions. Here, family candidates would compete under preestablished rules, such as minimum education and experience requirements (Kidwell, Eddleston, Cater, & Kellermanns, 2013) and need the explicit support from all families.
Finally, the study of institutions may contribute to explaining the benefits of MFFs for firm performance. We may argue that MFFs enjoy certain performance advantages in less institutionally developed contexts (Duran et al., 2018; Martin, 2014), where family ownership enjoys greater legitimacy in local environments (Miller, Breton-Miller, & Lester, 2013) and investors face poor protection laws. Thus, future research can help to disentangle firm-level effects from country-level effects. 6
Footnotes
Acknowledgments
The authors gratefully acknowledge comments and suggestions from Professor Franz W. Kellermanns (Editor) and two anonymous reviewers. We thank the support of Universidad Adolfo Ibáñez and Universitat Pompeu Fabra. Marcelo Ortiz acknowledges the support from CONICYT-PFCHA/Doctorado Nacional/2018-21180671.
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) received no financial support for the research, authorship, and/or publication of this article.
Notes
Author Biographies