Abstract
We propose a new perspective on family firms' puzzling under-diversification in product spaces: these firms first need to succeed in exploratory innovation so they may diversify into new product markets. We construct a large database of family ownership and patent records of U.S. public firms, and show that family firms produce more exploratory patents than others, a relation that is stronger among under-diversified family firms. In addition, we find that such innovation indeed helps family firms diversify business risks. A causal interpretation of our result is supported by (i) using the property division standard in state-level divorce laws as an instrumental variable and (ii) constructing a propensity score matched sample. This effect is also more pronounced among larger firms, older firms, and firms in industries with faster technology replacement. Our empirical evidence addresses family firms' under-diversification puzzle through the lens of innovation strategies.
1. Introduction
Family firms have played an important role in the global economy.1 Unlike typical publicly listed firms, risk diversification presents a key consideration for family firms, as retaining control requires controlling families' wealth and attention concentrated in a single firm (e.g., Faccio and Lang, 2002; Shleifer and Vishny, 1992). Moreover, controlling families have a long-horizon perspective (Kim et al., 2008; Miller and Le Breton-Miller, 2005), motivating them to pursue diversification when they identify long-term risks and challenges to their current business (Choo et al., 2009; Gomez-Mejia et al., 2007, Gomez-Mejia et al., 2010). Given the importance and practicality of risk diversification, it is puzzling that family firms often remain under-diversified in product spaces (Anderson and Reeb, 2003; Feldman et al., 2016; Gomez-Mejia et al., 2010).
In this paper, we revisit this puzzle by investigating how diversification influences family firms' development of technologies prior to product creation. As technology cycles accelerate and innovation competition intensifies, firms and managers must invest in R&D and innovative activities to survive and thrive in the long run (Ahuja et al., 2008; Sundaram et al., 1996). This persistent pressure may compel family firms to more actively pursue new technologies aligning with their long-term perspective (Gomez-Mejia et al., 2011). We hypothesize that the diversification incentives of controlling families may prompt their firms to particularly favor exploratory innovation. Exploratory innovation generates new knowledge that is substantively different from a firm's existing expertise through distant searches, radical experimentation, and revolutionary approaches (March, 1991; Levinthal and March, 1993). This enables firms to expand into new technological domains and unrelated business fields (Kim et al., 2024; Lavie et al., 2010). Consequently, exploratory innovation serves as the first step for controlling families to diversify their concentrated business risks.
To empirically examine our proposition, we construct a large database of family ownership of 3,391 U.S. public firms and link it to U.S. patent data, which sets our study apart from prior research that has focused on a smaller, selected sample or on index firms such as Fortune 100 or S&P 500. We define a patent as exploratory if it has a higher ratio of backward citations to prior patents that are outside of the filing firm's existing knowledge base. We regress a firm's number or ratio of exploratory patents filed in a year on the firm's family ownership (measured by a family-firm indicator variable or a ratio of family ownership in total shares), controlling for several firm characteristics, year fixed effects, and industry fixed effects. Our results suggest that family ownership is positively related not only to the number of exploratory patents but also to the ratio of exploratory patents in a firm's new patent portfolio. This relation is not only statistically significant but also economically sizable: when a nonfamily firm becomes family-owned, its number (ratio) of exploratory patents will increase by 37.2% (8.5%). In addition, a one-standard-deviation increase in family ownership is associated with an 8.0% increase in the filing of exploratory patents.
We further validate the diversification motive underlying the above relation. If the purpose of adopting exploratory innovation is to reduce risks, then less-diversified family firms should be more inclined to adopt exploratory innovation. This is supported in the data. We regress a firm's number or ratio of exploratory patents on the interaction term between an indicator for single-sector firms and the firm's family ownership. On the one hand, the significantly negative coefficients on the single-sector indicator suggest that firms with concentrated product lines are less likely to pursue exploratory innovation. On the other hand, the insignificant coefficients on family ownership suggest that family firms do not always pursue exploratory innovation. More importantly, the coefficients on the interaction term are significantly positive, confirming that only less-diversified family firms have the incentive to file exploratory patents.
Moreover, exploratory innovation should contribute to the intended outcomes of risk diversification, such as reduced volatility in profitability and stock returns. Our tests confirm this prediction: we find lower volatility in return on assets (ROA) and stock returns among firms filing at least one exploratory patent. Taken together, these findings support our hypothesis that family firms engage in exploratory innovation to achieve diversification, thereby mitigating risks to the monetary and socioemotional wealth of controlling families.
A general concern of our baseline analysis is that the family ownership-exploratory innovation relation may not be causal. For instance, confounding factors that influence both family ownership and innovation strategies could exist. To mitigate such endogeneity concerns, we consider the property division standard in state-level divorce laws as an instrumental variable.2 State divorce laws based on the community property standard benefit the poorer spouse (Voena, 2015) and thus can negatively affect the incentives for the wealthier spouse to accumulate family assets (e.g., Dnes, 1998; Voena, 2015; Fischer and Khorunzhina, 2019; Frémeaux and Leturcq, 2022). Theses property division standards are also plausibly exogenous to firm fundamentals (Roussanov and Savor, 2014). As such, state divorce laws help identify variations in family ownership. Consistent with its known negative impact on family assets, we observe that the community property standard negatively affects family ownership. More importantly, instrumented family ownership that is free of confounding effects still positively explains exploratory innovation output in the second stage, which supports a causal interpretation.
We also construct a propensity score matched sample in which family and nonfamily firms are homogenous in all firm characteristics using propensity score matching (Villalonga, 2004; Feldman et al., 2016). This sample mitigates the concern that family-owned firms might be different from nonfamily firms in terms of characteristics that are spuriously correlated with exploratory innovation. Also in this case, we observe a consistent, positive relation between family ownership and exploratory innovation.
We also examine the heterogeneous effects on the incentives of family firms to expand business opportunities. We first document that larger family firms are more exploratory than smaller ones in innovation activities. This observation is reasonable because family firms are more subject to financial constraints due to controlling families' need to retain control and reluctancy in seeking external capital (Morck et al., 2000; Morck and Yeung, 2003; Munari et al., 2010). Thus, larger family firms face fewer financial constraints when pursuing exploratory innovation (Xiang et al., 2019; Eng et al., 2021).
Next, we observe that older family firms tend to pursue more exploratory innovation. Two economic reasons may help explain this observation. First, older firms are more likely to rely on obsolete technologies, which gives them a stronger incentive to engage in exploratory innovation to ensure their long-term survival and technological diversification (Grandstrand, 1998). This incentive is likely stronger for family firms, which have longer investment horizons and require renewal through innovation to avoid failure (Gomez-Mejia et al., 2011; Laforet, 2013). Second, older firms tend to be controlled by later generation-families. These families have a greater incentive to preserve the value of their inherited assets, and see risk diversification as an important tool to achieve this goal. This effect aligns with recent findings that the second-generation successors tend to shift innovation trajectory toward exploration in China and Turkey (Erdoğmuş et al., 2017; Carney et al., 2019).
Lastly, we consider the effect of technology obsolescence. We find that the relation between family ownership and exploratory innovation is stronger in more R&D-intensive industries and industries with shorter half-life of backward citations, respectively. These results suggest that exploratory innovation matters more for family firms operating in industries characterized by more rapid technological changes. They are also consistent with our proposition because controlling families of high-tech firms should also prioritize innovation when technological changes accelerate (e.g., Sundaram et al., 1996; Trembley and Chênevert, 2008).
Our study is related to several strands of the literature. First, we provide a new perspective on the family ownership and diversification relation. Our findings on family firms pursuing exploratory innovation and technological diversification reconcile the puzzling contrast between the theoretical prediction of family firms' higher diversification and counter evidence from prior empirical analyses (Gomez-Mejia et al., 2011, pp. 667–668). Our evidence also echoes the proposition of Ahuja and Novelli (2017) that firms can be relatedly diversified in technologies but not in product markets. Moreover, our analyses for the boundary conditions including firm size, firm age, and technology pace offer additional insights on this literature.
Second, we provide new evidence to the literature on the determinants of exploratory innovation. Prior studies have examined how exploratory innovation is influenced by managerial capabilities (Levinthal and March, 1993; Smith and Tushman, 2005), alliances and networks (Phelps, 2010; Rothaermel and Deeds, 2004; Yang et al., 2014), mergers and acquisitions (Stettner and Lavie, 2014), and business environments (Kim et al., 2024; McGrath, 2001).3 Our empirical evidence for the effect of family ownership fills a gap in the literature of exploratory innovation (Brinkerink, 2018; Ardito and Capolupo, 2024).
Finally, our study also adds to the literature on the role that family ownership plays with respect to corporate innovation. Our proposition and empirical evidence suggest that major shareholders' diversification considerations could also fundamentally shape a firm's innovation choices. While some studies find that family firms are more engaged in innovation (e.g., Choo et al., 2009; Duran et al., 2016; Miller and Le Breton-Miller, 2005; Schmid et al., 2014), other studies point to the opposite direction (Morck et al., 2000; Morck and Yeung, 2003; Munari et al., 2010). In contrast to these studies and their focus on the level of innovation, we focus on the direction of innovation (i.e., exploratory innovation) that is closely related to firms' technological diversification and subsequent product diversification.
2. Literature review and hypothesis development
2.1. Family ownership and technological diversification
Family firms' strategic decisions are often analyzed through two non-mutually exclusive lens: agency theory and socioemotional wealth theory (Feldman et al., 2016; Gomez-Mejia et al., 2007, Gomez-Mejia et al., 2011). Agency theory suggests that controlling families choose to concentrate their wealth in one company (and bear the associated business risks) because they expect to gain extra economic rents through tunneling or private benefits (e.g., social status, empire building). Thus, since controlling families have objectives and interests different from those of nonfamily or minority shareholders, their decisions may not always maximize market value (Shleifer and Vishny, 1986; Villalonga and Amit, 2006).
Socioemotional wealth theory posits that controlling families derive noneconomic utilities from operating and/or monitoring their businesses (Gomez-Mejia et al., 2007). Such utilities include emotional attachment, involving family members in daily operations, the sense of legacy, and the intention to pass wealth and control to the next generation; thus, family firms will allocate significant resources toward facilitating intergenerational business transfers (Brunetti, 2006). The strategic choices of family firms are driven by a desire to maintain and enhance the controlling families' socioemotional wealth over other economic or monetary considerations.
Given rapid changes in the technology landscape and accelerated shifts in industry structures, today's firms must stay flexible and agile—largely by creating and adopting innovations. In other words, technological diversification, or the ability to develop and maintain a broad set of technologies, is important for firms that seek to survive technology and business cycles (Ahuja and Novelli, 2017; Choo et al., 2009; Miller, 2006). The risk of obsolescence due to rapid technological change and imitation can be attenuated if firms are technologically diversified (Grandstrand, 1998). For instance, Hsu et al. (2018) show that firms with more diversified patent portfolios are more capable of recovering from disruptions.4 Moreover, technological diversification promotes cross-fertilization of new inventions and prevents firms from being locked-in to particular technologies (Lee et al., 2023; Park and Lee, 2006; Makri et al., 2010; Makri and Lane, 2008). Technological diversification thus constitutes a critical direction of firms' strategic decisions, especially for those with major shareholders who are more long-term-oriented.
Both theories suggest that controlling families have relatively stronger incentives than other major shareholders to maintain firms' long-run operations but weaker incentives for pursuing short-term market prices (Miller and Le Breton-Miller, 2005; Wiseman and Gomez-Mejia, 1998). That said, there is limited evidence on the relation between family ownership and technological diversification (Gomez-Mejia et al., 2011, pp. 672–674). Gomez-Mejia et al. (2014) show that high-technology family firms are more willing to invest in R&D and engage in technological diversification as ROA declines. Choo et al. (2009) show that Korean chaebols have improved their performance by reducing investment inefficiencies and enhancing technological capabilities since the 1997 financial crisis. We contribute by proposing that controlling families are more motivated than other major shareholders to pursue technological diversification, so they may ensure the longevity of their firms and sustain their respective families' legacy.
2.2. Family firms' inclination toward exploratory innovation
Exploratory innovation is an innovation strategy that emphasizes new knowledge creation that substantially deviates from a firm's current innovation trajectories. To create such knowledge, firms employ remote searches and revolutionary experimentation (March, 1991; Levinthal and March, 1993). Such innovation may be costlier and its payoffs are of higher uncertainty. The recent literature highlights that exploratory innovation can especially help firms seeking to develop necessary technologies and expertise that will help them expand into unrelated business fields (e.g., Balsmeier et al., 2017; Kim et al., 2024; Manso, 2011). As suggested in March (1991), firms must engage in exploratory innovation to adapt to industrial structural changes and to survive in dynamic environments. In addition, exploratory innovation hence serves as the needed first step for controlling families to diversify their concentrated business risks.
Although exploratory innovation is riskier and more costly, family firms are more concerned about long-term survival than short-term pay-off (Gomez-Mejia et al., 2007; Palmer and Wiseman, 1999; Wiseman and Gomez-Mejia, 1998); as a result, family firms are more willing to invest in exploratory innovation and take short-term losses required to survive and thrive in the future.5 We thus form our primary hypothesis as the following:
Hypothesis 1a
Family ownership promotes exploratory innovation.
To validate the role of diversification in the context of the above hypothesis, it is important to further address both the purpose for and the intended outcomes of exploratory innovation. On the one hand, as controlling families face risks stemming from concentrated wealth tied to the operations of the firms they own, their motivation for engaging in exploratory innovation should be stronger when their firms are concentrated in specific product or market spaces, making them more vulnerable to business risks. On the other hand, exploratory innovation should help family firms achieve its intended diversification outcome, namely, the reduction of business risks. These considerations lead to the formulation of the following supportive hypotheses:
Hypothesis 1b
The effect of family ownership on exploratory innovation is stronger for firms with narrow product lines.
Hypothesis 1c
Exploratory innovation helps family firms reduce business risks.
The relation between family ownership and exploratory innovation may vary by different conditions related to technologies and family characteristics. The first conditional factor we consider is firm size, which reflects firms' financial resources and potential cross-fertilization of technological innovations. The literature has suggested that family firms are more subject to financial constraints due to controlling families' need to retain control and reluctancy in seeking external capital (Morck et al., 2000; Morck and Yeung, 2003; Munari et al., 2010), and such a choice unavoidably dampens innovation activities (Xiang et al., 2019; Eng et al., 2021). We thus propose that the positive effect of family ownership on exploratory innovation increases with firm size.
The second conditional factor is firm age for two reasons. First, older firms are likely to use obsolete technologies—particularly older family firms given their longer investment horizons—and therefore have a stronger incentive to engage in exploratory innovation to avoid failure (Grandstrand, 1998; Gomez-Mejia et al., 2011; Laforet, 2013). Second, older family firms tend to be controlled by later generations of families, whose incentives of preserving the value of their inherited assets likely prompt them to pursue exploratory innovation as a way of risk diversification (Erdoğmuş et al., 2017; Carney et al., 2019), aligning with evidence that a more conducive environment for generational continuity fosters greater investment in family firms (Ellul et al., 2010; Francis, 2012). These considerations suggest that the positive effect of family ownership on exploratory innovation is more pronounced in older family firms.
The third conditional factor we consider is the speed of technology obsolescence. Firms face the risk of being left behind in technology race, and such risk tends to be higher in industries with accelerating technology replacement (Sundaram et al., 1996). Since controlling families' top priority is to ensure the long-term survival of firms they own, they should be more incentivized to explore innovation opportunities under greater technology obsolescence risk (Trembley and Chênevert, 2008; Gomez-Mejia et al., 2011). Moreover, technological diversification associated with exploratory innovation prevents family firms' being stuck with outdated technologies that are more likely to occur in rapid-changing industries (Garcia-Vega, 2006). We thus propose that the positive effect of family ownership on exploratory innovation increases with technology replacement speed.
Aforementioned discussions lead to the following conditional hypotheses:
Hypothesis 2a
Family ownership promotes exploratory innovation, especially for larger firms.
Hypothesis 2b
Family ownership promotes exploratory innovation, especially for older family firms.
Hypothesis 2c
Family ownership promotes exploratory innovation, especially for firms facing rapid technology replacement.
3. Data
3.1. Exploratory innovation
We collect patent data from the Harvard Business School Patent Inventor Database of Li et al. (2014). These data include detailed information for all patents granted by the U.S. Patent and Trademark Office (USPTO) from 1976 to 2010. Following Benner and Tushman (2002), Katila and Ahuja (2002), and Gao et al. (2018), we define a firm's existing knowledge base as the patents it has filed over the past five years and the citations made by these patents. These patents and citations describe what has been created (in terms of new patents) and learned (from existing patents that are cited by these new patents) by a firm over a certain period. We categorize a patent as “exploratory” if 60 % or more of its citations are based on new knowledge from outside of a firm's existing knowledge base (i.e., do not cite the firm's recent patents or any citations made by those patents).
At the firm level, we can then define the variable Explorei,t in year t as the logarithmic value of one plus the number of exploratory patents filed by firm i in year t. To avoid any potential problem associated with this level variable, we also construct a percentage variable of exploratory innovation, %Explorei,t, defined as Explorei,t scaled by the total number of patents filed by firm i in year t. We set both variables to zero in a year for firms without any patent records or without exploratory patents in that year.
3.2. Family ownership
We identify family ownership for listed companies using ownership data from the Orbis database from Bureau van Dijk. This database traces the ownership and hierarchies of firms on a yearly basis from 2000 to 2011. For each public firm listed in the U.S. and covered in the Orbis database, we accumulate the voting rights of the firm across a possible pyramid structure (if any) until we are able to determine its ultimate owner. We also manually verify and augment identified ownership information by checking various information sources, including company annual reports, LexisNexis, and Factiva. Next, we follow Masulis et al. (2011) and manually check whether the ultimate owner of each firm is a family (e.g., biologically linked families, individual entrepreneurs, known alliances of families or entrepreneurs) or a nonfamily entity (e.g., governments, widely held firms, collective investment funds, widely held financial institutions). Due to our large sample size and the complexity of ownership identification, we manually verify the information on ultimate owners for four specific years in the sample period: 2002, 2005, 2008, and 2011. We then expand this verified information to the rest of our sample period by filling in information for the other years with that in the closest year.6
Using the preceding data and information, we construct two proxies for family ownership. The first proxy for family ownership is a dummy variable, Family Dummy, that equals one if the firm's largest shareholder is a family and if the family effectively controls (directly or through the holdings of affiliates) at least 20% of the firm's voting rights, and zero otherwise (Charumilind et al., 2006; Masulis et al., 2011). Based on this threshold, our sample includes 788 unique family firms, 16 of which experienced drastic transitions from dominant family ownership to dominant nonfamily ownership or vice versa. We also calculate the fraction of total voting rights ultimately owned by the controlling family of a firm (Family Ownership), which ranges between 0 (if the firm does not have a controlling family) and 1 (if the controlling family retains all voting rights).
3.3. Summary statistics
Our final sample consists of 21,949 firm-year observations (3,391 unique firms) in the sample period from 1998 to 2010.7 In Table 1, we present the summary statistics of our main variables for our baseline regression. Explore has a mean value of 0.18 (corresponding to an average of 0.56 exploratory patents per year) and a standard deviation of 0.50. %Explore has a mean value of 0.11 and a standard deviation of 0.29. In addition, the sample mean and standard deviation of Family Dummy are 10.0% and 30.0%, respectively, suggesting that 10.0% of sample firms are family-owned. In addition, the sample mean and standard deviation of Family Ownership are 4.3% and 14.4%, respectively, suggesting that 4.3% of outstanding shares are owned by controlling families.
Table 1. Summary statistics.
4. Analyses
4.1. Main results
We first examine the relation between family ownership and exploratory innovation by estimating the following pooled ordinary least squares regression for our baseline analysis:
Explorei,t+1 or % Explorei,t+1 =α0+β1 Familyi,t +λ Controlsi,t+FE+εi,t , (1)
in which Explorei,t+1 and Explorei,t+1 have been defined earlier and reflect the exploratory innovation of firm i in industry j in year t + 1, Familyi,t denotes the Family Dummy or Family Ownership of firm i in year t, and Controlsi,t denotes an extensive list of control variables8 as well as SIC 3-digit industry fixed effects and year fixed effects.9 We cluster standard errors by firm.
We tabulate the results of our baseline regression in Panel A of Table 2. We first observe that the family dummy is positively related to the development of exploratory innovation. In Models (1) and (2), the coefficient estimates of Family Dummy are 12.5% and 8.5% for Explorei,t+1 and %Explorei,t+1, respectively, and both estimates are statistically significant. In other words, when a nonfamily firm becomes family-owned, it will file 0.207 more exploratory patents (about 37.2% of the sample mean) and its ratio of exploratory patents increases by 8.5%.10 Similarly, a positive relationship can be found between Family Ownership and exploratory innovation in Models (3) and (4): the coefficients of Family Ownership are again highly statistically significant. In this case, a one-standard-deviation increase in family ownership is associated with 8.0% more exploratory patent filings and a 2.2% increase in the ratio of exploratory patents.11
Click to read more