1 Introduction

Family firms (FFs) are ubiquitous around the world and dominant in many countries (De Massis et al., 2018; Schulze & Gedajlovic, 2010), but some aspects of their behavior remain unexplored or only partially explained. One topic that deserves more attention is FFs’ innovation activities (Casado‑Belmonte et al., 2021). A number of recent studies in this area analyze the relationship between family ownership and innovation output (e.g. Aiello et al., 2021a, 2022; Asaba & Wada, 2019; Bannò, 2016; Block et al., 2013; Decker & Günther, 2017; Matzler et al., 2015), whereas others consider innovation input (Aiello et al., 2020, b; Block et al., 2012; Brinkerink & Bammens, 2018; Yang et al., 2019). However, while questions of whether, why, and how much FFs, and non-FFs differ in terms of innovation have been widely debated, the theoretical, and empirical evidence is inconclusive.

Recent reviews summarize the state of this debate (Block et al., 2022; Calabrò et al., 2019; Casado‑Belmonte et al., 2021; De Massis et al., 2013; Duran et al., 2016; Heider et al., 2022). Duran et al.’s (2016) influential meta-study shows that FFs produce more innovation output with fewer innovation inputs. To explain this finding they indicate that FFs, compared to non-FFs, have more efficient innovation processes because of both better monitoring and higher levels of tacit knowledge about their firm’s members, routines, and stakeholders. This allows FFs to better orchestrate their resources and thus convert innovation input into output with a greater degree of success (Duran et al., 2016). Conversely, Block et al. (2022) confirm that FFs use less innovation input than non-FFs but do not find systematic differences in output, suggesting the need to focus more on factors that explain differences among FFs, rather than only comparing FFs to non-FFs. Consistent with this line of reasoning, other studies argue that the contrasting findings across studies are the result of factors arising from firm heterogeneity that have not yet been properly considered in the literature (Chua et al., 2012; Daspit et al., 2021; De Massis et al., 2014a, b; Hernández-Linares et al., 2017; Patel & Chrisman, 2014).

Based on these analyses, more research on family businesses is needed to provide an analytical framework that can better incorporate heterogeneity across FFs. This is a difficult task because the differences among FFs are vast, even greater than the differences between FFs and non-FFs (Bennedsen et al., 2010; Chrisman & Patel, 2012). It is worth noting that previous studies examine heterogeneity by identifying unique categories. For instance, Nordqvist et al. (2014) focused on governance by identifying nine types of FFs. Others study variations in a particular condition, highlighting how FFs vary with respect to technological innovation (De Massis et al., 2013), social capital (Herrero, 2018), and transgenerational control intentions (Chrisman & Patel, 2012) to name a few examples.

In contrast to the existing literature, we extend and recharacterize research on innovation beyond the simple “family versus nonfamily” dichotomy to pursue a refined understanding of FF heterogeneity and enhance empirical precision. To this end, we consider two sources of firm heterogeneity, namely business dimension, and firm age, as they are the most debated firm-level structural factors influencing innovation activities (Balasubramanian & Lee, 2008; Leal-Rodríguez et al., 2015; Sørensen & Stuart, 2000). Here, we explore how, and why size and age affect FFs by better capturing, measuring, grasping, and gauging heterogeneity, with the goal of providing convincing answer to the following key research question: is the innovation gap between FFs and non-FFs (jointly) moderated by size and age?

The contributions of this study are threefold.

First, we improve the understanding of firm heterogeneity by using two dimensions rather than one, allowing us to gauge differences due not only to a single category (size or age) but also the combination of the two categories (size and age). This creates a more granular firm categorization, which helps us to better understand the multiple levels of the complex relationships within and between firm groups. As FF heterogeneity is explored across two interacting levels, we refer to a methodological approach that is best suited for simultaneously measuring differences among FFs and preserving the richness of data regardless of the level. In this way, the study offers additional insights to the increasingly frequent emphasis on FF heterogeneity (Daspit et al., 2021).

Second, using size, and age as key differentiators of firm types, this study contributes to understanding how and why these variables lead some FFs to be more successful than others. Some examples help to clarify the strength of using size and age jointly.

Large FFs are particularly sensitive to the influence of the generational stage (proxied by age) compared to small companies (Le Breton-Miller & Miller, 2013). Similarly, age matters when considering financing decisions: FFs suffer more from financial constraints due to their reluctance to use external resources in order to maintain control of the firm (Serrasqueiro et al., 2016). These constraints are binding for young firms but not for mature, established firms. In fact, being in the market for longer allows FFs to acquire a reputation that makes it easier to borrow (Diamond, 1989; Hall et al., 2004).

These considerations prove not only that FFs are far from homogeneous but also highlight how FFs innovate differently from non-FFs because of their specific characteristics (Carney et al., 2015; De Massis et al., 2013). For this reason, our hypothesis is that innovation performance changes across subgroups of firms, that is, combinations of FFs/non-FFs, mature/young, large/small firms, with the result that outcomes hold in some dimensions but not others. This potential variety of outcomes shows that age and size determine the context in which firms operate, offering more nuanced explanations about the determinants of innovativeness for FFs. In doing so, the research enriches the debate and helps to explain previous contradictory findings on FFs’ innovation gap (Chirico et al., 2020; O’Boyle et al., 2012). In this regard, it is worth noting that a number of studies analyze the impact of size and age on FF innovation (Aiello et al., 2022; De Massis et al., 2014a, b; Decker & Günther, 2017; Laforet, 2013; Werner et al., 2018) but provide inconclusive evidence because while some assess the individual effects of age and/or size, few analyze their role as moderators, and none considers their joint effect. To the best of our knowledge, this is the first study assessing if and to what extent size and age jointly affect the relationship between family ownership and patenting.

Third, by introducing socioemotional wealth (SEW) considerations as an important factor influencing firms’ innovation strategies (Chrisman & Patel, 2012; Gomez-Mejia et al., 2014), this study builds on SEW’s role in innovation and shows that this approach offers an appropriate theoretical lens through which one can interpret FFs behavior in depth, regardless of the firm type (i.e., young, and small; mature and large).

The analysis is based on a sample of 3500 Italian manufacturing firms observed over the period from 2010 to 2017. Data provided by the Bureau van Dijk was linked to the European Patent Office (EPO) PATSTAT dataset. Performance is gauged by the number of patent applications per firm per year (Griliches, 1990; Hall et al., 1986). The data show that it takes the EPO an average of roughly three years to grant a patent, measured from the time the application is submitted. Therefore, we considered patent applications from 2010 to 2017 that were granted by 2021. In this way, we avoid introducing biases into the analysis due to the time lag of the examination process at the patent office; we also limit the number of low-quality patents counted in the analysis (such as applications that were rejected). Here, it is worth mentioning that patents do not capture the entire output of innovation activities, as firms can develop nonpatentable innovations and/or can decide to protect their innovations with alternative appropriability strategies (e.g., trade secrets, other forms of intellectual property rights, complementary assets, etc.) even if they are patentable. However, compared to other innovation indicators patents avoid subjectivity in assessing what constitutes an innovation (Griliches, 1990). Furthermore, in consideration of the fact that patent distribution is highly skewed, reflecting overdispersion, our econometric analysis uses a zero-inflated nonlinear count model and several sensitivity checks are performed to assure the robustness of the estimates.

Our results are summarized as follows. On average, FFs obtain fewer patents than non-FFs, regardless of size, and age. When the individual effect of size on patenting is considered, the analysis indicates that size amplifies the disadvantages of familiness: FFs and non-FFs show similar levels of innovation performance when they are very small; however, beyond a certain threshold (sales of less than 2.8 million euros) FFs obtain fewer patents than non-FFs. Furthermore, firm age matters: first-generation enterprises benefit more from an age-related learning process and from the founder effect. Finally, and most importantly, FFs’ innovation performance is jointly influenced by size and age: FFs underperform when they are small and young, and when they are large and old, but we find no substantial differences compared to their nonfamily-owned counterparts in other cases. These findings cast doubt on the perceived negative influence of family ownership and shed additional light on the distinctive traits of FFs by extending the understanding of how specific attributes enable FFs to benefit from innovative activities.

The remainder of the study is organized as follows: Section 2 presents the theoretical background and the hypotheses; data, variables, and empirical strategy are described in Section 3; the results are presented in Section 4; and Section 5 discusses the results and offers some concluding remarks.

2 Theoretical background and hypotheses development

2.1 Family firms and innovation performance

In recent years, the literature on FFs’ innovation has grown enormously (Casado-Belmonte et al., 2021; Heider et al., 2022) and has emphasized how FFs behave differently when compared to a nonfamily counterpart, showing that their typical characteristics can exert either a negative or a positive influence on innovation (Block et al., 2022; Calabrò et al., 2019; Carney et al., 2015; De Massis et al., 2013; Duran et al., 2016; Röd, 2016). In this regard, Socioemotional Wealth (SEW) has been widely acknowledged as a driving force guiding FFs behavior and decision-making (Berrone et al., 2012; Gomez-Mejia et al., 2007; Gomez-Mejia et al., 2011; Gomez-Mejia et al., 2021), thereby representing a suitable theoretical framework to understand FFs attitude toward innovation (e.g. Calabro et al., 2018; Chrisman & Patel, 2012; Gomez-Mejia et al., 2014).

According to the SEW theory, family members benefit from a variety of nonfinancial and emotional outcomes associated with business, such as (a) enhancing and perpetuating family image and reputation, (b) maintaining family control of ownership and management, and (c) sustaining the family’s dynastic aspirations and ensuring that the business remains viable across future generations (Chrisman et al., 2012). In this regard, Miller et al. (2015) juxtaposed the SEW goals of FFs with the level of innovation needed to compete in the market, highlighting that family business socioemotional preferences are Janus-faced: they can facilitate innovation or do exactly the opposite.

On the one hand, FFs can build social capital resources that promote the innovation (Miller et al., 2015). Indeed, thanks to the high levels of social capital available (Arregle et al., 2007), they are able to gain more than non-FFs from knowledge flows. Social capital thus constitutes a valuable source of competitive advantage compared to non-FFs (Pearson et al., 2008) and is crucial in enhancing absorptive capacity (Andersén, 2015), which is the key factor in transforming knowledge into performance (Gkypali et al., 2018). Other elements that drive the innovation are family ties that reduce agency and coordination costs. Moreover, family values boost employee motivation, and family reputation helps in building external networks at lower costs than non-FFs (Bennedsen & Foss, 2015).

On the other hand, SEW goals do not always benefit FFs: they often conflict with their economic interests, resulting in a lack of talent and bringing agency problems (Chrisman et al., 2014; Verbeke & Kano, 2012). Indeed, family-directed altruism, jobs for incompetent family members, and the entrenchment of undeserving family executives can erode human, relational, and financial resources, and hence hinder innovation (Miller et al. 2015). Moreover, the risk aversion of FFs (König et al., 2013; Munoz-Bullon and Sanchez-Bueno, 2011), induced by a more general long-term orientation (Gomez-Mejía et al., 2007; Le Breton-Miller & Miller, 2006; Lumpkin et al., 2010), negatively influences activities to innovate. Bianco et al. (2013) argue that investments are significantly more sensitive to uncertainty in FFs than non-FFs. In these cases, FFs reduce their innovative activities to avoid the associated high risks and protect their SEWs (Gomez-Mejia et al., 2007).

Some empirical evidence confirms the validity of the SEW approach by pointing out how specific characteristics of family businesses can either foster or hinder the innovation. Some scholars have observed that the negative influence of familiness on innovation performance is due to the conservatism and risk aversion of family owners (Broekaert et al., 2016; Chen & Hsu, 2009; Nieto et al., 2015), to FFs’ limited resources (Muñoz-Bullón & Sanchez-Bueno, 2011; Nieto et al., 2015) and loss aversion concerning their nonfinancial goals (Gómez-Mejia et al., 2007, 2014). Nevertheless, some FFs are among the most innovative firms in the world (De Massis et al., 2013; Urbinati et al., 2017), and their long-term orientation acts as a stimulus to innovate (Diaz-Moriana et al., 2018).

Here, it is worth noticing that most studies have focused on innovation inputs (Aiello et al., 2020, b; Block, 2012; Brinkerink & Bammens, 2018; Chen & Hsu, 2009; Gomez-Mejia et al., 2007; Kotlar et al., 2014), while a few have considered outputs (Aiello et al., 2021a, 2022; Asaba & Wada, 2019; Bannò, 2016; Block et al., 2013; Chirico et al., 2020; Czarnitzki & Kraft, 2009; Decker & Günther, 2017; Matzler et al., 2015). Output-related papers highlight how FFs behave differently from non-FFs when managing the innovation, and particularly when they strategically protect intellectual property.

In accordance with the SEW perspective, there are numerous reasons that FFs might decide to patent or not. On the one hand, patents and the protection that they provide can be viewed as a means to preserve both the wealth for family business owners and their descendants (Duran et al., 2016; Kellermanns et al., 2008; Zahra, 2005) and their reputation, thus affirming the family name in the market. Moreover, the long protection of inventions assured by patents can be flavored by the long-term orientation typically found in FFs and their business transfer process through hereditary succession (Hauck & Prügl, 2015). On the other hand, as patenting is expensive because of the costs involved (i.e., application/renewal costs and litigation costs), external financial capital is required, thereby diluting the family’s ownership stake in the firm. Moreover, patenting entails disclosing information on the knowledge created, which can lead to losses of critical tacit knowledge, an important source of survivability capital and a critical condition for successful generational succession (e.g. Cabrera-Suàrez et al., 2001). Finally, patenting activities entail greater involvement of specialized human capital, managerial talent and expertize commonly not available within the family (e.g. Chrisman et al., 2014; Verbeke & Kano, 2012).

From the empirical perspective, the few studies examining the link between family ownership and patenting have provided somewhat mixed insights (Table 1). Some authors find that family involvement positively affects patenting (Duran et al., 2016; Matzler et al., 2015), especially in the first generation of FFs (Memili et al., 2015). In contrast, in Bannò (2016), family ownership does not impact the propensity to patent and the citation intensity, respectively. Finally, FFs are less likely than other firms to introduce green patents (Aiello et al., 2021a) or have a negative influence on the number of patents granted (Aiello et al., 2022; Decker & Günther, 2017) and on the number of patent citations (Aiello et al., 2022; Block et al., 2013). Following the theoretical arguments and the claims of most empirical literature, according to which FFs are more conservative, less prone to risk, and more likely to preserve the status quo in terms of SEW and firm control, we contend that family involvement in business can decrease the propensity to engage strategies to protect intellectual property. According to the above arguments, we postulate the following hypothesis:

  • Hypothesis 1: Family-owned firms patent less than non-FFs.

Table 1 An overview of the articles studying the relationship between family ownership and patents
Full size table

2.2 The moderating role of size and age

A potential caveat of prior studies is that some specific factors of firm heterogeneity are not properly considered. Thus, an extension of Hypothesis 1 comes from the awareness that firms are far from being homogeneous.

Starting from the fact that FF is a label that includes a variety of firms (Daspit et al., 2021), many scholars advocate that in order to better explain the performance of FFs, further research should focus on firm heterogeneity (Chrisman & Patel, 2012; De Massis et al., 2014a, b). In this respect, Block et al. (2022) argue that size and age are key variables in specifying the contexts under which FFs outperform or underperform compared to their nonfamily counterparts, thereby becoming good candidates to moderate the relationship between family ownership and innovation.

Although it has been well established that firm size (Acs & Audretsch, 1990, 1991; Cohen, 1995; Cohen & Klepper, 1996a, b; Griliches, 1980; Pavitt et al., 1987) and age (Balasubramanian & Lee, 2008; Coad et al., 2016, 2018; Rossi, 2016) affect innovation, the interest in family business research is limited (e.g., Aiello et al., 2022; Cruz & Nordqvist, 2012; Decker & Günther, 2017; Hillebrand, 2019; Rau et al., 2019; Werner et al., 2018; Zara 2015). Furthermore, as said in the Introduction and highlighted in Table 1, there is no paper that evaluates the joint moderating effect of size and age on the ownership-patenting nexus. In the following, we formulate some research hypotheses with the aim to fill this gap.

2.2.1 Firm size, family firm, and innovation performance

It is widely shown that large organizations have more resources to conduct R&D activities and adopt the innovation (Van de Ven, 1986). This is crucial for patenting as it entails several direct and indirect costs related to developing, attaining, and maintaining patent protection that might dissuade small firms from returning to practices of intellectual protection. Furthermore, larger firms are more likely to patent because of their higher organizational innovation capabilities (Majchrzak et al., 2004). These differences can lead to differing patent strategies.

While these considerations hold true independently of ownership, as far as FFs are concerned, we know that they innovate differently because of their specific characteristics, which can exert either a negative or a positive influence on innovation (cf. Section 2.1). For instance, financial and human capital are two factors that help to understand the link between size and the innovation capacity of FFs. Indeed, larger FFs mobilize human, and financial resources in a suboptimal fashion, thereby weakening their innovation performance compared to their nonfamily peers.

In particular, when 7a business increases in size, more funding is required to innovate, but the search for additional financing is limited by FFs’ reluctance to use external resources (Serrasqueiro et al., 2016). This is because FFs avoid sharing equity with nonfamily members (Sirmon & Hitt, 2003) in order to maintain ownership control in the family (e.g., Gómez-Mejía et al., 2007). Moreover, as a firm grows, any production process increases in complexity, and new skills are needed. Family-owned firms differ from others because they prevent recruitment from outside the family circle, with the consequence of maintaining direct control over strategic decisions and SEW. Nepotism and entrenchment of mediocre family executives limit human capital availability with the result that corporate innovation can suffer from their inadequate expertize (Bertrand & Schoar, 2006; Gomez-Mejia et al., 2001; Pérez-González, 2006). Conversely, recruitment in non-FFs is from the market, that is, from a very large pool of potential candidates who can inject fresh energy and resources that will boost innovation (Nordqvist & Melin, 2010).

From the empirical point of view, Table 1 indicates that the evidence is based on very few papers that analyze the moderating effect exerted by size and suggest that the conditions to patent work differently in large and small firms. For instance, Aiello et al. (2022) shows that Italian FF underperform compared to non-FFs when they are large. Werner et al. (2018) confirmed this for German firms. Evidence is ambiguous for small firms: Aiello et al. (2022) find that small FFs and non-FFs perform similarly, whereas Werner et al. (2018) reveal some advantages for small FFs. Based on this literature, we predict the following:

  • Hypothesis 2: The innovation patent-related gap between FFs and non-FFs increases with size.

2.2.2 Firm age, family firm, and innovation performance

When we consider age, some scholars argue that younger firms have a lower propensity to translate inputs in innovation (Coad et al., 2013; Helfat & Peteraf, 2003). This is due to the low absorptive capacity to use external knowledge (Zahra & George, 2002) and the lack of complementary resources (Bolívar-Ramos, 2017). As companies age, they gain more experience, thereby developing dynamic capabilities, and the ability to survive, with the expected result to be achieved over time (Rossi, 2016).

In looking at the literature focusing on FFs, it is worth mentioning that there is an additional narrative surrounding the role played by age. In many family-oriented papers, age is meant as a proxy of the generational stage (Beck et al., 2011; Fernández & Nieto, 2005; Fiss & Zajac, 2004; Kraiczy et al., 2015; Sánchez-Marín et al., 2020). The reason for this is that the participation of family generations in ownership and management is considered an important source of heterogeneity and thus an essential factor in understanding FFs’ strategic decisions and outcomes (Chrisman & Patel, 2012; Gomez-Mejia et al., 2007).

This literature highlights not only the positive founder effect (Adams et al., 2009; Barontini & Caprio, 2006; Villalonga & Amit, 2006) due to emotional attachment to the firm and its desire to preserve ownership but also shows that FFs benefit more from an age-related learning process because they usually start with a set of competencies and experiences smaller than non-FFs. As the learning process increases knowledge over time, firms improve their ability to adapt and reshape strategy choices as the external environment changes (Cucculelli et al., 2014).

This positive effect is reduced in the post-founder generation since generational involvement alters the dynamics among family members, heightening conflicts within FFs (Gersik, 1997; Le Breton-Miller & Miller, 2013). All this could restrict the learning capacity by impeding knowledge integration (Chirico & Salvato, 2008), thus affecting innovation performance. Similarly, family ownership impedes innovation output in companies as they become older, while nonfamily-owned firms do not suffer from this age effect (Decker & Günther, 2017). Rau et al. (2019) corroborated these findings. In contrast, Zahra (2005) points to a positive generation–innovation relationship by revealing that successors can contribute new knowledge to their FFs, facilitating the identification of novel market opportunities and innovation. The same is found in Cruz and Nordqvist (2012) and Hillebrand (2019). Finally, Arzubiaga et al. (2019) find mixed evidence: generational involvement positively moderates the relationship between explorative innovation and firm performance and, at the same time, negatively moderates the relationship between exploitative innovation and firm performance.

From these considerations, it emerges that firm age affects the relationship between FFs and patents. In particular, it is argued that first-generation enterprises benefit more from an age-related learning process and the beneficial effect of the founder. The hypothesis considering the post-founder generation is more controversial, in which firm age could negatively or positively affect innovation. Considering these diverse standpoints, the third hypothesis is as follows:

  • Hypothesis 3: The innovation gap between FFs and non-FFs decreases with age in first-generation firms. In the post-founder generation of owners, the gap could increase, or decrease as firms age.

2.2.3 Joint effect of size and age on the relationship between family ownership and patenting

What we have learned so far is that FFs have certain particularities that render them different from other businesses and that size and age can affect their innovative behavior in several ways. In line with the above, we recognize the importance of size and age in predicting the impact of family ownership on innovation, but we believe that this relationship is more complex than that highlighted by prior research. Additional insights can be gained by assessing the impact on performance jointly determined by ownership, size, and age.

Considering hypotheses H1, H2, and H3, we expect that the innovation gap between FFs and non-FFs will decrease for smaller and younger firms. On the one hand, the first generation of FFs benefits more from an age-related learning process and on the other hand, small FFs do not experience the disadvantages related to the suboptimal use of human and financial resources that they experience when they increase in size.

Conversely, we expect that the patent gap between FFs and non-FFs could increase for larger and mature firms. Indeed, in the post-founder generation, the beneficial effect of the founder disappears and generational involvement alters the dynamics among family members, heightening conflict within FFs, and thus affecting innovation performance. Conversely, nonfamily-owned firms do not suffer from this age effect. Furthermore, being a large company means to operate in a context that is characterized by higher complexity and formalization. In this case, the professionalization of the organization becomes imperative and more difficult to satisfy for the FFs because they tend to recruit inside the family circle, thereby making it hard for an FF to effectively innovate compared to a non-FF.

It is more difficult to anticipate any clear relationship between FFs and firm performance when we consider other ‘types’ of firms (for instance, smaller, and older, larger and younger) because of the tensions between advantages and disadvantages. Hence, we predict the following:

  • Hypothesis 4: The innovation gap between FFs and non-FFs decreases as the size and age increase in first-generation small firms, while it will increase as the size and age increase in post-founder generation large firms.

Figure 1 illustrates a summary of the conceptual model and hypotheses.

Fig. 1
figure 1

Conceptual model and hypotheses

Full size image

  • H1 Family ownership effect on innovation.

  • H2 Moderation effect: size effect on the family ownership-innovation nexus.

  • H3 Moderation effect: age effect on the family ownership-innovation nexus.

  • H4 The joint moderation effect: size and age effect on the family ownership-innovation nexus.

3 Methods

This section presents the data (§ 3.1) and describes the variables (§ 3.2) used throughout the paper.

3.1 Data and sample description

We used data obtained from the Orbis Europe database of Bureau van Dijk, which comprised an initial panel of about 26,000 firms that applied for at least one patent with the European Patent Office (EPO) between 1981 and 2017. The patents are from the Orbis Intellectual Property (Orbis IP) dataset (Bureau van Dijk), which was linked to PATSTAT released by the European Patent Office (EPO). The Orbis IP dataset makes available a unique firm identifier that allows matching between firm-level patents and balance sheet data contained in Bureau van Dijk’s Orbis Europe archive, which also provides information on the ownership structure of the firms. The sample is composed of 22,541 observations from about 3,500 companies observed from 2010 to 2017.

Table 2 shows the sample distribution among FFs (46.22% of the sample) and non-FFs (53.78%). As far as size is concerned, FFs are concentrated in the group of small and medium enterprises, while nonfamily companies are mostly medium and large. There are no differences between the two groups of companies when considering firm age: the sample is mainly composed of firms whose age is between 0 and 30 years (60.28%).Footnote 1 There is a high concentration of firms in the medium high-tech companies (49.55%), located mainly in the north of Italy (84.72%), which is the most industrialized area of the country. The data reveal that the distribution of FFs and non-FFs does not substantially differ when considering geography and industry composition.

Table 2 The distribution of the sample of firms
Full size table

3.2 Variables

The dependent variable to be analyzed is the yearly number of patent applications filed by firms. Although patents do not fully capture firm innovation, they are commonly used to measure firm-level innovation because they are a relatively homogeneous indicator of innovative activity (as innovations have to satisfy specific requirements to be patented).Footnote 2 The key explanatory variables are family ownership, firm size, and age (see the appendix Table 6 for a description). The literature proposes a variety of ways to classify FFs (Astrachan et al., 2002; Gersick et al., 1997). Although there is no universal consensus on how to classify a firm as an FF, most scholars agree that the family must exercise more authority over corporate decision making than others (Chrisman et al., 2005; Chua et al., 1999) and must subsequently hold a substantial fraction of the company’s shares. In this vein, the criterion used in this paper to differentiate FFs from non-FFs is the absolute ownership majority (i.e., more than 50% of capital shares), which assures control of the company. This criterion is commonly used for privately held companies (e.g. Arzubiaga et al., 2018; Broekaert et al., 2016; Classen et al., 2014; Memili et al., 2015; Meroño-Cerdán et al., 2018; Miller et al., 2013). The choice is well suited to our case because the structure of firm ownership in Italy is characterized by a limited number of shareholders with numerous block holdings (Baltrunaite et al., 2019), thereby implying that a 50% stake is enough to achieve control of the company (Miller et al., 2013). Similar conclusions can be drawn from data and the argumentation provided by Bianco et al (2013) and the Italian National Institute of Statistics (ISTAT, 2020).

While the log of sales (in millions of euros) is the main measure of firm size, other variables (total assets and number of employees) have been used to check the robustness of results. Firm age is expressed as the number of years since the company was established.

To control for other effects, regressions include the following controls: (a) an index of industry specialization, and to this end we classify manufacturing industries into categories based on technological intensity proposed by the OECD (2011); (b) the firm location at the level of macro regions (NUTS 1); and (c) time effects (year dummies).

Finally, the stock of patents is used to control for the effect of knowledge accumulationFootnote 3 and the debt ratio between total liabilities to total assets to control for the impact of firms’ financial constraints on patenting. Here it is worth pointing out that the variables size, stock of patents, and debt ratio are included with a one-year lag to consider the likelihood that these factors will affect the propensity to patent with some lag.

Table 3 displays the summary statistics of the variables used in this study, distinguishing between family-, and non—family-owned firms. The correlations between the variables are in the Appendix Table 7.

Table 3 Descriptive statistics
Full size table

From Table 3, it emerges that family-owned firms differ significantly from nonfamily-owned firms in terms of the number of patents (0.17 versus 0.56). This drives the differences in the stock of patents. The two groups also differ in terms of size: the sales are on average 11.94 million euro for FFs and 95.41 million euro for non-FFs. It merits to be said that size of FFs is less dispersed than that of non-FFs (the standard deviation is 18.65 against 547). When considering age and the debt ratio, the two subsamples are similar.

3.3 Empirical strategy

The use of patent data can be problematic since they are counts (i.e., nonnegative integers) and typically present a zero-inflated distribution (Hausman et al., 1984). We apply regression models that can account for these characteristics of the data. Among the two standard approaches, the Poisson regression and the Negative Binomial (NB) model, only the latter allows data showing overdispersion to be handled, i.e., the mean is smaller than the variance (in our study: mean = 0.38; variance = 2.89).

However, zero patents are probably from two different data-generating processes: (a) firms that do not innovate at all and (b) firms that attempt to innovate but fail to generate patents. It is important to distinguish between these two types to better understand the different innovation behavior of FFs compared to non-FFs.As our dependent variable exhibits several zeros, meaning that many companies did not patent over the entire period, we applied the Zero-Inflated version of the Negative Binomial model(ZINB).Footnote 4

In the ZINB, the probability of choosing to innovate or not is modeled in the inflated part through a logit model (B). The innovation outcome, i.e., the quantity of patents, is modeled using an NB process. The log likelihood function is thus constructed by combining the two processes. ZINB supplements NB density fNB(∙) with a binary process that has density fB(∙). When firms do not attempt to innovate at all (i.e., the number of patents has a value of 0 with a probability of 1), the binary process takes a value of 0 with the probability fB(0).

When firms attempt to innovate, i.e., the patent count takes values of 0, 1, 2, …, the binary process takes a value of 1 with the probability fB(1), of 1 − fB(0). The density function is defined as

$$g\left({pat}_i\right)=\left\{\begin{array}{l}f_B\left(0\right)+\left(1-f_B\left(0\right)\right)f_{NB}\left(0\right)\,if\;{pat}_i=0\\\left(1-f_B\left(0\right)\right)f_{NB}\left({pat}_i\right) \;\;\;\;\;\;\;\;\;if\;{pat}_i\geq1\end{array}\right.$$

where fB(.)can be parameterized through a binomial model like logit and fNB(.) is NB density given by

$${f}_{NB}\left({pat}_{i}\right)=pr\left(Y={pat}_{i}|{\mu }_{i},\alpha \right)=\frac{\Gamma ({pat}_{i}+{\alpha }^{-1})}{\Gamma ({\alpha }^{-1})\Gamma ({pat}_{i}+1)}{\left(\frac{1}{1+\alpha {\mu }_{i}}\right)}^{{\alpha }^{-1}}{\left(\frac{\alpha {\mu }_{i}}{1+\alpha {\mu }_{i}}\right)}^{{pat}_{i}}$$

where \(E\left({Y}_{i}|x\right)={\mu }_{i}=\mathrm{exp}({x}_{i}\beta ),\) and the parameter α is the NB overdispersion parameter.

To model the discrete choice to patent or not, we refer to a logit model:

$${f}_{B}\left(0\right)=G\left({z}{\prime}\gamma \right)=\frac{1}{1-\mathrm{exp}(-{z}{\prime}\gamma )}$$

where z includes variables that determine whether a firm chooses to innovate or not, and γ are the corresponding coefficients. The likelihood function becomes

$$L\left(\beta ,\gamma ,{x}_{i},z,{pat}_{i}\right)=\sum_{{pat}_{i}=0}ln{\left[\mathrm{exp}\left({z}{\prime}\gamma \right)+(1+\alpha \mathrm{exp}\left({x}_{i}\beta \right))\right]}^{{-\alpha }^{-1}}+\sum_{{pat}_{i}>0}{\sum }_{j=0}^{{pat}_{i}-1}\mathrm{ln}(j+{\alpha }^{-1})+\sum_{{pat}_{i}>0}\left\{-\mathrm{ln}\left({pat}_{i}!\right)-\left({pat}_{i}+{\alpha }^{-1}\right)\mathrm{ln}\left(1+\alpha \mathrm{exp}\left({x}_{i}{\prime}\beta \right)\right)+{pat}_{i}\mathrm{ln}\left(\alpha \right)+{pat}_{i}*{x}_{i}{\prime}\right\}+{\sum }_{i=1}^{n}\mathrm{ln}(1+\mathrm{exp}\left({z}{\prime}\gamma \right))$$

Vector x includes those variables used in the NB model and vector z in the logit model, both described in Sect. 3.2.

4 Results

4.1 Looking at the estimated coefficients

Before presenting the results, it is worth highlighting that we have employed the modified Park test to select the appropriate distribution family for patents (Manning & Mullahy, 2001a, b), with the result that the gamma distribution fits better the data.Footnote 5 This said, the following analysis is based on the estimations that have been obtained by performing a ZINB regression, which is the most suitable method to deal with our patent data (cf. footnote 4).

Table 4 presents the results associated with six model specifications (Models 1–6). The top part of the table presents the coefficients obtained when using the NB to study the number of patents (i.e., the intensive margin of innovation), whereas the middle part of the table presents the average marginal effects (AMEs) of the key variables (FFs, size, and age). The bottom part of the table reports the logit results of the inflated equation. It analyses the likelihood of observing (excess) zero patent counts and yields extensive margins of innovation, i.e., the likelihood of entering the patenting market.

Table 4 Patent equations: results from a zero-inflated model
Full size table

Column 1 of Table 4 refers to the baseline Model 1, which includes only the ownership variable for FFs and the controls. Model 2 adds size and age to capture their individual impact. Models 3 and Model 4 are augmented with the product terms FFs*Size and FFs*Age, respectively. Model 5 is a two-way interaction model with FFs*Size, FFs*Age, and Size*Age. To better explain how size and age jointly affect the relationship between FFs and patents, Model 6 considers three-way interactions.

In the following section we focus on the results of the second-stage NB distribution because it governs the actual realization of the outcome.Footnote 6 The estimates obtained from the Model 1 indicate that without controlling for size and age effects, family ownership has a significant and negative impact on the number of patent applications (Table 4). In particular, the FFs show a lower propensity to patent (-0.738), meaning that they obtain on average 0.478 times [0.478 = exp (-0.738)] as many patents as non-FFs.Footnote 7

In Model 2, size, and age enter as fixed controls. The estimated coefficients and the AMEs indicate that the intensive margin of patents increases with size and decreases with age, meaning that young firms have a higher ability to make the most of inputs in patenting. What is important in our discussion is that controlling for size and age does not alter the sign of the ownership effect, albeit the magnitude of the effect varies considerably (i.e., passing from Model 1 to Model 2, the AME associated with family reduces from -5.429 to -0.224).

Compared to Model 2, Models 3, and 4 assess the moderating role of size and age, respectively. It is found that size moderates the effect of family ownership on patenting negatively: in Model 3, the product term FFs*size has a negative and significant coefficient (-0.099), suggesting that the association between firm size and innovation performance will be more negative for family versus non-FFs as size increases. Conversely, the interaction of FFs and firm age has a positive result (0.114 in Model 4), meaning that, ceteris paribus, the patent gap of FFs reduces with age.

In Model 5, age, and size jointly moderate the ownership effect on patents. It is a two-way interaction model and thus, age, and size interact with each other. With respect to previous results, the effects of FFs, size, and age remain substantially unchanged, and the product terms Family*Size and Family*Age maintain the sign and the significance (the interaction Family*Size becomes -0.169 and the interaction Family*Age becomes 0.429).

The evidence related to the key research question, that is, how family, age, and size simultaneously impact patenting, is provided by Model 6. This is because compared to Model 5, the three-way interaction model is an improvement in terms of fit. Indeed, we performed a likelihood ratio test under the null hypothesis that the three-way regression does not result in a statistically significant improvement compared to the two-way model. The chi-squared statistic is 11.61 (p-value = 0.003), thereby indicating that the null hypothesis can be rejected and that Model 6 is the best performing regression. Furthermore, the three-way interaction term presents a negative and significant coefficient (-0.146), meaning that the ownership effect on patents varies across different combinations of size and age (column 6 of Table 3). The check on the significance of the coefficient associated with the triple interaction is an additional test to accept Model 6 as the best performing specification (Kaufman, 2019).Footnote 8

4.2 Family firms patent less than nonfamily peers

From the previous discussion, the first result to be highlighted concerns the ownership effect. Whatever the model, we find that FFs patent less than non-FFs: the AMEs reported in the middle part of Table 4 are always negative and significant in all models. Thus, we provide further empirical evidence to the previous literature on the negative effect of family involvement (Aiello et al., 2021a, 2022; Bannò, 2016; Block et al., 2013; Decker & Günther, 2017) and confirm that FFs can encounter greater difficulties in innovating (König et al., 2013).

Based on this, it emerges that Hypothesis 1 is confirmed and holds regardless of the model specification.

However, it is of interest to highlight that in a nonlinear model, the coefficient estimates and AMEs do not provide answers to many questions of interest (Ai & Norton, 2003; Karaca-Mandic et al., 2012; Mize, 2019). For instance, they do not tell us where (if anywhere) there are significant differences between FFs and non-FFs across the range of size and/or age. It is possible that there are these differences only at certain values or that they hold across all levels of the moderators. This is to be verified empirically by examining in depth what happens at any value of size and age. Additionally, even when the patent gap is confirmed, regardless of the moderators, it is important to investigate whether the gap in magnitude differs at distinct points of size and/or age. To address these issues, the following sections focus on the moderating role of size and age.

4.3 Size positively moderates the relationship between ownership and patenting

This subsection discusses the evidence concerning the moderating role of size in explaining the patenting gap of FFs. We refer to the estimates of Model 3 and convey the results of family size and of its interactions in a graphical interpretation (Fig. 2).

Fig. 2
figure 2

Factor effect of firm ownership moderated by size

Full size image

Figure 2 shows the factor-change effect of FFs as a solid line that slopes downward as size increases (it is traced by fixing age at its mean value). The interpretation is that the innovation gap increases as size increases, confirming the evidence from other studies (Aiello et al., 2022; Athreye et al., 2021; Hall et al., 2013; Pajak, 2016; Scherer, 1965; Symth et al., 1975). Interestingly, the vertical line at size = 2.5 marks the change from a nonsignificant to a significant moderating effect of size. Small firms (size < 2.5) perform similarly regardless of ownership. As size increases, FFs tend to have a lower number of patents than non-FFs, highlighting that size amplifies the disadvantages of familiness. For example, at a size value of 6, that is, with sales amounting to 400 million euro, FFs would be expected to have significantly fewer patents than non-FFs with about 0.7 times as many. These findings support Hypothesis 2.

4.4 Age moderates the relationship between ownership and patenting

Here, we refer to the estimates of Model 4 to present the age-moderating effect. Figure 3 shows the factor-change effect of FFs as a solid line that slopes upward as age increases, supporting the proposition that firm age has a positive moderating effect on patenting. In the same vein as what emerges when studying the size effect (cf. 4.3), Fig. 3 highlights that the age effect is not always significant. Indeed, the innovation gap decreases as age increases in the first-generation firms (the effect of FFs is significant up to the about 3.5 level of age, i.e., 33 years), while in the post-founder generation, age does not impact the patent gap. Stated differently, mature FFs, and non-FFs seem to behave similarly.

Fig. 3
figure 3

Factor effect of firm ownership moderated by age

Full size image

These results support Hypothesis 3 that the relationship between FFs and patents is conditioned by age and that first-generation enterprises benefit more from an age-related learning process and from the beneficial effect of the founder (Adams et al., 2009; Barontini & Caprio, 2006; Villalonga & Amit, 2006). It is of interest to note that this evidence holds under the condition that size is fixed at a given value (Fig. 3 is traced by fixing size at its mean value). In what follows, we explore what happens when size and age vary in their sample values.

4.5 Three-way interaction effect on patenting

What we have shown so far is that size and age distinctly moderate the ownership–patenting nexus. Although these are new results in the family business literature, the three-way interaction model allows us to obtain even broader evidence because it admits that the moderating effect is jointly rather than individually determined by size and age.

Therefore, the interest lies in detecting whether and to what extent FFs’ patenting activities differ from nonfamily peers at different values of size and age (Hypothesis 4). This more in-depth analysis is based on the estimates obtained from Model 6.

Table 4 displays the family ownership effect on the intensive margin of patenting when size and age vary over the sample range. To allow easier reading of results, regions of significance are shaded to differentiate between negative (lighter shades) and positive (darker shades) effects. The shaded area running on the diagonal from the top left to the lower right represents significant negative effects of FFs, that is, FFs' number of patents is less than non-FFs. This area is defined essentially by the combinations of size and age in the bottom half of their respective ranges or in the top half of their ranges.

For example, in the top-left corner, there are firms of minimal size and age. In such a case, the expected log count for FFs is 1.4492 lower than the expected log count for non-FFs, meaning that FFs obtained, on average, 0.2 times [exp (-1.4492) = 0.2347] as many patents as non-FFs.. At the opposite extreme, there are firms of the greatest size and age. The expected log count for FFs is 2.553 lower than the expected log count for non-FFs, meaning that FFs obtain, on average, 0.08 times [exp (-2.553) = 0.08] as many patents as non-FFs.

To convey the results meaningfully Fig. 4 provides a visual representation of the values of Table 4. Figure 4 displays different colorations—from lighter to darker shades—showing how the effect changes from negative to positive or at least to a smaller negative. Furthermore, large diamonds mark a significant effect while small circles denote no significant effect. The four lightest gray shades of the contours represent negative effects of FFs, with the lighter shades show more negative effect. The two darkest shades signify positive effects, with darker shade showing larger positive effect.

Fig. 4
figure 4

The effect of family ownership on the intensive margin of patenting moderated by the interaction of age and size

Full size image

What clearly emerges from Table 5 and Fig. 4 is that the innovation gap between FFs and non-FFs is substantial when size and age are both in the lower or upper parts of their domains. The gap diminishes when size and age are in the middle of their distribution. In other words, the negative family effect is largest for two subgroups of firms: the first includes very small and young firms, whereas the second subgroup refers to very large and older firms.

Table 5 Effect of family ownership on the intensive margin of patenting at different value of age and size.
Full size table

Here, it is worth highlighting that Table 5 and Fig. 4 allow the individual effect of size or age on the FFs to be detected as if they were treated as single moderators. This is similar to what Figs. 2 and 3 allow, but the additional benefit from Fig. 4 is to contemplate every value of each variable rather than only a given value, thereby providing a more general overview of any potential effect.Footnote 9

To isolate how age moderates the effect of FFs on the intensive margin of patenting, we fixed size at a reference value. It is found that the age-moderating effect varies according to firm size: for small companies, the patenting gap decreases with age, whereas for large firms, the gap increases with age. The proof of this is as follows: fixing size at a low reference value and tracing a horizontal path upward (say, i.e., at size = 1), the coloration changes from lighter to darker shades, showing how the family-ownership effect moves from negative to a smaller negative. The opposite occurs for large firms (with, i.e., size = 8). In such cases, the coloration changes from darker to lighter shades. These results enrich the evidence retrieved from Fig. 3, in which age was the only moderator and had a positive impact on the innovation gap.

Similarly, additional insights on the size effect come from comparing Fig. 4 to Fig. 2. On the one hand, it is easy to verify that for young firms (i.e., age = 1.5), the FFs’ patent gap reduces as size increases. This is in contrast with Fig. 2, as the moderating size effect was negative although not significant. On the other hand, for aged firms (i.e., age = 4.5), an increase in size determines a widening of FFs’ patent gap, thereby confirming what Fig. 2 shows.

A more detailed representation of the lower-left corner (smaller and younger firms) and the upper-right corner (larger and older firms) is provided in Fig. 5, which allows to grasp better the differences between firms and to clearly assess what happens if age and size increase jointly. It can be seen more clearly than in Fig. 4 that in the case of small and young firms (Fig. 5.a) the gap decreases (the coloration changes from lighter to darker shades), while in the case of large and mature firms (Fig. 5.b) the gap increases (the coloration changes from darker to lighter shades).

Fig. 5
figure 5

Focus on the effect of family ownership on the intensive margin of patenting moderated by the interaction of age and size

Full size image

In a nutshell, Figs. 4, and 5 help highlight three results. First, there are two areas, i.e., two combinations of age and size, in which FF underperform compared with non-FFs. This occurs for the groups of small and young firms and large and mature firms. Second, in a narrowed area of Fig. 4i, e., in the bottom-right corner, FFs perform more than non-FFs. Here firms are small and old. Third, in the group of large and young firms, i.e., the upper-left corner of Fig. 4, there is no difference between FFs and non-FFs.

These differences in patenting prove that size and age jointly influence the context under which FFs outperform or underperform compared with non-FF peers. As the family ownership effect on patents varies according to the typology of firms, the Hypothesis 4 is confirmed. A strong implication of this result is that the Hypothesis 1 must be validated considering firm heterogeneity. Indeed, this research demonstrates that the sign and magnitude of the ownership effect differ across firm groups and that a three-way interaction model helps to fully grasp the effect of the heterogeneity of FFs on patenting.

5 Discussion and conclusions

5.1 Discussion of empirical results

There are solid theoretical reasons to expect that innovation activities in FFs would differ from those in non-FFs. Differences emerge even among FFs due to their heterogeneity. Given this, our study investigates the relationship between FFs and innovation through the lens of SEW by adopting a context-based approach. In this respect, size, and age are seen as key contingent factors that can explain the variability in firm behavior. We take advantage of prior research that shows size affects the availability of resources and that age is associated with experience and the generational stage of a firm.

Departing from an assessment of the individual roles of size and age, the findings in this study indicate that these two factors exert a joint effect on patenting. As a result, we help to identify the conditions under which FFs are likely to outperform or underperform non-FFs. In this vein, by employing a proper research design to reflect FF heterogeneity we provide conclusions that complement, rather than contradict, prior studies of family business.

This study reveals that being family-owned can be either detrimental or beneficial to patenting. In fact, although we find a baseline negative effect on innovative performance for FFs, the results support the view that SEW constitutes a double-edged sword in FFs, depending on whether its positive or negative aspects prevail. In this respect, we find that firm size and age help to determine which aspect (i.e., positive or negative) of SEW dominates for different FFs.

In particular, we find that among small and young firms, FFs obtain fewer patents than non-FFs and that the innovation gap decreases with size and age within first-generation firms. Here, FFs benefits more from an age-related learning process because these businesses usually begin with a weak set of competencies and an inexperienced founder. Skills grow over time as founders acquire market knowledge, improve their ability to adapt, and reshape strategy choices as the external environment changes (Cucculelli et al., 2014). Additionally, owing to the high level of social capital available to them, FFs gain more from knowledge flows than non-FFs (Arregle et al., 2007). Building social capital takes time; therefore, the relationship between FFs, and patents is expected to vary with firm age.

The innovation gap persists and increases for large and post-founder generation enterprises. This can be explained by the following two considerations. First, as any business grows, its processes become more complex and new skills are needed. FFs avoid recruiting from outside the family circle; therefore, nepotism limits the range of human capital available to them and innovation activities can suffer due to their inadequate expertise (Bertrand & Schoar, 2006; Gomez-Mejia et al., 2001; Pérez-González, 2006). FFs have major financial constraints due to their considerable reluctance to use external resources (Serrasqueiro et al., 2016) in order to retain ownership control in the family (e.g., Gómez-Mejía et al., 2007). Innovation activities are costly and therefore unattractive for FFs with limited financial resources. Second, problems related to an FF’s generational stage can affect post-founder companies (Gersik, 1997). The family’s emotional attachment to the firm and its desire to preserve ownership will likely decline in the post-founder generation. Moreover, any negative effects that emerge among family members can be more damaging than when a respected powerful founder remains in charge, and it is possible (even likely) that some family members will continue to identify and be involved with the firm more than others, which can create conflicts that can compromise the viability of the business (Le Breton-Miller & Miller, 2013). Moreover, if the firm is very large, passing control to succeeding generations often results in a dispersed ownership structure (Van Aken et al., 2017). A principal-principal problem can arise due to conflicts between the members of different nuclear families within an FF. This phenomenon is known as family opportunism: family members make decisions based on their own nuclear household's welfare rather than the welfare of the extended family (Le Breton-Miller et al., 2011; Lubatkin et al., 2005; Miller et al., 2013). Furthermore, family conflicts in later generations can restrict the FF’s learning capacity by impeding knowledge integration (Chirico & Salvato, 2008), thereby affecting innovation performance. Thus, as the size of an FF increases, innovation activities are expected to decline compared to non-FFs. This is not expected in larger non-FFs (Werner et al., 2018).

In summary, SEW is an important reference point for family members when making strategic decisions (Schulze & Kellermanns, 2015; Wennberg et al., 2011; Zellweger & Sieger, 2012). This study shows that size matters, regardless of firm type. However, some factors that are typically related to firm age have a positive influence, while others have a negative effect on innovation performance, suggesting that a hybrid typology of FFs (combining characteristics related to both size and age) is most conducive to patenting. Taking generational stage into account, two facts emerge. For first- generation firms, the patenting gap between FFs and non-FFs decreases as size increases. The beneficial aspects of family leadership related to the founder’s learning process may balance the more complex competencies associated with larger firms. For subsequent generations of owners, increasing size can become a critical factor. As the resources and number of both family and nonfamily employees tend to grow as firm size increases, negative attributes of FFs in terms of family opportunism could dominate.

5.2 Theoretical and practical implications

This study contributes to the ongoing debate about the validity of SEW as a theoretical lens for studying FF behavior in depth. SEW is an umbrella concept that encompasses a range of noneconomic utilities, and much of the behavior of FFs mainly driven by their interest in preserving SEW (Calabrò et al., 2019; Gomez-Mejía et al. al., 2007). Although all FFs possess SEW, they are a heterogeneous group in which positive and negative characteristics of SEW coexist. As SEW theory cannot be uniformly applied, we conclude that contingency effects need to be explored.

Moreover, by stressing the role of contextual factors, our study indicates that to better understand FF innovation the roles played by both age and size must be considered. In this respect, we move the literature a step forward because the evidence presented here is based on how size and age jointly moderate the relationship between firm ownership and innovation performance. The theoretical implications are twofold. First, we shed light on an approach that can better analyze FF heterogeneity. Second, we contribute to an understanding of the characteristics that lead some FFs to be more successful than others in terms of innovation performance. In some ways, by emphasizing the heterogeneous nature of FFs we respond to increasingly frequent calls for a greater focus on FF heterogeneity (Chua et al., 2012; Daspit et al., 2021; Hernández-Linares et al., 2017).

The study also has important practical implications for FFs. FFs must grow in size, but the positive effects of that growth can be outweighed by conflicts between family members when the founder no longer leads the company. If the firm is large, passing control to succeeding generations will often result in a more dispersed ownership structure that can generate conflicts between the members of different nuclear families. Therefore, FFs should balance an emphasis on family knowledge and control with business logic. One strategy would be to recruit talented professional managers from outside the firm who would be charged with making decisions that favor innovation. This issue is particularly critical in Italy, where entrepreneurs are reluctant to formally hand over management of a family firm to outsiders, possibly resulting in severely degraded performance (Aiello et al., 2019; Baltraunaite et al., 2022). It will be important for future studies to replicate our findings in other countries to examine cross-national differences in the attitude toward FFs’ propensity to patent.

5.3 Limitations and future research directions

This study has some limitations that suggest interesting opportunities for future research. First, we use the number of patents as a measure of innovative output. Patents refer to specific innovations, which may not be entirely appropriate for SMEs that do not have the wherewithal to pursue patents, and low-tech industries where innovations are not typically patented and innovation dynamics are measured differently. Future studies could employ other definitions of innovation that use different innovation indicators (e.g., the number of new products and processes) and could distinguish between incremental and radical innovations. The impact of size and age on the innovation gap between FFs and non-FFs may vary in intensity depending on the type of innovation output considered.

Second, we use family ownership as the sole criterion to differentiate between family and nonfamily businesses. Our dataset does not allow us to investigate how results differ when considering founder CEOs, nonfamily CEOs, family-managed, and family-owned firms. This would be an important extension of the research to better understand which family business model is best for firms’ innovation strategies.

Third, cultural differences between FFs in different countries are worth considering. Our data are limited to Italian firms and our results may therefore depend this national context. For example, in Italy, significant weight is placed on family relationships. The current model could be tested in other contexts where family relationships may be less intense to verify the general validity of our findings.

Fourth, some of the concepts used in defining FFs’ advantages and disadvantages—for example, social capital, emotional attachment, and human capital availability—were not directly measured in the present study because of data limitations. Hence, future work could be based on quantitative measures of SEW elements, thereby reinforcing the explanations proposed here.

All of these limitations have their roots in the lack of relevant data, and it remains for future studies to address them.

5.4 Conclusion

In conclusion, this article helps to reconcile previous conflicting findings on the relationship between family ownership and innovative performance by considering firm size and firm age as key differentiators of different types of FFs. The results show how these two variables, individually and jointly, influence innovation performance, indicating the importance of a finer-grained approach to identifying the contexts under which FFs outperform or underperform compared to their nonfamily counterparts.