Family Firms Amidst the Global Financial Crisis: A Territorial Embeddedness Perspective on Downsizing

This study explores the downsizing propensity of family and non-family firms by considering their territorial embeddedness during both periods of economic stability and financial crisis. By drawing on a panel dataset of Spanish manufacturing firms for the period 2002–2015, we show that, all things being equal, family firms have a lower propensity to downsizing than non-family firms. When considering the effect of territorial embeddedness, we found that territorially embedded family firms have an even lower propensity to downsizing than their non-family counterparts. Furthermore, the concern of territorially embedded family firms for their employees’ welfare was particularly pronounced during the years of the global financial crisis. This result is explained by the existence of socially proximate relationships with the firms’ immediate surroundings, based on similarity and a sense of belonging, which push deeply rooted family firms to treat their employees as salient stakeholders during hard times. Overall, our study stresses the importance of local roots in moderating the relationship between family firms and downsizing.


Introduction
How and to what extent do firms resort to downsizing when an economic downturn occurs? This question becomes relevant because of the ethical aspects of firms' downsizing choices in response to adverse external contingencies (Neto & Mullet, 2018). Grounded in agency theory and stakeholder theory (Carver, 2004), the decision to downsize is driven by conflicting moral and legal obligations towards shareholders and stakeholders (Hopkins & Hopkins, 1999).
Thus, the ethical dilemma of downsizing may emerge as a trade-off between preserving psychological and social contracts that bind employers, workers and local communities together (Van Buren, 2000) and owners' financial goals and long-term firm sustainability (Hopkins & Hopkins, 1999). As an example, public opinion, in Spain, highlighted that many family firms survived the 2008 global financial crisis despite sacrificing profitability to save jobs (Arrondo-García et al., 2016). In this sense, the research stream focusing on the family nature of ownership and management has shown that family firms rely less on downsizing than non-family firms (e.g. Bassanini et al., 2013;Sanchez-Bueno et al., 2019;Stavrou et al., 2007) because of stronger non-financial motives, such as moral obligations (Bjuggren, 2015), reputation (Block, 2010) and long-term orientation (Kappes & Schmid, 2013).
Whilst organizational identity arguments have been largely put forward to explain family firms' greater concern for their employees' welfare (Block, 2010;Stavrou et al., 2007), current evidence on family firms and downsizing is essentially "place-less" (Shrivastava & Kennelly, 2013). Specifically, most of existing research (with a few exceptions, see Kim et al., 2019) has overlooked territorial "anchorage" (i.e. embeddedness) where both family and non-family firms are located as a contingent dimension to the decision to downsize. For instance, while family firms were found to provide greater employment stability during the latest global financial crisis (Ellul et al., 2018), evidence for this has been detached from any locational influence besides firm-specific characteristics, which is rather surprising given that firms do not exist in a vacuum devoid of connections with a place (Oinas, 1997).
To address the aforementioned research limitation, we link arguments from the organizational identity theory with the embeddedness perspective-which maintains that individual behaviours and choices are subject to social influences emerging from a flow of interactions and shifting relationships with others (Granovetter, 1985(Granovetter, , 2005-to hypothesize that the relationship between family firms and the propensity to downsize is contingent on the level of the firm's territorial embeddedness. It is defined as the firm's ties to a particular place and the degree to which it is conditioned by the economic and social dynamics occurring in the local milieu (Kalantaridis & Bika, 2006;Pallares-Barbera et al., 2004). Following this reasoning and adding the temporal dimension of crisis, we argue that it is especially in times of adversity, such as the global financial crisis that the deeply rooted connections of family firms with their home territory are likely to materialize (Imperiale & Vanclay, 2016;Smith, 2016), thereby further enhancing their commitment to safeguarding employment.
To test our hypotheses, we rely on a longitudinal dataset of Spanish manufacturing firms for the period 2002-2015. Spain represents a particularly suitable setting for the purpose of our study for two reasons in particular. First, because family firms account for nearly 90% of all firms and contribute to more than 66% of employment and 57% of the GDP (IEF, 2017). Second, Spain was hit badly-in terms of the unparalleled rise in unemployment-by the global financial crisis. While the recession hit European countries symmetrically, in Spain it was particularly acute because the banking sector crisis coincided with the bursting of the housing bubble (Carballo-Cruz, 2011).
Our findings indicate that, all other things being equal, family firms are 3 per cent less likely to downsize than their non-family counterparts. While territorial embeddedness per se does not affect the propensity to downsize, its combination with the family status of the firm provides a marker between the two types of firm. Indeed, territorially embedded family firms are found to be roughly 9% less likely of downsize than their territorially embedded non-family counterparts. However, while during the global financial crisis family and non-family firms were both forced to downsize, the former were 3 per cent less likely to do so. Finally, it is throughout the economic downturn that the influence of territorial embeddedness between the two types of firms was particularly evident. Specifically, we found that the propensity to downsize of territorially embedded family firms was nearly 12 and half percentage points below that of embedded non-family firms during the global financial crisis.
This study makes both theoretical and practical contributions. First, it contributes to the family business literature by unveiling the spatial and temporal contingent conditions (Shrivastava & Kennelly, 2013) in which the implicit contracts between family firms and their employees are honoured (Van Buren, 2000). From this perspective, our empirical findings confirm recent evidence (e.g. Kim et al., 2019) on the conditioning effect of family firms' local roots on the extent to which they resort to downsizing compared to non-family firms. Second, our study contributes to the ongoing debate regarding the territorial foundation of the firm's downsizing choices and corporate social responsibility (Attig & Brockman, 2017;Lähdesmäki & Suutari, 2012). Nevertheless, our findings highlight the need to discriminate firms on the basis of their family status. In this sense, our article sheds new light on the so-called family firm-territory "nexus" (Amato et al., 2021a), interpreted as the existence of embedded ties between family firms and their immediate surroundings (i.e. locality) which have a profound effect on their attitudes towards their employees (Dekker & Hasso, 2016).
Finally, this study has policy implications. The impact of the global financial crisis on the employment contraction has been highly asymmetric both across and within European countries, with some European regions and localities being more resilient than others (Giannakis & Bruggeman, 2017b). The findings allow us to speculate about the composition of the local business structure (i.e. prevalence of territorially embedded family firms) as the foundation of a region's ability to withstand economic downturns (i.e. regional resilience). Hence, public policies should take into consideration the heterogeneity and ubiquity of economic actors (e.g. family firms) when tailoring policy interventions.
This article develops as follows. First, by providing an overview of the literature on downsizing and family firms, and disentangling the effects of territorial embeddedness and the global financial crisis, we present the theoretical background to our hypotheses. Then, we describe our sample, variables and econometric design. Finally, we discuss the results, concluding with remarks and suggestions for future research.

Downsizing and Family Firms
Downsizing, understood as cutting jobs in the interests of firm's efficiency (Cameron, 1994) and profitability in a changing environment (Cascio, 1993;Datta et al., 2010), has received significant attention because of the ethical issues involved. On the one hand, particularly during hard times, moral and legal considerations towards shareholders oblige managers to secure the firm's financial health and long-term sustainability. On the other hand, concerns for employees' welfare prevail on managers to respect implicit and explicit contracts that tie employees to their firms (Hopkins & Hopkins, 1999). Hence, while downsizing is generally perceived as a violation of psychological and social contracts, which bind firms, individuals and their communities together (Coyle-Shapiro & Kessler, 2000), it may be considered fair as long as the decision is communicated to employees in a timely, appropriate and adequately motivated way (Hopkins & Hopkins, 1999). Notwithstanding that, the decision to downsize may also clash with the firm's need to preserve corporate reputation (Love & Kraatz, 2009). That is because firms are generally prone to making choices consistent with prominent aspects of their identities (Ashforth & Mael, 1989). From this perspective, a shared set of values among members of an organization provides meaning and guides the behaviour of its constituents and enables the organization to locate itself in the social environment (Scott & Lane, 2000). Nevertheless, firms differ widely in their set of values and consequently in their behaviours.
Family firms, being a specific type of enterprise where family members are involved in running the business, are in a unique position to develop a distinctive organizational identity because family involvement can affect organizational goals (Basco, 2013) and, consequently, choices such as the decision to downsize (Stavrou et al., 2007). The existing research on family firms and employee-related outcomes (see Table 1) holds that family firms are less likely resort to lay-offs than non-family firms. This hypothesis and, in some cases, the empirical evidence is based on the organizational identity and social identity theory, along with complementary perspectives such as socio-emotional wealth, stakeholder theory and embeddedness.
On the one hand, kinship ties, a shared family name and common history, which are inextricably tied to the firm (Salvato & Melin, 2008), create a sense of belonging and togetherness between family members and the firm, which is generally viewed as an extension of the family itself (Deephouse & Jaskiewicz, 2013). Family values, such as trust, loyalty and mutual support, are usually embraced by non-family members (Zellweger et al., 2010). This, in turn, results in a congruence of values and goals between the family and non-family employees, enhancing motivation and cooperative behaviour (Kappes & Schmid, 2013), fair pay and working conditions (Bingham et al., 2011), trust and empathy (Zellweger et al., 2013), and tighter implicit contracts (Bassanini et al., 2013;Sraer & Thesmar, 2007). For instance, the specific organizational identity created by the family involvement in the firm is reflected in higher employment growth (Diwisch et al., 2009), more employee-related investments (Kappes & Schmid, 2013) and a greater reluctance to cut wages (van Essen et al., 2015) and jobs (Sanchez-Bueno et al., 2019;Stavrou et al., 2007) than non-family firms.
On the other hand, the values and principles shaping a family firm's identity are not confined to the organizational boundaries, but extend outwards to relationships with external stakeholders, thus, becoming a "salient image of the firm in the eye of the public" (Zellweger et al., 2010, p. 58). Hence, heightened identification motivates family firms to engage in decisions that both enhance and preserve corporate reputation as an affective value of family members' selfesteem (Naldi et al., 2013). This is especially true when a firm bears the family name (Deephouse & Jaskiewicz, 2013) and higher exposure of the family firm to the public eye can be a source of either affective benefit or harm for family members (Berrone et al., 2010).
Consequently, based on social identity arguments applied to the family business and the empirical evidence, we infer that family firms are more inclined to preserve implicit contracts with their employees (Bingham et al., 2011;Deephouse & Jaskiewicz, 2013) and avoid reputation-damaging decisions such as downsizing (Block, 2010). Thus, our baseline proposition is as follows: P: The propensity to downsizing is lower for family firms than for non-family counterparts.
However, the aforementioned proposition is chiefly "place-less", since it ignores the local ties between the firm and its home territory potentially influencing family firms' caring attitudes towards their employees (Shrivastava & Kennelly, 2013). With few exceptions (e.g. Greenwood et al., 2010;Kim et al., 2019) current research has overlooked the socio-spatial conditions underlying downsizing decisions. Hence, to understand the differences in family and non-family firms' downsizing propensity, we next focus on territorial embeddedness to disentangle how and when it conditions family firms' downsizing propensity.

Territorial Embeddedness, Family Firms and Downsizing
Rejecting the "under-socialised" view represented by classical and neoclassical economics, the theory of embeddedness conceives economic action and decision-making as being subject to social influences stemming from a flow of repeated and accumulating interactions (Granovetter, 1985(Granovetter, , 2005. Embeddedness has gained growing interest among geographers and regional scholars when investigating the interplay between the firm, the territory and the extent to which the firm depends on its context (Ratajczak-Mrozek, 2017). As a result, territorial embeddedness evolved as a Multiple countries -

Institutional theory
Family ownership (at least 5% of firms' voting rights and is the largest owner) Wages level-Downsizing Family firms are more resilient than non-family firms, which is reflected in the lower probability to downsize or decrease wages in both crisis and pre-crisis periods system interrelated with a given place (Dicken & Malmberg, 2001;Oinas, 1997) incorporating both the firm's geographical location and the fabric of spatially bound relationships established therein (Filippi et al., 2011). The concept of territorial embeddedness is based on the firm's economic linkages with its locality. In this sense, the extent to which the firm's transactions take place mainly in the immediate surroundings reflects its economic dependence on and the strength of the firm's ties with the local milieu (Courtney et al., 2008;Kalantaridis & Bika, 2006). As economic bonds intermingle with the social ones (Uzzi, 1999), enduring and spatially bounded economic activities evolve into a set of localized socially proximate relationships (Lähdesmäki et al., 2019). These relationships are based on a sense of similarity between local actors and on a common feeling of belonging to the same environment, where these connections take place (Jones & Woods, 2013;Pallares-Barbera et al., 2004).
Similarity is understood as the outcome of a spontaneous and cumulative process generated by history and repeated socio-economic actions occurring in a given place (Capello, 2019). Shared social and cultural values enable the reproduction of a local "consciousness" (Paasi, 2002) as a firm's sense of being part of a given place that implicitly draws a boundary around a social community (Pallares- Barbera et al., 2004). Similarity creates a feeling of solidarity or "anchorage" to the local community, bringing together private and collective interests (Capello, 2019), and manifests itself either as informal rules or as tangible social actions towards the firm's immediate surroundings (Paasi, 2002). The logics of similarity and belonging are found to mould the localized network of relationships (Filippi et al., 2011) in the form of reciprocal and trust-based ties binding the community together (Cox, 1998).
Therefore, repeated economic interactions are expected to lead to social relationships based on familiarity, emotional closeness and a sense of obligation (Huber, 2012). This, in turn, influences the firm's consideration of the claims of stakeholders, in particular employees (Lähdesmäki et al., 2019) and its responsibility towards the surrounding community (Capello, 2019;Lähdesmäki & Suutari, 2012). As such, we expect that territorially embedded firms, based on their economic and social ties with the local environment, are less likely to downsize than non-embedded firms. Our first hypothesis is therefore: H 1 : Territorially embedded firms exhibit a lower downsizing propensity than non-embedded firms.
However, some firm-specific characteristics may explain the varying strength of anchorage to the territory which may either amplify or diminish the underlying feelings of similarity and belonging. One of the firm-specific characteristics is family status. Family firms are closely tied to the place where they operate (Bird & Wennberg, 2014), with these connections being not merely economic but also inherently social and affective (Amato et al., 2021a). For family members, the territory represents the place where they grew up, live together and where meaningful life experiences took place (Kalantaridis & Bika, 2006). The family history is inextricably tied to that of the firm, both sharing and concurrently influencing the same space of relationships (Smith, 2016). Hence, the overlapping of family and firm-related narratives contribute to strengthening the firm's territorial roots.
Because of their long-lasting and pervasive action as both entrepreneurs and members of the community, family firms are in a unique position to contribute to and to be influenced by a shared set of social practices, values and traditions prevailing in a given locality (Soleimanof et al., 2018). These become institutionalized in the local setting and the basis for the emergence of an intense local consciousness (Paasi, 2002). However, the awareness of being part of the local community comes with raised social expectations towards the family firm along with monitoring and sanctioning mechanisms on its behaviour (Lähdesmäki & Suutari, 2012). Hence, family firms' local roots turn into spatial loyalty that is a strong "sense of place" (Pallares- Barbera et al., 2004). It consists of an emotional connection stemming from family members' attachment to and identity with a given area, which becomes "closely tied to organizational identity" (Shrivastava & Kennelly, 2013, p. 96) as reflected in their social behaviour (Pallares- Barbera et al., 2004) and in concrete social actions towards their community (Zellweger & Nason, 2008). For instance, previous empirical evidence shows that family firms are frequently involved in philanthropic initiatives (Lähdesmäki & Takala, 2012), local investments (Mitchell et al., 1997), and have stronger corporate social responsibility concerns towards the local community (Albers & Suwala, 2021). Downsizing choices are no exception.
Given the aforementioned arguments, when the economic linkages of family firms take place mainly at local level (Courtney et al., 2008;Kalantaridis & Bika, 2006), their feelings of similarity and belonging to their home territory are heightened, and consequently their awareness of their local responsibilities. Therefore, while we expect territorially embedded firms to be more reluctant to make redundancies than non-embedded firms, we maintain that the influence of territorial embeddedness is stronger for family firms than for their non-family counterparts. Hence, our second hypothesis is as follows: H 2 : Territorial embeddedness influences the relationships between family firm status and downsizing propensity in such a way that the propensity to downsize is lower for territorially embedded family firms than for territorially embedded non-family firms.

Global Financial Crisis, Family Firms and Downsizing
Social identity theory maintains that particular circumstances, such as economic downturns during which firms resort to downsizing (Datta & Basuil, 2015), can trigger a collective response in which individuals rediscover "social identities" (Stets & Burke, 2000). Whereas changed economic circumstances may provide a justification to cut jobs, social identification-which is conventionally associated with intra-group cohesion, altruism and empathy (Ashforth & Mael, 1989)-acts instead as a safeguard for employees. This can be particularly true for family firms, whereby uncertainty and the threat of looming unemployment brought on by an economic downturn may further heighten the family firm's identification with the overall community (Smith, 2016).
Family cohesion is related to the emotional ties that bind relatives together and are necessary to build intragroup relationships and prevent fragmentation (Cabrera-Suárez et al., 2014). A common family identity based on a fundamental agreement about values, beliefs and ongoing interactions is a pre-requisite for such unity (Salvato & Melin, 2008). In adverse conditions we witness a reaction based on the inherent "ability of family members to rely on each other" (Sorenson, 2013, p. 127). Indeed, reciprocal support goes beyond the family domain to encompass those whose livelihood depends on the business (van Essen et al., 2015). This, in turn, strengthens intra-group cohesion and organizational identity. Over time, not only do non-family employees conform to the family values which become incorporated in the business (Sorenson, 2013), but they also become an extension of the owning family itself (Block, 2010). Conversely, massive downsizing may result in a significant loss of goodwill for family firms (Naldi et al., 2013), which might harm family harmony and corporate reputation (Deephouse & Jaskiewicz, 2013). Hence, empathic and altruistic behaviour may compel family firms to prioritize employees' claims even in the face of declining profits.
In the light of the abovementioned arguments, we infer that an economic downturn, such as the global financial crisis, heightens the family firm identity, resulting in tighter long-term implicit contracts to employees. Therefore, our third hypothesis is as follows: H 3 : During the global financial crisis the propensity to downsize is lower for family firms than nonfamily counterparts.

Territorial Embeddedness, the Global Financial Crisis and Downsizing in Family Firms
The impact of the global financial crisis has been highly asymmetric both across and within regions, with some regions and localities better able to withstand the social and economic repercussions (Giannakis & Bruggeman, 2017b). This imbalance has several reasons, including the peculiarities of highly embedded milieu such as found in rural areas, which show a greater ability to react to economic shocks than urban settings (Giannakis & Bruggeman, 2017a). In particular, whereas the social fabric of a given place (i.e. local social capital) is a key factor in regional and local development (Pike et al., 2016;Rutten & Boekema, 2007), the embeddedness of people and firms in territorial networks plays a crucial role in absorbing the impact of economic downturns (Cheshire et al., 2015).
Indeed, socially proximate relationships based on trust and reciprocity are deemed to emerge more vigorously in periods of crisis, as local actors share and mobilize their assets and resources to deal with the negative social and economic consequences (Imperiale & Vanclay, 2016). Local consciousness of being exposed to the same fate breeds solidarity and mutual assistance (Cheshire et al., 2015) and, hence, to a convergence of private and collective interests (Capello, 2019). Consequently, collective strategies enable those affected not only to cope with temporary adversity and distress, but also to reinforce their "sense of community, social cohesion, and social capital" (Imperiale & Vanclay, 2016, p. 216). Thus, both individuals and firms engage in pro-community initiatives involving the maintenance of employment levels. This is particularly true for family firms whereby the social and emotional bonds with their home territory show their worth in times of crisis. For territorially embedded family firms, the welfare of employees as stakeholders becomes a matter of considerable importance (Lähdesmäki & Suutari, 2012). Indeed, even though worsening market conditions may justify lay-offs, the sense of belonging towards the territory (Capello, 2019) can deter embedded family firms from cutting jobs (Kim et al., 2019). Altruistic behaviour during difficult times enhances the legitimacy of both the family and the business in the eyes of the local community (Lähdesmäki & Takala, 2012). And, this legitimacy, upon which the survival and success of the firm largely depends, constrains the employer's options as regards downsizing (Pallares- Barbera et al., 2004).
This constraint is primarily due to a mutual awareness of both family firms and non-family employees that they belong to the same territory, where social and economic relationships take place and are inextricably linked (Lähdesmäki et al., 2019). In particular, as adverse events such as the global financial crisis occur, profit motives yield to an emotional closeness of the family towards non-family employees and the immediate surroundings whose wellbeing and prosperity would be threatened in the event of massive lay-offs (Kim et al., 2019). Thus, despite economic disadvantages, territorially embedded family firms may feel morally obliged to honour implicit contracts (Sraer & Thesmar, 2007;Svenn-Age, 1998) and to support the welfare of the local community (Pallares- Barbera et al., 2004), where both the firm and the family have been established for generations (Salvato & Melin, 2008).
Previous studies investigating the behaviour of family firms during economic downturns found a distinctive influence of the territory on their decisions to downsize. For instance, D'Aurizio and Romano (2013) show that during the global financial crisis, family firms worked harder to preserve employment levels than non-family firms when operating closer to their headquarters. These findings confirm that family firms' intimate relationships with their home territory influence their decisions to cut jobs. More recently, Amato et al. (2020) found that, during the first years of the global financial crisis, family firms located in highly embedded contexts, such as small towns, resorted to making redundancies to a lesser extent than their non-family counterparts operating in the same area. Based on the abovementioned arguments, we can infer that the moderating effect of territorial embeddedness on the relationships between family employers and downsizing is further strengthened during the global financial crisis. Hence, our fourth hypothesis is as follows: H 4 : The moderating effect of territorial embeddedness on the relationships between family firm status and downsizing is higher in crisis periods than in periods of economic stability.

Sample and Data
To test our hypotheses, we have relied on longitudinal microdata of a sample of Spanish manufacturing firms. Spain is an ideal setting for our study for several reasons. First, family firms are the backbone of the Spanish economy, accounting for nearly 90% of private firms and contributing to the nearly 67% of employment and 57% of Spain's GDP 1 (IEF, 2017). Second, while family firms are quite widespread in all sectors, their incidence is particularly marked in the manufacturing sector accounting for 83% of the total (IEF, 2017). Third, Spain was among the European countries worst hit by the global financial crisis. After a period of sustained economic growth powered by the country's integration in the European Union, since the second half of 2007 the Spanish economy started to collapse as a result of the combined adverse effect of the U.S. sub-prime financial crisis and the bursting of domestic real estate bubble (Jimeno & Santos, 2014). As a result, the GDP growth rate contracted sharply (from 3.5% in 2007 to − 5.4% in 2013) with staggering increases in both national debt (from 27% of GDP in 2008 to 80% in 2013) and unemployment (from 8.3 to 27% in 2013) (Royo, 2020). Labour market rigidity and the prevalence of temporary contracts exacerbated further job losses in Spain throughout the period 2007-2013 (Jimeno & Santos, 2014).
Specifically, this study exploits data from the Survey on Business Strategy ("Encuesta sobre Estrategias Empresariales", ESEE) which is conducted yearly by the SEPI Foundation 2 on a representative sample of Spanish manufacturing firms. The ESEE aims to collect a wide range of information on the strategies, manufacturing and innovative processes, international scope and accounting data of firms. It additionally provides detailed information on firms' employment structure by differentiating their workforce according to the type of employment contracts, professional categories and education. For these reasons, the ESEE appears to be particularly suited for an investigation into downsizing. The ESEE is based on a representative sample because it includes all firms with more than 200 employees, whereas those employing 10-200 employees are selected through a random sampling method. 3 The initial sample consists of 89,056 firm-year observations distributed across 5,556 firms for the period 2000-2015. After deleting observations with missing data, we obtain a final sample of 21,643 firm-year observations for the period 2002-2015 consisting of 3,330 firms across 20 manufacturing industries at two-digit NACE Rev. 2 4 and 17 Spanish autonomous communities at NUTS-2 level. 5 Table SD1 in the supplemental online data shows the sample distribution by industry and by region across the years. Panel A indicates that Fabricated metal products (2,763), Food & Tobacco (2,247) and Non-metal mineral products (1,532) have the highest number of observations. Panel B shows that observations are mostly concentrated in Cataluña (4,555), in Community of Madrid (3,021) and in Valencian Community (2,920).

Dependent Variable
Following the previous literature (Block, 2010;Cascio et al., 1997), we measure Downsizing as a dichotomous variable taking value "1" if the firm's workforce decreased by more than 5% compared to the previous year, whereas positive or no changes are set at "0". Hence, in the latter case, firms are regarded as "stable employers". 6

Independent Variables
Family Firm To identify family firms, we relied on the socalled "demographic approach" (Basco, 2013), which considers the involvement of family members in ownership, governance and managerial positions as a sufficient condition to capture the influence of the family on the business. For each surveyed firm, the ESEE provides information on the number of family members in managerial positions. Following previous studies (Greenwood et al., 2010;Sanchez-Bueno et al., 2019), we measure Family firm as a binary measure coded "1" if one or more members of the owning family work in the top management team of the focal firm and "0" otherwise.
Territorial Embeddedness Given the nature, depth and the extent of a firm's ties to the local area (Greenberg et al., 2018;Kalantaridis & Bika, 2006), territorial embeddedness emerges as an elusive construct particularly for measurement purposes (Oinas, 1997;Ratajczak-Mrozek, 2017). Notwithstanding that, we follow Dekker and Hasso (2016) and Greenberg et al. (2018) who maintain that a firm's anchorage to its territory is stronger the more the firm depends on its immediate surroundings for its sales, whereas a firm's ties to its home base weaken as it expands further afield. Therefore, in accordance with Kalantaridis and Bika (2006), our measure mainly captures the market-oriented facet of 3 ESEE has paid attention at the attrition bias across years. This occurs wherever participants leave the study, resulting in systematic differences between the group that leaves and the one that continues the survey. For instance, if a firm answered the survey in 2002 and 2003 but refused to participate in 2004, this firm was surveyed again in [2005][2006]. The response rate for the period 2002-2015 was equal to 91%. 4 NACE is the abbreviation for "Nomenclature statistique des activités économiques dans la Communauté européenne" and represents the European standard classification of productive economic activities. Particularly, ESEE adopts the NACE Rev. 2 classification implemented in 2006. For more information on NACE classification, please refer to: https:// ec. europa. eu/ euros tat/ web/ nace-rev2. 5 NUTS stands for "Nomenclature of Territorial Units for Statistics" and represents the level of territorial division for statistical purposes. The Spanish territory is divided in the following levels: NUTS 1 consists of seven groups of autonomous communities (Agrupación de comunidades autónomas); NUTS 2 comprises 19 autonomous communities and cities (Comunidades y ciudades autónomas); NUTS 3 is made up of 59 provinces and islands, (Provincias, Islas). However, the ESEE excludes the autonomous cities of Ceuta and Melilla, thus, leaving 17 autonomous communities. For more information on NUTS classification, please refer to: https:// ec. europa. eu/ euros tat/ web/ nuts/ backg round. 6 Since the number of employees consists of self-reported information, some might rightly argue the risk of social desirability bias, which occurs whenever survey respondents (e.g. individuals and firms) intentionally provide answers that they consider socially acceptable with the aim of gaining approval from others (Roxas & Lindsay, 2012). In our case, firms may enhance their social desirability as good employers or to avoid disapproval in the case of massive downsizing by misreporting the number of employees. However, the anonymity of surveyed firms of ESEE mitigates the risk of social desirability bias in our study. Indeed, as survey research shows (Krumpal, 2013;Roxas & Lindsay, 2012), the guarantee of anonymity and confidentiality is an effective way to promote full disclosure of sensitive or stigmatizing information compared to non-anonymous methods, thereby considerably reducing the risk of selection bias in survey research. territorial embeddedness. Specifically, each firm taking part in the survey is required to indicate the geographical area of its main market by choosing between six alternatives, namely: (i) local; (ii) provincial; (iii) regional; (iv) national; (v) abroad and (vi) domestic and abroad. In line with Dekker and Hasso (2016), we operationalize Territorial embeddedness as a dichotomous variable taking value "1", if the local area is the main market for the company and "0" if it also operates in other markets. 7 We restrict the measure of embeddedness to the local level only because it delimits the closest administrative territory where both firms and people are situated and carry on their everyday working life (Paasi, 2002). 8 Additionally, and more importantly, local areas possess a so-called "imagined coherence" in the sense that their residents (e.g. individuals, families and firms) have a particular sense of identity tying them to the place, resulting in a perceived community with a shared pattern of behaviour (Jones & Woods, 2013).

Global Financial Crisis
Even though the shock affected the whole period 2007-2015 with severe consequences in terms of job losses, the unemployment rate in Spain was uneven both across the economy as a whole and within the abovementioned period. Specifically, the average unemployment rate of manufacturing was generally below the unemployment rate in agriculture, construction and services. Based on the Economically Active Population Survey (EAPS), 9 the number of unemployed increased systematically during the period 2007-2012 within manufacturing. The slow recovery of the labour market started in 2013. Therefore, for the purpose of our study a dichotomous variable named GFC which takes value "1" for the years 2007-2012 and "0" otherwise is built accordingly.

Control Variables
To deal with unobserved heterogeneity, we control for a wide set of variables potentially affecting downsizing. We control for firms' innovative activity as expressed by R&D expenditure to sales (R&D intensity) and innovation achievements in terms of both Product and Process innovations (dummy variable which takes value "1" if a firm introduced product/process innovations, "0" otherwise) (Kappes & Schmid, 2013). As a firm's international scope may affect the level of employment, we control for Export intensity as ratio of foreign sales to total revenues (Block, 2010). The variables Age (number of years since the firm was founded) and Size (logarithmic transformation of total assets) control for the effects related to the size and life cycle of the firm (Block, 2010;Sanchez-Bueno et al., 2019). As downsizing is usually a consequence of a firm's financial distress, we introduce Financial constraints as debt to total assets (Lins et al., 2013). To account for differences in performance, we control for Profitability, expressed as gross operating margin (Stavrou et al., 2007). The variables Listed (dichotomous variable which takes value "1" if the firm is publicly traded, "0" otherwise) and Group (dichotomous variable which takes value "1" if the firm is part of a corporate group, "0" otherwise) are introduced to account for constraints and requirements of employee-related policies and intra-group workers reallocation, respectively (Block, 2010). As businesses with high labour costs are more likely to downsize to reduce their overall expenditure, we control for Labour cost ratio computed as the ratio of labour costs over total sales (Sanchez-Bueno et al., 2019). To account for investment in human capital, we control for Employee T&D measured by total expenses in training and development on labour costs (Baù et al., 2019).
Furthermore, as the reliance on temporary and part-time contracts may provide flexibility during demand fluctuations in terms of employment adjustment, we control for Temporary workers ratio (ratio of temporary workers over total workforce) and Part-time workers ratio (ratio of part-time workers over total workforce) (Muñoz-Bullón & Sánchez-Bueno, 2014). Since downsizing may result from the competitive pressure of rival firms (Gandolfi & Hansson, 2011), we control for the number of competitors in the main market of the company (Competitors). Rural areas are regarded as localities particularly rich in terms of social capital where family and community networks result in localization advantages for family firms (Backman & Palmberg, 2015). Hence, we control for Rural as binary variable which takes value "1" if the firm is located in towns or places with less than 50,000 inhabitants, "0" otherwise (OECD, 2012). Finally, a series of dummy variables are introduced to control for the industry, region and year associated with each observation.

Empirical model
To test the proposed hypotheses, we used panel data analysis with random effects specification. Since we have an unbalanced panel, we checked for the stationarity of our data using the Fisher unit-root test. 10 The null hypothesis is that all the panels contain a unit root, that is, all series are non-stationary. As we obtained p values lower than 0.05, we rejected the null hypothesis with all series consequently being stationary.
To investigate the decision to downsize with respect to the family status of the firm, the existence of territorially embedded ties and the global financial crisis, we estimate the following model: where our dependent variable Downsizing represents the downsizing propensity; α 0 is the constant; β 1 , β 2 and β 3 represent the direct effect of our variables of interest on Downsizing; X i,t is a matrix containing all two-and three-way interaction terms for our three main regressors; γ is the corresponding vector of coefficients; C i,t-1 is a matrix of additional control variables that captures firms' heterogeneity potentially influencing downsizing choices, lagged by one year to lessen endogeneity from simultaneity bias (Block, 2010;Sanchez-Bueno et al., 2019); δ is the vector related to coefficients; T i , S i and R i are time-specific, industry-specific and region-specific dummy effects, respectively; τ, φ and ω are the vectors corresponding to coefficients; and ε i,t is the error term.
We estimate the above equation by using a linear probability model (LPM). LPM, a special case of OLS regression, is particularly suited to obtaining consistent and unbiased coefficients in the case of a dichotomous response variable (Wooldridge, 2010). One advantage of OLS over logistic models (e.g. Probit/Logit) is that it eases the interpretation of both the estimated coefficients, which directly represent the marginal effects and the interaction term which is rather problematic with logistic models due to the unclear signs or incorrect standard errors (Greene, 2010). Additionally, the marginal effect obtained from OLS model is very close to those obtained from non-linear models (Wooldridge, 2010). Finally, we deal with heteroscedasticity concerns by estimating robust standard errors.
While the sign and significance of the coefficients of the variables Family firm and Territorial embeddedness are related to the baseline proposition and Hypothesis 1, respectively, Hypotheses 2, 3, and 4 are operationalized by the following interaction terms: Family P Downsizing i,t = 1 = 0 + 1 Family firm i,t + 2 Territorial embeddedness i,t firm*Territorial embeddedness, Family firm*GFC and Family firm*Territorial embeddedness*GFC. For the sake of clarity, we interpret the interaction terms by group comparisons (Dinh et al., 2021). Given the existence of as many groups as possible combinations, a specific reference group is identified. 11 The sign and statistically significance of the marginal effect of a given group in comparison with the reference group provides straightforward evidence of differences across groups. Hence, to investigate the difference between family and non-family firm conditionally to territorial embeddedness, the global financial crisis and in both instances, the groups Non-family firm*Territorial embeddedness, Non-family firm*GFC and Non-family firm*Territorial embeddedness*GFC are compared with the reference groups Family firm*Territorial embeddedness, Family firm*GFC and Family firm*Territorial embeddedness*GFC, respectively. For the purpose of our study, while the choice of reference group leads to the same results, setting family firms as reference group also allows us to compare the same type of firms across different scenarios.

Descriptive Statistics
Panel 2A lists the descriptive statistics for the key regression variables. While 30% of firms resorted to downsizing, family firms account for nearly half of the businesses in our sample. 12 Only 6% of the firms claim to rely mainly on the local area for their sales and, hence, are considered territorially embedded. Finally, the firms in our sample have, on average, been in business for 30 years, with the labour costs accounting for almost 28% of total costs. Finally, more than 60% of firms are located in rural areas.
Panel 2B reports the sample differences depending on the family status of the firm. As the proportion of businesses engaging in downsizing is higher for family firms than non-family counterparts, the former are more territorially embedded than the latter. Additionally, family firms are both less innovative-as reflected in the lower R&D expenditure relative to sales and a lower propensity towards product and 12 The share of family firms out of the total of surveyed firms is fairly constant in each year of the time span considered (i.e. 2005-2015). Even in the years of the global financial crisis (i.e. 2007-2012), family firms' share amounts to 51% of the total, thereby ruling out concerns about sample composition change.
Footnote 10 (continued) type has a null hypothesis that all panels contain a unit root and it is implemented by the command "xtunitroot fisher". 11 In case of a dichotomous variable, Stata allows a change in the reference group with the command b(1))i. placed before the variable of interest. The lowest-numbered group (i.e. 0) is used as the reference group by default from the statistical package. The same applies both in case of single categorical variables and interaction terms. process innovations-and less internationally oriented than non-family counterparts. Finally, a higher proportion of family firms are located in rural areas (Table 2).
In Table 3, we report the correlation between our regression variables. While the association between family firm status and downsizing is positive but not statistically significant, territorial embeddedness is positively and significantly associated with downsizing. Both the global financial crisis and the labour cost ratio are closely related to downsizing. Inspection of the variance inflation factors (VIFs) displayed in the same table suggests that multicollinearity is not a concern in our data, as all VIF coefficients are below the tolerance threshold of 5 (Baù et al., 2019).

Regression Results
We report the results of our analysis in Table 4. 13 In Model 1 we introduce family firm variable along with control variables. The coefficient of Family firm is negative and statistically significant at the 0.1% level, suggesting that family firms-all things being equal-are more reluctant to downsize than non-family firms. Specifically, family firms are 3 per cent less likely to downsize than their non-family counterparts. By looking at the control variables, both process innovation (Process innovation) and R&D activities (R&D intensity) are negatively related to the propensity to downsize. Similarly, greater profitability (Profitability) and staff training (Employee T&D) are negatively linked to downsizing. While the size of the firm (Size) does not seem to have a significant impact on downsizing, greater levels of indebtedness (Financial constraints) do have a positive effect. As indicated by the coefficient of Labour cost ratio, a greater (relative) incidence of personnel costs is positively linked to downsizing. Likewise, we document a positive and negative coefficient of both Temporary workers ratio and Part-time workers ratio, which indicates that non-permanent employment contracts allow firms greater flexibility when deciding to lay-off.
In Model 2 we add the remaining two explanatory variables. The coefficient of Territorial embeddedness is positive but not statistically significant. Therefore, there is no evidence that the propensity to downsize is affected by the degree to which firms are anchored to their local area. Therefore, Hypothesis 1 is not supported. Finally, the coefficient of GFC is strongly positive and statistically significant at the 0.1% level. 14 In particular, the propensity to engage in downsizing during the years of financial crisis is more than 13% higher than in periods of economic stability.
In Model 3 we test Hypothesis 2 by comparing family and non-family firms at equal levels of territorial embeddedness. The marginal effect of the two-way interaction Non-family firm*Territorial embeddedness-as opposed to family firm counterparts as reference group-is positive and statistically significant at 1% level. This result suggests that, when it comes to downsizing, territorial embeddedness affects family and non-family firms in different ways, with territorially embedded family firms much being less likely to downsize. Specifically, for territorially embedded family firms the probability to downsize is 8.8% lower than their non-family and territorially embedded counterparts. Thus, Hypothesis 2 is supported. For a more straightforward interpretation of this result, we plot in Fig. 2a each pair of factors. It shows that the discrepancy-in terms of propensity to downsize-between family and non-family firms widens further when both types of firms are territorially embedded. Conversely, family firms themselves do not seem to differ on account of their embeddedness in the local milieu as shown by the marginal effect of the interaction Family firm*Nonterritorial embeddedness which, despite being positive, is not statistically significant.
In Model 4 we test Hypothesis 3 by comparing family and non-family firms during the global financial crisis. The marginal effect of the two-way interaction Non-family firm*GFC-as opposed to family firms as reference groupis positive and statistically significant at 1%, providing evidence that the two types of firms react differently as regards downsizing during the economic downturn. Particularly, family firms are found 3% less likely to downsize than their non-family counterparts. Additionally, Model 4 shows that the propensity to downsize of family firms was considerably higher (13.8%) during the global financial crisis than in periods of economic stability. Figure 2b depicts each pair of factors associated with the two-way interaction. Yet, even if the propensity to downsize increases for both types of firms during the global financial crisis, the downsizing propensity of family firms is always lower than that of non-family counterparts.
Finally, in Model 5 we test Hypothesis 4 by computing the marginal effect of the three-way interaction Non-family firm*Territorial embeddedness*GFC as opposed to family firms set as reference group. The marginal effect is positive Level of statistical significance + p < 0.10, * p < 0.05, ** p < 0.01, *** p < 0.001 a The Wilcoxon rank-sum test determines whether two samples were selected from populations having the same distribution (Sample 1: Nonfamily firms; Sample 2: Family firms) L Expressed in natural logarithm W Winsor at 1 and 99% tail   and statistically significant at 1% level, providing evidence that throughout the global financial crisis the propensity to downsize of territorially embedded family firms is lower than territorially embedded non-family firms. Specifically, the propensity to downsize of family firms is almost 12 percentage points below that of non-family counterparts. This result confirms that the strength of local roots of family firms is particularly marked during the global financial crisis, thus, supporting Hypothesis 5. To further illustrate this result, we plot the three-way interaction in Fig. 2c. The figure shows that during the crisis, the downsizing propensity of territorially embedded family firms is far lower than their non-family counterparts.

Robustness Check
We performed several robustness tests to corroborate our results. 15 First, we performed a subsample analysis by restricting the observations to the territorial embeddedness, the years of global financial crisis and jointly to the territorial embeddedness and years of crisis, respectively. In all the aforementioned subsamples, the family firm coefficient is negative and highly significant. Second, in lieu of the dichotomic measure of family firms (i.e. Family firm), we employed a continuous measure counting the number of family members in managerial positions. Third, we used an alternative definition of downsizing by raising alternatively to 7 and 10% the percentage threshold of employment reduction as compared to the previous year. Fourth, given the importance of time for firms-regardless of type-to develop and nurture their attachment to the local milieu (Lähdesmäki et al., 2019;Smith, 2016), we omitted firm age from the estimates. Fifth, we used a matching procedure to improve the covariate balance between the treated (i.e. family firms) and the control groups (i.e. non-family firms) to enhance the estimation of the causal effect. In so doing, we implemented the coarsened exact matching procedure (CEM) algorithm (Iacus et al., 2012). 16 Finally, we used logistic regression models (Probit and Logit) to estimate our coefficients with the marginal effects of both Logit and Probit models that are very close to those obtained with LMP. In all the abovementioned cases, we obtained similar results to those reported in the main analysis, thus, confirming our findings.   (Table SD2). 16 In general terms, matching consists of pruning observations that have no close match on pretreatment covariates in both the treated and the control groups. We performed CEM algorithm with the command "cem" on Stata (Blackwell et al., 2009) and matched on four variables capturing the age of the firm, the number of employees, the sales and the industry to which firms belong.

Discussion
Drawing on the perspectives of territorial embeddedness and organizational identity, this study investigates the downsizing propensity of family and non-family firms by considering the moderating effect of territorial embeddedness and the global financial crisis. Previous studies have shown that, when confronted with the ethical dilemma of preserving the firm's financial health or occupational levels (Hopkins & Hopkins, 1999), family firms are less inclined to downsize compared with non-family firms (e.g. Bjuggren, 2015;Block, 2010;Sraer & Thesmar, 2007). However, the joint influence of territorial embeddedness-understood as the embodiment of geographic and social proximity based on a sense of belonging and territorial identity (Lähdesmäki et al., 2019)-and external contingencies on downsizing choices has been largely overlooked. As firms "do not exist in a vacuum devoid of connection to actual locations" (Guthey et al., 2014, p. 259) and temporal context (Arrondo-García et al., 2016), anchorage to their home territory stands out both as a source of opportunities (Capello & Faggian, 2005;Uzzi, 1999) and constraints (Hess, 2004), with territorial embeddedness signalling a state of dependence on the place (Ratajczak-Mrozek, 2017). From this perspective, downsizing choices, both under normal circumstances and economic downturns, may be contingent on the set of economic, social and emotional connections that firms have established with their geographical and social milieu (Capello, 2019;Dicken & Malmberg, 2001).
The table presents linear probability model estimates based on a panel dataset over the period 2002-2015. The dependent variable is Downsizing which is coded "1" if the negative change in numbers of employees as compared to the previous year is higher than 5%. Positive changes are set to "0". Family firm is a dummy variable equal to "1" if one or more family members occupy managerial position in the firm and "0" otherwise. Territorial embeddedness is a dummy variable equal to "1" if the local area represents the main market for the firm and "0" if the firm also operates in other markets (provincial, regional, national, abroad, domestic and abroad). GFC indicates years 2007-2012 during which the global financial crisis occurred. Robust clustered standard errors are reported in parentheses Level of significance + p < 0.10, * p < 0.05, ** p < 0.01, *** p < 0.001 Our results highlight the importance of territorial embeddedness for a greater understanding of the differences between family and non-family firms' propensity to downsize, both in times of economic stability and subsequent downturns. Regarding the general stance towards downsizing, in line with previous studies (Bassanini et al., 2013;Stavrou et al., 2007), we corroborate that family firms have a lower propensity to downsizing than non-family firms. The family's identification with the business and family values, such as closeness, reciprocity and trust, is likely to extend beyond the family domain, thereby heightening the psychological and social bonds between family firms and their employees (Bingham et al., 2011;Deephouse & Jaskiewicz, 2013). While territorial embeddedness per se does not affect downsizing propensity, in the case of family firms it was found to play an important role. Specifically, the strong economic links and territorial identity of family firms turn into a spatial loyalty which further mitigates the propensity to downsize compared to non-family firms. Hence, as local roots provide locational advantages to family firms (Backman & Palmberg, 2015;Baù et al., 2019), our findings support earlier studies showing the association between the embeddedness of family firms in the local milieu and their pro-social behaviour (Berrone et al., 2010;Dekker & Hasso, 2016).
As the global financial crisis forced firms to cut jobs, our results reveal different responses, with family firms found to be less likely to downsize than non-family counterparts. This finding is at odds with that of Lins et al. (2013) who have shown that, during the early years of the global financial crisis (i.e. 2008-2009), family firms were equally likely to downsize as non-family firms. Conversely, this result is consistent with previous studies highlighting the altruistic behaviour of family firms towards their employees in the event of industry-specific shocks (Sraer & Thesmar, 2007) or economic downturns (Lee, 2006;van Essen et al., 2015). Specifically, our study confirms the findings of Arrondo-García et al. (2016) which reveal that during the most acute stages of the global financial crisis in Spain, family firms provided more employment stability than other firms.
However, considering separately the effects of territorial embeddedness and the global financial crisis on family versus non-family firms only offers a partial view of how the contextual dimensions influence firms' downsizing choices. Indeed, when territorial embeddedness and the global financial crisis are considered simultaneously, we found that the concern of territorially embedded family firms for their employees particularly stands out in times of adversity. Socially proximate relationships with the firms' immediate surroundings, based on similarity and affective bonds push family firms to consider the needs of their employees whose well-being and that of the local community would be endangered in the event of massive lay-offs. The difference in the propensity to downsize between territorially embedded family and non-family firms equals nearly 12%, showing that family firms downsized less than non-family firms. This result reveals the amplified spatial loyalty of family firms in time of crisis, which does not appear to the same extent when territorial embeddedness and the global financial crisis are considered separately (8.8 and 3%, respectively). These findings are consistent with D'Aurizio and Romano (2013) and Amato et al. (2020), who have shown the role of location in the decision to safeguard the employment levels of family firms during the early years of the global financial crisis. Hence, our study confirms the uniqueness of territorially embedded family firms in limiting the detrimental effects of economic downturns on employment.

Contributions and Policy Implications
Our study has theoretical and practical implications. First, it contributes to the family business literature examining the downsizing propensity of family and non-family firms (Amato et al., 2021b) based on embeddedness in the home territory and external contingencies (the global financial crisis). In this sense, we attempt to address the place-less research gap in family business studies (Stough et al., 2015). Following the debate into the locational effect on the prosocial behaviour of family firms (Dekker & Hasso, 2016;Kim et al., 2019), we reveal the conditions under which the favourable attitudes of family firms towards their employees are likely to materialize. While previous studies considered the characteristics of the territory where the family firm is located (Greenwood et al., 2010;Kim et al, 2019), we show that the anchorage of family firms in their local milieu clearly affects their propensity to downsize. In this way, our results reveal that the intimate connections between family firms and their immediate surroundings-arising from a deeper sense of place and togetherness with their home territory (Kim et al., 2019)-result in heightened awareness of the needs of their employees, as salient stakeholders, in times of adversity (Van Buren, 2000).
Second, this research contributes to the ongoing debate on the pro-social behaviour of businesses-with respect to safeguarding employment-and localities (Attig & Brockman, 2017;Lähdesmäki et al., 2019). As firms' local responsibilities are linked to geographical areas and socially proximate relationships, our study highlights the importance of distinguishing firms according to their type-that is between family and non-family firms-when the territorial foundations of corporate social responsibility are investigated. In particular, by introducing the spatial scope of embeddedness into the investigation of family firms' downsizing choices (Hess, 2004;Ratajczak-Mrozek, 2017), we corroborate family firms as a distinctive type of enterprise (Shrivastava & Kennelly, 2013), deeply rooted in a given place (Amato et al., 2021a) and committed to balancing survival and success with the well-being of the local community where they are placed (Berrone et al., 2010;Kurland & McCaffrey, 2020).
Finally, our research has implications for policy makers. As the negative repercussions of the global financial crisis spread across European regions in a markedly uneven manner (Groot et al., 2011), the concept of regional resilience, generally understood as the ability of a regional or local economy to withstand recessionary shocks, has attracted increasing interest (Fratesi & Perucca, 2018). In light of our findings, regional authorities should consider family firms' embeddedness and concentration in their home territories when tailoring either swift responses to the impact of economic downturn or place-based policies aimed at rebalancing the territorial effects of economic crisis (e.g. European cohesion policy) (Giannakis & Bruggeman, 2017b). In addition to this, our research highlights the need for more awareness by policymakers of the unique and valuable role of family firms when policies aimed at fostering regional and local growth are designed (Basco & Bartkevičiūtė, 2016;Pike et al., 2016).

Limitations and Avenues for Future Research
Our research has several limitations which could pave the way for future research. First, our definition of family firm does not consider either ownership thresholds or the role of the controlling generation. As previous studies have shown, the consideration of non-economic aspects affecting downsizing are heavily dependent upon both the percentage of common stock owned by the founding family (Block, 2010) and the generation in control (Arrondo-García et al., 2016). Hence, future studies should investigate whether and to what extent territorial embeddedness influences downsizing propensity in different ways depending on the thresholds of family ownership and between the first generation vs. multigenerational family businesses.
Second, as our study relies entirely on a demographic approach to define family firms, future studies should test the consistency of our results with multiple definitions of family firms, integrating the components of involvement and essence approaches accounting for "soft" factors such as the vision and intentions (Basco, 2013).
Third, alternative measures of territorial embeddedness could be employed to investigate whether family firms' downsizing propensity is sensitive to different operationalizations of local roots. Specifically, alternative and complementary measures-based on the distance from headquarters (Berrone et al., 2010;D'Aurizio & Romano, 2013), stakeholder proximity (Amato et al., 2021a), place tenure (Baù et al., 2019) and percentage of inputs and information coming from the local setting (Kalantaridis & Bika, 2006) among other factors. However, the most promising way to face the challenge of measuring a firm's local ties is to validate a "holistic" measure of territorial embeddedness entailing the multiple underlying dimensions, capable of fully grasping the conditioning effect of local milieu on a firm's decision-making and outcomes. 17 Fourth, as our study considers only one country (Spain), future research should expand the analysis to other countries to account for the influence of different institutional and cultural settings on the decision to downsize in response to recessionary shocks (Lins et al., 2013). Since our research does not consider differences in regional labour markets, which could potentially limit a firm's room for manoeuvre as regards downsizing (Svenn-Age, 1998), future research should investigate the sensitivity of territorially embedded family firms to differences in labour markets (Muñoz-Bullón & Sánchez-Bueno, 2014).
Fifth, as our study develops from a micro-level approach, future studies may step into a meso-level perspective to explore the role of the collective aggregate actions (i.e. family firms "density") as a source of regional resilience (Block & Spiegel, 2013). Sixth, in addition to economic downturns, other types of exogenous shocks such as natural disasters may adversely affect the economic and social structure of a given region or locality (Imperiale & Vanclay, 2016). Hence, it would be worth deepening the role of family firms' spatial loyalty in coping with such adverse events (Cheshire et al., 2015;Smith, 2016). Finally, the family firm-territory "nexus" stands out as a promising opportunity for investigation with qualitative research methods (e.g. case studies, in-depth interviews). Indeed, a qualitative approach may prove extremely useful in developing new theories and testing existing ones. Only through an in-depth investigation of the economic and socio-spatial dynamics in context, might it be possible to understand the genius loci of family firms, that is, how the sense of place is produced and evolves across time (Shrivastava & Kennelly, 2013), uniquely influencing the relationships between family firms, territory and prosocial behaviour.
of October 2018 at the Centre of Family Ownership and Enterprise (CeFEO) of Jönköping International Business School (Sweden), we would like to thank its director, Francesco Chirico, and his colleagues for their hospitality.
Funding Open access funding provided by Scuola IMT Alti Studi Lucca within the CRUI-CARE Agreement. This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.