Ownership structure and corporate social responsibility in an emerging market

While scholarship exploring the impact of ownership structure on corporate social responsibility (CSR) has investigated firms in developed markets, less work has examined how ownership in firms from emerging markets influences community-related CSR. Both internal and external forces potentially drive community-related CSR decisions. It is hence important to understand the role of internal constraints arising due to agency problems along with institutional pressures from external stakeholders in emerging markets in shaping CSR. In this study, we draw on agency theory and sociological perspectives of institutions to explore variations in the motivation of different owners to pursue a socially responsible agenda. Our analysis of a sample of Indian firms for the period 2008–2015 illustrates that business group and family ownership is beneficial for community-related CSR. Our theoretical arguments and results highlight the importance of combining multiple lenses to assess the influence of ownership structures on CSR in emerging markets.

to pursue private benefits from CSR in emerging markets. Combination of these theories can also help us better understand how distinct ownership structures result in variations in CSR responsiveness to various regulative, normative and mimetic institutional pressures. We theorize that firm ownership will exert a differential influence on receptivity and responsiveness to institutional pressures for CSR from internal and external stakeholders due to the extent of conflict between owners.
CSR is viewed as a multidimensional construct with four main constituting dimensions: economic, legal, ethical and discretionary (Carroll, 1991). More recently, Block and Wagner (2014) consider five CSR dimensions: community-related CSR, diversityrelated CSR, employee-related CSR, environment-related CSR, and product-related CSR and find that family-owned firms behave responsibly on some CSR subdimensions while being less responsive on other CSR dimensions. Since empirical research notes that firms tend to treat social and environmental issues quite differently in practice (Bansal, Gao, & Qureshi, 2014), we only investigate the influence of ownership structure on one component of CSR, namely, community-related CSR. We define community-related CSR as the extent to which a firm makes both charitable donations and community contributions (Mithani, 2017).
Some studies have begun investigating the impact of ownership structures on CSR disclosures in emerging markets such as South Africa (Ntim & Soobaroyen, 2013). Yet, there is a dearth of multi-theory research on the link between ownership structures and CSR expenditure in other emerging markets. Hence, in this study we examine CSR in a less investigated emerging market, India and seek to answer the following research question: How does ownership structure impact community-related CSR engagement by Indian firms? To answer this question, we adopt arguments from institutional theory along with agency theory.
To test our theory, we analyze a dataset of Indian firms currently listed on two Indian stock exchanges, namely, Bombay Stock Exchange (BSE) and National Stock Exchange (NSE) over an eight year period from 2008 to 2015. This empirical context is especially conducive for exploring our research question given that despite economic liberalization and significant changes in factor markets, different types of organizational forms continue to dominate the business landscape in India (Arevalo & Aravind, 2011). Due to India's mixed-plan economy with features of centrally planning, the central and state governments directly or through their associated institutions control several strategic industries. At the same time, a large private sector with business groups and family-owned firms where business group promoters and families play an influential role in organizational governance represent features of a market economy (Khanna & Palepu, 2000).
Moreover, Indian firms operate in an institutional context with poor socio-economic conditions such as a high illiteracy rate (37.2%), a high infant mortality rate (41.4 per 1000 live births), and a high percentage of people living below the poverty line (23.6%) (UNDP, 2015a, b, c). This creates high normative institutional pressures for corporations with significant resources to act in socially responsible ways to improve these conditions while maintaining their focus on financial performance. Additionally, community-related CSR in the form of charitable donations and community contributions constitutes a significant portion of CSR spending by Indian firms due to historical traditions, religious norms, a post-colonial national building ethic, and indigenous business-society relations (Fisman & Khanna, 2004;Mitra, 2012;Sivakumar, 2008;Worden, 2003).
Rising coercive institutional pressures also require Indian firms to spend a nominated portion of their net profits on CSR (Jain, Aguilera, & Jamali, 2017;Subramaniam, Kansal, & Babu, 2017 India, 2014). This creates high coercive pressures encouraging Indian firms to engage in community-related CSR.
At the same time, other researchers note that CSR in Indian firms broadly follows four models 1 : ethical, statist, liberal, and stakeholder (Kumar, Murphy, & Balsari, 2001). Although the influence of government ownership in Indian firms on CSR is less studied, it is plausible that it follows a statist model with public welfare objectives driving CSR initiatives. Indian business group owned firms have historically engaged in CSR to address institutional voids by providing education and employment in areas where they are most needed (Mishra & Suar, 2010). A major concern in family owned firms in emerging markets is that powerful families that control these organizations might expropriate minority shareholders to maximize their own benefits from CSR (Muttakin & Subramaniam, 2015;Oh, Chang, & Martynov, 2011). Due to their distinctive characteristics, CSR engagement can be expected to differ across Indian firms depending on whether they are government owned, business group owned or family owned. In summary, given that the influence of firm ownership on CSR in weak institutional settings remains largely understudied (Ararat, Colpan, & Matten, 2018), we believe that it is particularly important to assess whether variations in institutional and agency motives in Indian firms with different types of ownership leads to variations in their engagement in community-related CSR (Gomez-Mejia, Cruz, Berrone, & De Castro, 2011;Subramaniam et al., 2017).
The next section provides our theoretical background and hypotheses where we utilize institutional theory and agency theory arguments to explain differences in community-related CSR by Indian firms with different types of ownership structures. This is followed by the methods section, which details the research site and key variables. Finally, the results section is followed by a discussion of the implications and limitations of our study.

Theoretical development and hypotheses
Scholars have largely examined the role of ownership patterns in influencing CSR in developed market contexts using agency theory and stakeholder theory arguments (Dam & Scholtens, 2013). Research suggests that ownership structure has a differential impact on CSR engagement by firms and this impact is likely to vary across countries (Devinney, Schwalbach, & Williams, 2013). Studies in Europe have found that corporate, individual and employee ownership lower CSR engagement while bank, institutional and state ownership have a neutral impact (Dam & Scholtens, 2012, 2013. On the other hand, studies in the US suggest that greater insider ownership (manager and large block holders) mitigates the conflict of interest between different shareholders in firms investing in CSR since insider investors bear a larger cost associated with CSR (Barnea & Rubin, 2010). Other empirical research in the US reveals that different types of ownership structures can differentially influence different dimensions of corporate social performance (Johnson & Greening, 1999). Extant research in other Western contexts also indicates that companies with higher CSR ratings have less concentrated ownership (Brammer & Millington, 2008). Dam and Scholtens (2013) reason that even though a CSR practice might be socially optimal, developed market firms with concentrated ownership are less likely to prefer investments in CSR since benefits do not outweigh the costs.
In emerging markets, previous multi-theory studies of organizations with different ownership structures primarily investigate differences in voluntary CSR disclosures in annual reports. For example, Ntim and Soobaroyen's (2013) study finds that different types of ownership in South African firms has a differential impact on CSR disclosures which the authors explain utilizing a combination of resource dependence theory, stakeholder theory, legitimacy theory and agency theory.
Extant research suggests that it might be important to integrate institutional theory with agency theory arguments (Eisenhardt, 1988;Kostova, Roth, & Dacin, 2008) since key decision-makers of firms with heterogeneous governance structures have a greater capacity to avoid pressures from market and non-market actors to meet both their economic and social goals in these institutional environments. The presence of regulative, normative and mimetic institutional pressures in these countries should compel some owners to compete for economic efficiency by reducing CSR while others will increase CSR engagement to conform to expected social behavior to gain legitimacy with relevant stakeholders.
In emerging markets, most agency theory research assumes that since the government, family or business group is the pre-dominant shareholder and manages the firm or is closely related to the top management team, there is a natural alignment of interests between owners and managers, therefore, principal-agent problems are unlikely to surface (Jensen & Meckling, 1976). Instead, a different type of agency problem termed the principal-principal agency problem is pervasive due to conflicts of interest between majority and minority shareholders when majority shareholders disregard the interests of minority shareholders (Schulze, Lubatkin, Dino, & Buchholtz, 2001). Research suggests that this could be due to actions such as 'financial tunneling' i.e. transfer of assets at below market prices between publicly listed group affiliates to private group affiliated companies (Young et al., 2008).
Recent work notes that heightened CSR may be the result of 'social tunneling' (Kock, Lee, Min, & Park, 2016). The notion of 'social tunneling' highlights greater principal-principal agency conflicts and expropriation of non-controlling shareholders due to the divergent interests of controlling and non-controlling owners. While controlling owners in emerging markets tend to strive for high social performance to reach their non-economic goals, non-controlling shareholders have a greater preference for financial goals since they are less likely to enjoy the non-economic benefits of CSR.
Agency theory tends to focus internally on the opportunistic behavior of controlling shareholders of firms, while both internal and external forces drive community-related CSR decisions. Therefore, it becomes important to examine the role of institutional pressures that emphasize external forces along with principal-principal agency problems that highlight the role of internal constraints in shaping CSR in firms operating in emerging markets. Fig. 1 summarizes the model we elaborate and test empirically.

Government ownership and CSR
Governments in many emerging markets control critical resources and increasingly play a significant role in setting rules to create norms for the CSR behavior of organizations (Campbell, 2007). Historically, Indian government firms adopted 'implicit' CSR practices such as the provision of housing for their employees, lifelong employment, employee-friendly health care plans and sponsorship of social welfare projects to fill institutional voids in emerging markets (Mishra & Suar, 2010).
Institutional theory views governments as societal institutions with coercive power to regulate the behavior of organizations through laws and regulations. Social reform regulations in India have led to rising coercive pressures on government firms to prioritize community-related CSR (Muttakin & Subramaniam, 2015). In addition, government firms face growing normative pressures from a variety of non-market actors such as professional associations, non-governmental organizations and the media (Dhanesh, 2014). Institutional theorists emphasize the importance of conforming to regulations and social norms because doing so increases organizational legitimacy (Deephouse & Suchman, 2008).
Historical factors prior to pro-market reforms in India led government firms to be viewed as both economic and social providers of local communities. This increases the likelihood that government firms will be more attuned to normative pressures from external constituents to continue to provide social benefits similar to those they provided pre-transition (Mohan, 2001). Some studies have found that greater sensitivity and responsiveness to institutional pressures can lead government owners in India to increase government sponsored CSR programs (Kansal, Joshi, Babu, & Sharma, 2018) and disclosures of CSR directed towards the weaker sections of society (Muttakin & Subramaniam, 2015). From the agency theory perspective, government shareholders in India have a strong incentive to push for high social performance since enhancing financial performance might not be their primary objective (Subramaniam et al., 2017). In contrast, minority shareholders might have a greater preference for financial goals since they are less likely to enjoy the non-economic benefits of CSR. This makes the principal-principal agency problem in CSR terms more likely to occur. Due to the weak legal protection of minority shareholders in India (Sarkar, 2010) and the greater influence of the government on government firms, it is plausible that the principal-principal conflict will be resolved in favor of government owners.
Institutional theory arguments lead us to expect that the higher receptivity of government firms to rising institutional pressures for CSR from governments and external stakeholders will shift the balance towards greater prioritization of noneconomic/societal goals by government owners. Likewise, our discussion of agency theory problems also suggests that government firms are likely to safeguard the interests of the government by increasing community-related CSR. Thus: Hypothesis 1 Government ownership is positively related to community-related CSR

Business groups and CSR
Business groups are defined as a set of legally independent companies that are bound together by formal and informal ties (Chittoor, Kale, & Puranam, 2015;Khanna & Rivkin, 2001;Ramaswamy, Purkayastha, & Petitt, 2017). Business groups are a common feature in emerging markets since they are viewed as an efficient response to substantial institutional voids due to weak institutions and the poor infrastructure in these countries (Chari & Dixit, 2015;Elango, Pattnaik, & Wieland, 2016). Business groups are particularly successful in dealing with institutional deficient conditions since they have developed a variety of coping strategies and organizational mechanisms that provide them with an advantage over independent firms (Gaur & Kumar, 2009;George & Kabir, 2012;Purkayastha, Kumar, & Lu, 2017).
Citizens in many emerging markets do not have access to basic health care services and social security systems (Visser, 2008). To fill the institutional voids that stem from rudimentary social welfare systems to support markets, many Indian business groups with a post-colonial nation-building ethic engage in CSR. Case based studies highlight how business groups contributed to the socio-economic development of communities living around their industrial units long before pro-market reforms (Mitra, 2011;Worden, 2003). Older Indian business groups with a strong legacy of CSR continue to direct a significant portion of their group profits back to society through their charitable donations and community contributions (Sivakumar, 2008). Due to their historical socially responsible behavior, CSR by business groups has become a socially accepted norm and many Indian business groups have earned a favorable communityrelated CSR reputation (Mitra, 2011).
From an institutional theory perspective, normative pressures from local communities, financial and business associations should guide business group owners to view responsibilities to communities as inherently linked to their core organizational purpose. Because of their higher visibility and strong sense of obligation towards local communities, Indian business groups should prioritize visible community-related CSR strategies to respond to these rising institutional pressures. Doing so, allows them to alleviate stakeholder concerns while preserving their reputation with local constituents. Hence, it can be expected that their history of CSR engagement and efficiency with the use of group-level CSR resources (Sivakumar, 2008) will allow Indian business group promoters to derive socio-emotional benefits (Gomez-Mejia et al., 2011) by addressing institutional pressures for CSR. For example, congruence between stakeholder expectations and their community-related CSR engagement can help local communities to effectively cope with complex, deep-rooted social problems (e.g. poverty, malnutrition, illiteracy, etc.) while creating legitimacy gains for the business group promoter.
From an agency theory perspective, the complexity and opacity (i.e. incomplete disclosure on extent of insider ownership, extent of holdings by promoter type, etc.) of Indian business group ownership structures can interfere with efficient monitoring and lead to the expropriation of minority shareholders by promoters (Sarkar, 2010). One way minority investor expropriation can be undertaken is through social tunneling wherein business groups might strategically divert resources for CSR to group affiliates with higher visibility. Relatively weak investor protection and rule of law in India reduces the extent to which minority owners are able to act as a countervailing force against promoters. This creates principal-principal agency problems since business group promoters can collect majority of the benefits from CSR, while the costs of CSR engagement are disproportionately borne by minority shareholders (Kock et al., 2016).
Empirical work finds that over 69% of Indian business group affiliates have promoters as a chairperson or as a member of the board of directors (Sarkar, 2010). This indicates that business group structures provide promoters with greater discretionary power over corporate resources and a greater ability to appoint executives to management positions who are sympathetic to their objectives. Hence, we expect this principal-principal conflict to be resolved in favor of promoters. Based on our discussion above, group-affiliated firms will have stronger incentives to invest in communityrelated CSR to preserve their reputation with external stakeholders. Thus: Hypothesis 2 Business group affiliation is positively related to community-related CSR

Family ownership and CSR
Family firms are controlled by a family through involvement in management and ownership which is often coupled with a transgenerational vision for the firm (Chua, Chrisman, & Sharma, 1999;Zellweger, Nason, Nordqvist, & Brush, 2013). Since the family is easily identified by the public as owners of the firm, negative reputation can easily spillover from the firm to the controlling family (Block & Wagner, 2014;Liu, Shi, Wilson, & Wu, 2017). Most family firms tend to score higher on values like empathy, warmth and zeal (Payne, Brigham, Broberg, Moss, & Short, 2011). Altruistic sentiments which promote unselfish behavior between family members have been found to be dominant in family firms and can also play an important role in influencing decision-making processes related to external stakeholders (Berrone, Cruz, Gomez-Mejia, & Larraza-Kintana, 2010;Nekhili, Nagati, Chtioui, & Rebolledo, 2017).
Two prominent perspectives that argue for greater CSR performance by family firms include the 'long-term horizon' view (i.e. family firms assign a high priority to preservation of their reputation) and the 'expropriation' view (i.e. the principal-principal agency problem where the controlling family could expropriate minority shareholders to pursue private benefits). Empirical studies investigating the impact of family control on CSR in Western contexts have found support for the 'long-term horizon' view indicating that family firms engage in CSR to meet their non-financial or socio-emotional goals, namely, projecting a positive family image and reputation (Berrone et al., 2010;Sharma & Manikuti, 2005;Westhead, Cowling, & Howorth, 2001), gaining recognition through generous actions (Schulze, Lubatkin, & Dino, 2003), and enjoying prestige in the community (Corbetta & Salvato, 2004;Stafford, Duncan, Danes, & Winter, 1999). The 'expropriation' view suggests that controlling families engage in CSR by using their voting rights to divert resources away from other projects to CSR that benefits the family while financially expropriating non-family shareholders (Kock et al., 2016).
Both perspectives are important for CSR predictions of family firms operating in emerging markets. The 'long-term horizon' view suggests that family firms in emerging markets should respond to institutional pressures for CSR from local communities in a substantive manner in comparison to non-family firms in an attempt to preserve their high social status in the community and family-related identity. Involvement in community-related CSR can provide the controlling family with an opportunity to have an impact on the firm as more than just a business. Additionally, family firms might increase communityrelated CSR engagement due to their strong inclination to listen to family shareholders, many of whom are closely connected to the local community (Zellweger & Nason, 2008). Greater sensitivity and responsiveness to normative and mimetic institutional pressures increases the likelihood that family firms will engage in community-related CSR. A recent study utilizing CSR ratings of India's 500 largest firms provides support for the 'long-term horizon' view since family firms achieved higher levels of CSR engagement (Cordeiro, Galeazzo, Shaw, Veliyath, & Nandakumar, 2018).
Due to the institutional underdevelopment and weak legal protection of minority stakeholders in the Indian market (Lodh, Nandy, & Chen, 2014), agency theory arguments suggest that the 'expropriation' view might also apply to understanding CSR in family firms. Prior work notes that long-term relationships among family members in Asia provide the required trust for them to act opportunistically creating principal-principal conflicts (Sauerwald & Peng, 2013). Controlling families can be expected to divert greater resources to impose their CSR decisions on minority shareholders to pursue benefits for the family while ignoring non-family shareholder interests. Taken together, increased sensitivity of family owners to institutional pressures from stakeholders along with greater expected private benefits from directing resources to CSR suggest that greater family ownership might increase community-related CSR. Thus, we hypothesize that: Hypothesis 3 Family ownership is positively related to community-related CSR

Family-owned business groups and CSR
The Indian corporate landscape is dominated by business groups controlled by families (Sarkar & Sarkar, 2000). Higher levels of interaction within familyowned business groups increases trust and resource sharing between group affiliates (Manikutty, 2000;Zahra, 2010). Historically, a formidable challenge for emerging market firms was the difficulty in obtaining external funds. In such institutionally constrained environments, family-owned business groups provided group affiliates with access to an Binternal capital market^for funds (Khanna & Palepu, 1997). Besides, family-owned group affiliated firms were able to leverage their family reputation for easier access to external resources (Almeida & Wolfenzon, 2006).
Since controlling ownership is in the hands of a family, the family has greater discretionary power over corporate resources (Dharwadkar et al., 2000). Faced with rising institutional pressures, Indian family-owned business groups can be expected to take the 'long-term horizon' view (Ashwin, Krishnan, & George, 2015) and engage in community-related CSR activities for preserving their family image and reputation as a socially responsible business entity. Moreover, since business groups are viewed as excellent training grounds for developing talent (Mahmood & Mitchell, 2004), engaging in CSR projects across different group firms might allow family-owned business groups to act as incubators of CSR talent. Greater access to financial resources and CSR capabilities made available through multiple group affiliates coupled with greater discretionary power to use these for meeting their socio-emotional goals should encourage family-owned business groups to engage in greater community-related CSR in response to rising institutional pressures for CSR.
Prior research on business groups in India indicates that weak external governance due to the underdeveloped institutional environment results in principal-principal problems which allow controlling families to engage in 'financial tunneling' i.e. transferring profits from group affiliated firms where they have low cash flow rights to group affiliates with high cash flow rights (Bertrand, Mehta, & Mullainathan, 2002). Similarly, Indian family-owned business groups might indulge in 'social tunneling' (strategically engaging in community-related CSR mainly through their public affiliates rather than private affiliates) at the expense of minority owners because this approach effectively addresses institutional pressures from external stakeholders while creating reputation gains for their public as well as private business group affiliates.
From an agency theory perspective, greater mutual monitoring and trust between family members can facilitate the coordination and collective actions that are necessary to result in principal-principal conflicts. At the same time, institutional theory arguments suggest that higher normative and coercive institutional pressures will increase social benefits for family group affiliates from higher CSR performance. In light of both institutional theory and agency theory arguments, we hypothesize that Hypothesis 4 Family ownership of business groups is positively related to communityrelated CSR

Research context
India has a long tradition of community-related CSR consisting of corporate generosity directed towards addressing social problems (Husted, 2015;Sharma & Talwar, 2005). In the early modern era (early 1900s), community-related CSR began to be directed towards social causes such as education and women's rights (Mangaleswaran & Venkataraman, 2014). Nowrojee Wadia set up the first foundation in India, much before Carnegie and others in the United States (Sundar, 2013). During the 1950s, inspired by the Gandhian philosophy of 'trusteeship', businessmen viewed their business as a 'trust', held in the interest of the community at large (Balakrishnan, Malhotra, & Falkenberg, 2017). Post-independence, in the mixed economy framework, there was an increase in government sponsored CSR activities through government-owned firms (Mohan, 2001). Leading business groups like Bajaj, Birla, Godrej, Larsen & Toubro, Infosys and Tata played a role in both industrializing India as well as in its social and political development (Mohan, 2001). Post the 1960s, India witnessed a transition from community-related CSR being an implicit element of the institutional framework to becoming an explicit element of corporate policies (Matten & Moon, 2008). This shift was driven by the economic liberalization in the early 1990s and the growth in non-family companies.
Since late 2007, due to rising CSR regulations Indian firms have increased their attention to socially responsible behavior (Agrawal & Sahasranamam, 2016;Jain et al., 2017). Institutional reforms started with the then Prime Minister Dr. Manmohan Singh releasing the Ten Point Social Charter, which called for organizations to partner with the government to achieve Inclusive Growth (Singh, 2007). This was closely followed by the Corporate Social Responsibility Voluntary Guidelines issued by the Ministry of Corporate Affairs (Ministry of Corporate Affairs, 2009). In 2013, Section 135 of The Companies Act 2013 made CSR a mandatory requirement for companies having a net worth in excess of 5 billion INR, or turnover greater than 10 billion INR, or a net profit higher than 50 million INR during any of the last three financial years (The Gazette of India, 2014). These firms must invest at least 2% of their net profits in CSR. These rules went into effect on April 1st, 2014. Although rising CSR regulations mandate CSR, it is important to note that the Indian government has adopted a lenient view with regard to CSR implementation since organizations did not face any penalties for noncompliance with mandatory CSR guidelines until 2017 (ET Bureau, 2015). Therefore, for all practical purposes, there are no mandatory CSR requirements with associated stringent penalties for the entire period of our study.

Sample
We test our hypotheses using a sample of Indian firms currently listed on the two Indian stock exchanges, namely, Bombay Stock Exchange (BSE) and National Stock Exchange (NSE). Data for all our variables was collected from the Centre for Monitoring Indian Economy Pvt. Ltd. (CMIE) Prowess database which has been extensively utilized in past research on Indian firms (Chittoor et al., 2015;Chittoor, Sarkar, Ray, & Aulakh, 2009). Our analysis is for publicly listed Indian firms during the period from 2008 to 2015. We chose this time period since there was a global increase in consciousness around the CSR practices of firms following the financial crisis in 2008 (Giannarakis & Theotokas, 2011). Of the 5622 publicly listed firms, 1546 firms are business group affiliated and 100 firms are owned either by the central or state government. Of these publicly listed firms, our sample focuses on large firms with sales greater than 5 billion INR, since it is expected that such firms have enough resources to contribute towards CSR and new CSR laws enacted by the government mandate CSR for firms with sales greater than 5 billion INR. Based on this cut-off, our dataset consists of 768 business group affiliated firms, 718 family firms and 78 government owned firms. Due to missing data for some variables, we have an unbalanced panel dataset and firm-year observations are specified in the results table.

Measures
Community-related CSR Community-related CSR is operationalized as a ratio of charitable donations 2 and community contributions 3 made by Indian firms to their net sales. We consider both charitable donations and community contributions, because prior work has found that Indians firms are likely to inconsistently report these values such that community contributions from one year are reported as charitable donations in the following year, or vice versa (Mithani, 2017). We label this variable as communityrelated CSR since it represents monetary contributions directed towards charitable causes and community development as well as funds paid to non-profits (Mithani, 2017). Multiple surveys indicate that community-related CSR constitutes a significant portion of the CSR spending patterns of Indian firms (Chowdhry, 2017;KPMG, 2017). Fig. 2 illustrates the trend of yearly community-related CSR for publicly listed Indian companies (with annual sales in excess of 5 billion INR). A logarithmic transformation was performed since this variable was left skewed.
Ownership As mentioned earlier, we explore the impact of three forms of business ownership that are predominant in India, namely, state ownership, business group ownership and family ownership on community-related CSR.
Government ownership We operationalize state ownership as a dummy variable. It takes the value of 1 if the state or central government owns the firm. Classification of government ownership is available in the Prowess database. 4 We operationalize government ownership similar to several other studies (Ashwin et al., 2015;Bhaumik & Dimova, 2004).
Business group affiliation Business group affiliation is operationalized as a dummy variable. It takes the value of 1 if the firm is associated with a business group and 0 otherwise. This classification of business group affiliated firms is specified in the Prowess database (Khanna & Rivkin, 2001;Komera, Lukose, & Sasidharan, 2018;Lodh et al., 2014). However, there are certain limitations in the way business group affiliation is recorded (Chittoor et al., 2015;Komera et al., 2018). Firstly, historical information about changes related to business group affiliation are not provided. However, since business groups in India are not very active in the market for corporate control, this limitation is not of significant concern (Chittoor et al., 2015;Komera et al., 2018 ). Secondly, a firm is recorded as business group affiliated even when it is the only affiliate of that business group. This goes against the definition of business group wherein at least two firms needs to be affiliated with a group (Belenzon & Berkovitz, 2010). To overcome this problem, we use a timevarying dummy variable to examine whether the business group shrank or expanded its portfolio of affiliate companies below or above two (Chittoor et al., 2015). To check on the robustness of our results, we re-estimated all our models with the time invariant business group dummy.
Family ownership Family ownership is measured as the ratio of equity held by family owners or promoters to the total common stock (Ashwin et al., 2015). It is a 4 Additionally, we operationalize government ownership by manually calculating the ratio of equity held by government in each firm to the total common stock. Since the sample size for which this information was available, was very low, we choose to use the classification specified by Prowess. Prowess arrived at its classification of government ownership using data in the database along with market intelligence and its own judgment. combination of shares held by individual family owners or promoters. Additionally, for a robustness check, we created a dummy variable (family ownership greater than 20% = 1; otherwise = 0) (Lodh et al., 2014).
Control variables Based on prior research investigating the relationship between ownership variables and CSR, we control for the following variables, namely, firm size, firm age, financial slack, leverage and past financial performance McWilliams & Siegel, 2001).
Firm size. Since prior studies have found that firm size is positively associated with CSR (Johnson & Greening, 1999;Muller & Kolk, 2010), we control for firm size which is measured as the logarithm of total assets (Orlitzky, 2001).
Firm age. Based on previous research, firm age is likely to be positively (Moore, 2001) or negatively related (Cochran & Wood, 1984) with CSR. Hence, we control for age as a continuous variable by measuring it as the difference between current year and incorporation year.
Financial slack. Firms are more likely to make discretionary contributions like CSR when they have financial slack (Waddock & Graves, 1997;Xu, Yang, Quan, & Lu, 2015). We use current ratio as a proxy for financial slack (Mahmood & Mitchell, 2004).
Leverage. It is operationalized as the ratio of debt to equity. With higher levels of debt, firms are expected to spend their current cash flow in servicing the debt rather than investing in activities like CSR (Li & Zhang, 2010;Oh et al., 2011).
Past financial performance. We control for past financial performance, as poor prior financial performance can lead to conservative CSR behavior by firms (Cameron, Whetten, & Kim, 1987). We use lagged ROA (one-year lag) to capture prior financial performance.
We also control for industry and year using dummy variables. We created industry dummies using the 4-digit National Industry Classification (NIC) codes available in Prowess.

Results and analysis
Summary statistics for our dataset are provided in Table 1 as two parts. First, we present the summary statistics for the entire sample in part (a). From this, we can observe that the mean firm age of the sample is 30 years with the oldest firm being 150 years old. The large mean for business group affiliation suggests the significant presence of business group affiliated firms in India. In part (b), we present the summary statistics of dependent and control variables specific to the ownership categories. Further analysis revealed that there were 11 government owned firms and 13 business group firms, which had spent over 500 million INR in community-related CSR at least once during the period of our study. However, no family-owned firm had community-related CSR expenditure greater than 500 million INR. In Table 1 (b), we conducted 2-sided paired t-tests to assess whether there were significant differences in community-related CSR engagement based on firm characteristics. Our results indicate that there is a significant difference between the community-related CSR engagement of family and non-family firms, and between the community-related CSR engagement of government-owned and non-government owned firms. On the other hand, we do not find a significant difference between the community-related CSR engagement of business groups and non-affiliated firms.
The correlation matrix is provided in Table 2. We find a negative and significant correlation between government ownership and community-related CSR, and a positive and significant correlation between business group affiliation and community-related CSR. We checked for multicollinearity and found that Variance Inflation Factors were below 4. Therefore, multicollinearity is not a likely concern in our analysis (Neter, Kutner, Nachtsheim, & Wasserman, 1996). Table 3 presents the results for the relationship between ownership variables and community-related CSR. We used panel random effects regression for estimating the relationship. We use the random effect model because key explanatory variables in our models do not change (like group affiliation) or change slightly (ownership variables) over time (Kennedy, 1998). We also used robust standard errors clustered at the industry level.
In case the independent variables are endogenous, regression coefficients may be inconsistent. However, earlier research has argued that ownership structure in Asian and Indian firms remains relatively stable over time and endogeneity in ownership is not a concern for these firms (Ashwin et al., 2015;Jiang & Peng, 2011). Our research design helps mitigate concerns of omitted heterogeneity by the use of year and industry fixed effects (El Ghoul, Guedhami, Wang, & Kwok, 2016).
In Table 3, Model 1 represents the base model, which includes the control variables that influence community-related CSR. Among controls, we find that past financial performance has a significant and positive influence on community-related CSR. In Model 2, Model 3, and Model 4 we incrementally add the three ownership variables, namely, government ownership, business group affiliation, and family ownership. Model 5 includes all three ownership variables. Model 6 highlights the effect of business group affiliation, family ownership and its interaction. Finally, Model 7 includes all three ownership variables along with the interaction between family ownership and business group affiliation.
Model 2 presents the result for Hypothesis 1, which suggests that government ownership does not have a significant effect on community-related CSR. However, in Model 5 (β = −.702, p < .10) and Model 7 (β = −.607, p < .10), we observe a weak negative and significant effect of government ownership on community-related CSR. Considering that the predicted positive effect is not significant across all models, we do not find support for Hypothesis 1.
Model 3 shows the result for Hypothesis 2, which predicted that business group ownership is positively related to community-related CSR. In Model 3, the coefficient   for the effect of business group affiliation on community-related CSR is positive and significant (β = .152, p < .05) providing support for Hypothesis 2. The results presented in Model 5 (β = .240, p < .01), Model 6 (β = .369, p < .01) and Model 7 (β = .332, p < .01) also provide support for Hypothesis 2. The coefficient for the effect of family ownership on community-related CSR is positive and significant (β = .278, p < .10) in Model 4. This result holds well in Model 5 (β = .389, p < .05), Model 6 (β = .596, p < .05) and Model 7 (β = .586, p < .05). Therefore, we find support for Hypothesis 3.
Models 6 and 7 show the results for the interaction between family ownership and business group ownership on community-related CSR. We find that the moderating influence of business group affiliation on the relationship between family ownership and communityrelated CSR is not significant, which indicates that Hypothesis 4 is not supported.
In summary, we find that while business group affiliation and family ownership have a positive and significant effect on community-related CSR, government ownership and family-owned business groups do not have a significant effect.

Robustness tests
We conducted several diagnostic checks on the sample to determine the robustness of our results (presented in Appendix). First, in place of the time-variant business group dummy, we used the time invariant dummy and repeated the entire analysis. Here again, we found that the results were consistent. Second, we replaced family ownership (%) with a dummy variable (where family ownership greater than 20% = 1; otherwise = 0) and found that family ownership has a positive and significant impact on CSR. Finally, we check for potential issues arising from selection issues by repeating our analysis using a Heckman two-stage model (Komera et al., 2018). In the first stage of we created a dummy variable for whether the firm has undertaken community-related CSR or not and ran a probit regression model on the dummy variable, considering firm level predictors like firm size, firm age and financial performance. Using the results from this probit model, we calculated the inverse Mills ratio, which we subsequently add as a predictor to the overall regression model. Results from this robustness analysis are consistent with our original results.

Discussion and implications
This study shifts theoretical attention to the impact of ownership structure on communityrelated CSR in a less studied emerging market. We utilize institutional and agency theory arguments to hypothesize that sensitivity to rising coercive and normative institutional pressures from stakeholders for community-related CSR must be considered along with principal-principal agency costs to determine whether ownership structure influences community-related CSR. By integrating insights from both theories, we argue that the firm's sensitivity to institutional pressures from external constituents along with principalprincipal agency conflicts might enhance or reduce community-related CSR.
This study makes several theoretical contributions to the literature. First, it extends the CSR literature by examining the link between distinct types of ownership structures and CSR in firms operating in a difficult institutional environment. As highlighted earlier, in an attempt to overcome institutional constraints, organizations in emerging markets have developed distinct ownership structures such as state ownership, business groups and family firms (Ashwin et al., 2015;Khanna & Rivkin, 2001;Subramaniam et al., 2017). Since CSR activities in emerging markets are considered as a means for filling existing institutional voids (Chakrabarty & Bass, 2015;Su, Peng, Tan, & Cheung, 2016), this study contributes to our understanding of how differences in ownership structures in weak institutional environments might be associated with community-related CSR.
Second, our research outlines the ownership characteristics that influence receptivity to institutional pressures and drive engagement in community-related CSR. We find that institutional pressures that increase socio-political legitimacy (Young & Thyil, 2014) and preserve socio-emotional wealth (Gómez-Mejía, Haynes, Núñez-Nickel, Jacobson, & Moyano-Fuentes, 2007) encourage Indian business groups and family firms to engage in community-related CSR. This provides support for the 'long-term horizon' view indicating that business groups and family firms engage in CSR to maintain a positive family image and reputation (Berrone et al., 2010;Gomez-Mejia et al., 2011).
Owing to data limitations, we were unable to examine the effects of different percentages of family, business group and government ownership on community-related CSR. It is important to assess whether the percentage of ownership of Indian firms impacts community-related CSR differently. For instance, when the percentage of family or business group ownership is low, it is plausible that principal-principal conflicts might be resolved in favor of minority owners, such that they prioritize economic benefits over family or business group reputational benefits. Our study found that government ownership does not have a positive impact on community-related CSR. This suggests that principal-principal agency conflicts whereby government owners expropriate minority owners by engaging in community-related CSR might not be resolved in favor of government owners. This makes it particularly important to account for the impact of distinctive characteristics of owners of Indian government firms on their CSR engagement.
Since 1991, the Indian government has engaged in the partial sale of government equity in public sector enterprises (while retaining majority ownership) to improve their market competitiveness and efficiency. It is possible that the principal-principal conflict is resolved in favor of government owners at higher percentages of government ownership of government owned firms. On the other hand, at lower percentages of government ownership the principal-principal conflict might be resolved in favor of minority owners because government owners also choose to prioritize economic goals leading to lower engagement in community-related CSR. Our finding that government ownership does not have a positive impact on community-related CSR departs from evidence in other emerging markets such as China (Lau, Lu, & Liang, 2016) and South Africa (Ntim & Soobaroyen, 2013) where government ownership was observed to increase CSR engagement and CSR disclosure. Hence, our study suggests that it might also be important to account for differences between emerging markets in terms of their level of economic development and the institutional environment in exploring the association between government ownership and community-related CSR. Our results indicate that family ownership of business groups does not have a significant impact on community-related CSR. In light of recent work by Terlaak, Kim and Roh (2018), we believe that family ownership within Indian business groups by itself might not be sufficient to boost CSR. Instead, a combination of family ownership and family leadership within Indian business groups might be necessary to foster community-related CSR.
Third, our study contributes to the discussion on the effect of ownership on one CSR dimension. Unlike Block and Wagner's (2014) research in Western settings, which finds that family firms focus less on community-related CSR, our study finds that large Indian business groups and family firms prioritize the demands of community stakeholders by increasing their community-related CSR engagement. This could be because communityrelated CSR has become a culturally accepted norm in emerging markets such as India, which makes firms more sensitive to institutional pressures from stakeholders.
Fourth, our research examines the combined effect of institutional and agency arguments in influencing community-related CSR. Owing to both institutional pressures and plausible resolution of the principal-principal conflict in favor of group promoters in the Indian context, business groups are likely to exhibit greater attention to community-related CSR. Our study highlights the need to further explore the role of 'social tunneling' in business groups to better understand whether the expropriation of one shareholder group by another creates agency conflicts that heighten responsiveness to institutional pressures for CSR. Taken together, these findings advance recent work (Jain et al., 2017;Jamali & Karam, 2018) by suggesting that agency theory and institutional theory offer promising, complementary explanations for the CSR performance of organizations in emerging markets. Both internal constraints and external forces should be considered when analyzing the heterogeneous influence of dominant shareholders on CSR strategies.
More importantly, by examining the CSR practices of Indian firms this study integrates the corporate governance and international CSR streams of literature thereby addressing calls by IB scholars (Brammer, Pavelin, & Porter, 2009;Doh, Husted, Matten, & Santoro, 2010) and corporate governance scholars (Claessens & Yurtoglu, 2013;Strange, Filatotchev, Buck, & Wright, 2009) for greater synergy between IB and CSR research. Overall, our findings suggest that incorporating features of the institutional environment that shape the interests of stakeholders can provide a better understanding of the impact of ownership on CSR in international settings.

Future research and limitations
Our study opens up possibilities for further work in this area. First, our empirical investigation is based on one emerging market, India, which limits the generalizability of our findings to other emerging markets. Future research could replicate this study across other emerging and frontier markets in Africa, Asia and Latin America. Second, we study the role of ownership and CSR in a multi-industry context. Earlier research has found that institutional pressures on CSR engagement tend to vary across industry (Fernandez-Feijoo, Romero, & Ruiz, 2014;Imbun, 2007;Reverte, 2009) and within industry depending on the type of social performance (Brammer & Pavelin, 2006). Our research offers limited understanding on such nuanced differences and hence future research can explore whether the relationship between ownership structure and CSR varies in such instances in India. Third, comparing the community-related CSR performance of publicly-listed firms with that of smaller firms would be interesting since smaller firms face lower institutional pressures from a variety of stakeholders such as customers, NGOs and the government (Sahasranamam & Ball, 2018;Wickert, 2014;Wickert, Scherer, & Spence, 2016). Fourth, similar to Kock et al. (2016), future work in the Indian context can also explore the role of independent board monitoring in reducing social principal-principal conflicts. Such empirical research can contribute to a better understanding of whether governance mechanisms promote or hinder social performance in emerging markets. Fifth, since prior research has found that organizations can behave responsibly on certain CSR dimensions while demonstrating poor performance on other dimensions (Block & Wagner, 2014), researchers can examine how agency problems shape firm performance on visible CSR dimensions such as product-related CSR not examined in this study. Sixth, our research does not consider the exact ownership percentage of government and business groups owing to data limitations. Future research having access to such nuanced data will help to illuminate whether there are specific thresholds of ownership that influence community-CSR. Finally, the impact of CSR strategies of Indian firms on their financial market performance along the lines of research carried out in other emerging markets like South Africa (Arya & Zhang, 2009)  Though we include inverse Mills ratio in this robustness test, we acknowledge that in decisions like CSR it is difficult to differentiate which factors influence the decision to invest in CSR and subsequently the amount of CSR investment made Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made.