Multinational Enterprises’ Dual Agency Role: Formal Institutions and Corporate Social Responsibility in Emerging Markets

: This study systematically examines whether and how a nation’s extent of economic glob ‐ alization, manifested by the presence of multinational enterprises (MNEs), and formal institutional development may jointly influence a nation’s overall corporate social responsibility (CSR) involve ‐ ment across emerging markets. Drawing on institutional theory, we develop a dual agency model: on the one hand, MNEs take the role of CSR agents in demonstrating CSR practices and imposing direct influences on local firms in emerging markets; on the other hand, MNEs function as institu ‐ tional agents, conducting activities of institutional entrepreneurship and channeling global institu ‐ tional influences into emerging markets. Empirically, we conduct a bootstrapped test of mediation analysis across 83 developing countries. Our findings show that in the context of emerging markets, a more significant presence of MNEs in a nation positively links to its firms’ overall CSR involve ‐ ment, and the nation’s formal institutional evolvement partially mediates such a positive relation ‐ ship. This study ends with a conclusion and discussions on the contributions and implications.

Institutions are typically perceived as "the rules of the game in a society or, more formally, are the humanly devised constraints that shape human interaction" [19] (p.3).Institutions include both formal institutions (i.e., laws, government regulations, and other public policies that reflect national expectations toward the role of business) and informal institutions (i.e., widely adopted beliefs in a society) [20].Extant studies have found evidence that formal institutions are significant factors influencing firms' CSR involvement [18,21].
Nevertheless, although the relationship between formal institutions and firms' CSR involvement has been evidenced, the literature presents inconsistent and thus ambiguous findings.For instance, the earlier literature with developed countries as the research context shows two competing perspectives, i.e., institutional substitute perspective [6,22] that denotes that weak institutional environment leads to more CSR; and institutional complement perspective [23,24] that indicates strong institutional environment links to the growth of CSR.
The recent research with emerging markets as the research context also presents an evident relationship between formal institutions and firms' CSR involvement [25,26].However, such studies show fragmented findings, lacking systematic analyses and consistent findings [27].
Moreover, both research veins found evidence that besides the firms' embedded institutions, the extent of a nation's economic globalization, i.e., the integration of economic activities such as trade, investment, and financial capitals across national borders [28,29], also influence local firms' CSR involvement [30,31], although being sporadic.
Therefore, the extant literature reveals a critical question, i.e., whether and how a nation's extent of economic globalization and formal institutional development may jointly influence the firms' overall level of CSR in a country.In this study, we aim to examine this question systematically.Specifically, we focus on examining whether and how multinational enterprises (MNEs) may influence the nation's institutional environment in general and formal institutions in particular, and then further influence the firms' overall CSR involvement in a country.Following Dunning and Lundan [32], we view an MNE as a coordinated cross-border network of its value-creating activities conducted internally and externally with its business partners.Since the dominant vehicle of economic globalization is foreign investment [28,33] and MNEs are the main economic actors of foreign investment [34,35], we deem that the presence of MNEs is an appropriate proxy of the extent of economic globalization in a country and primarily rely on explaining MNEs' behaviors to reach our theoretical arguments.
Mainly drawing on institutional theory's [36,37] two theoretical perspectives, namely firms' institutional embeddedness and firm agents perspective, we argue that MNEs play dual agency roles in host countries.One is MNEs' role as CSR agents, which directly influence local firms regarding CSR practices.Another is MNEs' role as institutional agents, stimulating institutional changes, and then such institutional changes further influence firms' CSR involvement in host countries.
Empirically, to examine the relationship between economic globalization, institutional evolvement, and overall CSR involvement across countries, we set emerging markets as the research context.Emerging markets (e.g., Brazil, China, and India) are economies with low income but rapid growth, mainly resulting from economic globalization [38].Emerging markets have been experiencing not only increased economic globalization but also major institutional transitions, meaning "fundamental and comprehensive changes introduced to the formal and informal rules of the game that affect organizations as players" [39].Such evolving and dynamic institutional environments provide an ideal research context for answering the research question.
We built a longitudinal dataset covering a relatively large number of developing countries (i.e., 83 countries) from 2008 to 2017.We conducted a bootstrapped test of mediation analysis.Our findings show that in the context of emerging markets, a greater extent of economic globalization directly links to the nation's enhanced overall CSR involvement.Our findings also show that a greater extent of economic globalization leads to a stronger formal institutional environment, which positively relates to more extensive CSR involvement within a country.
We think that this study has unique contributions.First, this study contributes to the CSR literature by identifying and evidencing the extent of economic globalization, manifested by the presence of MNEs as an indicator of CSR, which may influence the nation's overall extent of CSR directly and indirectly via the mediating effects of the formal institutional evolvement in a nation.Second, we believe that this study enhances international business scholars' understanding regarding MNEs' dual agency roles regarding CSR diffusion in emerging markets.
Third, the evidence of MNEs' institutional agency role in emerging markets supports institutional theory's less studied perspective of firm agents, thus contributing to institutional theory.Last but not least, our findings offer implications for managers and policymakers of emerging markets about MNEs' positive transformational role in facilitating a stronger institutional environment and diffusing CSR activities in emerging markets.Consequently, better-informed public policies such as foreign direct investment policies can be further developed to accommodate more MNEs.

Literature Review
The extant literature has presented significant scholarly efforts in explaining how the institutional environment may influence firms' CSR involvement.Scholars have focused their efforts on classifying countries according to their variant business and institutional systems [40][41][42].The earlier scholarly efforts mainly focused on the research with the context of developed countries [6,43,44].These studies present two perspectives that suggest competing explanations.One perspective, the institutional substitute perspective, is that some scholars [6,22] argue that increased CSR is more likely to occur in countries with less regulated institutional environments but an ongoing and rapid process of economic liberalization [45].This perspective indicates that firms conduct proactive CSR to fill institutional voids (i.e., a lack of supporting institutions) in nations with weak formal institutions.Such CSR is considered firms' private governance to defend their power, obtain social control and gain legitimacy [46,47].
Another perspective, institutional complement perspective, is that some scholars assert that strong formal institutions lead to the rapid growth of CSR [23,24,44].It is argued that firms are more likely to conduct CSR activities [48] actively in countries such as Scandinavian countries that demonstrate well-established regulatory and political institutions.This perspective shows that firms' CSR activities complement or mirror the institutional settings of a well-regulated economy [17,49].
The existing literature provides empirical support for both perspectives mentioned above, keeping this question-how institutions may influence firms' CSR involvementambiguous. Indeed, the former perspective, the institutional substitute perspective, attributes the expansion of CSR to economic globalization and the lagged institutional development.The latter perspective, the institutional complement perspective, attributes the growth of CSR to the more established formal institutions in the context.
Both perspectives imply a few assumptions.First, static formal institutions where firms are embedded are assumed.Second, developed economies are broadly classified into two categories, i.e., economies with weak formal institutions and economies with well-established such institutions.Such a classification assumes no institutional variance within each type of country, i.e., countries with either strong or weak political/regulatory institutions.
Third, there is an assumed or expected relationship between economic globalization and the development of formal institutions.Similarly, a relationship between economic globalization and the growing CSR is also assumed.For instance, the institutional substitute perspective of CSR reflects the logic that because the formal institutional development in a nation is lagged behind economic globalization (i.e., an assumed failed relation between economic globalization and institutional development in economies with weak institutions), firms conduct CSR activities as private governance (i.e., an assumed successful relation between economic globalization and the growth of CSR) [50].
Recent research on firms' CSR involvement has been increasingly set emerging markets as the research context, and such research has been multiplicative [51][52][53].Some of the antecedents driving CSR involvement in emerging markets are investigated [25,54,55].The factor of formal institutions such as regulatory institutions is significant in influencing firms' CSR involvement [26,55].This finding indicates that CSR primarily reflects regulators' expectations in emerging markets.
Such a research expansion to examine firms' CSR involvement in emerging markets enriches our understanding of institutional influences on CSR across countries beyond developed countries [56,57].However, scholarly studies regarding institutional effects on CSR in emerging markets are still fragmented with inconsistent findings, and there is a lack of systematic effort to reconcile previous findings [27].
Meanwhile, some earlier and recent studies have also studied the influence of a nation's economic globalization, besides the influential national institutions.For instance, by conducting a comparative analysis in a context of 19 Western industrialized countries, Gjolberg [30] finds that both a nation's extent of global connectedness and its local institutions drive the nation's CSR practices, although possibly in different ways.By analyzing the content of company CSR initiatives revealed by two large British companies and two Danish companies, Brown and Knudsen [31] found that market pressures (i.e., global institutional influences) rather than domestic institutions determine the content of company programs.
Concurrently, given the context of emerging markets, scholars find evidence of the influence of a nation's extent of economic globalization on the overall CSR involvement within the individual nations [25,26,55].In this regard, Li et al. [58] shows that the influence of economic globalization, government, and supplier concentration are all positively related to firms' CSR involvement; Li et al. [59] highlights that institutional pressures from developed countries have pushed developing countries to promote sustainable practices in their business practices.
Therefore, thus far, the extant literature shows that further efforts are needed to figure out whether and how a nation's extent of economic globalization and its formal institutional evolvement may play out to influence the firms' overall CSR involvement within a country.In such a circumstance, we decide to conduct a systematic analysis at the country level in emerging markets to provide valuable insights for policymakers and enhance scholarly understanding.

Theoretical Framework and Hypotheses Development
By following studies with the context of non-Western countries [60][61][62], we view CSR as all business practices that firms conduct to improve social well-being and bring benefits for their stakeholders.We assert that such a broad conceptualization of CSR is necessary to capture the overall benefits that firms may contribute to the country's well-being.The following section, mainly drawing on institutional theory, develops our theoretical arguments by applying the two perspectives of institutional theory, i.e., firms' institutional embeddedness and firm agents perspective.

MNEs' CSR in Host Countries
Research has argued that foreign MNEs are capable and motivated to perform better in CSR than local firms in emerging markets.MNEs face more institutional tensions than uni-national firms, encountering uncertainties and complexities in the external environment [63,64].Given the multiple institutional pressures (i.e., global, and home and host country institutional pressures) and to gain social legitimacy in the host countries, MNEs are motivated to develop or adopt proactive CSR strategies and practices and transfer advanced practices to emerging markets regardless of the local regulations.For instance, Madsen [65], Christmann [66], and Garcia-Johnson [67] reported that MNEs in polluting industries tend to adopt the strictest environmental policies as their global environmental standards rather than simply following the environmental institutions of the individual countries where they have operations.In CSR, MNEs tend to be proactive to preempt future regulation by regulating themselves, e.g., generating codes of conduct or adopting some forms of standards such as the Dow Jones Sustainability Index [68,69].
Meanwhile, MNEs are more incentivized to adopt innovative technologies and practices to offset regulatory compliance costs due to the increasing legal and regulatory pressures [70].In particular, MNEs actively seek superior CSR practices, such as advanced environmental technologies, transferrable to subsidiaries in host countries to reach economies of scale and mitigate possible negative influences due to regulatory changes [71,72].
In addition, while MNEs continue to engage in CSR both at home and abroad, they accumulate and improve their knowledge and management capability regarding CSR.Thus, MNEs tend to be more capable and responsive to CSR issues.
Admittedly, some MNEs conduct CSIR or engage in CSR and CSIR concurrently, causing negative influences in host countries [73,74].We theoretically argue that the majority of MNEs proactively or reactively adopt advanced CSR and subsequently lead to a series of positive changes in host counties.
The above reasoning helps to reach our base argument that foreign MNEs perform better in CSR than local firms in emerging markets.Based on such a premise, we develop our hypotheses on MNEs' dual agency role to enhance the firms' overall CSR involvement in emerging markets.

MNEs as CSR Agents in Emerging Markets
MNEs face multiple institutional pressures [63,64].According to the institutional embeddedness perspective of institutional theory [36,37] that emphasizes how institutional forces influence firms, it is critical for MNEs to gain legitimacy in all the institutional environments.Thus, MNEs will need to balance the tension between local adaption and global integration [75] when conducting a business practice.To do so, MNEs often first adjust their strategies and structure to be more flexible and responsive to environmental uncertainties [68].Consequently, MNEs have increasingly adopted a decentralized structure with highly autonomous affiliates in host countries to be locally responsive while remaining internationally connected.With the internal structural network, MNEs tend to transmit internally developed best business practices or routines to the MNE affiliates in different countries [68].
In addition, due to global and home country institutional pressures, MNEs' usually superior CSR practices [76,77] essentially differ from local firms' CSR in emerging markets [23].Such differences bring varieties to the local institutional environments in emerging markets, and MNEs' intentional and unintentional transfer of influences yields spillover effects to local firms.Similar to the case of technology transfer, such spillovers occur via MNEs' demonstration and local firms' learning processes, the transfer of employees, and the intentional or unintentional knowledge sharing offered by MNEs to align the behaviors of local suppliers and distributors with their global standards [11,68,78,79].
In other words, such spillovers are a matter of course from MNEs' efforts to transfer best practices within their internal structural network, and the variant CSR practices demonstrated by MNEs' affiliates in host countries exert coercive, mimetic and normative isomorphic pressures [36] to other local firms, including MNEs' local direct partners and other firms.As Kogut [80] summarized, direct investment means extending the organizational principles and capabilities from the MNE's home country to other countries.We thus identify MNEs as CSR agents in emerging markets and reach our first hypothesis.

MNEs as Institutional Agents in Emerging Markets
In contrast to the firms' institutional embeddedness perspective of institutional theory, the firm agents perspective of institutional theory emphasizes how institutions themselves are influenced by firm agents, which bring changes.MNEs often transfer business practices internally within their structural network to adapt to the variant institutional constraints in different countries [68,81].However, the successful transfer of such business practices from MNEs to local firms requires supportive institutional infrastructure to grant firms necessary resources and legitimacy [82].In such a circumstance, MNEs tend to function as institutional agents and become associated with what [83] called "institutional entrepreneurship," i.e., "activities of actors who have an interest in particular institutional arrangements and who leverage resources to create new institutions or to transform existing ones" [84] (p.657).
As a set of political and regulatory rules, formal institutions may provide constraints and expectations for firms, such as imposing restrictions on firms regarding their CSR activities by influencing stakeholders' expectations [85].Meanwhile, formal institutions may also provide distinct and critical resources to enable certain practices [40], such as CSR involvement.
As MNEs often have substantial investments which are less mobile, MNEs have significant incentives to consistently affect the formal institutions in the host countries for favorable policies [86].Especially when the institutional environments are dynamic and display institutional voids, MNEs are more likely to engage in institutional entrepreneurship, imposing formal or informal influences on the surrounding formal institutions.
Indeed, in host countries, MNEs have many opportunities available at multiple levels to influence the process of forming or adjusting formal institutions.First, MNEs may have incentives to affect institutional evolvement at the international level.For instance, MNEs can partner with non-governmental organizations (NGOs) to generate and legitimize new standards [87] favorable to MNEs.Second, MNEs may initiate various negotiations and impose political influences to advance the public regulations that provide MNEs themselves with advantages over their competitors [88].Third, MNEs can engage in private self-regulation, creating and enforcing the standards to assess their practices [89].
Consequently, MNEs may be welcomed and granted legitimacy in emerging markets if they can introduce new institutional elements that are needed [90] and contribute to an ongoing institutional transformation process.The newly introduced institutional elements become part of the process of changing institutional structures in the host country.Thus, MNEs may intentionally change the interface between the government (and its agencies) and the business given all incentives.Indeed, the presence of MNEs creates new imperatives for changing and evolving local institutions, particularly formal institutions.Accordingly, we reach our second hypothesis.

Hypothesis 2 (H2).
In emerging markets, the extent of a country's economic globalization (or the presence of MNEs in a country) positively relates to the evolvement of formal institutional development of the country.

Formal Institutional Development as a Mediator
We have argued that MNEs may take their role as institutional agents to influence the evolvement of formal institutions in host countries.According to institutional theory, the formal institutions legitimized by governments (and their agencies) impose coercive isomorphic pressures on firms [37,91] in the local institutional environment.Governments may simultaneously offer incentives for organizational conformity [63].Moreover, governments as firms' unique stakeholders can also exert coercive pressures on firms by requiring them to comply with certain standards or procedures [63].Indeed, the literature has discussed and highlighted the relationship between formal institutions and firms' CSR decisions and behaviors [17,92].
In linking this relationship between formal institutions and firms' behaviors of CSR with our second hypothesis regarding the influence of a nation's economic globalization on the firms' overall CSR involvement, a mediating role of formal institutions in the link between a nation's economic globalization and the firms' overall CSR involvement is logically presented.
The extent of a nation's economic globalization affects the evolvement of formal institutions, which then influences the firms' overall CSR involvement.In other words, the evolvement of formal institutions represents a mediational pathway via which economic globalization affects the firms' overall CSR involvement in a nation.
However, as indicated in our first hypothesis, a nation's extent of economic globalization, mainly manifested by the presence of MNEs, may exert direct influences on local firms by demonstrating variant CSR practices and local firms' learning processes.Therefore, as a consequence, we predict a partially mediating effect of a nation's formal institutional evolvement on the influence of a nation's extent of economic globalization on the firms' overall CSR involvement within a country.We then reach our third hypothesis.

Hypothesis 3 (H3).
In emerging markets, the evolvement of the nation's formal institutions partially mediates the positive relationship between the extent of a country's economic globalization (or the presence of MNEs in a country) and the firms' overall CSR involvement.

Data Source
To examine the relations between economic globalization, institutional development, and overall CSR involvement across countries in emerging markets, we build a dataset composed of information mainly from data sources of the World Development Indicators Database and Global Competitiveness Report.The World Development Indicators Database collects time series data on various development indicators such as economic growth (e.g., GDP per capita) and education (e.g., school enrollment rate).It is available at the World Bank.
The Global Competitiveness Report is published by the World Economic Forum annually.This report assesses the extent of a nation's competitiveness by measuring its productivity by using its available resources [93].It ranks countries based on the Global Competitiveness Index (since 2004, the Global Competitiveness Report ranks countries based on the Global Competitiveness Index developed by Xavier Sala-i-Martin and Elsa V. Artadi.Before 2004, there were two sub-indexes.The macroeconomic ranks were built on the Growth Development Index developed by Jeffrey Sachs, and the microeconomic ranks were based on Business Competitiveness Index developed by Michael Porter.The Global Competitiveness Index integrates the above two sub-indexes of competitiveness into a single index).This index measures national institutions, policies, and other factors that reflect the nations' current and medium-term potential for growth [93].The report covers about 150 countries and consists of more than 110 variables.One-third of the variables come from publicly available sources (e.g., the United Nations).The rest is based on the Executive Opinion Survey, i.e., an annual survey among a representative group of business leaders across countries.The survey captures valuable information (e.g., the quality of the educational system, overall ethical practices of firms within a country) that is otherwise unavailable on a global scale.The variables are categorized into twelve pillars (the twelve pillars are institutions; appropriate infrastructure; stable macroeconomic framework; good health and primary education; higher education and training; efficient goods markets; efficient labor markets; developed financial markets; ability to harness existing technology; market size-both domestic and international; production of new and different goods using the most sophisticated production; and innovation), and each pillar represents an area regarded as a critical determinant of national competitiveness.

Dependent Variable-National Ethical Practices
Constructing a truly representative measure of CSR is challenging because of its multidimensionality [94], and measuring CSR from a single or only a few dimensions may lead to distorted understanding of CSR [95].Researchers face even more difficulties when measuring CSR at the national level [5].Thus, many studies [96][97][98] have highlighted the importance of the managerial perceptions of CSR activities when trying to understand the overall extent of firms' CSR behaviors.Well-known organizations such as United Nations Global Compact and PricewaterhouseCoopers have frequently utilized such managerial perceptions.These organizations conduct surveys among managers worldwide to make sense of CSR activities across countries.Similarly, the Global Competitiveness Report's annual executive surveys provide information on executives' opinions about the overall ethical practices conducted by firms within their respective countries (the question that was asked about the ethical practices within a country was "In your country, how would you rate the corporate ethics of companies (ethical behavior in interaction with public officials, politicians and other firms)?".A seven-point scale was used, with 1 indicating "extremely poor-among the worst in the world" and 7 indicating "excellent-among the best in the world.").We use such an indicator as the proxy for national ethical practices.

Independent Variable-Economic Globalization
We use the de facto measure of the economic dimension of the KOF index as the proxy of economic globalization in our analyses.Initially introduced by Dreher [99] and updated by Dreher et al. [100], the KOF Globalization Index is a composite index and has been broadly used in the literature [101][102][103].Economic globalization ranges from 1 to 100, with higher values indicating a greater level of globalization.The latest version of the KOF index covers the annual information of 209 countries for the period 1970-2015.
The KOF Globalization Index measures globalization along three dimensions: economic, social, and political.Gygli, Haelg, Potrafke, and Sturm [104] recently revised the index to include de facto measures (i.e., variables that indicate actual flows and activities) and de jure measures (i.e., variables indicating policies that enable flows and activities) for each dimension.Among the three dimensions, "Economic globalization characterizes long-distance flows of goods, capital, and services as well as information and perceptions that accompany market exchanges" [104].Thus, we think that the dimension of economic globalization well reflects what we want to measure as the extent of the presence of MNEs in a country.

Proposed Mediating Variable-Institutional Development
We use World Governance Indicators (WGIs) reported by the World Bank to measure political-institutional development.We intend to use this variable to indicate a nation's formal legal and political-economic institutions and national/public governance ability.We are assured that WGIs are the comprehensive indicators for our measurement purpose.First, WGIs indicate a nation's ability to successfully provide the institutions to ensure the economy functions smoothly [105], which is consistent with what we intend to measure.Second, the WGIs are based on hundreds of variables that consistently measure six dimensions of national governance (i.e., voice and accountability, political stability and absence of violence, government effectiveness, regulatory quality, the rule of law, and control of corruption ("(1) Voice and Accountability (VA)-capturing perceptions of the extent to which a country's citizens can participate in selecting their government, as well as freedom of expression, freedom of association, and a free media.(2) Political Stability and Absence of Violence/Terrorism (PV)-capturing perceptions of the likelihood that the government will be destabilized or overthrown by unconstitutional or violent means, including politically-motivated violence and terrorism.(3) Government Effectiveness (GE)-capturing perceptions of the quality of public services, the quality of the civil service and the degree of its independence from political pressures, the quality of policy formulation and implementation, and the credibility of the government's commitment to such policies.(4) Regulatory Quality (RQ)-capturing perceptions of the ability of the government to formulate and implement sound policies and regulations that permit and promote private sector development.(5) Rule of Law (RL)-capturing perceptions of the extent to which agents have confidence in and abide by the rules of society, and in particular the quality of contract enforcement, property rights, the police, and the courts, as well as the likelihood of crime and violence.(6) Control of Corruption (CC)-capturing perceptions of the extent to which public power is exercised for private gain, including both petty and grand forms of corruption, as well as "capture" of the state by elites and private interests."(Kaufmann, Kraay, and Mastruzzi, 2010.p.4)) (refer to Kaufmann, Kraay, and Mastruzzi [106] for detailed information).Third, the WGI report covers a relatively large number of countries (i.e., more than 200 countries) over a long time period (i.e., since 1996), allowing us to avoid reducing our sample size.Moreover, all values are standardized and centered (ranging from −2.5 to 2.5) to facilitate the comparison across countries.Therefore, we collected all values of the six dimensions and calculated a weighted average of the six dimensions.

Control Variables
Firms are embedded in one or more national business environments [24,68,107].Some scholars [5,24,108,109] conceptualized the national business environment into four dimensions, including the political system, the education and labor system, the financial system, and the cultural system.Among these four dimensions, Whitley [109] (p.48) explains that the political system is essentially featured with the power of the state, indicating the extent to which states "directly or indirectly regulate market boundaries, entry and exit, as well as set constraints on the activities of economic actors" via laws and regulations.It critically influences firms' outcomes, especially firms' social performance [5].Whitley's [109] conceptualization of the political system is consistent with our measurement of the proposed mediating variable, a nation's institutional development.To comprehensively reflect the business environment that firms are embedded in, we set the other dimensions identified by Whitley [109] (i.e., the education and labor system, the financial system, and the cultural system) as control variables.
A country's education system, especially higher education, directly "develops and certifies competencies and skills" [109] (p.50).We thus include school enrollment (tertiary), a ratio of eligible population to the age group within a country, as a control variable.This variable indicates the near-future availability of high-quality human resources.It is available at the World Development Database.
Due to the pressures from employee associations such as labor unions [5,24] and the incentives of positive output such as mitigating workplace adverse behavior [110], firms strive to engage in CSR, especially employee-related CSR, to maintain a good relationship with employees.We thus infer that the employer-employee relationship, primarily indicated by firms' involvement of employee-related CSR, is a barometer of the labor system in a country.We thus include the Cooperation in Labor-Employer Relations Index (the question that was asked about the Cooperation in Labor-Employer Relations Index within a country is "How would you characterize labor-employer relations in your country?"A seven-point scale was used, with 1 indicating "generally confrontational" and 7 indicating "generally cooperative."),an index of the employer-employee relationship as a control variable.It is available in the Global Competitiveness Report.The values range from 1 to 7, with higher values indicating better cooperative employer-employee relationships.
The initially identified four dimensions of national culture, namely individualism, power distance, uncertainty avoidance, and masculinity, are included in our analyses.We employ the updated measures proposed by Taras, Steel, and Kirkman [111] (the metaanalysis by Taras, Steel, and Kirkman [111] provides a set of updated national cultural scores along Hofstede's cultural dimensions with separate indices for different decades to reflect cultural changes over time.This set of updated cultural scores is based on 451 empirical studies that employ comparable approaches with Hofstede's to reach their participants' cultural values.These studies contain more than 2000 samples consisting of more than half a million individuals from 49 countries and regions) in their meta-analysis to minimize concerns (i.e., culture changes over time and Hofstede's measurements are more than four decades old) about using Hofstede's original measurements [112].As much fewer countries were examined for those two dimensions [113], we do not include the later developed two dimensions (i.e., long-term orientation and indulgence) to avoid reducing our sample size substantially.
A better regulated or structured financial system with fewer constraints has been considered more efficient and effective in allocating capital resources to reward firms' CSR activities [114][115][116][117].We incorporate a nation's financial system as a control variable to test similar effects in emerging markets.We use the market capitalization of listed domestic companies (% of GDP) to indicate the ability to mobilize the capital of a nation's financial system.The market capitalization ratio is the value of listed shares of domestic firms divided by GDP, available at the World Development Database from the World Bank.Scholars frequently employ this ratio to measure stock market size, which positively correlates with firms' ability to obtain capital [118].
In addition, we incorporate another control variable, namely the natural logarithm of gross domestic product (GDP) per capita, to indicate the extent of a country's economic development.This dataset is available at the World Development Database from the World Bank.

Dataset
Following Preacher's [119] (pp.827-828) suggestion "at a minimum, some time should elapse between a putative cause and its associated effect to allow for the effect to occur or unfold," we use a one-year lag to avoid simultaneity bias between the explanatory variables and dependent variables.For the explanatory variables, following Kwok and Tadesse [120], we calculated decade averages to reflect the accumulative influence during the one decade.We, in the end, reached a dataset with a time window of 2008-2017 for the dependent variable.The sample covers 83 developing countries.Table 1 shows the summary statistics and the correlation coefficients of the variables.The data reveal wide variations, especially in the economic globalization, financial system, and education system.For instance, Bangladesh showed the lowest economic globalization, 21.39 in 2008, and Lebanon had the highest economic globalization of 85.57 in 2016.The dependent variable, national ethical practices, significantly correlates with the other variables, except the education system.National ethical practice negatively correlates with national culture (individualism and uncertainty avoidance).The correlations between the proposed mediating variable, institutional development, and other variables are significant, except national culture (in terms of individualism).The correlation between institutional development and national culture (power distance and uncertainty avoidance) is significant but negative.
Since some variables highly correlate with others (e.g., the education system highly correlates with GDP per capita with a correlation r = 0.72), we conducted a variance inflation factor (VIF) analysis to check possible multicollinearity.The highest value (VIF of the education system) of the analysis is 4.97, which is less than the threshold value of 10 proposed by Hair, Anderson, Tatham, and Black [121] (the VIF values of other variables are: VIF value of GDP per capita is 4.43, masculinity is 3.29, uncertainty avoidance is 3.23, power distance is 2.56, individualism is 2.17, economic globalization is 2.04, skilled labor is 1.97, the national institution is 1.79, and financial system is 1.66).

Method
Tests of mediation effects often follow the stepwise procedure proposed by Baron and Kenny [122] and Kenny, Kashy, and Bolger [123].This method assumes normally distributed mediation effects (i.e., indirect effects) [124,125].Shrout and Bolger [125] proposed an alternative method, namely the bias-corrected bootstrap method, that does not require a priori assumption, i.e., the assumption of normal distribution [126].This alternative method is a nonparametric statistical tactic "based on bootstrap data-resampling procedures to establish confidence intervals for testing the statistical significance of an indirect effect" [124] (p.372).It estimates the sampling distribution by repeatedly resampling the data.This analytical technique estimates the direct effect, indirect effect (i.e., mediation effect), and the total effect of the independent variable.It also reaches the endpoints of the confidence interval.A further bias correction is provided in cases of small samples (e.g., n < 80) to adjust for possible distribution asymmetry caused by bootstrap estimates [124,127]."Of the resampling methods compared, Shrout and Bolger's bias-corrected bootstrap method performed relatively best" [124] (p.373).Given the merits of this mediation analysis method, we decided to follow this method.A total of 5000 bootstrapping samples was reached for each model.

Results
Table 2 shows the results of five bootstrapped tests of mediation effects.The sample size is variant due to the availability of data points.Bootstrap sample size = 5,000; LL = lower limit; CI = confidence interval; UL = upper limit.path A: the effect of independent variable on mediating variable; path B: the effect of mediating variable on dependent variable; path C: the effect of independent variable on dependent variable.* p < 0.1; ** p < 0.05; *** p < 0.01.
The first model is the base model in which we include one control variable (i.e., the logarithm of GDP per capita) besides the dependent variable, proposed mediating variable, and independent variable.In the rest models, we gradually added the other control variables.
Although methodologists have suggested that the significance of the first step, Path C (the influence of the independent variable on the dependent variable), is not required for testing a mediation effect [123,125,128], we found Path C effects of the first four models are significant at the 0.01 level (p < 0.01) with positive coefficients.Meanwhile, the Path C effect of model 5 is significant at the 0.05 level (p = 0.020) with a positive coefficient.In the second step, all Path A effects (the influence of the independent variable on the proposed mediating variable) show a significant and positive relation (p < 0.01).
For the third step, i.e., Paths B and C', we found that once we include the mediating variable in the model, the independent variable becomes insignificant in model 1 (p = 0.579) and model 2 (p = 0.368); the independent variable remains significant at the 0.1 level (p = 0.080) in model 3, at the 0.01 level in model 4 (p = 0.000) and the 0.05 level in model 5 (p = 0.019).All coefficients of the independent variable and proposed mediating variable are positive.
We then use the bootstrapped method with bias-corrected confidence estimates [125] to test the mediation effects.In our study, a 95% confidence interval of the indirect effects was obtained by retesting the 5000 bootstrapped samples.Results of the mediation analyses confirmed the mediating effect of institutional development in the relationship between economic globalization and nations' overall ethics practices (b = 0.499) with a 95% confidence interval excluding 0 [0.002, 0.007], indicating a significant mediation effect [129].The mean indirect effect from the bootstrapped analyses was positive and significant (0.004).
In addition, the results indicate that the direct effect of economic globalization on a nations' overall ethical practices remained significant at the 0.05 level (b = 0.003, p = 0.019) when controlling for the education system, labor system, financial system, national culture, and economic development.Such a result suggests a partial mediation effect of institutional development.This partial mediation model accounts for 57% explained variance (adjusted R2).Therefore, our three hypotheses are all supported.The results indicate that the proposed mediator, institutional development, exerts a partial mediating effect.
According to our final full model (including all control variables based on our base model), a nation's education system has a statistically significant but negative effect (b = −0.024,p < 0.01) on the nation's overall ethical practices.Such a result indicates that the adequacy of available skilled labor releases a firm's pressures of competing for human resources by conducting CSR activities.This result is consistent with previous findings by Ioannou and Serafeim [5], who found that firms from countries with higher availability of skilled human resources perform worse in CSR.
The control variable of the labor system shows a significant and positive effect (b = 0.317, p < 0.01), as we expected, indicating that firms from countries that value sound employer-employee relationships will consistently perform well in CSR, especially employee-related CSR.
Consistent with the findings or arguments of previous studies [5,24], the effect of the financial system is positive and significant (b = 0.004, p< 0.01), indicating that the more effective and efficient a nation's financial system is, the more likely firms will conduct CSR to signal they are being socially responsible for gaining financial resources from the financial market.
For the control variable of national culture (i.e., individualism/collectivism, power distance, uncertainty avoidance, and masculinity), surprisingly, individualism does not have a significant influence in Path C (b = 0.001, p = 0.994).In Path A, individualism has a significant (at the 0. Such results about cultural dimensions show that, in general, a country's culture significantly influences the diffusion of business practices and the formation of institutions.However, when the proposed mediating variable, institutional development, is introduced to the model (i.e., when it comes to Path B and C'), we found that three (i.e., individualism, power distance, and uncertainty avoidance) of the four cultural dimensions were insignificant.Only the cultural dimension of masculinity is significant and positive (b = 0.266, p = 0.051).This result indicates that formal institutions take a significant role in guiding firms' behaviors of CSR once they are formed in the context of emerging markets.In this case, the cultural factors become insignificant, except the masculinity dimension, indicating that developing countries valuing competition and pursuing achievement are actively engaged in CSR activities beyond the signals or guidance of the formed institutional frameworks.
The control variable of GDP per capita (natural logarithm) shows a significant and positive influence throughout all models, except model 1.This result indicates that a nation's initial economic development affects the nation's subsequent capabilities of adopting CSR demonstrated by MNEs.

Robustness Tests-Instrumental-Variables Methodology
In our analyses, we take a one-year lag between the values of a nation's economic globalization level and the nation's institutional development, which is one year behind the overall extent of ethical practices.This approach mitigates the possible concerns about reverse causality.However, one may still argue that the past economic activities such as international trade and foreign investment that are counted as the nation's level of economic globalization are inversely influenced by the prediction of future institutional environment and ethical practices in the nation.
The possible reverse causality can be tested with an instrumental-variables methodology, employing instruments to disentangle the link between economic globalization and institutional development and the link between economic globalization and ethical practices.The ideal instruments should be variables that affect institutional development and ethical practices but are less likely to be inversely affected.The instruments such as economic history and legal origin commonly suggested in the finance and economic literature [120] are not applicable in this study because this type of instrument is likely to be inversely influenced and is also likely to influence the independent variable, economic globalization.Thus, we exclude this type of instrument.
We then seek variables that may influence economic globalization but are less likely to link to institutional development and ethical practices.The literature shows that international economic activities (e.g., foreign direct investment) are facilitated by the macroeconomic environment such as infrastructure (e.g., energy, communication, and transportation) [130].We then chose electric power consumption (per capita) as the instrumental variable.Table 3 presents the results with the chosen instrumental variable.Model 10 shows the partial mediating effect of institutional development (b = 0.462, p = 0.000 in Paths B and C') between the chosen instrumental variable (b = 0.286, p = 0.012 in Paths B and C') and national ethical practices.This result confirms the major findings from model 5. Bootstrap sample size = 5,000; LL = lower limit; CI = confidence interval; UL = upper limit.path A: the effect of independent variable on mediating variable; path B: the effect of mediating variable on dependent variable; path C: the effect of independent variable on dependent variable.* p < 0.1; ** p < 0.05; *** p < 0.01.

Discussion and Conclusions
In this study, drawing on institutional theory, we present a dual agency model of MNEs, arguing that MNEs play both a CSR agency role and an institutional agency role in emerging markets.Taking a CSR agency role, MNEs exert direct influences on domestic firms' CSR involvement via spillover and demonstration effects and local firms' learning processes.Taking an institutional agency role, MNEs exercise activities of institutional entrepreneurship to influence formal institutional evolvement in emerging markets, which then impose coercive isomorphic pressures on local firms regarding their CSR involvement.
This dual agency model integrates MNEs' adaptation and creativity with institutional change at the macro level, jointly influencing local firms' CSR involvement.Our empirical analyses based on a bootstrapped mediation analysis evidence that in emerging markets, a greater presence of MNEs directly links to the nation's enhanced overall CSR involvement, and the nation's formal institutional evolvement mediates such a positive relationship.
By proposing and verifying such a dual agency model of MNEs, this study clarifies the direct and indirect channels through which MNEs influence local firms regarding CSR activities.Therefore, on the one hand, this study enriches international business scholars' understanding of MNEs' dual agency roles regarding institutional evolvement and CSR diffusion in emerging markets.
On the other hand, this study contributes to the CSR literature.First, although the literature has presented that foreign MNEs usually invest more in CSR, resulting in greater CSR performance in emerging markets [65][66][67][68][69], these studies are largely MNEs focused.Extending this vein of research, we examine the influences of MNEs' CSR from the perspective of local firms and the institutional environment in emerging markets.We investigate whether local firms can learn from foreign MNEs and benefit from MNEs' knowledge spillover regarding CSR.Our findings confirm such spillovers from MNEs in emerging markets.This result helps to portray a more comprehensive picture of the influences of FDI in host countries and partially answers the question raised by Cantwell, Dunning, and Lundan [68], "how should we evaluate the contribution of MNEs to the transformation of the host countries?".Second, with a research context of emerging markets featured with overt institutional transmission and rapid economic development, this study reconciles the contradictive perspectives revealed in previous studies, i.e., the institutional complement perspective and institutional substitute perspective based on the research context of developed countries.The findings of this study reveal that in emerging markets, the firms' overall CSR involvement is simultaneously influenced by the extent of economic globalization and the formal institutional evolvement in a nation, and the extent of economic globalization positively relates to the formal institutional evolvement.
Furthermore, this study utilizes the widely studied perspective of institutional theory, namely firms' institutional embeddedness perspective, and highlights the less studied perspective, namely firms' agent perspective.From this perspective, this study also contributes to institutional theory, especially providing empirical evidence regarding MNEs' institutional entrepreneurship in emerging markets.
In addition, the robust evidence of MNEs' dual agency role in emerging markets contributes to the long debate on the effects of globalization.For instance, in the environmental management area, some scholars [131] argue that firms tend to "race to the bottom," escaping from countries such as advanced economies with stricter environmental regulations and rules to other countries such as emerging markets with lax environmental regulations.However, some scholars assert that MNEs indeed "race to the top," introducing innovative technologies and better management systems from advanced economies to meet various stakeholders' requirements in host countries [132,133].Our findings substantially support the positive effects of globalization embodied by the presence of MNEs.
As a result, this study provides valuable implications for managers and policymakers.Managers of local firms in emerging markets are suggested to take advantage of every opportunity to learn CSR from MNEs to better adapt to the institutional evolvement, as institutional evolvement is an inevitable result of economic globalization in emerging markets.For policymakers, better-informed foreign direct investment policies can be further developed to accommodate more MNEs with good CSR performance in emerging markets.Meanwhile, policymakers can also learn from and cooperate with qualified MNEs to facilitate a better institutional environment for both MNEs and domestic firms to conduct CSR.
Lastly, this study is not without limitations.This study is based on a national-level dataset, making it possible to conduct a systematic analysis across a set of developing countries.However, this approach sacrifices some detailed insights that can only be reached based on a fine-grained dataset.Hence, the merits of this study also imply some future research opportunities.For instance, future studies can examine whether and how different configurations or compositions of MNEs with variant entry modes, size, and country of origin in host countries may lead to possibly different consequences regarding the host countries' institutional evolvement and overall CSR involvement.
05 level) and positive influence (b = 0.270, p = 0.014).The cultural dimension of power distance has a significant and positive influence in both Path C (b = 0.011, p = 0.015) and Path A (b = 0.121, p = 0.010).The cultural dimension of uncertainty avoidance has positive and significant effects in Path C (b = 0.348, p = 0.006) and Path A (b = 0.429, p = 0.000).The cultural dimension of masculinity also have significant and positive influences in Path C (b = 0.409, p = 0.005) and Path A (b = 0.286, p = 0.011).

Table 1 .
Descriptive statistics and correlations.

Table 3 .
Results with instrumental variable.