Sustainability ‘Best Practice’ Spillovers

Abstract

Current research has made significant progress in exploring the sustainability efforts made by domestic and foreign firms, respectively, but paid insufficient attention to the interaction of these firms in the form of sustainability practice spillovers. This paper aims to fill this gap by discussing the spillovers of ‘best practices’ of corporate environmental sustainability (CES) from multinational enterprises (MNEs) that have made increasing investment in green production in recent decades to local forms in host developing countries where environmental protection is relatively weak. In line with internalization theory, we contend that MNEs have to internalize CES ‘best practices’ in their affiliates across the globe, and such practices can spill over to local firms in host developing countries with poorer CES practices. The level of development of press freedom in the host developing country conditions the CES practice spillovers. This study tests hypotheses against firm-level data from a large-scale survey and finds robust evidence to support our argument. This study takes a quantitative approach to unveil the existence and boundary conditions of CES practice spillovers and suggests that policymakers need to facilitate the spillovers of such practices and that scholars need to further advance research in this area.

1. Introduction

The nexus between national institutions, international business actors, and sustainability practices has been a focus of international business (IB) research for a long time. Earlier research has focused on how the institutional void, particularly the lack of environmental laws and regulations, in developing countries has induced multinational enterprises to transfer environment-unfriendly technologies to affiliates in developing countries to create ‘pollution heaven’. Recent research started to examine the ‘halo effect’ that MNEs help improve the natural environment of host developing countries [1,2,3].

This paper aims to examine the global diffusion of corporate environmental sustainability (CES) practices from MNE affiliates to local firms in host developing countries through externalities. CES practices refer to the applications of a construct whereby companies integrate environmental concerns in their business operations and, as such, treat the environmental ‘subset’ of corporate social responsibility in the discourse on the global responsibility of MNEs [4] (p. 565) and [5,6]. A focus in the discourse has been on whether and how MNEs improve or damage the natural environment of host developing countries [3,6]. Despite the dominance of the ‘pollution heaven’ thesis that MNEs transfer environment-unfriendly technologies to developing countries where environment protection is weak in the early debate, recent research has increasingly observed a ‘halo effect’ that MNEs help improve the natural environment of host developing countries [7,8,9]. MNEs have significantly increased CES investment under pressure from domestic and international stakeholders in recent years [10,11].

Advocates for the ‘halo effect’ have so far focused on two ways by which MNEs help improve the natural environment: (1) MNEs directly transfer CES practices to their affiliates operating in host developing countries; (2) MNEs contribute to the economic growth of host developing countries, which, according to the environmental Kuznets Curve, translates into better environmental protection [1]. Accordingly, they have estimated the ‘halo effect’ by examining the positive association of FDI presence with ‘green’ measures such as pollutant emission or energy savings in a host developing country or region as a whole [1,2]. Despite its insight, this research stream neglects an important channel by which the ‘halo effect’ may take place—the spillovers of CES practices from MNEs and their affiliates to local firms in host developing countries.

This paper aims to fill this research gap by investigating the existence and boundary conditions of CES practice spillovers. We draw on internalization theory and examine CES practice spillovers that are linked to MNEs’ decisions to internalize CES in their affiliates. We contend that in the internalization process, MNE affiliates develop superior CES practices that can spill over to local firms in host developing countries where the level of CES practices is relatively low [8,11]. As a mechanism to detect and convict environmentally irresponsible behavior, press freedom in the host country plays a key role in moderating CES practice spillovers. We test our theoretical framework against recent firm-level data, and find supporting evidence: (1) the level of CES practices is significantly higher in MNE affiliates than local firms in a host developing country—a precondition for CES practice spillovers; (2) a greater presence of FDI is associated with a higher level of CES practices of local firms in a host developing country—an indicator that CES practice spillovers have taken place; and (3) press freedom in a host developing country helps enhance CES practice more in local firms than MNE affiliates to narrow the CES practice differential, and strengthens CES practice spillovers from MNE affiliates to local firms. These findings have important public policy implications.

2. Theoretical Framework and Hypotheses

2.1. Internalization and CES Differential

We draw on internalization theory, as explicated in Buckley and Casson’s 1976 book and their subsequent writings, to form the theoretical base of this paper [12,13,14,15,16]. Buckley and Casson [16] (p. 33) noted that firms maximize profit in a world of imperfect markets. When markets in intermediate products are imperfect, there is an incentive to bypass them by creating internal markets in intermediate products. The internalization of markets in intermediate products across national boundaries generates MNEs. Buckley and Casson were particularly interested in the internalization of markets for a particular kind of intermediate product, intangibles including CES, in the emerging new international business landscape in which externalized and semi-externalized activities gain importance compared to vertically integrated activities [8,12].

We take this internalization approach and consider CES to be intangibles that MNEs need to own, control, and internalize in the affiliates they establish in different parts of the globe. MNEs invest in CES initiatives to develop intangible global CES brands. In marketing terms, a brand is a name, symbol, or any other feature that identifies one seller’s good or service as distinct from those of other sellers [17]. We define a CES brand as the CES component of a corporate brand [18,19]. Consumers are increasingly environment-conscious and are attracted to products or services associated with a better CES brand [19]. This ‘ethical consumerism’ has made a global CES brand an effective way to differentiate product offerings to attract customers [20] (p. 281) and [21,22,23]. Many MNEs have taken it as an important strategy to integrate green programs in building a global brand [23]. They broadcast their green initiatives with the ultimate goal of building global CES brands [24,25].

A corporate brand may comprise many components already examined in conventional research, each of which has its own brand equity, such as super advanced technology, super reliability and quality, super customer service, and super delivery convenience [19,23]. The CES brand equity differs from these conventional brand equities in an important way. That is, products or services associated with most conventional brand equities are either search goods with qualities that consumers can assess before purchase (e.g., super advanced technology) or experiential goods with qualities that consumers can assess after purchase (e.g., super reliability and quality, super customer service, and super delivery convenience). In contrast, products or services associated with CES brand equity are credence goods with qualities that consumers cannot assess even after repeated purchases [26]. Instead, consumers assess a firm’s CES commitment by evaluating the firm’s CES behavior. As MNEs operate through global value chain, one CES breach at one of the affiliates in the global network might destroy the entire CES image of the MNE even if all other affiliates under its brand behave in an environmentally responsible manner [26,27].

As a brand owner, therefore, an MNE needs to internalize the global CES brand it owns throughout the entire organization across national borders to avoid or, at least, minimize the possible damage caused by CES negligence in any of its affiliates [8,12]. As in the internalization of other intangibles, the internalization of a CES brand can be achieved by focusing on compatible CES practices when an MNE establishes affiliates through a greenfield venture, acquisition, or joint venture [12,13,14,15,16]. More importantly, the internalization can be achieved through the internal diffusion of CES practices, i.e., the intraorganizational transfer of CES practices from the MNE headquarters to the affiliates they have established, and thus require additional investment support [28,29,30,31]. On the part of MNE affiliates, they have incentives to embrace the transferred CES practices [32,33,34,35].

In consideration of the effort made by both the MNE headquarters and their affiliates in the internalization of CES practices and the abundant financial and technological resources they have to invest in CES initiatives, we expect that the level of CES practices is higher in MNE affiliates than local firms of a host developing country where CES standards are typically low—a precondition for CES practice spillovers to the local firms of host developing countries. This expectation is based on internalization theory, and consistent with recent research in IB [36,37,38]. We thus develop the following hypothesis.

Hypothesis 1.

MNE affiliates have a higher level of CES practices than local firms in a host developing country.

2.2. Spillovers of CES Practices to Local Firms

In the internalization of CES intangibles analyzed above, MNEs can capture the benefits of CES practice improvement in their affiliates to varying degrees. More often than not, however, the benefits of improved CES practices in MNE affiliates may be captured by local firms in the host developing country. That is, the improved CES practices in MNE affiliates spill over to the local firms. CES practice spillovers are externalities that occur when the presence of MNEs, often in the form of the affiliates they establish via FDI, in a host country leads to improvements in the CES practices of local firms such that the focal MNEs cannot capture all the benefits created by the CES practice improvements. Similarly to technology spillovers that have been extensively examined in IB research, CES spillovers are informal and concern the nonmarket diffusion of information, knowledge, and practices from MNE affiliates to local firms, including business partners and competitors [39].

Moreover, as in the case of technology spillovers, CES practice spillovers arise from the differential in the level of CES practices between MNE affiliates and local firms, and are based on the assumption that MNE affiliates have introduced better CES practices and have obtained a higher level of CES practices than local firms as, explicated in our Hypothesis 1. Unlike technology spillovers, which are reflected in the improved productivity of local firms, CES spillovers are reflected in improved CES practices in local firms. As noted previously, MNE affiliates tend to make great effort to adapt the CES practices transferred from MNE headquarters to the local context and thus make the CES practices of the affiliates more ‘localized’, easier to spill over, and more likely to be accepted by local firms [30] (p. 390) and [31,32].

Compared to technology spillovers, which often involve industry-specific or even project-specific techniques and know-how, CES practice spillovers are often generic, involving general CES routines and measures such as the use of green resources and materials, reduction in wastes, and monitoring, auditing, and target setting for energy and emissions, which can be applied to most, if not all, industries. Therefore, while technology spillovers take place mainly within the same industry [intra-industry spillovers) or at most between forward-linked and backward-linked industries (inter-industry spillovers), CES practice spillovers can be location-bound or country-bound—a focus of the present study [39]. Consequently, while research on technology spillovers uses FDI at the industry level as the unit of sources of technology spillovers, research on CES practice spillovers can use country-level FDI as the unit of sources of CES practice spillovers. If CES practice spillovers take place, there should be a strong association between the presence of FDI in a host developing country and the level of CES practices of local firms in the country. We propose the following hypothesis.

Hypothesis 2.

The presence of FDI in a host developing country is positively related to the level of CES practices of local firms in the country.

2.3. The Role of Press Freedom

The diffusion of CES can encounter difficulties in MNEs’ host countries in the developing world where many business behaviors and practices that are ‘environment-unfriendly’ by international standards have been taken for granted or even necessary [6,37]. In particular, with the rapid industrialization and urbanization in the global integration process, many local firms in developing countries get into businesses that harm the natural environment. The demand for, and the unsustainable use of, energy resources has placed stress on the ecosystem. This has led to growing concerns over serious ecological and health crises in the wider public of developing countries.

Local firms are reluctant to address these problems because these environment-unfriendly business practices are so deeply imbedded in the peculiar competition context that they are, to a large extent, institutionalized. To local firms, addressing these environmental problems involves high costs in financial resources, hurts the short-term interests of their business and shareholders in competition with rivals, or even makes their own business simply unviable [37]. Even if the CES practice spillovers from MNE affiliates to local firms are ‘free of charge’, local firms are often unwilling to accept them. In other words, local firms in developing countries have strong motivation not to follow the international CES standards, and not to embrace the CES spillovers from MNE affiliates in order to achieve short-run financial gains at the cost of the natural environment and the wide public.

We propose that free press serves as an external detection and conviction mechanism and can mitigate the reluctance and resistance to the diffusion of CES practices in developing countries. Environmental irresponsibility involves more moral and ethical issues than legal and criminal ones. Nevertheless, in the context of growing awareness of CES and public health consequences of environmental pollution in developing countries, the offenders of CES norms are similar to offenders of laws and public rules to the extent that they are risk-inclined [4]. Accordingly, they are deterred more by the probability of detection and conviction via free press than by punishment if detected or convicted. That is, free press can serve as an effective weapon to curb environmentally irresponsible behavior and promote CES diffusion in developing countries. Thus, press freedom is likely to play a key role in moderating the diffusion of CES practices in developing countries [4,27]. Specifically, we expect that press freedom pushes local firms to improve CES standards and thereby helps narrow the CES practice differential between MNE affiliates and local firms. We also expect that press freedom pushes local firms to imitate and learn CES practices from MNE affiliates and thereby helps strengthen the CES practice spillovers. We propose Hypotheses 3a and 3b.

Hypothesis 3a.

Press freedom narrows the differential in the level of CES practices between MNE affiliates and local firms in a host developing country.

Hypothesis 3b.

Press freedom strengthens the association between the presence of FDI and the level of CES practices of local firms in a host developing country.

As shown above, the three hypotheses are designed to address the gap observed in the current research on CES practices in terms of (1) whether and how internalization generates a CES differential between MNE affiliates and local firms that serves as the basis of CES practice spillovers from the former to the latter; (2) whether the CES practice spillovers indeed occur; and whether and how a free press may moderate the spillovers of CES practices. Next, we will test these hypotheses to see whether our arguments are supported by evidence.

3. Methodology

3.1. Data Source

The sample comes from the EBRD-EIB-World Bank Group Enterprise Survey (the Survey hereafter) conducted by the European Bank for Reconstruction and Development (EBRD), European Investment Bank (EIB), and the World Bank in 19 countries in Eastern Europe, Central Asia, and the Middle East in 2018–2019, including Albania, Belarus, Bosnia, Bulgaria, Kazakhstan, Kyrgyz, Lebanon, Moldova, Mongolia, Montenegro, Morocco, North Macedonia, Russia, Serbia, Tajikistan, Turkey, Ukraine, Uzbekistan, and the West Bank. According to the economy classification of both the World Bank and IMF, the 19 economies are not on the list of advanced economies and are, therefore, considered to be in the group of developing economies in this study (These economies are classified as middle- or low-income economies by the World Bank, and emerging and middle-income economies or low-income and developing economies by IMF. For the purpose of simplicity in terminology, we name them developing economies for short.).

In addition to the usual firm-level information provided in the Enterprise Survey in previous years, the survey includes a follow-up ‘Green Economy Module’ specifically related to firm-level CES practices. The Green Economy module is the main data source of this study. The dataset contains 13,655 firms. The whole sample was used to examine the differential in the level of CES practices between MNE affiliates and local firms. In addition, we compiled a local firm sample and used it to examine CES practice spillovers to local firms. There are 12,652 local firms and 1003 MNE affiliates. Due to missing information for some variables in some firms, the actual sample size differs depending on the variables included in individual regression models. Missing data were removed from the regression, and there were no outliers in the dataset.

3.2. Measurement

3.2.1. Dependent Variable

CES practices include the management routines a firm adopts to address environmental concerns as well as the measures a firm takes to actually reduce damage to the natural environment in daily operations [4,7]. The survey includes several questions on how the firm routinely managed CES in energy and emission. The questions ask whether the firm (1) ‘monitored its energy consumption’, (2) ‘completed an external audit of its energy consumption’, (3) ‘had targets for its energy consumption’, (4) ‘monitored its CO₂ emissions’, (5) ‘completed an external audit of its CO₂ emissions’, (6) ‘had targets for its CO₂ emissions’, (7) ‘monitored its emissions of pollutants other than CO₂’, (8) ‘completed an external audit of its emissions of pollutants other than CO₂’, and (9) ‘had targets for its emissions of pollutants other than CO₂’. We constructed a binary variable for each of the nine questions, with 1 denoting Yes, and 0 otherwise.

The survey also includes a question on whether the firm took measures to improve the usage of energy and emissions of CO₂ and other pollutants in daily operations in nine areas, including (1) ‘heating and cooling improvements’, (2) ‘more climate-friendly energy generation on site’, (3) ‘machinery and equipment upgrades’, (4) ‘waste minimization and recycling’, (5) ‘air pollution control’, (6) ‘water control’, (7) ‘upgrades of vehicles’, (8) ‘improvements to lighting systems’, and (9) ‘other pollution control measures’. We constructed a binary variable for each of the nine areas, with 1 denoting Yes, and 0 otherwise. We added the 18 binary variables to construct a variable, named CES practice, to represent the aggregate level of CES practices of the firm. The variable ranges from 0 to 18.

3.2.2. Independent Variable and the Moderator

We followed the United Nations Conference on Trade and Development [40] and constructed a binary variable, MNE affiliate, with 1 denoting firms with more than 10% of foreign owned assets in a host country, and 0 otherwise. This variable is used as an independent variable in regressions to test the differential in the level of CES practices between MNE affiliates and local firms. Two alternative threshold percentages of foreign-owned assets were used in the robustness test.

Given that MNE affiliates are established through FDI, the presence of FDI has been widely used as the source of technology spillovers from these affiliates to local firms in empirical studies [39]. Following this approach, we constructed a variable, FDI presence, using the average ratio of FDI inflows to gross fixed capital formation in the host country in the past four years from 2016 to 2019. This variable was taken as an independent variable in regressions to test the spillovers of CES practices from MNE affiliates to local firms. Data of country-level FDI inflows came from UNCTAD [40].

For the moderator, press freedom, we used the press freedom index compiled by Reporters without Borders, which ranks the degree of press freedom of countries from 1 to 100, with 1 representing the highest level of press freedom. To facilitate interpretation, we reversed the scale, with 100 representing the highest level of press freedom.

3.2.3. Control Variable

We need to control for the influences of other factors. Unfortunately, there is little research on the factors that influence CES practice spillovers. In selecting control variables, we primarily draw on existing studies on the factors that influence the diffusion of CES practices within an MNE and the literature on innovation and technology spillovers. At the firm level, we constructed a variable to control for firm age using the logarithm of the difference between the survey year and the year when the firm was established [39]. We constructed a variable to control for firm size using the logarithm of the number of employees [39]. We constructed a variable to control for manager experience using the logarithm of the number of years when the general manager had held their position in the current firm [39]. We used the reversed percentage of the main product in total sales to control for product diversification [39]. We constructed a binary variable to control for R&D based on the question of whether the firm invested in R&D [39]. We also constructed a binary variable to control for green strategy based on the question of whether the firm had strategic objectives that mention environmental or climate change issues. Moreover, the moderator, press freedom, was included as a control in regression without its interaction with the independent variables.

At the industry level, we controlled for industry affiliation using dummies based on the single-digit United Nations ISIC Rev.3.1 classification. At the national level, we constructed dummies to control for country of origin. The descriptive statistics and correlation matrix of major variables in the whole sample and the local firm sample are reported in

Table 1. Mean, standard deviation, and correlation of variables for the whole sample.

Variable	Mean	SD	1	2	3	4	5	6	7	8
1. MNE affiliate	0.070	0.261
2. Firm age	2.644	0.771	−0.008
3. Firm size	3.353	1.372	0.206 **	0.245 **
4. Manager experience	2.689	0.775	−0.049 **	0.482 **	0.088 **
5. Product diversification	0.120	0.188	−0.009	0.039 **	0.010	−0.024 **
6. R&D	0.080	0.264	0.074 **	0.056 **	0.181 **	0.011	0.038 **
7. Green strategy	0.160	0.364	0.104 **	0.097 **	0.219 **	0.000	0.082 **	0.168 **
8.CES practice	3.820	2.698	0.106 **	0.105 **	0.271 **	0.006	0.043 **	0.206 **	0.341 **
9. Press freedom	4.059	0.163	0.041 **	0.299 **	0.029 **	0.219 **	0.093 **	0.062 **	0.096 **	0.038 **

** p < 0.01 level.

and

Table 2. Mean, standard deviation, and correlation of variables for the local firm sample.

Variable	Mean	SD	1	2	3	4	5	6	7	8
1. FDI presence	0.171	0.416
2. Firm age	2.645	0.768	0.111 **
3. Firm size	3.274	1.335	−0.038 **	0.250 **
4. Manager experience	2.700	0.772	0.065 **	0.498 **	0.097 **
5. Product diversification	0.121	0.187	0.017	0.033 **	0.018 *	−0.028 **
6. R&D	0.070	0.255	−0.001	0.060 **	0.179 **	0.011	0.031 **
7. Green strategy	0.150	0.353	0.006	0.099 **	0.205 **	0.006	0.082 **	0.161 **
8.CES practice	3.740	2.640	0.034 **	0.110 **	0.260 **	0.013	0.043 **	0.202 **	0.336 **
9. Press freedom	4.057	0.163	0.200 **	0.310 **	0.019 *	0.229 **	0.094 **	0.054 **	0.093 **	0.038 **

** p < 0.01 level; * p < 0.05.

, respectively.

3.2.4. Estimation Strategy

To test Hypothesis 1 about the level of CES practices of MNE affiliates in comparison to that of local firms in the host country, we estimate the following equation.

$${CERPractice}_i=\mathrm{a}+{\beta }_1{MNEaffiliate}_i+{\beta }_2{X}_{ijc}+{\epsilon }_i$$

(1)

Here, i stands for firm; _j stands for industry; and c stands for country. a represents the intercept, whereas ${\epsilon }_i$ indicates the error term. X stands for a vector of control variables. In the empirical test, therefore, CES practice in firm i is regressed on the MNE affiliate dummy in addition to a vector of control variables. Here, ${\beta }_1$ represents the differential effect of being an MNE affiliate on CES practices.

To test Hypothesis 2 about CES practice spillovers from MNE affiliates to local firms, we estimate the following equation.

$${CERPractice}_{di}=\mathrm{a}+{\beta }_1{FDIpresence}_c+{\beta }_2{X}_{dijc}+{\epsilon }_i$$

(2)

Here, di stands for domestic firm. In the empirical test, therefore, CES practice in local firm di is regressed on FDI presence in the host country in addition to a vector of control variables. ${\beta }_1$ represents the effect of FDI presence in the host country on the CES practices of local firms.

To test Hypothesis 3a about the role of press freedom in moderating CES practice differential, we constructed an interaction term of MNE affiliate with press freedom for Equation (1). To test Hypothesis 3b about the role of press freedom in moderating CES practice spillovers, we similarly constructed an interaction term of FDI presence with press freedom for Equation (2). To mitigate multicollinearity in regressions with interaction terms, we transformed the continuous variables in the interactions, press freedom, and FDI presence into z-scores by normalizing the variable to mean zero and standard deviation one as shown in Equation (3):

$${z_v}_i={\displaystyle \frac{v_{i-}{\overline{v}}_i}{{{\sigma }_m}_i}}$$

(3)

where ${z_v}_i$ is the z-score of the variable v_i in firm i. ${\overline{v}}_i$ is the unweighted average of variable v_i across all observations. ${{\sigma }_m}_i$ is the standard deviation of variable v_i across all observations.

Since the dependent variable, CES practice, is measured by count data, it is not appropriate to use the OLS approach in regressions. We employed the Poisson approach instead. In estimating these equations, we corrected all estimates for heteroskedasticity using White’s heteroscedastic consistent standard errors.

4. Result

4.1. Hypothesis Test for CES Differential and Spillovers

The results of the hypothesis tests are reported in

Table 3. Results of Poisson regression for hypothesis test.

Dependent Variable: CES Practice	Whole Sample to Test CER Differential		Local Firm Sample to Test CER Spillovers
	(1)	(2)	(3)	(4)
Firm age	0.0425 ***	0.0421 ***	0.0369 ***	0.0369 ***
	(0.009)	(0.009)	(0.010)	(0.010)
Firm size	0.0839 ***	0.0840 ***	0.0879 ***	0.0877 ***
	(0.004)	(0.004)	(0.005)	(0.005)
Manager experience	0.00106	0.000759	0.00299	0.000455
	(0.009)	(0.009)	(0.009)	(0.009)
Product diversification	0.107 ***	0.107 ***	0.0763 *	0.107 ***
	(0.030)	(0.030)	(0.031)	(0.031)
R&D	0.208 ***	0.210 ***	0.210 ***	0.205 ***
	(0.020)	(0.020)	(0.021)	(0.021)
Green strategy	0.394 ***	0.394 ***	0.407 ***	0.412 ***
	(0.015)	(0.015)	(0.016)	(0.016)
Press freedom	0.0655 ***	0.0713 ***	0.0678 ***	0.119 ***
	(0.008)	(0.008)	(0.008)	(0.011)
MNE affiliate	0.0820 ***	0.0597 **
	(0.021)	(0.022)
MNE affiliate × Press freedom		−0.0614 **
		(0.023)
FDI presence			0.0318 ***	−0.117 ***
			(0.005)	(0.021)
FDI presence × Press freedom				0.304 ***
				(0.042)
Constant	0.531 ***	0.534 ***	0.573 ***	0.532 ***
	(0.039)	(0.039)	(0.041)	(0.041)
Industry dummy	Yes	Yes	Yes	Yes
Country of origin dummy	Yes	Yes	Yes	Yes
Observations	12,128	12,128	11,270	11,270
Pseudo R2	0.0774	0.0775	0.0732	0.0747
Wald chi2	3342.45 ***	3354.12 ***	2922.23 ***	2979.39 ***

Notes: * p < 0.05, ** p < 0.01, *** p < 0.001. White’s heteroscedastic consistent standard errors are in “( )” parentheses.

. Hypothesis 1 states that MNE affiliates have a higher level of CES practices than local firms in a host developing country. We estimate Equation (1) using the whole sample. As shown in column 1, the coefficient of MNE affiliate on CES practice is positive and statistically significant (β = 0.0820; p < 0.001). The results fully support Hypothesis 1, indicating that the CES practice score of MNE affiliates is on average 0.0820 points higher than that of local firms.

Hypothesis 2 posits that the presence of FDI in a host developing country is positively related to the level of CES practices of local firms in the country. We estimate Equation (2) using the domestic firm sample. As shown in column 3, the coefficient of FDI presence on CES practice is positive and statistically significant (β = 0.0318; p < 0.001). The results fully support Hypothesis 2, indicating that an increase of one standard deviation in the FDI inflow ratio is, on average, associated with a 0.0318 point increase in the CES practice score of local firms in the host developing country.

It should be noted that there may be reversal causality in estimating Equation (1). That is, an MNE may establish affiliates by acquiring the share of local firms with a higher level of CES practices. As explicated in our theoretical model earlier, however, the reversal causality is an integral part of the CES internalization process and does not alter what we aim to establish in the test—MNE affiliates have obtained a higher level of CES than local firms. The CES practice differential may result from CES internalization either through the intraorganizational transfer of CES practices from MNEs to the affiliates they have established or through the acquisition of shares of local firms with compatible CES practices when MNEs are establishing affiliates. Reversal causality is not a concern in the empirical estimation since it is inherent in our theory. In estimating Equation (2), moreover, the independent variable, FDI presence in a host country over the past four years, is considered to be exogenous—it would be unreasonable to contend that the current level of CES practices in individual local firms influences the accumulated level of FDI inflows into the host country in the past four years. In addition, we include country dummies to minimize omitted variable biases at the country level—another source of endogeneity regarding country-level FDI. Thus, as in existing research on technology spillovers, endogeneity is not an issue in the empirical estimation of CES practice spillovers via FDI.

4.2. Hypothesis Test for the Role of Press Freedom

Hypothesis 3a posits that press freedom narrows the differential in the level of CES practices between MNE affiliates and local firms in a host developing country. We include an interaction term of MNE affiliates with press freedom in estimating Equation (1). As shown in column 2, the coefficient of the interaction term is negative and statistically significant (β = −0.0614; p < 0.01), indicating that press freedom helps enhance CES practices of local firms more than MNE affiliate and thus narrows the CES differential. On average, a one standard deviation increase in press freedom leads to a 0.0713 point increase in CES practices in local firms, but only a 0.0099 point increase in CES practices in MNE affiliates. This does not come as a surprise given that MNE affiliates typically have a higher level of CES practices and are less subject to the influence of a free press in CES behavior than local firms.

Hypothesis 3b posits that press freedom strengthens the association between the presence of FDI and the level of CES practices of local firms in a host developing country. We include an interaction term of FDI presence with press freedom in estimating Equation (2). As shown in column 4, the coefficient of the interaction term is positive and statistically significant (β = 0.304; p < 0.001), indicating that press freedom helps strengthen the CES spillovers from MNE affiliates to local firms. Specifically, a one standard deviation increase in press freedom is associated with a 0.304 point increase in the effect of FDI presence on CES practices of local firms. Moreover, not surprisingly, the coefficient of press freedom is positive and statistically significant in all regressions, indicating the critical role it plays in enhancing firm-level CES practices.

4.3. Robustness Test and the Effect of Control Variables

In the empirical test, we distinguish between MNE affiliates and local firms by a threshold of 10% of foreign-owned assets. It may be argued that this threshold may not accurately reflect differences between the two kinds of firms. To address this issue, we use 0% and 50% of foreign-owned assets as a threshold to distinguish the two kinds of firms, respectively, and rerun the regression analysis. As shown in

Table 4. Results of Poisson regression for robustness test (local firms without foreign share).

Dependent Variable: CES Practice	Whole Sample to Test CER Differential		Local Firm Sample to Test CER Spillovers
	(1)	(2)	(3)	(4)
Firm age	0.0420 ***	0.0417 ***	0.0369 ***	0.0368 ***
	(0.009)	(0.009)	(0.010)	(0.010)
Firm size	0.0837 ***	0.0839 ***	0.0876 ***	0.0875 ***
	(0.004)	(0.004)	(0.005)	(0.005)
Manager experience	0.00122	0.000951	0.00256	−0.00000383
	(0.009)	(0.009)	(0.009)	(0.009)
Product diversification	0.107 ***	0.107 ***	0.0783 *	0.110 ***
	(0.030)	(0.030)	(0.031)	(0.031)
R&D	0.208 ***	0.209 ***	0.209 ***	0.204 ***
	(0.020)	(0.020)	(0.021)	(0.022)
Green strategy	0.394 ***	0.395 ***	0.408 ***	0.413 ***
	(0.015)	(0.015)	(0.016)	(0.016)
Press freedom	0.0654 ***	0.0719 ***	0.0687 ***	0.121 ***
	(0.008)	(0.008)	(0.008)	(0.011)
MNE affiliate	0.0802 ***	0.0569 **
	(0.020)	(0.022)
MNE affiliate × Press freedom		−0.0651 **
		(0.023)
FDI presence			0.0320 ***	−0.118 ***
			(0.005)	(0.021)
FDI presence × Press freedom				0.307 ***
				(0.042)
Constant	0.533 ***	0.536 ***	0.576 ***	0.535 ***
	(0.039)	(0.039)	(0.041)	(0.041)
Industry dummy	Yes	Yes	Yes	Yes
Country of origin dummy	Yes	Yes	Yes	Yes
Observations	12,128	12,128	11,240	11,240
Pseudo R2	0.0773	0.0776	0.0722	0.0737
Wald chi2	3346.91 ***	3354.32 ***	2862.39 ***	2920.89 ***

Notes: * p < 0.05, ** p < 0.01, *** p < 0.001. White’s heteroscedastic consistent standard errors are in “( )” parentheses.

and

Table 5. Results of Poisson regression for robustness test (local firms with 50% or less of foreign share).

Dependent Variable: CES Practice	Whole Sample to Test CER Differential		Local Firm Sample to Test CER Spillovers
	(1)	(2)	(3)	(4)
Firm age	0.0419 ***	0.0418 ***	0.0379 ***	0.0379 ***
	(0.009)	(0.009)	(0.010)	(0.010)
Firm size	0.0857 ***	0.0857 ***	0.0889 ***	0.0885 ***
	(0.004)	(0.004)	(0.005)	(0.005)
Manager experience	0.000150	−0.0000852	0.00117	−0.00162
	(0.009)	(0.009)	(0.009)	(0.009)
Product diversification	0.106 ***	0.106 ***	0.0724 *	0.104 ***
	(0.030)	(0.030)	(0.031)	(0.031)
R&D	0.209 ***	0.210 ***	0.210 ***	0.204 ***
	(0.020)	(0.020)	(0.021)	(0.021)
Green strategy	0.396 ***	0.396 ***	0.406 ***	0.412 ***
	(0.015)	(0.015)	(0.016)	(0.016)
Press freedom	0.0662 ***	0.0695 ***	0.0664 ***	0.118 ***
	(0.008)	(0.008)	(0.008)	(0.011)
MNE affiliate	0.0571 *	0.0430 †
	(0.024)	(0.025)
MNE affiliate × Press freedom		−0.0472 †
		(0.027)
FDI presence			0.0330 ***	−0.117 ***
			(0.005)	(0.021)
FDI presence × Press freedom				0.306 ***
				(0.041)
Constant	0.536 ***	0.537 ***	0.567 ***	0.527 ***
	(0.039)	(0.039)	(0.040)	(0.040)
Industry dummy	Yes	Yes	Yes	Yes
Country of origin dummy	Yes	Yes	Yes	Yes
Observations	12,128	12,128	11,489	11,489
Pseudo R2	0.0771	0.0772	0.0753	0.0768
Wald chi2	3310.77 ***	3307.64 ***	3106.62 ***	3172.86 ***

Notes: ^† p < 0.10, * p < 0.05, *** p < 0.001. White’s heteroscedastic consistent standard errors are in “( )” parentheses.

, the basic results remain virtually unchanged.

Also, for count data, there is the potential problem of overdispersion, which can lead to inefficient or biased estimates if not addressed. Following common practice, we ran a negative binomial regression to address the potential problem of overdispersion as an additional robustness test. As shown in

Table 6. Results of negative binomial regression for robustness test.

Dependent Variable: CES Practice	Whole Sample to Test CER Differential		Local Firm Sample to Test CER Spillovers
	(1)	(2)	(3)	(4)
Firm age	0.0414 ***	0.0409 ***	0.0352 ***	0.0352 ***
	(0.008)	(0.008)	(0.010)	(0.010)
Firm size	0.0843 ***	0.0842 ***	0.0892 ***	0.0890 ***
	(0.004)	(0.004)	(0.005)	(0.005)
Manager experience	0.0009	0.00074	0.0028	0.0004
	(0.008)	(0.008)	(0.009)	(0.009)
Product diversification	0.109 ***	0.109 ***	0.0778 *	0.115 ***
	(0.020)	(0.020)	(0.033)	(0.034)
R&D	0.212 ***	0.214 ***	0.213 ***	0.208 ***
	(0.030)	(0.030)	(0.031)	(0.032)
Green strategy	0.398 ***	0.398 ***	0.411 ***	0.416 ***
	(0.018)	(0.018)	(0.018)	(0.018)
Press freedom	0.0668 ***	0.0724 ***	0.0685 ***	0.121 ***
	(0.009)	(0.009)	(0.009)	(0.014)
MNE affiliate	0.0833 ***	0.0599 **
	(0.023)	(0.024)
MNE affiliate × Press freedom		−0.0616 **
		(0.025)
FDI presence			0.0324 ***	−0.121 ***
			(0.006)	(0.025)
FDI presence × Press freedom				0.308 ***
				(0.048)
Constant	0.533 ***	0.539 ***	0.579 ***	0.538 ***
	(0.039)	(0.039)	(0.043)	(0.043)
Industry dummy	Yes	Yes	Yes	Yes
Country of origin dummy	Yes	Yes	Yes	Yes
Observations	12,128	12,128	11,270	11,270
Pseudo R2	0.0778	0.0779	0.0735	0.0749
Wald chi2	3351.32 ***	3356.18 ***	2928.34 ***	2981.41 ***

Notes: * p < 0.05, ** p < 0.01, *** p < 0.001. White’s heteroscedastic consistent standard errors are in “( )” parentheses.

, the results remained virtually unchanged.

We followed recent practice and do not discuss control variables in the results to save space. However, we would like to highlight that five control variables, firm age, firm size, product diversification, green strategy, and R&D, have a positive and statistically significant coefficient in all the regressions. The findings are consistent with prior research on the vital role of these variables in corporate responsibility practices.

5. Discussion

5.1. Contribution to the Literature

Global warming has attracted growing attention in scholarly research, as does the role MNEs play in the worldwide battle against it. Despite progress in research on the significant increase in the investment MNEs have made to develop CES brands across the affiliates they have established in recent years, little is known about the spillovers of CES practices from MNE affiliates to local firms in host countries across the developing world. We believe that the lack of research in this area is attributable to the disconnection between CES research and IB theory. On the one hand, CES issues have been discussed as if they were not closely related to IB theory and have not been adequately integrated into conceptual models of IB theory. On the other hand, IB theory has paid insufficient attention to CES issues, particularly CES externalities, and has not adequately incorporated these issues into theoretical modeling.

The most important contribution of this paper is to incorporate CES into internalization theory. Treating investment in CES as efforts to develop global brands, we posit that the need for the internalization of intangible CES brands drives MNEs to pay attention to CES practices when they establish affiliates, and transfer CES practices to the affiliates once they have been established. Comparing a CES brand with conventional intangibles, we contend that the nature of credence goods associated with the CES brand equity makes it imperative for MNEs to focus on the development of CES practices in every affiliate they establish across the world. Coupled with the advantages MNEs have in financial and technological resources for CES initiatives, the ‘deep’ internalization of CES results in a higher level of CES practices in MNE affiliates than that of local firms in developing countries and thereby creates the precondition for externalities in the form of spillovers of CES practices to the local firms [30,38].

Importantly, we examine for the first time the similarities and dissimilarities between CES practice spillovers and conventional technology spillovers. Similarities include the following: (1) both of them arise from the differential between MNE affiliates and local firms in the level of CES practices (for CES practice spillovers) or technology (for technology spillovers); (2) both of them occur through a competition effect, a linkage effect, a demonstration effect, and an employment effect; (3) both of them are externalities, i.e., the benefits of the economic activity of the focal MNE received by a third party; and (4) for both of them, the receiving third party may include business partners and competitors [39]. Dissimilarities include the following: (1) CES practice spillovers may occur in a more cooperative than competitive manner compared to technology spillovers since climate change needs collective action; (2) CES practice spillovers are more generic and less industry-specific or project-specific in nature than technology spillovers since they often involve general routines and measures that apply across industries; and (3) whereas technology spillovers are mainly reflected in improved productivity of local firms, CES practice spillovers are mainly reflected in improved CES practices of local firms [6,39]. These distinctive features make it easier for CES practices to spill over from MNE affiliates to local firms than technology, and make it easier for local firms to adopt the ‘spilled’ CES practices than the ‘spilled’ technology.

Given the widespread violation of ‘green’ business conduct in local firms and the growing awareness of the human cost of such violation in developing countries, external mechanisms that help curb environmental misconduct become an important factor that affects the decision of local firms to adopt CES practices. While attention has been paid to the role of nongovernmental organizations in this regard [41] (p. 24), we contend that a free press plays an important role in influencing public opinion and serves as a detection and conviction mechanism to push local firms to improve CES standards. Specifically, we conceive and demonstrate that press freedom enhances the level of CES practices of local firms more than that of MNE affiliates in developing countries and thereby narrows the CES practice differential in CES practices between the two kinds of business entities. Also, press freedom pushes local firms to imitate and learn CES practices from MNE affiliates, and thereby strengthens CES practice spillovers.

5.2. Public Policy Implications

Generally speaking, externalities occur when the action of one party positively affects another party without benefiting from it or negatively affects another party without paying for the damage. A core task of public policy is to address externalities [11]. Prior research on environmental sustainability has examined the negative externalities of MNEs in developing countries, particularly the ‘pollution heaven’ effect, and called for public policy measures to combat the environmental damage caused by foreign investment in developing countries [10] (p. 3). Recent research has observed positive externalities in the form of a ‘halo effect’ that the presence of MNEs is positively related to improvement in the natural environment of the host developing countries, but did not explain the mechanisms by which the positive externalities take place and the public policies that need to be taken to promote the positive externalities [1,2,3].

The present study suggests that the positive externalities take place in the form of spillovers of CES practices from MNEs and their affiliates to local firms [39]. That is, the investments MNEs make in CES are not only directly transferable to their affiliates but also indirectly diffusible to local firms in host countries across the developing world. Policymakers at the national and international levels need to formulate policies to support not only MNEs’ investments in CES practices but also to enhance the spillovers of such practices to local firms in the developing world. These policies may include preferential tax treatments to MNEs who invest in CES practices in their affiliates in host developing countries, and who help local firms in these countries to access and learn such practices [24].

Moreover, this paper indicates that the press freedom of a host developing country facilitates the development of CES practices in local firms. The impact of press freedom is particularly strong in local firms in which local owners hold more than 50% of assets. Importantly, press freedom is found to facilitate CES practice spillovers from MNE affiliates to local firms. It is crucial for public policymakers at the national and international levels to promote the development of press freedom in developing countries in order to enhance CES in local firms and facilitate CES spillovers from MNE affiliates to local firms in host developing countries [7,42]. Policymakers need to collaborate with international media organizations such as Reporters with Border to this end. Scholars may conduct more detailed research in these regards.

5.3. Limitation and Direction for Future Research

Externalities from CES internalization are frontiers in IB research. As a pioneering work, this paper is subject to a number of limitations. To begin with, existing research has pinpointed the importance of an institutional distance between home and host countries to the adoption of corporate responsibility practices [43,44]. CES practice spillovers may occur across the borders of nations with similar or dissimilar institutional matrices [44]. The similarities and dissimilarities in institutions between home and host countries may be an enabler or barrier to CES practice spillovers from MNE affiliates to local firms. Due to data constraints, this paper does not examine the impact of institutional distance on CES practice spillovers to local firms in developing countries. Future research may move in this direction when data become available to unveil the home–host country distance dynamics.

Moreover, existing research has suggested that corporate governance matters to the adoption of corporate responsibility practices [30]. The way by which an organization interacts with another may stimulate or hinder CES practice spillovers, as does the way by which an organization is structured [45]. It is possible that a more flexible and flat organizational structure provides managers with greater autonomy and power in adapting CES practices to local contexts and thus facilitates CES practice spillovers to local firms [46]. In contrast, a more rigid and hierarchical organizational structure may constrain the ability of managers to make flexible adjustment and hinder the development of locally adapted CES practices and the spillovers of such practice to local firms [45]. Given data constraints, this paper does not examine the impact of corporate governance structure on CES practice spillovers to local firms in developing countries. Future research may move in this direction when data become available to unveil the organizational-level dynamics.

Furthermore, existing research has shown that individuals play a crucial role in the adoption of corporate responsibility practices. The spillovers of CES practices involve managers who differ in CES-related experience. Managers of local firms with a high level of CES-related skills, knowledge, and capabilities are likely to do well in embracing the CES practices spilled over to their firms. In particular, the boundary-spanning skills these managers have may help facilitate the connection of the CES practices of MNE affiliates with the existing practice of local firms [47]. Given data constraints, this paper does not investigate the impact of these factors on CES practice spillovers to local firms in developing countries. Future research may move in this direction when data become available to unveil the individual-level dynamics.

Finally, there are some measurement issues. The CES practice variable has 18 binary variables. Some practices might be more impactful or require more significant effort than others. As it is difficult, if not impossible, to add weights to each of these practices, we implicitly assume that each of these practices has an equal contribution to the overall level of CES practice. Future research may address this issue when appropriate approaches to weight these practices have been developed. Also, the FDI presence variable is constructed using an average ratio of FDI inflows to gross fixed capital formation in the host country over the past four years. Although this is a standard practice in empirical research on spillovers, alternative measures may be used in future studies. This study relies on secondary data from the Enterprise Survey and other publicly available datasets (UNCTAD, Reporters without Borders). While using large-scale survey data has advantages, the researchers have no direct control over the data collection process, the design of the survey instrument, or the potential biases inherent in the data. This is a general limitation of using secondary data. Future research may collect primary data to test whether the findings are robust. In addition, the practical value and significance of this study need to be further explored when new data become available (we thank an anonymous reviewer for this point).

6. Conclusions

Drawing on internalization theory, this paper examines for the first time externalities in the form of CES practice spillovers from MNE affiliates to local firms in developing countries. This paper tests hypotheses against firm-level data and finds robust evidence of CES practice spillovers. Moreover, this paper finds that the press freedom of a host developing country is conducive to not only the development of CES practices of local firms but also CES practice spillovers from MNE affiliates to local firms. This paper enhances our understanding of the mechanisms of the international diffusion of CES practices in the developing world, enriches internalization theory, and provides fresh perspectives on the critical role public policy plays in the global battle against climate change.