Abstract
This study investigates the effects of cultural, institutional and natural ecosystems on corporate response to climate change. We find that national cultural tendencies towards future orientation, uncertainty avoidance, gender egalitarianism and humane orientation strengthen corporate performance, whereas certain other cultural dimensions, such as in-group collectivism, pose barriers to optimal carbon performance. We suggest that culture provides an incremental explanation for corporate climate behaviours beyond socioeconomic or regulatory determinants. Our study contributes to the carbon literature by comprehensively examining the role of a country’s national culture in determining corporate carbon performance, and its findings may help in the implementation of international climate accords (such as the Paris Agreement of 2015) in countries with heterogeneous cultural values and natural environmental contexts.
JEL Classification:
1. Introduction
There is increasing empirical evidence that carbon risks and carbon performance are relevant to investor decision making and that lower carbon emissions (or superior carbon management and performance) are associated with a higher firm value (Chapple et al., 2013; Choi et al., 2021; Choi and Luo, 2021; Clarkson et al., 2015; Griffin et al., 2017; Matsumura et al., 2014; Nguyen, 2018). However, climate change is a very complex issue, characterised by considerable inherent uncertainty (IPCC, 2018; Stern, 2006). This means that it is challenging for managers and investors to evaluate climate risk accurately, predict public policy and anticipate developments in renewable energy technologies. Determining the economic effects of carbon reduction initiatives involves substantial judgement and appraisal by managers. Therefore, managers and various stakeholders are more likely to respond to a climate threat in accordance with their cultural prescriptions. When underlying standards, regulations and norms are uncertain, culture plays a more critical role in managers’ carbon decisions (Leung et al., 2005). In addition, managers must act in conformity with stakeholders’ expectations and perceptions of carbon issues, which tend to rely on cultural contexts as well. Hence, our research question is this: Does national culture as a soft institution have an impact on corporate carbon performance? If yes, do hard institutions (e.g. laws and regulations) play a role in moderating this relationship?
We expect that national culture is likely to be a determinant of heterogeneity in firms’ carbon strategy and performance. However, previous studies have not systematically explored this association. The carbon literature largely focuses on the determinants of voluntary carbon disclosure (see a detailed review in He et al., 2021). Other researchers explore the relationship between carbon disclosure and carbon management/performance (Luo, 2019; Luo and Tang, 2014b; Qian and Schaltegger, 2017) from the perspectives of signalling theory and legitimacy theory. However, to transition efficiently to a low-carbon economy, it is important to have a thorough understanding of how different dimensions of national culture play a role in determining corporate attitudes towards addressing climate change. Climate change is an international issue that requires collaboration from organisations in countries with diverse cultures. Our research attempts to fill this gap in the literature by examining how a firm’s carbon performance is affected by various country, industry and firm-level factors and elucidating differences in the ways in which managers in distinct cultural contexts respond to climate change. The findings could be helpful for investors evaluating carbon risks and opportunities for long-term decision making and for policymakers designing effective carbon regulations and tailoring them based on cultural tendencies.
Our sample includes large companies from 22 countries targeted by the CDP 1 (previously the Carbon Disclosure Project) and publicly disclosed carbon emissions data for 2008–2014. We rely on dimensions of national culture developed from the indices of the Global Leadership and Organizational Behavior Effectiveness (GLOBE) study (House et al., 2004). We define carbon performance as the outcome of a company’s carbon policies and its actions to manage its impact on ecosystems, measured as the intensity of carbon emissions. We find that high values for future orientation (FO), uncertainty avoidance (UA), performance orientation (PO), gender egalitarianism (GE) and power distance (PD) stimulate superior carbon performance. By contrast, our evidence suggests that the cultural dimension of in-group collectivism (IGC) inhibits corporate carbon performance. These findings indicate that national culture acts as a soft and implicit institution (Ioannou and Serafeim, 2012), as opposed to a hard and explicit one (such as laws and regulations), and has incremental influences beyond economic and regulatory incentives. These empirical results hold when confounding influences and sample selection bias are controlled and when alternative measures are used for national culture and carbon performance.
Our article differs from prior research in several ways. First, it fills a void in the literature on culture and corporate climate-related decisions. Prior studies examine the impact of the implementation of an environmental management system, stakeholder pressure and corporate governance on environmental performance (e.g. Haque, 2017; Kumarasiri, 2017; Tung et al., 2014). However, carbon accounting is still underexplored, despite a growing literature in this field (He et al., 2021). Carbon accounting is a unique dimension of corporate social responsibility (CSR), and greenhouse gas (GHG) emissions are different from other pollutants (such as water or chemical waste). Carbon reduction requires specific capabilities and long-term investment (Luo et al., 2013). Thus, extending the focus of CSR studies to carbon is necessary because underlying associative patterns may differ (Chatterji et al., 2009; Strike et al., 2006; Walls et al., 2011).
Second, of the limited carbon studies, only a few consider the impact of cultural factors on corporate climate decisions. A notable exception is Luo and Tang (2016b), who investigate the influence of national culture on corporate carbon disclosure. Both Luo and Tang (2016b) and the current study shed light on carbon issues and provide the critical and comprehensive insight that managers’ climate-related decisions are influenced by multifaceted cultural values. Some cultural values sharpen the trade-off between financial priorities and ecological considerations (e.g. PD), whereas others help reconcile or mitigate the tension between financial performance and carbon performance (e.g. GE or femininity). Our article makes a unique and incremental contribution beyond that of prior studies (e.g. Faisal et al., 2018; Luo and Tang, 2016b; Mohamed Adnan et al., 2018), which narrowly focus on carbon disclosure or CSR disclosure. This study specifically focuses on carbon performance. Although disclosure and performance are somewhat related, they are different constructs. We cannot readily generalise the findings of the extant carbon disclosure studies to the carbon performance context, which deserves a separate study. Carbon performance refers to ‘how well firms do at managing and controlling carbon emissions’ (He et al., 2021: 23), whereas carbon disclosure refers to a description of carbon reduction actions and outcomes. By nature, their difference reflects a difference between reality and a description of reality (Velte et al., 2020). García-Sánchez et al. (2021) highlight that voluntary carbon disclosure is often decoupled from actual carbon performance, particularly when firms have poor carbon performance or negative news. Although carbon disclosure intends to improve the transparency of carbon activities, enhance the confidence of users and strengthen firm relationships with stakeholders, some prior studies find that carbon disclosure and carbon performance are not always aligned or synchronous (e.g. Hrasky, 2012; Luo, 2019). Thus, to the best of our knowledge, we are the first to explore and provide insightful evidence on the relationship between national culture and carbon performance. Third, unlike Luo and Tang (2016b), we use an integrated framework of macrolevel institutional factors and firm-level operational factors to assess carbon performance, which is consistent with the prior literature (Pérez-López et al., 2015). Using this approach, we conceptualise and explore the dynamic relationships between national culture, institutions and natural environmental factors on one hand and corporate carbon performance (CP) on the other hand. In particular, we make a theoretical contribution supporting the validity and applicability of the institutional theory. We distinguish soft from hard institutions, and we show how soft institutions contribute, beyond hard institutions such as an emissions trading scheme (ETS), to environmental decision making.
Last but not least, our study provides consistent evidence of the effects of dimensions of national culture on carbon performance while also taking into account geographic and climate variables. This is necessary because carbon emissions constitute a threat to ecological systems and may be the primary trigger that activates a response by management. This issue is not considered by Luo and Tang (2016b) and is still largely ignored in contemporary literature. Our findings suggest the natural environmental factors play a role in shaping managers’ behaviour with regard to the green transition to a carbon-neutral future. To the best of our knowledge, this article is the first to document critical evidence of the influence of natural factors on corporate carbon performance. In addition, our analysis is broader than that of Luo and Tang (2016b). For example, we find more significant interactive relationships between various regulatory factors (e.g. ETS, legal system) and cultural factors that influence corporate carbon decisions. In sum, global warming jeopardises the environment, and the detrimental impacts are exacerbated by excessive anthropogenic GHG emissions, which motivate firms to reengineer their existing carbon management processes to solve the problem. Our evidence of the synergetic effects of cultural, institutional and natural environmental factors on corporate climate decisions provides critical insights for policy and has implications for theory.
Our findings have important implications for a range of stakeholders. For example, because the Paris Agreement replaces the Kyoto Protocol as the main international climate policy instrument, it is essential that public policymakers understand the role of national culture as they develop appropriate implementation mechanisms. In addition, multinational enterprises often operate in multiple societies that are culturally disparate. It is challenging when domestic operations are managed by individuals with dramatically distinct cultural values (Franke et al., 1991). To reconcile separate domestic carbon policies into a single, integrated, corporate-wide carbon strategy, directors must incorporate cultural effects into their creation of a consolidated approach that is compatible with companies’ unique cultural backgrounds. In sum, our findings indicate that heterogeneity in the carbon tendencies and performance of companies may be associated with institutions that are exogenous to them (such as macrolevel climate policies, carbon regulations, sector membership and geographic factors), endogenous capacity (e.g. environmental resources and corporate carbon governance) and managers’ attitude and philosophy regarding environmental and economic optimisation, which are closely related to national culture.
2. Theoretical framework
Culture refers to patterns of beliefs and values that manifest in behaviours, practical norms and various artefacts shared by members of a society (Hofstede, 1980). There is a general consensus that culture is a very complex phenomenon. To enhance theoretical understanding of the relationship among individuals, culture and philosophy, and the natural ecosystem, we draw on a metaphor introduced by Coleman (1994), colloquially known as Coleman’s boat. The basic argument is that national culture influences the philosophy of individuals (e.g. managers, the public, stakeholders) on environmental protection, which affects their choice of a proactive or reactive climate strategy, which will ultimately lead to differences in carbon performance.
Figure 1 shows an adaptation of this framework (Dow, 2017). In the upper left corner (Ac) are cultural differences at the country level, and in the upper right corner (Df) are outcome variables, here carbon performance at the firm level. The fundamental realisation of the Coleman boat metaphor is that most relationships between two aggregate-level variables actually move through a chain of individual actions and events (i.e. the bottom of the boat). Cultural differences (Ac) are related to firm-level outcomes (Df) because they affect individuals’ perceptions of cultural differences (Bi). In turn, these individual-level perceptions of differences (Bi) influence particular courses of action (Ci) via a mediating variable (MedB-C). Mediating factors (e.g. perception and management of carbon risks) have consequences for individuals’ preferences for particular activities (Ci). Finally, when the individual is an executive, these preferences influence firm-level outcomes (Df), or carbon performance in our study. This metaphor illustrates the gaps in theories of how and why cultural differences affect climate decisions and their results. We focus on the underlying mechanisms to conceptualise the influences of culture on carbon decisions (Maseland et al., 2018). As already alluded to, the bottom of Coleman’s boat can be understood from a number of different theoretical perspectives, depending on the particular dimension and circumstances. We suppose that culture can play a role in influencing carbon performance by influencing the underlying incentives and motivations of managers. In the following discussion, we synthesise the theoretical insights of institutional theory and show how differences in national culture can influence firm-level carbon decisions.
Coleman’s boat. From Dow (2017).
The impact of culture on carbon performance can be examined through legitimacy and organisational institutionalism (Kostova and Zaheer, 1999; Scott, 2013). Organisational survival depends on continued support from stakeholders (DiMaggio and Powell, 1983). Legitimacy is thus defined as ‘a generalized perception or assumption that the actions of an entity are desirable, proper, or appropriate within some socially constructed system of norms, values, beliefs and definitions’ (Suchman, 1995: 574). Firms now face challenges to their legitimacy related to the threat of climate change. Firms typically use market-based coping mechanisms, such as changing operations or making asset-specific investments in clean energy, enhancing disclosure, or reducing emissions footprints or taking similar actions to reduce potential carbon liability. Culture can play a role here in a number of ways. The legitimacy problem is largely related to the expectations and preferences of stakeholders. Culture can increase or decrease perceptions of climate legitimacy in relation to the urgency, salience and importance of climate change. Legitimacy costs and issues can be perceived in different ways in different institutional and cultural contexts (Zaheer and Mosakowski, 1997). Thus, culture is regarded as a source of differences in stakeholder attitudes towards climate change (Zaheer, 1995). For example, ignoring climate change may not be perceived as a legitimacy problem in a masculine country, whereas excessive GHG emissions could be viewed as a liability in a community with a humane orientation (HO) in which quality of life and the physical environment are perceived as more important than financial success.
In addition, business practices converge into certain types of models in particular industries. This convergence is called isomorphism, and institutional differences (such as cultural differences) can create isomorphic pressures in multiple institutional contexts (Kostova and Zaheer, 1999; Meyer et al., 2011). In our context, firms tend to engage in isomorphic behaviour, such as adopting the carbon strategies of their peer firms, to overcome a legitimacy gap. Cultural factors can accelerate the isomorphic process. For instance, because a collectivist manager tends to act together with peer colleagues and business partners, isomorphism with regard to a green strategy is more likely to take place in a collectivist country. Similarly, leaders and executives in high-PD societies are able to facilitate green isomorphism by more quickly implementing governmental climate change policies and decarbonisation strategies than managers in low-PD nations.
3. Literature review and hypothesis development
Cultural values affect various firms’ financial decisions (Barnea and Schwartz, 2008; Chui et al., 2002; Grinblatt and Keloharju, 2001; Han et al., 2010; Kanagaretnam et al., 2011; Kwok and Tadesse, 2006; Morosini et al., 1998; Schwartz and Bardi, 2001; Shao et al., 2010). In addition, increasing studies also suggest that CSR is an outcome of managerial discretion and cultural values (Buhr and Freedman, 2001; Husted, 2005; Luo and Tang, 2016b; Van der Laan Smith et al., 2005; Waldman et al., 2006; Williams, 1999).
Despite the contributions of prior literature, research on the influence of culture on CSR in general, and on carbon decisions in particular, is very limited. Due to the inherent uncertainty in climate change, a corporation’s carbon response is likely to be sensitive to national culture (He et al., 2021; Luo et al., 2018; Luo and Tang, 2016b). Matten and Moon (2008) argue that different cultural systems have generated disparate assumptions about society, business and government. Crossland and Hambrick (2011) show that cultural traits have an observable impact on managerial discretion. Thus, we would expect a priori that there would be differences in how firms respond to climate change in societies with distinct cultural values. A priori factor – national culture – can influence the cognitive processes that underlie managers’ judgements and attitudes towards the importance and relevance of climate risks for companies’ operations and performance (Chui and Kwok, 2008; Luo and Tang, 2016b; Peterson et al., 2018; Tung and Stahl, 2018). Nevertheless, different cultural dimensions may have different effects, so we examine the dimensions identified by the GLOBE study. In doing so, we provide the most comprehensive analysis of the effects of culture on CP to date.
3.1. Future orientation
Future orientation (FO) is ‘the degree to which a collectivity encourages and rewards future-oriented behaviours such as planning and delaying gratification’ (House et al., 2004: 282). The propensity to save now to invest in the future is an expression of a high FO, which also incorporates the tendency to value future rewards, as seen in characteristics such as perseverance and thrift. FO in a country is similar to FO in business (Orij, 2010). Those with a low FO focus on immediate results, but those with a high FO stress the creation of long-term strategies to gain strength and achieve more sustainable development (Freeman et al., 2010; Hofstede et al., 2010; Trotman and Bradley, 1981). In the case of a green transition to a low-carbon future, GHG mitigation initiatives involve investments with a relatively long-term payback. This is because climate stabilisation is a time-consuming process (Luo and Tang, 2016b). Therefore, a preference and tendency towards green business policies is expected in countries that put stock in FO. Managers in high-FO countries tend to mitigate the magnitude of damage from global warming, which leads to long-term planning, green investment and high CP. Hence, we formulate a hypothesis as follows:
H1. The FO dimension of national culture is positively associated with corporate CP.
3.2. Uncertainty avoidance
House et al. (2004) refer to uncertainty avoidance (UA) as ‘the extent to which a society, organization, or group relies on social norms, rules, and procedures to alleviate the unpredictability of future events’ (p. 30). This dimension of national culture specifically assesses the extent to which members of a country perceive themselves to be threatened by uncertainty and ambiguity and attempt to avoid them. Company managers in weak-UA countries tend to tolerate uncertainty and risks. By contrast, those raised in strong-UA countries are more sensitive to unstructured or novel situations and attempt to control the future (Offermann and Hellmann, 1997). They are prone to greater anxiety and risk aversion and tend to be more nervous when faced with uncertainty. Thus, a more predictable environment is preferred in high-UA countries (Kanagaretnam et al., 2014). Managers in such countries rely on formalised policies and procedures, take moderate and carefully calculated risks, and strongly resist social or environmental changes (Hofstede et al., 2010).
Scientific evidence indicates that uncontrolled climate change may cause temperature fluctuations, raise the sea level, provoke extreme weather and lead to other adverse outcomes (Stern, 2006). However, the timing, scope and severity of these impacts are very uncertain. Thus, managers’ climate strategy and actions are likely to be linked to the sensitiveness of members of the community to the risks and uncertainty arising from climate change. If managers do not care too much about the risks of climate change, they are not expected to undertake serious policies to mitigate these risks. In contrast, stakeholders in higher UA countries tend to be risk-averse, so they are uncomfortable with the unpredictable consequences of global warming (Riddle, 1992). A proactive corporate climate strategy is more compatible with a national culture characterised by high UA. Such a community would opt for GHG mitigation initiatives that would promise more stable future weather. Thus, another hypothesis is as follows:
H2. The UA dimension of national culture is positively associated with corporate CP.
3.3. Performance orientation
Performance orientation (PO) captures the extent to which an organisation or society encourages and rewards its group members for performance improvements and excellence (House et al., 2004). High-PO countries value high competitiveness and materialism, and people in these countries reward accomplishment and emphasise results. In contrast, individuals in low-PO societies show a preference for societal and family relationships, emphasising greater harmony with the natural environment and valuing the merits of integrity and loyalty over material achievements.
It can be argued that this dimension may have some influence on carbon performance. However, the impact of PO on carbon performance depends on whether the focus of the expected performance is limited to financial results or extends to nonfinancial results (e.g. social or environmental performance). From a firm’s perspective, management always faces a dilemma given limited resources. If resources are used to pursue economic growth, fewer resources will be available for environmental protection, including stabilisation of climate change. If an organisation emphasises economic output only, PO will likely have an inverse correlation with CP. However, if a company has specific, measurable carbon reduction targets and operates in a system with monetary or nonmonetary incentives for employees who achieve superlative CP, this cultural dimension can be expected to have a positive impact. Thus, it is difficult to predict the direction of the association with CP. We propose the following nondirectional hypothesis:
H3. The PO dimension of national culture is significantly associated with CP.
3.4. Gender egalitarianism
Gender egalitarianism (GE) is defined as the extent to which an organisation or a society minimises differences in gender roles and gender discrimination (House et al., 2004). Gender differences are related to the distribution of emotional roles between the genders (e.g. Hofstede, 1980). Countries high in GE tend to reduce gender discrimination, upgrade social status and unlock the potential of female members. Generally speaking, feminine countries are inclined to emphasise life quality more than economic growth (Hofstede et al., 2010). A focus on the quality of life can lead to a preference for environmental protection because the quality of life is closely related to the quality of the living environment (Luo and Tang, 2016b). Accelerating dangerous global warming via excessive GHG emissions will degrade the quality of living conditions (Stern, 2006). Thus, managers of companies located in countries with greater gender equality are more likely to be aware of and cautious about their effects on climate (Van der Laan Smith et al., 2005). It is well recognised that the negative impact of climate change falls more heavily on women than men. This is because women often have access to fewer resources and enjoy less resiliency to the natural disasters caused by global warming (UNFCCC, 2012). Therefore, managers in countries that have high regard for gender equality are encouraged to participate in the green transition. Managers in these countries are able to play a more active role in addressing climate change challenges and have access to more resources and funding, so they can make greater contributions to carbon performance (e.g. Williams, 1999). Thus, we formalise the following hypothesis:
H4. The GE dimension of national culture is positively associated with CP.
3.5. Human orientation
Human orientation (HO) refers to the degree to which organisations and communities encourage and reward individuals for being fair, altruistic, friendly, generous and caring (House et al., 2004). This dimension resembles the dimension of kind-heartedness described by Hofstede and Bond (1988). HO is associated with a tendency to be sensitive to the needs of others, and those with higher degrees of HO are expected to concern themselves with the quality of others’ living conditions. Managers in such countries are less likely to ignore the threat posed by climate change and more likely to act to avoid its harmful effects. Thus, it can be expected that CP will be higher in companies that operate in high-HO countries because stakeholders in such countries have stronger feelings about the suffering of others and their happiness. This orientation will be extended to the conditions and quality of the natural ecosystem in which the given people are living. This means that the general public will be less tolerant of excessive GHG emissions and will pressure the government and corporations for more ambitious carbon reduction targets. Thus, we form the following hypothesis:
H5. The HO dimension of national culture is positively associated with CP.
3.6. Institutional collectivism and in-group collectivism
Institutional collectivism (IC) refers to the degree to which institutional practices of organisations and society discourage and punish the collective distribution of resources and collective action (House et al., 2004). By contrast, in-group collectivism (IGC) is the degree to which individuals express pride, loyalty and cohesiveness in their organisations or families. As IC and IGC are closely linked, we generally consider whether any national tendency towards collectivism versus individualism has any impact on corporate CP.
In individualist countries, there are loose ties among individual members of a group, whereas countries that are high in collectivism tend to be integrated into strong, cohesive in-groups. Thus, it is expected that individualism and collectivism may have opposite effects on carbon performance. On the one hand, it can be argued that less collectivist countries tend to be more transparent, allowing nonprofit and nongovernmental bodies to participate in the development of public climate change policies and pushing the government to adopt climate-friendly measures (Jaggi and Low, 2000). Managers in such collectivist countries are able to develop key initiatives to enhance public awareness of climate change, and firms are under greater pressure from the community to discharge their ecological accountability by improving carbon performance (Husted, 2005). In addition, countries characterised by high levels of individualism typically allow for greater individual initiatives, which may encourage innovation, a key factor in the development of low-carbon technology. Thus, individualism is positively associated with CP.
On the other hand, Buhr and Freedman (2001) contend that more collectivist cultures encourage organisations to adopt more cooperative approaches, and stakeholders may expect decision-making processes to be broader, more participatory and more consultative. Thus, managers who commit to a green transition tend to use cooperative approaches to engage various stakeholder groups to achieve ambitious climate targets. Ringov and Zollo (2007) suggest that environmental interests can be ignored in individualist countries, where managers attempt to prioritise financial benefits. Unless it is consistent with their recognised self-interest, such managers are less motivated to take voluntary actions to reduce their GHG footprint. In sum, the net effects of this dimension are somewhat ambiguous; thus, we propose nondirectional hypotheses.
H6a. The IC dimension of national culture is associated with CP.
H6b. The IGC dimension of national culture is associated with CP.
3.7. Power distance
Power distance (PD) is a cultural variable that has a strong influence on people’s behaviour. It indicates the degree to which members of an organisation or society expect and agree that power should be unequally distributed (House et al., 2004). Institutions in high-PD countries are more stratified socially, economically and politically, so people in positions of authority expect to receive unconditional obedience from low-level members of the community (Javidan et al., 2006). In high-PD cultures, power is concentrated in the hands of a few privileged groups of people, and there is a greater hierarchy between superiors and subordinates. In sharp contrast, low-PD countries encourage a free exchange of ideas, promote informational transparency and do not allow manipulative uses of power, which leads to policy adaptability and less economic and political inertia.
Generally speaking, global warming affects everyone but impacts the poor even more negatively (Luo and Tang, 2016b). Therefore, those with power are likely to be less concerned about the environment. There is a lack of equal opportunities for minorities and professional development in high-PD countries (Carl et al., 2004). Managers are inclined to consolidate their power and feel less responsible for contributing to the welfare of the community. In high-PD countries, carbon policy is less likely to be a topic of open debate, so the public is less likely to be aware of the consequences of climate change, and governmental climate policy is more likely to be deficient. It is difficult to trace the carbon footprint of an organisation in such a country because of the lack of adequate channels for sharing climate information (Luo and Tang, 2016b). Managers who favour the status quo are less likely to set ambitious emissions reduction strategies and targets (Waldman et al., 2006). High-PD countries tend to entrench hypocritical and narcissistic managers and strengthen information asymmetry between insiders and outsiders to protect vested commercial and industrial interests. By contrast, in low-PD countries, the public and various stakeholders are able to protest against companies’ climate policies and actions that are not friendly to the natural environment (Husted, 2005). For this reason, managers are less likely to abuse their power to pursue personal benefit, which encourages them to build long-term relationships with key stakeholders or feel more responsible for the broader social welfare. Thus, the final hypothesis is presented as follows:
H7. The PD dimension of national culture is associated with CP.
4. Research design
4.1. The measurement of carbon performance
Two methods are generally used to measure carbon performance: process-based and outcome-based measurement (Busch and Hoffmann, 2011). Process-based measurement considers corporate commitments or managerial efforts to address underlying climate change. It typically involves carbon management initiatives or carbon abatement projects, which require substantial time and are subject to great uncertainty before outcomes can be observed. In contrast, outcome-based measurement focuses on firms’ actual carbon emissions, which have a direct impact on society and the environment and reflect the efficiency of their production processes. Busch and Hoffmann (2011) contend that process-based measurement neglects concrete resource consumption and emissions optimisation. Therefore, we use an outcome-based measure of corporate carbon performance. Luo (2019) and Luo and Tang (2014b) argue that the intensity of carbon emissions is an observable, objective and quantifiable measure of carbon performance and can assist stakeholders in their decision-making process.
We calculate our measure of CP (CEP1) by applying the negative sign of the natural logarithm of the ratio of total Scope 1 and Scope 2 GHG emissions to total sales (Clarkson et al., 2008; Patten, 2002; Sutantoputra et al., 2012). The use of natural logarithmic transformation corrects for the non-normality of the CP variable (Luo and Tang, 2014b). The negative sign is applied to facilitate interpretation. Higher values for CEP1 suggest that a firm uses its resources, particularly its energy, efficiently and therefore has good carbon performance. In addition, we use the intensity of carbon emissions instead of absolute emissions to control for variation in output to ensure that data are more comparable across different firms and reporting periods (Hoffmann and Busch, 2008).
4.2. Sample selection
Our initial sample includes GHG-related data disclosed by companies responding to the CDP survey over 2008–2014. The CDP is a nongovernmental organisation, headquartered in the United Kingdom, that collects comprehensive climate change information from leading listed firms (more than 5000 per year) worldwide. We choose CDP data because it results from the use of a standardised questionnaire, although the disclosure is voluntary (Luo et al., 2012). The standardisation of the disclosure effectively minimises incomparability and suppresses attempts at greenwashing that discretionarily foreground good news in carbon presentation. The CDP database is relatively consistent, complete and comparable and uses a set of well-recognised methods. The CDP is recognised as having the highest credibility among sustainability ratings and rankings (SustainAbility, 2013), and its data are widely used in studies of corporate decarbonisation (e.g. Chapple et al., 2013; Ioannou et al., 2016; Krishnan, 2003; Luo and Tang, 2014a; Tang and Luo, 2014). Note that the CDP’s multinational data set offers wide variation in the constructs of interest and reflects the global nature of responses to climate change, which strengthens the validity of our investigation. We delete observations that have missing values for financial variables and that are from countries with fewer than 30 observations. The final sample includes 6673 observations from 22 countries and 10 Global Industry Classification Standard sectors (Table 1).
Sample selection process.
CDP: Carbon Disclosure Project.
4.3. Empirical model
We perform ordinary least squares (OLS) regression. To reduce heteroscedasticity, we follow Han et al. (2010) and correct standard errors using the White (1980) procedure 2
Here, CEP1 stands for CP, and FO, UA, PO, GE, HO, IC, IGC, PD, and ASV are the nine cultural dimensions identified by the GLOBE study (House et al., 2004).
4.3.1. Control variables
We control for factors that might influence a firm’s CP. Larger firms are under greater pressure to achieve better CP and possess more resources and capabilities to do so (e.g. Ioannou et al., 2016). We thus control for a firm’s size, measured as the logarithm of total assets (SIZE). The exact effects of leverage (LEV) depend on creditors’ preferences. On the one hand, a debt covenant may restrict funding for CSR projects. On the other hand, creditors may wish the company to go green. We do not predict the sign of the influence but instead allow the empirical results to inform us of the impact of leverage. Furthermore, we use return on assets (ROA) to control for the effects of financial performance (Bewley and Li, 2000; Cormier and Magnan, 1999; Li and McConomy, 1999; Luo et al., 2013), and we use Tobin’s Q (TOBINQ) as a proxy for controlling economic growth (Ioannou et al., 2016). Capital intensity (CAPINT) may also influence CP. Firms with higher capital intensity might set less ambitious targets because reducing carbon might be difficult for them (Ioannou et al., 2016). Newness (NEW) is a proxy for technologies that are expected to improve operational efficiency (e.g. Clarkson et al., 2008). Finally, we control for the possible sector and year effects by including 15 dummy variables, 9 for the sector and 6 for the year (Montiel and Husted, 2009).
4.3.2. Country-level control variables
The nature of the legal system of a country is related to its stakeholder/shareholder orientation and thus is associated with CSR (Simnett et al., 2009). We thus include a dummy variable (LEGAL) that equals 1 if the firm belongs to a common law country and 0 otherwise. ETSs (ETS) are emerging as the primary policy instrument used in response to climate change, and such schemes are a cost-efficient way to reduce emissions. Thus, the existence of an ETS in a country is expected to be associated with corporate decarbonisation in that country’s enterprises. Finally, we use gross domestic product per capita (GDPPC) to control for the economic development of the country, which might be correlated with its green aspirations (Adhikari and Tondkar, 1992; Luo et al., 2013). We also consider the influence of the natural environment, geographic and climate factors in particular, on carbon decisions. It seems intuitive that corporate managers’ decision is likely to be influenced by aspects of the natural environment that are not under managers’ control (Huang et al., 2017). We consider two climate and geographic variables, namely, mean elevation above sea level (in metres) and average annual temperature (in degrees Celsius). ELEV is the country’s mean elevation above sea level (in metres). TEMP is the mean annual temperature of the country (in degrees Celsius). The definitions of these variables are presented in Appendix 1. Financial variables are obtained from DataStream (measured in millions of US dollars). We winsorise all continuous variables at the 1% and 99% levels to reduce the influence of outliers.
5. Empirical results
5.1. Descriptive statistics
Table 2 summarises the descriptive statistics for the variables. The mean CEP1 is −4.352, which implies an average corporate carbon intensity of 76 tonnes per million dollars of sales in our sample firms. Approximately 58% of the firms operate in a common law (i.e. shareholder-oriented) country, and 40.5% are domiciled in a country with an operating national ETS. The mean SIZE is 8.665, which suggests that the average total assets is US$5.796 billion, which is consistent with the fact that the CDP targets the largest companies. The average return on assets (ROA) for the sample is 5.1%, and leverage (LEV) is 25.6% of total assets, which is comparable to prior studies (Clarkson et al., 2008; Luo and Tang, 2016b).
Descriptive statistics (N = 6673).
P25 and P75 = 25th and 75th percentiles, respectively.
Table 3 indicates that the number of firms participating in the CDP increases from 479 in 2008 to 1183 in 2014, which indicates an enhanced awareness of climate change among our sample companies. In addition, year-by-year data demonstrate that the mean CEP1 increases from –4.602 in 2008 to −4.194 in 2014, which suggests that improvements occur in the CP of our sample firms. Table 4 shows that industrials, materials and consumer discretionary have the most observations of all sectors, accounting for 23.36%, 15.05% and 14.91% of the sample, respectively. Utilities, materials and energy have the highest carbon emissions. Table 5 shows that the United States (25.64%), the United Kingdom (13.74%) and Japan (10.97%) have the most observations. Firms in developing countries exhibit less decarbonisation, probably because of their inefficient consumption of energy (Cao, 2003).
Means for variables by year.
Please see Appendix 1 for detailed definitions and sources of variables.
Means for variables by GICS sector.
GICS: Global Industry Classification Standard.
Please see Appendix 1 for detailed definitions and sources of variables.
Means for variables by country.
ISO3 = ISO 3166-1 alpha-3 code. The third column lists the percentage of firms in a particular country out of the full sample. Please see Appendix 1 for detailed definitions and sources of variables.
Table 6 shows that six out of the nine GLOBE cultural dimensions are significantly associated with CEP1 in Pearson correlation analyses. We also find that the cultural variables are significantly correlated with one another: the Pearson and Spearman correlations between UA and FO are 0.72 and 0.53, respectively. Hence, we calculate the variance inflation factor (VIF) for each independent variable in equation (1). Moreover, the pairwise correlations between other firm-level control variables are generally low, which suggests no severe problem with multicollinearity when the variables are included in the regression.
Correlation matrix (N = 6723).
Pearson (Spearman) correlation coefficients are below (above) the diagonal. Financial variables are winsorised at the 1% level and are measured in millions of US dollars. *, ** and *** represent significance at the p < 0.10, p < 0.05 and p < 0.01 levels, respectively (two-tailed). Please see detailed definitions and sources of variables in Appendix 1.
5.2. Multivariate analyses
Before we estimate the coefficients of OLS regression, we check and analyse several statistical assumptions of the OLS regression, such as normality, heteroscedasticity, multicollinearity and autocorrelation. Following the central limit theorem, we assume that a normal distribution does not affect the quality of our results because the sample size is sufficiently large. As a sensitivity test, we perform quantile regression as an alternative model, which does not require a normal distribution. To alleviate heteroscedasticity and autocorrelation, we estimate standard errors using the White (1980) procedure and control for industry and year fixed effects in all models. To check for the existence of multicollinearity between the independent variables, we compute the VIF for each independent variable in the equation as well as the mean VIF. We find that the mean VIFs for all equations are below 3 and the maximum VIF is below the critical threshold of 10, so multicollinearity does not appear to be a concern (Kennedy, 2008).
We report the results for equation (1) in Table 7. For all specifications, we estimate the coefficients using OLS regression with robust standard errors and controlling for industry and year fixed effects. Column (1) contains only control variables, and in Column (2), all nine cultural variables are included. Column (3) includes all variables, including cultural ones. We discuss the differential effects of individual cultural dimensions on carbon performance based on the results of Column (3).
OLS Regression based on GLOBE national cultural factors.
OLS: ordinary least squares; GLOBE: Global Leadership and Organizational Behavior Effectiveness.
An OLS model is used to test the impact of cultural values on carbon performance. t statistics (based on robust standard errors) are in parentheses. *, ** and *** represent significance at the p < 0.10, p < 0.05 and p < 0.01 levels, respectively (two-tailed). Financial variables are winsorised at the 0.01 level and are measured in millions of US dollars. Please see detailed definitions and sources of variables in Appendix 1.
First, the coefficient of FO is positive and significant (β1 = 0.730, p < 0.01), which suggests that CP is better in countries with a tendency towards FO, consistent with H1. The dominating characteristic of this cultural dimension is that it is forward-looking and values long-term outcomes. Forward-looking managers are more likely to expect increased future benefits arising from stability in the climate. Second, the coefficient of UA is positive and significant at the 5% level, as shown in Column (3). These results provide support for H2. Third, PO has a positive and significant link to CP. This result supports H3, which means that PO leads to a focus on nonfinancial outcomes, and this orientation is also a notable stimulus for achieving an ambitious CP target.
Fourth, the estimated regression coefficient of β4 (GE) is positive and significant at the 1% level. This suggests that firms in countries with higher GE have better CP. This result is consistent with H4 and resonates with previous studies showing that firms with gender diversity on their boards are more sensitive to the quality of life and, hence, more willing to protect the integrity of the climate (Liao et al., 2015). Fifth, the effects of HO on CP are reflected in the coefficient of HO (β5), which is positive but not significant (β5 = 0.216, p > 0.1). Sixth, the coefficient of IGC is negative and significant (β6 = −0.938, p < 0.01), but the coefficient of IC is not significant though negative (β7 = −0.049, p > 0.1). This is consistent with H6b but not H6a. The results for PD show that this dimension has significant positive impacts on CP, which is consistent with H7.
The effect of the control variable SIZE on CEP1 is negative and significant, which indicates that larger firms tend to have higher carbon emissions, which is consistent with our expectation. The coefficient of LEV is negative and significant in all models, which suggests that more highly leveraged firms face greater pressure from debtholders to keep from diversifying their funding into green investment. This result can be explained by stakeholder theory, where debtholders form part of a stakeholder group. In addition, the significant positive coefficient of TOBINQ suggests that growing firms in our sample tend to have better CP. The coefficient of CAPINT is negative and significant, as expected. However, NEW is negatively related to performance, which contradicts our expectation and suggests that investment in technology is unrelated to energy efficiency. We find that the coefficient of GDPPC is positive and significant in all models, which suggests that CP is better in countries with a higher gross domestic product per capita. This is consistent with resource availability theory (McWilliams and Siegel, 2001; Waldman et al., 2006; Williams, 1999). The negative coefficient of LEGAL in all models corroborates stakeholder theory (Luo and Tang, 2016a; Simnett et al., 2009) in the sense that management may respond to the request of broad stakeholders, beyond immediate shareholders, with high CP. Contrary to our expected result, we do not find a significant relationship between the presence of a national mandatory ETS and CP. One possible reason for this is that cultural factors may have a greater influence than regulatory institutions on CP. The coefficients of ELEV (elevation) and TEMP (temperature) are all negative and significant, which is consistent with our expectation of a significant influence. Our empirical results suggest that people who live in different climate conditions feel differently about climate change. In particular, people in countries with cold weather are more sensitive to climate change than people living where warm or hot weather predominates. In sum, the results resonate with the claim that culture affects corporate decarbonisation, even when we control for the influences of geographic and climate factors.
5.3. Robustness tests
We also run some sensitivity tests. First, because the United States accounts for a large portion of our sample observations, it may be that our results are biased. To address the uneven distribution of observations across countries, we exclude the United States from the analyses. In addition, because the United States, the United Kingdom and Japan together account for approximately half of the observations, we also test the robustness of our results when observations from these three countries are excluded. The results (not shown) are qualitatively similar to our main inferences.
Second, to test the robustness of our results, we use the alternative measure CEP2 for carbon performance. CEP2 is sector-adjusted CEP1, calculated as firms’ CEP1 minus the mean for their sector in a particular year (Luo and Tang, 2014b). The formula is
where Njk is the total number of firm observations in sector j and year k; j is energy, materials, industrials, consumer discretionary, consumer staples, financials, health care, information technology, telecommunications or utilities; and k is the year 2008, 2009, 2010, 2011, 2012, 2013 or 2014. Higher values for CEP2 represent good CP relative to sector peers. The results (not reported) are consistent with our main inferences.
Third, because our final sample includes only those companies that disclose their carbon emissions through voluntary participation in the CDP (Matsumura et al., 2014), it is subject to self-selection bias. To address this issue, we use Heckman’s (1979) two-stage estimation. Following Luo and Tang (2016b), the first stage involves an assessment of the firm’s decision to publicly disclose emissions information, which we estimate using the following probit model:
DIS equals 1 if a company responds to the CDP by answering the questionnaire and makes its responses publicly available and 0 otherwise. The second stage involves OLS regression and controls for both industry and year effects specified in equation (1). The results (not shown) do not alter our inferences.
Fourth, because the financial sector is subject to different regulations and engages in different operations, we exclude it from our analyses. Fifth, in place of OLS regression, we use a random-effect model for panel data because of the nature of our data. The results (not shown) show that our inferences still hold when using this alternative model specification. To further address some assumptions of OLS regression, we complement the OLS analysis with the use of quantile regression models. Unlike OLS regression, quantile regression does not assume specific parametric distributions, nor does it assume a constant variance for the residuals. Quantile regression provides a more robust estimate of the effect of the cultural values on each of the quantiles of the distribution of CP. Specifically, we estimate the regression with bootstrap standard errors, which retains the assumption of independent errors but relaxes the assumption of identically distributed errors. Except for PO, the coefficients of the cultural variables remain significant with the expected signs. The results show that our main inferences are qualitatively unchanged.
Finally, in our main analysis, we use the average elevation and temperature to capture the most common types of physical risks and threats a country faces through climate change. We acknowledge that these two measures are specifically related to sea rise and weather-related physical risks at the country level and may not fully capture all of the consequences of these environmental factors. To alleviate this issue, we use an alternative measure, namely, the Global Climate Risk Index (GCRI), to proxy for the overall physical risk faced by a country. It is developed and published annually by the NGO Germanwatch (https://germanwatch.org/en/cri) and measures a country’s fatality rate and economic losses due to a wide range of extreme weather events such as storms, floods and heatwaves. This measure was used in prior studies to reflect a country’s physical risks that affect manager decisions (e.g. Huang et al., 2017). Note that this measure reflects a consequence of global climate change that may worsen under certain climate and geographic conditions. The results are reported in Column (4) of Table 7. We find that some national cultural dimensions become more significant than the main analyses. This confirms our main inferences and provides even stronger evidence when we adopt GCRI as a proxy for country-level physical risk.
6. Additional analyses
6.1. The dynamic relationship between culture, carbon performance, and institutional and climate and geographic factors
Cultural factors are unlikely to be the sole influences on managers’ behaviour and business decisions. Thus, our final analyses explore the dynamics of the relationship between culture, corporate carbon activity and national institutional variables on one hand and the climate and geographic variables identified in the previous section on the other hand.
6.1.1. Moderating effects of carbon institutions on the relationship between culture and CP
We first consider the role of carbon institutions in the relationship between culture and CP. An ETS is a governmental policy that establishes a market that allows firms to trade emissions allowances or rights. This mechanism sets a price for carbon, which internalises the cost of emissions into the participating firms. It can be argued that an ETS can amplify or diminish the influence of cultural factors, as it can modify, complement or substitute the effects of culture on CP. Thus, we include nine interaction variables (e.g. FO×ETS denotes the interaction between FO and the ETS dummy) one by one in Columns (1)–(9) of Table 8. The coefficient of FO×ETS in Column (1) is not significant, which suggests that FO has an impact regardless of the presence of an ETS. In Columns (2)–(9), the coefficients of the remaining interaction variables are all significant, which implies that the degree and direction of the effects of these cultural dimensions on CP depend on the existence of an ETS. For example, in Columns (2) and (4), the coefficients of UA×ETS and GE×ETS are negative, which suggests that the effects of UA and GE are more pronounced in countries without an ETS. This is perhaps because without an ETS, there is an institutional void; in such a case, cultural tendencies, such as UA and GE, appear to fill the void (Leung et al., 2005). Column (3) demonstrates that the coefficient of PO×ETS is positive and significant, which suggests that an ETS gives performance-oriented managers an additional incentive to improve CP. Similarly, the positive coefficient of PD×ETS in Column (8) indicates that an ETS leads directors in high-PD countries to feel a stronger obligation to stakeholders (Ioannou and Serafeim, 2012). In sum, these findings suggest that an ETS is a moderating influence on the relationship between cultural factors and CP.
Moderating effects of an emissions trading scheme.
An ordinary least squares model is used to test the moderating effects of an emissions trading scheme on the relationship between cultural values and carbon performance. t statistics (based on robust standard errors) are in parentheses. CULTURE represents a specific dimension of national culture as indicated in the column. For example, CULTURE*ETS in column (1) represents the interaction item between cultural dimension of future orientation and emission trading scheme (FO*ETS). *, ** and *** represent significance at the p < 0.10, p < 0.05 and p < 0.01 levels, respectively (two-tailed). Financial variables are winsorised at the 0.01 level and are measured in millions of US dollars. Please see detailed definitions and sources of variables in Appendix 1.
In addition, few countries in our sample adopt a national ETS, with the EU ETS the most prominent one. However, the EU ETS during our sample period is characterised by its low emissions price. We thus replace the ETS with the OECD Environmental Policy Stringency Index (EPS). EPS measures the degree to which environmental policies impose a price or cost on polluting or other environmentally harmful behaviours (Choi and Luo, 2021). The results are presented in Columns (1)–(9) of Table 9. The sample size decreases from 6673 to 6469, as data for EPS are only available for OECD countries. The interactions between most of our focused cultural variables and EPS are significant, and the signs are consistent with those reported in Table 8. Although some interaction variables (PO×EPS, GE×EPS and HO×EPS) are nonsignificant, this does not affect our main inference that environmental/carbon policies play a significant role in determining the magnitude of cultural values on corporate carbon performance. It is important to note that how and to what extent different cultural values/dimensions can influence firms’ managers differ quite a lot.
Moderating effects of the stringency of environmental policies.
An ordinary least squares model is used to test the moderating effects of the stringency of environmental policies on the relationship between cultural values and carbon performance. t statistics (based on robust standard errors) are in parentheses. CULTURE represents a specific dimension of national culture as indicated in the column. For example, CULTURE*EPS in column (1) represents the interaction item between cultural dimension of future orientation and the stringency of environmental policies (FO*EPS). *, ** and *** represent significance at the p < 0.10, p < 0.05 and p < 0.01 levels, respectively (two-tailed). Financial variables are winsorised at the 0.01 level and are measured in millions of US dollars. Please see detailed definitions and sources of variables in Appendix 1.
6.1.2. Moderating effects of culture on the relationship between the legal system and CP
There is a general consensus (Hope, 2003; Jaggi and Low, 2000; Luo, 2019; Luo and Tang, 2016a; Zhou et al., 2016) that firms in code law countries are under greater pressure from stakeholders to reduce carbon emissions than firms in common law countries, which are generally oriented towards the interest of shareholders (Luo, 2019). Thus, we ask whether culture can moderate the influence of the national legal system on CP. To answer this question, we create nine interaction variables (i.e. FO×LEGAL, UA×LEGAL) in Columns (1)–(9) of Table 10. The coefficients of eight of the nine interaction variables are significant at least at the 5% level. These results suggest that cultural factors have moderating effects on the relationship between the legal system (a proxy for shareholder vs stakeholder orientation) and CP. In other words, culture strengthens or weakens the impact of the legal system on carbon activities in one way or another. For example, Columns (1)–(3) show that the coefficients of FO×LEGAL, UA×LEGAL and HO×LEGAL are negative. This means that the effects of the legal system on CP are more pronounced in low-FO, low-UA and low-HO countries.
Moderating effects of legal system.
An ordinary least squares model is used to test the moderating effects of legal system on the relationship between cultural values and carbon performance. t statistics (based on robust standard errors) are in parentheses. CULTURE represents a specific dimension of national culture as indicated in the column. For example, CULTURE*LEGAL in column (1) represents the interaction item between cultural dimension of future orientation and the legal system (FO*LEGAL). *, ** and *** represent significance at the p < 0.10, p < 0.05 and p < 0.01 levels, respectively (two-tailed). Financial variables are winsorised at the 0.01 level and are measured in millions of US dollars. Please see detailed definitions and sources of variables in Appendix 1.
6.1.3. Moderating effects of geographic factors on the relationship between culture and CP
Many differences in culture and climate change policy between different nations can be traced to historical and geographic factors. Thus, we examine the impact of culture on CP in a geographic context. Column (1) of Table 11 shows that six of the nine interaction variables created from the nine cultural variables have negative and significant relationships with elevation. Only three have nonsignificant relationships, which means that the influence of those aspects of culture is indifferent to natural conditions. This means that the influence of culture on climate change policy is generally stronger in low-elevation countries because people living at low elevations are more vulnerable to climate change and rising sea levels, so their natural systems and socio-cultural tendencies are more likely to accelerate the speed of decarbonisation.
Moderating effects of geographic factors on the relationship between national culture and carbon performance.
This table reports coefficients of interactions between geographic factors and cultural variables. A robust ordinary least squares regression model is used to obtain these coefficients by introducing interaction variables one by one into equation (1). *, ** and *** represent significance at the p < 0.10, p < 0.05 and p < 0.01 levels, respectively (two-tailed). Financial variables are winsorised at the 0.01 level and are measured in millions of US dollars. Please see detailed definitions and sources of variables in Appendix 1.
Column (2) of Table 11 reports whether the impact of the average temperature on CP is conditional on various cultural factors. The coefficients of almost all interaction variables are significant, which indicates that culture has a significant moderating effect on the relationship between geographic factors and decarbonisation. Four interactions are positive and significant, four are negative and significant, and one is not significant. This confirms our prediction of interactive effects. Although it is difficult to conceptualise interactive effects, our study provides strong and consistent evidence of the existence of dynamic relationships of culture, along with natural factors and social and economic systems, on corporate responses to climate change. These results are consistent with those of Buhr and Freedman (2001), who find that geographic factors can lead to heterogeneity among countries in their political systems, legal environments, and business climates and in national features of the accounting profession. We show that this effect can be influenced by aspects of national culture.
6.1.4. Time effects
The literature suggests that culture does not have a monotonous influence over time. We thus divide the whole sample period into two sub-periods: the Kyoto Protocol period (2008–2012) and the post-Kyoto period (2013 and 2014). The regression results are reported in Table 12. We find that the coefficients of FO, GE and IGC are consistently significant with the expected sign, which suggests that these cultural factors play a role in affecting managers’ carbon decisions in both periods. However, UA and PO only significantly affect decarbonisation in the Kyoto Protocol period, which shows that these two factors may be effective at improving performance through their influence on regulatory and institutional reform, and this effect can change over time.
Time effects.
An ordinary least squares model is used to test the impact of cultural values on carbon performance. t statistics (based on robust standard errors) are in parentheses. *, ** and *** represent significance at the p < 0.10, p < 0.05 and p < 0.01 levels, respectively (two-tailed). Financial variables are winsorised at the 0.01 level and are measured in millions of US dollars. Please see detailed definitions and sources of variables in Appendix 1.
7. Conclusion
In the study, we explicate the core links among culture, CP, carbon institutions, and climate and geographic conditions. Culture is ‘the integrated, complex set of interrelated and potentially interactive patterns characteristic of a group of people’ (Lytle et al., 1995: 170). We adopt a more dynamic view of culture to explore its situation-dependent and dynamic nature (Tung and Stahl, 2018). Instead of adopting a fragmented, piecemeal approach (Tung and Stahl, 2018), we bring a multidisciplinary and holistic approach to bear and consider simultaneously all of the dimensions of culture identified in the literature in their financial, legal and geographic contexts (Tung and Stahl, 2018). Thus, we produce the most comprehensive picture, as well as the most nuanced insight, so far of the effects of culture on corporate CP.
We conclude that carbon activities are framed in relation to a social and geographic context and thus are influenced by the prevailing climate conditions in such contexts (Jackson and Apostolakou, 2010). We explore variation in the CP of large firms worldwide and find evidence that differences in cultural tendency can amplify or mitigate corporate responses to the challenge posed by climate change. Overall, we find that firms in countries with an affirmative FO outlook show better CP, individualism is positively correlated with superior CP and a relative cultural tendency towards PO enhances performance. However, masculinity or higher inequality of gender roles acts as a barrier, with an inhibitory or preventive effect on CP. Although HO provides strong proactive motivation, countervailing incentives exist in countries with the opposite cultural tendency. Finally, potential friction from differences in outlooks towards power and authority, uncertainty, and even quality of life or cooperation, as embodied in more feminine societies, can have implications for the development of carbon strategy promoting more sustainable corporate operations low-carbon economy. In sum, the greater cultural difference between the actors involved in a business decision typically translates into greater information asymmetries, higher legitimacy pressure, greater perceptions of carbon risk and more severe isomorphism pressure. Certain cultural preferences may motivate managers to beaver away on energy-saving technology and enable them to think positively: mitigation and adaptation measures elicit innovation and long-term resilience and sustainability, whereas the beneficial effects of using fossil fuel may be transitory in dynamic environments. Averting climate change will involve a short-term financial cost, but shifting away from carbon may eventually enrich the economy.
Our central conclusion is that a corporate decision to respond to climate change cannot be attributed to a single factor but rather is influenced by multiple contingent variables. Most importantly, we present evidence that underscores the importance of geographic and climate factors for corporate response modes in today’s radically changing, post-Kyoto legislative environment is striving towards a carbon-neutral economy. Our message is relatively simple and intuitive: Climate change must be managed by people, and people are influenced by the cultural values of the nation where they live. The institutional void in climate change response is created by the sort of cataclysm exemplified by extreme weather, characterised by inadequate regulatory systems and international arrangements. It is also associated with economic and political inertia and institutional isomorphism. In such circumstances, aspects of cultural heritage are likely to be the most immediate tools from which the country’s members can draw. Explicit institutions do not determine what will happen in a country in very predictable ways but only interact with cultural heritage in complex ways to shape the business context. Our results show that culture plays a role in creating the conditions in which firms can compensate for institutional voids. This trade-off is largely overlooked in the literature. Our findings should support implementing the recent international climate accord (Paris Agreement of 2015) in societies with distinct cultural preferences.
One of the limitations of this study is that we rely on CDP-reported emissions data to measure carbon performance. Carbon emissions are voluntarily disclosed via an annual survey. For those companies that do not participate in the CDP, we are not able to get emissions data. Although we use Heckman’s two-stage estimation to address sample selection issues, caution should be exercised when generalising our results to non-CDP companies. Finally, our study focuses on the direct and moderating role of culture on corporate carbon decisions. We believe there are plenty of opportunities for future research in this area. For example, it can be argued that policy decisions in a given country could be influenced by its culture, so the social values of the country are reflected in voting and political decisions on climate legislation. In addition, culture may have mediating effects on corporate carbon performance. For instance, firms in different cultures may interpret an international accord such as the Paris Agreement in different ways. This means that the rules and agreement may affect corporate carbon performance via culture as a mediating mechanism.
Footnotes
Appendix
Summary of variables.
| Variable | Definition | Source |
|---|---|---|
| Dependent variables | ||
| CEP1 | Negative sign of the natural logarithm of the ratio of total Scope 1 and Scope 2 greenhouse gas emissions to total sales | CDP |
| CEP2 | Sector-adjusted CEP1, calculated as the firm’s CEP1 minus the mean for its sector in a particular year | |
| Variables of interest from the GLOBE National Culture Study | ||
| FO | Future orientation society value | House et al. (2004) |
| UA | Uncertainty avoidance society value | |
| PO | Performance orientation society value | |
| GE | Gender egalitarianism society value | |
| HO | Humane orientation society value | |
| IC | Institutional collectivism society value | |
| IGC | In-group collectivism society value | |
| PD | Power distance society value | |
| ASV | Assertiveness society value | |
| Control variables | ||
| Geographic variables | ||
| TEMP | Mean annual temperature in the country (in degrees Celsius) | https://www.pdx.edu/econ/country-geography-data |
| ELEV | Mean elevation above sea level (in metres) | |
| GCRI | Global Climate Risk Index | Germanwatch (https://germanwatch.org/en) |
| Firm-level variables | ||
| SIZE | The natural logarithm of total assets | DataStream |
| LEV | Total debt (measured as short-term debt, current portion of long-term debt, and long-term debt) divided by total assets | DataStream |
| ROA | Net income before extraordinary items/preferred dividends divided by total assets | DataStream |
| TOBINQ | Total market value based on year-end price and the number of shares outstanding, plus preferred shares, the book value of long-term debt, and current liabilities, divided by the book value of total assets | DataStream |
| CAPINT | The ratio of gross property, plant and equipment to total assets | DataStream |
| NEW | The ratio of net property, plant and equipment to gross property, plant and equipment | DataStream |
| Country-level variables | ||
| ETS | An indicator variable that equals 1 if the firm is located in a country that has an operating national emissions trading scheme and 0 otherwise | Website |
| LEGAL | An indicator variable that equals 1 if the firm is located in a common law country and 0 otherwise | La Porta et al. (1998) |
| GDPPC | The natural logarithm of the gross domestic product per capita | The World Bank |
Acknowledgements
We thank the attendees at the 1st International Symposium on Carbon Accounting and Finance held on 29 and 30 November 2017, in Wuhan, China, the participants in research seminars at Western Sydney University and Nanjing University of Finance and Economics and the anonymous reviewers and participants at the Annual Conference of the Accounting and Finance Association of Australia and New Zealand 2018, held in Auckland, for their useful suggestions and comments.
Final transcript accepted 22 July 2021 by Mark Wilson (AE Accounting and Auditing).
Funding
The author(s) disclosed receipt of the following financial support for the research, authorship and/or publication of this article: L.L. also acknowledges the financial support of the Faculty of Business and Law at the University of Newcastle. Q.T. also acknowledges the financial support of the School of Business at Western Sydney University.
