The effect of home country national culture on companies’
corporate social performance
Abstract
This study investigates the cultural drivers of Corporate Social Performance (CSP) by using Hofstede’s four cultural dimensions: masculinity, power distance, uncertainty avoidance and individualism. Using a sample of global companies from 34 countries, an annual composite CSP index is constructed based on environmental, social and corporate governance metrics. The findings suggest that home country individualism has a significant and positive
association with CSP. Additionally, one of the studied moderators, national competitiveness, was found to be significant and weaken the relationship between home country power distance and CSP.
Aurelia Grajdieru 10395954 29-06-2015 2014/ 2015 Supervisor: D. Waeger
Statement of Originality
This document is written by Student Aurelia Grajdieru who declares to take full responsibility for the contents of this document.
I declare that the text and the work presented in this document is original and that no sources other than those mentioned in the text and its references have been used in creating it.
The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.
1. Introduction
Nowadays, with the onset of globalization increasingly more companies have started to expand beyond their home countries and along with this expansion, their corporate citizenship has transcended the borders as well. Pressured by multiple stakeholders, firms have begun to realize that economic profitability does not suffice and they have to act on environmental and social matters as well. This issue is especially relevant since most multinational enterprises from highly industrialized countries export their operations and production to developing countries, which are willing to tolerate poor social and
environmental performance because of their desperate need for capital investments (Husted, 2005). Therefore, corporate social responsibility (CSR) has to be a priority on multinationals’ agenda, considering the immense impact that they could potentially have on sustainability, especially since the private sector holds much larger resources than government programs (Husted, 2005). Unfortunately, despite the criticalness of the issue few companies perform responsibly. According to the 2013 United Nations Global Compact Accenture CEO Study, only 33% of CEOs believe that business is doing enough to address global sustainability challenges. Despite all the controversy surrounding the issue of corporate social performance, academics have reached little consensus regarding what CSR is exactly and how it can be defined and measured. One specific issue within this debate is that the very concept of CSR is context-specific; with national culture influencing the expectations society has from
businesses.
Despite the obvious complexity of CSR little attention has been paid to identifying its causes, the main focus remaining on CSR as an antecedent for measures such as financial performance or consumer behavior and satisfaction. A recent study conducted by Ioannou and Serafeim (2012) revealed that of the total variance in firms’ CSR engagement 55% is explained by firm effects, 10% by industry effects and the remaining 35% of the explainable variance is attributed to nation-level institutions. These nation-level effects can be further divided into formal institutions (financial, political and legal institutions), which have been considerably more researched than the informal institutions (local culture and customs). What is more, there are inconsistent findings among the studies that analyzed the effects of culture on CSR. Therefore, this paper will further explore the relationships between national culture and corporate social performance by posing the following central research question: To what extent does home country culture impact a firm’s corporate social performance (CSP) and how is this relationship moderated by the type of industry a firm belongs to and by the degree
of national competitiveness?In order to answer this question, an empirical analysis is conducted using CSP data on the Global Fortune 500 companies.
The remainder of the paper is structured as follows. First previous literature on the topic of CSP is reviewed, followed by a conceptual framework in which the development of the hypotheses is presented. Subsequently, the research design and methodology are
explained, after which the results of the analysis are described. Finally, the paper concludes with a discussion and conclusion in which the findings are explained in light of past research and limitations and avenues for future research are outlined.
2. Literature review
Although previous research has attempted to empirically and theoretically operationalize CSP for several decades, little consensus has been reached regarding its definition, scope and nature (Ho, Wang & Vitell, 2011). For the most part, academics have made self-standing conceptual developments about the topic but failed to systematically integrate these, which resulted in diverging ideas (Wood, 1991). The first study, which attempted to combine multiple conflicting perspectives on CSP, was conducted by Caroll (1979), who integrated CSR along three dimensions: CSR, corporate social responsiveness and social issues. According to Caroll the social responsibility of a business “embodies the economic, legal, ethical and discretionary expectations that society has of organizations at a given point in time” (1979). His model was further extended and specified by Wartick and Cochran (1985), who discussed the principles of CSR, the process of social responsiveness as well as the policies management can use in order to solve social issues. Although the literature diverges on the approach and definition of CSP, Wood has proposed the most integrative and widely accepted definition which will be used in the present paper: “a business organization’s configuration of principles of social responsibility, processes of social responsiveness, and policies, programs, and observable outcomes as they relate to the firm’s social relationships” (1991). Moreover, most academics agree that CSP is essential to the good functioning of businesses, which therefore have to understand how to uncover and address social issues. Nonetheless, researchers have failed to identify precisely which social issues fall within the scope of CSP (Ho et al., 2011). In the present paper, I follow the study of Ioannou and Serafeim (2012) and consider the issues related to: environmental, social and corporate governance (ESG).
Aside from theoretically defining CSP, academics have also explored a wide range of effects and relationships that CSP has on a variety of variables relevant to firms. The most
prominent relationship in the stream of research is between CSP and the financial performance of companies. Specifically, CSP and financial performance appear to be positively associated. This was also the finding of Orlitzky, Schmidt and Rynes (2003) in their comprehensive meta-analysis that integrated 30 years of research and was conducted across different industries and study contexts. Additionally, Cheng, Ioannou and Serafeim show that superior CSR performance provides firms with better access to financing and lower capital constraints, because of increased transparency and stakeholder engagement (2014). Another type of effect studied extensively is between CSR and indirect performance measures such as: consumer attitudes towards a firm’s CSR, as well as their purchase behavior (Sen & Bhattacharya, 2001), consumer satisfaction (Luo & Bhattacharya, 2006), stakeholder perceptions and awareness of CSR (Sen, Bhattacharya & Korschun, 2006) and brand reputation (Smith, Smith & Wang, 2010). Specifically it was found that CSR activity positively influences the attitudes and identification of stakeholders with the firm, customer’s intent to purchase and their satisfaction.
It is obvious that CSP has an influence on key determinants of firm performance and business success. Therefore, it is of utmost importance to understand what in turn causes CSP and its variations. Another stream of research has done precisely this, analyzing CSP as a dependent variable and focusing to uncover the factors that determine variations in CSP across industries, countries and regions. For example, according to Chapple and Moon (2005) CSR activities that multinational companies perform depend on the specific local and
national environments in which they operate. Consistent with the idea that national factors could explain CSR dynamics is Maignan and Ralston’s study conducted on firms based on two continents, which revealed systematic and significant variations in stakeholder pressures and management incentives to act responsibly (2002). In an earlier study Maignan found that consumers from France and Germany are more likely to support responsible organizations than their counterparts in the US (2001). This disparity could be explained by diverging perspectives on stakeholder relationships (Jurgens, Berthon, Papania & Shabbir, 2010). While the US stakeholder model is centered on the maximization of shareholder value, the European model concentrates instead on meeting the needs of important stakeholder groups. Similar findings were presented by Aguilera, Rupp, Williams and Ganapathi, who build up on this idea by arguing that in the Anglo-American model CSR initiatives are taken only if they bring short-term benefits, while the European model embraces long-term strategies (2007).
Another significant explanation of cross-country CSP differences was provided by Ioannou and Serafeim, who found the political, labor, educational and cultural systems to be the most relevant types of home country institutions that affect CSP (2012). Specifically, their results show the importance of the following factors: for the political system,
competition and regulation, level of corruption, insider self-dealing and the type of political ideology. For the education and labour system, availability of skilled labour and union density. And for the cultural system individualism and power distance. Additionally, they found no significant effect for the financial system. Although, the link between culture and CSP has already been investigated in prior literature, little consistency and completeness can be observed across the existing studies, with only two papers including in their analysis all four dimensions of national culture: individualism, power distance, masculinity and uncertainty avoidance (Ringov & Zollo, 2007; Ho et al., 2011). The remaining considered papers incorporate only two cultural dimensions: individualism and power distance
(Waldman et al., 2006; Ioannou & Serafeim, 2012). These empirical studies have resulted in conflicting findings, ranging from negative to positive relationships. Additionally, different methods for measuring culture were observed: some rely on Geert Hofstede’s national culture studies (Ringov & Zollo, 2007; Ho et al., 2011; Ioannou & Serafeim, 2012) while others employ the Global Leadership and Organizational Behavior Effectiveness (GLOBE) research program (Waldman et al., 2006). Despite the fact that the two models share apparent
similarities, as Shi and Wang reported, the distinctive research designs that they employ can cause different results when applied to the same field, in this case CSR (2011). First, the GLOBE studies operationalize national culture using two indicators: values based and practices based, and find a negative relationship between these two groups. This contradicts the main assumption of Hofstede according to which cultural values and practices are positively associated (Javidan, House, Dorfman, Hanges & Luque, 2006). Second, while Hofstede’s research includes only one company (IBM) from one industry, the GLOBE studies cover multiple firms in three different industries (telecommunications, financial services and food processing) and despite the fact that both studies conduct surveys at the individual level, they diverge significantly in how they analyze and aggregate the scales to the societal level (Peterson & Castro, 2006). This final difference has made academics question the empirical and theoretical implications of such multilevel studies (Tang & Koveos, 2008). Considering the value assigned to Hofstede’s framework over time and the limitations of the GLOBE scores in the present study the focus will be on Hofstede’s cultural
measurements. Because the latter were more extensively used this would also allow for a more thorough comparison with previous literature on CSP.
3. Theoretical framework
As mentioned in the literature review, the main purpose of this paper is to evaluate the reason why companies that are headquartered in different countries, but are nonetheless comparable in size, financial performance and active in the same industry, show significant variations in CSP ratings. For example, French automakers Peugeot and Renault have much higher CSP scores than their South Korean competitors Kia Motors and Hyundai Motors. In this study it will be argued that these variations in CSP between countries such as France and South Korea are due to differences in cultural systems.
National culture represents one of the core determinants of the differences between countries. As defined by Hofstede “culture is the collective programming of the human mind that distinguishes the members of one group from those of another (…), it is a system of collectively held values ” (1994). Although the concept of culture has existed for thousands of years, it has become increasingly important for companies, with the onset of globalization, when the facilitation of international mobility and free trade has emphasized the differences in mindsets, value systems, customs, beliefs, behaviors and traditions. As Hofstede has shown through three research projects, two conducted with samples of students from 10 and respectively 23 countries and the other one in the subsidiaries of IBM a multinational company, these differences hold true not only for individuals but for organizations as well (1994). What is more, previous research has indicated that culture has a clear impact on personal ethics as well as on business ethics. Specifically, relevant aspects such as ethical attitudes, sensitivity, perceptions and judgments seem to be strongly susceptible to cultural variations (Ho et al., 2011). Therefore, in the following subsections we continue this line of thought by developing a set of hypotheses that link Hofstede’s cultural dimensions to CSP.
3.1 Masculinity
The masculinity-femininity dimension reflects the distribution of roles between genders and the way gender-related beliefs direct the society towards values such as assertiveness and material success as compared to values focused on social relationships, such as caring and quality of life (Hofstede, 1994). In a feminine culture (such as Sweden, Norway, the
Netherlands, Denmark) there is little competitive behavior between genders (Ho et al., 2011). On the contrary, societies that are masculine (such as Japan, China, Austria) are built around
competitiveness and greed and display low willingness to engage in cooperative behaviors, which in turn leads to weaker ethics (Vitell & Festervand, 1987). This is also supported by Hofstede, who argues that a strong focus on competition, personal financial gains and economic growth hinders any display of corporate social behavior (2001). Thus, it is hypothesized that there is a negative relationship between home country masculinity and CSP.
H1: Firms based in more masculine countries will score lower on the CSP index.
3.2 Uncertainty avoidance
Uncertainty avoidance, as defined by Hofstede, represents “the extent to which members of a culture feel threatened by uncertainty or unknown situations“ (1994). It describes the degree to which a society is ready to accept risk and ambiguity and suggests how it will react in unstructured situations (Husted, 2005). Hofstede found that cultures that score low on uncertainty avoidance (such as Sweden, Switzerland, Singapore, UK, US) encourage differences of opinions, show preference for few rules and loose laws and thus are more prone to risky behaviors (1984), which has been established to be an antecedent for unethical behavior (Rallapalli, Vitell, Wiebe & Barnes, 1994). On the other hand, high uncertainty avoidance societies (such as Russia, Turkey, Spain, Mexico) value structured environments, clear hierarchies, rules, formal institutions and social norms, which allow them to minimize and prevent uncertainty (Ho et al., 2011). Considering that CSR engagement could help companies establish sustainable relationships with their key stakeholders, it may constitute an effective business strategy for reducing future environmental uncertainty (Peng, Dashdeleg & Chih, 2014). Thus, it is expected that home country uncertainty avoidance will be positively related to CSP.
H2: Firms based in countries with higher uncertainty avoidance will score higher on the CSP index.
3.3 Power distance
Power distance, as defined by Hofstede, represents “the extent to which the less powerful members of institutions and organizations within a country expect and accept that power is distributed unequally“ (1994). This indicates that a country’s inequality level is approved and supported not only by leaders but by their followers as well. In countries with low power
distance (such as Austria, Denmark) managers have less power in decision-making and are only seen as facilitators in the process, therefore, they face higher normative constraints (Ioannou & Serafeim, 2012). On the contrary, in countries with high power-distance (such as China, Malaysia, Russia) a significant dependence of subordinates on their superiors is witnessed. The latter are perceived as “well-meaning autocrats“ who offer favors to their subordinates in exchange for their devotion (Hofstede, 2001). In addition, such managers may be inclined to pursue their individual interests and benefits rather than invest time in establishing meaningful relationships with key stakeholders and catering for the social welfare of the society at large (Ioannou & Serafeim, 2012). The respect and trust that people in high power distance societies associate with their superiors leads to decreased initiatives to debate and to question decisions (Husted, 2005). This means that people in high power distance cultures accept unethical business conducts easier and even go as far as start viewing them as being ethical, while the managers that endorse these conducts are held accountable less than they would in a low power distance society (Cohen, Pant & Sharp, 1996).
Therefore, taking into account the prior empirical evidence described above, a negative relationship between power distance in the home country and CSP is expected.
H3: Firms based in countries with higher power distance will score lower on the CSP index.
3.4 Individualism
The individualism-collectivism dimension reveals the extent to which people are integrated into groups and the degree to which decisions are taken by individuals or by groups (Husted, 2005). Thus, societies that score low on individualism (such as Colombia, South Korea, Taiwan) are considered collectivistic in the sense that their members are integrated into unified and strong groups and they place higher value on the welfare and interest of the group rather than on their personal self. Accordingly, the decision-making process in such societies is expected to be more broad, consultative and participatory (Ioannou & Serafeim, 2012). On the other hand, as Husted observed, in highly individualistic cultures (such as Australia, Canada, UK) the bonds between people are loose, with less formally pre-arranged groups and more voluntary association (2005). Everybody is expected to endorse individual initiative and is more tolerant of unilateral decision-making (Ioannou & Serafeim, 2012). What is more, members of such cultures assign greater significance to personal achievement, interest and freedom as opposed to collective achievement and collective effort (Ho et al., 2011).
Translating these findings into the CSP field, Akaah (1990) discovered that organizations based in individualistic cultures display lower levels of ethics than those in collectivistic societies. Consequently, it is predicted that there is a negative association between individualism in the home country and CSP.
H4: Firms based in more individualistic countries will score lower on the CSP index.
3.5 National competitiveness
National competitiveness, as defined by the World Economic Forum in its annual Global Competitiveness Report (2013), represents “the set of institutions, policies, and factors that determine the level of productivity of a country“. Productivity level determines the level of prosperity and income that an economy can reach, as well as the rates of return on the
investments made in that country, which are in turn directly related to the growth rates of the economy. Therefore, the more competitive an economy is, the higher its potential to achieve rapid future growth. It has been widely acknowledged that the economic development of nations has a direct effect on CSR (Husted, 2005). Beyond that, Hofstede argues that economic development also has an effect on culture and thus interacts with his four cultural dimensions in multiple ways (2001). First, wealthy countries generally score higher on variables such as individualism and lower on dimensions such as power distance. Second, the high level of development has allowed the majority of the population of wealthier countries to transcend the basic needs for survival, security, education and healthcare. Since these countries are very resourceful, they have a real ability and possibility to invest in CSR. The issue whether they choose to do so and whether companies based in these highly developed countries choose to have CSR on their agenda depends partly on their cultural values. For instance, according to Husted, power distance and individualism in particular affect political pluralism and public debate (2005). Thus, in rich countries characterized by strong debate and pluralism it is much likely for social and environmental issues to arise and be treated with priority, since the more basic needs and issues have already been solved. On the other hand, wealthy societies with high collectivism and power distance are less likely to commit to CSR issues, considering that pluralism and the potential for debate are discouraged by
intrinsic cultural characteristics. Correspondingly, less economically developed countries that have yet to deal with issues such as poverty, irrespective of the presence of pluralism and debate, are naturally forced to concentrate on more basic needs (Vogel & Kun, 1987). Therefore, it is expected that home country national competitiveness will weaken the
expected negative relationship between power distance, individualism and masculinity, while strengthening the expected positive relationship between uncertainty avoidance and CSP.
H5a: National competitiveness moderates the negative relationship between home country masculinity and CSP in such a way that the relationship is weaker for higher levels of national competitiveness.
H5b: National competitiveness moderates the positive relationship between home country uncertainty avoidance and CSP in such a way that the relationship is stronger for higher levels of national competitiveness.
H5c: National competitiveness moderates the negative relationship between home country power distance and CSP in such a way that the relationship is weaker for higher levels of national competitiveness.
H5d: National competitiveness moderates the negative relationship between home country individualism and CSP in such a way that the relationship is weaker for higher levels of national competitiveness.
3.6 Industry type
The level of pollution and environmental harm varies across different types of industries, with the chemical, utilities and manufacturing industries being consistently rated as the most “dirty“, while industries such as consulting are rated as “clean“ (Banerjee, 2003).
Considering that the degree of social and environmental risk varies with the type of industry, it will be considerably stronger in the case of dirty industries than in that of clean ones. A specific industry that has been at the forefront of social and environmental scandals in the past decade is the oil and gas industry. Recent disasters such as the Deepwater Horizon spill in the Gulf of Mexico or the Exxon Valdez oil spill have brought to our attention the
magnitude of the risks that oil and gas production assumes. These risks include health and safety risks, liability risks and environmental risks. Therefore, companies in this industry impose numerous and great costs on the society, such as personnel injuries and deaths, oil spills, air pollution, which leads us to hypothesize that membership in the Oil and Gas industry will strengthen the predicted negative relationships between home country power
distance, individualism, masculinity and CSP, while weakening the predicted positive relationships between home country uncertainty avoidance and CSP.
H6a: Membership in the Oil and Gas Industry moderates the negative relationship between home country masculinity and CSP in such a way that the relationship is stronger for companies of the Oil and Gas Industry.
H6b: Membership in the Oil and Gas Industry moderates the positive relationship between home country uncertainty avoidance and CSP in such a way that the relationship is weaker for companies of the Oil and Gas Industry.
H6c: Membership in the Oil and Gas Industry moderates the negative relationship between home country power distance and CSP in such a way that the relationship is stronger for companies of the Oil and Gas Industry.
H6d: Membership in the Oil and Gas Industry moderates the positive relationship between home country individualism and CSP in such a way that the relationship is stronger for companies of the Oil and Gas Industry
4. Research design and Methodology 4.1 Sample and Data Collection
To answer the research question and test the hypotheses an empirical quantitative analysis was conducted. Because of time constraints, the research was cross-sectional, using recent data from 2013. This choice of time horizon facilitates the comparison of results with previous studies, the majority of which are similarly cross-sectional, although they utilized older data. Because this thesis is exploring the relationship between culture and CSP, which are both publicly available and quantified measures, a secondary analysis of data using electronic databases was appropriate. For this purpose, multiple sources and datasets were combined. First, a sample of 500 global firms was selected from the 2013 Global Fortune 500 ranking. Because this ranking is based on measures of total revenue for the fiscal year 2013 it allowed us to have a sample of comparable companies and thus a more valid analysis.
Secondly, firm-level data was retrieved using Datastream, which is a database that combines data from several important sources such as national governments, Worldscope, OECD, IMF, MSCI. Specifically, data on firm environmental, social and governance (ESG) indices was
derived from Thomson Reuters ASSET4 database, while accounting and industry metrics were obtained from WorldScope. Given that for some observations no data was available, these were excluded from the sample. After a data inspection for such missing values, the final sample consists of 359 companies. This data set represents 34 countries across the world and 4 continents, which allows for considerable variance in the cultural scores. Additionally, country-level data on national competitiveness was taken from the 2013 Global
Competitiveness Report issued by the World Economic Forum; data on global GDP per capita was retrieved from the World Development Indicators found on the World Bank website and finally the data matrix with the four cultural dimension scores was obtained from Geert Hofstede’s website.
4.2 Dependent Variable
Given the multidimensional character of CSP, previous literature has proposed a number of different approaches for measuring the construct, which include databases and reputation indices such as the Fortune Index, Bloomberg, The Kinder, Domini (KLD) and Lydenberg (Turker, 2009). In constructing a CSP index, this study utilized the Environmental, Social and Governance (ESG) ratings from Thomson Reuters ASSET4, which is a comprehensive database with scores dating back to fiscal year 2002. The primary data of this database is publicly available and objective and is collected by specially qualified analysts. Specifically, it consists of more than 250 key performance indicators that have been compiled using data sources such as: CSR and annual reports, stock exchange filings, websites of
non-governmental organizations, and various news sources. For environmental factors the data typically describes information such as energy usage, CO2 emissions, water and waste recycled, spills and pollution disputes, while for social factors issues such as health and safety controversies, training hours, employee turnover, accidents, injury rate, donations and women employees are considered. These indicators are further structured into four pillars: environmental performance score, social performance score, corporate governance score and economic performance score. For each of the four pillars firms receive a z-score on a yearly basis which serves as a benchmark for their ESG performance and allows for inter-firm comparisons.
Given the lack of consensus across literature on the definition and measurement of CSP, the components used to construct the CSP index vary as well. For example, Ioannou and Serafeim (2012) operationalized CSP as a measure of environmental and social
metric. For the construction of the index we followed the latter method of Cheng, Ioannou and Serafeim (2013) and included in it the environmental, social and corporate governance metrics. Subsequently, a decision had to be made regarding the aggregation of the index components. Since there is no theory on how to distribute the weights between the measures, I followed Waddock and Graves (1997) and Waldman et al. (2006) and assigned equal importance and therefore equal weights to each pillar.
4.3 Independent Variables
As independent variables the four dimensions of national culture were used, namely Masculinity, Uncertainty Avoidance, Power Distance and Individualism. These are continuous variables measured using Hofstede’s indices (1985). Additionally, in order to correctly estimate the relationship between the independent and dependent variables, control variables have to be accounted for. These are factors that have been found by previous literature to have significant correlations with the main regression variables and in our case are represented by three types of control variables: country-level, industry level and firm level. At the country level I controlled for national prosperity, measured using GDP per capita, which was recommended by Tang and Koveos for researches that use Hofstede’s country scores as independent variables (2008). At the industry level, I controlled for the type of industry a firm belongs to. At the firm level, I controlled for company size, measured by Total assets, and for financial performance, operationalized as Return on assets, as it is expected that CSP will be higher for better performing firms and for larger firms (Ioannou & Serafeim, 2012). Additionally, six dummy variables denoting the world region in which the company is headquartered were created in order to control for regional differences that might influence CSP. Specifically, the home countries were classified in six major regions: Asia, Africa, Europe, North and Central America, South America and Australia and Oceania. Because of multicolinearity issues SPSS excluded in the regression the dummy Australia and Oceania. Therefore, in the present analysis the coefficients of the remaining dummies were compared to the base category Australia and Oceania.
4.4 Moderating variables
In order to measure national competitiveness the Global Competitiveness Index (GCI) was used, which represents a comprehensive tool that integrates the microeconomic and
macroeconomic factors of national competitiveness. The advantage of this tool is that it captures a wide range of factors, each evaluating a different facet of competitiveness and
groups them into 12 pillars: institutions, infrastructure, macroeconomic environment, health and primary education, higher education and training, goods market efficiency, labor market efficiency, financial market development, technological readiness, market size, business sophistication and innovation. The index itself is subsequently calculated as a weighted average of these 12 pillars. Although to a certain extent all the 12 pillars matter for each country, the weights are being allocated differently according to the stage of development that a country is in. Thus, the GCI has a dynamic character and accounts for the way countries evolve in their development path. In the first stage of development economies are considered to be factor-driven and compete using their factor capabilities such as natural resources and unskilled labor, while companies compete based on price, selling mainly commodity products. As the competitiveness increases, wages rise and countries enter the efficiency-driven stage characterized by efficient production processes and product quality. Finally, when the innovation-driven stage sets in, the only way to sustain continuously increasing wages and living standards is for businesses to compete with novel and unique products. Therefore, given the stage of a country’s development the GCI attributes higher relative weights to the pillars that are more important in that certain stage. This allows for a fairly accurate measurement of competitiveness.
For industry type the Industry Classification Benchmark (ICB) was used, which is a globally recognized taxonomy that classifies the market into industries, supersectors, sectors and subsectors within the macro economy. For the purpose of this paper only the industry category was considered, which consists of 10 types of industries denoted by a code: oil and gas, basic materials, industrials, consumer goods, healthcare, consumer services,
telecommunications, utilities, financials and technology. The companies in my sample were classified among these nine industries using dummy variables. Because of multicolinearity issues SPSS excluded in the regression the dummy for the financial industry. Therefore, in the analysis the coefficients of the remaining dummies were compared to the base category financials.
4.5 Method
In order to evaluate the strength of relationship between multiple numerical independent variables and a numerical outcome variable, the multiple regression technique was performed using the statistical software SPSS, but initially the variables were tested for normality. In order to assess normality the skewness and kurtosis of the distributions were analyzed. Given that with increasing sample sizes the standard error decreases, z-tests tend to be easily
rejected even for distributions that do not differ considerably from normality. Therefore, without considering z-scores the absolute values for skewness and kurtosis were examined against the reference values of 1. Thus, any values greater than 1 were considered to denote substantial non-normality. In particular, the variables CSP index and GCI score were found to be negatively skewed, while Total Assets was found to be positively skewed. This
non-normality issue was resolved by log transforming the substantially skewed variables. It is important to note that the transformation of the negatively skewed variables reversed the relationship with the original variable. Therefore, in the analysis and discussion this will be accounted for. Afterwards, four hierarchical multiple regressions were performed each one testing a different independent variable. This type of regression allows us to hold constant additional variables that have been proven by prior studies to be related with CSP, this way avoiding their influence on the studied relationship. This was done by first adding the block of control variables (Model 1), followed by the examined independent variable (Model 2). Additionally, in order to further examine the relationship between culture and CSP, two moderators were included: Oil and Gas industry sector (Model 3) and national degree of competitiveness (Model 4). The interaction effect between each predictor variable and each moderator was tested one at a time, first by mean centering the continuous variables in order to mitigate potential multicolinearity issues (Aiken & West, 1991). Subsequently, the
interaction term was computed by multiplying the two centered variables whose interaction is studied. After all the required transformations were executed hierarchical regression was applied.
5. Results
5.1 Sample characteristics
There were 359 companies in the sample with average total assets of 219m and average ROA of 4.68 (SD = 5.23). Altogether, these 359 cases covered 4 world regions: Asia, Europe, North and Central America, South America, Australia and Oceania. Although Africa is also an important geographic region it was not considered because none of my observations belonged to that group. As can be seen in Table 1, the most common region was Asia (33.4%) followed by North and Central America (32.6%) and Europe (29.8%). On the other hand, South America and Australia and Oceania were underrepresented with only 2.2% and respectively 1.9% of the total sample. Country wise, as Table 3 shows, most companies in my sample are headquartered in the United States (29.8%), followed by Japan (15.6%) and China (7.5%). Moreover, the distribution of the companies in the sample among the nine industries
can be seen in Table 2. The most common industry sector the companies belong to is the financial sector (19.5%) followed by industrials (16.4%), consumer goods and services sectors (14.2%; 13.4%) and the oil and gas sector (11.1%).
Table 1: Number of companies and average CSP scores by region
World region Nr. of companies % Mean CSP SD Asia 120 33.4 54.86 18.10 Europe 107 29.8 81.10 10.65
North & Central America 117 32.6 73.99 18.42
South America 8 2.2 65.30 13.26
Australia & Oceania 7 1.9 86.73 3.10
Table 1 also presents the average CSP per world region. It can be seen that Europe has a high average score (mean = 81.10; SD = 10.65), while Asia has a low score (mean = 54.86; SD = 18.10). Similarly, Table 3 displays the average CSP index scores by country. A significant variation in CSP scores can be observed across the 34 countries. Overall in my sample the scores range from 7.03 to 93.99, with a mean of 69.77 (SD = 19.51). Companies in Australia, France, Ireland and United Kingdom receive high CSP scores (Mean > 82), while firms in Russia, China and Hong Kong receive low scores (Mean < 54).
Table 2: Number of companies and average CSP scores by industry
Industry Nr. of companies % Mean CSP SD
Oil & Gas 40 11.1 70.76 18.15
Basic Materials 20 5.6 71.03 71.03 Industrials 59 16.4 68.33 21.53 Consumer Goods 51 14.2 70.87 15.14 Health Care 19 5.3 64.74 27.52 Consumer Services 48 13.4 71.14 15.87 Telecommunications 16 4.5 71.26 24.16 Utilities 17 4.7 69.00 17.32 Financials 70 19.5 67.41 22.40 Technology 19 5.3 77.61 12.37
In a similar manner, Table 2 shows the average CSP ratings by industry. It can be observed that the technology industry has the highest average CSP (mean = 77.61; SD = 12.37), followed by telecommunications (mean = 71.26; SD = 24.16) and consumer services (mean = 71.14; SD = 15.87). In contrast, healthcare (mean = 64.74; SD = 27.52), financials
(mean = 67.41; SD = 22.40), industrials (mean = 68.33; SD = 21.53) and utilities (mean = 68.99; SD = 17.31) industries have low CSP scores.
Table 3: Number of companies and average CSP scores by country
Country Nr. of companies % Mean CSP SD AUSTRALIA 7 1.9 86.73 3.10 AUSTRIA 1 .3 73.26 0 BELGIUM 3 .8 67.56 15.07 BRAZIL 7 1.9 62.86 12.24 CANADA 9 2.5 65.71 27.37 CHINA 27 7.5 40.99 21.98 COLOMBIA 1 .3 82.33 0 DENMARK 1 .3 65.14 0 FINLAND 1 .3 87.85 0 FRANCE 25 7.0 82.80 6.91 GERMANY 19 5.3 74.49 10.28 HONG KONG 7 1.9 44.82 21.31 HUNGARY 1 .3 79.98 0 INDIA 5 1.4 72.04 8.99 IRELAND 2 .6 89.47 1.64 ITALY 7 1.9 80.63 27.47 JAPAN 56 15.6 59.08 12.23 KOREA (SOUTH) 11 3.1 59.86 18.33 LUXEMBOURG 1 .3 84.33 0 MALAYSIA 1 .3 48.47 0 MEXICO 1 .3 27.90 0 NETHERLANDS 7 1.9 81.46 8.91 NORWAY 1 .3 83.61 0 RUSSIA 4 1.1 53.39 15.31 SAUDI ARABIA 1 .3 62.79 0 SINGAPORE 2 .6 68.97 3.27 SPAIN 6 1.7 81.32 7.78 SWEDEN 2 .6 82.65 6.14 SWITZERLAND 9 2.5 82.80 5.67 TAIWAN 4 1.1 57.54 21.15 THAILAND 1 .3 72.77 0 TURKEY 1 .3 69.90 0 UNITED KINGDOM 21 5.8 85.90 4.61 UNITED STATES 107 29.8 75.12 16.96 5.2 Regression
Four hierarchical multiple regressions were performed in order to examine the main effect of home-country cultural dimensions on CSP, as well as the moderating effects of the oil and gas industry type and national competitiveness on this main relationship. In the first step of
the hierarchical multiple regression (Model 1) five control variables were entered: company size, company financial performance, home country national prosperity, industry type and world region. Most of the control variables obtained significance in the predicted direction. This first model was statistically significant (F (16, 342) = 16.100; p-value < 0.001) and explained 43% of the variance in CSP. As can be seen in all four regression tables, financial performance (β = 0.010) and firm size (β = 0.174) were significant at 1% level and both showed a positive relationship with CSP, which is in line with the theory of prior literature. For GDP per capita on the other hand no significance was found. At the industry level, when compared to the reference financial industry dummy, I found that companies in the
technology industry have a CSP 0.442 points higher (p-value < 0.01), those in
telecommunications 0.283 points higher (p-value < 0.01) in industrials 0.223 points higher, and in consumer goods 0.218 points higher (p-value < 0.01). Similarly, firms in the basic materials industry have a CSP 0.222 points higher than the financials industry (p-value < 0.05) and so do utilities firms with a CSP index 0.169 points higher (p-value < 0.10). On the other hand, for those in healthcare (β = 0.082), utilities (β = 0.169) and consumer services (β = 0.071) no significance was found. Moreover, for the world region, when compared to the base region Australia and Oceania, I found that companies based in Europe have a CSP 0.345 points higher (p-value < 0.01), companies with a home country in South America 0.492 points higher (p-value < 0.01) and firms based in North and Central America have a CSP index 0.252 points higher (p-value < 0.10). For the remaining two regions Asia (β = 0.154) and North and Central America (β = 0.252) no significance was found.
5.2.1 Masculinity
In Table 4 can be observed how after the addition of the independent variable masculinity in the second model of the regression, the total variance in the CSP index explained by the second model was 43.2% (R2 change = 0.003; F (17, 341) = 15.285; p-value < 0.001). This shows only a minor increase in explanatory power, which suggests that masculinity does not offer significant additional predictive power to CSP. This is further supported by the
individual coefficient, which reveals no statistical significance for masculinity (β = 0.001, p-value = 0.192). Therefore, there is not enough evidence to reject the first hypothesis
according to which firms based in countries with higher masculinity score lower on CSP. In Model 3 the moderation effect of the oil and gas industry sector was tested. This model extends the second one with the interaction term between masculinity and the oil and
Table 4: Regression analysis for Masculinity
1 2 3 4
Constant 3.022*** (0.387) 3.091*** (0.390) 2.962*** (0.386) 3.160*** (0.394)
Independent Variable
Masculinity (MAS) -0.001 (0.001) 0.001 (0.001) -0.002* (0.001)
Oil and Gas x MAS 0.001 (0.003)
National competitiveness -1.886* (0.976) National competitiveness x MAS -0.010 (0.040) Control Variables Company size - 0.174*** (0.043) -0.173*** (0.043) -0.172*** (0.044) -0.172*** (0.043) Company financial performance -0.010*** (0.004) -0.011*** (0.004) -0.011*** (0.004) -0.011*** (0.004)
GDP/capita -9.926E-7 (0.000) -4.717E-7 (0.000) -4.224E-7 (0.000) -3.554E-6 (0.000)
Dummy variables: Industry
Oil & Gas -0.256*** (0.076) -0.265*** (0.076) -0.261*** (0.078) -0.252*** (0.076)
Telecommunications -0.283*** (0.093) -0.278*** (0.093) -0.277*** (0.093) -0.272*** (0.093) Technology -0.442*** (0.095) -0.447*** (0.095) -0.446***(0.096) -0.462*** (0.095) Healthcare -0.082 (0.094) -0.076 (0.094) -0.075 (0.095) -0.086 (0.094) Utilities -0.169* (0.092) -0.171* (0.092) -0.170* (0.092) -0.173* (0.092) Industrials -0.223*** (0.073) -0.228*** (0.073) -0.227*** (0.073) -0.234*** (0.073) Basic Materials -0.222** (0.090) -0.225** (0.090) -0.225** (0.090) -0.226** (0.090) Consumer Goods -0.218*** (0.074) -0.221*** (0.074) -0.220*** (0.220) - 0.228*** (0.074) Consumer Services -0.071 (0.082) -0.070 (0.082) -0.069 (0.082) -0.078 (0.082) Dummy variables: World Region Asia 0.154 (0.117) 0.175 (0.118) 0.177 (0.119) 0.119 (0.123) Europe -0.345*** (0.126) -0.364** (0.127) -0.366*** (0.127) -0.382*** (0.132)
North & Central America -0.252* (0.129) -0.260** (0.129) -0.262** (0.129) -0.301** (0.135) South America -0.492*** (0.181) -0.511*** (0.181) -0.514*** (0.182) -0.428*** (0.192) Model fit N 359 359 359 359 R2 0.430 0.432 0.433 0.439 Adj R2 0.403 0.404 0.402 0.407 F-stat 16.100 15.285 14.398 13.955 P-value 0.000 0.000 0.000 0.000
*** p < 0.01, ** p < 0.05, *p < 0.10 Values in parentheses are standard errors
gas industry sector. This addition leads to an explained total variance of 43.3% (R2 change = 0.000; F (18, 340) = 14.398; p-value < 0.001). Such a minor increase in explanatory power coupled with a decrease in the adjusted R2 from 0.404 in Model 2 to 0.402 in Model 3
indicates that there is no significant interaction (β = 0.001, p-value = 0 .827). Therefore, there is no evidence to reject hypothesis 6a, according to which the oil and gas industry acts as a moderator for the relationship between home country masculinity and CSP.
Similarly model 4, which explained 43.9% of the total variance in the dependent variable CSP (R2 change = 0.006; F (19, 339) = 13.955; p-value < 0.001), showed no significant interaction effect between masculinity and national competitiveness (β = -0.010, p-value = 0.801). Given this inconclusive evidence, hypothesis 5a which predicted a
moderation effect for home country national competitiveness cannot be rejected.
5.2.2 Uncertainty avoidance
In the second regression the predictor variable uncertainty avoidance was tested. As can be seen in Model 2 from Table 4, a 43% total explained variance indicates that uncertainty avoidance has no significant effect on the dependent variable CSP (R2 change = 0.000; F (17, 341) = 15.110; p-value < 0.001). This is further supported by the decrease in the Adjusted R Square from 0.403 to 0.401, which suggests that uncertainty avoidance offers no significant additional predictive power to the set of control variables. The t-test result for the individual regression coefficient of uncertainty avoidance (β = 0.000, p-value = 0 .905) also reveals that there is no basis on which to reject the second hypothesis, according to which home country uncertainty avoidance has a positive effect on CSP.
Furthermore, Model 3 that explains 43.2% of the total variance in CSP (R2 change = 0.002; F (18, 340) = 14.347; p-value < 0.001) shows that there is no significant interaction effect between uncertainty avoidance and the oil and gas industry sector (β = 0.003, p-value = 0.270). Given this lack of evidence, hypotheses 6b that predicted a moderation effect of the oil and gas industry type on the relationship between home country uncertainty avoidance and CSP cannot be rejected.
Finally, the moderation effect of national competitiveness was tested in Model 4, which explained 43.3% of the total variance in CSP (R2 change = 0.004; F (19, 339) = 13.646; p-value < 0.001). The individual regression coefficients reveal no significant
0.939). Thus, there is insufficient evidence to reject hypothesis 5b, which predicted a moderation effect of national competitiveness.
5.2.3 Power Distance
After entering power distance in the second model of the third hierarchical regression, the total variance in CSP explained by the model was 43.1% (R2 change = 0.001; F (17, 341) = 15.184; p-value < 0.001). This model indicates only a minor increase in explanatory power as compared to the block of control variables, however, because adjusted R2 decreases from 0.403 to 0.402 it means that the additional explanatory power is less than would be expected
Table 5: Regression analysis for Uncertainty Avoidance
1 2 3 4 Constant 3.022*** (0.387) 3.025*** (0.389) 2.962*** (0.386) 3.153*** (0.398) Independent Variable Uncertainty Avoidance (UAI) 0.000 (0.001) -8.562E-5 (0.001) 0.000 (0.001)
Oil and Gas x UAI 0.003 (0.002)
National competitiveness -1.370 (0.963) National competitiveness x UAI -0.002 (0.028) Control Variables Company size - 0.174*** (0.043) -0.173*** (0.044) -0.170*** (0.044) -0.174*** (0.044) Company financial performance -0.010*** (0.004) -0.010*** (0.004) -0.010*** (0.004) -0.010*** (0.004)
GDP/capita -9.926E-7 (0.000) -9.464E-7 (0.000) -8.908E-7 (0.000) -3.524E-6 (0.000)
Dummy variables: Industry
Oil & Gas -0.256*** (0.076) -0.254*** (0.077) -0.258*** (0.077) -0.245*** (0.077)
Telecommunications -0.283*** (0.093) -0.282*** (0.094) -0.279*** (0.094) -0.283*** (0.095) Technology -0.442*** (0.095) -0.441*** (0.096) -0.439***(0.096) -0.456*** (0.096) Healthcare -0.082 (0.094) -0.081 (0.095) -0.082 (0.095) -0.092 (0.095) Utilities -0.169* (0.092) -0.168* (0.093) -0.161* (0.093) -0.172* (0.093) Industrials -0.223*** (0.073) -0.223*** (0.073) -0.223*** (0.073) -0.226*** (0.073) Basic Materials -0.222** (0.090) -0.222** (0.090) -0.218** (0.090) -0.222** (0.090) Consumer Goods -0.218*** (0.074) -0.216*** (0.075) -0.212*** (0.075) -0.225*** (0.075) Consumer Services -0.071 (0.082) -0.069 (0.083) -0.069 (0.082) -0.079 (0.083) Dummy variables: World Region
Asia 0.154 (0.117) 0.152 (0.118) 0.158 (0.119) 0.103 (0.127)
Europe -0.345*** (0.126) -0.349*** (0.129) -0.344*** (0.129) -0.351*** (0.131)
North & Central America -0.252* (0.129) -0.257* (0.136) -0.251* (0.136) -0.273** (0.137) South America -0.492*** (0.181) -0.497*** (0.186) -0.496*** (0.186) -0.420** (0.200) Model fit N 359 359 359 359 R2 0.430 0.430 0.432 0.433 Adj R2 0.403 0.401 0.402 0.402 F-stat 16.100 15.110 14.347 13.646 P-value 0.000 0.000 0.000 0.000
*** p < 0.01, ** p < 0.05, *p < 0.10 Values in parentheses are standard errors
by chance. Additionally, the t-test for the β coefficient returns a value of 0.859 at a
significance of 0.391. Therefore, I conclude that there is not enough evidence to reject the fourth hypothesis, which stated that home country power distance has a negative effect on CSP.
The third model, which explains 43.1% of the total variance in the dependent variable CSP (R2 change = 0.000; F (17, 341) = 15.184; p-value < 0.001) reveals that the expected moderation effect of the oil and gas industry sector on the relationship between the independent variable power distance and CSP is not significant (β = 0.001, p-value = 0.636). Considering this lack of evidence, hypothesis 6c, which predicted that the oil and gas
industry sector strengthens the relationship between home country power distance and CSP, cannot be rejected.
The final model that tested the interaction effect between national competitiveness and power distance was overall significant and explained 44.4% of the total variance in the dependent variable (R2 change = 0.013; F (19, 339) = 14.266; p-value < 0.001). Because this model accounted for significantly more variance than the model without the interaction term, a potentially significant moderation between power distance and national competitiveness might be taking place. I start by recalling that the main effect of power distance on CSP was not significant, but the main effect of national competitiveness on CSP was significant (β = -2.106, p-value = 0.035). A look at the individual coefficients indicated that the predicted moderation was indeed negatively significant at a 10% level (β = -0.071, p-value = 0.055). Examination of the moderation plot displayed in Figure 1 shows that for high levels of national competitiveness the negative correlation between power distance and CSP is weaker
(R2 = 0.241) than for lower levels of national competitiveness (R2 = 0.247). Therefore, hypothesis 5c is supported.
5.2.4 Individualism
In the final regression, the independent variable individualism was tested. Contrary to the expected negative direction, Model 2, with an explained variance of 45.4 %, showed that individualism has a statistically significant positive effect (R2 change = 0.025; F (17, 341) = 16.692; p-value < 0.001) on the dependent variable CSP. Furthermore, the adjusted R2 showed an increase from 0.403 to 0.427, which confirms that individualism improved the model more than would be expected by chance. The t-test statistic for the individual
regression coefficient showed (β = 0.007, p-value = 0.000) that the predictor was significant Table 6: Regression analysis for Power Distance
1 2 3 4
Constant 3.022*** (0.387) 2.922*** (0.404) 2.972*** (0.384) 3.119*** (0.392)
Independent Variable
Power Distance (PDI) 0.002 (0.002) 0.002 (0.002) 0.002 (0.002)
Oil and Gas x PDI -0.001 (0.003)
National competitiveness -2.106** (0.995) National competitiveness x PDI -0.071* (0.037) Control Variables Company size - 0.174*** (0.043) -0.176*** (0.043) -0.176*** (0.043) -0.180*** (0.043) Company financial performance -0.010*** (0.004) -0.011*** (0.004) -0.010*** (0.004) -0.012*** (0.004)
GDP/capita -9.926E-7 (0.000) -1.496E-7 (0.000) -1.816E-7 (0.000) -2.008E-6 (0.000)
Dummy variables: Industry
Oil & Gas -0.256*** (0.076) -0.256*** (0.076) -0.251*** (0.077) -0.242*** (0.076)
Telecommunications -0.283*** (0.093) -0.286*** (0.093) -0.288*** (0.093) -0.285*** (0.092) Technology -0.442*** (0.095) -0.439*** (0.095) -0.439***(0.096) -0.449*** (0.095) Healthcare -0.082 (0.094) -0.082 (0.094) -0.081 (0.095) -0.103 (0.094) Utilities -0.169* (0.092) -0.170* (0.092) -0.171* (0.092) -0.161* (0.091) Industrials -0.223*** (0.073) -0.226*** (0.073) -0.227*** (0.073) -0.227*** (0.073) Basic Materials -0.222** (0.090) -0.222** (0.090) -0.221** (0.090) -0.229** (0.089) Consumer Goods -0.218*** (0.074) -0.220*** (0.074) -0.221*** (0.074) -0.219*** (0.074) Consumer Services -0.071 (0.082) -0.073 (0.082) -0.071 (0.082) -0.080 (0.081)
Dummy variables: World Region
Asia 0.154 (0.117) 0.147 (0.118) 0.148 (0.118) 0.101 (0.126)
Europe -0.345*** (0.126) -0.339*** (0.126) -0.338*** (0.126) -0.340*** (0.126)
North & Central America -0.252* (0.129) -0.242* (0.129) -0.241* (0.130) -0.274** (0.130)
South America -0.492*** (0.181) -0.487*** (0.181) -0.483*** (0.181) -0.357* (0.186) Model fit N 359 359 359 359 R2 0.430 0.431 0.431 0.444 Adj R2 0.403 0.402 0.401 0.413 F-stat 16.100 15.184 14.321 14.266 P-value 0.000 0.000 0.000 0.000
*** p < 0.01, ** p < 0.05, *p < 0.10 Values in parentheses are standard errors
Figure 1: Moderation effect of national competitiveness on the relationship between power distance and CSP
at 1% level. Although the effect is not very strong it still indicates that companies based in more individualistic countries have better CSP than those based in less individualistic countries. Consequently, the third hypothesis is rejected.
Similarly to the previous model, Model 3 explained 45.4% of the total variance (R2 change = 0.000; F (18, 340) = 15.724; p-value < 0.001). Additionally, the adjusted R2 went down from 0.427 to 0.425. This indicates that the model with the interaction term between individualism and the oil and gas industry sector did not increase the amount of variance that the model accounts for. Furthermore, the individual coefficient (β = 0.236, p-value = 0.813) showed that the interaction term between individualism and the oil and gas industry sector was not significant. It follows that there is not enough evidence to reject hypothesis 6d, according to which the oil and gas industry acts as a moderator for the main relationship between home country individualism and CSP.
Finally, Model 4 explaining 46.5% of the total variance in CSP showed that the predicted moderation effect of national competitiveness in the case of individualism was not significant (R2 change = 0.011; F (19, 339) = 15.490; p-value < 0.001). This is also indicated by the individual coefficients (β = -0.024, p-value = 0.440). Therefore, there is inconclusive evidence to reject hypothesis 5d, according to which national competitiveness moderates the relationship between home country individualism and CSP.
Table 7: Regression analysis for Individualism
1 2 3 4
Constant 3.022*** (0.387) 3.259*** (0.384) 2.798*** (0.379) 2.958*** (0.387)
Independent Variable
Individualism (IDV) -0.007*** (0.002) -0.007*** (0.002) -0.007*** (0.002)
Oil and Gas x IDV -0.000 (0.002)
National competitiveness -2.192** (0.888) National competitiveness x IDV -0.024 (0.031) Control Variables Company size - 0.174*** (0.043) -0.176*** (0.042) -0.176*** (0.043) -0.176*** (0.042) Company financial performance -0.010*** (0.004) -0.011*** (0.004) -0.011*** (0.004) -0.012*** (0.004)
GDP/capita -9.926E-7 (0.000) -6.412E-7 (0.000) -6.243E-7 (0.000) -2.270E-6 (0.000)
Dummy variables: Industry
Oil & Gas -0.256*** (0.076) -0.256*** (0.074) -0.254*** (0.075) -0.238*** (0.074)
Telecommunications -0.283*** (0.093) -0.290*** (0.091) -0.291*** (0.091) -0.282*** (0.091)
Technology -0.442*** (0.095) -0.443*** (0.093) -0.444***(0.094) -0.457*** (0.093)
Utilities -0.169* (0.092) -0.174* (0.090) -0.175* (0.090) -0.173* (0.090) Industrials -0.223*** (0.073) -0.238*** (0.071) -0.227*** (0.073) -0.242*** (0.071) Basic Materials -0.222** (0.090) -0.216** (0.088) -0.216** (0.088) -0.219** (0.088) Consumer Goods -0.218*** (0.074) -0.220*** (0.072) -0.221*** (0.073) -0.224*** (0.072) Consumer Services -0.071 (0.082) -0.057 (0.080) -0.056 (0.080) -0.064 (0.080) Dummy variables: World Region Asia 0.154 (0.117) 0.124 (0.115) 0.122 (0.115) 0.067 (0.126) Europe -0.345*** (0.126) -0.148 (0.133) -0.149 (0.134) -0.132 (0.133)
North & Central America -0.252* (0.129) 0.047 (0.148) 0.046 (0.148) 0.021 (0.148)
South America -0.492*** (0.181) -0.217 (0.190) -0.216 (0.191) -0.078 (0.197) Model fit N 359 359 359 359 R2 0.430 0.454 0.454 0.465 Adj R2 0.403 0.427 0.425 0.435 F-stat 16.100 16.692 15.724 15.490 P-value 0.000 0.000 0.000 0.000
*** p < 0.01, ** p < 0.05, *p < 0.10 Values in parentheses are standard errors
6. Discussion
From the results outlined above it can be observed that most relationships analyzed in this study were found to be non significant. Regarding my main regression, which tested the effects of Hofstedes’s four cultural dimensions on CSP, masculinity, uncertainty avoidance and power distance turned out to be non significant. This high proportion of insignificant outcomes is surprising considering that previous literature, which analyzed the same
relationships, has consistently found significant results. Nonetheless, it is worth to note that the findings of this past literature are largely contradictory when it comes to the direction of the relationships between Hofstede’s cultural dimensions and CSP. For example, Ringov and Zollo (2007) found a negative effect for power distance, while Ho et al. (2011) and Ioannou and Serafeim (2012) found a positive effect. Similarly, for masculinity Ringov and Zollo (2007) found a negative relationship, while Ho et al. (2011) found a positive relationship. Only for uncertainty avoidance the literature seems to agree that it influences CSP in a positive way (Ringov & Zollo, 2007; Ho et al., 2011). This variation in previous findings along with the non-significance of results in the present paper indicates that more research is definitely needed in order to clarify the relationship between culture and CSP and reconcile all the conflicting results and opinions.
A possible explanation for my insignificant findings is related to the construction of the sample. While former studies used large samples of more than 10,000 companies, the present study is limited to only 359 observations, obtained after the exclusion of 28.2% of the observations in the initial sample due to incomplete information. This may have caused an issue of low statistical power. Additionally, in my study some countries and world regions such as South America and Australia and Oceania have very low numbers of companies compared to the remaining regions, while the African region is omitted entirely given a lack of observations. Taking into consideration that the type of analysis conducted is cross-national, the countries and regions considered should be better represented in order to get more reliable results.
The only significant independent variable out of the four cultural dimensions was individualism. My findings reveal a significant opposite direction than was predicted. An explanation for the observed positive relationship between home country individualism and CSP may be that because in individualistic societies economic actors are provided with a greater discretion, they tend to engage more in voluntary and explicit CSR (Ioannou & Serafeim, 2012). As Porter and Kramer note, explicit CSR is often also strategic (2006). This means that the explicitness of the decisions and actions of managers from individualistic societies is marked by their pursuit of recognition (Ioannou & Serafeim, 2012). In the case of CSR this may be recognition from various stakeholder groups that expect companies to act responsibly. On the other hand, as was suggested by Ioannou and Serafeim, societies that are more collectivist would prefer implicit CSR because according to them altruism should be performed discreetly rather than be displayed selfishly (2012). In such societies implicit CSR is usually institutionalized (Jackson & Apostolakou, 2010). Considering that CSP is defined as a measure of explicit and voluntary CSR initiatives, this may explain why higher levels of home country individualism lead to higher CSP scores.
Furthermore, no significant interaction effect was found for any of the four cultural dimensions and the oil and gas industry sector. Similar lack of significance was observed for the interaction effect of national competitiveness, however, only for three out of the four cultural dimensions: individualism, uncertainty avoidance and masculinity. Power distance, on the other hand exhibited a significant interaction with national competitiveness, which was in line with my prediction. Specifically, I found that home country national competitiveness acts as a buffer for the effect of power distance on companies’ CSP. Thus, for higher levels of home country national competitiveness, power distance has a weaker negative effect on CSP than it would for lower national competitiveness. Interestingly, power distance alone did not
appear to have a significant effect on CSP, thus, it could be that CSP depends on home country power distance only for some levels on national competitiveness. However, more research is needed in order to elucidate this relationship.
As both of my moderation analyses turned out to be largely non significant, I
attribute this fact to the same problem of statistical power as was previously explained in this section. In the case of moderation, however, apart from total sample size, the size across moderator-based subgroups could also influence statistical power. As was indicated by Aguinis (1995) having unequal sample sizes across moderator-based subgroups could be detrimental to the power of the moderated regression and this was the case in the present study. Moreover, it is important to note that previous research has not looked at factors that could moderate culture and instead has explored the moderating effects of culture itself. Because it is impossible to compare my moderation analyses it is complicated to conclude whether the insignificant results are due to issues arising from my method or whether the assumed moderators are in fact not associated with my predictors in the studied context of CSP.
Aside from the main variables studied, a number of control variables were also included. The majority of the controls were found to be significant in the hypothesized direction, as claimed by previous studies. Therefore, at the firm level the effect of firm size and financial performance was found to be positive and significant, which indicates that larger and better performing firms have higher CSP than smaller ones, thus confirming prior work by Ioannou and Serafeim (2012). Furthermore, at the industry level I find that
companies in certain industries such as technology and telecommunications display higher CSP than companies in other industries. Similar results were found for world regions, with regions such as Europe having higher CSP scores than others. However, at the country level, national prosperity appears to have no significant influence on the CSP variance. While this contradicts the findings of Tang and Koveos (2008), it might be because in their study the variable was log transformed due to highly skewed distribution, which was not the case in the present paper.
7. Conclusion
The main objectives of this paper were to uncover the nature of the relationship between the cultural dimensions of a company’s home country and its CSP, as well as to investigate whether the strength of this relationship changes with different levels of national competitiveness and whether it is different for companies belonging to the oil and gas
industry sector. In order to achieve this, multiple regression analysis was applied to a sample of 359 global companies from 34 countries. Unexpectedly the results show that out of four cultural dimensions only individualism has a significant positive effect on CSP. Additionally, I find that for higher levels of home country national competitiveness the negative
relationship between power distance and CSP is weakened. The remaining relationships were found to be non significant. Considering my largely insignificant results and the fact that across literature the findings regarding the effects of culture on CSP are inconsistent, I
acknowledge that a more systematic research is needed in order to clarify these contradictions as well as uncover additional possible factors that might affect CSP but have been omitted or underexplored so far. For example, in my sample most companies operate in multiple
countries, however, the presented empirical analysis does not account for this level of internationalization. It may be the case that companies pursuing a global expansion engage more in CSR, as they depend on international reputations and international financial markets. Additionally, it may be that a company’s operations abroad influence its CSP. Third, I note that my sample contained several state-owned companies, which have a largely domestic production. The records of these companies are usually less scrutinized by civil groups, which could imply that they may have less pressure to perform responsibly. All these aspects were outside the scope of this paper, however, future research could follow these avenues in order to shed more light on the complex topic of CSR.
Accenture. The UN Global Compact-Accenture CEO Study on Sustainability. (2013). Accessed on 1 June 2015.
(https://acnprod.accenture.com/~/media/Accenture/Conversion- Assets/DotCom/Documents/Global/PDF/Strategy_5/Accenture-UN-Global-Compact-Acn-CEO-Study-Sustainability-2013.pdf)
Aguilera, R. V., Rupp, D. E., Williams, C. A., & Ganapathi, J. (2007). Putting the S back in corporate social responsibility: A multilevel theory of social change in
organizations. Academy of management review, 32(3), 836-863.
Aguinis, H. (1995). Statistical power with moderated multiple regression in management research. Journal of Management, 21(6), 1141-1158.
Aiken, L. S., & West, S. G. (1991). Multiple regression: Testing and interpreting interaction. Thousand Oaks, CA: Sage.
Akaah, I. P. (1990). Attitudes of marketing professionals toward ethics in marketing research: A cross-national comparison. Journal of Business Ethics, 9(1), 45-53.
Banerjee, S. B., Iyer, E. S. & Kashyap, R. K. (2003). Corporate Environmentalism:
Antecedents and Influence of Industry Type. Journal of Marketing, 67, pp. 106-122. Carroll, A. B. (1979). A three-dimensional conceptual model of corporate performance.
Academy of management review, 4(4), 497-505.
Chapple, W., & Moon, J. (2005). Corporate social responsibility (CSR) in asia a seven-country study of CSR web site reporting. Business & society, 44(4), 415-441. Cheng, B., Ioannou, I., & Serafeim, G. (2014). Corporate social responsibility and access to
finance. Strategic Management Journal, 35(1), 1-23.
Cohen, J. R., Pant, L. W., & Sharp, D. J. (1996). A methodological note on cross-cultural accounting ethics research. The International Journal of Accounting, 31(1), 55-66. Deanna Wang, H. M. (2010). Corporate social performance and financial-based brand equity.
Journal of product & Brand management, 19(5), 335-345.
Geert Hofstede. Accessed on 1 June 2015 (http://geerthofstede.nl/research--vsm). Ho, F. N., Wang, H.-M. D., & Vitell, S. J. (2011). A global analysis of corporate social
performance: The effects of cultural and geographic environments. Journal of Business Ethics, 107(4), 423-433.
Hofstede, G. (1985). The interaction between national and organizational value systems [1]. Journal of Management Studies, 22(4), 347-357.
Hofstede, G. (1994). The business of international business is culture. International business review, 3(1), 1-14.
