Literature review
Consumer responses to multinational corporation ethical behaviour can arguably correlate to CSR. Significant work has been undertaken on understanding these responses and how they differ depending upon the view of the corporation’s behaviour, and involves analysing CSR policies and their effect on customer’s opinions and understanding. Wan et al [1] explain how research indicates a positive relationship between financial and social performance. Moreover, reaction from consumers to CSR policies is largely positive and includes paying more for products or services, such as Fairtrade products, where purchasing gives a feel-good factor.
The definition of face-concern according to [1], is an individual’s concern or consideration of their actions in regard to keeping face, which is a considered factor. Those with high face-concern would consider carefully any association with a brand that behaved unethically, that could reflect poorly back upon them. Consequently, a customer’s personal beliefs and concerns for face will directly affect their standpoint on corporations and their actions. Those with higher face-concern are more sensitive to the CSR activities and its effect on the reputation of a corporation, than someone who has less face-concern.
The idea that personal beliefs relate directly to consumer response is supported by Abdeen et al [2]. They propose four belief categories; economic, legal, environmental and philanthropic and demonstrate the different significance each has on the response of the consumer; each person weights these factors differently, affecting the response from each consumer. Using statistical and regression based analysis on consumer survey data from New Zealand, the authors found each of the beliefs had different effects on support intention and purchase behaviour of consumers. Support intention is the conscious idea to support organisations who behave ethically. The idea of direct and indirect effects is presented; the only direct effect on purchase behaviour that was significant, was ethical beliefs. The indirect effect of philanthropic and legal beliefs were found to be significant, supporting the idea of a mediation effect of support intention between these beliefs and purchase behaviour.
Additionally, some work has identified cultural differences as a factor affecting responses, such as Maignan, who tested cultural-based hypotheses using statistical techniques on a French, German and U.S. consumer survey [3]. By breaking down the elements of CSR into economic, environmental, social and philanthropic areas, it demonstrated that different cultures value the importance of these differently. For instance, Germany and France were more concerned with legal issues concerning a corporation, whereas the U.S. were more concerned with economic issues. Germany and France were reasonably aligned in their views and the importance of economic, environmental and social concerns of a corporation. Results indicated that consumers from all countries are willing to pay for goods or services from corporations deemed to have acted in an ethical manner.
Traditionally, customer demographics have been considered as one of the most influential factors in consumer response to activities of corporations and is a significant factor in the consideration of CSR. One paper investigated traditional factors of age, gender and education level, as well as psychological factors such as customer support for CSR, customer collectivism and customer novelty seeking [4]. Using statistical analysis on a survey from customers of Spanish banking services, it was found that both gender and education level were rejected as being significant factors however the hypothesis that age is an influential factor in CSR perception and response, could not be rejected. Although the survey is applied in specific context, the results can be considered for multinational corporations who may apply the same methods towards CSR activities and consumer segmentation.
Research also indicates consumers treat multinational corporations differently to smaller companies and have different expectations. Green and Peloza studied the effect of company size on consumer response, in terms of consumer trust and expectation towards corporations [5]. To do this, a diverse range of consumers across North America were interviewed regarding CSR and corporation size. Expectations of large corporations were much higher than for small corporations and seen as a minimum requirement, whereas for small corporations, the expectation is more ‘do your best’. Additionally, consumers felt small corporations were already socially responsible; they are often local firms employing local people, boosting the local economy. They are also trusted and forgiven more than large firms; therefore, their ethical behaviour has less impact on consumer response. Consumers view small-firm unethical behaviour as ‘needs must’, whereas large firms are suspected to have ulterior motives; their motives are often questioned, more so than for small firms. However, the paper finds that local initiatives often help consumers believe in the integrity of large firms. These findings suggest that consumers see economic and resource constraints on small firms that they don’t see for large firms.
From the literature review, 3 hypotheses are formulated:
1: Consumer response to ethical behaviour of multinational corporations through CSR is directly related to the consumers’ own beliefs and ethical standpoint.
2: Consumer response varies culturally, depending on consumer citizenship.
3: Young people, women and people with high-level education are more expectant in terms of the ethical behaviour of multinational corporations and are therefore more sensitive in response to multinational corporations’ activities.
Analysis
Data analysis was undertaken in SAS and cleaning and updating of data was required to obtain a suitable format for analysis. Proc print was used to identify data requiring cleaning. Missing numerical data, excluding age, was replaced with the mean using a proc standard command. Sub-parts of each question were aggregated using a simple sum and the sub-part variables dropped. Most issues in the data consisted of incorrect spelling or incorrect data entry; records where incorrect data made the record hard to interpret, such as gender, were deleted, and incorrect spelling or errors cleaned, using a series of if statements. Most data cleaning was undertaken on the Country of Birth, Citizenship, Income and Education fields as there were different interpretations of the same response. Labels were also assigned to the variables. Following data cleaning, some variables were re-coded, such as gender, personal income, household income, and Religious affiliation. The purpose was to be able to perform analysis, such as regression, if required.
The first hypothesis relates to the relationship between consumer’s personal ethical beliefs and ethical beliefs about corporations. Research indicated high personal ethical values often result in high ethical response towards corporations. However, this was found to be untrue in the data set given. Applying a regression model to Q15, (ethical beliefs of personal actions), against Q1-4, (consumers’ opinion of corporations ethical behaviour), the model was an extremely poor fit, with an adjusted R-squared value of 0.0159, although the p-value for personal actions was below 0.0001 which implies significance. The fit plot of the total ethical behaviour score against personal behaviour score, shown below, indicates little trend or relationship appearing. This is supported by the correlation statistic where Pearson, Spearman and Kendall’s correlation coefficients ranged between 0.08 to 0.13 to 2 decimal places, thus implying a small level of positive correlation but not the substantial amount expected. A better fitting model was actually achieved using just consumers’ beliefs on corporations, shown in figure 2. However, it is still fair to reject the proposed hypothesis, since the data does not support a strong relationship.
Figure 1- Fit Plot for Observation Score Regression Model
Figure 2-Fit Plot for Belief Regression Model
The second hypothesis relates to consumer cultural differences and whether this affects responses. To test this, a significance test can be performed on consumers’ response scores to test whether the means of the respective cultural groups are different; this is firstly performed against all the different citizenships. Citizenship was chosen as it infers the consumer has spent a period of time there and thus probably has more influence on the consumer than their place of birth, where they may not have lived for long. The F-Test from the ANOVA tables indicates that with an F-Value of 5.98, the means are significantly different which therefore implies that culture affects consumer response. However, explanatory analysis shows some citizenships contain minimal observations, thus groups are diluted, hence it was decided to group by continent; this was mostly done geographically except for one or two special cases. An ANOVA table was used to analyse the differences in the mean response scores across these groups; an F-Value of 7.81 was obtained, implying the means are significantly different. Of the different combinations of culture groups under pairwise t-test, groups involving Europe or North America were significant, in terms of difference in mean, at the 0.05 level indicating these groups in particular, respond differently to other groups. This supports conclusions drawn by Maignan [3]. The South American group had the lowest mean response scores indicating they are more influenced in their response towards ethical behaviour through CSR initiatives. Based on the evidence presented, the second hypothesis can be accepted.
Figure 3 – Box and Whisker plot for Distribution of Observations
The third hypothesis relates to consumer demographic. Research indicated that expectations were higher depending on age, but not for gender or education level. To evaluate whether these consumers are more sensitive to CSR activities and thus respond more extremely than others, a series of statistical analyses were carried out. An ANOVA test on education level produced a p-value smaller than 0.001, indicating the means are significantly different in the groups. Furthermore, pairwise t-tests revealed most group pairs had statistically different means at 0.05 significance level. Observing the box and whisker plots produced, doctoral, masters and college/university degrees produced lower scores on corporation ethical behaviour questions, meaning they are more likely to respond positively to ethical behaviour and negatively to unethical behavior. This result contradicts results found in the literature review. Moving on to gender, a t-test was performed to compare male and female responses. The results clearly indicate that the means are different at 0.05 significance level. Female responses produced a lower mean and smaller range, shown by figure 6, suggesting their response is more sensitive to ethical behaviour of multinational corporations. Grouping response data by consumer age, an ANOVA table was also used to compare the mean responses of the groups. It was found that the means were different at the 5% significance level, but the 17-25 group had the highest mean of all the groups, shown by figure 7. There is also negative correlation between age and response values. This contradicts the research and results presented in [4]. A regression model was also produced using these characteristic variables, and all were found to be significant, all remaining in the regression equation. However, the model was a bad fit, with an adjusted R-squared value of 0.024 (2.4%), suggesting other variables exist, that have a significant impact on the observation value from the survey questions. Overall, the third hypothesis cannot be wholly accepted. Although the evidence above supports the arguments made for gender and education level, it does not support the argument that young people are more sensitive in their response to corporation behaviour.
Figure 4 – Box and Whiskers Plot of Observation
Figure 5 – Gender Comparison of Observation
Figure 6- Box and Whiskers Plot of response variable Observation
Figure 7- Output from Regression Model