Assessing the structural model

Before assessing the actual paths and their probability, the multivariate assumptions are analysed via observing outliers and multivariate influentials, and multicollinearity.

Outliers and influentials

In order to identify any multivariate influential outliers a Cook's distance analysis (See Cook 1977) was ran in SPSS 25 with linear regression analysis and chart builder tool. No cases below 1 was identified concerning both CHAFOC and ENREFOC and their independent variables. And only one case was greater than 0.1 (0.128) when comparing to latent dependent variable CHAFOC, but when comparing to ENREFOC it was far less than 0.10.

Cook's distance analysis was calculated also between the final dependent brand reputation variable (named as COMPADV7 in the data) and latent CHAFOC and ENREFOC variables. As said earlier brand reputation variable is the single item (observed) variable in this study, hence only included in the last step of SEM. No a Cook's distance greater than 1 was observed. Only one case was differing from the others (0.120) but when comparing to CHAFOC and ENREFOC as dependent variables a Cook’s distance was below 0.05. Consequently, multivariate influential outliers should not exist or affect significantly on this research.

Multicollinearity

Variable inflation factors (VIFs) were examined in SPSS 25 for all predictors of two dependent variables, CHAFOC and ENREFOC (See Table 6) in order for analysing multicollinearity.

Regression analysis was again used in SPSS 25, in which linear regression and collinearity diagnostics were used.

Table 6: Variance inflation factors and tolerance with CHAFOC and ENREFOC variables.

As recommended by Kock and Lynn (2012) VIFs greater than 3 were not observed. Only two variables (PINST and LCOM) have VIF greater than 2 which is recommended by Zuur et al (2010).

Additionally, COMPADV7/Global brand reputation variable was examined with its two predictor variables, CHAFOC and ENREFOC, and also their VIFs are less than 2 (See Table 7). Therefore, no multicollinerity should affect the SEM.

Dependent Variable: COMPADV7/

Global brand reputation

Independent Variable Tolerance VIF

CHAFOC 0,833 1,2

ENREFOC 0,833 1,2

Table 7: Variance inflation factors and tolerance with global brand reputation variable.

Dependent Variable:

CHAFOC

Dependent Variable:

ENREFOC

Independent Variable Tolerance VIF Independent Variable Tolerance VIF

CHAMAG 0,889 1,125 ENREMAG 0,937 1,067

CUSOR 0,806 1,24 CUSOR 0,814 1,228

FINCO 0,889 1,125 FINCO 0,874 1,144

INSTBAR 0,752 1,33 INSTBAR 0,749 1,336

PINST 0,447 2,238 PINST 0,447 2,237

LBUS 0,748 1,337 LBUS 0,752 1,331

LCOM 0,487 2,054 LCOM 0,506 1,977

Path model, path coefficients and probability

The path model and its path coefficients and probabilities are illustrated in Figure 8. In total 16 different paths were hypothesized. The path model involves company characteristics (including philanthropic and environmental CSROs that are positioned separately in the model) and their relation to GCSR activity, institutional factors and their relation to GCSR activity, and GCSR and its relation to global brand reputation. Both statistically significant (bold solid lines in the figure) and non-significant paths (dashed lines in the figure) can be seen with their path coefficients.

INSTITUTIONAL

CUSOR = Export customer orientation, FINCO = Financial constraints, INSTBAR = Institutional barriers,

PINST = Political institutions, LBUS = Local businesses, LCOM = Local communities, CHAMAG = Philanthropic CSR orientation, ENREMAG = Environmental CSR orientation, CHAFOC = Global focus of charitable and philanthropic activity, ENREFOC = Global focus of environmental sustainability activity

CHAMAG

The hypotheses, paths, path coefficients as betas (β) and probabilities as p-values can be observed from Table 8, from which can be seen whether the hypothesis is supported or not (conclusion). The results are summarized briefly and after that the actual main findings, theoretical conclusions and managerial implications are discussed (See Chapter 5).

H5a: Barriers based on institutional communities on the export market, and the level of these relationships impact

H8b: The extent and frequency of the

Table 8: Paths, path coefficients and probabilities of the structural model.

Company characteristics

It was suggested that if export companies are engaging in philanthropic CSR orientation, including charitable and social activities, it would affect positively the global focus of philanthropic charitable activity (path CHAMAG à CHAFOC). However, the result show that H1 was not supported, since its p-value is 0.072 (< 0.10), which is rather close to 0.05 but still requires that the hypothesis is not supported. Its path coefficient was neither that strong (β = 0.073). Hence, if an export company is following philanthropic CSR orientation it does not signify that the company’s charitable and philanthropic activity would be continued into global level: no significant positive effect was found between these variables. Nevertheless, when examining H2 that focuses on an export company’s engagement in environmental CSR orientation, including environmental sustainability activity (path ENREMAG à ENREFOC), the results are very significant, p < 0.001 with β = 0.232, which is the strongest path coefficient among the company characteristics drivers. It can be then stated that when an export company is environmental oriented it impacts positively on the global focus of its engagement in environmental sustainability activity.

Export customer orientation was suggested to affect both philanthropic and environmental GCSR intitiatives. H3a indicates that when an export company has highly developed export customer understanding capability (customer relationship) it impacts positively on global focus of an export company’s engagement with charities and philanthropic activity (path CUSOR à CHAFOC). The

results show a significant positive effect (p < 0.01 with β = 0.161). On the contrary, H3b was not supported indicating that highly developed export customer understanding capability (customer relationship) does not affect positively the global focus of an export company’s engagement with environmental sustainability activity (path CUSOR à ENREFOC).

From angle of financial constraints H4a is not supported, whereas H4b is supported. The results show that when an export company has financial constraints it does not impact negatively on global focus of an export company’s engagement with charities and philanthropic activity (path FINCO à

CHAFOC, p = 0.122 with β = -0.059), but when an export company has high level of financial constraints it impacts negatively on global focus of an export company’s environmental sustainability activity (path FINCO à ENREFOC, p < 0.05 with β = -0.122).

Institutional factors

Relation with barriers based on institutional factors among formal institutions (e.g. court systems, political instability, widespread corruption, crime and theft) on the export market and GCSR activity was not identified as statistically significant resulting in as both H5a (path INSTBAR à

CHAFOC, p = 0.817 with β = -0.008 ) and H5b (path INSTBAR à ENREFOC, p = 0.392 with β = 0.041) are not supported. Thus, institutional barriers based on formal institutional structures do not affect significantly on global philanthropic or environmental CSR activity.

When focusing on an export company’s relationships with political institutions including e.g.

national and local governments, regulatory and funding bodies, boards, banks and tax services, the reserach findings are highly significant from both philanthropic and environmental angle.

Consequently, H6a (path PINST à CHAFOC, p < 0.001 with β = 0.158) is supported indicating that the extent and frequency of the relationships between companies and political institutions on the export market, and the level of these relationships impact positively on global focus of an export company’s engagement with charities and philanthropic activity. Similarly, H6b (path PINST à

ENREFOC, p < 0.001 with β = 0.358) is supported indicating that these relationships between export companies and political institutions have a significant positive impact on global focus of a company’s environmental sustainability activity. Thus, these relationships that business managers have built with political institutions as informal institutional structures are very significant drivers of GCSR activity.

On the contrary, when informal institutional factors are changed into relationships with group of businesses and local managers the results are opposite: both H7a (path LBUS à CHAFOC, p = 0.894 with β = -0.005) and H7b (path LBUS à ENREFOC, p = 0.262 with β = -0.058) are not supported. The results actually show a small negative path coefficient in both CHAFOC and ENREFOC. Therefore, the extent and frequency of the relationships between export companies and other businesses, and the level of these relations do not impact positively on global focus of an export company’s engagement with charitable and philanthropic activity or environmental sustainability activity.

However, the perspective of local community groups brings different results. The extent and frequency of the relationships between companies and local communities on the export market (e.g.

newspaper editors, reporters, opinion leaders, sponsors or local sports clubs), and the level of these relationships impact positively on global focus of an export company’s engagement with charities and philanthropic activity. That is to say, H8a is supported (path LCOM à CHAFOC, p < 0.001 with β = 0.166). Yet, when observing H8b (path LCOM à ENREFOC, p = 0.263 with β = 0.071) the results are not statistically significant indicating that business relationships with local community groups on the export market do not have a positive significant effect on export companies’ global environmental sustainability activity.

GCSR and Global Brand Reputation

Finally, hypotheses concerning GCSR and its effect on export companies’ brand reputation reveal interesting findings between philanthropic and environmental GCSR dimensions. First, H9a (path CHAFOC à GLOBAL BRAND REPUTATION/COMPADV7, p < 0.001 with β = 0.394) is supported indicating that global focus of an export company’s engagement in charitable and philanthropic activity impacts positively on global brand reputation. These results are then highly significant with the strongest path coefficient value in this SEM .

Contrary to the highly significant results of the previous hypothesis (H9a), H9b (path ENREFOC à

GLOBAL BRAND REPUTATION/COMPADV7, p = 0.977 with β = -0.002) was not supported, which signifies that global focus of environmental sustainability activity does not have a positive significant impact on global brand reputation. The difference of probability between these two GCSR dimensions is remarkable (H9a with p < 0.001 versus H9b with p = 0.977).