4.1. The Measurement Model Assessment
This study tested the validity and reliability of the measurements and the presence of common method bias. We tested Cronbach’s alpha for construct reliability [
44].
Table 2 shows that the construct reliability for all variables, Cronbach’s alpha, is above 0.8. The cut-off is below 0.7. In addition, exploratory factor analysis was conducted using the Varimax rotation method to confirm the constructs of measurement items, and the factor loadings were found to be 0.85 or higher, only for the factors with an eigenvalue of 2.475 or higher, through factor analysis. As shown in
Table 3, all variables were comparable to the composite reliability.
In addition, we tested confirmatory factor analysis (CFA) to avoid low factor loading.
Table 3 shows the results of CFA. The overall model fit is found to be X
2 = 510.09 (df = 303), GFI = 0.890, AGFI = 0.862, CFI = 0.976, NFI = 0.944, TLI = 0.972, and RMSEA = 0.048. In addition, the t-values for factor loadings of constructs were found to be 14.0 or higher, which indicates that the measurement items for innovativeness, proactiveness, risk taking, competitive aggressiveness, autonomy, firm performance, and non-firm performance were valid.
To evaluate whether the measurement items were representative of this study, the average variance extracted (AVE) and the conceptual reliability were analyzed. The conceptual reliability of the study unit exceeded the recommended standard of 0.80, and the AVE exceeded the recommended standard of 0.50. As illustrated in
Table 4, these values are suitable. As shown in
Table 4, the items of this study were found to be representative of the research units. In addition, these values supported the convergent validity of the composite in the model [
45].
Table 5 shows that the discriminant validity is acceptable because the square root of the AVE values is greater than the square of the correlation coefficient between the composite and all other variables in this model [
45]. All composites in the model are statistically verified, and the measures are different from each other. Therefore, discriminant validity is valid.
4.3. Results of SEM Analysis
The relationship between entrepreneurial orientation and firm performance was examined using path analysis. Additionally, to confirm the findings of this study, SEM was conducted to examine the relationship between entrepreneurial orientation and firm performance.
Table 7 and
Figure 3 show the results of the SEM. First, the goodness-of-fit between the data and the model was X
2 = 537.16 (df = 304), GFI = 0.885, AGFI = 0.856, CFI = 0.973, NFI = 0.941, TLI = 0.973, and RMSEA = 0.051.
The path coefficient value innovativeness to firm performance was found to be −0.065 (p > 0.05) and not statistically significant. The path coefficient value from proactiveness to firm performance was found to be −0.067 (p > 0.05) and not statistically significant.
The path coefficient value from risk taking to firm performance was found to be 0.155 (p < 0.05) and statistically significant. The path coefficient value from competitive aggressiveness to firm performance was found to be 0.095 (p > 0.05) and not statistically significant. The path coefficient value from autonomy to firm performance was found to be 0.293 (p < 0.01) and statistically significant.
The path coefficient value from innovativeness to non-firm performance was found to be 0.011 (p > 0.05) and statistically not significant. The path coefficient value from proactiveness to non-financial firm performance was found to be 0.055 (p < 0.01) and statistically significant.
The path coefficient value from risk taking to non-financial firm performance was found to be −0.028 (p > 0.05) and not statistically significant. The path coefficient value from competitive to non-financial firm performance was found to be 0.057 (p > 0.05) and not statistically significant.
The path coefficient value from autonomy to non-financial firm performance was found to be 0.077 (p < 0.05) and statistically significant. In addition, autonomy has a positive effect on a firm’s non-financial performance. Autonomy has a direct impact on financial and non-financial performance.
This factor indicates that employees and owners participate more actively to improve the financial and non-financial performance of their businesses. Therefore, additional analysis confirmed that there was no significant difference in the research results obtained by using path analysis and SEM.
4.4. Moderating Effect of Creating Shared Value on the Relationship between Entrepreneurial Orientation and Firm Performance
This study divided the samples into CSV (high) and CSV (low) based on the mean (4.528) of the scores on the CSV scales.
Table 8 shows the result of group classification. Among the 294 total respondents, the number for CSV (high) was 140, and the number for CSV (low) was 154.
This research conducted an X2 difference test to find the moderating effect of CSV (high, low). To analyze the differences in the path coefficients indicating the causal relationship among innovativeness, proactiveness, risk taking, competitive aggressiveness, autonomy, financial firm performance, and non-financial firm performance according to CSV (high, low), the causal relationship path among the baseline models, such as innovativeness, proactiveness, risk taking, competitive aggressiveness, autonomy, financial firm performance, non-financial firm performance, was selected for the free model. For the restricted model, a model that restricted the path coefficient of CSV (high) and CSV (low) to be the same was selected. Next, the change in the X2 value was measured between the two models.
Table 9 and
Table 10 show the results of differences by CSV classification and comparisons, such as the high and low groups, respectively. In the relationship test between innovativeness and firm performance, the difference between the two models was found to be significant (ΔX
2 = 7.894 > ΔX
20.05 (1) = 3.84). Thus, it is confirmed that a moderator effect of CSV (high, low) occurs in the relationship between innovativeness and a firm’s financial performance as economic benefits. In addition, the path coefficient value of CSV (low) was not significant, while that of CSV (high) was significant, except for proactiveness in FP and risk taking in NFP. Almost all high CSV groups were better than low CSV groups. CSV (high) was statistically related to positive firm performance, but CSV (low) was not statistically significant. Thus, H6a is supported since there are statistically significant positive effects on firm performance.
In the relationship test between proactiveness and firm performance, the difference between the two models was found to be not significant (ΔX2 = 0.884 < ΔX20.05 (1) = 3.84). It is not confirmed that a moderator effect of CSV (high, low) occurs in the relationship between proactiveness and a firm’s financial performance. Thus, H7a is not statistically supported.
In the relationship test between risk taking and firm performance, the difference between the two models was found to be significant (ΔX2 = 4.220 > ΔX20.05 (1) = 3.84). Thus, it is confirmed that a moderator effect of CSV (high, low) occurs in the relationship between risk taking and a firm’s financial performance. In addition, the path coefficient value of CSV (low) was not significant, while that of CSV (high) was significant; thus, H8a is supported at the 0.1 statistical level.
In the relationship test between competitive aggressiveness and firm performance, the difference between the two models was found to be significant (ΔX2 = 5.095 > ΔX20.05 (1) = 3.84). Thus, it is confirmed that a moderator effect of CSV (high, low) occurs in the relationship between competitive aggressiveness and a firm’s financial performance; therefore, H9a is supported. In addition, the path coefficient value of CSV (low) was not significant, while that of CSV (high) was positively significant. High CSV was positively related to firm financial performance at the 0.05 statistical level.
In the relationship test between autonomy and firm performance, the difference between the two models was found to be significant (ΔX2 = 7.475 > ΔX20.05 (1) = 3.84). Thus, it is confirmed that a moderator effect of CSV (high, low) occurs in the relationship between competitive aggressiveness and a firm’s financial performance. In addition, while the path coefficient value of CSV (low) was not significant, that of CSV (high) was significant. H10a for high CSV was supported because there was a statistically significant positive effect on financial performance.
In the relationship test between innovativeness and a firm’s non-financial performance as social benefits, the difference between the two models was found to be significant (ΔX2 = 5.406 > ΔX20.05 (1) = 3.84). Thus, it is confirmed that a moderator effect of CSV (high, low) occurs in the relationship between innovativeness and a firm’s non-financial performance. In addition, although the path coefficient value of CSV (low) was not significant, that of CSV (high) was significant. This means H6b is supported because there are statistically significant positive effects on non-financial performance.
In the relationship test between proactiveness and a firm’s non-financial performance, the difference between the two models was found to be significant (ΔX2 = 13.452 > ΔX20.05 (1) = 3.84). Thus, it is confirmed that a moderator effect of CSV (high, low) occurs in the relationship between risk taking and a firm’s non-financial performance. In addition, while the path coefficient value of CSV (low) was not significant, that of CSV (high) was significant.
In the relationship test between risk taking and a firm’s non-financial performance, the difference between the two models was found to be not significant (ΔX2 = 0.731 < ΔX20.05 (1) = 3.84). Thus, it is not confirmed that a moderator effect of CSV (high, low) occurs in the relationship between proactiveness and a firm’s non-financial performance; therefore, H7b is statistically supported.
In the relationship test between risk taking and firm performance, the difference between the high and low CSV groups was found to be not significant (ΔX2 = 0.888 < ΔX20.05 (1) = 3.84). It is not confirmed that a moderator effect of CSV (high, low) occurs in the relationship between risk taking and a firm’s non-financial performance. Thus, H8b is not statistically supported. However, CSV (high) is a positive direction.
In the relationship test between competitive aggressiveness and a firm’s non-financial performance, the difference between the two models was found to be significant (ΔX2 = 4.872 > ΔX20.05 (1) = 3.84). Thus, it is confirmed that a moderator effect of CSV (high, low) occurs in the relationship between competitive aggressiveness and a firm’s non-financial performance. In addition, the path coefficient value of CSV (low) was not significant, while that of CSV (high) was significant. H9b for CSV (high) is supported because there are statistically significant positive effects on a firm’s non-financial performance.
In the relationship test between autonomy and a firm’s non-financial performance, the difference between the two models was found to be significant (ΔX2 = 6.094 > ΔX20.05 (1) = 3.84). Thus, it is confirmed that a moderator effect of CSV (high, low) occurs in the relationship between competitive aggressiveness and a firm’s non-financial performance. In addition, the path coefficient value of CSV (low) was not significant, while that of CSV (high) was significant. Therefore, H10b is statistically supported.