CHAPTER 6: DATA ANALYSIS AND DISCUSSION
6.2. Empirical results – testing of reciprocal correlations between intellectual capital
Results in Table 6.2 support hypotheses H1a – H1c for the intellectual capital variables in the first model. These results show how human capital link to relational capital (β = 0.552; p value < 0.01) and how relational capital links to structural capital (β = 0.256 – 260; p value < 0.01) for all regression models with four types of endogenous variables in terms of corporate performance. All above paths connections are higher than 0.2 to be considered relevant to warrant managerial attention (Chin, 1998a).
Human capital exerts a pronounced (β = 0.371 – 0.376) and significantly (p value <
0.001) direct effect on structural capital. The next step focuses on the significance of the indirect effects from human capital to structural capital. As determined in Table 6.2, the indirect relationship from human capital to structural capital is moderate (around 0.141) and statistically significant (p value < 0.1). Since the direct and indirect effects are both positive and significant, it is concluded that relational capital represents complementary mediation of the relationship from human capital to structural capital. It is a partial mediation as human capital have a direct correlation with structural capital and relational capital partially mediates this correlation.
To assess the predictive power of each of the structural model, the adjusted R2 values of each endogenous construct must be considered. They are follows; relational capital (30.1 – 30.5 percent in all four models tested), structural capital (30.0 – 30.8 percent in all instances). The adjusted R2 value of around 0.3 for the endogenous constructs of relational capital and structural capital can be described as moderate to measure predictive accuracy (Hair Jr & Hult, 2016).
In general, the results of this study re-affirm that human, structural and relational capital reciprocally circulate and affect each other. Human capital has the positively direct effects on relational capital and structural capital, as well relational capital positively affects structural capital. In addition, for the relationship between human capital and structural capital, relational capital serves as a complementary mediator. Higher levels of human capital efficiency increase structural capital efficiency directly but also increase relational capital efficiency, which in turn leads to structural capital efficiency. The results generated are very consistent, within a wide range and support previous IC-based research.
Table 6.2. Summary of the results of the first hypothesis testing
H1
Models with endogenous
construct
Structural path relationship
Expect- ed sign
Direct effect
t value
Signi- ficance
(p <
0.1)?
Indirect effect
t value
Signi- ficance
(p <
0.1)?
Total effects
t value
Signi- ficance
(p <
0.1)?
Type of mediation
Test result
H1a ATO RCE HCE + 0.552 8.618 *** – – – 0.552 8.618 *** – Support
INVEFF + 0.552 8.372 *** – – – 0.552 8.372 *** – Support
ROE + 0.552 8.373 *** – – – 0.552 8.373 *** – Support
TOBINQ + 0.552 8.819 *** – – – 0.552 8.819 *** – Support
H1b ATO SCE HCE + 0.371 2.679 *** 0.141 1.992 ** 0.512 6.120 *** Partial Support
INVEFF + 0.376 2.912 *** 0.143 2.116 ** 0.519 6.576 *** Partial Support
ROE + 0.374 2.778 *** 0.142 1.940 * 0.516 6.577 *** Partial Support
TOBINQ + 0.371 2.808 *** 0.141 1.977 ** 0.513 6.455 *** Partial Support
H1c ATO SCE RCE + 0.256 2.024 ** – – – 0.256 2.024 ** – Support
INVEFF + 0.260 2.184 ** – – – 0.260 2.184 ** – Support
ROE + 0.258 2.011 ** – – – 0.258 2.011 ** – Support
TOBINQ + 0.256 2.017 ** – – – 0.256 2.017 ** – Support
Note: Significant at: *10, **5 and ***1 percent levels (2-tailed)
Source: Calculated by the author in SmartPLS 3.1
6.3. Empirical results – testing of the correlations between intellectual capital components and strategic management accounting practices (H2)
Table 6.3 summarizes the results examining SMA practices that are expected to be influenced by IC components’ intensity. H2a, H2b, H2c are accepted in more details below.
In terms of the SMA-related results, in Table 6.3, the direct impact of human capital on SMA practices of respondent firms are considered. All β-path coefficients are positive and validate in statistics at 5 percent level in four research models about corporate performance as follows; asset turnover (β = 0.081, p value = 0.019), investment efficiency (β = 0.085, p value = 0.010), return on equity (β = 0.082, p value = 0.016), Tobin q (β = 0.081, p value = 0.018). Although these relationships are significant, the absolute value size of a path coefficient may be too small to warrant managerial attention. Nonetheless, under the mediating role of relational and structural capital, the path coefficient value of 0.552 is enough substantial to generalize a managerial implication, showing H2a is supported here. Following the mediation analysis procedure in Table 6.3, relational and structural capital serves as complementary mediators. Higher levels of human capital increase the level of SMA practices directly but also increase relational capital, which in turn leads to structural capital, which afterwards causes a rise in the level of SMA practices.
This result shows that firms which rely on human capital are likely to consider SMA practices such as balanced scorecard, non-financial measures to be leading indicators that provides sophisticated information to make strategic decisions. One possible reason for the acceptance of SMA is that the better internal knowledge within firms owing higher human capital is the prerequisite conditions to move into an advanced management accounting system where the capability of management accountants is able to utilize more beneficial outputs to add more firm values.
Data in Table 6.3 shows that t-statistics of four paths around 2.369 to 2.456 are greater than 1.96, which means the β-path coefficients are significant at 5 percent level and indicates a significant correlation between structural capital and SMA practices the four testing models. This is a direct correlation, showing H2b is supported. This result is supported by the prior studies indicating that the selection of strategic management accounting practices is not dictated by technical criteria or economic efficiency alone, but also a cultural and political process that concerns legitimacy, power and organization
structure system (Collier, 2001; M. Hussain & Hoque, 2002), are called structural capital.
Firms chasing with strategy orientation are willing to develop efficient routines and processes, marked by structural capital, as the infrastructure to collect and transform customer-focused and market-driven data into the inputs that are then processed by SMA techniques to make managerial decisions. For example, firms with structural capital including customer databank have a convenient condition to improve SMA practices in terms of customer accounting. As Anderson (2007) who is discussing strategic management accounting, much of information required is occurring outside the accounting function, hence it requires an effective infrastructure to implement SMA.
Turning to relational capital, the direct relationship with SMA practices is significantly positive (β-path coefficients around 0.722 to 0.725) in all four research models, indicating that employees with more connections with externals and internals have more better opportunities to access different resources to enhance SMA practices to some extents because much of the inputs used in SMA techniques is required to occur outside of the accounting function. Therefore, H2c is supported. More profoundly analysis, in the indirect relationship with SMA practices via the mediating role of structural capital, the empirical t values of the indirect effect (0.034 to 0.037) for the RCE → SMA relationship is around 1.519 to 1.558, yielding p value higher than 10 percent level, thus mediation may not exist at all. It means that the information obtained by external relations are directly applied into SMA techniques by each individual, but it is not capable of transforming into explicit knowledge to be disseminated throughout the formal organization system.
With relations to the predictive power of each model, the adjusted R2 value of SMA practices construct is enough substantial to conclude the fitness of four structural models.
Specifically, they are as follows; ATO (0.714), INVEFF (0.711), ROE (0.713) and TOBINQ (0.718) (see Appendix 24 to 27). In summary, the positive relationships between each of IC components and SMA practices are all accepted with high coefficients to alarm managerial attention. In addition, this study also detects the mediating role of structural capital to mediate the association between human capital and SMA practices. These results confirm that the higher value of each IC components firms own, the more appropriately firms utilise the infrastructure of IC components in order to implement an advanced management accounting like SMA approach.
Table 6.3. Summary of the results of the second hypothesis testing
H2
Models with endogenous
construct
Structural path relationship
Expect- ed sign
Direct effect
t value
Signi- ficance
(p <
0.1)?
Indirect effect
t value
Signi- ficance
(p <
0.1)?
Total effects
t value
Signi- ficance
(p <
0.1)?
Type of mediation
Test result
H2a ATO SMA HCE + 0.081 2.350 ** 0.472 8.056 *** 0.552 10.244 *** Partial Support
INVEFF + 0.085 2.587 ** 0.467 8.081 *** 0.552 10.109 *** Partial Support
ROE + 0.082 2.427 ** 0.470 8.033 *** 0.552 10.463 *** Partial Support
TOBINQ + 0.081 2.378 ** 0.471 7.946 *** 0.552 10.470 *** Partial Support
H2b ATO SMA SCE + 0.143 2.369 ** – – – 0.143 2.369 ** – Support
INVEFF + 0.129 2.419 ** – – – 0.129 2.419 ** – Support
ROE + 0.137 2.409 ** – – – 0.137 2.409 ** – Support
TOBINQ + 0.142 2.456 ** – – – 0.142 2.456 ** – Support
H2c ATO SMA RCE + 0.722 18.807 *** 0.037 1.538 No 0.758 21.604 *** No Support
INVEFF + 0.725 18.968 *** 0.034 1.558 No 0.759 21.557 *** No Support
ROE + 0.723 19.120 *** 0.035 1.538 No 0.758 22.068 *** No Support
TOBINQ + 0.722 17.242 *** 0.036 1.519 No 0.758 20.771 *** No Support
Note: Significant at: *10, **5 and ***1 percent levels (2-tailed)
Source: Calculated by the author in SmartPLS 3.1