6 RESULTS
6.10 Findings when ROA as a Dependent Variable and Summary of
Variables: Tobin’s Q and ROA
To assess the separate dependent variable (ROA) all the models were run for which results are reported in Appendix I(a), Appendix I(b), and Appendix I(c) with FSTS, FATA, and composite DoI respectively. The section following this discusses these findings with a comparison to the findings when Tobin's Q is used as a dependent variable.
The findings when DoI is measured as the composite of FSTS and OA as a dynamic capability is measured as a relative exploration are summarized in Appendix I (a). In the tables, apart from Betas, the parentheses include standard errors. Superscripts indicate the level of significance (+ p < 0.1 level, * p < 0.05 level, ** p < 0.01 level, *** p < 0.001 level). The model 1 reports the control model where lagged dependent variable, R&D intensity, and size have been used as control variables. The main effects of relative exploration on Tobin's Q (see model 2) has first the negative slope (beta= -6.34***) and the square of exploration has a positive slope (beta=2.80*).
As hypothesized, the betas for DoI and performance show non-significant effects.
However, these effects become significant in the presence of FSAs (both R&D intensity and SGA intensity). The model 3 has the first positive slope
(beta=2.69***), then negative slope (beta=-1.45***), and finally negative slope (beta=-0.91***). Similarly, the model 4 has the first positive slope (beta=5.73***), then negative slope (beta=-3.59***), and finally negative slope (beta=-0.63***). The findings suggest that there is an inverted U-shaped relationship between OA and Tobin's Q and an S-curve relationship between DoI and performance which supported the hypotheses postulated in chapter 4.
The findings when DoI is measured as the composite of FATA and OA as a dynamic capability is measured as a relative exploration are summarized in Appendix I (b). In the tables, apart from Betas, the parentheses include standard errors. Superscripts indicate the level of significance (+ p < 0.1 level, * p < 0.05 level, ** p < 0.01 level, *** p < 0.001 level). The model 1 reports the control model where lagged dependent variable, R&D intensity, and size have been used as control variables. The main effects of relative exploration on ROA (see model 2) has first positive slope (beta=3.10) and negative slope (beta=-2.48).
As hypothesized, the betas for DoI and performance show non-significant effects.
However, these effects become significant in the presence of FSAs (both R&D intensity and SGA intensity). The model 3 has the first negative slope 5.24***), then positive slope (beta=7.48***), and finally negative slope (beta=-3.24***). Similarly, the model 4 has the first negative slope (beta=-7.37***), then positive slope (beta=4.28***), and finally negative slope (beta=-1.07***). The findings suggest that there is an inverted U-shaped relationship between OA and Tobin's Q and an S-curve relationship between DoI and performance which supported the hypotheses postulated in chapter 4. The moderating effects are reported in subsequent models—model 3 reports the effect of technology differentiation, model 4 reports the effect of marketing differentiation, model 5 reports the effect of cost leadership (insignificant), and model 6 reports the effect of hybrid strategies.
The findings, when DoI is measured as the composite of FSTS and FATA and OA as a dynamic capability, is measured as a relative exploration are summarized in Appendix I (c). In the tables, apart from Betas, the parentheses include standard errors. Superscripts indicate the level of significance (+ p < 0.1 level, * p < 0.05 level, ** p < 0.01 level, *** p < 0.001 level). The model 1 reports the control model where lagged dependent variable, R&D intensity, and size have been used as control variables. The main effects of relative exploration on ROA and DoI on ROA are shown in model 2 where first there is a positive slope (beta=5.62) and a negative slope (beta=-6.93). The findings suggest that there is an inverted U-shaped relationship between OA and Tobin's Q.
As hypothesized, the betas for DoI and performance show non-significant effects.
However, these effects become significant in the presence of FSAs (both R&D intensity and SGA intensity). The model 3 has the first positive slope (beta=4.88***), then negative slope (beta=-12.41***), and finally positive slope (beta=5.74***). Similarly, the model 4 has the first positive slope (beta=13.43***), then negative slope (beta=-26.35***), and finally positive slope (beta=12.78***). The findings suggest that there is an S-curve relationship between DoI and performance which supported the hypotheses postulated in chapter 4. The moderating effects are reported in subsequent models—model 3 reports the effect of technology differentiation, model 4 reports the effect of marketing differentiation, model 5 reports the effect of cost leadership (insignificant), and model 6 reports the effect of hybrid strategies. These findings are in line with internalization theory (Buckely and Casson 1976) which suggests the role of intangibles as FSAs as conditions for the emergence of multinationals.
The findings in a nutshell for all measures of DoI and two dependent variables—
Tobin's Q and ROA are presented in Table 24. Cost leadership does not have a moderating effect on the relationship between relative exploration and Tobin's Q across measures. The concluding thoughts on these foregoing analysis and plotting supports the notion put forward by Contractor (2007) in the following paragraph adapted for authenticity and emphasis on the findings:
“the 3-stage model posits two relatively short periods (Stage 1 and Stage 3) where incremental internationalization produces a net negative effect on profits, and a longer middle Stage 2 wherein the effect of international expansion is (in net terms) positive. Overall, the theory thus posits a sigmoid M/P function. (In empirical practice, the statistically fitted curves may turn out to be U-shaped if Stages 1 and 2 predominate in the sample firms; or Inverted-U-shaped if Stages 2 and 3 are heavily represented in other company samples;
or indeed S-shaped if all three stages are well represented as was found in Contractor/Kundu/Hsu 2003, and Thomas/Eden 2004). (This is discussed and depicted later in Figure 3). One inescapable fact remains: In virtually all empirical M/P studies, whether we see a U, Inverted-U, or S-shape, there is embedded in the results a positively sloped leg over some part of the Degree of Internationalization range, thus empirically supporting the notion that international expansion is “good” over some or much of the range. (See Figure 3)” Contractor (2007:459).
The composite DoI measure with Tobin's Q supports this notion but for individual standalone measures such as FSTS and FATA have different shapes
(inverted U shape and negative inverted U-shape). I analyzed the standalone versions also to compare with similar studies using similar measures. This assertion is noteworthy compared to similar other studies and mixed findings. It is noteworthy with similar other studies because current study handles CMV, unobserved heterogeneity, and endogeneity. It is noteworthy compared to meta-analyses such as Krica et al. (2011) and Kirca et al. (2012) because it does not just aggregate the similar findings based on designs which were flawed. The interestingness of the findings on the premise of existing research gap suggests that some studies like Berry and Kaul (2016) are arguing that in the presence of endogeneity the multinationality effect evaporates. Therefore, current study fulfills the major goal of increasing the understanding of antecedents leading to sustainable performance.
Table 24. Summary of Hypotheses Testing with Relative Exploration and DoI: Comparing Three Measures of DoI for both Dependent Variables–Tobin’s Q and ROA
1Non-significant results reported in the tables for the main effects become significant in the presence of moderating factors such as FSAs and competitive strategies.
S=supported, NS=Not supported
1Non-significant results reported in the tables for the main effects become significant in the presence of moderating factors such as FSAs and competitive strategies. Therefore, it is a mixed finding.