Page 148 - Ebook HTKH 2024

had the best model fit. In detail, most of the fitness index meets the acceptable
benchmark: CFI=0.867, TLI=0.829, GFI=0.912, NFI= 0.872 (above 0.8) (Bauldry,

2015; Brett, 2010). This is the expected result consistent with literature review.
Meanwhile, ROC is a profitability ratio that measures a company’s profit using
both its debt and equity capital. Therefore, a latent variable combined of ROC and ROE
is more suitable for comprehensively representing a company’s performance.
The positive impacts of economic distance on MNE’s global subsidiaries’
performance have been confirmed within the model. Additionally, the negative impacts
of cultural distance and the positive impact of legal institutional distance on MNE’s
global subsidiaries’ performance are endorsed (p<0.001). As such, hypotheses H1

(Home-host country cultural distance has significant negative impacts on MNE
subsidiary performance within manufacturing industries), H2 (Home-host country
institutional distance has significant positive impacts on MNE subsidiary performance
within manufacturing industries) and H3 (Home-host country economic distance has
significant positive impacts on MNE subsidiary performance within manufacturing
industries) are accepted.

In terms of control variables, the empirical results also revealed that type of
industry (p=0.000) and firm size (p=0.009) have a positive influence on performance
of MNE’s global subsidiaries; however, experience of doing business at a local market
(age) and performance of MNE’s global subsidiaries appears to be not supported
(p=0.9>0.05)
4.3. Robustness check
To check if the results are robust, we examine an alternative model. Some

estimated paths have been accidentally added into the current model to create an
alternative model for Robustness check in this technique. In detail, while all the
constructs and items associated to each construct are unchangeable, new paths have
been formed between cultural distance (CUL) and economic distance (ECO); legal
institutional distance (INS) and economic distance (ECO), cultural distance (CUL) and
legal institutional distance (INS).

The result shows that the suggested model has a better model fix than the new
model (on every goodness-of-fit index) and the concerned relationships are still
acceptable in the new model with the same tendency pointed out by the suggested
model. To conclude, the suggested model is the best model to represent the relationships
among the research contexts.
Table 1. The regression path coefficient of the alternative model
Path coefficient p Result
CUL → PER -2.475 0.000 Supported
INS → PER 1.249 0.000 Supported

ECO → PER 1.842 0.000 Supported
Industry → PER 1.609 0.000 Supported
LTE → PER 0.376 0.006 Supported

140