“…All models are subjected to multiple regression analysis using the fixed-effect method, which was chosen based on the panel data model selection test (Gujarati and Porter, 2009;Wooldridge, 2016). The fixed-effect method yields the most efficient estimation, as shown in the results in Appendix 2 (Gujarati and Porter, 2009;Arora and Sharma, 2016;Wooldridge, 2016;Mayur and Saravanan, 2017;Santos et al, 2019;Barros et al, 2020;Tahir et al, 2020;Tran, 2021;Khursheed and Sheikh, 2022;Noor et al, 2022). In addition, we also applied a robust standard error to all the models (Koester et al, 2016) and to all the regression analyses because there were pieces of evidence that the residual of the models suffered from non-normality, heteroscedasticity and autocorrelation (in Appendices 3 and 4).…”