Climate change, geography and trade agreements: A perspective of Asian bilateral trade

PLOS One | https://doi.org/10.1371/journal.pone.0320363 April 25, 2025 15 / 23 PLOS One Trade agreement, climate change and geography existed. Previous studies (e.g., Hussain et al., 2020; Friel et al., 2020) also confirm our findings, and argue that trade agreements have a remarkable impact on bilateral exports between the trading partners. In addition, WTOx provisions are positively incline towards bilateral exports. To avoid collinearity, each provision is investigated separately by not examined into single equations. However, each estimation has been measured with a dummy variable, suggesting that whether a PTA endures or not. The results demonstrate that the coefficient of capital mobility (CPM) clause in PTA has positive and significant impact bilateral exports, which implies that a 1% increase in CPM upsurges bilateral exports by 24.2% and 32.1% for exporting and importing countries models respectively. Subsequently, the coefficient of competition policy (CPP) has also significant positive impact on bilateral exports. It implies that 11.1% bilateral exports are increased due to a 1 percentage change in CPP for exporting countries model. Conversely, a 1% change in CPP upsurges bilateral exports by 9.1% for importing countries model. Labor mobility (LMB) provisions stimulate bilateral exports approximately 19%, suggesting that pro- vision of facilities for movement cause an increase in export volume. It is worth notable that migration stock variable added as control variable and isolates the certain outcomes of labor

mobility. Conversely, bilateral exports are increased about 17.3% due to a one-percentage point in LMB. The magnitude of environmental standards (EVS) has also positive effects on bilateral exports, which implies that 20.1% and 13.1% bilateral exports are due to one unit change in EVS for exporting and importing countries models respectively. The outcomes suggest that environmental standards in PTAs persuades efficiency, innovation and business effectiveness, ensuing in exports. Considering the simultaneous effect of GDP and CO2 emissions, the coefficient of inter- action term (GDP*CO2) adversely affect the bilateral exports. This implies that a 1% change in GDP*CO2 reduces bilateral exports by 3.5% in the case of domestic geographic countries. On the contrary, the outcomes for the case of partner geographic countries also confirm the negative correlation with bilateral exports; a 1.3% decrease in bilateral exports is due to a 1% change in interaction terms. This suggests that economic GDP and CO2 emissions are simul- taneously deteriorating bilateral exports. The reason may be that an increase in economic growth causes carbon dioxide emissions, resulting in extensive production and consumption levels that directly or indirectly affect bilateral exports based on geographic locations. Fur- thermore, the marginal impact of bilateral exports with respect to CO2 emissions is positive for bilateral exports. The outcomes indicate that a 1% increase in economic GDP increases the marginal impact by 14.1%, suggesting that CO2 emissions affect bilateral exports through

economic GDP over time. Robustness checks. Geographical factors could also be determined by trade, the presence of natural resources, and human economic activities. Thus, the potential reverse causality problem could be presented. However, the large investment in human and natural geography to create economic activities and discover natural resources can provide great returns to countries through production and export channels. The expectations of causality on the side, export level, and export probability can be directly influenced by geographic factors compared with other ways [70,88]. The indicators of geography do not greatly change their values concerning time, especially natural factors, reflecting that any unexpected change in geographical factors has not happened because of export flow. Thus, we report the potential endogeneity issue by exploiting the lack of geography as a new measurement of distance. Our analysis deals with heteroscedasticity (constant elasticity models) and checks the consistency of the baseline estimate with OLS and the pseudo-maximum likelihood estima- tor, as suggested by [89–91]. They explain that the lowest bias is produced by the Poisson Pseudo-Maximum Likelihood (PPML) for several heteroscedasticity patterns. The results are