Nonlinear effects of global value chain participation on environmental quality in developing countries

Despite the growing interest in the environmental consequences of global value chain (GVC) integration, the potential nonlinearity of this relationship remains unexplored. This study addresses this gap by examining the regime-dependent relationship between GVC participation and CO 2 emissions across 41 developing countries over the period 2007–2020. Unlike the majority of extant studies, which assume linearity, this paper employs a Panel Smooth Transition Regression (PSTR) model, complemented by linear and dynamic System GMM estimations, to explicitly account for nonlinearities. The findings provide substantial evidence that the environmental impact of GVC integration is not uniform across levels of participation. Specifically, GVC participation significantly reduces CO 2 emissions at low-to-moderate levels of integration; however, this emission-reducing effect weakens and loses statistical significance once GVC participation crosses the estimated threshold, as expansion-driven scale effects come to offset the initial efficiency gains. This pattern therefore provides only conditional support for the pollution halo hypothesis, confined to the earlier stages of integration. These findings imply that the environmental benefits of GVC participation are concentrated in its earlier stages rather than emerging only once integration becomes substantial. Additionally, energy intensity and economic globalization are found to exacerbate emissions, while digitalization plays a modest mitigating role. Overall, this study contributes to the existing academic literature by demonstrating that the GVC-environment nexus is inherently nonlinear and influenced by structural factors. It also offers policy-relevant evidence that upgrading within global value chains, rather than merely increasing participation, is more likely to support sustainable environmental outcomes. These findings underscore the importance of accounting for threshold effects in future research.