The literature establishes a bidirectional relationship between energy consumption and participation in GVCs. ( Amri, 2019 ) examined the link between renewable and non-renewable energy consumption and trade in 72 developed and developing countries over the period 1990-2012. They find a bidirectional link between both types of energy consumption and trade in developed and developing countries. ( Ben Jebli & Ben Youssef, 2015 ) with panel cointegration techniques find the same results. For some, energy is an indispensable resource for a country seeking to integrate GVCs. For others, it is rather participation in GVCs that influences energy consumption.

The former highlights the effects of energy on GVC participation. Kan et al. (2019) argue that the prosperity of the global economy is largely achieved at the expense of the overwhelming global consumption of primary energy. Furthermore, ( Makarov et al., 2022 ) studied the relationship between countries’ energy indicators and GVC participation indices. It is found that countries with higher energy consumption and fuel-exporting countries have higher participation in downstream GVCs. Energy-importing countries and more energy-intensive countries have higher participation in upstream GVCs. Similarly, Wu et al. (2020) find that renewable energy consumption is positively related to GVC participation.

The latter highlights the effects of participation in GVCs on energy consumption. At this level, two literatures emerge. The first stipulates that participation in GVCs can reduce energy consumption ( Achabou et al., 2017 ; Khattak et al., 2015 ; Örgün, 2014 ) through new energy technologies, knowledge and improved learning and innovation capacities. This literature considers that integration in GVCs allows countries to have access to knowledge and improved learning and innovation capacities ( Pietrobelli & Rabellotti, 2010 ). The second literature states that participation in GVCs can increase energy consumption ( Kaltenegger et al., 2017 ). ( Wu & Chen, 2017 ) believe that participation in GVCs can lead to energy leakage or waste. Given the production process spread across multiple economies, energy extracted in country X can be exported to country Y for the production of goods and services that will ultimately be consumed by country Z. Using the panel vector autoregressive (PVAR) model and the system generalized method of moment (System-GMM), ( Wu et al., 2020 ) studied the dynamic causality between GVC participation, renewable energy consumption, and carbon dioxide (CO₂) between 1990 and 2015 for 172 countries. The results show that GVC participation leads to negative renewable energy consumption.

The economic literature establishes a bidirectional relationship between competitiveness and participation in GVCs. The first literature is based on the effects of competitiveness on participation in GVCs. Furthermore, competitiveness is considered as an indicator of participation in GVCs. The most competitive firms in the market are those that export the most. The second literature is based on the effects of participation in GVCs on the competitiveness of economies. This literature considers that participation in GVCs promotes competitiveness through the reduction of production factor costs, access to new varieties of quality inputs and access to technology.

Through participation in GVCs, firms can acquire high-quality foreign inputs. It can also facilitate technology transfer. The acquisition of quality inputs and technology can reduce production costs and increase productivity. Global value chains offer the opportunity to share production processes with global leading firms, which in turn share their knowledge, capital, management practices and technical assistance with local firms ( Boffa et al., 2021 ). This will contribute to improving product quality and firm competitiveness. ( Bas, 2012 ), analyzing the effect of input trade liberalization on export decisions in Argentina, finds that trade in intermediate inputs allows firms to increase their efficiency, increase their exports and improve their profitability and competitiveness.

The participation of firms in GVCs affects the quality of exports through imported inputs ( Fang et al., 2023 ). In this sense, ( Manova & Zhang, 2012 ) show that Chinese firms are relatively efficient in export markets thanks to better quality inputs. ( Giovannetti et al., 2014 ) explores the impact of belonging to a supply chain on the internationalization of firms. They find a positive and significant impact of integration into supply chains on the probability of exporting. On the other hand, another literature considers participation in GVCs as a brake on countries’ competitive policies. Thus, integration in GVCs creates a dependence of countries on the outside. When a country integrates the international production network, its production will depend not only on local resources but also on imported intermediate goods. Also, policies to protect local industries will be ineffective in GVCs. Therefore, the competitiveness of the national economy will increasingly depend on foreign partners.

To assess the joint effect of energy consumption and competitiveness on participation in global value chains (GVCs), we use the following econometric model:

$GV{C}_{it}={\text{α}}_0+{\text{α}}_{1}GV{C}_{it-1}+{\text{α}}_{2}IG{C}_{it}+{\text{α}}_{3}Energ{y}_{it}+{\text{α}}_4{X}_{it}+{\text{γ}}^i+{\text{θ}}_t\text{}+{\text{ε}}_{it}$ (1)

where $GV{C}_{it}$ is the GVC participation index, upstream and downstream participation, $IG{C}_{it}$ the overall competitiveness index, $Energ{y}_{it}$ represents the two types of energy measurement variables (i.e., energy consumption per capita and electricity production per capita) , $X_{it}$ represents the vector of control variables, namely, GDP per capita, trade openness, inward foreign direct investment, weighted average tariff applied to imported goods. To account for initial participation in GVCs, and unobserved country-specific effects, we include the dependent variable (GVC) lagged by one period and a country ${\text{θ}}_t$ and time-specific effect ${\text{γ}}^i$, in addition to the idiosyncratic error term ${\text{ε}}_{it}$.

The estimation method used in this study is the system GMM method. Three reasons prompted us to adopt the GMM method. The first contains the conditions for using the system GMM method, while the latter two show why this method is suitable for our model. First, the number of countries (N = 35) is larger than the number of years (T = 19), which helps control bias in the dynamic panel ( Baltagi, 2005 ; Bond, 2002 ; Roodman, 2009 ). Therefore, the N>T condition of the GMM method is satisfied. Second, our estimation models reveal a dynamic relationship between the explanatory variables and GVC participation, with past GVC deepening influencing present GVC deepening. However, the inclusion of the lagged dependent variable and the short observation period indicate a potential endogeneity bias. Third, the estimation method addresses the problem of reverse causality and endogeneity for all variables (dependent and explanatory). Therefore, the use of instrumental variables is necessary to correct the endogeneity bias.

Indeed, there are two types of GMM estimators applicable to dynamic panels (system GMM estimator and first-difference GMM estimator). The former is both efficient and robust to heteroscedasticity, while the latter is inefficient and lacks robustness to heteroscedasticity ( Roodman, 2009 ). However, the system GMM estimator is more suitable. On the other hand, the application of system GMMs sometimes leads to the risk of too many instruments, which can invalidate the set of instruments and distort the test results. However, the application of two-stage system GMMs can improve the efficiency of the latter. Moreover, a large number of instruments can lead to a downward bias in the two-stage system GMM standard errors due to poorly estimated weighting matrices. Windmeijer (2005) highlights this problem, suggesting a correction that accounts for variability in the weighting matrix. To minimize bias, Windmeijer’s (2005) correction is implemented in this analysis. Finally, to test the validity of the instruments, we use and present two model diagnostics: Hansen’s test for overidentification and a test for serial autocorrelation.

The data used in this study come from four sources: the UNCTAD-Eora database for GVCs, the World Economic Forum indicators for the Global Competitiveness Index (GCI), the Global Carbon Project (GCP) for CO2 emissions, and the World Development Indicators (WDI) for control variables ( Table 1 ). The dependent variable represents the index of participation in global value chains (GVCs), backward and forward participation. It is measured by the percentage of gross exports and indicates the share of foreign inputs (backward participation) and domestically produced inputs used in third-country exports (forward participation) ( De Backer & Miroudot, 2014 ; Miroudot, & Ye, 2021 ; Casella et al., 2019 ).

Our variables of interest include the Global Competitiveness Index (GCI), primary energy consumption per capita and energy intensity level. The Global Competitiveness Index (GCI), which takes scores from 1 to 7 (a higher average score means a higher degree of competitiveness), is calculated as a weighted average of several components of competitiveness grouped into 12 pillars. Energy is represented by both types of measurement variables (i.e. energy consumption per capita and energy intensity level). The list of countries considered in this work is provided in the Appendix ( Table A1 , Figures A1-A3 ).

Notes: GCP means Global Carbon Project; WDI means World Development Indicators; WEF means World Economic Forum. Source: Author.