Service trade network structure and its determinants in the Belt and Road based on the temporal exponential random graph model

Service trade has become an important engine of world economic growth. Using complex network method and temporal exponential random graph models, we constructed the Belt and Road service trade networks (BR‐STNs) and global service trade networks (G‐STNs) in 2010–2017, systematically analysed the features of BR‐STNs compared with G‐STNs, and tested the factors affecting their evolution. We find that (a) the scale of BR‐STNs is small, but potential is huge. The BR‐STNs present the features of small‐world, larger degree disassortativity, lower reciprocity, and smaller heterogeneity than G‐STNs. (b) The BR‐STNs are divided into three communities and their pattern is more stable than G‐STNs. (c) Luxembourg, Italy and Singapore are the core in BR‐STNs, but there is a gap compared with the USA, the UK and Germany. (d) Reciprocity, structural embeddedness, the clustering effect and stability play important roles in the evolution of BR‐STNs and these effects increased after the proposal of the BRI. Economy‐pairs with similar trade openness, urbanization rates and quality of institution, and with different GDP and per capita GDP are more likely to trade. Countries with higher GDP, trade openness or quality of institution are active. Sharing a language, religion, colonial relationship or border, signing regional trade agreements and using a common currency facilitate service trade between countries in BR.