Geographical and economic characteristics help to predict Regional Trade Agreements (RTAs), and we here argue that Global Value Chains (GVCs) are one of these predictors. We estimate the time-varying probability that a country pair share a trade agreement, identify pairs who should sign an RTA and have not yet done so, and among these potential RTAs those mostly driven by the participation of the country pairs in GVCs. Using a General-Equilibrium model of an endowment economy, we construct counterfactuals where the geography of RTAs is redesigned according to our predictions and assess the trade and welfare consequences of phasing-in such pending RTAs. We last simulate the impact of a slowdown of GVCs on trade and welfare.