Reproducibility package for Beneath the Surface: Deep Trade Agreements and the Potential of a Deeper AfCFTA

This paper examines what makes some trade agreements more effective at promoting trade than others, and applies this question to the case of the African continent. To address it, we use a clustering algorithm to classify worldwide trade agreements into three categories and estimate their trade effects within a structural gravity framework. This allows us to rank PTA clusters according to their estimated trade effects as Shallow, Medium, and Deep.
We then use feature-attribution methods and counterfactual policy combinations to identify the provisions and policy configurations most strongly associated with Deep classification, and apply this approach to the African Continental Free Trade Agreement (AfCFTA). Although the AfCFTA, which entered into force in 2019, is intended to harmonize existing regional agreements into a single continent-wide framework, it is classified as Medium in our baseline. Our general-equilibrium counterfactual analysis shows that, if implemented as a Deep agreement, the AfCFTA would generate substantial additional trade gains, along with an average GDP gain of 0.7% for the African economies in our sample. The classification analysis further indicates that tighter disciplines on trade-defense instruments, stronger legal enforceability, and visa-related provisions are among the policy dimensions most closely associated with Deep agreements.