Preferential Trading in Agriculture: New Insights from a Structural Gravity Analysis and Machine Learning
Preferential Trade Agreements (PTAs) may yield benefits in agricultural trade liberalization. Still, empirical findings quantifying these benefits are ambiguous, partly because of the complicated and diverse provisions that may promote or hinder agricultural trade. Modern PTAs contain hundreds of provisions in addition to tariff reductions in areas as diverse as competition policy, quantitative restrictions, Technical Barriers to Trade (TBT), Sanitary and Phytosanitary (SPS), or intellectual property rights. Existing research has struggled with overfitting and severe multicollinearity problems when trying to estimate the effects of these provisions on trade flows. This paper uses the plug-in Lasso, a machine learning approach, to select the most related provisions and quantify their impact on agricultural trade flows. The results show that PTA provisions related to anti-discriminatory policies, SPS and TBT measures, and geographical indication promoted agricultural trade between partner countries. However, these effects are inconsistent over different income levels, calling into question the current PTA regime and its development directions for sustainable integration of global agriculture.
*The latest version can be found here!