Quantifying Trade Law: New Perspectives on the Services Trade Restrictiveness Index

Abstract

Measuring the restrictiveness of applied services trade policies is far from straightforward. In addition to identifying policy measures of interest, there is also the problem of weighting and aggregating them into Services Trade Restrictiveness Indices (STRIs). This paper tackles that problem, which has traditionally been solved by using weights determined by analysis or expert judgment. The approach here is novel: a machine learning algorithm is used to determine the weights that have the best predictive power for bilateral trade costs. This alternative approach produces an index with significantly greater explanatory power for bilateral trade than the Organisation for Economic Cooperation and Development (OECD) STRI, using the banking sector as an example. A quantitative simulation shows that the alternative methodology makes a major difference in policy terms: the global impact of a 10% reduction in the restrictiveness of applied services policies is about ten times larger than the estimated impact using the OECD’s STRI.