Haosen Ge (Princeton University)
Abstract: Existing studies show that regulations are one of the major barriers to global economy. Nonetheless, identifying and measuring regulatory barriers remains a challenging task for scholars. I propose a novel approach to quantify the level of regulatory barriers at the industry level. Utilizing information from annual reports of publicly listed companies in the U.S., I can identify regulatory barriers encountered by business practitioners all around the world. The reported barriers are identified by first using a cutting-edge neural language model trained on a hand-coded training set. The final prediction accuracy on the test set is around 90%. Then, I use a dynamic item response theory model to estimate barrier level of major industries in 122 countries while controlling for various channels of confounding. The estimated barrier level returned by this approach should be much less likely to be contaminated by unobserved confounders such as WTO politics. Therefore, it is well-suited for future political science research.