Using tariff line data for the 20 largest importers, Theresa Carpenter and Andreas Lendle estimate that only 16.3% of global trade flows are given preferential treatment. (That estimate is an upper bound since they don’t observe preference utilization and therefore measure eligibility.) That seems small, and the authors conclude that “there is not much trade with high preferential margins.”

On the other hand, “on aggregate, preferences reduce the global trade-weighted average tariff from 3.2% to 2.1%.” (Their calculations don’t account for the endogeneity of trade volumes, so these 1.1 percentage points probably overstate the preferential margin. But remember that trade-weighted average tariff measures underestimate protection, so we’re taking the difference of two fuzzy numbers.) Policies equivalent to 1/2 of remaining global tariff barriers strike me as big. (Though tariff barriers are never big when compared to non-tariff barriers like labor mobility restrictions!)