How Big Are (or Aren't) the Gains From Trade

One of the things trade economists have traditionally enjoyed doing to hammer home the gains from trade is to cite seemingly outrageous numbers about the positive impacts of trade. For instance, there is this 2003 report from the US International Trade Commission that estimates the net costs from trade restrictions for 1990 to be about $170,000 per job saved (but yes, jobs were saved by protection!). Or, there are these calculations that take a more nuanced approach sector-by-sector:

Industry	Annual loss to economy from barriers = Cost	Net employment loss if barrier is removed = jobs "saved"	Annual cost PER JOB SAVED
Textile and apparel	$ 10.04 billion	55,000	$ 182,545
Maritime transport	$ 2.79 billion	2,450	$ 1,138,775
Dairy	$ 1.01 billion	2,083	$ 484,878
Motor vehicles	$ 710 million	3,400	$ 208,824
Sugar	$ 661 million	1,694	$ 390,200
Meat	$ 185 million	100	$ 1,850,000
Steel mills	$ 162 million	1,265	$ 128,063
Nonrubber footwear	$ 147 million	1,316	$ 111,702

From McConnell and Brue, 2002

Source: compiled from United States International Trade Commission data released December 1995. Data are for 1993.

New research has called these claims into question, at least somewhat. Jonathan Dingel summarizes some of the new research on trade (which includes contributions by Arkolakis et al., Ossa, and the opinions of Prescott) using more advanced trade models here. Arkolakis, et al. find that the gains from trade in these "new models" is "not much". But, these gains depend on the estimation of elasticity coefficients, which may vary by industry. When these coefficients are estimated separately for different industries, the gains from trade increase by about 6 fold, to somewhere between 24 and 42% of GDP.
Thus, the gains from trade may well be large after all.