Another interesting consequence of using the national distribution of income framework to analyze international trade (as Pettis and others do) is that trading with countries with a higher household share of national income (and hence higher consumption) is better than trading with countries with low a household share of income, because the high consumption countries can import more back from you. So for instance, free trade with Mexico--say, building low labor cost factories there--will put those US laborers out of work, but they can get jobs in different industries and sell those products to Mexico where consumption is 66% of GDP so they buy a lot, too.

But free trade with China, where consumption is just 36% of GDP, means they cannot buy very much back from us, and so a lot of the money we send to China ends up in the hands of businesses and the government which they use to buy dollars to keep their currency weak, rather than buying goods which then employs the Americans who make those goods, as would happen if we traded with Mexico.

Yes we have a deficit with Mexico in spite of that, but that's because of the convenience of the shared border, where surplus countries like China or Japan export intermediate goods to Mexico which then go into the US. Mexico on the whole runs a trade deficit, meaning it contributes to global demand rather than shrinking global demand as China's economic structure does and reduces employment in the US via buying dollars with all the economic impact I just described.