Kevin Drum writes

THE STOLPER-SAMUELSON THEOREM....Over at Max's place, Josh Bivens tells us about something called the Stolper-Samuelson Theorem, which predicts that workers without a college degree always get screwed by expanded trade:

There was a big debate about this in the economics profession in the early 1990s. Not one single economist argued about the direction of trade's effect — it was universally agreed that it was negative for these workers. Some said that trade's effect was small, even very small. Some said it was large. But again, there was absolute unanimity that the net effect of trade on these workers was negative, and that trade had exacerbated inequality.

Obviously it matters a lot whether the effect is large or small or very small, but I didn't know that there was unanimity among economists that, regardless of the size of the effect, it's always negative "in absolute (not just relative) terms, and permanently (not just through tough 'transitions')."

That certainly puts a different spin on the standard thesis that free trade agreements are good for growth, doesn't it? If "growth" mean GDP growth, it's probably true. But if "growth" means growth in median wages, as I think it should, then it might not be. You learn something new every day.

I reply

The Stolper Samuelson theorem also implies that trade is good for foreign workers without college diplomas. The effects are symmetric. The USA is, by world standards, a highly educated country so college educated workers are not scarce in the USA and should receive relatively low incomes without trade. At the other extreme, manual laborers are even less scarce over there. Trade should, in theory, equalise payments to different people over the world and thus be good for manual laborers in other countries and college graduates in the USA.

Thus, if you care about equality and eliminating poverty, according to Stolper and Samuelson you should be rabidly pro free trade. The utter selfishness of looking only at the effect on Americans has no place on this blog.

As to growth, the model assumed 0 growth. If generalised, it assumes that growth is exogenous and that, in the long run, economies must always grow at the same rate no matter what the policy is. Thus Stolper and Samuelson (and I assure you the economists who agreed on the sign of the effects salaries of people without college diplomas) assumed that there was no effect on growth, or rather tried to measure effects after conditioning on economic growth.


I want to make two points. If you look closely at economic research, you usually find much less there than you expect as important questions are not answered and set aside using strong assumptions. More importantly, when you talk about trade you really must decide if non Americans count too. Looking at the USA only, it is reasonable to guess that free trade increases average real income and increases inequality. Looking at humanity as a whole, it is reasonable to guess that trade increases average real income and decreases inequality.

Ah yes that reminds me of an anecdote. One of the important figures in the 90s consensus on the sign of the effect of trade on salaries of college non graduates in the USA is Larry Katz who definitely concluded that the effect was small. Later Katz was undersecretary of labor for research (or something) and was in Mexico negotiation labor protection bits of NAFTA. Brad DeLong told me that Larry Katz told him that he saw Mexican workers protesting and had to restrain himself from reciting the Stolper Samuelson theorem to them.

Thanks to an anonymous commenter who pointed out that I failed to post this second hand anecdote on the first try.