Dani Rodrik suspects that they are small than Josh Bivens thinks they are:
Dani Rodrik's weblog: The pains from trade: The workhorse model of international trade (the 2x2 Heckscher-Ohlin model) has very stark implications for the effect of trade with poor, labor-abundant countries. Low-skilled workers in rich countries (read the U.S.) must end up as losers--not in relative terms, but in absolute terms. Moreover, the larger the overall gains from trade, the bigger must this adverse distributional effect be. In that world, it is inconsistent to claim there are large gains from globalization while downplaying the distributional impacts. Which is why many economists teach the model in their classrooms, but shift to other, more complicated models when they engage in the public debate about the effect of trade on wages.
A recent paper by Josh Bivens carries out a quantitative simulation of the basic 2x2 model which suggests that the increase in U.S. trade with the developing countries between 1995 and 2006 would have reduced labor earnings by 4 percent while increasing the payment for skills by 3 percent, for a 7% increase in the differential altogether. This is an interesting exercise, to be compared to that undertaken by my colleague Robert Lawrence. The latter gives much less of a role to trade, in large part because it finds there is little reduction in real earnings once adjustments for productivity, prices and benefits are made.
One clear difference between the two perspectives is the extent to which one thinks of trade with developing countries as competing directly with U.S. production. Lawrence says that developing country exports hardly displace any domestic production anymore because much of that activity has already shut down. So whatever adverse impact may have existed in the 1980s, the current situation is much more benign. (But of course, less impact on productive re-allocation means fewer overall gains from trade too!). In Bivens' world (and that of the standard 2x2 model), by contrast, head-to-head competition is critical in driving the distributional effects.
I am on Dani's side, only more so. Two additional points are important, I think:
-
For competition to be head-to-head, the two countries have to be making very similar goods with similar production processes. Hand-spinners in Pakistan don't compete with labor here in the United States but with the capital embodied in our large automated spinning mills.
-
What trade does to our distribution of income can be undone by normal domestic redistributionist policies. The right way to deal with the issue is to (a) maximize the third world's ability to take advantage of our demand to spur its own growth, and (b) use domestic redistribution here to compensate for any adverse distributional
