Here is his new paper, but start first with this Mark Thoma summary, and two graphs from Brad DeLong. The main point is that some U.S. imports may be more labor-intensive and less skill-intensive than previous classifications had indicated. Here is one key paragraph (p.20):
But what are we to make of NAICS 334, Computer and Electronic Products? In U.S. data it ranks as the most skill-intensive of industries, yet it is also an industry in which more than three-quarters of imports come from developing countries, especially China.
If these sectors count as "importing labor," we can find that trade is creating more downward pressures on U.S. wages than we had thought.
I don’t think Krugman is quite right to claim: "the apparent sophistication of imports from developing countries is in large part a statistical illusion." I would sooner say that China and some other Asian countries are specializing in new (and sophisticated) techniques of cooperation, made possible by long-term historical investments in human capital and social norms. At least in certain sectors, they are combining complementary labor inputs, with complementary capital inputs, more effectively than before; it’s hard to explain that change in the impoverished vocabulary of the substitution-obsessed Heckscher-Ohlin model. The skill is in the combination not in the people themselves. "Capital-intensive" vs. "labor-intensive" or "skilled" vs. "unskilled" are not simple either/or questions.
So I think Krugman is confused on the semantics, but in the final analysis this perspective supports and perhaps even strengthens his point. If the paper looked at wages and employment in northern Mexico, following the move of China onto the world stage, the revisionist conclusions would fall more easily into place. On one hand Chinese competition hit Mexico (a home of unskilled but relatively cooperative labor) very hard; on the other hand northern Mexico responded successfully by moving up the value chain rather than by folding and losing. Both developments suggest that the Chinese competition is not just a simple example of skill-intensive labor.
You might say: "Chinese competition with northern Mexican textiles and plastics isn’t at all like Chinese competition in the hi-tech sector." I would sooner say: "The Chinese are applying common production techniques across the board." Of course the phenomenal Chinese levels of both personal savings and labor migration to urban areas also support the overall interpretation of complementarity.
Addendum: Here is a good paper on the changing nature of Chinese exports.