By popular request, my new article at The American Conservative makes the same basic point that Karl Smith made back in November, but I think I spelled it out here very clearly to isolate Krugman’s mistake. Contrary to some of you guys in the comments, I don’t think the issue on this one was nominal vs. real, I really think it was marginal vs. inframarginal. I tried to show that by reproducing Krugman’s error in a story involving direct payment in apples. An excerpt:
In terms of our story, we can illustrate the problem by imagining that two more workers show up in town, looking for work picking apples. Through reasoning similar to that above, we conclude that in the new equilibrium, one of the newcomers goes to Smith’s orchard, the other to Jones’s, and they both get paid 10 apples per hour. However—and here is the crucial part—the original workers now only get paid 10 apples per hour, too. Because the four workers are all interchangeable, they have to get paid the same amount, and we know that the marginal product of the new guys is only 10 apples per hour.
Step back now and survey the whole picture. There are a total of 60 apples being harvested every hour among both orchards, while the four workers are only being paid 40 apples total. Thus the orchard owners are enjoying a “surplus” of 10 apples each, for every hour the laborers are at work. Thus, it is not true to say that the workers in this hypothetical world are “getting paid what they put in,” in the sense that Krugman means. Yet even so, the workers are being paid their marginal product, according to textbook price theory.