Read the whole thing. Here is the conclusion:
We hope that readers will take from our paper three main conclusions about the recent trends in U.S. real and relative incomes. First, to date there is little evidence that globalization through the classic channel of international trade in goods, intermediates, and services has been raising inequality between more-skilled goods, intermediates, and services has been raising inequality between more-skilled and less-skilled workers. Second, there is at least suggestive evidence that globalization has been boosting the real and relative earnings of superstars. The usual trade mechanisms probably have not done this, but other globalization channels—in particular, the combination of greater tradability of services and larger market sizes abroad—may be playing an important role. Third, our analysis sheds new light on the sobering fact of pervasive real-income declines for the large majority of Americans in the past decade. These real-income declines may be part of the same globalization and innovation forces shaping returns to superstars and to capital.
These conclusions must be placed in the proper context, which is “there is so much more we need to know from future research.” A good deal of recent empirical work investigates the effects of trade on the adjustment process of particular workers, occupations, and industries (which simple models ignore), and documents workers, occupations, and industries (which simple models ignore), and documents (the sometimes long-lasting) adverse effects. Our goal here, however, has been to advance some basic models describing the economywide evolution of, for example, widespread real-wage declines but rising earnings of superstars. Of course, future research will hopefully explore not only the experience of the United States but that of many other countries as well—both developed and developing.
For superstars, we do not yet fully understand product prices in sectors that employ superstars relatively intensively. This is both because existing industry data do not distinguish highly talented individuals well (if at all), and because many of the sectors in which we presume superstars are concentrated consulting, athletics, and entertainment do not have reliable data on product prices (or much else). Nor do we have good data on personal attributes that make individuals potential superstars. We suspect that for at least some of these superstar intensive industries, globalization has played an important role in boosting demand for their services—both via the information technology revolution reducing their natural trade costs and thus boosting their tradability, and via fast economic growth around the world boosting demand for their services. But these conjectures await additional analysis.
With regard to the sobering falls in real income for the large majority of Americans, our framework does add some new insights. We agree with Autor (2010a) that explaining falling real income for so many American workers remains a daunting empirical challenge. Much research to date has focused on income inequality, not income levels. We argue that this focus should change, because the post-2000 real-income declines are pervasive, new, and troubling. Our enriched trade framework offers some possible explanations for how globalization and/or innovation work offers some possible explanations for how globalization and/or innovation can boost superstar real earnings yet reduce real earnings of so many others.The last paragraph is particularly, as the author's say, sobering. But it also suggests that the outsourcing debate is more or less irrelevant--I doubt that China and India (other than Bollywood) are much in the superstar business yet.