This is an email from Paul Krugman:
I think it's really important to realize that we have only a modest amount of direct evidence that technological change is driving increased income inequality. That is, while there have been a few studies showing some connection between increased use of IT and changes in the wage structure, very little of the conventional wisdom that technology is the culprit is based on those studies.
So why is technology given the credit? Basically because it's the residual category - and as Bob Solow said about the role of technology in growth, the residual is the measure of our ignorance. We estimate the effects of the stuff whose effects we know how to measure - taxes and globalization, mainly - and then attribute the rest to technology.
The point is that it's all too possible that we're attributing to technology rising inequality that may be largely due to hard-to-quantify political and institutional change.
There are several reasons to think that politics plays a big role. One is the broad correlation between the political climate and trends in inequality, which I pointed out in the Times. (By the way, Larry Bartels in Princeton's politics department shows that there's a strong correlation between party control of the White House and inequality trends even in the short run; see http://www.princeton.edu/~bartels/income.pdf. It's kind of a mysterious result, but worth pursuing.)
Another piece of evidence is the wide difference in inequality trends between the US and to a lesser extent the UK, on one side, and everyone else.
Yet another piece of evidence, which I think is very suggestive, is the discontinuous nature of the Great Compression. If you go back to the original Goldin and Margo paper, http://www.nber.org/papers/W3817, they found that there was a drastic reduction in wage inequality over the course of just 5 or 6 years in the 40s, which then stuck for another 30 years. In the paper, they struggle to reconcile this with a supply-and-demand framework, but it sure looks like a change in norms which had sustained effects on market outcomes.
So what are the mechanisms? Unions are probably top of the list; I believe that there's a qualitative difference between wage bargaining in an economy with 11 percent of workers unionized, which is what we had in the early 30s, and one with 35 percent unionization, which is what emerged from World War II. That's discontinuous change, partly driven by a change in political regime. And the process went in reverse under Reagan.
An overall climate of public scrutiny may matter too, especially at the top of the scale.
And don't forget that some taxes affect the pre-personal-tax distribution of income. Taxes on corporate profits went from a minor inconvenience before FDR, to a major source of revenue under Eisenhower, and back again.
The bottom line is that the view that rising inequality reflect forces beyond the reach of politicians may sound sensible, but it's actually a supposition based on very little evidence, and there's a lot of evidence on the other side.
Update: Andrew Gelman at the Statistical Modeling, Causal Inference, and Social Science blog reminds me of his excellent discussion of the Bartels paper referenced above. The discussion is at