Paul Krugman writes that politics matters for the income distribution, citing the long-term trends in inequality and their close correlation with long-term political trends. Brad DeLong says he thinks this is wrong, political changes can and do have a large impact on after-tax income distribution but the trends show up strongly in pre-tax income. "I can't see the mechanism by which changes in government policies bring about such huge swings in pre-tax income distribution."
I note for the edification of readers that one thing I've learned since arriving in DC is that a difference of opinion on this subject is a major divide within the progressive economic policy community. Most mainstream economists -- including most liberals -- agree with DeLong. Politics and policy affect the secondary distribution (after tax and transfer) and what happens with the primary distribution is just out there. Leftier economists tend to say this is mistaken.
I would side with Krugman on this. The trend data is too striking to be ignored. If you have a phenomenon and are having trouble identifying the cause, the thing to do is to try harder to identify the cause, not assert that the phenomenon isn't happening. But what is the cause? I can think of some plausible stories.
One thing to say is that tax policy impacts pre-tax distribution. When the top income tax rate was very high -- 70 percent or above -- this not only meant that rich people paid a lot in taxes, it also meant that there were a broad range of circumstances where it didn't necessarily make much sense to offer well-compensated people even more compensation. When you have a very progressive rate structure, an employer can get a lot more bang for his buck by directing his employment budget at middle-income people than at rich people. As you flatten the tax structure, this becomes less-and-less the case.
Similarly, very high tax rates encourage high income people to engage in more leisure and less work whereas right now we have the somewhat odd situation where highly compensated people tend to work more than do the moderately compensated.
All this, I think, makes a big difference.
Then the other factor to note is probably unionization which is much more impacted by policy decisions than people often seem to realize.