Hoisted from comments:
Grasping Reality with Every Limb: Brad DeLong's Semi-Daily Journal: Intellectual Garbage Collection: The Unreliability of Alan Reynolds: The problem with Reynold's critique is that he latches on to a few correct points about the limitations of the IRS SOI data, then he:
- ignores the work others have done to correct these problems;
- way overstates the bias from these problems;
- ignores other data that do not suffer from these problems; *having ignored all the inconvenient evidence to the contrary, misleadingly concludes there's no inequality problem.
He's not nuts. Some changes in inequality as measured by the IRS SOI data (the main source for Pik&Saez) are induced by income shifting due to tax changes. But while that can explain a spike in one year to the next, it doesn't explain longer trends in income concentration found in that and every other data set we have.
Also, analysts at IRS have worked hard to try to deal with some of these shortcomings. Take a look at this paper--the figures and tables clearly show an increase in inequality since 1988 income measures that correct for some of the inconsistencies.
I won't belabor the CBO points others have made, other than to say that these data go the furthest towards addresses all the concerns raised by Reynolds. Sorry if these tables don't align correctly--I just pasted them in from excel.
They show real household income changes, since 1988. 2000 was an economic peak and the year before the bursting financial bubble took a big bite out of high incomes (large capital losses), so you see more inequality if you stop there. But I've include the latest CBO year too--02-03--to show that the old pattern is returning.Pretax 1988-2000 1988-2003 2002-2003 Bottom fifth 15.6% 9.6% -1.3% Mid fifth 11.2% 7.5% -0.4% Top fifth 35.3% 19.0% 2.3% Top 1% 68.4% 25.1% 5.9% Posttax 1988-2000 1988-2003 2002-2003 Bottom fifth 17.7% 13.7% -1.4% Mid fifth 12.9% 13.1% 0.7% Top fifth 30.9% 20.0% 3.9% Top 1% 60.4% 22.1% 8.2% Source: CBO
You simply can't write about this stuff in good faith and leave all this information out, unless you're trying to push an agenda that's dependent on misleading.