Yesterday's New York Times predicted that the United States Conference of Mayors would shortly issue a report on income inequality, and sure enough it promptly turned up here.  The Report was actually prepared for the Conference by a consulting firm called IHS Global Insight.  Friends of mine tell me that that firm is generally well respected in the field of economic consulting.  Perhaps that is true, but you would never know it from this Report.

According to the Times, Mayor de Blasio of New York is the "chairman" of the "task force" that is "tackling" the issue of income inequality for the Conference of Mayors.  The task force was apparently formed at the prior meeting of the Conference in Dallas in June, but its next meeting took place yesterday in New York, seemingly coincidentally with the issuance of the Report.  I guess they just left it up to de Blasio to come up with what to tell IHS, because the Report more or less parrots de Blasio talking points on the issue.

The problem with this Report is that it completely accepts U.S. Census income data at face value, without considering what is included and not included in those data and whether those data are suitable for the purpose for which they are being used.   The result is a series of howlers, occurring both in the sections on diagnosis of the issue and in the prescriptions for solution.

The centerpiece of the income inequality discussion is a chart of income by quintile of the population for selected years since 1975.  The chart is derived from U.S. Census data.  It seems to lose some formatting in my copying, but I think it is still legible, so here it is:

FIGURE 4: SHARES OF INCOME BY QUINTILE

            Quintile                 1975          1985          1995          2005          2012   
 
% of income
Lowest 20%             4.3            3.9            3.7            3.4            3.2
Second                10.4           9.8            9.1            8.6            8.3
Third                  17.0          16.2          15.2          14.6          14.4
Fourth                 24.7          24.4          23.3          23.0          23.0
Highest 20%            43.6          45.6          48.7          50.4          51.0
           Top 5%                16.5          17.6          21.0          22.2          22.3   

The Times summarizes the takeaway as follows:

From 1975 to 2012, the highest-earning 20 percent of households markedly outpaced the lowest-earning 20 percent in America. In 1975, the wealthiest households captured 43.6 percent of the nation’s income, while the poorest had a share of 4.3 percent. In 2012, low-income households saw their share drop to 3.2 percent while the high earners saw their share jump to 51 percent.

Now for starters, is it too tacky to mention that this Census data is pre-tax, so that the top earners lose up to half to income taxes while at the bottom they get a refund from the EITC?  That would change these numbers very dramatically, but no, it is not mentioned either in the Report or in the Times coverage.

And how about that 3.2% share of income in 2012 for the bottom 20%?  Those who know anything about this subject know that government in-kind handouts are also not counted in the Census data.  Are those enough to skew the answer?  Yes, hugely.   The most recent compilation I can find from the Heritage Foundation puts the annual cost to the taxpayers of all government means-tested benefits at $940 billion in FY 2011.  That would mean that they have to be well over $1 trillion today.  Well, that's about 7% of all personal income in the country, most to almost all of it going to that bottom quintile.  So how can we possibly take at face value the 3.2% share of the bottom quintile in national income?  Yes, a few of the government handouts are counted in the Census definition of income (e.g., TANF, SS disability), but the large majority are not (housing, food stamps, Medicaid, EITC).  Of course you can't actually get the real data anywhere, but my estimate is that the handouts would approximately triple the share of the bottom quintile in national income, and also completely reverse the direction of change for the share going to that quintile, from a small decline since 1975 to a large increase.  And no, neither IHS nor the Times mentions this.

And of course when you have no idea what's in the data you are using, you are very likely to come to ridiculous conclusions.  For example (from the IHS/Conference Report):

An existing program which may be expanded to directly reduce the after-tax income
inequality is the Earned Income Tax Credit, which also encourages work and builds
experiences and skills.

Well, yes, but after six more rounds of increasing the EITC, won't people like you still be quoting the pre-tax income inequality numbers, without saying so, to convince the ignorant to support yet more EITC redistribution?  That's certainly what you're doing here.  And even that's not as bad as then including a paragraph praising increased pre-K education programs as a way of reducing income inequality.