Some Real Information On Poverty And Income Inequality

On numerous occasions on this blog, I have pointed out that the government's data on poverty and income inequality are systematically fraudulent.  For starters, they define "income," for purposes of determining both poverty and income inequality, in a way to arbitrarily exclude well over a trillion annual dollars of government transfers and benefits, leading to results that are entirely misleading.  And then those intentionally misleading results are used to advocate for yet more government programs and transfers, all of which will again be excluded when measuring poverty and inequality in the next round.  For a few examples of my previous posts on this subject, see here, here and here.  If you have time, I would recommend reading those for background.

What I have not previously done is attempt to go through all the uncounted government programs and quantify the effect that including them would have on the reported rates of poverty or income inequality.  One reason I have not done that is that it is a lot of work.  Another reason I haven't done it is that even correcting for all the omitted government programs would only be a start at the project of getting a handle on the real rate of poverty in the United States, that is, poverty in the sense of actual physical deprivation.  Even if all government benefits and transfers get included in the "income" of the recipients, and the statistics for poverty and income inequality get corrected accordingly, there would still be very large amounts of resources available to the "poor" that would remain uncounted.  The most obvious example would be the unreported illegal economy (estimated in this 2011 study at approximately $2 trillion annually, or about 12% of the economy, which is even more -- almost double -- the amount of uncounted government benefits).  And then there's the provision of resources by families and extended families.  Nevertheless, doing a study to figure out what the quantitative effect of including all these previously-excluded government benefits would be on the poverty and inequality statistics is certainly a worthwhile project.

And thus into my mailbox this week floated exactly such a study, by a guy named John Early at the Cato Institute.  The study is titled "Reassessing the Facts about Inequality, Poverty, and Redistribution."   Although I only got it this week (in the snail mail), it has a publication date of April 24.  Early is identified as a former assistant commissioner in the Bureau of Labor Statistics, which probably is a good indication that he knows how these numbers are put together (although the income and poverty numbers come from the Commerce Department rather than BLS).

So first, John, could you give us a list of some of the government benefits and handouts that are excluded when the government measures "poverty" and "income inequality"?

Census money income estimates explicitly exclude the following:4

  • The Earned Income Tax Credit (EITC)
  • The monetary value of benefits from the Supplemental Nutrition Assistance Program (SNAP), more commonly known as food stamps
  • Free or subsidized medical care such as Medicaid and the Children’s Health Insurance Program (CHIP)
  • Free, subsidized, or controlled rent or other “affordable housing” schemes
  • Heating subsidies
  • Free or reduced-fee social services such as daycare, tax preparation, or meal services

That footnote 4 goes to a Census Bureau site with the official government definition, so if you are so inclined you can check to see that he's right..  And, besides government benefits and handouts, are there any other large categories of spendable resources that are completely legitimate and above-board that are systematically excluded in measuring poverty?  Yes.  Biggest categories:

  • Capital gains!  (What is the possible reason for this exclusion?)
  • "Lump sum" (as opposed to annualized) withdrawals from retirement savings!

Here's Early on the latter:

The Census Bureau acknowledges that retirement income is underreported.5 The underreporting results in part from excluding lump-sum payments. . . .  The Census Bureau’s money income statistic excludes these lump sums, and the subsequent withdrawals from the bank are not counted either.

And then, of course, at the high end, the Census "income" numbers make no adjustment for taxes paid.  This omission is rather highly relevant to the calculation of "income inequality."   

Early then notes that other government entities -- notably the Congressional Budget Office and the Congressional Research Service -- put out their own income estimates, with their own definitions.  Bizarrely, the estimates from these entities include some of the things omitted by Census, but still exclude plenty of other things.  Why?  Nobody knows!  For example, as to CBO:

In addition to the transfers included by the Census Bureau, the CBO adds food stamps, Medicaid, and CHIP.  But it still excludes about $900 billion in other transfers.  The CBO subtracts 93 percent of federal taxes. It does not subtract state and local taxes, and it counts the EITC as a negative tax [thus not added to "income" of the recipients].

Well, some might think that some $900 billion of annual omissions, plus the EITC (another $80 billion or so) comes to a real number when you are trying to determine if people have enough resources to live on.  Can you give us an idea of what the omissions are that add up to such huge numbers?

The Congressional Research Service (CRS) has listed 83 federal welfare programs with total appropriations of $746 billion. The CBO estimates [of "income"] include only seven of them.10 Many of these programs require state matching funds. The CRS identified only the federal portion. The Senate Budget Committee (SBC) staff identified $283 billion in state matching funds for these same 83 programs, yielding a grand total of $1.03 trillion per year, with most of it going to lower-income households.11

If you want a list of all of those 83 programs, you can follow those footnotes, or go to this link, which is the technical appendix of Early's study.  The list of 83 "need-based" programs (the first 7 of which are included in CBO's "income" estimates while the other 76 are excluded) is at pages 10-13 of the appendix.  The list is way too long to copy in full here, but examples of omissions include:  "8. Public Housing, . . . 10. Low Income Home Energy Assistance Program, . . . 20. National School Lunch Program, . . . 36. Pell Grants, . . . 49. Supportive Housing For The Elderly, . . . 53. Homeless Assistance Grants, . . . 73. Emergency Food and Shelter Program, . . . 74. Legal Services Corporation . . . ."  And on and on and on.  You get the picture.  There's no human need you can think of that doesn't have its very own program, benefit or handout, none of which count as income to the recipient when they report statistics about "poverty" or "income inequality."

And now for the bottom line:  If you now include in "income" all of the benefit and handout programs, plus all of capital gains and retirement income, and you include taxes refunded (mostly EITC) at the low end and subtract taxes paid at the high end, what results do you get for poverty and income inequality?

First, income inequality.  Measured by the GINI coefficient (1.00 being perfect inequality and 0.00 being perfect equality), the U.S. GINI coefficient goes from officially-reported 0.4 down to about 0.22.  And by the way, that 0.22 is lower than the GINI coefficient reported for any of the other G7 countries, including holier-than-thou places like France, Germany and the UK.  By quintile, the lowest income quintile goes from 2.2% of income all the way up to 12.9%, which is really not much lower than the percents for the next two quintiles (13.9% and 15.4%).  And the top quintile comes down from 57.7% to 39.3%.

And how do these same adjustments affect reported poverty?  Including all the benefits at cost to the taxpayers, and leaving the poverty "thresholds" the same, would reduce the poverty rate from an officially-reported 12.7% (2016) all the way down to about 2.8%.  Early also suggests that measuring the poverty "thresholds" by applying the implicit price deflator, rather than the CPI, to bring the levels forward since they were established in the 1960s would further reduce measured poverty to about 2%.

Here's what Early never gets to:  Who would that remaining 2% be?  They are the residual of people who have little to no income that anyone can identify even after accounting for all officially measured income and all government benefits and handouts.  The funny thing is, it is highly unlikely that that group would include any significant number of people that you would actually think of as "poor."  I mean, under this methodology, if a person started out in true physical-deprivation poverty, all that person would have to do would be to go apply for the benefits, and that act would suddenly lift him or her out right of the poverty.  Available benefits in essentially all cases are a multiple of the "poverty" thresholds.  The people that you think of as "poor" -- like the people you see living on the streets, the beggars, the people in the shelters and those in the low income housing -- all of them, if they ask, have full access to benefits that, if counted, will immediately lift them out of measured poverty.  Now, could there be some who, even though entitled to such benefits, somehow fall through the cracks because they fail to apply or to take advantage of what is offered?  Of course.  On the other hand, there are government workers out there every day looking for these people to try to help them.  The remaining "real poor" can't actually be a significant part of the residual 2%.

So who would be left?  If you stretch your brain, you can think of some legitimate possibilities.  Maybe bush people in Alaska or mountain people in the Appalachians, who live "off the grid" and entirely out of the market economy; perhaps heiresses living off family money; maybe some early retirees withdrawing from savings.  Could such categories really come to a full 2% of the population (about 6.5 million people)?  Nothing about the government data will tell you the answer; but personally, I find it unlikely.  

Let's face it:  most of this residual 2% has to be people who are in the underground economy.  They don't apply for benefits because doing so would risk discovery of their illegal situation.  Likely large categories are illegal immigrants and people making relatively big money illegally (e.g., successful drug dealers).    

Perhaps you are offended that counting the vast array of government benefits toward "income" is going to cause essentially all of the recipients to be removed from measured "poverty."  After all, you can observe these people on a daily basis, and many of them are in obvious distress.  They certainly look poor!  But availability of government benefits far in excess of the poverty threshold manifestly fails to alleviate their distress.  Would doubling, or tripling, those benefits lead to any better result?