Wednesday, September 21, 2022

Gramm, Early and the Unfixable Problem

Phil Gramm and John Early have a new WSJ oped, based on their smashing new book. Both are based on an astounding fact: The numbers used by the Census Bureau, and countless following researchers, to define income inequality and poverty do not include taxes, which reduce income of the rich, and transfers, which increase income of the poor. The latter, obviously, matters to just how many Americans fall in the Census Bureau's definition of poverty.

Specifically, in the oped, the new refundable tax credit cannot, by arithmetic, do anything to alleviate measured child poverty because 

"the income numbers used to calculate the official poverty rates don’t count refundable tax credits as income to the recipients. "

This is wonderful for advocates of ever larger transfer programs, as it creates a problem that can never be measured to be fixed! 

The more general issue 

The Census Bureau fails to count two-thirds of all government transfer payments to households in the income numbers it uses to calculate not only poverty levels but also income inequality and income growth. In addition to not counting refundable tax credits, which are paid by checks from the U.S. Treasury, the official Census Bureau measure doesn’t count food stamps, Medicaid, the Children’s Health Insurance Program, rent subsidies, energy subsidies and health-insurance subsidies under the Affordable Care Act. In total, benefits provided in more than 100 other federal, state and local transfer payments aren’t counted by the Census Bureau as income to the recipients

The book goes on to show how this startling omission overturns just about everything you've heard from the hyperventilating classes about income inequality. Granted, spending zillions on rotten health insurance that people value much less than a dollar per dollar is not quite the same as cash, but there are lots of cash or cash equivalent transfers in there. 

A question I do not know the answer to: Do means-tested programs count as "income" the transfers from other means-tested programs? If a program is only available to, say, those with less than $50,000 per year income, does that figure include any other means-tested programs?  Even the ones that send cash, rather than in-kind transfers such as rent, energy, and health insurance subsidies? I suspect largely no. If not, the incentives for means-tested programs are far worse than even they appear. Facts welcome.

One might easily respond that ok, but evil capitalism created wider pre-tax pre-transfer inequality, and only by the grace of larger and larger transfers has some measure of stability been restored. Well, which is the cause and which is the effect -- wider pre-tax pre-transfer inequality, or the large expansion of means-tested programs, all of which add to the stupendous marginal tax rates facing Americans with less opportunity? The book goes on to argue convincingly the latter. I'll cover that later. Noting here, they anticipate the argument. 


  1. The local municipality provides subsidies to residents who would otherwise not be able to afford social services delivered by the municipality's recreation and cultural services department. Specifically, the subsidies are provided through discounting program fees by allowing households with children to enroll their children in sports programs and cultural programs (dance, art, music, &c.) for a nominal amount (set by the family member), and by allowing senior citizens to enroll in seniors' programs for nominal amounts (det. by the senior citizen) on application. The applications are maintained in strict confidentiality, and other families and seniors are not made aware of the enrollee's need or the amount of the payment made. The shortfall in program revenues is made up from the municipality's general fund revenues (princ. property taxes).

    The subsidies are not considered to be "income" to the recipients. The purpose this program which is not widely advertised is to ensure that members of the community who cannot otherwise afford to participate in the community's programs are able to do so without taint of receiving handouts or accusations of free-riding.

    One can imagine the reasons behind the outrage expressed by Phil Gramm and John Early, but a statistics bureau that 'fudged' the numbers by including costs of government programs delivered to low income households ("transfers") would just as easily be accused of misleading government decision makers as the extent and nature of the distortions in gross income distribution amongst the population at large. As a decision-maker, you want to see the full extent of the need before application of the remedies, not after. Governments by and large already know how much they expend on transfers to individuals and families. They already know how much they expend on subsidies to firms and corporations. What they would want to know is how much of the economic pie is being shared by which population cohorts reporting income by income category. How much tax to collect; how much of that tax is to be transferred to the less well-off in society; what good is it doing; what harms arise from it? You can't get there without data free of the transfer program spending as "income". It's simply a case of proper census practice. To assert otherwise, is to assert an untenable position premised on political concerns. Gramm and Early are pursuing a misguided path.

    1. Actually, you would like to see the entire break down so that you can compute all the statistics yourself. Why would I only want data on income before taxes and transfer, or income after taxes and transfers? If you give me the entire break down, I can create whatever category best fits my application -- and derive things like poverty rates over time myself as I see fit if I need them.

    2. The issue is not what the economic analyst is interested in, but what measure of poverty yields the greater benefit to which interest faction, political party or social activist advocating for more (less) social spending (transfers between households).

      Gramm and Early argue that income used to define the poverty threshold should include such social spending programs as Medicaid, "SNAP" (Food stamps), "Obama-care" subsidized health care insurance premiums, &c.

      The Census definition (status quo) defines income for poverty level excluding those social spending categories. From a census perspective, excluding the enumerated social transfer payments, subsidies and supports to low income households and persons is best practice. If the census looks only at gross income, excl. social welfare and subsidies, then a true picture of the distribution of income is obtainable and threshold (cut-off line) is objectively practicable.

      The (political) focus is on what the government defines as income, not what the idealized economic analyst would define as income.

    3. "The issue is not what the economic analyst is interested in, but what measure of poverty yields the greater benefit to which interest faction (...)."

      So, you didn't understand a word of what I was saying, or you didn't bother reading. I am saying that the proper measure for doing any number of things either the government or someone like me would do is context-dependent so that everyone would be better served if that choice was made by the end users.

      And your response is "no, I'll make it for you and it's political." That's very amusing.

    4. Government is political.

      You're at liberty to search the government databases and create whatever statistical measures you think are important to whatever object you are pursuing. Who's stopping you? Not I.

      The subject matter is not what the individual analyst would like to do, given sufficient time and resources, but what political decision-makers determine they need before they can make a decision concerning programme thresholds, budget appropriations, &c.

      Gramm and Early are pursuing a political objective with their series of essays and articles. Piketty and Sais got out ahead of Gramm and Early, and G & E are playing catch-up. One reason to focus on gross income to the exclusion of programme transfers after tax is that programmes rely, for their continuity, on political decisions in Congress and the Administration, and as such whatever "levelling-up" that redistribution of income by government may be at the present moment can be changed with a change in government and is therefore not permanent. If you object to social security, and feel that a 'guaranteed basic income' would be more to the purpose, and you have the political power to effect discontinuance of social security and replacing it with $10,000/yr/person 'guaranteed basic income', then whatever the income value of social security is now becomes nil, but the gross income before transfers remains unaltered.

      It's political because dividing national income between interest groups and sectors and regions is political. Economics cannot be divorced from politics, try as some might try. The tax code is all about politics. The 'War on Poverty' is entirely political, from conception to realization. Piketty and Sais are political economists. Gramm is a former congressman. Politics runs through it.

    5. Census publications re: poverty level in the U.S. can be found here:

  2. I think the general public should be given both strict income and tax+transfer data when being told there is a major problem with increasing inequality.

  3. In general, I agree with this post.

    However America has also followed "cheap labor" policies for the last 60 years in regards to de facto open borders for labor and imports.

    Add on, lots of "expensive housing" policies due to property zoning.

    1. In general, policy analysts should cease trying to measure "poverty" by the accountants' measure of "income." Inequality should be measured by something like "household wealth" and include an index for skills and labor wealth. Replace one accouting fiction with another, since all that is accomplished here is a more "precise" measure of social inequality (income and wealth are both just proxies for the generalized idea that some people contribute more and other people receive more from the social pie).
      This is known as False Precision and can yield all sorts of nonsense.

    2. The famous Apples + Oranges aggregate.

  4. I think the article (not the blog post) is slightly misleading. A couple things to consider:

    It is correct that the official poverty measure by Census does not include tax & transfer income. This is how the measure was originally created, for better or worse. A reason it hasn't changed is bc it would make numbers less comparable across time.

    This is why Census also the Supplemental Poverty Measure (SPM)-- which does include transfer income. Both measures include cash transfer income (like Social Security, UI benefits) but the SPM does also income tax transfers (the EITC) and includes the value of non-cash benefits as well (like SNAP & Medicaid).

    There are other differences too, but the Census isn't ignoring transfer income -- it created an additional measure just for this reason while maintaining their historical measure as well.

    1. This is true but the question then is do these high profile inequality studies/ hyper ventilating news stories use these other more complete measures (SPM etc.) of income when they calculate inequality? Btw the WSJ article does mention there measures that include some of the transfers.

    2. So, the Census Bureau is basically already doing what I explained somewhere else they should be doing. I was kind of puzzled as to why they wouldn't be doing something like that -- and did it sound from the post that they weren't doing it.

      As for Anonymous, I'm an economist and I wasn't aware that all nicely curated series existed for this because this isn't the kind of data I tend to work on. You can easily imagine journalists wouldn't know. As for studies on inequality, I'm not familiar with that literature, but I can tell you that there are various reasons cited in studies for using peculiar kinds of definitions: it can be an issue of trying to make international comparisons and data availability is often a problem in that context, it can be trying to work on a theoretical model and an odd definition best fits the model, and it can also be that they have a very specific question that's best addressed by this definition, or in some case you want to be comparable to other studies so you use their definitions.

      Economics isn't exactly the kind of field where top journals publish ideological drivel: the field is just too ideologically diverse for that, so I'm guessing they have some kind of reason like that for using it.

  5. I made this same point in my 2015 paper responding to Piketty, "Capital and Finance in the 21st Century" (

    In addition, wealth inequality statistics ignore the major source of wealth for most people -- the present value of their pension, social security, and Medicare benefits.

    I support a UBI to replace all of these transfer payments. As you point out, dollar-for-dollar, people value cash more than these benefits, so it would increase aggregate utility. Since they could spend the cash as they wish, prices would be better signals for producers, increasing economic efficiency. It would also reduce inflation in subsidized industries (health care, education, child care, etc.). Finally, it would reduce the size of government, end the demeaning and inefficient process of applying for and receiving these benefits, and provide a cushion for workers who are displaced by creative destruction.

    I would set the UBI at the poverty level -- roughly $10k per person (or $40k for a family of four) -- eliminating poverty in one fell swoop. But, as with all government programs, we should probably test this one out on a small sample before imposing it on the country at large (ie, the scientific method).

    1. Follow that link! I just read Bob's excellent, concise, clear review of Piketty. A must-read,

    2. would ever accept that it should go to everyone and no one elshould receive everything extra. I've tried explaining this to people before se that everything after that point becomes a luxury and it does not register. For example, living in London is expensive, i.e. luxury. Therefore, if you're on UBI and nothing else, you're goingprobably to have to move out of the expensive city to one cheaper.

    3. At $10,000 per individual to replace Social Security and Medicare currently relied upon by seniors who do not have private pension plans, your preferred policy option would impoverish a large fraction of the population too old to work, or to medically incapacitated to hold full-time employment. It would certainly solve the federal government's pay-as-you-go unfunded liabilities, but at a cost of great impoverishment, only offset by a jump in the death rate, and a decline in the average life expectancy statistic. In other words, it would regress back to the 1930s and earlier living standards for those not fortunate enough to have solvent defined benefits pension plans. It would also reduce consumption, in favor of higher precautionary lifetime savings.

    4. Good stuff gentlemen - but no such thing as aggregate utility. Theory of Games and Economic Behavior (1944) - von Neumann Morgenstern

  6. I bought the book and found what I was looking for: The marginal tax rate on average income in the bottom quintile moving to average income in the second quintile is ------ 93%. 'Twould make a rich man blush!

  7. "Governments by and large already knOld Eagle Eye comment: ow how much they expend on transfers to individuals and families. They already know how much they expend on subsidies to firms and corporations."

    I dispute that governments know how much they expend on transfers. The For example, tchool district my wife works in is 25% Hispanic. Some legal, som and illegal Mexican and Guatemalan. There'overall s no accounting figure providdetermined ed uch of the student population their fall below the threshold tonts d localon't pl system, i.e. a transfer paid for by rest of the tax and businesses locatedpayers in the school corp for the school corporation to cover its costs of operating.

  8. To Grumpy Economist:
    I agree that “what to count” is an important question and it may have more than one answer. In the article, $4.5 trillion for 2021 for transfer payments was reported. But it is back to about $3.914 trillion for Q2 2022, still a high number. This figure came from a Federal Reserve report: See Table 8 at this link:

    But another Federal Reserve report shows a different figure for Q2 2022. The FRED chart shows Transfer Payments to be only $2.874 trillion. Why the difference I do not know. (If I remembered all of National Income accounting I probably could answer my own question.)

    It sure would be nice to have a table that shows all of the relevant line items and the formula used to reach a sub total or total.
    I tied to find the details or line items that total $2.870.049 trillion. See the pdf entitled “Effects of Selected Federal Pandemic Response Programs on Personal Income, 2022Q2 Second.”

    Given the high amounts of Transfer Payments, it might be useful to have a report that lists the categories and show the math behind the figures
    Any help will be appreciated.

    Thank you,

    Joseph V. Smith
    Houston, TX


Comments are welcome. Keep it short, polite, and on topic.

Thanks to a few abusers I am now moderating comments. I welcome thoughtful disagreement. I will block comments with insulting or abusive language. I'm also blocking totally inane comments. Try to make some sense. I am much more likely to allow critical comments if you have the honesty and courage to use your real name.