Minor Blog Wars – my part in their genesis

As a regular reader of Brad DeLong I was slightly alarmed at a recent post reporting an outbreak of unpleasantness about which OECD country has the most progressive tax system.

Brad DeLong linked to an article by Jonathan Chait which rather sharply criticised Veronique de Rugy of the Mercatus Centre for arguing that “Contrary to common belief, the United States already has a more progressive tax system than do the most industrialized democracies worldwide. ”

Chait pointed out that “the amount of taxes paid by the rich is a function not only of their relative tax rate but also their relative share of the income. If rich people earn a far larger share of the income in the U.S., which they do, then they may pay a higher share of the tax burden even if the U.S. income tax system is less regressive.”

Responses have subsequently ping-ponged across the internet, as is common in controversies of this sort. Veronique de Rugy has responded twice, and  Jonathan Chait  had another go.  Paul Krugman and Clive Crook take opposing sides

Brad DeLong came back again and reported the results of the latest OECD study Divided We Stand which shows that the USA reduces inequality less than most other countries.

The figures initially quoted by Veronique de Rugy came from the chapter on redistribution in an earlier OECD study “Growing Unequal” published a few years ago.  Readers, I wrote that chapter – hence my concern that people understand what it means.

In fact, the chapter quoted by Veronique de Rugy found that the USA did have the most progressive distribution of direct taxes in the OECD, as we can see in the chart below. This includes income taxes and employee social security contributions, but not indirect taxes.

As Jonathan Chait correctly points out, part of the reason why the US has a very progressive tax system is that it also has a very unequal distribution of market income.  An identical tax scale would collect more money from rich people in one country if they had a higher share of income than the rich in another country.  However, we also calculated what was the effect of greater inequality in the USA, and found that the direct tax system in the USA was now the second most progressive (after Ireland).

Another objection to this finding is that by looking at only direct taxes, we leave out the less progressive taxes that may impact more heavily on low income households. It is correct that taking account of employer payroll taxes and sales taxes would make the overall US system less progressive, but employer payroll taxes and VAT are much heavier in Europe, and also less progressive than their direct taxes, so the US ranking would almost certainly not change.

This finding is not new -  Peter Lindert has pointed this out before – English-speaking countries tend to have more progressive tax systems, but lower overall levels of taxation, while Europe has less progressive and higher levels of taxation.

It is important to understand what progressivity means in this context. Some time ago Don Arthur pointed to earlier debates about this.

Progressivity measures the differences in taxes paid at different income levels.    Progressivity is measured as the rate of increase in taxes, not the level of taxes.  This is simply the widely accepted standard for measuring progressivity used in the technical literature.

In the OECD report and in the figure above progressivity was measured by the concentration coefficient of household taxes, which is the Gini coefficient but calculated with households ranked by their disposable income rather than their tax payments.

It is the distribution of taxes that is used in calculating progressivity, but this tells you nothing about the level of taxes. The extent to which the tax system redistributes income is determined by both the level of taxes and their progressivity.

For example, in the USA, the taxes paid by the richest 10% amount to 41% of their income, while the taxes paid by the poorest 10% are about 12% of their income. If you take Sweden, the richest 10% pay 58% of their income in taxes, nearly one and a half times what the corresponding group in the USA pay.

However, in Sweden the poorest 10% pay nearly 25% of their income in direct taxes, or more than twice the rate for low income Americans. Put another way the richest 10% in the USA pay 3.5 times as much in taxes as the poorest 10%, while in Sweden the richest 10% pay 2.3 times as much in taxes as the poorest income group.

So taxes in the USA are more progressive than in Sweden, but rich people in Sweden pay much higher taxes than rich people in the USA.

Most importantly, progressivity is not the same as redistribution. In a simplified sense, countries have four fiscal tools for addressing income inequality – the first is the overall level of taxes, and the second is the progressivity of the tax structure, the third is the overall level of social spending, and the fourth is structure of spending, which can also be defined in terms of progressivity.  As I have pointed out before, Australia has the most progressive distribution of benefits in the OECD. However, for assessing redistribution you need to look at the combined effect of all four factors together. But it is also useful to look at each component separately, in order to understand how specific policy changes impact on income inequality.

Because European countries collect a lot more in tax they spend a lot more on social security transfers and government services. It is through the spending side that other countries achieve much greater redistribution and reduce inequality to a larger extent than the USA. In fact, the USA has one of least redistributive social security and welfare systems in the OECD – only South Korea is less redistributive. The USA is the only OECD country which reduces inequality more through the direct tax system than through transfer spending.

What is important, however, is the combined effect of taxes and spending, and when you put the two together the USA is about the fourth least effective OECD country in terms of reducing inequality, as shown below.

Another way of visualising this is look at the distribution of taxes and transfers together.  This means that we treat benefits as negative taxes and add them to the positive taxes that households pay.

When you do this, the picture looks  like this.  For simplicity, I have only included the USA, Sweden and Australia.  In order to standardise across countries net taxes are expressed as a percentage of overall average household income in each country. In all three countries, tax rates are negative for most of the bottom half of the income distribution.  and become increasingly positive in the top half of the distribution.  What stands out most, however, is that the USA has the lowest negatives in the bottom half of the distribution.  For the bottom decile, net transfers in the USA are about two-thirds the level of Sweden and only half the level of Australia.  This difference is entirely due to the transfer side of the equation – remember that taxes on the bottom decile are much higher in Sweden than in the USA.


So it is certainly not correct to say that the USA has the most progressive tax-transfer system, even it is correct that it has the most progressive direct tax system.  So my reading is that Veronique de Rugy is correct, although not comprehensive.

It is worth noting that part of the explanation for the higher progressivity of the US direct tax system is that they provide support through the tax system in the form of the earned income tax credit (EITC) and refundable child tax credit, which here in Australia we do through the family payments system as cash transfers.

As an outsider, I find it puzzling to see the US discussion focus only on the progressivity of taxes.  This is important, but it is only one of the tools that can be used to reduce inequality.  The reason why most other OECD countries reduce income inequality more than the USA is that they place greater emphasis on spending – and collect the taxes necessary to pay for it.

24 thoughts on “Minor Blog Wars – my part in their genesis

  1. Great post. I have one comment which can start based on this bit:

    “What is important, however, is the combined effect of taxes and spending, and when you put the two together the USA is about the fourth least effective OECD country in terms of reducing inequality, as shown below.”

    Not being an economist, perhaps I have a weird view on all of this, but I wonder if this would be a better place to start than the standard debate which always seems to start at levels of redistribution — i.e., starting at the extent that your system happpens to get you to some level of inequality over time, rather than the extent your system happens to move money around. For example, it may well be harder to reduce inequality in some places than others for many and very varied reasons (e.g., entrenched poverty, family structures, cultural attitudes to education etc.). If this is the case, then comparing one country with another and finding out they have similar rates of redistribution is not the most meaningful figure on Earth — Perhaps the US really does move money around more than some other countries, but if inequality in the US is harder to change, then the difference between the amount they need to move around would need to be greater. Thus the difference between the amount of redistribution done and the amount needed is more important than the absolute level of redistribution that occurs.

  2. Thanks Peter. Very informative. It makes me rethink my approach. I might use some of this in a tax teachers’ association paper I am re-doing called Reason in revolt now thunders to end the age of neoliberal tax cant? I just took the ACTU figures – the top 20% pay 34.5% of their income in tax; the bottom 20% pay 26.7%. The top 20% now over 60% of Australia’s wealth; the bottom 20% less than one percent. So much for progressivity. But having read your piece I’ll need to rethink and rework this.

  3. Defence spending in the USA is part of explanation, but not all of it

    The more recent OECD report – to which I did not contribute – notes about the USA: “Redistribution of income by taxes and benefits is limited. Over the long run, these offset less than 10% of the increase in inequality of market incomes – gross earnings, savings and capital taken together.

    The limited redistributive effect in the United States is to be found on the benefit side rather than the tax side: benefits represent just 6% of household income, while the OECD average is about 16%. Income support for the unemployed has become less generous over time prior to the 2008-09 financial crisis. The gap between in-work and out of work income has increased for lone parent families and couples with children particularly. The income of a lone mother with 2 kids, who had full unemployment insurance and earned around the average wage, is less than 40% of her former take-home pay – in 1995, this was over 50%.”

  4. @Pedro “Is defense spending part of the explanation?”

    The average OECD nation has ~30% more revenue (%GDP) than the US. Scandinavian countries’ revenues are over 50% higher. That extra money goes to cash transfers, as well as healthcare and education subsidies (which the figures here don’t account for, I believe). The extra 2-4% that the US uses for defense goes on top of that, but most of the difference owes to higher revenues.

    Reading de Rugy’s original article, which mentioned that (1) her higher US progressivity figures were somewhat skewed by income inequality and (2) US tax progressivity is increased by tax credits, I suppose Chait’s mostly in the wrong. However, de Rugy’s article is dishonest. First, she addresses US tax credits, but not cash transfers which are effectively similar, not accounted for in this statistic, and result in the US being one of the most regressive OECD members. Second, she references higher US progressivity for the purpose of arguing that the US cannot bridge the budget gap by increasing progressivity without noting that top earners pay much less than in other OECD nations. If the US adopted Scandinavian rates for the top 5 or 10% of incomes, it would much more than solve the budget gap (paired with reforms that bring medical costs in line with the rest of the world).

  5. John Passant:

    “But having read your piece I’ll need to rethink and rework this.”

    Some sort of Internet first: an expert puts up a reasoned backgrounder to some overheated Internet debate, and someone who has taken one side or the other says they’ll need to rethink their position in the light of the new information.

    Anyway, this is a trend to be encouraged, and a great credit to both John and Peter. Though having read a few of Peter’s graceful and clear explanations over the years, I would have picked him as the expert most likely to provoke this reaction.

  6. A Australian acquaintance of Swedish origins paints a picture of Sweden that is not all sweetness and light … Sweden has a very high rate of Absenteeism – might be the highest in the world – and many swedes take every opportunity to get on to ferry’s to various Baltic states and get totally -reallytotally- pissed as quickly as possible , land in ‘latvia’, load up with as much aquavit as the boot can hold and then pour back on to the boat.

    He was not that impressed by the ‘paradise’.

  7. Pretty much all the Scandy countries have alcohol issues. With those long, dark cold winters and the accompanying seasonal affective disorder, that’s probably no surprise, especially since drinking culture is embedded through culture more broadly (look at Australia).

  8. Oh Peter, I’m afraid you have assaulted Veronique de Rugy’s reputation rather seriously here. You conclude that de Rugy is technically correct, but I’m afraid in the process of your explanation you have painted her as a snake oil salesman.

    If we now go back to the original column, armed with your definition and explanation, and now read it, we immediately notice things start to unravel:

    (1) The readers of this op-ed will be thinking not in terms of your definition but progressivity measured over absolute incomes rather than deciles.

    (2) The chart de Rugy includes is an apples-to-oranges comparison, since among other reasons the top decile in France is not at all like the top decile in the USA.

    (3) De Rugy’s conclusion, that making the tax code more progressive by raising the top marginal rate will neither help inequality nor help with the debt, is completely unsupported and quite arguably total bupkis.

  9. Jon

    I don’t normally see myself as being in the business of assaulting people’s reputations. I think she is technically correct in terms of expression, but doesn’t provide a comprehensive discussion of the issues.

  10. Jon S: “Peter, I’m afraid you have assaulted Veronique de Rugy’s reputation rather seriously here”

    My impression was the opposite – that Peter had been very careful to avoid assaulting reputations, while carefully unpicking most of the differences between the various commentators. As my post above indicates, I’d like to see more comments like Peter’s in blog discussions, and less of the other type: “X is obviously a biased fascist/Trotskyite/self-satisfied/delusional/megalomaniacal/inadequate idiot”.

  11. David

    Thanks for your kind comments here and elsewhere. Like Nick I am a believer in the powers of pursuasion – hope that’s an accurate picture of your views, Nick. Which is not to say that I don’t get annoyed by what I regard as very misleading posts.

    I think Ms de Rugy could be described as selective in her picking of which facts to emphasise, but as I indicated in the original post I am also puzzled as to why the American left emphasise direct tax progressivity over all other considerations.

    Generally speaking what the experience of the Nordic welfare states show is that what is most important is “how you spend it”.

  12. John Passant

    Thanks for your comments also. I think that my arguments are standard analysis from the UK social policy literature and approach.

    I think that you should also have a look at the two different ways the ABS present the distribution of wealth in their recent income survey publication. It is certainly true that when you rank wealth holders by their wealth, then the distribution of wealth is spectacularly unequal. However, when you rank people’s wealth by their income then this is mich less the case – in fact when ranking by income, wealth is less unequally distributed than income! This is because of the lifecyle accumulation of wealth – old people have more wealth than younger people, but they have lower incomes. It is also because the largest single component of wealth is owner occupied housing, which most prominently assists lower income older people – put another way some low income people are not as badly off as they appear.

  13. As Jonathan Chait correctly points out, part of the reason why the US has a very progressive tax system is that it also has a very unequal distribution of market income.

    On what basis is that a forward causality? Could just as easily point in the opposite direction. The economy is after all a feedback system.

    It is correct that taking account of employer payroll taxes and sales taxes would make the overall US system less progressive…

    Payroll taxes are not government money, they are money that the government holds in trust for retirements… much the same as compulsory Super, or a lifelong savings account. No different than taking money to the bank and putting it in as a deposit. This was a promise made to the people of America by President Franklin Roosevelt (some say he was one of those “progressive” guys, whatever that means, presumably not the same as whatever “progressive” might mean today).

    If the Social Security system is taking those people’s money and ripping them off, then I’d suggest someone take a very close look at how it is managed. If on the other hand Social Security is correctly paying back the investment as originally promised, there is no reason to concern about those people being unfairly taxed.

  14. “This was a promise made to the people of America by President Franklin Roosevelt.”

    No it wasn’t. There’s a promise to today’s workers of a future safety net in exchange for paying for the safety net of today. What you get out is a function of what you put in, but it’s a progressive one and one that’s changed over time. What bank inversely pegs returns to the size of an investment? What bank takes your money and then repeatedly says, “You know what… the returns I promised aren’t enough. Here’s some more!” A massive trust fund is a recent invention of the Greenspan Commission brought about to deal with Baby Boomer retirement.

  15. nice post, Peter. This stuff is much harder than it would seem from 1st year econ from which you usually get the misguided impression that tax progressivity is all about income tax. I think that underlies a lot of the confusion you signal.

  16. OK, FDR described it as an “insurance” system, and Social Security have described their accounts as an “insurance accounts”, so there’s a transfer aspect taking money from those with more capacity to pay, and giving it to those with less capacity to pay (and any compulsory insurance system will have a transfer aspect to it, otherwise it wouldn’t need to be compulsory).

    Thus the bank account analogy should really be an insurance account analogy. My apologies for being so sloppy.

    At any rate, pretending that payroll tax is the same as any other tax, and therefore regressive is downright dishonest.

    People paying in may not know exactly what they will get out, but they do expect they will get *something* out of it, because they expect to either grow old and retire, or lose their job and get unemployment, etc. FWIW here’s the relevant FDR quote:

    1. Unemployment compensation.

    2. Old-age benefits, including compulsory and voluntary annuities.

    3. Federal aid to dependent children through grants to States for the support of existing mothers’ pension systems and for services for the protection and care of homeless, neglected, dependent, and crippled children.

    4. Additional Federal aid to State and local public-health agencies and the strengthening of the Federal Public Health Service. I am not at this time recommending the adoption of so-called “health insurance,” although groups representing the medical profession are cooperating with the Federal Government in the further study of the subject and definite progress is being made.

    In the US context, “annuities” basically means insurance payments.

  17. Pingback: Is the U.S. tax system more progressive than those of most other rich countries? « Consider the Evidence

  18. Thank you for Peter’s very useful posting. We also need to think about the effect of taxes and transfer payments on inequality of lifetime consumption, or even potential consumption (the latter would argue for greater equality in quality of schooling. Transfers to elderly Americans who have accumulated spendable wealth due to significantly past income but who have low current income is very different from transfers to young people with disabilies that seriously impair their earning potential.

    Since in the U.S. compared to other countries a disproportionatley large fraction of transfers go to elderly people with low incomes relative to their lifetime consumption, this methodology would almost surely show a much less progressive system in the U.S. rlative to other countries. The other issue is consumption versus income inequality. At the high end of the income distribution, and for owners of businesses, income is seriously understated because capital gains are only counted on realization rather than on accrual.

    Many of these problems would be mitigated by having a progressive consumption tax, a highly progressive inheritance tax (rather than estate tax) and a substantial social safety net to provide for education, health, food and housing.

  19. Andrew

    I agree with your points – but unfortunately we don’t have good estimates of lifetime consumption across countries.

    I would describe the USA now as a small Bismarckian welfare state – its public spending as you say is now focused on social insurance for older people. Having said that much of continental Europe and particularly southern Europe probably spend a higher proportion of their higher social spending on the elderly. Also Japan.

    As you say capital gains are only counted on realisation not accrual. This suggest to me that you need a longer measure of resources over time to correctly rank people.

    I entirely agree with your last paragraph.

    Having said this, what makes the USA distinctive is its very unequal distribution of wages. Before the great recession this was partly offset by its higher employment rates, but not now.

  20. As an outsider, I find it puzzling to see the US discussion focus only on the progressivity of taxes.

    To the first order, this is because because the general party of the center-left tends to defend the existing transfers to the middle and upper middle class if anything, more than the center-right party. The Republicans have actually on several occasions proposed cutting middle and upper middle class benefits while leaving benefits to the poor alone, which the Democrats reject. There’s an enormous concentration on helping the middle class. To the extent that there’s any emphasis on redistribution, it indeed focuses on taxes alone.

    But to a second order, that’s because that’s the where the political equilibrium is.

  21. Pingback: IRS Data: Millionaires Pay 4X As Much As Middle Class In Taxes - Democrats, Republicans, Libertarians, Conservatives, Liberals, Third Parties, Left-Wing, Right-Wing, Congress, President - Page 18 - City-Data Forum

Leave a Reply

Your email address will not be published. Required fields are marked *

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>

Notify me of followup comments via e-mail. You can also subscribe without commenting.