(1) How regressive is the French system?
I found the French system to be pretty flat -- deciles 2-8 all paying in the low 50s percent of their incomes -- with the exception of deciles 9 and 10. For example, the average tax rate for the 10th decile was about fifteen percentage points less than the rates for deciles 2-8.
As a matter of arithmetic, most of the dip at deciles 9 and 10 comes from the fact that France gets so much revenue from the payroll tax, and my assumption that the decile-pattern of payroll tax collections is the same in France as the CBO estimated for the U.S., with, for example, the 10th decile paying at half the rate that deciles 1-8 do. (For these purposes, I consider France's "Contribution Sociale Generalise" as a separate tax with the same rate for all deciles).
As a qualitative matter, I suspect that my French result is correct because much of the payroll tax in France was capped at less than 3000 Euros per month, and thus below the average (Piketty and Saez reported a French average household income of 38,665 euros per year). Moreover, the payroll tax does not tax capital income, and the high deciles are the more capital income intensive.
Piketty and Saez (2007) looked at France's 10th decile's payroll tax rate. Their study was not designed to consider the lower deciles, and therefore not designed to calculate an all-decile average, but they reported a 10th decile rate that was 87 percent of the average (here I have subtracted off the "Contribution Sociale Generalise," which they considered as part of the payroll tax), as compared to 63 percent the CBO found for the U.S. payroll tax. You might think that 87 and 63 are different because the countries are different, but interestingly Piketty and Saez found the U.S. 10th decile payroll rate to be 100 percent of the average rate, which means (a) they disagreed with CBO about the U.S. and (b) by this metric they find the U.S. payroll tax to be less regressive than France's.
(2) Do Low Income French Really Pay that Much More?
In case you are dubious that below-average-income French pay more tax than below-average-income Americans, consider a Frence household that earns 24,000 Euro per year (inclusive of payroll taxes paid by employers on this household's members' behalf -- the same income concept used by Piketty and Saez), entirely labor income at a rate of 14 Euro per hour, and spends 20% of that on food (4,800) and spends everything else (after taxes) on consumption items. (this calculation is entirely separate of the data I used to calculate tax rates by income decile).
That French household has an annual income after employer contributions of 17,805 Euros, because French employers pay almost 35 percent payroll tax (8.3 percent to the old age program, 1.6 percent to the survivor's program, 13.23 percent to the health insurance program, 2.26 percent to the work injury program, 4 percent to the UI program, and 5.4 percent to the family allowances program; my source is here) on employees earning below the cap. Then the employee has another 9.9 percent taken out of his check for his "contributions," for a total of 7,957 Euros in payroll tax -- a third(!) of what this employee would earn if he were exempt from payroll tax.
To be conservative, I will assume this already includes the roughly 5 percent flat rate "Contribution Sociale Generalise," and this household owes ZERO individual income tax.
That leaves the household with 16,043 Euros, on which it will pay another 250 Euro on food VAT (the rate is 5.5 percent) and 1,842 on other VAT (the rate is 19.6%). So from payroll and VAT alone that is a tax bill of 10,050 Euro, or 42 percent of its income.
Compare to an American who would pay 16 percent in payroll taxes and probably less than 4 percent on retail sales taxes for a combined payroll and sales tax liability of less than 20 percent of its income.
This is yet another way to see that tax burdens on below-average-income French must be double, and probably more, the tax burdens on below-average-income Americans.
(3) Treatment of the estate tax
Last week I explained how the estate tax might be interpreted as a tax on capital income, and therefore ultimately born by labor. That's a reasonable approximation, especially in a closed economy analysis in which most of the capital employed in a country was owned by residents of that country. At the other extreme, one could think of wealthy French and wealthy Americans as owning essentially the same capital portfolios, so that the estate tax in one country would actually hurt workers all over the world, and not particularly workers in the country levying the tax.
In the open economy case, one might assign estate tax incidence according to the decile of the payer, at least for the purpose of cross-country comparisons. This adjustment would have no perceptible affect on my results for the country tax rate gaps at deciles 1-9, but it would widen the France-U.S. gap at decile 10 by 1.5-2 percentage points. As one can see in the chart I posted today, it would still remain the case that the France-U.S. gap at the 10th decile is a small fraction of what it is at deciles 1-8.