EDIT: Previous post had the axis backwards.
Here is the correlation between share of income and the share of taxes within the OECD
The US is the point right at the tip of the trend line.
The two upper outliers are Poland on the left and Italy on the right. In any case the US is more or less right on target.
Oh and I know that the lack of any context makes this blog even more insidery and opaque than it already is but, it is what it is. I wanted to make the the point but am trying to finish something completely different.
16 comments
Comments feed for this article
Monday ~ March 21st, 2011 at 4:30 pm
Adam Ozimek
I think the axis are reversed, e.g. in the table the richest pay 45% of the income taxes in the U.S.
Monday ~ March 21st, 2011 at 4:35 pm
James Oswald
I’m surprised the majority is above the 1:1 line. I would have guessed a far more progressive tax structure. Does the chart include capital income?
Monday ~ March 21st, 2011 at 4:36 pm
James Oswald
Edit: It makes more sense now that the axis have been fixed.
Monday ~ March 21st, 2011 at 4:36 pm
Travis Wiebe
I think Yglesias gets a good point in on the analysis: http://goo.gl/2znaJ Basically, you can’t do a country by country comparison of overall tax burden by only looking at US Federal Income Tax. Skipping FICA, state, and local taxes takes away the ability to draw firm conclusions.
Monday ~ March 21st, 2011 at 8:05 pm
Reid Kelley
Except that he misstates the fact that the chart does include FICA taxes, while using it to claim that Mankiw is deliberately attempting to mislead people.
Monday ~ March 21st, 2011 at 6:12 pm
marmico
CBO data for 2007 indicates that the top decile reports 42% of pretax income and pays 55% of all federal taxes. Although the CBO percentages are higher than those in the OECD report, the ratio of federal taxes to income is in the ballpark.
The federal tax/income ratio has been relatively consistent since 1979.
Mulligan was attempting to quantify state and local taxes in the NYT Economix last December.
Tuesday ~ March 22nd, 2011 at 12:00 am
William Bruntrager
Axes are fixed but why would you run the regression that way? If you run it the other way the US is an outlier. What did I do wrong?
http://tatulln.files.wordpress.com/2011/03/decile_regression.png
Tuesday ~ March 22nd, 2011 at 12:47 pm
RSA
I think that’s exactly right. If we’re interested in whether the top 10% of households are paying more or less in federal taxes than what’s the case for the OECD, relative to their share of income, then the x-axis should be the latter. Further, we might run the regression excluding the U.S.; it turns out it’s even more of an outlier (depending on your definition of outlier).
Tuesday ~ March 22nd, 2011 at 2:15 pm
William Bruntrager
Right, excluding the US in the linear regression probably makes even more sense. Here is the chart and OLS line with US excluded:
http://tatulln.files.wordpress.com/2011/03/deciles_2.png
Tuesday ~ March 22nd, 2011 at 3:06 pm
Johnnie Linn
William:
Same question as for Area Man just below: since income of top 10% income earners is not defined for x < 10%, does intercept term of regression have meaning? Would it be better to run the regressions as percentage of tax paid versus ***taxable income*** of top 10% income earners?
Tuesday ~ March 22nd, 2011 at 10:30 pm
William Bruntrager
Don’t really see the problem, Johnnie Linn. No, the intercept doesn’t have meaning here, it also wouldn’t mean much to extrapolate out to where tax share paid by the top decile is more than 100%. But for making predictions within the range of the sample I don’t see these as problems. Running the regression you suggest would make it hard to compare across countries.
Tuesday ~ March 22nd, 2011 at 1:59 am
Area Man
I don’t understand what “on target” is supposed to mean in this case.
Obviously, if you increase the share of income going to the richest 10%, the share of total taxes this group pays goes up, regardless of the progressiveness of the tax structure. If you have a progressive tax structure, the percentage of total income paid by the rich should not scale proportionally, so countries where income in skewed more toward the wealthy should have increasingly higher percentages of total taxes paid by that group.
Maybe the US is similar to other countries in regards to the progressiveness of its tax structure, and maybe not, but this graph can’t answer that.
Tuesday ~ March 22nd, 2011 at 9:40 am
Johnnie Linn
This regression has an intercept value of approximately 16%. How do we interpret this intercept term, since the vertical axis by definition cannot have a value less than 10%? Does a linear regression imply no lower bound to the vertical axis?
Tuesday ~ March 22nd, 2011 at 9:43 am
Barry
Kevin Drum’s comments are in: http://motherjones.com/kevin-drum/2011/03/who-you-calling-progressive
My version is shorter: nothing Mankiw says is trustworthy, until verified by a non-hack.
Tuesday ~ March 22nd, 2011 at 11:58 am
Area Man
For what it’s worth, 33% of income going to the richest 10% in the US doesn’t sound right. Every other source I’ve seen puts it at 45-50%. Even the Tax Foundation’s own chart has it at 45.8%:
http://www.taxfoundation.org/news/show/250.html
Friday ~ April 8th, 2011 at 8:55 am
spragus
-The US has a high gini coefficient – much higher than say Sweden, in fact it’s more like a third world country level. This seems to be reflected in the data/chart. So like Willie Smith said, “that’s where the money is”. The IRS seems to have figured that out.
-I think one has to be sure the data are comparable and measure what you think it measures. Reading the note by the fellow who created the data for the table, suggests to me that what he may have had a more narrow goal than I might want to measure.
-In line with this, does Italy, for instance stand out as an outlier as a result of its (by reputation) large underground economy? Unreported income does not get taxed. And how should bribes paid to government officials be treated?