Charts Of The Day

Kevin Drum sees this and weeps.


My god! The rich aren’t paying any taxes any more! Look at how steep those lines are! Those greedy rich bastards are getting off easy!

Or, maybe if you rescale the image, you make a different point (my apologies for an inability to make a chart, this in Excel is about the limit of my skillz):


Ahh, now I see. It hasn’t changed much, has it?

But that whole line is far higher than it should be! Those greedy Washington bastards think they deserve 30% of our hard-earned money!

  • Jorge Camoes

    Actually, his chart is better than yours. We may argue that a 1% change is relevant enough to use a better chart resolution, like he does. Yours is lying because it uses an artificial scale.

    To have a clear picture of what’s really happening, series for other relevant household groups should also be displayed.

  • Brad Warbiany


    Mine does not use an artificial scale. One would state that natural bounds of percentage tax rate of a state over an individual’s income would be 0% and 100%. (One can argue that “effective tax rate” could conceivably be negative due to subsidy or over 100% due to taxation of assets/other that isn’t counted in “income”, but that’s splitting hairs).

    But that’s neither here not there. I think you’re missing the point.

    The point of the post is that fancy-looking charts tend to add weight to a blogger’s post. But when creating a chart, it must be understood that the chart is specifically designed to make a certain point.

    I suspect that Kevin Drum wanted to show a calamitous drop in tax rates paid by the top 1%. His chart certainly appears to show that. I saw through this, and posted a similar chart that shows that the drop is there but tax rates on the rich haven’t fallen off a cliff.

    A 1% drop in effective tax rate may be relevant, or it may not. Its impact on the federal budget and its impact on the rich are not addressed by Kevin Drum’s post (or by mine). You can claim Drum’s chart overstates the relevance and mine understates it — which is probably true. It’s obvious that I’m doing this deliberately. It may be less obvious whether Drum is doing this deliberately.

  • Jorge Camoes

    Brad,you are using an artificial scale (deliberately…). You can argue that the Y scale should start at zero by definition, but you can’t extend the upper limit to its “natural bounds” just to flatten the trend. I would say that, for a single chart (there may be special cases) if the upper limit is more than 20% the highest data point then the scale is not correct (for whatever purpose).

    I know what the point is. But the first chart is not misleading because of the way it manages the Y scale. It is misleading because what it shows is a half true.

    Did you know that the total effective federal tax rate for the lower 20% was in 2006 53% of the tax rate in 1993? On the other hand, did you know that the tax rate for the top 20% was in 2006 96% of the tax rate in 1993? There was almost no change in this case.

    But if you split the top groups, you’ll see that the tax rate changed to 96% for the top 10%, 95% for the top 5% and 90% for the top 1%. Chart the time series and you’ll see that something changed in 1997 regarding the tax rate for the top 1%, compared with the top 20%, top 10% and top 5%. After 1996 they do get a more favorable tax rate.

    Bottom line: Look at the data and tell your own story.

  • tarran


    I think your argument is silly:
    1) You argue that Brad’s chart is “lying” for using a scale of 0 – 100, which constitutes the entire possible range of values.
    2) You argue that Brad’s chart wouldn’t be lying if it ranged from 0 – 43 satisfying your (arbitrary) 1.2 X highest data point rule.
    3) You argue that Kevin Drum is being accurate while bemoaning the size of the change in some value while scaling it so that the change comprises the entire y-axis.

    If Brad’s scaling is misleading, then Ezra Klein’s scaling is absolutely misleading; if he had applied the same process to a change of 0.5 percent between the max and the min values, you’d still get the appearance of a dramatic drop.

    I don’t think Brad’s scale is misleading; it demonstrates quite effectively how small the change really is, especially when considering that historically that tax rate was up around 95 percent.

  • Brad Warbiany


    Again, my point is that Kevin Drum chose the scale to suit the point he was trying to make (actually, I think he pulled that chart from another source, but again he used it because it showed the point he was trying to make). I chose the scale I chose to suit the point that I was trying to make — that changes in the effective tax rate are overstated by use of a graph like Drum’s.

    And I agree with you that there are a lot of far more substantive issues with tax policy where some are getting far better deals than others (especially folks such as that top 1%, who are more capable of lobbying the tax code than people who are merely top 5% or 10%). If you spend more time here at The Liberty Papers, you’ll see that we’re more than willing to discuss such matters — I didn’t want to get into it on a post.

    And I clicked over to your site — I’ll grant that you’re far more of an expert on charts than I am. But I think you’d agree that charts are not simply a visual representation of data, charts are an active component of the message. In this case, Kevin Drum and I are trying to put forward very different messages, and the charts we would produce are built for that purpose.