So, last year around this time, I had a mini dust-up over at Hit and Run with my old pal Andrew Chamberlain about the Tax Foundation’s “Tax Freedom Day“—the national one’s today, if you’re inclined to celebrate—which is supposed to mark the dividing-line in the year between “days Americans work to pay taxes” and “days they work to support themselves.” So far, in other words, we’ve been collectively toiling away for Uncle Sam; henceforth, we’re working for ourselves. Now, my original beef was that this is highly misleading given the massively skewed tax burden, such that people really have massively different “tax freedom days.” Andrew replied with a number of objections, which I’ll let speak for themselves, and which dovetail with the Tax Foundation’s rejoinder to some similar criticisms from the (leftish) Center on Budget and Policy Priorities. (The foundation typically characterizes the suggestion that TFD focus on the median household’s tax burden as a demand to count “the tax burden of the middle one-fifth of earners and ignore the taxes paid by the remaining four-fifths of Americans.”) I’m going to revisit the issue now to try and clarify why I still think the way they go about computing this is somewhat bogus.

TF calculates Tax Freedom Day by measuring the country’s total tax burden as a percentage of GDP Net National Product, and then translating that into a portion of the year. So this number represents the time it takes “the economy as a whole” to meet its tax burden. (They also do a breakdown by state.) And if we’re very clear on what the significance of the number here is, that’s all well and good. But it’s also sort of uninteresting: If you just want to look at the total magnitude of America’s tax burden, why is a date any more illuminating than just giving a percentage of GDP? Well, it isn’t, but let’s take a look at how TFD is typically covered in the press. Here are some clippings from this year. Here’s The Pocono Record (similar language was used by PA’s CBS station):

In other words, the average working Pennsylvanian had to labor from January 1 to April 27 — a total of 117 days, or nearly one-third of the year — just to pay federal, state and local taxes.

Here’s Doug Bandow in The American Spectator:

Americans stop paying for government today. According to the Tax Foundation, “Tax Freedom Day” is April 30. That’s when local, state, and federal officials stop fingering the average American’s wallet.

Here’s The Colorado Springs Gazette:

Because April 15 this year fell on Sunday, April 17 became the actual deadline for filing income taxes. But the day you should really be circling on your calendar is April 30. That’s Tax Freedom Day — the day average Americans quit working for the government and start working for themselves.

In other words, assuming that every dollar you made beginning Jan. 1 went to the government, your federal, state and local taxes won’t be paid off until the end of this month. Take a moment to think about the implications.

Here’s Georgia’s Walton Tribune:

Have any idea how many days you have to work to pay off your taxes here in Georgia? [….] Tax Freedom Day, the last day a taxpayer has to work before having his or her tax bill for the year covered, will come April 22 in Georgia, nearly a full month before Tax Freedom Day in other states with higher taxes and eight days before the national Tax Freedom Day.

Here’s Tax Foundation’s own descripton—which, for reasons that should be clear in a moment, is quite cagey:

Does Tax Freedom Day measure the “average” American tax burden?
Yes. The mathematical average of any group is found by adding up all the values and dividing by the number of values. Tax Freedom Day totals up tax collections and divides by national income. The result is the average tax burden for the U.S. economy as a whole. Including high-income taxpayers tends to increase the average tax burden – just as including low-income taxpayers decreases it. To be an objective mathematical average, Tax Freedom Day must include all taxpayers.

I’ll stop, since that’s already a bit tedious, but I could pull up dozens of equivalent phrasings from press around the country. Notice anything important about these accounts? Yep: TF will accurately talk about the time “the economy as a whole” takes to pay its taxes. But these accounts almost universally use language like “the average taxpayer” or “the average wage earner.” These are totally different. Any claim that TFD somehow measures this latter figure is just unambiguously false. TF’s own claim is just barely honest in a highly pedantic sense thanks to their judicious placement of quotation marks: They’ve measured the “average” American tax burden but damn if that isn’t worded as to almost invite the misreading that they’re measuring the “average American’s” tax burden. They’re not, despite widespread press coverage giving that impression, which TF seems to feel no need to correct. The easiest way to prove it is by using an imaginary population of taxpayers to illustrate.

Suppose we have 100 citizens: 20 make $20,000 per year and are taxed at 10 percent. A big middle class, 70 people, make $50,000 and are taxed at 20 percent. The top 10 earners all make $200,000 and are taxed at 30 percent. When is “Tax Freedom Day” for “the average citizen”? Well, in this case, the modal and median taxpayer are the same, and all you have to do to calculate that is take their effective tax rate as a percentage of the year: about 73 days.

But we don’t want to “ignore” the other taxpayers. So here’s another alternative: We could calculate the number of days each citizen works to pay taxes (by the same method), add them all up, then divide by the number of citizens to get a mean “Tax Freedom Day” for an “average” member of the group. Scribbling quickly on the back of a napkin here, I come up with 69.3 days as the mean for this notional population. Either of these could arguably be described as “the amount of time it takes the average American to earn enough to pay her taxes.” But you know what can’t? The number you get by the Tax Foundation method: just a hair under 83 days. Again, for our fictional population fully 90 percent of taxpayers require less time than this to pay their taxes. And the mean of the times required by all individuals is less still. Which, to my kooky way of thinking, makes it a little queer to call this the amount of time “the average taxpayer” needs.

Now, of course, I’ve just used numbers I pulled out of thin air. But if you check out the data on the actual distribution of the federal income tax burden, there’s ample reason to think that my example significantly understates the disparity between the TFD you’d get by looking at either the median household or by taking the arithmetic mean of each individual’s TFD. That’s because the more “topheavy” the actual distributon of the tax burden, the more the TF method of calculating a date will yank the number up. (Play with the numbers yourself if you like.) The distribution shown on the Cato chart linked above is skewed far more topheavy than that in my example, though it doesn’t include state taxes or more regressive taxes like FICA. But in a way, this is beside the point: As the example above should make clear, if the TF number happens to coincide with either of the numbers that could actually, defensibly be described as the amount of time “the average American” needs to meet his tax bill, it’s just that: Coincidence.

Now, I guess, as Andew charged last year, I’m not “advancing liberty” (which is to say, mobilizing people to be outraged at their tax burden) by pointing this out. But I think we have enough good arguments with which to “advance liberty” that we needn’t resort to bullshitting people with misleading figures whose significance is cloaked in Clintonian wording, on the assumption that most journalists will—whoops, not our fault! certainly not our job to correct the ubiquitous misreportings!—easily confuse the carefully-worded, technically-true claim with the more attention-grabbing but indisputably false one.

So, Tax Foundation: Would someone there be willing to post a clarification? I’d like to see someone state outright that TFD is not, on any plausible reading of the phrase, a measure of when “the average American” has met his or her tax bill, to ensure that journalists in the future can’t “misinterpret” it.

Update: Two clarifications:
(1) James Joyner has a post which seems to think I’m quibbling about whether a mean or a median is the better interpretation of “average.” That’s not it at all. Even if what you want is the mean—that is, the mean individual Tax Freedom Day—that’s not what the TF number is giving you. As I say above, that’s the number you’d get if you calculated all the personal TFDs and then averaged them, and it’s no the number you get by the TF method.

(2) Andrew replies in the comments to say he’s calculated the median American’s TFD, and in fact it’s quite close to their number—at least at the national level; I don’t know whether they’d line up as neatly for the state-level TFDs, which TF also calculates and which are also widely reported. (This similarity, by the way, interesting in itself as a measure of how little progressivity there actually is in the tax system as a whole.) Well, great. But I don’t think, as Andrew suggests, that it’s therefore ridiculous of me to make “all this fuss” about it. It’s nice that the national-level median TFD and their TFD happen to be quite close this year, but this amounts to a “fake but accurate” defense. They’re still putting out a number that they have to know is being all but universally reported as representing something that it isn’t. If it doesn’t matter because they’re so close, great: then actually give reporters the mean individual TFD, or actually give them the median American’s/household’s TFD. But if you don’t want to, or the former turns out to be impossible to calculate, then make it clear enough that’s not what you’re doing that we don’t see it misreported the majority of the time.