Fishy Figures

Chatterbox at Slate is fascinated by this New York Times article, which claims to show that -- once all forms of taxation are taken into account -- all income groups in this country pay approximately the same average tax rate. I won't take issue with that claim, because it might well be true; after all, Social Security and other payroll taxes are well known to be regressive, as are excise taxes aimed at products disproportionately consumed by the poor (such as liquor and tobacco).

What's been bothering me, though, is this chart attached to the article, which purports to show the average tax rates for all income quintiles. Notice that the very highest percentage is still only 19%, implying that nobody in this country -- neither the rich nor the poor nor the middle class -- pays more than a fifth of their income in taxes of any form to any government. Now, that sounds awful fishy to me, because the usual estimates run at least into the low thirties. According to the Tax Foundation, Tax Freedom Day in 2002 fell on April 27 last year, because Americans paid 32.1% of their income to federal, state, or local government. (April 27 is just about 32% of the way through the year, you see.)

What could explain the discrepancy? I must confess that I've failed to nail down the answer. I figure it has something to do with different sources of data; the NYT graph draws on the BLS's Consumer Expenditure Survey, while the Tax Foundation's figures come from the BEA's National Income and Product Accounts. Also, they are probably using different notions of what constitutes "income." The Tax Foundation uses Net National Product to calculate Tax Freedom Day. I have no idea what the NYT article used, but for it to have produced lower average tax rates, it would have to have employed a broader definition of income. But NNP is already pretty broad (broader than, for instance, Net Income or Personal Income, two other figures provided by the NIPA). GNP and GDP are larger, but not, I think, enough larger to justify the substantial difference in implied tax rates.

The discrepancy might have something to do with exclusion/inclusion of taxes on business. But in order for the NYT's figures to result in lower average tax rates, they would have to rely on a *narrower* definition of taxes -- and that doesn't make sense, because the whole point of the article was to discuss the burden of taxes from the most inclusive standpoint possible, taking into account all forms of double taxation. The chart's subtitle starts with, "Taking all types of government taxes into consideration…," implying an inclusive approach. Including sales taxes, property taxes, sin taxes, etc., while leaving out business taxes would be disingenuous at best.

So I'm still stumped on finding an explanation for the discrepancy. I will say this: something is mighty strange about the Consumer Expenditure Survey's results. For instance, this table indicates that the average survey respondent had pre-tax income of $47,507 and after-tax income of $44,587, implying a tax rate of only 6.2%. Even in the richest quintile, the table implies only an 8.7% tax rate. As that short dude in "The Princess Bride" would have said, "Inconceivable!" (Of course, he eventually found that word didn't mean quite what he thought it meant.)

Like most disputes of this nature, I imagine this all comes down to a matter of definition. Most likely, it has something to do with the exclusion/inclusion of certain types of tax, plus some bizarre definition(s) of income, plus the unavoidable weirdness that comes from any survey-based data. But damned if I can pin it down. As I've wasted far more time thinking about this than I can afford, I'm posting it here in the hope that someone else will figure it out.