Saturday, October 15, 2011

How to Lie With Statistics: Tax Rates

A good deal of the recent rhetoric in support of Democratic proposals for raising taxes is designed to make it sound as though rich people pay federal taxes at a lower rate than everyone else. That, as one can easily check by looking at the published figures from the Congressional Budget Office, is not only false but wildly false. Most people in the bottom half of the income distribution pay no federal income tax at all, although they do pay payroll taxes and, arguably, some of the cost of corporate income tax passed on in higher prices or lower wages. On the CBO calculations, the ratio of total federal tax paid to income rises pretty much monotonically with income.

The less extreme claim, which has been getting a good deal of press of late, is that a quarter of the households with an income of at least a million dollar a year pay taxes at a lower rate than the ten percent of those with incomes of under $100,000 who pay at the highest rate. 

I have not seen any detailed explanation of how those numbers are calculated, but presumably they are based on income and tax for a single year. If so, although the claim may be literally true, it is also highly misleading—an elegant example of how to lie while telling the truth.

Income and tax liability vary for each individual from year to year. If you take a large capital loss one year, part of it carries over to reduce your taxes, but not your income, in the next year. If you have a large capital gain in one year, your taxes go up for that year but your average tax rate goes down, since capital gains are taxed at a lower rate than ordinary income. 

Some of the 25% of high income taxpayers paying at the lowest rate are people who regularly pay less taxes than most, some are taxpayers who happen to be paying a lower rate than average this year. Some of the 10% of middle income taxpayers paying at the highest rate are people who regularly pay more taxes than most, some are people who happen to be paying a higher rate this year than most years. So the widely reported calculation overstates, by how much I have no way of knowing, the spread of both distributions, both the number of middle income taxpayers who on average, year after year, are taxed at a higher rate than the bottom 25% of high income taxpayers and the number of high income taxpayers who on average are taxed at a lower rate than the top 10% of middle income taxpayers.

If the logic is not clear, consider betting on the races. Each day, a significant fraction of the bettors—say a quarter—make money. A few of them make money because they really are much better than most at guessing which horse will win. Most of them make money because that was the day that they happened to be lucky. If you looked only at the day's results, you would conclude that the top quarter make money at the races. If you looked at the year's results, you would come up with a much smaller number.

Just as, if you looked at the tax rates paid by any group of taxpayers over a period of years, you would get fewer paying a rate that was unusually high or unusually low than if you look at them for a single year.

And for readers interested in a more general account of how to lie with statistics, I have a book to recommend.

----
P.S.

Here are two summaries of federal tax incidence, one from the Tax Policy Center of Brookings and the Urban Institute, one showing the figures from the Congressional Research Service. The former shows figures for the top one percent and top tenth of a percent. At least by its calculation, the effective rate rises monatonically with income.

P.P.S.

I linked to the figure showing the Congressional Research Service numbers, which I found on Google+. That apparently didn't, or at least doesn't, work. Here is the figure:



39 comments:

Doc Merlin said...

"The less extreme claim, which has been getting a good deal of press of late, is that a quarter of the households with an income of at least a million dollar a year pay taxes at a lower rate than the ten percent of those with incomes of under $100,000 who pay at the highest rate. "

Its capital gains income versus wage income which are taxed at different rates. Its completely bogus as capital gains are double taxed via corporate taxes.

Brandon Berg said...

It's also worth noticing that the fixation on rates can in and of itself be misleading. A lot of people seem to lose sight of the fact that 25% of $1,000,000 is more than eight times as much as 30% of $100,000. Consequently you end up with people saying stupid things like "Warren Buffett pays less in taxes than his secretary."

People routinely gloss over the huge, orders-of-magnitude difference in the actual sums paid and focus on minor differences in tax rates.

We don't really do this with anything else, though. Nobody complains about the horribly regressive price structure of the market for automobiles---how a new car costs a middle-class person 25% of his annual income and a rich person only 2% of his annual income.

Anonymous said...

I apologize for my comment not being germane to your post, but I'm curious to know what you think of Herman Cain's 9-9-9 tax plan. Would the poor (however ill-defined that term is) wind up paying more in taxes? And even if that's so, would the economy overall be much more prosperous? Thanks in advance.

Brandon Berg said...

Its completely bogus as capital gains are double taxed via corporate taxes.

The jury's still out on that. There's an argument to be made that, due to the low barriers to capital mobility, labor bears some of the cost of corporate income taxes.

However, it's worth noting that the overlap between people who complain about low tax rates on investment income and the people who favor abolishing the corporate income tax is pretty much nil. Leftists have an extremely strong anti-corporate reflex that cannot be overcome without extensive training.

Tony P. said...

A good deal of recent rhetoric in opposition to Democratic proposals for cutting taxes is designed to make it sound as though not-rich people are under-taxed. Maybe they are; that's a value judgement. But the honorable thing, if you really think so, is to say so, frankly and openly.

What does David mean by "raising taxes", exactly? He is honorable enough to mention, if only in passing, that people in the "bottom half of the income distribution" do pay payroll taxes -- a fact that less-honorable rhetoricians like to ignore by talking as if the personal income tax is the only federal tax that matters. "Democratic proposals" include extending the payroll tax CUT as well as imposing a "millionaires' surtax". It's not "taxes" that Democrats propose to raise; it's certain particular taxes.

That rich people are under-taxed is a value judgement just like the one about not-rich people. Statistics are value-free. We could agree on every last statistical detail and still come to different value judgements about the "fair" way to divvy up the tax burden between the rich and the not-rich. Nothing dishonorable in that. But it is at best disingenuous to pretend, on either side, that our argument is about statistics.

--TP

Anonymous said...

This kind of discussion can be carried on at two different levels:

(1) Nominal tax rates -- where we don't consider that fact that all taxes change prices and wages.

(2) The actual incidence of taxes after factoring changes in prices and wages caused by the taxes.

Now, all the claims being made in popular debates on this subject use approach (1) -- nominal rates. So it's natural to answer the claims also by recourse to nominal rates only.

This is significant when we try to decide who pays the corporate tax. By approach (2), various different parties (workers, customers, investors, etc.) pay the corporate tax.

By approach (1), only the shareholders of a corporation pay the corporate tax (since they own the corporate funds).

Now, it's very important not to mix approaches (1) and (2) in the same discussion. Otherwise, one side gets an undue advantage. (For example, treating the payroll tax as if it's paid totally by the worker (approach(1)), but treating the corporate tax as if it's paid by various parties (approach(2)).

Now, given that the usual claims in the public debates on this subject stick to approach (1), it's appropriate to also answer those claims by sticking to approach (1).

Thus, the corporate shareholders DO pay the corporate tax (about 40% on average for US corporations). This substantially jacks up shareholders' tax rates above the personal rate of about 15%, radically altering the discussion.

Taxation of fictitious inflationary gains and restrictions on deducting capital losses also mean that shareholders pay higher effective tax rates than those shown in the tax tables.

Much of this discussion comes from Warren Buffett's claim that he pays a lower rate than his office staff.

Buffett admitted in a Charlie Rose interview that he added the payroll tax rate to his office staff's rates (~15.5%). Now, since those payroll taxes go into personal accounts (Social Security, Medicare) that will pay off in the future, one can argue that the present value of those future payments cancels out the "contributions" -- meaning that it is incorrect to include those tax rates in the calculation.

Buffett also totally ignored the corporate tax rate -- as if it were being paid by Martians.

Anonymous said...

"Thus, the corporate shareholders DO pay the corporate tax (about 40% on average for US corporations)."

That would be hard since the maximum rate is 35%. The actual average is, I believe, around 25%.

Alan said...

Anonymous wrote: "since those payroll taxes go into personal accounts (Social Security, Medicare) that will pay off in the future, one can argue that the present value of those future payments cancels out the "contributions" -- meaning that it is incorrect to include those tax rates in the calculation."

This might be valid if anyone who is working today still believes that Social Security and Medicare will be there when they retire. It also ignores the fact that lower income people are less likely to live long enough to collect these benefits anyway, and because they are taxes rather than contributions, their family will never recover those assets either.

Peter said...

I wonder if they included the inflation tax. If you get a 3% return with 2% inflation and 30% tax rate, you are effectively taxed 90% (you got 1 dollar after inflation and paid 90 cents in taxes).

Nightrunner said...

Who makes the claim that rich people pay federal taxes at a lower rate than *everyone*else*? Straw man?

David Friedman said...

Nightrunner asks: "Who makes the claim that rich people pay federal taxes at a lower rate than *everyone*else*?"

Nobody. Which is why what I wrote was "rhetoric ... is designed to make it sound as though," to describe something short of an explicit claim.

I note, for instance, that the headline of the story I linked to is "One-Fourth of Millionaires May Break 'Buffett Rule,' Study Says." That certainly makes it sound as though it is talking about people who pay at a lower rate than "middle class Americans," not at a lower rate than the rate payed by a small minority of the most highly taxed middle class Americans. That is an example of how you can imply something without actually saying it--the story itself makes it clear what the real claim is.

David Friedman said...

Tony asks: What does David mean by "raising taxes", exactly?

I mean raising taxes. The particular proposal I was referring to is what is being described as the "millionaire's tax."

I did in fact discuss payroll taxes in two blog post some time back, including the effect of cutting them, but that isn't the subject of this post. If you want to read those posts, search the blog for "payroll tax".

David Friedman said...

Anonymous asks my view on Herman Cain's 9-9-9 tax plan. As best I can tell, it would make federal taxes considerably less progressive. So far as its overall effect on the economy, that's hard to guess. It would eliminate lots of distortions due to the present tax system, but taxing firms on gross income rather than on profit would introduce a new set of distortions.

As to how the revenue it brought in would compare to the revenue from the present system I have no idea, and doubt Cain does.

David Friedman said...

Tony writes:

"But it is at best disingenuous to pretend, on either side, that our argument is about statistics."

I can't speak to "our argument," but my post was indeed about statistics--specifically about the statistical trick by which a particular statistic is used to make the system look less consistently graduated than it is.

Joe said...

"Income and tax liability vary for each individual from year to year."

IMHO that statement is less true among the lower income ranges than higher ones. Most (80%?) wage slaves in the lower group don't have investments and never or rarely report a capital gain or loss. Therefore, their incomes from year to year, and taxes, are much stabler than for those in the higher income groups.

It could also be argued that, although some in the higher groups may experience losses, others in the group have offsetting gains, so on average, the upper group's overall liability may not fluctuate that much except as it follows the business cycle.

Neolibertarian said...
This comment has been removed by the author.
Neolibertarian said...

In this instructional post, David Friedman does an excellent job demonstrating how to lie with "statistics". I use the word statistics in quotes here because for a post purporting to be about statistics, there is a paucity of actual data.

It is no surprise that an ideologue might accuse others of lying with statistics while writing even more egregious falsehoods. As several commenters have noted, different tax rates for different incomes plays a role here, one that DF omits. Despite discussing capital gains in anecdote, he fails to even note that the rate for long term capital gains is less than half the top marginal rate.

Like many right wing ideologues discussing taxation, the first misrepresentation DF commits is to discuss only federal tax rates, ignoring other governmental taxes faced by individuals. Without looking at the entire tax system, selecting one aspect or another amounts to cherry picking, also known as lying with statistics.

If DF did any research, he'd know that local taxes are in fact quite regressive. The Institute on Taxation and Economic Policy found that for 2002 state/local tax rates on the bottom quintile were 11.5%, but only 7.5% on the top quintile.

Looking at the complete tax burden, Danel Altman showed that the total tax burden for 2003 was actually quite flat, with by quintile rates of; 18%, 14, 16, 17, 19. The top quintile dose pay the highest rate, but the bottom quintile pays the second highest rate.

Addressing DF's specific topic of federal taxes, Emmanuel Saez, Thomas Piketty, and Edward Wolff analyzed the Federal Reserve's Consumer Finance Survey and found that for 2004 the tax rate those making more than $10 million was 20.1%, less than the 20.6% paid by those making $100k-200k, and the 22.3% paid by those making $1-10 million.

DF's own atrocious "analysis" of the tax system eschews using facts, figures and data and instead tries to pass of a fictional anectdote replacement for actual statistics.

George Bittlingmayer said...

Thanks to one and all for interesting points. My caution: "how to lie with statistics" assumes something that may be true, but may not. A more accurate description of what goes on might be "how to fool yourself and muddy the debate with statistics." People sort through the data until they find what they want, and really believe they're illuminating the issue. Perhaps not different from lying in its effects, but clearly something else.

Albert Ling said...

But why don't economists that measure such statistics correct for these sampling errors by average out the last 10 years of earnings or using some other statistical trick to smooth things out... Taxes seems like a pretty straightforward thing to measure since all the data is so clean and available.

NeoLibertarian, all taxes that tax based on a percentage of income are progressive. A "flat tax" is not actually flat since the rich pay more. I propose that the rich pay a bit more, since the services they get are a bit better than average by living in better policed neighborhoods, or getting into better public schools. But nothing related to income. What does total income have to do with payment of services?

Neolibertarian said...

Albert Ling:

But why don't economists that measure such statistics correct for these sampling errors...

You seem to be under the impression that there is gross variability in the aggregate tax rate based on... what I'm not sure. Economic cycles? I suspect that the variability in tax rate is more sensitive to changes in tax law than it is in long term economic cycles. The sort of economic cycles that can have significant effect on tax receipts are generational, not annual. You're welcome to present data that supports your hypothesis.

What does total income have to do with payment of services?

As I'm sure you know, one of the primary responsibilities of government is management and enforcement of property rights. To the extent that it is correct to tax income, it is only as a proxy for total property. I doubt you'll get any insurance company to charge you a flat rate regardless of the value of the property you are insuring. Likewise, the same holds true for governments.

David Friedman said...

"It could also be argued that, although some in the higher groups may experience losses, others in the group have offsetting gains, so on average, the upper group's overall liability may not fluctuate that much except as it follows the business cycle."

Indeed it could. But the statistic in question isn't about the overall liability of high income taxpayers. It's about the tax rate paid by the 25% of high income taxpayers for whom it is lowest. The fact that in the same year the same effect means that the tax rate paid by the 25% for whom it is highest is higher than average doesn't offset that. Increasing the variance of tax rate lowers the rate paid by those for whom it is lowest and raises the rate paid by those for whom it is highest.

David Friedman said...

Joe suggests that variability of tax rate from year to year is less in the lower income ranges, which for all I know is true. But the statistic I've been discussing is about the ten percent of people with incomes below $100,000/year who pay the highest tax rates, and I doubt very many of them are low income.

Anonymous said...

It all depends on the source of your income. At the personal level, "earned income" (wages, etc.) is taxed higher than dividends and long-term capgains. This feature was deliberately introduced into the system to compensate for over-taxation of stock investors. Yet, it is now being "discovered" that these stock investors pay a lower personal rate. We discovered statistically that some million-dollar people pay lower rates than some 100k-dollar people. Buffett presents this as an amazing discovery. He pays a lower rate than his own office staff! Wow, we better correct this! But once we've "corrected" it, we will likely find that tax revenue from capgains and dividends actually falls (Laffer effect). Capgains can be deferred merely by not selling. Companies can decide to reduce or eliminate dividends, due to high taxation. The Laffer effect for capgains and dividends has actually been observed in the past. So, again, a call will come to reduce the tax rates on these items. So we thrash endlessly. It's the old saw that those who are ignorant of history repeat it.

Meanwhile, capgains on home sales are generally tax-free. That must introduce a pretty distorting effect into the price structure. In-kind income from a home is also tax-free, although Milton Friedman jokingly called for a "2nd IRS" to solve that problem. The system is filled with arbitrary rules.

Nightrunner said...

> I note, for instance, that the headline of the story I linked to is "One-Fourth of Millionaires May Break 'Buffett Rule,' Study Says."

Are you going to join Brad Delong in bemoaning the incompetence and mendacity of the American mass media?

Neolibertarian said...

Regarding the addendum, DF does another excellent job of cherry picking data. His post specifically discusses the extremely rich comparing the most tax advantaged millionaires to the least tax advantaged five figure wage earners in 2006. But when it comes to data, he presents quintile data from the future (2012), which shows only aggregates irrespective of tax advantage. Of course, it's worth noting that the despite the aggregation issues, the same prescient data broken down by income shows that those making $500k-$1 million pay a lower rate than those making $200k-$500k.

If DF actually looked at the "the report by the non-partisan Congressional Research Service dated Oct. 7" mentioned in the article, he would know that tax rates are not constant over quintiles or income categories, and the most tax advantaged in an income category pay significantly less than the average tax rate.

It appears that DF has done a remarkable job of demonstrating his inability to go beyond his superfical understanding aggregate metrics in discussions of how different distributions relate to one another.

Bravo.

David Friedman said...

"If DF actually looked at the "the report by the non-partisan Congressional Research Service dated Oct. 7" mentioned in the article, he would know that tax rates are not constant over quintiles or income categories, and the most tax advantaged in an income category pay significantly less than the average tax rate."

If neolibertarian actually looked at the graph I linked to, he would notice that the vertical bars show not merely that tax rates are not constant over quintiles but just how unconstant they are, since the bars run from the tenth to the ninetieth percentile for each group.

If neolibertarian understands the meaning of average, he would not think it necessary to comment that people who are below average in a group are below average--and the graph shows by how much.

Cal said...

David, I can't fathom why your blog seems to attract a noticeable number of especially unintelligent and hostile comments (eg neolibertarian ITT). Kudos for replying to them anyway.

Neolibertarian said...

If neolibertarian actually looked at the graph I linked to

If you're referring to your link to somewhere inside google+, that is not accessible. Assuming it's from the CRS report that I linked to, you're free to specify the page number. Better yet, why not actually read it.

The next time you try to offer up a response to the work of an actual economist, maybe you can do better than basing that response on your own fictional anecdote.

For someone who's father made his name using emprical evidence to further the development of economics, you'd think you'd know that a single pretend data point just doesn't cut it.

David Friedman said...

Apparently linking to a figure on Google+ doesn't work, or at least doesn't keep working; my error. I have now pasted the figure into my blog post.

Tony P. said...

Just for curiosity, and referring to "Cluster A" and "Cluster B" in the latest graph:

How many people are in each cluster?

"Under $100K" surely describes a vast number of Americans. Only a small fraction of that vast number, I assume, fall into the "Cluster A". But it's not clear how small a fraction.

"Over $5M" surely comprises a very small number of Americans. I have no clue whether a large or small fraction of them fall into "Cluster B". Maybe most of them do not.

It's hard to tell, in other words, how many people we're talking about in each "cluster", i.e. how much of a statistical fluke (or "lie", if you prefer) the two clusters amount to.

Still, $5M is FIFTY TIMES what $100K is. It is ONE HUNDRED TIMES the median income. Statistically, the "typical" American who makes 100x (and more) the median income may indeed pay a few more percentage points of effective tax rate than the just plain typical American. But 100x the income ought to be SOME source of consolation for being so grievously put upon, no?

--TP

David Friedman said...

"But 100x the income ought to be SOME source of consolation for being so grievously put upon, no?"

The point of my post was not whether the present system is or isn't just or efficient or whatever—I don't see any good arguments for what tax rates are or are not juster than others. It was to point out that a particular factoid was misleading for an interesting reason. Quite a lot of commenters seem to have missed that.

What the graph appears to show is that effective federal tax rate rises with income up to an income of about $350,000/year, and is roughly constant thereafter. Also, that for any income group except the lowest shown, rate varies within the group by about ten percentage points from the 10th to the 90th percentile.

And, finally, that Buffett's claimed tax rate is far below the rates reported for the top income group.

Anonymous said...

I don't understand why Neolibertarian is bragging about being an economist. That's like bragging about being an alchemist.

marmico said...

There is no evidence that neoliberatarian was bragging about be an economist, but there is plenty of evidence that most of the posters, including the host, to this thread are incapable of comprehending high school arithmetic, let alone english.

1. Buffett's WSJ op-ed (Charlie Rose interview) were based on taxable income (TI) not adjusted gross income (AGI). The Oracle from Omaha only gave $20 million in charitable donations last year to reduce his AGI.

2. Friedman conflates his late to the game PPS "Buffett heart" income point with a cribbed Congressional Research Service (CRS) chart.

3. Dumbos like Merlin, Berg and their anonymous benefactors don't know that Friedman's chart (the cribbed figure 2, CRS) allocates corporate tax to individual dividends and capital gains.

4. Buffett's individual effective federal tax rate (income, payroll and corporate allocation) was less than Bosanek.

5. Buffett didn't lie. Friedman says that even though 10% of the rich pay less tax 90% of the moderates, it is a statistical lie. I say bullshit!

David Friedman said...

"Friedman says that even though 10% of the rich pay less tax 90% of the moderates,"

Actually, the claim being discussed is that a quarter of the rich pay tax at a lower rate than the 10% of the under $100,000 who pay at the highest rate. Judging by the graph, the tax rate that 90% of the lower group pay more than is about 9%, which is much less than half the rate paid by the tenth percentile of the high income group.

And, of course, paying the same rate corresponds to the rich paying much more tax.

I gather you haven't bothered to try to understand why I describe it as a case of lying with statistics.

Neolibertarian said...

I gather you haven't bothered to try to understand why I describe it as a case of lying with statistics.

It was you who fabricated the anecdote. While some people's understanding of the report may be incomplete, you are the only one presenting fabrications as reality here.

David Friedman said...

"It was you who fabricated the anecdote."

I gather that you also haven't bothered to try to understand why I describe it as a case of lying with statistics.

Neolibertarian said...

I gather that you also haven't bothered to try to understand why I describe it as a case of lying with statistics.

The idea that you can demonstrate that someone is "lying with statistics" by fabricating anecdotal data is beyond absurd. If your goal is to demonstrate your intellectual laziness, you've done an excellent job. By all means go ahead and read the report and address it's content directly with concrete discussion, instead of just making things up to suit your ideology.

fort worth accounting firms said...

The effective tax rate is important because it shows the percentage of tax an individual or company is paying on its taxable income. It differs from marginal tax rate because effective tax rate is the average percentage paid on all income rather than the percentage of tax paid on the last dollar earned. Thanks.

income tax prep said...

The VAT is considered a general tax because it is applied to all commercial activities that involve the provision of a service or the distribution of a good. However, it can also be considered a consumption tax, because it is the end user or final consumer of a product or service who ultimately pays the VAT. Thanks.