Nobody From Nowhere (@i8dc)

Occasional Common Sense

Warren Buffett’s Secretary – Fact Checking the Fact Checkers

with 6 comments

This post builds on this previous one.

It took a presidential speech announcing as-yet-undefined “Buffett Rule,” but the Internets are now abuzz with commentary about Warren Buffett’s NYT op-ed wherein he says his average federal tax rate is 17.4% and the next-lowest in his office is 33%. Here’s a sampling:

“Warren Buffett’s anecdotal “evidence”… seems pretty far-fetched and not consistent with: a) average federal income tax rates available from the IRS, nor b) average tax rates for all federal taxes paid, from the CBO.” — Professor Mark Perry, CARPE DIEM

“The entire Buffett Rule premise is false, as the nearby table shows. In 2008, the last year for which such data are available, the IRS reports that those who made more than $1 million in adjusted gross income paid an average income tax rate of 23.3%.” — Wall Street Journal

“[B]oth the White House and Buffett are misleading you, because IRS data show the rich already pay more in federal taxes than the lower brackets.” — Elizabeth MacDonald, FOXBusiness

“President Barack Obama makes it sound as if there are millionaires all over America paying taxes at lower rates than their secretaries.” — Stephen Ohlemacher, Associated Press [note – the AP revised their lede replacing the above quote with this: “President Barack Obama says he wants to make sure millionaires are taxed at higher rates than their secretaries. The data say they already are.”]

“He repeated the old half-truth about millionaires not paying as much in taxes as their secretaries. (In reality, the top 10 percent of earners pay nearly 70 percent of all income taxes, according to the I.R.S. People in the richest 1 percent pay 31 percent of their income to the federal government while the average worker pays less than 14 percent, according to the Congressional Budget Office.)” — David Brooks, New York Times

“Multiple times in recent weeks, President Obama has justified the need for higher taxes by saying that Warren Buffett pays a lower tax rate than his secretary. Of course, this is a preposterous talking point.” — Mark Hemingway, The Weekly Standard
When you add up all of the various taxes, and look at the effective tax rates, it is clear the tax system is already pretty progressive. Everyone pays some tax, even those who pay no federal income taxes, and the wealthiest pay a larger percentage share of taxes. ” — Glenn Kessler, Washington Post Fact Checker

The problem is, not a single one of these pundits got the story right. Some are dyed-in-the-wool Republicans whose ideology causes them to reflexively scream out “you lie!” whenever the president speaks. But others are careful and thoughtful, and just didn’t read closely enough, dig deep enough, or think hard enough to piece it together.

I wonder about the hubris of some of these folks, too. I mean, when they didn’t think things added up at first glance, did they really think that Buffett had something published in the New York Times that wasn’t adequately fact-checked?

“Buffett should know better than to make that kind of claim without checking his facts first.” — Ed Morrissey,

Oh. Apparently so. But I guess that’s better than claiming Buffett just got confused about the difference between average or effective tax rates and marginal tax rates.

“[T]he 41 percent top tax rate he ascribed to his fellow workers appears to be a marginal rather than an average rate. ” — Roberton Williams, Tax Policy Center

Ahem. Okay, so they think he’s gotten confused in his old age. At least they’re not saying that he made the numbers up.

“It’s almost as if Barack Obama and Warren Buffett made those numbers up.” — Stephen Moore on CNBC’s The Kudlow Report, via Mark Perry

Sigh. The criticism seems to boil down to three:

  1. Buffett either lied or made grievous errors in his op-ed.
  2. Buffett’s op-ed may be factually accurate, but his effective tax rate is extraordinarily atypical of the super-rich.
  3. The “Buffett Rule” would raise taxes on all making more than $1M/year.

Criticism #1: Buffett either lied or made grievous errors in his op-ed.

First, lest you think I’m just a partisan hack of a different stripe, I’ll start by showing that Buffett did fudge the numbers. He just did it in a way that not one pundit I’ve read has picked up on. Here’s what he did:

  1. The percentages in Buffett’s op-ed are taxes divided by taxable income. This is okay when talking about income tax (which is based on taxable income), but not when talking about payroll taxes (which are based on gross income).
  2. Buffett adds in payroll taxes. But payroll taxes are based not on taxable income, but on gross income, so there’s an apples-to-bananas problem here. And the apples of the payroll tax appear to get bigger as the bananas (taxable income) gets smaller. The more deductions and exemptions, the greater the impact of payroll taxes on Buffet’s effective tax rate figure.
  3. Buffett includes not just the employee side of payroll taxes, but also the employer share. This is fairly standard and supportable in principal, but it doubles the distortionary effect of using taxable income as the denominator.

These sleights of hand tend to improve Buffett’s bottom line numbers. On to the math.

Buffett’s 17.4% is easy. Based on the $6,938,744 he says he paid in tax and oversimplifying a bit, I can estimate that about $5 million of his income was regular and taxed at up to 35%, and about $35 million was long-term capital gains, taxed at 15%. A miniscule portion of his income would be subject to payroll taxes.

Now on to folks in Buffett’s office. As I said, the inclusion of all payroll taxes and use of taxable income are big deals here. Take for example an unmarried secretary with a $50,000 salary and no other income who takes the standard deduction; her taxable income is $40,650 and her income tax is $5,944 after the Making Work Pay credit (using the tax calculator here). She pays $3,650 in payroll taxes, as does her employer. Here’s how the various effective tax measures break down:

Taxes Included

Divided by

Effective Rate

Income Only



Income Only

Taxable Income


Income and Payroll



Income and Payroll

Taxable Income


Income, both sides of payroll



Income, both sides of payroll

Taxable Income


This last one seems to be the figure Buffett used.

Do the same math for a married person with a spouse at home taking care of two kids, with a salary of $100,000 and $20,000 of itemized deductions. Taxable income is $65,400, payroll taxes are $7,650, and income tax owed is $6,173 after the $800 Making Work Pay and the $2,000 child tax credits. Same calculations as above:

Taxes Included

Divided by

Effective Rate

Income Only



Income Only

Taxable Income


Income and Payroll



Income and Payroll

Taxable Income


Income, both sides of payroll



Income, both sides of payroll

Taxable Income


This one really has a dramatic different between using AGI and taxable income as the denominator, showing how important using taxable income was to Buffett’s argument. One more, just to show that the 41% of taxable income quoted is possible: Single taxpayer making $50,000 who just bought a house, has itemized deductions of $19,000. Taxable income is $27,350, payroll taxes are $3,825, and income tax owed is $3,284.

Taxes Included

Divided by

Effective Rate

Income Tax Only



Income Tax Only

Taxable Income


Income and Payroll



Income and Payroll

Taxable Income


Income, both sides of payroll



Income, both sides of payroll

Taxable Income


Again, note the great disparity between the percentages when taxable income is used instead of AGI.

Also note that it’s not critical to the argument – Buffett’s assertion was that the government collects a lower percentage of his income than it does of people in the middle class. In all cases above, when both halves of payroll taxes are included and AGI is the denominator, Buffett’s assertion holds.

The fact-checking conclusion here is mostly true or one pinocchio. What Buffett wrote was factually true. But using taxable income and payroll taxes together is not a legitimate way to get there, and unnecessarily overstated the case.

Criticism #2: Buffett’s op-ed may be factually accurate, but his effective tax rate is extraordinarily atypical of the super-rich.

Most of the critiques here are falling short because they don’t look at enough data. Here’s what the Post’s Glenn Kessler showed, from CBO data:

Lowest quintile (23.4 million taxpayers), zero to $18,900: 4.3 percent

Second lowest quintile (22.4 million), $18,900-$32,100: 10.2 percent

Middle quintile (22.9 million), $32,100-$47,400: 14.2 percent

Fourth quintile (23 million), $47,400-$71,200: 17.6 percent

Highest quintile (23.6 million), above $71,200: 25.8 percent

Top 10 percent (12 million), minimum income of $98,100: 27.5 percent

Top 5 percent (5.9 million), minimum income of $134,400: 29 percent

Top 1 percent (1.1 million), minimum income of $332,300: 31.2 percent

The problem is that the data at the top aren’t granular enough. When you break out the fractions of the top 1%, you start to see that Buffett’s not an outlier at all. The following are average effective income tax rates, not including payroll taxes at all, as a percentage of AGI based on 2008 IRS data from here and here.

Income range Returns Percentile Tax/AGI
All returns 142,450,569 13.6%
No adjusted gross income 2,489,989 0.0% -2.2%
$1 under $5,000 11,638,707 1.7% 5.0%
$5,000 under $10,000 12,139,638 9.9% 2.6%
$10,000 under $15,000 11,702,056 18.4% 2.9%
$15,000 under $20,000 11,076,002 26.7% 3.8%
$20,000 under $25,000 9,866,247 34.4% 5.2%
$25,000 under $30,000 8,743,581 41.4% 6.2%
$30,000 under $40,000 14,554,280 47.5% 6.8%
$40,000 under $50,000 11,087,123 57.7% 7.5%
$50,000 under $75,000 19,196,461 65.5% 8.5%
$75,000 under $100,000 11,729,485 79.0% 9.3%
$100,000 under $200,000 13,851,341 87.2% 12.7%
$200,000 under $500,000 3,476,747 96.9% 19.6%
$500,000 under $1,000,000 577,618 99.4% 24.1%
$1,000,000 under $1,500,000 140,635 99.8% 24.8%
$1,500,000 under $2,000,000 59,460 99.87% 25.0%
$2,000,000 under $5,000,000 86,329 99.91% 24.8%
$5,000,000 under $10,000,000 21,390 99.98% 24.0%
$10,000,000 under $109,736,000 13,080 99.991% 22.2%
$109,736,000 or more 400 99.9997% 18.1%

At about $2M, the average tax paid by successively higher income groups declines. This is completely hidden when larger groups, like the top 1% of taxpayers, are used. And the extent of the shift is hidden even when the Top 0.1% is used; you can’t see from that data that the average tax rate paid by the Top 400 taxpayers is lower than the average tax paid by taxpayers making between $200K and $500K.

Paul Krugman highlighted a release yesterday from the Tax Policy Center that gets even more granular, but unfortunately only at the lower income levels. It’s useful, though, because it estimates the distributions within each income group. Krugman’s description:

“Consider, in particular, the estimates of the combined income and payroll tax by income class; I’ve deleted a couple of columns to make the thing fit with more or less legible type:

“Here’s how to read this: 40 percent of taxpayers with incomes between 30K and 40K pay more than 12.9 percent of their income in income and payroll taxes; meanwhile, 25 percent of people with incomes over $1M pay less than 12.6 percent of their income in these taxes. This suggests that there are a lot of very-high-income guys paying a lower tax rate than their secretaries.”

So perhaps Buffett is an exception; but as Krugman demonstrates, there are an awful lot of exceptions out there.

Criticism #3: The “Buffett Rule” would raise taxes on all taxpayers making more than $1M/year.

Buffett clearly advocated raising taxes on incomes in excess of $1M. Obama hasn’t yet passed the trial balloon stage, but it seems that what’s evolving is an AMT-like rate floor on incomes above certain thresholds (like $1M and $10M). So no longer will the wealthy’s capital gains be taxed at the 15% rate that’s been in effect since 2003. Note that the 15% rate was unprecedentedly low in the post-WWII era, so increasing it would hardly be revolutionary. Top-end capital gains were taxed at 28% under Reagan.

Bottom line:

Most of these pundits are just fulfilling their normal partisan niches when they write this stuff. But others should know better. In particular, Ph.D. economists should be better than this. I’m also a little disappointed that The Post’s Glenn Kessler apparently didn’t reach out to Buffett’s people for some help understanding the numbers. Because Buffett’s on the WPO board, perhaps nobody at the Post calls him. But given that Kessler was writing after this storm had arisen and dozens in the media were off on wild tangents, an Omaha high priest or priestess might have been forthcoming with a record-straightening conversation.

As always, stay skeptical.

Update, 9/23/11

An otherwise disagreeable poster going by “methinks” at Mark Perry’s Carpe Diem blog provided a link to a Greg Mankiw post that linked to this CBO analysis of 2005 data, which breaks down total federal tax impacts in the top end of the spectrum. Unsurprisingly, when the impacts of all investment taxes are included, the tax share and effective tax rates at the top increase. It remains true that the 2003 cuts in individual income taxes on capital gains and dividends have made the system regressive at the top, but only at the very top; the top 0.01% pay a lower overall rate than do those in the top .01-05% range.

Krugman talks about this data today.


Written by David Clayton

September 22, 2011 at 7:01 am

Posted in Debunkery, Punditry

6 Responses

Subscribe to comments with RSS.

  1. I followed the link here from Carpe Diem. I am obviously not a professional economist, but I do appreciate the use of some basic math in this discussion, so thank you. I have a few questions/comments on your analysis of Criticism #1:

    1) I understand that you were trying to demonstrate how Buffett could have reached the effective tax numbers for his employees, but how can a calculation of effective tax rates be based on taxable income? Just your examples show how much taxable income can vary for two people with identical salaries.

    2) Why is the employers contribution of the payroll tax counted as tax but not also as income? I understand the underlying assumption that the employee actually pays the employers tax, but treating it as a tax only significantly distorts the effective tax rate numbers especially when already distorting it by using taxable income in the denominator.

    3) How do person 1 and person 3 both have the same salary but different payroll taxes?

    Buffett is not an idiot; he didn’t get his numebrs confused. He was intentionally trying to be misleading. His point is valid. (criticism #2 shows this – the chart with percentiles is informative. 1st time I have seen the data presented that way) His point, which I think is:

    “The tax code allows very rich people, who get a significant portion of their income from capital gains and dividends, to pay a much lower effective tax rate on their total income than someone whose income is from wages or salary”

    Right? Why did he need to confuse the issue by fudging the numbers and using a confusing, opaque, unverifiable, anecdotal story about his secretary?

    I doubt you care, but I’ve added this to one of the blogs I follow.


    October 3, 2011 at 12:26 pm

    • Thanks for your comment. You’ve got good points.
      Whenever anybody talks about tax rates, it’s important to pay attention to definitions. Usually, “effective tax rate” is defined as all taxes divided by taxable income. So, from a strict definition, Buffett’s math is easily defensible, since “taxable income” generally means AGI less deductions and adjustments.
      But for payroll taxes, taxable income is different. In one way, it’s 100% of earned income (assuming for this note that all wages/salary is under the ~$107K ceiling on Social Security taxes). In another way, as you note, it’s 107.65% of earned income – you can’t add the employer’s share to taxes paid without also including it in employee compensation.
      Buffett’s fudge is that he uses the common taxable income definition while adding in taxes that have a different taxable income basis. One could likely create a logical blended taxable income measure, but the simplicity of the argument is one of its strong points.
      Buffett’s been making this argument for years. Personally, I think the fudge was a mistake, as the argument’s strong without it. That Buffett’s effective tax rate is lower than those who work for him is shocking enough; that it’s less than half doesn’t add that much to the story.
      The real take-away from all of this discussion is that the 15% tax rate on capital income has resulted in regressivity at the top of the income scale. There’s a solid counter-argument to this, but it’s NOT:

      • Buffett’s math was wrong.
      • Buffett should just send a check to the government if he’s for higher taxes.
      • Buffett’s an outlier.
      • Buffett’s an out-and-out liar.

      Your point #3 looks right – I’ll take a look at the numbers later and post a correction.

      And I care quite a lot that you’ve chosen to spend your precious electrons following this blog. Thanks.

      David Clayton

      October 3, 2011 at 1:31 pm

      • “Usually, “effective tax rate” is defined as all taxes divided by taxable income. So, from a strict definition, Buffett’s math is easily defensible, since “taxable income” generally means AGI less deductions and adjustments.”

        I’m not trying to sound like a 3rd grader, but that is a stupid definition. Who came up with that. Normal people who pay taxes and try to figure out what percentage of their income went to the federal government do a much simpler calculation:

        How much did I send to the government / How much did I make

        The fact that the more deductions (tax loopholes as the people like the call them) you use the higher your “effective” tax rate is makes no sense? This is why I’m an engineer not economist.

        While it is possible that people like Warren can pay a lower effective rate than many lower income tax filers, it’s not the most significant example of unfairness in the tax code. Your household examples better illustrate the inherent injustice in the tax code. Why do 1 and 3 pay such a different tax rate (no matter how you calculate it) even though they have the same salary and why does household 2 pay the lowest rates of the 3 even though they have the highest income. That’s the problem, and this debate over whether to have a Buffett plan or not is a distraction.

        As for Warren, you’re still being too kind. He knew that he was oversimplifying and being intentionally misleading. If he really wanted to clarify what he meant, CNBC would let him host their entire lineup for a week and he could easily take out a full page add in the NY Times and Wall Street Journal. He knows exactly what his is doing and has an agenda.


        October 3, 2011 at 7:15 pm

      • Effective tax rate criticism – can’t argue with you. As the tax code has gotten more and more convoluted, the less this method for effective tax rate makes sense. For the record, it’s not my definition – it’s quite common if not universal. The problem in this instance is that taxable income for payroll taxes is gross earned income.

        As for agendas, it sounds so sinister. Can’t we hold a belief without having some self-serving, hidden objective driving us? Buffett’s been making the same argument for years. He’s been consistent in questioning tax policies that don’t make sense to him – remember when he caused a small problem for Schwarzenegger (whom Buffett was advising) when he spoke out about the foolish property tax law in California?

        David Clayton

        October 4, 2011 at 12:25 am

      • David, the request for restraint in impugning motives is honorable, and would also reflect charitably upon you had you not unleashed such viscera in your original post. As is, it just makes you seem hypocritical.

        FWIW, a simple-minded lay person like myself would interpret the phrase “effective tax rate” as “how much tax I paid” / “how much I earned”. I struggle to find any useful value in the calculation you describe.

        Earle Williams

        October 13, 2011 at 6:37 pm

      • This comment refers to another post.
        I don’t know, such viscera? My intent was that Perry’s chart showed a significantly distorted view of tax burdens, and could only have occurred if he was being dishonest, lazy, or thoughtless (though “unthinking” would have been a better word choice). My subheading could have been read to mean that I meant Perry is generally dishonest, lazy, and/or thoughtless, so I modified it so that the description clearly applied only to that chart. I modified nothing else in the post (as opposed, for example, to Perry’s modification of this post, where he deleted his original last sentence: “Instead of raising tax rates, we should probably figure out what kind of loopholes allow Warren Buffet to pay taxes of only 17.4% on his $40 million income last year.” This sentence, of course, is excellent support for a Buffett rule.)

        But I stand by my assessment of the chart. I provided a discussion of my thought process there, and invite other explanations for Perry’s motivation.

        David Clayton

        October 15, 2011 at 1:19 pm

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: