Turning up the heat on mainstream fact checkers

Dustin Siggins' post at the conservative blog "Hot Air" serves notice that I'm not alone in pointing out the failure of mainstream fact checkers in testing claims about effective tax rates.

Siggins:

The Obama campaign is running an ad claiming that Mitt Romney pays a lower tax rate than that of the average American.  Following the ad’s release, PolitiFact published an article rating it as “Half-True.”  From the article:

There are two main ways to make this calculation, and they lead to opposite conclusions. While we believe that including payroll taxes in the calculation offers a more accurate picture of what the American public pays the IRS, it’s also true that the Obama ad didn’t specify which measurement it was using, and in fact used a figure for Romney — 14 percent — that was based on income taxes alone. On balance, then, we rate the claim Half True.

Unfortunately for PolitiFact, their analysis completely misses the boat.

Siggins links to an analysis from "Just Facts," a not-for-profit independent fact checker:

Specifically, CBO found that households in the middle 20% of the U.S. income distribution paid an effective federal tax rate of 11.1% in 2009. Using CBO’s new estimate for allocating the burden of corporate income taxes, Just Facts and Ceterus calculate that Romney’s federal tax rate was 23.3% in 2010, which is twice the middle-income tax rate in 2009.

Just Facts published the above the same day I published my critique of Farley's fact check (Aug.7).  Their fact check goes much deeper than mine--I simply noted that Farley had short shrifted an entire body of evidence indicating that Romney almost certainly paid a higher effective tax rate than the average American. 

I criticized PolitiFact along the same lines on Aug. 10, but Siggins takes note of something I missed:

(T)he PolitiFact analysis ignores data cited by its own resource.  The article cites the Tax Policy Center to look at what tax levels are at for all income quintiles.  However, PolitiFact fails to note that the Center’s chart (the same one cited in the article) shows that the top 1% (which Romney definitely falls into) pay 7.7% of their income into the corporate tax structure.

So the PolitiFact researcher, Louis Jacobson, had the information staring him in the face and missed it or ignored it.  That's on top of somehow missing the CBO studies on effective tax rates.

Not good.


Correction Aug. 15, 2012:  Apologies to Dustin Siggins for consistently finding ways to put the vowel "a" in his name where it doesn't belong.

Former PolitiFact writer Robert Farley still spreading misinformation at Factcheck.org

Writer Robert Farley worked at PolitiFact before catching on with the more respectable Annenberg Fact Check (factcheck.org).

I've hoped that Farley would step up his game.  But the folks at Annenberg are, after all, liberally biased, so it's no surprise that Farley continues to fact check in a style similar to that he used at PolitiFact.

Our case in point:

A new ad from the Obama campaign claims that Mitt Romney “paid only 14 percent in taxes—probably less than you.” That depends. Romney paid a federal income tax rate that is higher than the income tax rate paid by 97 percent of tax filers. But if you include a combination of income taxes and payroll taxes — which make up the bulk of federal taxes for most taxpayers — the ad is accurate.

Liberal fact checkers tend to ignore a key evidence while fact checking claims about comparative income tax rates:  a report from the Congressional Budget Office that estimates effective tax rates while factoring in the effects of corporate taxes.  The tendency to ignore the CBO's report occurs despite the fact that fact checkers tend to believe the CBO almost without reservation, the classic example coming from the CBO's static analyses of the Affordable Care Act.

Google "effective tax rates" and the CBO report is right there (now on the second page of hits after a long time on the first page).  The report reveals that very high income earners pay much higher federal tax rates than liberal fact checkers claim.  Farley actually hints at the truth, perhaps without fully realizing it.

Farley (bold emphasis added):

There are all sorts of ways to slice tax data. According to the administration’s Economic Report of the President, the median effective tax rate for the middle 20 percent of U.S. taxpayers in 2012 is 13.3 percent when you include income, payroll and corporate taxes (Table 3-1). That also puts the ad’s claim in the right ballpark. But in order to get there, you have to compare Romney’s income tax only to the rates others pay in combined income, payroll and corporate taxes.

The second of the two sentences in bold makes false the claim that many people pay a greater effective rate of federal income tax (payroll taxes included) with corporate income taxes included in the mix.  Farley buries the key sentence with the lead-in "That also puts the ad's claim in the right ballpark."  The story ends up with one sentence that strongly upsets the argument about Romney's supposed low income tax rate, set in an arrangement that camouflages it sufficiently so that few are likely to recognize its meaning. 

Take a gander at Table 3-1:

click image for enlarged view

The bottom line across from "Top 1 percent" represents something of a bell curve showing the effective tax rates for the top 1 percent of earners.  It's likely that Romney pays a rate somewhere near the middle, though apparently he also gives a great deal of money to charity, which decreases his tax bill because of the charitable deduction.  One has to hunt on the chart to find effective tax rates higher than the median rate for the top 1 percent (29.6 percent). 

Yet here's the kicker quote/paraphrase Farley uses to wrap up:

No matter how you slice it, Toder said, Romney’s tax rate is very low for someone with his level of income. The average income tax rate for the top 0.1 percent (which is where Romney falls) is 23.6 percent.

Slice us up some of that horse hockey, Farley.

Fact checking is in a sad state.  But a least Annenberg Fact Check is still better than PolitiFact.

A fact-checking travesty

During the course of my most recent "Grading PolitiFact" item, I questioned a figure PolitiFact used to estimate the portion of an individual's average tax rate.

PolitiFact asked a pair of Center for Tax Policy experts to estimate the effects of payroll taxes on the tax rate of a person making about $50,000 per year.  And mine were not the only eyebrows raised by the estimate returned by Roberton Williams and Rachel Johnson.

From PolitiFact's Facebook page:

Zwack asks a good question, and we're still waiting for any sort of response from PolitiFact.

An Annenberg Fact Check (factcheck.org) story linked through the PolitiFact website helps confirm one of my two proffered solutions to the payroll tax percentage mystery.

The Annenberg item, unlike the PolitiFact story, took its calculations out of the black box and put them on display:

In all cases, when calculating combined payroll and income tax rates, we have followed the same method used by Buffett in his New York Times piece (and in general by economists and tax analysts as well). That is, we have counted the employer-paid portion of both Social Security and Medicare taxes as though they were paid by the worker, and we have also added those amounts to the worker’s income when calculating the overall rate. This is based on the economic theory that employers see the taxes as a part of total compensation, and that they would pay it as salary if the worker paid that half of the tax instead.

The above justification makes sense.  But even the Annenberg researchers do not make clear that the calculations were performed correctly.

To illustrate the problem, consider the person making $50,000 per year.  Most of us think of that figure as the person's salary--the gross amount on the paycheck stub before taxes and other deductions.  If the person making $50,000 pays a 6.2 percent payroll tax on that amount, then the effective tax portion stemming from the payroll tax is 6.2 percent--an easy calculation.  But the employer's share works differently.  If the employer's share of the payroll tax (a match of the 6.2 percent) counts as employee income then the employees is making 6.2 percent of his salary added to his $50,000 salary:  $53,100.

That total income figure, in turn, affects the person's average tax.  Instead of adding the employer's match to the employee's share to reach 12.4 percent for the share of payroll tax, we take the total tax ($6,200) and take that as a share of the $53,100 total.  Which comes to about 11.7 percent.

The same principle appears to apply to any other aspect of employee compensation, and there are quite a few of those potentially adding $10,000-$20,000 to the total compensation of the employee earning a $50,000 salary.

It isn't clear that the Annenberg fact check took all these things into account.  It's pretty clear that the PolitiFact fact check did not:

We asked two researchers at the Urban Institute-Brookings Institute Tax Policy Center, Roberton Williams and Rachel Johnson, for their advice on how to factor in payroll taxes. They estimated that combining the workers’ share of the payroll tax with the employer’s share -- the usual practice among economists -- would mean an extra 15 percentage points for our hypothetical middle-class worker, and less than 2 additional percentage points for the high-income taxpayer.

Adding these to the percentages we previously found for the income tax alone produces a new, "final" rate of 22 to 23 percent for the construction worker and 20 to 30 percent for the $50 million earner.

Williams and Johnson very probably were not responsible for the faulty equation PolitiFact uses for its calculation.  But if either Tax Policy Center expert reads their press then what stops them from pointing out PolitiFact's error?

And what stopped PolitiFact from considering total compensation in certain other fact checks from the past?

Regardless of the reasons, PolitiFact botched the calculation and called it fact checking.

The St. Petersburg Times enters the political fact-checking game

Yes, the newspaper that puts its readers supposedly "In the Know" has joined Annenberg Fact Check (factcheck.org) and Media Matters in the political fact-checking game.

The folks at Annenberg have established a solid reputation for their fact checking enterprise. Media Matters skews left to the point where the site is near worthless.

I expect "PolitiFact" (catchy name?) to perform somewhat worse than Annenberg, but better than Media Matters by quite a bit.

One feature I do not like is the graphic they use to rate the accuracy of a claim overall. The graphic tends to predispose the reader to judgment regardless of the justification used in the explanation. Looks like a decision influenced by the design department. The Annenberg method of providing a brief explanation of the findings risks less distortion.

I had heard about the project awhile back, but I was prompted to visit after the SPT posted one of its PolitiFact summaries on the front page. It featured Rudy Giulianni speaking on cancer and nationalized healthcare. I'm not sure I trust their rating.

It will also be interesting to see which fact checks rate a location on page 1.

*****