Correlation Proves Causation, David Cay Johnston Edition

September 27, 2010

David Cay Johnston used to report on the IRS for the New York Times; now he blogs at, the public website of commercial tax publisher Tax Analysts.

Johnston’s latest piece has garnered quite a bit of attention in the tax webosphere. He uses IRS income statistics between 2000 and 2008 to conclude that the Bush tax cuts were a failure, or as he claims, to prove that they didn’t live up to their promise. Unfortunately, regardless of where you stand on the tax cuts, Johnston’s piece should be taken about as seriously as a political party press release.

Let me start by saying that I agree with Johnston that tax cuts are not the correct response to every economic situation, and I do not believe that letting the Bush tax cuts expire would cause an economic armageddon. If the federal government’s proclivity for deficit spending can’t be curbed by reducing tax revenue – the “starve-the-beast” approach – then permanently extending the Bush tax cuts for any and all taxpayers is a worse policy than letting the cuts expire because the country will drive off the fiscal cliff even sooner.

But any honest broker that reads Johnston’s piece is going to shake his/her head in wonder that such a celebrated author could point to a mere statistical correlation and call it proof that either the Bush tax cuts failed or that they didn’t live up to their promise. Now Johnston claims in responses to comments made about correlation/causation that he wasn’t trying to prove in his article that the tax cuts failed but rather that the tax cuts didn’t live up to the promises that were made by their supporters.

But that defense is both irrelevant and a lie. It’s a lie because throughout the piece Johnston explicitly claims that the tax cuts were bad, regardless of what the promised benefits were. For example, Johnston says this: “But despite that one sliver of good news about low six-figure incomes, the data show overwhelmingly that the Republican-sponsored tax cuts damaged our nation.” That is not a claim assessing promises. That is a statement about the failure of a policy.

Right after that, Johnston then says “Examining performance against the promises, what do we find? Overwhelming evidence that the tax cuts of 2001 and 2003 made us much worse off.” Saying that “the tax cuts made us much worse off” is not “examining performance against the promises.” Claiming that the tax cuts made us worse off has nothing to do with the promises, whether they were too rosy or too pessimistic. Using the word “made” means you are totally implying causation.

But regardless of whether Johnston was really assessing promises made or the economic benefits of the Bush tax cuts by themselves, the numbers he cites do not definitively prove anything. For example, Johnston does the ridiculous Republican-style comparison of revenues before and after the tax cuts, comparing 2000 with other years. Regardless of whether you think 2000 should be the comparison point, a comparison of revenue performance across years like Johnston does is evidence of shoddy work. In fact, had Johnston done this the right way, he could have actually made the case stronger that the Bush tax cuts lowered revenues. The methodology that Johnston uses to “prove” revenues fell by comparing revenue in Year A to Year B would be laughed out of the building at JCT or CBO.

Johnston then goes on to make the economic failure argument:

“The tax cuts did not spur investment. Job growth in the George W. Bush years was one-seventh that of the Clinton years. Nixon and Ford did better than Bush on jobs. Wages fell during the last administration. Average incomes fell. The number of Americans in poverty, as officially measured, hit a 16-year high last year of 43.6 million, though a National Academy of Sciences study says that the real poverty figure is closer to 51 million. Food banks are swamped. Foreclosure signs are everywhere. Americans and their governments are drowning in debt. And at the nexus of tax and healthcare, Republican ideas perpetuate a cruel and immoral system that rations healthcare — while consuming every sixth dollar in the economy and making businesses, especially small businesses, less efficient and less profitable.”

Whether Johnston is claiming this as proof the tax cuts failed or proof that the tax cuts didn’t live up to their promises, these numbers do neither. But Johnston says right after this:

“This is economic madness. It is policy divorced from empirical evidence. It is insanity because the policies are illusory and delusional. The evidence is in, and it shows beyond a shadow of a reasonable doubt that the 2001 and 2003 tax cuts failed to achieve the promised goals.”

In Johnston’s world, correlation proves causation. Nevermind that there were millions of other factors at play affecting the economic variables cited above besides tax cuts including a financial crisis that had little to do with tax policy (a lot to do with regulatory policy). Let’s be fair: Republicans do this same trick as Johnston with the stimulus bill, whether you are talking about proving that a policy has failed or proving that a policy failed to live up to its promises. For example, nevermind that CBO says things would be worse had the stimulus bill not been passed, Republicans cite the current economic struggle as evidence the stimulus failed. And they constantly cite the now infamous Christina Romer incorrect projection of unemployment with/without the stimulus as evidence that it didn’t live up to its promises, despite the fact that you should be looking at the delta caused by the stimulus. According to this logic of Johnston/Republican stimulus critics, since the Miami Heat are favored to win the NBA championship this year, if they don’t win, it’s because Lebron James was brought in, even though it could be the case that without Lebron, the Heat would have not even made the playoffs.

A rigorous analysis of either the tax cuts or the stimulus would assess what those economic variables would have been with and without the tax cuts and could assess the claims of those who supported the tax cuts. But Johnston would have none of that. For him, looking at before and after is sufficient. Not even a disclaimer about the possibility that there were other factors at play is in Johnston’s piece.

Johnston concludes his piece by saying the following: “How much more evidence do we need that we made terrible and costly mistakes in 2001 and 2003?” According to Johnston’s response to the commenter, that closing statement is somehow about promises and not a blanket statement about the Bush tax cuts that implies he has proven with the evidence presented that the Bush tax cuts were a mistake. Yeah right.

For those interested in a solid critique of the Bush tax cuts, I would suggest reading the works of Brookings economist William Gale or former OMB Director Peter Orszag.

Apparently David Cay Johnston and Terence Jeffrey took the same stats course, and it wasn’t this one:

Related Articles