Delicious take-downs are always a treat. From my perspective they rarely get better than when one professional accurately describes the deficencies in a peer's writing, the glaring flaws in their assumptions - and do so with numbers.
Rarer yet the take-down which in which the proponent pins, firmly, to their opponent a label which should disqualify them from ever being a trusted source of analysis.
In two recent posts on his blog Brad DeLong has done exactly that to Martin Feldstein. In essence, DeLong pins firmly to Martin Feldstein the label "math challenged" - which is not a label anyone who works with numbers would ever want aptly describing their writing. Certainly not an economist, a Professor of Economics no less. At Harvard. With prior gigs advising Presidents past, present, or aspirational.
The two posts Brad DeLong made are in response to an piece which Martin Feldstein wrote for the WSJ. The essay presented Feldstein's argument for why the Romney budget would, and could, generate as much revenue as needed to cover the tax cuts proposed. For the rich. Again.
Some highlights from DeLong's essentially brutal take-downs below the florid orange swirl - along with Feldstein, in the original.
Before We Go Further Another Oddity in Feldstein's Article
First I'd like to address another oddity in Feldstein's piece. I must confess when I read the original at WSJ the numbers were not the only thing which didn't immediately add up for me. Feldstein asserts that in choosing a model, which he terms speculative, the Tax Policy Center's analysis of the Romney plan is inherently flawed. Thus Feldstein chooses to use an actual year of tax returns for his baseline - because that will avoid any speculative trap. The irony that any analysis of Romney's plan is inherently speculative, because the plan is mostly hand-waving and lacks important details isn't lost on me, and surely you the reader. But, rather than the contesting arguments presented by Feldstein's analysis and that of the Tax Policy Center, what also struck me as odd was his assertion as to why he chose, of all years, 2009 for his baseline.
To avoid the resulting uncertainties, I decided to analyze the Romney plan using the most recent IRS data, which is based on tax returns for 2009 and published in the current issue of the IRS quarterly publication. (Although 2009 was a low-income year because of the recession, using that year is preferable to looking back to some earlier period.)
That struck me as odd. I'm used to an assertion of preference being supported, at least in passing. Here Feldstein simply asserts his preference as plain on its face. So it begs the question. Why, precisely, would 2009 be preferable to any other recent earlier year? Would the extent of the revenue loss from implementing the Romney plan be worse in another earlier year - thus making the shortfall yet more difficult to make up through exclusions of various tax credits or deductions? Would it make the Romney plan (as faithfully divined by Feldstein) appear insanely good? Perhaps someone reading this might have a reasonable interpretation - beyond it being the last year for which fully published data is available. Which last rationale is the only one plainly on view in Feldstein's WSJ article.
Brad DeLong's Post 1 - In Which He Points Up TPC's Take-Down of Feldstein's Flawed Assumptions
It was not merely the math that Brad DeLong called into question. In the first of his posts DeLong gives voice to concerns about assumptions made or implied in Feldstein's WSJ writing. DeLong is not alone in taking issue with some sloppiness on display at the WSJ and copies in responses from analysts for the Tax Policy Center to his post - Tax Policy Center Feldstein's Analysis Confirms TPC Findings highlighting issues with those assumptions.
Feldstein [also] employs several questionable assumptions that understate the revenue loss of Governor Romney’s tax cuts and overstate the revenue gains from reducing tax breaks and deductions…. Feldstein’s version of the Romney proposals would not be revenue-neutral; instead it would result in large revenue losses. Specifically:
1 - Feldstein assumes 30c revenue from each dollar of itemized deductions no longer available to affected taxpayers - but Romney's
maximum tax rate for the affected parties is 28% not 30%.
2 - Feldstein assumes the loss of both itemized deductions AND the standard deduction - which is argued nowhere else, not even by Romney.
3 - Feldstein neglects to pay for the repeal of the estate tax - itself a $21 billion loss of revenue.
For these and other technical reasons the TPC concludes:
Taking the estate tax and other effects into account, Feldstein’s proposals come up at least $90 billion short of revenue-neutral….
and further that:
Finally, the debate over what is or isn’t possible distracts from the more important question of what the Romney plan actually is. The governor could settle this issue quickly simply by describing how he’d pay for his tax cuts.
Which last clarifying revelation is surely going to happen on the same day the Romney releases his own tax returns!
Brad DeLong - Round Two: Feldstein and Math Fail
After noting that Feldstein allows there are many known unknowns in Romney's budget "plan" - we get the shortened version of the math Feldstein imposes to make the Mitt plan no longer a miss, 30%.
It is impossible to calculate the exact effects… since Gov. Romney hasn't specified what he would do… taxpayers with adjusted gross incomes over $100,000… made itemized deductions totaling $636 billion… paid marginal tax rates of 25% to 35%… apply a 30% marginal tax rate to the $636 billion…
DeLong then lets loose with both barrels.
First the pure arithmetic as assumed:
That is Feldstein's point:
•Cost = $186B
•Maximum Possible Revenue = 30% x $636B = $191B
•$191B > $186B
But that point is simply, mathematically, arithmetically wrong.
DeLong then points up how wrong Feldstein is in his math - that with Romney's proposed spread of top end rates being 20 - 28% (down 1/5th from 25 - 35%) the nominal median value to impose would be 24%. So, rather than $191 billion the Romney plan could only hope to generate $152 billion. This seems to me to be such a glaring error that it deserves the title used by DeLong -
Martin Feldstein Accidently Proves Either (i) 152 > 186 or (ii) It Is Mathematically Impossible for Romney to Keep His Tax-Policy Promises
Or, as DeLong concludes: $152B < $186B
QED.
But DeLong isn't finished fisking Feldstein. First he notes that even in order to get to the $152 billion number he has to grant Feldstein four party favors. Allowing him to claim big behavioral responses when they help his argument, yet to limit them when they do not; one to excuse the failure by Feldstein to winkle out of the Romney campaign details that could flesh out their budget plan, and then this last one on a definition of what constitutes a high-income taxpayer.
But refuting the Tax Policy Center's assertions… only requires knowing if enough revenue could be raised from high-income taxpayers to cover the $186 billion cost. The IRS data show that taxpayers with adjusted gross incomes over $100,000…
Hitherto in the tax debate "high-income taxpayers" has meant has meant a reported AGI of more than $200K/year on their return--not $100K/year. Hitherto it never struck anybody in this debate that "high-income taxpayers" could be defined as Feldstein defines it. And the Tax Policy Center explicitly defines high-income taxpayers as those reporting AGI of more than $200K/year.
The fact that Feldstein cannot use the TPC's definitions is proof that the TPC is correct.
DeLong - after taking apart each of those four "passes" granted Feldstein then concludes:
I think that Feldstein was working quickly trying to produce a piece that would be helpful to a candidate, Mitt Romney, who he thinks would make a much better president for America over the next four years than Barack Obama.
I think Feldstein took the standard economist's shortcut of just doing the first-order effects, and then in the press of deadlines failed to check whether the second-order interaction effects were large enough to make a difference.
So - a clear exposition of Feldstein's flawed assumptions, a clear declaration of mathematical failure - and a demotion from vaunted levels of academia to mere "standard" economist.
It almost makes me feel sorry for the guy.