DeLong finds some fuzzy math in Martin Feldstein’s WSJ op-ed (and I found some more)

by pdxblake

Martin Feldstein provoked Brad DeLong’s ire with a Wall Street Journal op-ed that tries to show how Mitt Romney’s tax cut plan could lead to no increase in the deficit or the need to raise taxes on middle- and lower-income Americans, and takes issue with five points, including one where Feldstein makes a mathematical error by counting savings from removing tax deductions as if the current tax rates were in place, rather than Romney’s proposed rates.  DeLong rushes through the first four objections (which he gives Feldstein a pass on since the fifth is the most easiest and makes Feldstein’s point moot, since $152 billion > $168 billion is a false statement).  Here are a few of the four other items

First, Feldstein argues that Romney’s tax cuts cost $168 billion, once you add in the ‘dynamic scoring’, which is wonk-speak for magical supply side effects where tax cuts benefit the economy so as to raise GDP enough that the taxes on the additional GDP offsets some of the cost of the tax cuts.   The CBPP has a good summary of the arguments against using dynamic scoring in budget estimates which can best be summarized as the effects are highly uncertain and are likely to be small.  The love of conservatives for dynamic scoring is that, when fed in high estimates of the feedback from tax cuts to growth that are not well supported in empirical study (they are high than reality), they show that tax cuts have a more stimulative impact on economic growth than they actually do, and lead to the conclusion that tax cuts are much less costly than they actually are.

Second, Romney has not provided any specific deductions that he would propose eliminating and the Tax Policy Center report (pdf), which he criticized unfairly as being ‘biased’, makes an overwhelmingly friendly assumption that deductions on high-income could politically be eliminated (a very rosy assumption in favor of Romney’s plan).  By assuming this rosy political outcome, it effectively lowers the impact on the deficit by assuming more deductions will be removed for high-income people than can be credibly assumed.

Thirdly, Feldstein’s plan ‘moves the goalposts’ as DeLong points out by including in the amount of deductions those claimed by everyone with more than $100,000 (a much larger group than Obama’s focus for not extending the Bush tax cuts for only those people with incomes of $250,000).  This leads his estimate of total deductions at $636 billion to be much larger than any plausible scope for deductions to be eliminated since it includes the mortgage interest deduction for people earning between $100,000 and $250,000.

While one can argue the merits of the deduction of mortgage interest for this segment of earners (and for all earners), there is no way in hell that any politician would vote for a bill to do this any time in the near future (remember, it would have to pass the House whose members have to go try and get re-elected every 2 years).  The conservative Tax Foundation has data showing that of all tax returns filed, 6.8% of those are filed by people with incomes between $100,000 and $200,000, and include an average of almost $9,000 in deductions from the mortgage interest deduction.  With 150 million returns filed annually, the back-of-the-envelope amount of deductions that are being added into Feldstein’s number by moving the goal post just from the mortgage interest deduction is approximately $91 billion, or about one-seventh of the value of the deductions he thinks (with questionably support) could be eliminated.

The final straw for Brad DeLong was when Feldstein estimated the budgetary impact from eliminating $636 billion in deductions, where he used an average rate of 30%, which DeLong explains:

Taxpayers making more than $100K/year in AGI had marginal tax rates of 25%-35% in 2009–an average tax rate, Feldstein assumes, of 30%.

After Romney’s 1/5 reduction in tax rates they will have tax rates of 20%-28%–an average tax rate of 24%.

Multiplying not the wrong 30% but the true correct 24% marginal tax rate by the $636 billion in itemized deductions gets us not $191B but $152B.

$152B < $186B

DeLong makes a great point, but even giving a pass on the average tax rate being 30% (rather than what it would be, closer to 24%), and instead focusing on just moving the goalposts from $200k down to $100k (for the income levels where the deductions start to be eliminated) and just focusing on the impact of the mortgage interest deduction, we have Feldstein’s questionable ‘savings’ from eliminating deductions of $191 billion reduced by $27 billion ($91 billion in deductions * 30%), which puts Feldstein’s math wrong again because he is now saying:

$164B < $186B

And as DeLong pointed out, there are so many dodgy assumptions that get Feldstein to his original numbers that it is more and more ridiculous for anyone to defend the idea that Romney’s budget can either a) not increase the deficit, or b) not lead to increased taxes on people earning under $250k / $200k / $100k (pick your favorite).

UPDATE: Feldstein responds and starts to say, “While I still believe the assumptions that I used in my analysis…” before Brad DeLong interrupts and says “No! No! Ten thousand times no!  If the maximum marginal tax rate is 28%, you cannot cut $1 of itemized deductions and increase revenue by 30 cents. It is not possible. The assumption that you can is unbelievable. Nobody should believe it.”