From a quick look, here are 3 reasons why they come to different conclusions:
1. Static Revenue loss of Romney plan: $180B vs $360B
- Feldstein takes the 2009 IRS numbers for non-capital gains income tax revenues of $900B and multiplies it by 20% for the Romney 20% cut, resulting in roughly $180B to make up via base broadening
- TPC uses a 2011 current law policy baseline and estimates the cuts will cost $360B in 2015.
The difference in timing (i.e. 2009 vs 2015) accounts for some of the gap and 2009 was an usually bad year for income and thus income taxes (as Feldstein himself admits). While 2009 is the latest year IRS has available, using it underestimates the revenue cost. As we will see, this is not the major factor driving the difference between Feldstein and TPC (especially since Feldstein’s deductions are in 2009 dollars).
2. Behavioral Responses:
- Feldstein assumes an elasticity of .5 for tax rates, which means some of the $180B doesn’t have to be paid for via base broadening
- “History shows that a tax cut that raises the after-tax share of earnings that an individual keeps by 10% raises taxable income by about 5%. This implies that the revenue loss from the 20% tax cut would be $148 billion, not $181 billion.”
- This is higher than the best estimates of labor supply elasticities in the literature (see Saez and Grueber), especially if you consider fiscal externalities from income shifting.
- TPC notes that the CBO 2003, 2005 & JCT 2005 estimated small revenue growth effects. Even if you use the Mankiw and Weinzierl (2006) assumption of 15% growth 5 years later, TPC shows their conclusion still holds. However, applying the Mankiw and Weinzierl assumption gets you close to Feldstein’s $150B, so the key behavioral assumption isn’t the elasticity on tax rates – it is the elasticity on deductions. Feldstein assumes that people don’t respond to having their deductions lowered, even though they will be paying more taxes (relative to just getting the 20% cut in rates without base broadening) and potentially lower marginal rates (if the deduction pushes them into a different tax bracket).
- With that said, elasticities are small and some of these deduction removals are lump sum levies (as housing decisions have already been made for instance) so this is not the major factor driving the difference between Feldstein and TPC.
3. Excluded Deductions:
- To make tax changes revenue neutral, base broadening of either $150B or $360B is required
- Feldstein simply argues that itemized deductions above $100K in 2009 amounted to $636B, so 30% is $190B which can cover the cost of the cuts (i.e. the $150B). Since $190B>$150B, he concludes that Romney’s plan is mathematically possible. Note that eliminating the estate tax costs about $20B in 2009 revenue, so this gap isn’t huge.
- TPC argues using all deductions is highly unrealistic for 3 reasons:
i) 1/3 of tax expenditures promote spending and investment
ii) 10% of them are for hard to eliminate
iii) Lowering rates by 20% makes deductions less valuable. “For example, eliminating $1 of deductible mortgage interest raises $0.35 when the top rate is 35 percent, but only $0.28 when the rate is 28 percent.”
- These three factors result in only $550B in tax expenditures instead of $1.3T or ~42%. Based on this logic, Feldstein shouldn’t use the $636B base but 42% of it (i.e. $270B). Based on the Feldstein 30% of the tax expenditures approach, you’d get only $80B, which is insufficient to cover the $150B for cutting rates. This mirrors the TPC conclusion, which shows that you’d have to use 2/3 of the the realistic tax expenditure base to pay for the Romney cuts.
In short, the main difference is based on what’s included in the base of tax expenditures. “Protecting” the savings and investment incentives, as well as the 10% in the really hard to cut category, and adjusting for the effect of lower rates on the value of deductions brings us back to the mathematically impossible category (in terms of preserving the distribution of tax burdens).