Last night, Republican presidential nominee Mitt Romney made news by substantially “etch-a-sketching” the tax policy he had been running on since the GOP primaries began. In making up policy on the fly, he promised that his tax cuts would be entirely revenue-neutral, that he would cut taxes on the middle class, and that he would not cut taxes on high-income earners. Taken together with his specific tax cutting plans, these pledges violate basic rules of arithmetic.

Early on in the debate, Romney disputed President Obama’s claim that the former governor’s central economic plan was a $5 trillion tax cut on top of extending the Bush-era tax cuts:

“First of all, I don’t have a $5 trillion tax cut. I don’t have a tax cut of a scale that you’re talking about. My view is that we ought to provide tax relief to people in the middle class. But I’m not going to reduce the share of taxes paid by high-income people.”

I’ve gone over these numbers before, but it’s worth a quick refresher about the broad thrust of what’s wrong with this claim: Romney is hugely specific about just how he’ll cut taxes (mostly for high-income earners) but refuses to specify any real-world offset though the “base-broadening” that he’s promising.

Romney initially proposed repealing the estate tax; eliminating capital gains, dividends, and interest taxation for households with adjusted gross income under $100,000 ($200,000 for married taxpayers filing jointly); cutting the corporate income tax rate from 35 percent to 25 percent; eliminating the corporate alternative minimum tax (AMT); and repealing new taxes from the Affordable Care Act. This $2.7 trillion package of tax cuts would be entirely deficit-financed—indeed repealing business tax loopholes would only be used to offset further corporate income tax rate reductions below 25 percent. Romney subsequently proposed reducing all individual income marginal tax rates by 20 percent and fully eliminating the AMT, claiming that this round of tax cuts would be both revenue-neutral (i.e., financed with “base-broadening”) and distributionally neutral. If this second round were fully deficit-financed, the cost of Romney’s tax plan would increase by $3.4 trillion, rising to $6.1 trillion. (The $5 trillion figure being cited on the campaign roughly multiplies the Tax Policy Center’s (TPC) revenue estimate for 2015 by 10, as budgets are usually analyzed over a 10-year window, whereas I have indexed tax cuts to nominal GDP.)

The only specifics on “base-broadening” the campaign has released are possibly capping itemized deductions at $17,000 (which, by the way, could easily violate the pledge not to raise taxes on the middle class—living with a big mortgage in a high tax state would likely do the trick) and that the tax preferences for capital gains and dividends will absolutely not be repealed.

Then as he reinvented his tax policy on the fly last night, Romney dug himself even deeper into a mathematical black hole:

“My number-one principal is, there will be no tax cut that adds to the deficit. I want to underline that: no tax cut that adds to the deficit. But I do want to reduce the burden being paid by middle-income Americans.” (Emphasis added.)

What this means is that the amount of “base-broadening” Romney has promised just soared from $3.4 trillion to $6.1 trillion—a major transformation. Note The Washington Post‘s Fact Checker Glenn Kessler got this totally wrong: He dismissed the $5 trillion figure as income tax cuts, estate tax repeal, and AMT repeal that Romney would supposedly make “revenue neutral,” inaccurately conflating the second round of tax cuts and related (unidentified) offsets with the (over $5 trillion) bill for both rounds of tax cuts. Previously, only the individual income tax cuts and AMT repeal were meant as revenue-neutral, leaving at least $2.7 trillion in deficit-financed tax cuts that only now must be paid for.

Can this be done through “base-broadening?” The Congressional Research Service estimates that the feasible range for eliminating individual income tax preferences—totally ignoring distributional concerns that would violate Romney’s promises—range somewhere between $100 billion and $150 billion, or between $1.3 trillion and $1.9 trillion (again indexing to nominal GDP growth). And the Joint Committee on Taxation recently estimated that revenue-neutral corporate tax reform could only lower the top statutory rate to 28 percent assuming all major tax expenditures were eliminated—cutting the corporate rate to 25 percent and eliminating the corporate AMT without adding to the deficit would require eliminating individual income tax expenditures to subsidize corporate tax cuts. And on top of that, the middle class is promised a tax cut. The math simply doesn’t add up.

If you cut taxes for the middle class and your overall tax package is revenue-neutral, both relative to current policy, that necessarily means you’re raising taxes on the poor and/or upper-income households (above and below one’s definition of “middle class”). TPC’s analysis does indeed assume Mitt Romney would on average raise taxes on households earning under $30,000 by ending recent expansions of the Earned Income Tax Credit, Child Tax Credit, and American Opportunity Tax Credit. But even cutting refundable credits and jacking up income taxes won’t soak enough revenue from lower-income households, because that’s not where the income is and we’re looking for $6.1 trillion in offsetting tax increases.

That leaves upper-income households. Romney’s concretely identified tax cuts, ignoring unspecified offsets, would reduce effective tax rates for the highest-income 1 percent of households by 7.8 percentage points relative to current policy, raising their after-tax income by 11.2 percent, according to TPC. A recent TPC report found that itemized deductions lower after-tax income for the top 1 percent of households by 3.0 percent—fully eliminating, not just capping itemized deductions, would still leave upper-income households with a tax cut. Similarly, tax exclusions lower after-tax income for these households by 6.4 percent—their elimination wouldn’t be near enough to finance the specified tax-cuts either. As another recent TPC report made unequivocally clear, you can’t offset these upper-income tax cuts without raising taxes on the middle class or adding to the deficit if you preserve the capital gains and dividends preferences: Romney’s tax cuts for households earning more than $200,000 total $251 billion in 2015, whereas the total value of tax expenditure (excluding saving and investment preferences) total only $165 billion. (The capital gains and dividends preferences are the most regressive in the tax code—75 percent of their benefit goes to the top 1 percent.) The authors concluded that “any revenue-neutral individual income tax change that incorporates the features Governor Romney has proposed would provide large tax cuts to high-income households, and increase the tax burdens on middle– and/or lower-income taxpayers.” And that was before the corporate income tax cuts were also supposed to be revenue-neutral and before the middle class was supposed to get a net tax cut. Looks like the poor are on the hook for the entirety of these tax cuts.

Lastly, Romney cited “growth” as a source for revenue under his plan, which is also in the too good to be true category—so much so that Doug Holtz-Eakin, director of the Congressional Budget Office under President Bush, rejected the use of dynamic growth effects in scoring tax proposals. (See Jason Furman’s evisceration of the dynamic scoring cop-out.) But growth won’t come to Romney’s rescue: Any near-term demand boost from cutting taxes would be dwarfed by the reduction in aggregate demand from spending cuts required to meet his proposed cap on federal spending at 20 percent (which requires somewhere in the ballpark of $6.2 trillion in cuts to nondefense spending over a decade). The one escape valve for Romney’s arithmetic not adding up on a static basis is bunk.

Mitt Romney’s tax plan didn’t add up before last night, and it’s further into the mathematically impossible today. This myriad of promises is not a credible plan; this is fantasy.