New York Times columnist Ross Douthat caused a bit of a stir by proposing that we not just extend the payroll tax holiday, but eliminate the payroll tax altogether. Let's set aside the merits of such a proposal for the future sustainability of Social Security and Medicare, not least because a major part of Douthat's case is that eliminating the tax makes it easier to cut, or at least change, the former. What would such a repeal cost, and how could one pay for it?

Borrowing: At least in the near term, it could make sense to repeal the payroll tax and replace it with nothing at all. With interest rates on even long-term federal debt hitting negative levels, it's actually cheaper right now for the government to spend money it borrows than money it gets from taxes. So if ever there were a time to run up the deficit, it's now.

So how much debt would eliminating the payroll tax produce? According to the OMB, the Social Security payroll tax will raise $707 billion next year, and $9.15 trillion over the next 10 years. Getting rid of the Medicare payroll tax as well would bump that up to $921 billion and $12 trillion, respectively. A Social Security-only abolition would double the deficit over 10 years, and a Medicare-inclusive one would do that and throw on another $2 trillion for good measure.

Add a bunch of new taxes: Another option would be to replace one or both payroll taxes with a wholly new tax, or set of taxes. But cobbling together enough to make up the lost revenue would be tough. According to the CBO, payroll taxes will bring in about $1.4 trillion in 2020. To make that up while keeping revenue where it'd be under current policies, you'd need to let all the Bush tax cuts expire (including those on income under $250,000), and then add another $950 billion.

Using the options the CBO has presented, it's hard to make up that difference. Adding both a 5 percent value-added tax and a carbon tax wouldn't be enough; indeed, that wouldn't even get you halfway there. You'd also have to limit tax breaks across the board, while eliminating those for employer-provided health care, state and local taxes and mortgage interest. And change the tax status of certain investments, like state and municipal bonds and life insurance, and sharply reduce the charitable deduction. And raise the gas tax.

Even if you did all that, it wouldn't replace the payroll tax as a dedicated funding stream. It'd be hard to carve out the money from eliminating certain breaks and dedicate it to Social Security and Medicare. You could dedicate the VAT and carbon tax, and the gas tax increase, to the programs and then mandate that the rest be paid for out of general revenue. But that leaves the programs vulnerable to the vicissitudes of income tax revenue.

Impose a big VAT: To fully replace the payroll tax as a dedicated source of funding, you'd probably need a much bigger VAT. For example, the Tax Policy Center estimates that a 12.3 percent VAT would raise about $1.2 trillion in 2015, more than payroll taxes are set to take in that year. You could probably get the rate a little lower if you paired it with a carbon tax, though not by a lot. A carbon tax would raise about $105 billion in 2015, which couldn't replace much of the VAT revenue.

Keep the tax and provide a credit: Given how much revenue you'd have to raise to fully repeal the payroll tax, it might be more feasible to keep the tax but provide a credit so people don't pay it on the first portion of their income. The Earned Income Tax Credit is meant to play this role for some low earners, and is a major reason why workers making less than $20,000 get more back from the income tax than they pay in the payroll tax.

But the EITC is indexed to inflation rather than wages, so payroll tax burdens are growing faster than it is. So some, notably conservative tax wonk Robert Stein, have proposed expanding the child tax credit, letting it count for the payroll tax and pegging it to wage growth, so as to offset more of the tax than the EITC and child credit currently do, and to prevent them from getting less effective as years go by. Stein would pay for this by converting the standard deduction and personal exemption to a credit, which would primarily hit upper-middle and upper-class taxpayers.

Alternately, economist Kenneth Scheve and Matthew Slaughter suggest exempting workers earning below the national median from the payroll tax, and making up the difference by raising the tax's rate, introducing higher brackets, or increasing the tax's wage cap.

Such an exemption would be tricky to implement, especially for workers juggling a number of part-time jobs. For example, suppose the tax rate is 10 percent and does not apply to the first $20,000 one earns. A worker who gets $10,000 each from three different jobs would owe $1,000 in tax ($30,000-20,000 * 10 percent = $1,000) but wouldn't have paid that at any of his jobs, and so would be hit with a hefty tax burden at the end of the year. That's why a credit approach like Stein's may be more viable.