Monday, April 12, 2010

What Krugman gets wrong on Social Security taxes…

Over at AEI's enterprise blog I write:

Paul Krugman, responding to folks who pointed out that having nearly half the population not pay income taxes might not be quite right, says

"The thing to bear in mind is that overall, the US tax system isn't actually that progressive: the payroll tax is regressive, as are most state and local taxes, which largely offsets the progressivity of the income tax."

I'll limit myself here to payroll taxes. (It's not clear to me that state income and property taxes are regressive, although sales taxes are.) Social Security taxes, as Krugman may or may not be aware, fund a thing known as Social Security benefits. Analysts of this system consider them together in order to determine whether the program is or isn't progressive.

One way they do it is by calculating what's called the "net tax rate"—that is, the statutory 12.4 percent Social Security tax paid by workers minus the benefits they receive from the program. If you receive benefits equal to your taxes, then your net tax rate is zero. If you pay more in taxes than you receive in benefits, your net tax rate is positive; likewise, if you receive more benefits than taxes your net tax rate is negative.

This chart, which I calculated using the Policy Simulation Group's GEMINI model, reports net tax rates for recent retirees, broken down by quintiles of lifetime earnings. What it shows is that the highest-earning 20 percent of individuals—the fifth quintile—pay a positive net tax rate of 3 percent of lifetime earnings.

The other 80 percent pay negative net tax rates, meaning that they tend to receive more in benefits than they receive in taxes. (For anyone looking forward to this in the future, just bear in mind that the program is insolvent.) The lowest-earning fifth of the population has a negative net tax rate of almost 27 percent, which implies that they receive far, far more from Social Security than they pay into the program.

Call me picky, but what I'd like to see from tenured Ivy League professors and Nobel Prize winners is that it takes more than, oh, 30 seconds to figure out the obvious flaw in their arguments. Maybe next time.

Here's another way to put it: if we're looking at Social Security taxes and benefits in isolation, then yes, Social Security taxes are regressive. But then, Social Security benefits would have to be considered a form of welfare, rather than earned social insurance. The key to the program from FDR's time, I've argued, was that there was some link between the two. If so, we should think about them in those terms.


Bruce Webb said...

Lots of states don't have income taxes. And property taxes on investment property/rentals while officially being incidental to the owners are actual paid by renters (the argument used by conservatives to explain why corporate taxes are paid by consumers and so are not progressive at all). It is amazing what parts of taxes are considered elastic and which not when it comes to cumulative tax burden.

Andrew G. Biggs said...

I'm no expect, but in my experience there are exemptions and rebates on property taxes for low earners and seniors. And to the degree that people's property values correlate to their incomes, even a flat tax with no rebate isn't regressive -- it's just flat.
More broadly, while state taxes are hardly inconsequential, federal taxes are several times larger and the main federal ones -- income taxes, Social Security taxes, and Medicare taxes -- are all quite progressive when coupled with their benefits. I'm not saying the entire tax code is as progressive as the income tax schedule, but for Krugman to say it's not that progressive overall has got to be wrong.

JG said...

The strange social arithmetic of Paul Krugman:

Start with a progressive income tax, then add a progressive Social Security program, and you make the situation more regressive.

Progressive + Progressive = Regressive.


Anyhow, the OECD has a different opinion:

the U.S. "has the most progressive tax system and collects the largest share of taxes from the richest 10% of the population."
-- cite and data via the Tax Foundation.

PK keeps saying the US should be more like Europe. Maybe we should be even more regressive?

Joe said...

Where are the "net tax" calculations for every program? If a billionaire pays taxes used for defense and has his billions defended, and a poor person pays taxes used for defense, joins the military, goes to war and gets killed, what is the "net lifetime tax" paid by each for defense?

Andrew G. Biggs said...


While you've got a legit point in that it's hard to measure the benefit people get from each dollar of taxes they pay, I think the military analogy is flawed, not least because we've got an all volunteer force. And for what it's worth, based both on discussions with people in the military and statistics, poor people who join tend to go into positions where you learn new skills, which tend not to be front-line combat; middle and upper class people, by contrast, join for the adventure and tend to go into higher risk positions. (While race is hardly a perfect proxy, you would find for instance that minorities in the military tend to have a lower than average death or injury rates while whites have higher ones.) So I get your point, but I'm pretty confident the overall system remains progressive.

Joe said...

I am pretty sure that service in the military is a net positive for those who serve, but in a time of war I wouldn't be so sure that I really understood the true costs of sefvice.

To say we know how the benefits of "national defense" are distributed or even how defense spending solely in dollar terms is distributed in a given year is asking a lot. Maybe someone actually figured this out. Do you know beyond mere armchair analysis?

The simple fact is that the tax to benefit ratio in social security is explicit while it is implicit but no less real in every other area of taxing and spending. Then there are the non-quantifiable benefits. Income replacement for a lifetime low earner has to be more valuable than for a corporate executive with a golden parachute. Similarly many of the rules in our tax and regulatory systems are design to benefit those with the most, not to protect those with the least. The rules of the government's September 11th Victim Compensation Fund gives this whole argument a peculiar twist. Richer people's lives are worth more.

The bottom line is that both you and Krugman are making incomplete and incompatible points because your purpose isn't really enlightenment. You are just trying to claim ownership of another barren hilltop in you ideological warfare. Your reprehensible use of the word "welfare" --or should I say your appeal to the negative connotations of the word "welfare"--makes that clear.

But rather than getting into a pissing contest, care to talk about what aspects of the social security benefit systems is more or less progressive. My understanding is that disability benefits are far more progressive than retirement benefits. That women's benefits are far more progressive than men's, and that the social phenomenon of differential mortality is more important to the degree of progressivity than the details of the benefit formula.

Thomas said...

I would be interested to know if the calculation of net tax rate assumes any rate of return on social security contributions or if the calculation assumes no return. If the calculation compares only contributions and benefits in flat dollar terms (with no rate of return), my guess would be that with an assumed rate of return on contributions most social security recipients would have a positive tax rate and that the differential of the rate from top to bottom is increased due to the compounding effect. Mr. Biggs, could you clarify the methodology?

Andrew G. Biggs said...

Thomas, The calculation basically incorporates rates of return. Since taxes and benefits are discounted at the govt bond interest rate, that means that if your discounted taxes and exactly equal to your benefits you earned a return on your taxes equal to the bond rate. If your taxes had a smaller present value than your benefits, your return was above the bond rate. I could have done the same exercise using rates of return and it would show the same thing.

Thomas said...

Thank you for the clarification. I guess use of the government bond (TBill?) rate makes sense because the social security trust funds end up getting loaned back to the Treasury for general purposes.

Andrew G. Biggs said...

Actually, the use of the government bond rate is appropriate because Social Security benefits carry a risk of default (or cuts) roughly equivalent to that of government bonds -- it's a safe, but low-returning, 'asset.' We can quibble about the proper discount rate one way or another, but the bond rate is approximately correct. At this point, it's also just convention to do it that way -- SSA, CBO and others calculate net tax rates and other 'money's worth' figures by discounting at the bond return.

Thomas said...

Perhaps I was a bit too subtle---the comment was intended to be somewhat facetious.