Monday, July 21, 2008

Would raising the retirement age be regressive?

It's sometimes commented that increasing the Social Security retirement age would be regressive. This makes sense at first glance, since lower earners tend to have shorter life spans that higher earners. However, an increase in the normal retirement age is equivalent to an across the board benefit cut – for each year the retirement age rises, benefits are reduced by around 6.7 percent. So there's no particular reason to think raising the retirement age would be any more regressive than any other benefit cut.

However, to check it out further I did a quick run using the Gemini model, where I compared replacement rates (annual benefit divided by an annualized income derived from the present value of lifetime earnings) by lifetime earnings percentile. I calculated replacement rates for current law and for a policy that would eliminate the hiatus in the retirement age increase from 66 to 67, and further raise the retirement age to 70 by the year 2060. (Although I examined the policy for the 1960 birth cohort, meaning their retirement age would not have risen all the way to 70 by the time they retired.)

The figure below compares replacement rates by earnings percentile for current law and an increased retirement age.

While this isn't decisive, visually there doesn't appear to be any strong deviation by earnings level. (Bear in mind that while the sample population was over 47,000, when examined by individual earnings percentile sample sizes were smaller.) There are other ways to calculate things, but at this point this seems to confirm my prior belief that progressivity wouldn't change very much.


Jim Glass said...

Since I'm one who's always saying here that raising the retirement age is a regressive benefit cut, I'll clarify my meaning -- and the fact that I'm talking about a potent, existing political argument, not particularly my own opinion.

It's not that it would reduce the size of benefits of low earners relative to those of high earners. It's that any equal % benefit cut for both low-wealth individuals and high-wealth individuals is felt much more by the low-wealth individuals. Ms. Casey-Kirschling off sailing on the First Boomer and low-income individuals depending entirely on Social Security for all their income take the same % cut in benefits, but the cut is felt much more painfully by the latter -- and they are the persons SS is supposed to serve as "social insurance".

Delaying the retirement age is simply a benefit cut of another sort. While most people may be able to afford it and/or work longer, it's the people who can't afford it and can't work longer for whom it is a a real hardship -- and those are the people for whom "social insurance" was created.

I.e., it is the degree of hardship resulting from the benefit cut that lands regressively.

(Look at this way: the SS tax is often decribed as "regressive" because it applies at the same flat rate to rich and poor -- the very same thing is true of a uniform benefit cut.)

In my experience, as people realize this they tend to become less sympathetic to deferring the retirement age and more sympathetic to means testing of some sort. I'm not being ideological about this, it's just one person's observation, FWIW.

ISTM that one of the basic factors that determines one's attitude towards this issue is what one conceives Social Security actually to be.

If one sees SS as a pension system, then a unform benefit cut of some sort to bring the system into fiscal balance makes perfect sense. Keep everyone on the same terms proportionately as before. That's only fair.

But if SS is a "social insurance" system, then it's avowed first and foremost purpose is to help the poor and unfortunate. (Including the well off, from Warren Buffett down through the upper middle class to the likes of Ms. Casey-Kirschling, is merely political strategy to keep the better off invested in the system, to prevent them from turning into "welfare". Why the same argument doesn't apply to give the rich the Earned Income Credit I don't know ... but I digress).

Note how over on left-side blogs and sites people now are overtly comparing Social Security to "home insurance" -- DeLong and Thoma compare it to fire insurance .

As the return on contributions diminishes to the negative -- making SS objectively ever worse as a "basic pension" and creating a stronger argument for private accounts providing higher returns -- these people answer that, hey, Social Security as "social insurance" isn't supposed to provide a positive return any more than any other kind of insurance. It's purpose is to protect against poverty and misfortune. (Paul Sameulson's description of the "beauty of social insurance" having received the memory wipe treatment ... but I digress again.)

To the extent that one buys this "social insurance" argument that SS's real purpose is "fire insurance against poverty", it's pretty hard to think that a change to the system landing equally on the poor as on Warren and the girls at the yacht club, but felt much more painfully by the former, isn't regressive. And a lot of people buy this argument, and are well-organized around it.

Me, FWIW, I see SS as a defined benefit pension plan that has become a very poor deal for today's workers through the dropping of the sunk cost of SS's one-time $14 trillion "backward transfer" on them through regressive payroll taxes that reach only a minority portion of national income. As Milton Friedman noted, we didn't pay off World War II with a payroll tax on lower-wage workers. The backward transfer has to be paid some way -- but workers' basic SS pension could be made hugely more attractive by paying it off some other way. But that's a whole 'nother story.

Jim Glass said...

BTW, this Social Security as Fire Insurance argument shows that there's a spot there suffiently sore to the left to drive normally smart people like DeLong and Thoma nuts when touched -- they can't even read right when ranting about it.

Thoma writes:

Social Security is mainly a means of insuring against economic risk. It is fundamentally an insurance program, not a saving program, and as such it is not welfare. Just because an economic activity transfers income from one person or group to another does not make it welfare. Fire insurance transfers income. Some people pay premiums for their whole lives and collect nothing. Others, the unlucky few who suffer a fire, collect far more than they contribute. Does that make it welfare? Of course not. Social Security is no different

Social Security will be "no different" than fire insurance when fire insurance becomes a tax-funded goverment program that starts paying an annuity at retirement age based on one's earning record -- regardless of whether one has had a fire. (Rather like a defined benefit pension.)

Or, alternatively, when Social Security pays no benefit unless one falls as an unfortunate victim of financial risk, while "some people pay premiums for their whole lives and collect nothing".

And Social Security will be "mainly a means of insuring against economic risk" when for the first time ever in its 70-year history any loss from a "risk event" is even considered in its benefit formula. You know, like with fire insurance. (What loss from what risk did Warren Buffett suffer to be recovering from it with the maximum benefit? One larger than that received by millions of persons poorer than him?)

My gosh, what drives the left to say such crazy things? Something senstive was touched!

It was Robert Samuelson applying the word "welfare" to Social Security...

Welfare is a governmental transfer from one group to another for the benefit of those receiving. The transfer involves cash or services (health care, education). We have welfare for the poor, the old, the disabled, farmers and corporations. Social Security is mainly welfare

It drove Thoma into such a froth he couldn't read the words in front of him...

Is gambling welfare? Gambling transfers income from one person to another. Does that make it welfare? Loaning money transfers income when the loan is paid back with interest. Are people who receive interest income on welfare?...

Is gambling, or loaning money, or fire insurance, "a governmental transfer"? ;-)

Another thing is, I'm old enough to remember when liberals were proud of creating a "welfare state". Loud about it even. Now ... ouch!

... by this definition, farm subsidies are welfare. Steel tariffs are welfare. R&D tax credits are probably even welfare

Well, yeah. And from now on Thoma and DeLong (and according to Thoma, apparently Angry Bear) are barred and forbidden from complaining ever again about "corporate welfare" in the form of subsidies, tariffs and tax breaks.

Such extreme political sensitivity today could say something about the likely future politics of SS as well.

shoffy22 said...

Thanks for posting these very interesting results on the effect on progressivity from raising the NRA! It's awesome to see this detailed analysis up on the web!

One comment - it seems like the scale of the graph might have an influence so that it looks like the option has a very modest effect on progressivity, since the Y axis goes up to 400% to account for the very high replacement rates for those in the lowest 10% of lifetime earnings.

I wonder if the Y axis was capped at 100%, which looks to be about the replacement level for the 10th lifetime earnings percentile and above, if then on this new scale the differentials between current law and NRA70 across the earning spectrum would look more pronounced and the option's effect on progressivity would seem more substantial?

Andrew G. Biggs said...

Jim -

Re your first comment, you're right that a uniform percentage reduction in Social Security benefits (such as raising the retirement age) would have a disproportionate effect on the poor's overall retirement income because they're disproportionately dependent on Social Security. I'll run some new numbers to show that. This was the motivating theory behind progressive price indexing; a paper by Schieber, Shoven and Pozen argued that everyone should bear a similar reduction in their overall retirement income, which implies progressive reductions in Social Security.

Re your second comment, Thoma and DeLong are wrong (in my opinion) to characterize Social Security as insurance, at least as far as retirement benefits are concerned. Two reasons: first, a lot of the retirement benefit is purely earnings-based; and second, even the part that is progressive only haphazardly so: low earners do better on average than high earners, but there's a ton of variation in treatment even among people with similar earnings levels. If that's insurance, it's not particularly well designed.

Shoffy - capping replacement rates at 100% may tend to flat line the chart from the 10th earnings percentile on down, which would make it seem as if low earners didn't get hit at all. However, capping at some level makes sense when dealing with average (as opposed to median) replacement rates, since on the low end there will be some folks with very high rep rates. Will check into it. Thanks!

Bruce Webb said...

Jim I'll just point out that there is not in fact any $14 trillion "backward transfer". You seem to be confusing the Present Value of 'Unfunded Liabilities' over the Infinite Future Horizon with some actual transfer of dollars in the past. You could sum the column yourself but just by eyeball total expenditures on Social Security since program inception would not total $14 trillion Table VI.A4.—Historical Operations of the Combined OASI and DI Trust Funds,Calendar Years 1957-2007 [Amounts in billions] And of course the vast majority of that has been contributers recaptuing their own premiums via insurance payouts.

Perhaps you could explain this further, but you seem to be time shifting a future benefit claim on incomes that for the most part have not yet been earned to some sort of legacy cost. It would be understandable, albeit odd, to cast the current Trust Fund balance of about $2.2 trillion as a "sunk cost", we are in fact shifting a liability forward, with interest, even as the utility of that excess past contribution is depreciating. But I can see no justification for deploying that $14 trillion figure.

As for the rest of your argument you have seemingly discounted the disability and survivors portion of Social Security to zero. These components are clearly insurance by any definition.

OASI is an insurance policy which converts to a inflation adjusted annuity at retirement age. DI is an insurance policy which terminates at retirement age (at which point beneficiaries are transferred to OASI). If you use a significantly narrow definition of 'insurance' you can in fact exclude 'annuity' but I think most people would find this kind of odd. And absent that your argument rather falls to pieces.

And means testing is simply a Trojan Horse being introduced into the discussion by people who are ideologically opposed to Social Security to start with, the word 'concern troll' springing to mind. When I see conservatives who are perfectly happy cutting heating oil subsidies for seniors today and refusing to fund an extention of SCHIPS without provisions that would keep kids uninsured for a year before they could switch over suddenly expressing their belief that the first priority of Social Security should be to protect the interest of the poor over Warren Buffett and Bill Gates I get a little suspicious.

'Welfare' as commonly thought of is not just a transfer from one to another, it is a one sided transfer for which the transferor gets no benefit. In that sense Social Security is clearly not welfare and any attempt to convert it rhetorically or financially towards that will be resisted as being the first step towards eliminating it altogether.

Sometimes it seems that people who lean towards liberatarianism think that leftists did not in fact have their own adolescent flirtation with Objectivism. Well most of us did read The Fountainhead and Atlas Shrugged and some went on to read We the Living and Anthem and some of us got starry eyed for awhile. Only to wake up and realize these people if they could would take Social Security away from my widowed grandmother on grounds that while she might starve at least she wouldn't be on The Road to Serfdom. (If I wanted to cruelly satirize libertarianism I can think of no better way than reducing TRTS to an 18 panel cartoon and so transform Hayek from an economist/philosopher to an outright loon. To find out that the Right went about this project on its own and actually feature it on the website of the Austrian mothership almost defies belief. Yet there it is. (I particularly like panel 17 which equates calisthenics to totalitarianism and golf to freedom)).

Libertarians often will generally try to layer their argument up but when you scratch down through you mostly end up with some sort of apologia of Atlas Shrugged with the expositor casting himself as John Galt. And some of us can spot that right from the outset.

Andrew G. Biggs said...


I think you're not right regarding the backward funding (meaning net transfers to early cohorts), and the table you cite tracking trust fund balances wouldn't really give a yes/no answer in any case. If you look at this table ( you'll see that past/present participants have received $17.4 trillion more in benefits than they paid in taxes. Future participants, even under scheduled benefits, will pay $1.5 trillion more in taxes than benefits. Add in the current trust fund balance and net things out and you have the infinite horizon shortfall of $13.6 trillion.

The upshot is that the infinite horizon shortfall that people deride (based on the 'infinite' part) isn't a function of over-payments to future generations, but to past ones. We already know that future cohorts must get out a lot less than they pay in, because past ones got out a lot more.

The insurance question is complicated. Clearly DI and SI are pretty pure insurance, but retirement benefits are a lot less clear-cut. A good part of the retirement benefit is totally earnings based, meaning it's forced saving rather than insurance. Also, the progressive part is very poorly targeted, meaning that many low earners get low replacement rates and many high earners get high ones. As insurance, the current benefit formula is often like auto insurance that doesn't pay off when you have an accident but does pay off when you haven't. This doesn't mean that insurance against low lifetime earnings can't or shouldn't be done, only that the current benefit formula doesn't do a great job at it.

Jim Glass said...

Jim I'll just point out that there is not in fact any $14 trillion "backward transfer".

Bruce, you are correct only in that the 2008 Trustees Report updates the number to a larger figure, as our host points out above.

When you get this most simple fact straight, after it has been pointed out to you repeatedly, we can perhaps move on to your opinions of libertarians and readers of Ayn Rand -- though what they would have to do with a response to me in your imagination I don't know, as I've never read the woman, and don't give much thought to libertarians, so you seemingly are far ahead of me in both regards.

Thus, perhaps instead it would be more productive then to move on to a discussion of how a good understanding of the "backward transfer" is the first thing someone interested in the subject of SS reform should try to obtain -- as the backward transfer is the very root and first cause of all the several problems of Social Security that drive a need for reform.

I've explained how this is so for these various problems, with a helpful illustration from Milton Friedman that makes it easier for many people to see how it is so, in a bit of posting that is probably too long to fit comfortably in a comment here, so I've posted it over at my house.

Paul Lawin said...

Andrew -

It makes sense to me that the windfall gains from the early generations, adjusted for interest, have to equal the losses incurred by the last generation. However, I'm struggling with the numbers.

You referenced Table IV.B.7. The $17.4 trillion there is the "present value of future costs less future taxes for current participants". However, you describe it as "past/present participants have received $17.4 trillion more in benefits than they paid in taxes".

The description in the Trustees report seems to make the $17.4 trillion subject to future decisions. That is, Congress could make the $17.4 trillion go to zero by either raising taxes or decreasing benefits. But clearly, the past excess of benefits over taxes can't be reset to zero by an act of Congress.

Paul Lawin said...

Andrew -

No need to reply, I somehow missed the "current" in the description.
So the future taxes only include the taxes that would be paid by the current participants. The $17.4 billion is the amount they would lose if we shut the program down today, which should equal the past "profits" from earlier generations.

Andrew G. Biggs said...

The distinction between past/present participants (which ranges from someone who died in 1940 to someone who entered the workforce today) vs future participants isn't the greatest breakout I could think of, but it's what the actuaries give us. There are papers that look at net transfers cohort by cohort (an economist at SSA named Dean Leimer has done the best work). I may try to break things out in clearer ways, since most of that $17 is for folks who are already retired or dead.

That said, the termination cost of the system (benefits that have been earned but not yet paid) is also around $17 trillion, meaning that if we shut the system down now we'd still owe a ton of money. The fact that that two numbers are very similar is, I think, coincidental.