Arnold Kling  

Milton Friedman on Social Security

PRINT
Moore's Law... The Airline Union Game...

My choice for last week's comment of the week generated a lot of comments! It refers to this article by Milton Friedman, who describes a conceptual scheme for transitioning out of Social Security. Based on the comments, I can tell that people are challenged by Friedman's idea. Let me try to explain it.

Suppose that I reach retirement age today. How much is in my Social Security account?

Nothing. All I have is a promise from the government to pay benefits.

Now, suppose that instead of a promise to pay benefits, the government gives me cash equal to the present value of the promised benefits. The present value can be calculated using the formulas that life insurance companies use to convert cash to annuities. In fact, I could take my cash, walk into a life insurance company, and purchase an annuity. Thus, I would go from a government program to a private program with the same characteristics.

If the government gives me cash instead of promises, then it has successfully transitioned me out of Social Security. To find the cash, however, it has to borrow money. This debt can be repaid immediately by raising taxes now, or it can be repaid with taxes collected over a long period of time. Those taxes could be payroll taxes, or they could be other revenues such as income taxes.

(For younger workers, a cash-out transition also can work. However, the younger worker's expected Social Security benefits are out in the future, so the present value has to be adjusted. Furthermore, as we abolish Social Security, the younger worker's one-time cash payment should be reduced by the present value of the payroll taxes that the worker will no longer be paying.)

The day that the government cashes everyone out of the Social Security system, the debt on its balance sheet will balloon, because the debt that funds the cash payments counts as debt, but the promises of future benefits are not accounted for. This is a matter of sloppy bookkeeping in the present system.

If you were to count future Social Security liabilities on the balance sheet, then a cash-out program would leave total government liabilities unchanged. To a first approximation, the only difference between Milton Friedman's cash-out plan and the current Social Security system is that the bookkeeping of Friedman's program is transparent, while the bookkeeping for Social Security is opaque.

Note that Friedman's cash-out concept should not be confused with any "privatization plans" that have been placed on the table in the past few years. Privatization plans often are misleadingly presented as if they were solutions to the long-term Social Security budget problem, which they continue to hide. Instead, Friedman's approach would put the long-term budget problem out in plain view.

Friedman's approach is simply a shift in timing. Instead of an annuity, you receive a one-time cash payment. It relates to privatization in the sense that if you cash everyone out of the system today, then the government can choose to get out of the tax-and-transfer business with future workers. However, as a matter of theory, the government could stay in the tax-and-transfer business by making new cash payments to people as they enter the labor force. As a matter of practice, the foolishness of this would be transparent.

For Discussion. Democratic economists, including Brad DeLong, "bang their heads against the wall" trying to make it clear that the country needs fiscal discipline because of the long-term outlook for Social Security. Should they support Friedman's idea as a way of making the liabilities in the entitlement programs more transparent?


Comments and Sharing


CATEGORIES: Social Security



COMMENTS (23 to date)
Jim Glass writes:

Sure they should, everybody should want transparent accounting on a bipartisan basis.

The federal government's cash basis accounting is illegal for any private business with >$1 million in revenue that has inventories. Private business owners who used the government's accounting methods would be in jail, literally, not as a rhetorical statement.

Alan Greenspan in his recent testimony encouraged the government to adopt accural accounting, like the entire private sector uses, saying it is quite doable. IIRC, the world's central bankers at one of their conventions out at Aspen a few years back said adopting accural accounting was the best reform of fiscal governance that nations could adopt -- and at no more cost than keeping an honest set of books.

If the government did use it, nobody would have believed for a moment that it ever was running a surplus. Everyone would know that Social Security, instead of having $1 trillion in the "trust fund", has a projected funding deficit discounted to current value of $11 trillion. Everyone would know that Medicare has a similar shortfall of $13 trillion -- up $3 trillion in only the last two years! And how would Congress justify adding a prescription drug benefit then? ... Well, that's why it uses cash accounting.

(And speaking of drug benefits, heads go banging against the wall so often over at Prof. DeLong's site that I suspect he has an ulterior motive of having the gov't pick up the cost of his headache medicine. Either that or he owns stock in a wallboard maker.)

Jane Galt writes:

Are you reducing the accrual or the present value of probable benefits? If the former, why take out the future payroll taxes? You'll get back any expected taxes in the future on the pension income.

Jim Glass writes:

The numbers for Social Security and Medicare represent the amounts by which the expected cost of future benefits exceeds expected revenue available to pay for them, over the next 75 years, discounted current value.

See the Analytical Perspectives of the Budget, http://www.whitehouse.gov/omb/budget/fy2004/
Table 3-3 at page 48.

Note that the deficits are much larger when projections go beyond 75 years:

"Limiting the calculations to 75 years understates the deficiencies, because the actuarial calculations omit the large deficits that continue to accrue beyond the 75th year. The understatement is significant, even though values beyond the 75th year are discounted by a large amount.

"For Social Security, the present value of benefits less taxes in the 76th year alone is nearly $0.1 trillion, so the omission of these distant benefits amounts to several trillion dollars of present value....

"Extending the calculation beyond the 75th year would add many trillions of dollars in present value to Medicare’s actuarial deficiency, just as it would for Social Security..."

I.e.; If taxes were raised tomorrow sufficiently to close these 75-year deficits entirely, next year, when the 76th year out became the 75th, these programs would be back in actuarial deficit again.

The fact that these programs are *actuarially unsound* in a *big way* is what people like Prof. DeLong should be banging their heads about, rather than the Bush defict increase, which is tiny in comparison. I mean SS & Medicare's accrued deficiencies went up by $4.4 trillion in current value over just two years(!), many times the amount of Bush deficit increases -- so if one is really worried about future "national bankruptcy" which of these should be inspiring the head banging?

Now, if the government kept its books on an accrual basis -- as it requires everybody else to do -- the voters would have known all this all along. There never would have been any reported "surplus" for all the politicians to take credit for, and to justify their big surge in discretionary spending that we just enjoyed. The issue of enacting a further major increase in entitlement spending with a new drug benefit would have to be framed a whole different way ...

GT writes:

I've posted enough on the other thread (Jim there's a new post for you!) but I'd like to make a couple of points.

1) I would give zero importance to any projections about year 76. The idea that we can project with ANY degree of certainty demographic and economic changes 75 years out is ridiculous. Do you think in 1900 we could have predicted the demographics on 1975? No. In fact any projections more than 40-50 years are probably useless.

2) Friedman's idea does not simply make clarify our government accounting problems. If that's what you wanted to do you could simply pass a law changing how the government runs its books. Friedman is proposing terminating SS which is a pretty different thing. So no, I don't think any economist worried about fiscal transparency should support Friedman's idea. There are better ways that don't result in ending SS. Economists should only support Friedman's view if they agree with him that we need to terminate SS and not replace it with anything.

3) As for the fact that our SS liabilities do not show up as debt being sloppy accounting that's true but that's also true of most other expenditure items. We know we will have to spend trillions of dollars in defense in the future. Why don't we include that as debt? Or how about education? Does anybody really think that claims on justice and police functions are really any less than claims on SS? Any of these can be changed by law at any time by the very entity that must make the payments (unlike the pensions liabilities of a company which is why I find comaprison with private sector accounting less than illuminating) so that simply means that at the end of the day it's a question of political preferences. Or are we going to add every single line item we expect in the future to the debt calculations? If sop we should add all the expected tax revenues to the asset side.

Patrick R. Sullivan writes:

The post on the other thread GT refers to says:

" ...now we implement the Friedman plan. Overnight SS taxes fall by $600 billion which means that income taxes or some other tax must rise by that same amount. ....Can you imagine a $600 billion tax increase in JUST ONE YEAR? "

Not with the above arithmetic, I can't. I.e., 600B-600B= 0.

Also, earlier in that post on the other thread GT wrote:

" If I owe you $100 the cost can be paid by me, by paying you the $100, or by you, by forgiving the $100, or a combination of both."

Which again clearly makes the point that there is no "transition cost".

GT writes:

Patrick,

You should make an effort and read the whole thing.

GT writes:

Just to be clear and using the $600 billion example.

Taxpayers that pay the $600 billion in SS taxes this year are doing so in the expectation of getting it back and with interests. And although we know that given how SS is set up right now there won't be enough to give them back ALL the money we do know that there will be enough to give them back some $450 billion.

Of course paying $600 billion and only getting $450 billion back is not a great deal unless you compare it with Friedman's alternative. In Friedman's model taxpayers will pay $600 billion and get nothing back.

That is the transition cost.

GT writes:

Oh, and one last point.

I fully understand that on a net basis comparing across generations even the $450 billion cost disappears since the money that some taxpayers 'lose' by not getting the $450 billion back is money that other taxpayers 'win' by not having to pay for it.

But it doesn't change the fatc that it's a cost to the current taxpayers, the ones who have to pay up yet give up their $450 billion claim. I don't think telling them, "But it's not really a cost since the next generation won't have to pay" will convince them.

Jim Glass writes:

"In Friedman's model taxpayers will pay $600 billion and get nothing back."

Then the $600 billion that is collected from taxpayers is paid to whom?

GT writes:

Jim,

Other taxpayers get that $600 billion back.

But the tax payers that paid for it don't even though under the current SS they would get , at least, $450 billion back.

I've always accepted that if you take the economy or taxpayers as a whole then there is zero cost since whatb one pays another gets.

Of course using that argument the US government costs basically zero as well since everything we pay to the government the government pays back to us and others in form of wages, buying goods and transfers ( we can leave out the small amounts that are paid to non-residents).

Are you willing to say the US government costs zero?


I have always been focusing on a specific group that would bear the cost of the transition and that's what I always meant by transition costs.

Jim Glass writes:

"(Jim there's a new post for you!)"

And a major effort too. I have a reply drafted that I'll probably post over there, but first here to the nub of things....

"Your mistake is that you forget, or leave out, that under the current SS plan the tax dollars that pay for $X CREATE new claims that will be paid by future generations (and we know there will be enough to pay back at least 75% of that back in the future). In Friedman’s plan the tax dollars that pay for $X create no claims whatsoever. That’s why those transition costs ONLY show up if SS is terminated. If SS continues those that pay $X will get 75% back. But if they terminate SS they get 0% back. That difference is the transition cost..."

Nope ... your mistake is that you forget, or overlook, that under Friedman's plan when SS terminates workers get *much more* than 0% back. What the workers who would otherwise be paying payroll taxes get back is:
1) Full promised benefits on every dollar of payroll tax they've paid *to date*, rather than diminished benefits ; *plus*
2) Relief from having to pay a 12.4% of pay payroll tax perpetually *into the future*, of which they would otherwise recover only 75%.

Add these two up and they total *savings* 0f exactly $11 trillion dollars of current value -- obviously, since the termination stops them from losing that $11 trillion over the next 75 years through the combination of taxes which they now won't pay, and diminished benefits which now won't be diminished but paid in full. That seems a considerable savings to me, a good deal more than $0.

You also seem to be forgetting that the cost in *dollars* under the status quo to payroll tax payers of getting only 75% of payroll taxes back is the exact same as the cost in *dollars* under the Friedman plan to income tax payers of picking up the cost of $x and getting $0 back for it. (Remember, the 25% SS funding deficiency is *entirely* due to $X -- and that's what both have to pick up. ) When the dollar cost either way is the same, and the dollar cost is the issue, comparing %s is specious.

Rather, your argument "If SS continues those that pay $X will get 75% back. But if they terminate SS they get 0% back" sure sounds like "getting back 75% of a bad investment is better than getting back 0% of no investment" which is of course the same as saying losing 25% of a bad investment is better than losing nothing.

Now, Friedman's plan doesn't result in a net "transaction gain" either, obviously, because the cost of paying off the already accrued benefit obligation $X is merely shifted to income tax payers. $X is the exact same whether paid through payroll or income taxes. So income tax payers lose the exact same amount that payroll tax payers gain, as this sunk cost of $X is shifted to them -- but income tax payers are richer and better able to pay it, so what the heck.

However, to say the people who have been carrying the payroll tax load under the current system get $0 from the termination of SS on paying off $X is just false. They are the WINNERS from the change, because they cease paying off $X. They cease paying anything. Which saves them from losing 25% of what they would otherwise have paid.

Now if that's not plain enough I'll spell it out in clearer (if not excruciating) detail in my post after yours on the other thread.

And after that, which ought to be more than enough on this, I'm sure we'll be able to agree to disagree. ;-)

Jim Glass writes:

"As for the fact that our SS liabilities do not show up as debt being sloppy accounting that's true but that's also true of most other expenditure items. We know we will have to spend trillions of dollars in defense in the future. Why don't we include that as debt?"

I assume you mean "as an accrued liability". For the same reason that General Motors does not include the billions of dollars in salaries it expects to pay in the future as an accrued liability -- because it's not one. Salaries are a current operating expense, and those are matched against current income. Neither current expenses nor current income are accrued in any fixed known amount regarding the future -- they can and will be managed to balance each other in real time, as things develop.

But if GM legally commits to pay $X in pensions and retiree medical benefits in the future at a known time, that's a legally enforceable expense determined *now* that will have to be paid in the future from current operating earnings then *in addition* to current operating expenses. Thus, the present value of $X, call it $P, reduces GM's value as an entity. Legally, GM can account for $P in two ways: It can fork over $P of assets to an independent trust fund to finance the future pension benefits, or it can decide to ultimately pay the $X out of income when it comes due while currently incurring a liability of $P on its books. Either, obviously, reduces the value of its stock by $P.

But suppose it did neither -- it instead decided to pay $X when it came due from current income, but not to record the liability for $P and take the charge for it. Then today's investors would overpay for its stock by $P -- and wind up, um, "unexpectedly poorer" when $X came due. While the current owners would gain by selling their shares for $P more than they were worth. Ooops. Hiding liabilities like this is *exactly* why Fastow of Enron is facing 100 felony counts and headed to the slammer.

And it is also is exactly the same accounting the federal government uses, so politicians can benefit by talking of "surpluses" (and by spending them, through both new programs and tax cuts) while liabilities are going up $2 trillion a year.

There are several easy fixes, of course. As per Friedman, the gov't could simply issue $24 trillion in new bonds plus $2 trillion or so more each year to recognize the growing accrued liabilities for SS and Medicare. But can you imagine the collective *stroke* so many people would have at the sight of that exploding pile of debt? Yet it wouldn't increase the govt's real debt in the form of its obligations by even a penny. People would just *see it* that's all. Which is why Congress doesn't do it. To prevent all those strokes from driving up medical costs. ;-)

"Does anybody really think that claims on justice and police functions are really any less than claims on SS?"

Sure, they are current operating costs, not accrued future liabilities. Look, *everybody* uses accrual accounting except the federal gov't, including state and local governments that run police forces and court systems. Generally Accepted Accounting Principles exist for governments too -- except Congress has excepted itself, for some reason.

"Or are we going to add every single line item we expect in the future to the debt calculations?"

Only according to GAAP rules, like the state and local governments that follow GAAP do.

"If so we should add all the expected tax revenues to the asset side."

Of course not. But we would include all accrued assets -- e.g, liabilities owed *to* the federal gov't, such as bonds issued by others and owned by the federal gov't.

GT writes:


Jim,

I'll start with the easier one. You say that : "Neither current expenses nor current income are accrued in any fixed known amount regarding the future -- they can and will be managed to balance each other in real time, as things develop.

But if GM legally commits to pay $X in pensions and retiree medical benefits in the future at a known time, that's a legally enforceable expense determined *now* that will have to be paid in the future from current operating earnings then *in addition* to current operating expenses. "

I agree 100% with everything you wrote. My only point is that when talking about central governments the situation changes a little. GM's pension commitments, as you point out, are legally enforceable. But the SS obligations are not. That's a pretty big difference IMHO. Congress can pass a law changing SS benefits tomorrow if they wish and get rid of the unfunded benefits overnight. Of course there may be a political cost but that's a different story.

SS benefits are just like defense or justice in that sense. They are simply the result of political decisions of our elected representatives that can and will be changed periodically. GM can't say to its pensioners "Guys, I don't have enough money so I'm cutting your benefits by 20%". The US government can and there's no legal recourse to that (other than voting new representatives).

I do think we need to take into account what we expect to pay in the future and I agree we should reflect it in our current budget (and if we had that in place today there probably wouldn't have been any taxcuts) so I do agree that we need to use accrual accounting. At the same time I also think that we have to remember that there is this fundamental difference between the US Govt and GM in that one of them can unilaterally change the rules while the other can't. When the time comes to pay for unfunded SS the government will have to balance that against all the other claims (justice, defense, welfare, etc..) and politically decide what we pay and to whom based on available resources.

That's why I talked about adding tax revenues to the asset side. Since the money for SS and every other govt payment comes from the same source-taxpayers- if there is more available in the future, because the economy grows faster than expected say, than our liabilites may disappear.

This flexibility is not available to GM or any other company. That's why I find accrual accounting a bit misleading when talking about the US. Just a bit though.

Jim Glass writes:

"GM's pension commitments, as you point out, are legally enforceable. But the SS obligations are not."

What? What? The goverment's SS obligations aren't *real* obligations? Next you'll be telling me that the SS trust fund isn't a real trust fund, Congress just calls it that because ... gosh, why would they do that? ;-)

One of the benefits of accrual accounting -- and of Friedman's idea of issuing enough bonds to fully fund SS openly -- is that we find out if Congress is serious about these things, actually willing and intending to pay for what it promises.

Shouldn't politicians who *profit* from creating new financial obligations for the future be on the hook to put aside appropriate funding for them? If they aren't, don't they have a clear incentive to overpromise, and to be .. um ... disingenuous about it all?

I mean, isn't it disingenuous for a politician to say, "Vote for me, and I *promise you* this very valuable benefit in the future" ... and then after he gets elected on that promise turn to the accountants and say, "Hey, we don't have to account anything for the cost of that, because we can always cut it the way GM cuts workers in a bad year."? Would he have told the voters that??

"That's a pretty big difference IMHO."

You're right, having Congress take accountability for the cost side of its promises so they would become meaningful would be a big difference from the status quo. Among other things, there wouldn't be a 25% funding shortfall imperiling the future of SS. There'd also be a lot less damage to the wallboard over at Brad DeLong's site due to the impending "national bankruptcy" in 2030 or 2050 so.

"That's why I talked about adding tax revenues to the asset side. Since the money for SS and every other govt payment comes from the same source -taxpayers - if there is more available in the future, because the economy grows faster than expected say, than our liabilites may disappear.

"This flexibility is not available to GM or any other company. That's why I find accrual accounting a bit misleading when talking about the US. Just a bit though."

If future revenue is larger than expected liabilities do not disappear, there is just more money to pay for them. Then if you want to use the money to accrue even larger liabilities, fine, you know how you'll pay for them. And that is the same situation that GM and other entities that use accrual accounting are in.

You are right, though, that there's an obvious difference between the federal gov't and GM in that the government is "above the law" since it makes the law, and so can't really legally bind itself in the future, generally.

But the point of having it use accrual accounting is to make its true financial situation more transparent to everyone, and thus to discipline it by making politicians answerable to better-informed voters.

It can't be a bad thing to give voters much better information about the state of the government's finances, the credibility of politicians' promises, and the willingness of politicians to take responsibility for paying for them.

GT writes:

As I said, I agree with accrual accounting.

My point about the 'flexibility' is that, at the end of the day, these are political decisions that must be made.

I certainly agree about politicians needing to be more honest in their accounting.

I take it you did not vote for Bush?

:)

As for Friedman's idea, leaving aside the 'cost' issue (I'll try to respond in a day or so, at least to finish it) is that it's not about making costs clear, or at least it's not only about that. Friedman wants to get rid of SS. That's a different thing.

Oh and your right. I didn't mean to say that more revenues made our liabilities disappear. Only our unfunded ones.

Patrick R. Sullivan writes:

Just when I thought GT had finally surrendered to the inevitable, by writing:

" Congress can pass a law changing SS benefits tomorrow if they wish and get rid of the unfunded benefits overnight. Of course there may be a political cost but that's a different story.

" SS benefits are just like defense or justice in that sense. "

since the above had been pointed out to him in posts in the earlier thread (by Paul Zrimsek and myself). Hence, there is no "transition cost".

But now I see that GT apparently doesn't grasp the implications of what he admitted. Because he's back to:

"...it's not about making costs clear, or at least it's not only about that."

In fact, that is exactly what Friedman's scheme does. And what it doesn't do is create (absent some minor transaction or clerical costs) a "transition cost".

Then, oddly, GT says:

" Friedman wants to get rid of SS. That's a different thing."

A thing GT has said he favors too! But it should be obvious that Friedman's wanting to end SS, DOES NOT create a "transition cost" either.

Finally:

" Oh and your right. I didn't mean to say that more revenues made our liabilities disappear. Only our unfunded ones."

No, it doesn't. GT needs to brush up on his accounting.

Jim Glass writes:

"As for Friedman's idea, leaving aside the 'cost' issue ... it's not about making costs clear, or at least it's not only about that. Friedman wants to get rid of SS. That's a different thing..."

No, the point of Friedman's article is to make the real cost of SS more clear and understandable, and clear up the current confusion in the public debate. As he concludes: "I believe that the ongoing discussion about privatizing Social Security would benefit from paying more attention to fundamentals.."

It's true that he personally would like to end SS altogether. But that's not the point of the article, which is that there is *no transaction cost* involved in moving from the current SS system to either no system or *any other* system (privatized *or* governmental) that is fully funded, if current obligations are fully paid off -- and that once people *realize* this, the illusion that we are "locked in" to the present system somehow via the payroll tax mechanism will go away, and we'll realize we have a completely free hand to act to improve the system however we wish.

"Oh and your right. I didn't mean to say that more revenues made our liabilities disappear. Only our unfunded ones."

Not true. Increases in revenue generally don't decrease future unfunded liabilities. For SS in particular, you'd need an increase in future *payroll tax revenue* -- and that wouldn't do it either, because SS benefits are indexed to wages subject to payroll tax. So if the wages/tax receipts go up unexpectedly, benefits owed do too.

Jim Glass writes:

"As for Friedman's idea, leaving aside the 'cost' issue (I'll try to respond in a day or so, at least to finish it)..."

I've decided not to consume our host's bandwidth by putting up a 100,000 word response to your last post in the other thread. But I still don't understand where you think the "transaction cost" is coming from. Here I'll give what I think our differences are, as briefly as I can. Then you can have the last word, and we can agree to disagree and move on with our lives.

Friedman and I (and I think Paul and Patrick) say: The cost of all benefits earned but unfunded under SS as of today is $X. This payment must come from taxpayers in the future. The cost of all future retirement benefits earned as of tomorrow forward is $Y and must also be paid by taxpayers in the future. Ergo, taxpayers of the future must pay $X & $Y but receive only $Y, and thus are going to be out $X.

This $X is a *sunk cost* -- it is unavoidable, somebody *must* pay it either through increased taxes or SS benefits that are reduced from the currently promised level, or both. But we have a free hand to decide who pays it and how. E.g.:

(a) we can keep the status quo and have SS participants take an $11 trillion loss over the next 75 years by receiving only a 75% recovery of payroll taxes they pay, or
(b) we can close the funding gap by supplementing payroll taxes with $11 trillion of income taxes over the next 75 years, shifting the loss to income tax payers while making payroll tax payers whole, or
(c) we can end the current SS program tomorrow and have income tax payers pick up everything owed under it, which is $X, and then start a new system from scratch (or not).

Now, under all three options $X=$X=$X, and there is no other cost that kicks in under only one of them, so future taxpayers must pay $X WHATEVER option we choose, and these in substance are the ONLY options possible.

Ergo, we have a free hand to modify SS as we may wish, at a net cost to future taxpayers of $X in current dollars, no less and NO MORE, and thus with no transaction cost. (Although $X continually *increases* as we delay doing so -- and this *is* a real cost).

OK. Now as I understand your replies to date, your objections have been...

1) While those who pay $X under Friedman's "terminate SS option" (the income tax payers) get "0% on their return" for doing so, under current law payroll tax payers will get 75% back on their payroll taxes, which is much better than 0%, so the Friedman option thus imposes an extra cost.

But this is the same as arguing that since getting 75% back of $44T gives a much higher rate of return than getting back 0% of $11T, it is a *less costly* option. This is false, obviously. The dollar cost is the same either way.... and/or

2) Individuals in the near future will pay much less under current law than under the Friedman option, since SS is a "rollover machine" that is "fully funded until 2044", while if SS is terminated today income taxpayers will have to pick up $600B a year starting tomorrow and get nothing for it.

But that's not necessarily so. Treasury bonds financed with income taxes are a "rollover machine" as well. If SS is terminated and $X is funded with T-bonds, we can choose to roll them over so the *net* amount of income tax collected to pay off $X on a year-by-year basis exactly matches that to be collected under current law from payroll taxes. That is, the loss to income tax payers can exactly match the loss that otherwise would occur to payroll taxpayers every year. Again, we have just a *shift* of cost to income tax payers from payroll tax payers, with no change in the total cost at all even year-by-year .... and/or

3) The cost $X of the accrued liability of SS today which must be paid off in the future somehow actually *differs* -- and is *less* -- if we continue SS under current law, compared to if SS is terminated today as per Friedman.

You stated this when you said "You claim that $X and $U [unfunded SS liabilities for the next 75 year] are both about $11 T. You also say that 'The two $11 trillion costs are not coincidence, obviously.' This is a big point. If I read you correctly you are saying that $X and $U are related, maybe even the same thing. But they aren’t. They are two separate, even unrelated concepts. An example will make this clear. Suppose that SS was perfectly balanced and expected to be perfectly balanced forever. In this example the demographics don’t change and the beneficiary-worker ratio never changes. What would $U be? Well, zero!..."

But this is mistaken. Though in my midnight postings I may have garbled things by going back and forth between the 75-year and permanent horizon. Apologies for that. Let me try to make it clear now:

With any pension system funded only by employee contributions (say, taxes) and only paying benefits going forward, the value of benefits paid and taxes collected must equal each other when discounted to current value, because they are the *same money*. A dollar of tax collected *is* a dollar of benefits. (If any taxes are invested in bonds in a trust fund until needed, the interest earned on them provides the discounting adjustment).

So if you discount both taxes and benefits to current value they will match to the dollar. However, if you subtract a chunk of money collected through the taxes and transfer it to another purpose (such as giving it to a generation around before the system started, that never paid any tax) obviously the discounted current value of the future benefits is reduced by that chunk, while that of the future tax side is not reduced.

So future taxpayers will receive less than they put it -- take a loss -- that can be measured at current value. And the amount of their future loss is an *identity* to the amount of the past transfer. It's just arithmetic. First T=B, then subtract X from B, now T>B by X.

Now if one just looks at an arbitrary limited time period (like 75 years) it's not an identity any more, because the later period is cut off, there are demographic effects, and whatnot. But any benefit deficit/loss that people suffer in that time period still results entirely, 100%, from the transfer, and from *nothing else*.

This is easy to see. Imagine there is *no* transfer of receipts out to the earlier generation -- as with FDR's own originally enacted SS program, which didn't provide such a transfer and wasn't going to pay full pension benefits until the late 1970s(!) contrary to all popular myth. With all contributions accumulating in the trust fund for 40 years before full benefits even *begin* being paid, obviously no "funding gap" will occur.

More to the point, with each person's lifetime contribution accumulating in the trust fund before that person retires, no funding gap *can* occur *regardless of demographics*. The worker/retiree ratio can plunge like a rock -- every retiree's benefits will be -- MUST be -- fully funded! Even if the number of workers drops to zero!!

So the past transfer creates 100% of the future benefit shortfall -- and discounted to current value they are an *identity*. Without the past transfer the future funding shortfall would be $0. Demographics do NOT create the future funding shortfall, but play a role only with the timing and rate at which the dollar cost of the past transfer is born in the future -- $Z of loss to SS taxpayers within Y years.

Note that demographics have been against SS from the start with the worker/retiree ratio declining for decades. The reason SS has provided a positive return until now is because Congress has steadily *raised taxes* to do so -- from 3% to 12.4%, which in % terms is quite an increase. But that game has ended. Krugman called SS "a Ponzi game ... in which each generation takes more out than it put in. Well, the Ponzi game will soon be over". And he is right, SS is fiscally unsound BECAUSE "each generation takes more out than it put in", i.e., the backward transfer, and that can't go on. Krugman's not a dummy (on economics).

The next generation will take an $11 trillion loss over 75 years *driven 100% by* the transfer to the past generation. And if you go far enough into the future beyond that the real future loss, discounted to current value, becomes an *identity* with the past transfer $X. The future loss can come from nothing else! It's arithmetic.

Now, it is true that demographics may *defer* the realization of the loss by payroll tax payers resulting from their financing of $X, spreading the loss over a long period into the future if there is a rollover effect. (Or condense the loss into a short period if there are not enough taxpayers to finance a rollover.) BUT, as already noted, financing the loss with T-bonds can produce the *exact same* rollover and deferral, if we wish. So the rollover effect washes out.

Either way, the cost that is being rolled over is $X at current value. There is nothing else that can go into the future unfunded liability, and there is *no way* to reduce that shortfall resulting from $X in current value terms, only to defer paying it.

Ergo, financing that transfer is all the cost there is in terminating SS, and we must incur that cost ANYHOW even if it isn't terminated, as there is no fourth alternative to the options (a) (b) and (c) above.

So there is no transition cost. We are free to end SS outright, to replace it with a private plan or a fully funded public plan, or to keep it going as is, with no net "transition cost" incurred among those different options.

Thus, we are free to enact a new plan for purposes of equity and economic efficiency, undeterred by any extra, new, net transition cost.

I vote for one that on a going-forward basis creates fully-funded private accounts from which everyone will receive a positive return (rather than a negative one) and in which people have property rights, while SS's accrued liabilities of $X are fully paid off in the future with the cost being shifted from poorer payroll tax payers (as per now) to richer income tax payers. And I vote to do this soon, before $X grows a whole lot larger to all our cost.

There -- not quite 100,000 words but plenty enough. Finis.

GT writes:

Jim,

Yes, I’ll post one last time and we can leave it at that.

On the impact of a growing economy on SS’s unfunded liabilities: I recall reading that if the economy grew in the next 75 years at the same rate it grew in the 96-2000 period (about 3.5% a year compared to the 2.5% a year we had between 73 and 95) then most, if not all, of SS’s unfunded liabilities will disappear. I’m not sure how that works. Maybe greater labor participation rates? Maybe the SS benefits formula is not linear so that those that pay 100% of the possible SS tax (which maxes out for salaries around $85K I think) get proportionately less? Or maybe I was wrong or what I read was wrong. If so I stand corrected. But that’s what I was referring to.

On Friedman’s point not being about shutting down SS I have to wonder if you are kidding. Friedman makes it very clear that that’s his final objective. Arnold Kling’s question was very clear. He asked “Should (Democrat economists that want greater fiscal discipline because of the long-term outlook for Social Security) support Friedman's idea as a way of making the liabilities in the entitlement programs more transparent?”

The answer is clearly no. If you want transparency there are other ways to do so. Friedman’s solution goes way beyond that and results in the end of SS, which I suspect most Democrat economists don’t support. That’s like saying that the solution to a house pet that has been tearing up the garden is to give him away. That solution only makes sense if you wanted to get rid of the pet or if that was the ONLY way to protect the garden. Otherwise there are other solutions, like fencing the garden, keeping the pet on a leash etc… Similarly there are many other solutions to the problem of transparency like a change in the government’s accounting that can solve the problem Democrat economists are worried about without getting rid of SS.

As for the transition costs I won’t repeat all I wrote. It seems to me you are using different definitions than I am and I certainly don’t agree with how you define the unfunded liabilities.

Rather than go over what I already wrote I will simply say something I should have focused more on the beginning. Many, many people have written about transition costs. Krugman has many times accused the Bushies of ignoring the transition costs in their SS proposals and, with the exception of Friedman’s article, I have yet to see anybody counterargue that there are no transition costs. Not even in the blogging world where posters have attacked Krugman for the silliest things, including the title of the chapters of one of his books.

The CBO has looked at transitions in other countries (like Chile or Argentina) and how they handled the transition costs. In other words, they recognize there are transition costs.

Finally even very strong supporters of SS privatization like the Cato Insitute recognize the transition costs (although they argue that they are offset by the potential benefits of the transition, something I have always agreed with).

See, for example this:
“The major political challenge to effective Social Security reform involves managing the transition costs from a 100-percent government-funded system to some form of privatized investment. Critics argue that the reduced payroll taxes coming from workers who elect to start managing their funds privately would cause a shortfall to the Social Security trust fund that pays benefits to current retirees and those workers who do not opt out of the system. How could this shortfall be made up?
Studies conducted by the Cato Institute and other researchers have shown how transition costs to a privatized pension system can be managed.”
In other words there is a cost (my point all along) but it may be more than made up by other benefits.

The Cato report is here: http://www.cato.org/pubs/ssps/ssp13es.html.

Patrick R. Sullivan writes:

Since GT is now appealing to the authority of Cato, perhaps he'll want to deal with:

http://www.cato.org/pubs/ssps/ssp13c.pdf

where he can read:

" One of the more common concerns about
transforming Social Security’s pay-as-you-go
financing into a market-based structure is the
transition cost. Critics claim that people would
be unduly burdened because they would have to
pay twice—once for their own retirement and
once for those already retired. This double
expense would be so prohibitive, it is argued, as
to warrant rejecting privatization even if it were
meritorious on other grounds.

" Such arguments ignore the enormous
unfunded liabilities of the current Social
Security system. Any valid discussion of the
costs of moving to a market-based Social
Security system must compare those costs with
the costs of maintaining the current system,
including the costs of meeting those unfunded
liabilities.

" The mechanisms for paying those costs
remain the same whether one attempts to prop
up the existing system or shift to a new,
market-based system....

" Therefore, redesigning Social Security as a
market-based system of personally owned
retirement accounts does not actually entail any
new costs."

IOW, just what Jim, Paul, and I have been telling you.

GT writes:

Patrick,

I long ago accepted that on a net basis there may be no costs. In fact I would agree with Cato that on a net basis a privatization may actually result in a net benefit.

I have not been discussing this on a net basis which requires comparisons across time. I've always focused just on the current impact WITHOUT taking into account the possible benefits 20 or 30 years from now because it is now that the political difficulties need to be resolved and addressed.

Cato recognizes that. They realize there is an upfront cost that is more than compensated by the benefits in the future. I have never disagreed with that.

So no it's not what I have been debating with Jim since I never disagreed with this.

GT writes:

Patrick,

If you read the whole article you may have noticed the conclusion Cato reaches?

"Transforming Social Security to a marketbased system of individual accounts does require a transition period, a cost, and a change in the timing of cash flows. However, any cost is reasonably less than staying with the present law. Little if any reason exists to retard necessary reform because of this issue. Indeed, the relative transition benefits should be another incentive to move quickly toward reasoned
reform."

Notice the conclusion. There is a transition cost (my point all along) but it is a cost that is more than offset by the benefits of the change (something I have agreed with all along).

My recommendation for next time? Read the whole article, not just the first page.

Also note how Cato talks of "a change in the timing of cashflows" which is another way of saying that the transition costs have to be paid today while the SS unfunded costs wouldn't have shown up until much later in the future which means it's not exactly the same people that will pay the costs even if on a net basis it all washes out (another one of my points).

Patrick R. Sullivan writes:

Hmmm. So, now it's not about the benefits being "given up", as you claimed earlier, numerous times? It's the costs described in the Cato paper as:

" An illustration is the tradeoff in refinancing a home mortgage. Costs, such as points, title insurance, title search, attorneys’ fees, credit report and the like, incur a cost over and above present law....

" but these transaction costs...."

However, Jim Glass said:

" In fact, that is exactly what Friedman's scheme does. And what it doesn't do is create (absent some minor transaction or clerical costs) a 'transition cost'."

So, I'm wondering why you didn't bother to inform Jim of this when he raised it? (And Jim is correct, those costs are trivial compared with the unfunded liabilities)

And, did you notice the chart in the upper right hand corner of page three, where the cash flow shortage is depicted WITHOUT a transition?

Comments for this entry have been closed
Return to top