Wednesday, December 01, 2004

More From Max on Framing the Social Security Debate

In response to my last post about Social Security reform, Max Sawicky asks:

[I]sn't it radically inconsistent to say (cash-flow) deficits matter, but that a long-term in-go (e.g., lower projected Social Security benefit payments) washes out a short-term out-go (a payroll tax diversion to individual accounts, replaced by additional borrowing)? Haven't we heard in the past that some current added borrowing will pay for itself with some gain in the future? If you think that, for instance, the returns to pre-natal care are subject to dispute, what about the two-thirds of the Social Security Trust Fund shortfall derived from years beyond the next 75, lo unto eternity?
I don't see any inconsistency. Let me try to address all the pieces of Max's post. (If casual readers feel their interest starting to wane in all of these details, feel free to skip down to the last three paragraphs to see where I would like the debate to go next.)

1) Do deficits matter?

Yes. Every dollar that the government borrows today must be repaid, with interest, in some later period. The deficits matter today because they imply a repayment next period, and resources have to be found to make that repayment. All deficits matter--whether they are on-budget or off-budget, whether they are run in the current year or future years. (The question, "Do deficits matter?" is often asked in the context of whether they increase interest rates. This is an interesting but, for the purpose of this discussion, less relevant consideration than the simple fact that they matter to the people who have to repay them, regardless of what they did to interest rates in the interim.)

2) Can a long-term in-go wash out a short-term out-go covered by borrowing?

Yes. The interest rate tells us the amount of in-go we would need in the future to cover a dollar of out-go this year. If we run a $1 deficit this year, and the interest rate is 3%, then we need $1.03 next year to balance the books. Or we would need 1.03*1.03= $1.0609 in two years, or 1.03^30 = $2.43 in thirty years, etc.

3) Haven't we heard in the past that some current added borrowing will pay for itself with some gain in the future?

I cannot stipulate to everything that Max has heard, so I'll give this a qualified "not on my blog." I discussed something that might be confused with this assertion in the context of Senator Kerry's campaign statements about the reform plan analyzed in Chapter 6 of the Economic Report of the President. That reform plan--Commission Model 2 under the assumption that 100 percent of eligible workers opt for the personal accounts and that debt is issued where needed to cover expenses--does generate some current added borrowing. It also gets repaid, as part of the reform, but in no way is it "paying for itself." It gets repaid, as in #1 and #2 above, precisely because individuals who opt for the personal accounts while working agree to receive lower benefits from the pay-as-you-go system later in life and because the reform plan includes reductions in benefits for future retirees relative to current law quite apart from the personal accounts.

4) If you think that, for instance, the returns to pre-natal care are subject to dispute, ...

I confess that I'm not sure what Max means by the reference to the returns to pre-natal care, so I'll just state for the record that I strongly believe in spending money to encourage pre-natal care. Also, breastfeeding and doulas, but those are topics for a different post.

5) ... what about the two-thirds of the Social Security Trust Fund shortfall derived from years beyond the next 75, lo unto eternity?

I have referred a couple of times to the $10.4 trillion shortfall in the Social Security system. This is the present value of the excess of projected benefits less projected revenues, and it is estimated in this table of the 2004 Trustees Report. The table shows that, evaluated only through 2078, the present value of the excess of projected benefits less projected revenues is $3.7 trillion, or about 1/3 of the $10.4 trillion. Max is referring to the roughly 2/3 of this $10.4 trillion that is due to deficits incurred more than 75 years into the future.

Well, what about them? Is Max suggesting that it would be inappropriate to consider the deficits that are incurred during the lifetimes of our youngest citizens? I don't think so. Is Max suggesting that it is appropriate to count the taxes that people pay on their earnings in, say, 2040, without also counting the benefits that they earn as a result, to be paid in, say, 2080? I certainly hope not. I favor the $10.4 trillion measure because it is the most comprehensive summary of the program.

With any projection, there are two important concerns. The first is whether it is biased. Is it systematically over- or understating future deficits? Have the actuaries made defensible assumptions about the key underlying parameters? Interested readers can go to Section V of the 2004 Trustees Report for the detailed explanations. I served on a 1999 technical panel of the Social Security Advisory Board to offer guidance on assumptions and methods used by the actuaries. From that experience, my general view is that improvements in mortality are understated in the projections and productivity growth is probably a bit too low. (The latter depends on how permanent we believe the post-1995 uptick in productivity will be. It is looking more permanent as time goes by.) These two deviations tend to offset each other. The 2003 technical panel made similar recommendations. So I am inclined to take the $10.4 trillion number as a reasonable starting point.

The second concern is that a projection, even if unbiased or right on average, may be imprecise. This concern gets more important as the projections are made for more and more distant years (and I think this is Max's broader point here). The 2004 Trustees Report addresses this imprecision in an appendix, in which simulations are presented that attempt to account for the historical variation in the key parameters (like the fertility rate, the mortality rate, and productivity growth). The end result is this figure, which shows that there is a 97.5 percent probability that the annual deficit will be greater than 1 percent of payroll in the 75th year (with larger deficits in prior years). It also shows a 10 percent probability of an annual deficit of greater than 11 percent of taxable payroll in that year. To me, reasonable attempts to take the imprecision of the central projection into account suggest more, not less, of a reason to address this problem as soon as possible.

In the interest of moving the debate along, I'll say that I am not wedded to the infinite horizon estimate of a $10.4 trillion shortfall as the only metric for gauging reform. Here's an alternative criterion that Max and other folks who are "going to the mattresses" might find more constructive.

Choose a projection period over which you feel confident in the accuracy of the underlying economic and demographic assumptions, subject to the constraint that it is long enough to cover the retirement of the baby boom generation. Provide specific reforms to the system such that the Social Security trust fund is positive and trending upward in the last years of that projection period. Do not use any gimmicks related to benefits or costs in those last few years.

I have seen exactly one such plan [from people who criticize personal accounts], and the Democrats in Congress refuse to endorse it. I would be happy to see another from Max, and even happier still to see his commitment to use the reputation of his organization among Democrats to move it in Congress.

9 comments:

Anonymous said...

Keep in mind that Social Security, even on the conservative assumptions of the Trustees, is in balance--i.e., can pay full benefits without any increase in total government debt--through 2042 or 2044 (I've seen both years listed recently as the so-called "day of reckoning"). By that date any living Boomers will be in their 80's, at least. The CBO projects SS to be in balance through 2052. So with no changes, we are going to be able to almost completely cover the retirement of the baby boom generation, unless the economy goes completely into the tank. So the question is, to what degree should present day fiscal policy decisions be drive by forecasts of deficits to be incurred starting no less than 38 years from now, and probably more than that? I would argue that the dark forces of time and ignorance will inevitably defeat efforts to make accurate forecasts with such long time horizons, and that therefore the best course for now is to do nothing, because the very strong likelihood is that any action we take now, with such minimal knowledge, is going to be wrong. If it is politically necessary to do something to make the "future deficit" number smaller (e.g., if large majorities are persuaded by the arguments of economists arrogant enough to believe that tney know the unknowable), the first step should be to simply eliminate the cap on earnings subject to the payroll tax.

Anonymous said...

Provide specific reforms to the system such that the Social Security trust fund is positive and trending upward in the last years of that projection period.

Tax the rich.

Anonymous said...

[Dean Baker -- http://www.cepr.net] In response to Andrew's defense of his use of an infinite horizon for assessing the projected Social Security shortfall, it would probably be helpful to keep in mind what is going on here.

Suppose that we write down a schedule of taxes and benefits in which the taxes stay constant as a share of GDP and the benefits rise at the rate of
0.1 percentage point a decade. Obviously this system will eventually run a deficit, and at some point a very large deficit.

This is effectively what we have done with Social Security. The reason the benefits keep rising through time is that life-spans are projected to increase for the indefinite future.

The question is, how do we think about a deficit over the indefinite future.

The Social Security trustees give us projections over the program's 75 year planning period, which we know to be 0.73 percent of GDP over this period.

Andrew prefers the dollar measure (@ $3 trillion) and would have us look over an infinite horizon, which takes it to more than $10 trillion. The question is, how concerned should we be about the deficit that is projected beyond the 75-year projection period.

To me, this is purely a question of education. If our children and grand-children (or great grand children) are too dumb to figure out that at some point they will either have to increase their taxes or cut their benefits (which could include raising the retirement age), then we will have failed them terribly. Note, the issue is not a transfer from future generations to anyone who is alive and working now, the question is a potential transfer from the distant future generations to less distant future generations.

Of course, we don't get to make policy for future generations anyhow; they will make that for themselves. So, if we ensure that we maintain decent edcuational standards, I don't see any reason to believe that our children and grandchildren won't be able to design a properly functioning Social Security system for the 22nd century. Unfortunately, I think we have more important issues to worry about right now.

JG said...

Anonymous wrote:

"Keep in mind that Social Security, even on the conservative assumptions of the Trustees, is in balance--i.e., can pay full benefits without any increase in total government debt--through 2042 or 2044"

This is false.

The moment SS obligations start being paid from the SS trust fund the public debt must start going up dollar-for-dollar (or the equivalent: taxes must rise or other spending must be cut dollar-for-dollar)

After all the, government must get the money to pay off the SS bonds from somewhere!

Note that all the trust fund bonds amount to government public debt of $0, because they are issued by the gov't to itself, and so are both an asset and liability to it of equal amount netting to $0 on its balance sheet. They are exactly the same as an unfunded, unsecured, mere promise to pay money that will be raised later.

Thus, raising $1 trillion (and more) later to pay off the promises represented by the trust fund bonds will increase the government's public debt by $1 trillion (and more).

"The CBO projects SS to be in balance through 2052. So with no changes, we are going to be able to almost completely cover the retirement of the baby boom generation, unless the economy goes completely into the tank. So the question is, to what degree should present day fiscal policy decisions be drive by forecasts of deficits to be incurred starting no less than 38 years from now, and probably more than that?"

This is completely false.

GAO projects just the *interest* on the national debt will reach 20% of GDP -- as much as the entire federal government today -- in about 35 years, due to unfunded entitlement spending (SS, Medicare, Medicaid) and ended its 75-year projection there, saying 'government will end', further projection is implausible.
http://www.gao.gov/new.items/d01385t.pdf

And that was a year 2000 projection assuming we'd still be running surpluses in 2004.

JG said...

"Note, the issue is not a transfer from future generations to anyone who is alive and working now,"

No, it is.

"the question is a potential transfer from the distant future generations to less distant future generations."

Not so. The break point is the year 2000. All annual cohorts retired before 2000 get back *more* from SS than they put it. (Which is why it is so popular with them). All retiring after 2000 will get back *less* than they put in (putting the future popularity of SS very much in doubt) to pay for those who are alive and well and getting back more now, plus the more that went to earlier retirees.

C'mon: Someone who retired in 1990 started off paying 3% payroll tax on a low wage base, to get full benefits today.

Someone young who began work in 1990 started off paying 12.4% tax on a much larger wage base, to get benefits that were *reduced* for him (but not for the 1990-retiree) by the 1980s legislation. And even *those* benefits are 30% underfunded and so will have to be cut in the future.

Why do today's young get such a worse deal than today's retirees?

Because SS went broke in the 1980s, so Congress decided to hit the workers of the next generation with a bad deal to fund the great deal that had been promised to older workers and retirees *then.*

After all, if you were a politician then, which generation would you have wanted to please?

Anonymous said...

This is false.

The moment SS obligations start being paid from the SS trust fund the public debt must start going up dollar-for-dollar (or the equivalent: taxes must rise or other spending must be cut dollar-for-dollar)

After all the, government must get the money to pay off the SS bonds from somewhere!

Note that all the trust fund bonds amount to government public debt of $0, because they are issued by the gov't to itself, and so are both an asset and liability to it of equal amount netting to $0 on its balance sheet. They are exactly the same as an unfunded, unsecured, mere promise to pay money that will be raised later.

Thus, raising $1 trillion (and more) later to pay off the promises represented by the trust fund bonds will increase the government's public debt by $1 trillion (and more).
This is false. The government currently pays interest on the bonds held by the SSA, so they are part of the national debt. When the SSA starts to redeem those bonds, they will be retired by issuing new Treasury bonds. The government's debt service burden will be unchanged. Red-herring arguments about notional "government balance sheets" can't change that fact.

JG said...

Anonymous wrote:

"The government currently pays interest on the bonds held by the SSA"

False. Who does it pay the interest to?

Just to itself. When you pay interest to yourself, how much are you paying?

"so they are part of the national debt."

Really? The Treasury says the public debt was $4.41 trillion as of yesterday. With no SS bonds ($0.00)listed in that. Why not?

Go to the Treasury's consolidated balance sheet of the US and see if the SS bonds are listed as debt there.

And btw, if the SS bonds were *really* going to be used in lieu of taxes or other new borrowing to finance SS, they'd have to be an *asset*, not debt.

"When the SSA starts to redeem those bonds, they will be retired by issuing new Treasury bonds. The government's debt service burden will be unchanged."

So what you're claiming here is that if you want to buy yourself, say, a $100,000 vacation residence when you retire, you can write yourself an IOU of $100,000 now, and pay interest to yourself on that debt to finance it. Then when you retire you'll cancel that debt to yourself and replace it with a $100,000 mortgage from a bank.

And then the bank mortgage, because it will merely replace the debt in the same amount that you owe to yourself, will *not* increase your personal debt or your interest costs at all.

Cool. We should all finance ourselves this way. I'm going to write an IOU to myself for $100,000 right now to pay for my kids' future college costs! That way all those future college loans won't increase my debt.

Oh, but have you figured out what stupid mistake GAO made, at that link I provided, in projecting the *interest* on the national debt rising to 20% of GDP in just 30 years?

"Red-herring arguments about notional 'government balance sheets' can't change that fact."

Ah, the rules of accounting and balance sheets are "red herrings". ;-) Well, they were for Enron. GAO budget projections are red herrings too?

Anonymous said...

When you pay interest to yourself, how much are you paying?According to the Social Security Administration, in the fiscal year just ended, they received nearly $88 billion in interest payments on the roughly $1.6 trillion in treasury securities which they currently hold. I guess you better set them straight on that, Jim--make sure they understand that that $88 billion doesn't exist, and that they are deluded to think that they have been using it to help meet their expenses:)

Really? The Treasury says the public debt was $4.41 trillion as of yesterday. With no SS bonds ($0.00)listed in that. Why not?The problem here is that you don't understand the difference between the public debt, the total dollar amount of debt held by private individuals and institutions, with the national debt, which is the total bonded indebtedness of the US government, regardless of who holds it.

The Treasury pays interest on the entire national debt, Jim. That means their interest expense would not be affected when the SSA bonds are redeemed by selling new bonds to the public--it's just another example of rolling over debt, something the Treasury does all the time.

Here's a thought experiment to help clarify this issue. Your claims imply that the debt held by the SSA isn't "real." If the debt "isn't real," then the Treasury would be able to repudiate it without consequence--they could announce today that as of January 1, say, they no longer consider any bonds held by the SSA to be obligations of the US government. The financial gain from doing so is obvious--they could at a stroke of a pen wipe out over 20% of the national debt, and reduce the current budget by $88 billion. So why don't they?

The reason this couldn't happen, of course, is that it would shatter the trust bondholders have in the US government. The bonds issued by the Treasury have no collateral--bondholders have no claim to any other assets should the Treasury default. The only backing they have is the credibility of the US government.

Should the government default on any part of its debt, regardless of the holder, it would destroy this credibility. The Treasury would suddenly find itself having to pay incredibly high interest rates to sell bonds and finance the enormous budget deficits being run by the Bush Administration. Current bondholders all over the world would abandon US Treasuries. Moreover, confidence in the US dollar would also be demolished, and the dollar would drop like a stone in foreign exchange markets (just as the Russian ruble collapsed in August 1998, after the Yeltsin government suspended payments on their debts). Given the importance of the dollar to the world economy, the outcome could well be a global economic downturn like nothing since the Great Depression.

So the debt held by the SSA, and by other government agencies like the Fed, is a genuine obligation of the US Treasury, and I am entirely correct in saying that when the SSA-held bonds are redeemed, the government will still be paying interest on the same amount of debt.

JG said...

Anonymous wrote:

"I am entirely correct in saying that when the SSA-held bonds are redeemed, the government will still be paying interest on the same amount of debt."

No you are entirely and completely wrong.

The tip-off to you about this should be your need to dismiss GAO reports, the Treasury's own financial statements, the rules of accounting, and plain logic about the value of interest paid to *yourself* to support your spurious belief -- and to replace them all with a "thought experiment" with holes that one could drive a 16-wheel truck though.

Look, here's something else from the government you can dismiss: its own Analytical Perspectives on the Budget, http://www.whitehouse.gov/omb/budget/fy2004/pdf/spec.pdf , page 48, quote:
~~~~

[Trust fund assts] do not consist of real economic assets that can be drawn down in the future to fund benefits...

.... the assets in the trust funds are special purpose financial instruments issued by the Treasury Department.

At the time Social Security redeems these instruments to pay future benefits, the Treasury will have to turn to the public capital markets to raise the funds to redeem the bonds and finance the benefits,
just as if the trust funds had never existed.

From the standpoint of overall government finances, the trust funds do not reduce the future burden of financing Social Security...
~~~~

OK? Now do you see those words "just as if the trust funds had never existed" and "the trust funds do not reduce the future burden of financing Social Security" ??

They mean exactly *what they say*: That when the government starts redeeming those trust fund bonds to pay SS benefits it will have to *increase* its debt, interest, and tax costs *" just as if the trust funds had never existed"* Get it?

Now you are arguing with the US Treasury here, not me. If you want to keep telling it that it doesn't understand its own finances, be my guest.

As to my not understanding the difference between public debt and total debt including intragovernment, ha, ha, I repeat my little thought experiment for you that you ignored.

If you *really believe* that converting debt one has issued *to oneself* into debt owed to others *doesn't* increase your real economic costs, try saving for your own retirement this way!

Issue IOUs to yourself now for a few hundred thousand dollars that you'll need later. Pay yourself full interest on them! Then when you retire, convert the IOUs to yourself into a debt you owe to a bank in the exact same amount.

Do your loan payment and interest costs go up now that you are paying them to *someone else*??

The US Treasury knows the answer to this question. Why don't you?

If you still refuse to see the answer, I'm afraid no one can help you.