Saturday, August 8, 2009

CBO releases new Social Security projections

The Congressional Budget Office yesterday released its projections for the future of Social Security. Like the annual Trustees Report, which in May showed a worsening in the program's financial outlook due to the recession and a projected increased in life expectancies, the CBO also shows the program's deficits growing larger. The long-term actuarial deficit is projected at 1.3 percent of taxable payroll, or 1.5 percent using (to my mind) more realistic assumptions regarding the growth of income taxes. (I discussed these baseline issues when last year's CBO report was released). Like the Trustees Report, CBO's projected shortfall grew by 18 percent from the 2008 through the 2009 projections.

Click here to read the full report and here to read a blog entry by CBO director Doug Elmendorf. CBO continues to do terrific work in this area, both on the technical end as well as presenting the material in an understandable way.


Jack said...

Thanks for providing the link to Elmendorf's blog giving a tidy summary of the CBO report. Otherwise one would have had to take your brief analysis as a sign of terrible things to come, though not for another 34 years. We all know how accurate economic projections three decades to the future tend to be.

"Without changes in law, CBO expects that the Social Security trust funds will be exhausted in 2043. If that point is reached, the Social Security Administration will not have the legal authority to pay full benefits and the amounts that could be paid would be about 17 percent less than those scheduled under current law."
D. Elmendorf
That is a really scary concept, in 34 years the Trust Fund MIGHT be shy 17%. Not broke folks, but a little light. In 34 years from now. Seems to me that most boomers, those greedy bastards that are going to break the SS Trust Fund, will be dead or near to it in 2043. That should take some of the strain of the system.
On the other hand Elmendorf notes that we need only raise the FICA a bit, 1.3%. He'd like to start now, but doesn't lay out any rationale for not waiting ten, twenty or thirty years to get a better picture of what's going to happen in 34 years. You see, Andrew, Mr. Elmendorf himself notes that these long term projections, while so much fun to make, are a bit fanciful as long term economic projections do tend to be. Does the term "irrational exuberance" sound familiar to you?

As Mr. Elmendorf notes,
"Many of the factors that will affect Social Security’s long-term finances are subject to significant uncertainty. Thus, a full exposition of projected finances includes both the expected outcomes and the inherent uncertainty surrounding such projections."

Significant uncertainty is the most certain part of the reported projections regarding the Social Security system three decades down the road.

Andrew G. Biggs said...


All good points. Some context:

First, while the trust fund is projected to be solvent until 2043, as you point out, the program itself will begin running deficits in the near future. That means that to the government as a whole, and therefore to the taxpayer, Social Security's rising costs are not a distant problem. As the CBO itself noted in this brief (
"Trust fund holdings, as internal liabilities between government accounts, are not assets of the government. Nor do they represent money owed to program recipients individually; payments to Social Security recipients and beneficiaries of other social insurance programs are based on a variety of rules set by law unrelated to trust fund holdings. A federal trust fund is basically an accounting device that measures the difference between the income designated for a specific program and the expenditures made to its beneficiaries. The accumulated difference, or balance, often represents a reserve of future "spending authority" for the program, but it is not a reserve of money for making payments."

Second, you treat uncertainty regarding future projects as a reason to discount projected deficits or to delay action to address them. But this ignores the fact that uncertainty works in two directions: outcomes are just as likely to be worse than projected as to be better. And since risk-averse people place greater emphasis on bad outcomes than good ones, uncertainty is an argument to reform sooner rather than later.

Third, while a 1.3 percent of payroll deficit sounds small, even as such it would amount to an 11 percent increase in what is already the largest tax most people pay. Moreover, the 1.3 percent figures relies on what is (in my view) a very unrealistic baseline regarding income tax rates, in which income taxes effectively double relative to GDP. This is historically unrealistic (taxes have remained fairly constant over time). Using a baseline in which income taxes, and thus Social Security revenue from income taxes levied on retirement benefits, remain constant relative to GDP, the deficit rises to 1.5 percent. And, this tax increase would not produce truly permanent solvency; it would imply levying higher taxes on workers over the next 75 years, but then having the program become insolvent in the 76th, as many of them retire. A sustainable tax increase, such that the program would remain solvent indefinitely without further changes, would be roughly twice the size of the 75-year shortfall.

Bruce Webb said...

Well yes but there is no particular reason why that roughly double tax increase could not be phased in over twenty, thirty of forty years. For example the Northwest Plan devised by a few people at Angry Bear shows that a 0.2% increase in 2011 added to with another 0.1% increase in 2012 fills the immediate need, while an additional 10 year series of 0.2% increases starting in 2026 fills the medium term need.

The spreadsheet does not extend past the standard 75 year window but extrapolation shows perhaps an additional five year series of comparable increases starting in around 2090 fills the remaining gap and all from incomes projected to be much larger in real terms than those of today.

If we step back a bit we can see that most of the challenges faced by Social Security after mid-century or so will be the product of current generations putting in the investments that will result in those future generations living longer and healthier lives. We can agree to disagree on how much of the future gap was due to over-generosity to the Greatest Generation in the past who only won a war for us and established the basis on which all that future prosperity was built, but really under the circumstances only a generation of over-indulged whiners could possibly complain about some perceived 'intergenerational inequity'.

2054: "Gosh, look at that ROI they got! And all I got is a wall size TV, a countertop supercomputer with full access to every book published or artwork produced since the dawn of time, and a 300mph supertrain. Oh and had to save a little more to compensate for my twenty-five years of life in retirement. Man, life is so unfair."

Boo-hoo Gen-X and Y.

JG said...

"Third, while a 1.3 percent of payroll deficit sounds small"...

I don't understood why "percentage of taxable payroll" is always used to measure the shortfall, when...

(1) Few understand it (as exemplified by the need to explain "it sounds small, but..."),

(2) It omits the very real tax that is going to land on people to finance the SS Trust Fund, and

(3) On current law payroll tax is not the tax that is ever going to go up on people to fund SS.

ISTM easier-to-understand and more accurate to measure the shortfall in terms of promised benefits minus expected payroll tax receipts -- that is, the financing gap to be covered by general revenue income tax increases (or additions to the national debt) needed for Trust Fund operations and after.

Happily, the Trustees give us that number in % of GDP terms, and translating it to income tax terms is easy.

By 2030 the cost of Social Security rises by 1.43 percentage points of GDP. (From a surplus of 0.13% of GDP this year to benefits exceeding payroll tax by 1.3% in 2030). Then cost stabilizes.

In 2007, the last year data wasn't distorted by recession, total income taxes (personal and corporate combined) equaled 11.2% of GDP.

Since 1.43 /11.2 = 0.1277, we can estimate that an across-the-board income tax increase of 12.8%, on both individuals and businesses, or the equivalent, will be needed to pay for SS by 2030 and forever after.

Now, many people lately have dismissed the underfinancing of SS as minor, not a real problem on its own, so why bother?

But for persepctive the Clinton 1993 tax increase was only 0.83% of GDP -- and got through the Democratic-controlled Senate only on vice-president Al Gore's tie breaking vote, and Democratic-controlled House by only 218-216 (a single voter's difference), even though it was 42% smaller than the tax hike needed for SS.

So maybe the financing cost of meeting SS's cash flow needs isn't so trivial after all!

Especially when realizing the tax hike for SS will be coupled with other tax hikes about twice as large (if not larger) to cover Medicare, Medicaid, unfunded federal pensions, etc.

Really, saying "X% percentage of payroll deficit" doesn't convey any of this. Moreover ...

It wrongly counts the SS TF as an asset funding SS's solvency as if it holds funds available for making payments (even though you yourself quote CBO saying it is not so) -- which is exactly what lets so many people deny that there'll be any money problem with SS until after the TF expires decades from now...

It implies that if payroll tax were increased by that amount today SS's financing problems would be over -- when the TF would remain fully unfinanced, and if a SS surplus were re-created it would be re-blown just as in the past, so things would be largely unchanged...

And it has totally unrealistic implications that a payroll tax hike is relevant for SS's long distance, post-TF future. Look -- by 2030 Congress is going to increase income taxes to pay all benefits (in which case a future payroll tax hike is unneeded and irrlevant), or will cut benefits to avoid increasing taxes (same), or a VAT or something will be created with other program changes (same).

So why is SS always "explained" in terms of "% of covered payroll"?

Reality is: the shortfall to be covered = promised benefits minus future receipts from current payroll tax rate, as % of GDP -- however that shortfall is to be covered.

So why don't the good people at SSA explain this in plain English, such as...

"Social Security's financing shortfall is of a size that needs to be financed by a 13% across-the-board income tax increase, or the equivalent, starting soon and fully implemented by 2030."

That's clear and accurate (unless one has a better percentage number).

Then let the actuaries discuss "percentage of covered payroll" among each other if they want to.

Jack said...

Why is there no acknowledgment of the relationship between the Trust Fund, the excess FICA payments which have been contributed by tax payers for more than twenty years in recognition of the eventual need to supplement FICA receipts at some point in order to pay out social security benefits and the use of Trust Fund revenues to supplement the general budget deficit? Is that question too long to deserve to be answered? Yes, general tax revenues will likely have to be increased in order that the general budget pays back to the Trust Fund what it has been utilizing to support both excessive tax cuts in the recent past and profligate spending in general. What groups have benefited most from the use of Trust Fund revenues to support the general budget over the past 20 plus years? Tax cuts for the wealthiest tax payers, wars of adventure with no legitimate purpose, but plenty of corporate benefits. Yes, a revision of purpose is certainly called for, but the Trust Fund revenues didn't go to some secretive purpose. Replenish the Fund with tax revenues from those sectors of our society that have been enjoying the benefit of those revenues to the greatest extent.

Bruce Webb said...

On a constant dollar basis the gap between SS revenue excluding interest and cost in 2030 is projected by SSA to be $1.308 tn - $1.028 trillion or $280 billion. On a current dollar basis the respective figures are $2.309 tn - $1.816 tn or $439 bn.
Taxable payroll will be $13.8 trillion out of an overall GDP of $37.8 trillion. Currently Social Security FICA has incidence around 84% of individual income meaning that total individual income will be somewhere around $16.4 trillion. (for simplicity I am not including corporate income)

$439 billion represents right on 0.25% of $16.4 trillion. Now I suggest that most people would look at:
"Social Security's financing shortfall is of a size that needs to be financed by a 13% across-the-board income tax increase, or the equivalent, starting soon and fully implemented by 2030."
to mean that everyone was faced with an average increase in their rate of 13%, because after all that is what 'across-the-board' implies. Which would overstate the impact by a factor of 50X. Instead your calculation seems to represent a 13% increase in collections, most of which (as we are ceaseless reminded in other contexts) comes from the very top income brackets.

Meaning there is no 'across the board' about it, your 13% number has exactly zero relevance to say families of four earning under the median household income who under current law pay little to nothing in income taxes to start with.

On the other hand 1.3% of payroll is something people can calculate and appreciate in real terms and if enacted would come out of the first dollar.

So I don't see that 13% adding any clarity at all, if anything it obfuscates the discussion by adding a number that really doesn't apply to any particular individual, nor as far as I can see does it clearly apply to any given marginal rate. All it does it suggest that people will have to hand over 13% of their income to finance what is mostly simply a repayment with interest of money borrowed from workers from 1983 to around 2017.

Ceteris paribus people who pay income taxes will pay relatively more once this particular source of cheap borrowing dries up and has to be repaid. Big deal. In 2030 we as a people will have to find about one quarter of a percent of national individual income to fulfil our promises to what will be then 71 million seniors and millions more survivors and disabled people (out of 371 million residents).

Somehow I don't think the country will come to a crashing halt.

Bruce Webb said...

Little math error. That $439 billion is 2.5% and not 0.25% of income. . Still not that much considering that it will be going to feed and shelter around 20% of the U.S. population

JG said...

"Why is there no acknowledgment of the relationship between the Trust Fund, the excess FICA payments which have been contributed by tax payers for more than twenty years in recognition of the eventual need to supplement FICA receipts at some point in order to pay out social security benefits..."

If you are asking why no mention of the intention of the creators of the Trust Fund for it to hold surplus FICA taxes to fund future retiree benefits -- it's because there never was any such intention.

To quote Robert Myers, chief actuary of SS who was executive director of Greenspan Commission that created today's Social Security, in his oral history at
"Q. ... some people rationalize the [trust fund] financing basis by saying that it's a way of partially having the baby boomers pay for their own retirement in advance. You're telling me now this was not the rationale. Nobody made that argument or adopted that rationale?

"Myers: That's correct. The statement you made is widely quoted, it is widely used, but it just isn't true. It didn't happen that way ... The main thing that was talked about was how do we fix up the short-range problem. Are you sure we aren't going to have another crisis in 2 or 4 years?..."

The idea of a "trust fund" never even occurred to anybody at the time -- much less that it would be used to fund anybody's future benefits.

The politicians' big concern was to set the FICA tax rate high enough so they wouldn't suffer the trauma of raising it again on their own watch. The rate they set turned out to be higher than expected compared to benefit costs. So they spent the difference. Sen. Moynihan, who was a member of the Commission, later repeatedly tried to cut the payroll tax rate to the cost of benefits, but Congress preferred to keep spending the money. That's the story.

When the current SS tax scheme was set up nobody had any intention of funding anybody's future benefits with any trust fund. Since then, it's just been the politics of politicians spending money they were happy to find unexpectedly coming in.

"Replenish the Fund with tax revenues from those sectors of our society that have been enjoying the benefit of those revenues ..."

Which is *not us*, right?! ;-)

I'm sure a whole lot of people are going to be saying that when the tax bill arrives.

But it's not so easy to drop it on "those sectors". The biggest gainers of the deal were the retirees of the 1980s who got their benefits protected by dropping big benefit cuts, tax increases and net losses from SS on their youngers, and the income tax savings and spending benefit of the FICA surplus too -- but they're gone now.

Then all the politicians who spent all the extra FICA money -- over Moynihan's objection -- to buy votes certainly profited from it, but how much are you going to get back from their pockets? ... Then you have all the citizenry who for all the decades since 1983 have benefitted from having the surplus spent on their behalf, but they will be retirees in the 2010s -- do you want to drop the cost on retirees? AARP won't like that.

Yes, we'll all agree about "them" being the ones who should pay, but a lot of "them" are gone from the scene, and among them that are left many may not agree, like the AARPers, and some of them may even think "them" isn't them but you and me.

In any event, both the arithmetic and the politics are obvious: It's going to be the young workers who go no benefit from all that past politicking who'll get clobbered with the cost of it.

JG said...


Regarding your first four paragraphs, you can simplify greatly and avoid such mistakes by just going along with the fact that the financing of SS by the federal govt is simplest thing possible: cash in = cash out, real time.

Thus if SS benefits rise by $X then federal revenue used to finance them must go up by $X at the same time. QED. That's all. That means taxes must go up by $X or debt must go up by $X -- which means taxes go up by $X on an deferred-plus-interest basis -- or some combination. Which means if you know $X all else follows, very simply.

The Trustees tell us $X for 2030 relative to today is 1.43% of GDP, and stabilizes then there. (CBO tell us the % is a little higher, but I'll stick with the Trustees). Given a current income tax level of 11.2% of GDP, a 1.43% of GDP income tax increase amounts to 13% rounded.

As the Trustees say cost basically stabilizes from there, it follows that a 13% increase in income tax from today's level by 2030, maintained permanently from then, solves SS's funding problem permanently. Obviously, the same result comes from an equivalent increase in any other tax that provides the same 1.43% of GDP increase in revenue.

That's it. What could be simpler?

Of course, if you want to make things a whole lot more complicated than this for your own edification, enjoy yourself.

"... $439 billion represents right on 0.25% of $16.4 trillion ... ...mean that everyone was faced with an average increase in their rate of 13% ... because after all that is what 'across-the-board' implies. Which would overstate the impact by a factor of 50X..."


"Instead your calculation seems to represent a 13% increase in collections..."

You finally got that from the "Since 1.43 /11.2 = 0.1277" ?

"On the other hand 1.3% of payroll is something people can calculate and appreciate in real terms and if enacted would come out of the first dollar."

Except that ....

1) Federal bonds, such as those in the SS Trust Fund that will be redeemed to pay the 1.43% of GDP in 2030, are financed with income taxes, not payroll taxes.

Which makes disussing their cost in income tax terms much the more realistic. Have you heard of any proposal to change the law and service the Trust Fund bonds with payroll taxes instead?

2) 1.3% of covered payroll is a lot less than 1.43% of GDP, only about 1/3rd as much. So it doesn't begin to accurately reflect the tax cost coming to meet the "cash in" needs of financing the SS trust fund. (Because the "payroll cost" calculation omits the cost of TF operations). So it is simply the wrong number to use when computing the increase in the future tax cost of financing SS.

I'll skip puzzling over the philosphical question of whether a tax increase of a given % can be described as "across the board" if it applies to everyone who pays the tax but there are people in the world who don't, as whatever the answer is doesn't change the 1.43% of GDP cost by a penny.

Of course, neither does the fact that many lower-income individuals today don't pay income tax, about 48% of all individuals in fact.

But I will note that given the well known damage that high and rising marginal rates (which result from taxing the costs of the many to the incomes of the few) cause to the economy via the deadweight cost of taxes -- which rises not with the tax rate by by the square of the increase in the tax rate -- there is ample and growing talk in Washington these days of moving to keep top marginal rates down by moving to financing social insurance programs with a VAT.

If that's enacted, the many will be back to covering their own costs.

But who knows what the future holds?