Monday, March 17, 2008

Are the Trustees too pessimistic?

In the comments, Bruce Webb raises a common argument, that Social Security’s Trustees have generally been too pessimistic about the program's finances. I wrote on this issue a long time ago, and posted my little solvency widget in part to let people play out their own scenarios. But this issue is an evergreen and deserves some more attention.

Bruce argues that the Trustees have been overly pessimistic based on their short-term forecasting record. Bruce compares GDP growth as projected in a Trustees Report with the realized figure one year later. He argues that since 1997 GDP growth has generally been understated and that the Low Cost projections have been more accurate. From here, he concludes that the Low Cost 75-year projections may be more accurate.

A couple points: First, it is hard to draw firm conclusions about the Trustees’ 75-year forecast from the accuracy of their single-year forecasts, since these are very different animals. The SSA Office the Chief Actuary is geared more toward long-term forecasting, while the Congressional Budget Office and private sector forecasters concentrate more on the short-term. Moreover, in many ways long-term forecasts are easier than short-term forecasts, since the long term relies more on underlying fundamentals and less on year-to-year fluctuations. For instance, I cannot say with accuracy whether the weather one week from now will be cooler or hotter than today’s, but I can say with great accuracy that the weather six months from will be hotter simply because of the fundamentals.

Second, Bruce uses one method for assessing the Trustees’ accuracy, but others – perhaps better ones – are available. Here I turn to Chuck Blahous, the number two at the White House National Economic Council and the Bush administration’s chief policy man on Social Security. In a paper presented last year at the American Enterprise Institute, Chuck applied a rigorous test to the Trustees accuracy.

First, he devised two benchmarks of Social Security’s financing health: a), the annual cash balance (surplus or deficit) as a percentage of payroll; and b), the trust fund ratio – the trust fund balance as a percentage of annual costs. At any given time, these are good measures of how well Social Security is funded, clearly more so than GDP growth alone. In 2005, the last year for which data was available when the paper was being written, the annual Social Security cash surplus was 1.55% of taxable wages, and the Trust Fund ratio was 318%.

Next, Chuck looked back over the previous Trustees Reports ranging back to 1983 and analyzed how accurately they projected these outcomes for 2005. Specifically, he asked which set of projections – Low Cost, High Cost, or Intermediate – was the most accurate in projecting the cash balance and trust fund ratio for 2005.

Figure 1 details the Trustees projections of the annual cash flows. Of the 22 Reports, the Low Cost was most accurate in only 4, the Intermediate Cost in 7, and the High Cost in between 8 and 11 Reports (this is due to changes in how the Reports were structured).

Figure 2 repeats the process for the Trust Fund ratio. Here the Intermediate Cost projections have been most accurate, being the best projections in 12 of 22 Reports. The Low Cost projections were most accurate in 6 years, and the High Cost in 4 years.

Clearly other tests could be applied. But one test – summing the squared errors of the different methods, then discounting to control for the difficulty of prediction as time increases – shows the summed errors for the Intermediate Cost projections were only one-fourth the size of those of the Low or High Cost scenarios.

Chuck summarizes the results in this way:

A reasonable summary of the 1983-2004 projections’ accuracy is that the Trustees’ Intermediate projections most accurately predicted the cumulative value of the Social Security surpluses over the totality of the 1983-2004 period. With respect to predicting the balance of annual system operations in 2005, the High-Cost projections were the most accurate in the highest number of reports, although the Intermediate projections exhibited the lowest total error. On balance, by both standards taken together, the Intermediate projections have been the most accurate, and the Low-Cost projections have been the least. Though this does not necessarily mean that the superiority to date of the Intermediate projections is significant for the future, the data clearly contradict the widespread misconception that the Low-Cost projections have in the past been best.

Anyone considering the argument that the Trustees are overly pessimistic must account for the arguments and evidence Chuck Blahous presents in this paper.

In addition to the paper itself, PowerPoints from Chuck and discussants Steve Goss (SSA) and John Sabelhaus (CBO) and video of the entire event are available here at the AEI website.


Bruce Webb said...

More later.

Economic performance from the time of the Greenspan Commission/Reagan-Monihan compromise of 1983 to 1993 tracked actual economic performance to an astonishing degree with the desired result: Social Security in 1993 finally climbed back to a Trust Fund Ratio of 100 in that year. Hurrah! And in all honesty credit was due all around, any claims my side has against Reagan/Bush on this being unfair in numeric context, a carefully crafted compromise delivered what it was supposed to do: short term actuarial balance. Trying to average those spot on results with those of the understated projections of the last eleven years is to give too much weight to the trailing tail compared to actual performance over the last ten years.

Andrew G. Biggs said...

If you're looking to find (or disprove) a general trend, you want as many data points as possible, since with a small sample there's always the possibility that the general conclusion -- 'the Trustees are overly pessimistic' -- is based on chance.

As it happens, restricting your data only to the past 10 years may be biased because the mid-90s were a point when, as the Blahous paper points out, the Trustees genuinely were too pessimistic about the program's finances. They took a while to accept higher rates of productivity and wage growth. But it's hard to extrapolate from those years to the coming 75 years; there isn't any strong consensus among economists that productivity will be higher than the Trustees currently project. said...


So, how is it that in six of the last ten years, the economy has done better than the GDP forecast of the low cost projection, but somehow the chart you show has the low cost projection always doing better than the actual cash balances?

I note that the intermediate cost projection has a very pessimistic projection regarding immigration and demography more generally. Is immigration really going to collapse to near zero in the near future?

Also, I point out that even if the threatened "bankruptcy" occurs as forecast in the intermediate projection, will not the payments recipients get after that still exceed those of current recipients in real per capita terms by more than 120%.

Barkley Rosser

Andrew G. Biggs said...

Barkley asked:

"So, how is it that in six of the last ten years, the economy has done better than the GDP forecast of the low cost projection, but somehow the chart you show has the low cost projection always doing better than the actual cash balances?"

If I'm understanding the question, it's asking how recent low cost estimates overstated the 2005 cash flows even though GDP growth exceeded the low cost projections.

The answer (I believe) is that GDP growth didn't surpass the Low Cost projected levels. For instance, in the 2002 Report the Low Cost projected cash flow for 2005 was a surplus of 2.66 percent of payroll. Projected real GDP growth for 2002, 2003 and 2004 was 1.6%, 4.6% and 4.6% respectively. Realized real GDP growth for those years was 1.6%, 2.5% and 3.9%, respectively. So the economy didn't live up to the Low Cost projections. (This is only for one example; other factors may be at work here. For instance, disability cases have been rising quickly; I'm sure they rose faster than the Low Cost projections predicted.)

Bruce Webb said...

Which of course presents a challenge for privatizers.

I made up a little ditty back in 1997 or so and I think it is as valid today as it was then:

If Privatization is Possible, it Won't be Necessary.
If Privatization is Necessary, it Won't be Possible.

And this was the kernel of Dean Baker's 2004 No Economist Left Behind challenge. How do you get 6.5% real returns using a productivity assumption of 1.6% (then) or even with the current ultimate 1.7% productivity and 2.0% Real GDP starting in 2013? Certainly Intermediate Cost is a possible outcome, perhaps as some argue a median projection, but then how do workers manage to fund a PRA with a Real Wage limping along at 1.1%?

That is some people tried to pass the NELB challenge by assuming a portfolio that was capturing higher productivity overseas, which may work for the traditional investor class. It is just hard to see how it works for the new worker/investor whose wages and hence ability to invest are based on domestic productivity.

But privatizers can't bump their assumptions up too far to improve returns. 2.0% productivity and 1.6% real wage gets you right to Low Cost and fully funded Trust Fund (other things being equal)

And while their may not be a 'strong' consensus there seem to be some whack jobs at some place called ' B L S' that have a different take:

"Productivity assumptions. As mentioned earlier, one striking aspect of recent U.S. economic history has been vigorous growth in labor productivity. High productivity growth allows for a mix of higher wages and profits and lower consumer prices. Together, these permit a higher standard of living and quality of life. Since 1995, the U.S. economy has had the fastest productivity gains in 30 years.
However, beginning in the second quarter of 2005, productivity growth slowed.
It is uncertain whether the productivity slowdown is a lull or is an end to the productivity boom of the late 1990s and early 2000s, but what is clear enough is that productivity growth is one of the critical influences on the economy’s long-term growth potential and increases in living stand-
ards. With steady GDP growth projected over the next 10 years, the BLS has assumed that productivity will grow at a pace of about 2.2 percent annually between 2006 and 2016.

And their figure for Real GDP over that period? 2.8% (1.9% per capita)
And maybe someone could have a stiff word with the boys at OMB, in Table 12-2 of the Analytical Pespectives on the Presidents 2007 Budget they were projecting real GDP above 3.1 percent for the entire 2006-2011 period (p. 182) as were the CBO and the Blue Chip Consensus. Nor does the OMB suggest sharp slowdowns in productivity (p. 162)

This has been a pattern for a long time, when you chase down the economic numbers under the budget and tax policies they always end up a lot more optimistic than the ones the Trustees use. Presumedly Treasury Secretary Trustee has access to the budget and Labor Secretary Trustee access to numbers coming out of her shop. Do they actually not examine the Reports before they sign them.

It is kind of hard not to be paranoid when different parts of the same executive branch seem to be able to use different sets of assumptions for different purposes. Plug either the BLS or OMB assumptions into the Social Security models and you have quite a different outcome.

Bruce Webb said...

"Projected real GDP growth for 2002, 2003 and 2004 was 1.6%, 4.6% and 4.6% respectively. Realized real GDP growth for those years was 1.6%, 2.5% and 3.9%, respectively."

Well I would like a link to the source of that first set of numbers. I looked at the 2001, 2002, 2005, 2007 Social Security Reports and found nothing like them.
Table V.B2 from the 2002 Report projects .7% 3.8% 3.5% Real GDP for 2002 to 2004
The same table from the 2005 Report shows actual 1.9% 3.0% 4.4%
Now what do you know the 2007 Report does show 1.6% 2.5% 3.9% but that is after significant revisions introduced with the Q2 2006 productivity numbers. Unless the authors of the 2002 and 2005 Reports knew in advance the results of the 2005 Census of Manufacturers what you are looking at there is apples and oranges. None of that really explains the year over year underestimations typical of all the Reports from 1997 to 2004. said...

I just tried to post the actual numbers, and they do not seem to be here.

Biggs: Bruce's question is relevant. I question the numbers in Blahous's report. Are they lies?

The numbers you have posted on GDP are off. Are you lying?

Barkley Rosser said...


I apologize for the harshness of my language, but the hard fact is that there have been a lot of wild misrepresentations by people declaring social security in crisis.

Barkley Rosser

Arne said...

As I have looked at the SS reports I have concluded that the actual ratio of workers to beneficiaries will be the driving factor. There will be a surge in retirees that will begin using the interest from the TF in about 2017. The IC says we start drawing on the principal about 10 years later (depending on which years report you use), which is also about when the ratio gets down to 2.2 to 1. In the LC model, the ratio never gets below 2.3, so the principal is never touched.

So, how accurate is the SSR in predicting the number of workers from 20-64? One data point: the 1995 LC projection underestimated it all the way from 1995 to 2004. The 1995 IC was off by about 5 percent with th gap widening.

On top of that, the report tables give people 20-64 for the entire 75 year period, even though the retirement age is rising.

In 10 years we will have a better handle on the real impact of Boomer retirements. Then we should reassess tax rates, caps and retirement ages. Unless we stop being a nation of couch potatoes, we won't be outliving the current assumptions.

Andrew G. Biggs said...

To answer Bruce's initial question on the GDP numbers, the Low Cost projections for real GDP growth from the 2002 TR are here:

I compared these to historical figures on GDP growth from the 2007 TR here: Unless I've made a mistake, those numbers are correct.

Andrew G. Biggs said...

Arne's question deals with the accuracy of the demographic projections. He's right that the ratio of workers to beneficiaries is the driving factor in the long run. I believe those estimates in the TR have been pretty accurate. I'd have to double check, but I believe that if you look at the 1983 TR's projections of the worker-retiree ratio for today they'll be pretty close. Where the older TRs went wrong was on a) overestimating future economic growth; b) underestimating the future disability rolls (though understandable, since the eligibility rules were slackened in the mid-80s); and c) new modeling methods from the actuaries, in particular dealing with the age structure of immigrants. Really only the economic growth projections were the 'fault' of the Trustees, so overall they didn't do a bad job of looking 25 years in to the future.

But I disagree with Arne's conclusion that we should wait and see. Again, the demographic side is easy to predict. And every year we wait makes the burden of reform larger on future cohorts. A better approach is to adopt a flexible reform that raises taxes or cuts benefits as needed and can adapt to changing conditions. This gets around the objections from the Bruce Webb/Dean Baker crowd, since then you won't be in danger of overbalancing if the problem turns out to be smaller than projected. I'll write more on this at another point.

Arne said...

I can actually get on board with automatic changes to the retirement age as life expectancies increase, but I would note that the 2 year increase we have already means that a person retiring at age 67 in 2027 will have fewer years in retirement (per the IC estimate) than a person who retired in 2002 at age 65. (15.6 versus 16.2).

If you want to give people 20 years to plan, it is time to calculate the retirment age for the 1961 cohort. If you want to give them the same number of years of work per number of years in retirement (2.2 as a target for men and women combined), then the correct change is: none at all for at least 10 years.

Anonymous said...

the ratio of retirees to workers is the same as the ratio of retired years to working years (other things being equal which they almost are).

and it's largely irrelevant to the "problem,"

because there is no problem.

if i am going to live 20 years in retirement after a working career of 40 years and i want to have 40% of my lifetime average real monthly wage to live on, i would need to save 20% of my wage while working.

which works out very nice, because the boss pays half of that, and the government guarantees it all against inflation by the miracle of pay as you go financing with wage indexing.

you can wear a blindfold and convince yourself that the elephant is really a smelly cave, but if you take your blinders off, you will see that he is really a rather elegant animal well able to care for himself if he can survive the zoo's doctors.

begin with the simple fact that the worker will be setting aside from 6.2% to maybe as much as 10% of his wages in order to guarantee... as good a guarantee as humanly possible... a basic minimum to retire on. this will not change.
it really can't change. and all this fancy dialogue is meant only to obscure the reality from the people who both pay for Social Secutity and need it. said...

It is not true that the demography is so clear. The serious loose end is immigration, and the intermediate cost projection is very pessimistic about that.

I would also note that the awful ratios we will supposedly experience in 2030 are about what we see in many western European countries right now, some of which, such as Germany, have substantially higher levels of government pension payments than does the US. Nevertheless, these countries are all paying these pensions reasonably well, with Germany currently running current account surpluses and having a rising currency compared to that of the US.

Also, Andrew, you have not answered my point that even if the threatened "bankruptcy" were to occur and the intermediate projection actually came to pass, with no adjustments made, the payments recipients would get would exceed 120% in real terms of what current recipients get. This is a crisis worthy of doing anything about, much less making it into the great cause celebre to endlessly beat politicians and others over the head with? (I also add that almost nobody under 30, and most over 30, do not know that fact I just stated.)

Barkley Rosser
Professor of Economics and Kirby L. Cramer, Jr. Professor of Business Administration
Editor, Journal of Economic Behavior and Organization
James Madison University

Andrew G. Biggs said...


Sorry for missing your question on immigration. This is where the sensitivity analysis from the Trustees Report can be helpful. Also helpful is the solvency 'gadget' a couple weeks ago, which makes these kinds of comparison easy (and hopefully more fun).

The TR sensitivity analysis for immigration ( says that "Each additional 100,000 net immigrants increases the long-range actuarial balance by about 0.07 percent of taxable payroll."

The baseline number is 900,000 immigrants annually and the current deficit is 1.95 percent of payroll. Divide 1.97/.07 and you get around 28, which multiplied by 100,000 equals 2.8 million on top of the current 0.9 million. So (roughly) 3.7 million immigrants per year, every year, will get you 75-year solvency. I believe that last year net immigration was a bit under 1.1 million, so you'd need to more than triple it. I know that's not the only solution you'd propose, but this does give you the scale of things.

Regarding the real level of benefits, it's strange that people say that no one on the right mentions this. It's of course true that the post-insolvency benefit is higher than the average benefit today. Folks on the right have argued that we can afford to give people benefits at least as high as today's, but we can't afford to give them real benefits that rise with wage growth each year. Policies like price indexing simply smooth the evolution of benefit changes so we don't get the large cut in benefits post-2040.

Strangely, while people toward the left often cite level of post-2041 real benefits to claim there's no 'crisis', they get very upset with any attempt to lower scheduled benefits which, in 2041, would be around 30 percent higher than today's levels.

If it's no crisis to have future benefits as high as today's, then I'm sure folks on the right can quickly agree with you on a reform plan. Alternately, if we must have full scheduled benefits forever, then we do face a big funding problem since scheduled taxes won't be nearly enough to pay for them.

Anonymous said...

post insolvency

There is no insolvency. The trust fund, in its present form, was created (they say) to bridge the baby boom funding problem. by 2040 when most of the boomers are dead, the trust fund will be retired, having done its job.

and social security, fully solvent on a pay as you go basis can continue forever.


don't be so eager to add 2 years to the sentence of other workers dying to retire. it's their own money. they paid it in. they should get to retire on it. and if you get all confused by the pay as you go feature, you probably don't know much about what a bank does with your deposit either.

you all tend to talk about this as though money were more real than people. and you never address the size of the money "problem" or who pays it or why they pay it.

oh, gosh, you say... look at that trillions of dollars of debt, lets raise the retirement age. how many of you stop to say... hmmm that trillions of dollars amounts to twenty bucks a week paid by the very people who want to retire when they get to be 62 and are tired out??? can i say it again.. it's their own damn money. not the governments. not yours.

and, yes, the younger generation will be glad to pay it, if you let them understand how it works and their turn will come.

Arne said...

anonymous 5:20

You misread my post or are unaware that the retirement age has already been increased. In was born in 1960, so my retirement age is 67. If people live longer, SS needs to have more money to pay the benefits longer, but the added 2 years is already enough to pay for the increased life spans gained from 1983 to 2027.

Anonymous said...


i think i read you correctly. i am aware that retirement ages have been increased 2 years already.

but you said you would be on board with "automatic increases" so i just picked the 2 years as a place to start talking.

the point i was making was that adding any years to someone elses retirement age is cruel and dishonest. the people pay for their own retirement.

if i can keep the same benefits and the same retirement age, even though my life expectancy is longer, and i might be healthier, just by paying an extra twenty dollars a week... out of a pay that is 300 dollars a week more than it is today... that would seem to me to be the best possible choice for me.

and for lots of others. now, if you have a great job as, say, a professor of economics, you may not want to retire. no one is forcing you.

or maybe some aging assistant professor will poison your tea if he gets tired of waiting.

Anonymous said...

i'd be glad to give up being anonymous

but i can't figure out the sign in routine.


Arne said...


You need to sign up first. Right now we work about 2.5 years for each year in retirement. If we start living 3.5 years longer we need to have 1.0 years more of income to pay for the same benefits. Alternately, we need to increase taxes. Right now we also have the option of earlier retirement on reduced benefits. I would stay away from increasing taxes, and I think that implies supporting raising the age for full retirement benefits - which I think would be best done on a schedule tied to actuarial tables (better then getting legaslative agreement on how much each decade). Reduced benefits at 'early' retirement would also be tied to actuarial tables. (I would not raise the age from 62). I recognize that if Bruce is right, that would actually mean reducing taxes in the future.

Anonymous said...


would you care to explain why you would raise the retirement age rather than raise taxes?

i can't imagine any real worker choosing another year of servitude when he can buy his freedom for an extra 20 dollars per week out of a pay increase of 300 dollars a week (total average income of 1000 dollars a week as projected by the Trustees).

and of course if he's feeling good about his job, no reason he can't keep working.


Bruce Webb said...

Well with apologies to Mr. Biggs I think the syntax tripped me up somewhat, though we can point out that the 2002 LC numbers are really 1.6%, 4.6%, and 4.3% (and not 4.6%)

But in any event this is to miss the point. The fundamental question is not whether or not Low Cost is too pessimistic in recent Reports but whether Intermediate Cost is. I don't need Low Cost numbers to come in each and every year, I only need outcomes consistently better than Intermediate Cost. And we are getting that or were until the Bush economy took productivity over the cliff in Q4 2005 (by coincidence (?) eighteen months after the growth empowering 2003 tax cut). Even now TF balances came in ahead of projections despite the 2.2% GDP figure.
As to the immigration question. No one is proposing that the total solution is immigration, only that people who present the problem using covered worker ratio figures that include capped numbers for immigration are speaking out of both sides of their mouths. To the extent that the problem is a shortage of workers going forward we can solve that as we have for 300 years-by admitting immigrants.

Bruce Webb said...

"And every year we wait makes the burden of reform larger on future cohorts."

Hmm. No. In the face of inaction the payroll gap has stayed even or shrank in 9 of the last 11 years. There is a word for a problem that left unaddressed tends to shrink in size and move out in time at a rate of more than a year per year. That word is not 'crisis'.

The 'we can't afford to wait' crowd needs to bring some numbers to the table because nothing in the recent historical record supports their case. said...


Thank you for responding. I agree with Bruce Webb that the point was not that immigration alone will solve everything. The point is that the baseline is about half of most estimates of what immigration is, which means that the baseline is too pessimistic. The current reality is below what "alone" would "save" social security, but it is about halfway there from the baseline.

Your point is well taken that one should be prepared to accept reductions in future benefit increases if one is going to say a cut to over 120% in real terms of what current recipients is not a crisis. However, the problem is the public presentation and perception of all this. So, when Bush spoke on this he spoke of a cut to around 70%, not telling anybody that this would be a cut that would leave people so well off. Of course, such a drastic one time cut would never occur. As it is, more young people think they will see a flying saucer than that they will get any benefits at all. This sort of delusion is fed by these kind of distorted speeches. I and three colleagues took a survey in our classes and found that out of 250 economics majors, not a single one was aware of the facts. All held far too pessimistic views of the situation. As it is, there is indeed plenty of time to make adjustments, if indeed it begins to appear that the intermediate projection will come true and a deficit will appear in or about 2017.

In the meantime, although I cannot find him having posted on it, Bruce Webb has sent me numbers for 2006. In that year, the overall balances were better than the low cost projection, not to mention even better still than the intermediate cost one.

Actual balance: $2.0492 trillion
low cost: $2.0380 trillion
intermediate cost:$2.0353 trillion.

Barkley Rosser