Thursday, September 10, 2009

New SSA paper: “Unfunded Obligation and Transition Cost for the OASDI Program”

Social Security's Office of the Chief Actuary has released an update on their paper "Unfunded Obligation and Transition Cost for the OASDI Program." This paper expands on some of the measures included in the annual Trustees Report while providing additional information that's not in the Report.


JG said...

Excellent information. Thanks.

lawipau said...

I think the "Maximum Transition Cost" is a particularly interesting number. If I understand the calculation correctly, it is approximately the amount that we would need in individual accounts today if the current SS benefit schedule were fully prefunded.

It's "approximate" because we need to add the current trust fund and the PV of future FIT on SS benefits to get the full amount. We could also debate the discount rate.

Regardless, the fully funded requirement is a very big number - something on the order of $20 trillion. For comparison, the market capitalization of the Wilshire 5000 stocks (aka the "total US stock market") varied between $18 trillion and $12 trillion during 2008.

coberly said...

apologies for tone and style, but this IS a science fiction blog, right?

"excellent information"

if you have a stomach for filling in the corners of Thanksgiving dinner with empty calories.

reflect upon "open group," "obligation" "infinite horizon" and "uncertainty," and if you still think this kind of "analysis" is helpful, or has anything to do with Social Security at all, your taste for fantasy will be amply rewarded right here now and in the future.

but if you can understand that Social Security is a way of paying for "our daily bread" by the magic of pay as you go financing with wage indexing, and that if future costs rise they can be met by future taxes that look for any future any of us can see with a straight face like they will amount to about an 80 cent per week increase in some but not all years while incomes are going up ten dollars per week in all years (on average), then, well, then you can relax and say "sufficient unto the day..."

no. i didn't think you could.

Andrew G. Biggs said...

lawipau: That's basically right: the maximum transition cost represents how much we'd need to have on hand today if we wished to stop the current program and shift to a pre-funded one. Obviously that's a very big number, which is one (probably the biggest) impediment to passing a carve-out personal accounts plan. People speak approvingly of 'prefunding' the system, but that demands we actually pony up the funds. No one really wants to do it.

coberly said...



my point exactly.

coberly said...


i used to work with the sort of engineers who would use their factory supplied calculation sheets and their wonderful calculators to arrive at an answer that was an order of magnitude wrong. they never had a clue because nothing in their education had taught them to suspect that an answer so correctly arrived at could be wrong.

you are doing the same thing. you never suspect that when you get an answer that says the staggering debt will destroy the economy, the sky will fall, and the universe will come crashing in on us.. that there is something wrong with the calculation.

WilliamLarsen said...

It is truly amazing how things have changed in the past 30 years. In 1984 using an early model computer, I revised my Fortran SS-OASI computer model program using the recently passed 1983 SS reform. It was in 1984 that I calculated that 2037 ~November that SS-OASI would be unable to pay full benefits.

Many people laughed at me. Since 1984 my projected date has not changed much, now it is early 2038, but then I have not run the program in the past year and it could go back to late 2037.

What has amazed me all these years is how people like to throw out all these modeling techniques such as Monte Carlo, but in the end they are searching for the normal outcome, the average of all inputs, nominal values. Sure you can use abnormally high rates of return, low rates of inflation, excessive wage growth and more to achieve a specific outcome. This is exactly what SSA did in 1983 when it projected well past 2064 and later as early as 2029.

In 1998 I wrote a paper "Cold Turkey" I stated then the following;

"The calculated unfunded balance for COLD TURKEY is $10,712,445,155,000 or $10.712 Trillion. This number is arrived at by subtracting the projected OASI fund balance at end of 1999 of $740,298,964,000 or $740.3 billion from the COLD TURKEY BALANCE IN 1999 of $11,452,744,119,000 or $11.45 Trillion. "

That was 11 years ago. The amount has doubled. In fact it is worst because the Treasury rate earned by the Trust fund has dropped from 5.5% to less than 3%.

Madoff was able to keep his ponzi scheme going for twenty years and was shut down. When will we have the fortitude to admit a huge mistake and shut Social Security down?

SS-OASI is very simple to model. The benefit is defined by an exact formula. The formula is comprised of wage growth. Benefits are not determined by taxes paid, but by wages earned subjected to SS. COLA is paid based on inflation, not taxes paid or wages earned.
As for the potential number of workers, that is pretty simple. Birth rate of 2.1 per woman for the past 30 years, we know the number births for all past years; this provides potential workers and retirees. Best case is full employment. Worst case is high unemployment. I always did best case analysis know full well that if this did not produce a viable program result, then nothing would.

There is no manner of changes to Social Security that can produce full benefits to all! You either pay higher payroll taxes for the same promised benefit; wait longer to receive benefits resulting in paying longer and more, for the same promised benefit: or pay the same for reduced benefits.

Those born after 1985 can expect at most to receive 29 cents back for each dollar of payroll tax and interest at the US Treasury rate in benefits. The last birth year to pay less in combined payroll taxes and earned Interest at the US Treasury was 1938.

The legacy tax being thrown around to pay for the extravagant benefits to those born prior to 1938 is a way to political spin to accept that Social Security is failed program. There was no generation promise ever made. There were no guarantees.

Anyone interested in writing a computer model of SS-OASI, feel free to contact me. I will give you insight in how to do it. Hopefully I made an error in the past 40 years and it is not as bad off as I think it is.

Andrew G. Biggs said...

Tell all that to Bruce Webb!

coberly said...


your analysis contains a number of errors we are both too lazy to track down.

here is a very simple formula to put the problem within some parameters. note... it is simplified. the complications will not materially change the result.

Worker works for 40 years paying 12% of his wages into social security. equals same worker living 12 years collecting 40% of his (inflation adjusted average) wage.

Worker works for 40 years paying 20% of his wages into Social Security. equals same worker living 20 years collecting 40% of his wages.

add in some factors for growth of wages (if any) and payouts for death and disability, and growth or decline in population and you will change the results, but not by much.

the sky is not going to fall.

and your 29 cents in benefits for each dollar of taxes is off by at least a factor of three. and the difference ... whether a real gain of 2.5% as Biggs calculates.. or a "present value" loss of 0.5% as Biggs calculates (I think. he will correct me.) matters to you is a function of your estimate of the risks, and of course your ability to tolerate large losses.

the general experience of the country over the past 70 years has been that most people and the country as a whole do better with Social Security than they do with "the markets."

coberly said...


ah, i read too fast and did not notice that Larsen was including "earned interest" in his calculation. it is not clear he included "inflation" in his calculation, but to be honest it's not the sort of calculation i am interested in spending too much time with.

As he notes (i think) he could have made this calculation in 1936 and saved us all the bother of Social Security and all those old people who had enough money to live indoors following their retirement.

I thought your reply to lawipau showed you understood this. well, no accounting for what people understand.

coberly said...

further note to larson

i did some calculations of my own a while back. best i remember is that the "return on investment" to social security beneficiary born 1943 or so would range from about 10% to about 2% (nominal) (but note that if inflation is higher, the roi on social security goes up automatically. can't do that with your "safe" investments) depending on lifetime income, married or single, self employed or not... leaving out any considerations of disability or death etc. and assuming average life expectancy.

the sheer number of variables makes such calculations of only limited usefulness.

you could, and Andrew has, fudge up some kind of "average person" and find an average roi, but what's the point. Social Security was not designed as an investment vehicle for the average person, who would no doubt do slightly better than social security... as an investment. but Social Security was designed as an insurance policy in case you don't quite do that well. for those who fall into that category, the investment is priceless.

Bruce Webb said...

Bruce Webb read it and is still scratching his head.

The date of TF depletion has moved in and out as SSA and CBO have adjusted their methodologies in relation to economic changes over the decades since 1984. That some guy wrote a Fortran program that at 25 year intervals managed to give the same result as the SSA Intermediate Cost alternative impresses me no more than a blind pig finding an acorn. Or in this case two acorns.

The impact of Boomer Demographics was baked into the pie by 1964. Only so much variation from the mean of any calculation was ever possible.