Saturday, May 24, 2008

Response to questions on Angry Bear

Over at Angry Bear, Bruce Webb posts a number of questions on Social Security financing and reform put together by Arne. These raise a number of important and interesting issues which deserve treatment at some length, so rather than post in the comments I thought I'd do a fresh post here.


  1. What should we do if the SSA intermediate forecast is right? When and how would we change?

    The intermediate forecast is the Social Security trustees and actuaries "best guess" of the program's finances going out into the future. CBO also puts out a forecast based on its own best guess, which is slightly more optimistic than SSA's but qualitatively similar.

    If we assume the intermediate forecast is correct, what should we do? Actually, the easier question to answer is 'when should we do it?', the answer to that being that we should start immediately. Why? Because any solution will be both economically more efficient and generationally fairer if we spread it across as many cohorts as possible. The deadweight losses associated with tax increases are more than proportional to the tax rate, so it's better to have a small tax increase over a long time than a large tax increase focused on a smaller number of cohorts. Likewise, it is easier to adjust to small benefit reductions than to large ones, so starting today means everyone make adjust a bit.

    How should we fix Social Security? This demands a long answer, but to make it short I would: a) focus on reducing promised benefits rather than increasing taxes. Why? Principally because we're going to need all the taxes we have, plus some, to fix Medicare; it's relatively easy for people to make up for reduced Social Security benefits by saving more, while it's tougher on the health care side. b) I would introduce some measure of prefunding; this is designed to smooth cost burdens over generations, such that today's generation pays more so future ones can pay a bit less. I'm fine on this being a 'carve-out' financed from the existing payroll tax so long as the associated transition costs are 'paid for', meaning not borrowed. Otherwise, an add-on funded with new contributions is preferable. c) I'd change the tax/benefit structures to simplify the program to make it more understandable; equalize treatment between different household types (why should single earner couples get a better deal than dual-earners?); and encourage delayed retirement (say, by increasing the early retirement age, reducing the payroll tax at older ages to encourage work, and increasing the normal retirement age). This is a rough sketch, but more or less lays things out.

  2. Which is really more important, productivity or demographics?

    Demographics are almost certainly more important than productivity (a view I believe most experts on either side would share). Because retirement benefits are indexed to wages, higher productivity/wages translates to higher benefits. Because benefits for current retirees are indexed to inflation, not wages, increased wage growth will reduce costs relative to the wage base for a while. But once the higher wages start to filter through to new benefits, costs stop declining.

    Fertility and mortality (i.e., life expectancies) play a more direct and longer-lasting role. Increased life expectancies have only a small positive effect through taxes (a few working age people live longer and pay in more) but a large negative effect through benefit costs (if people live longer, we must pay them for a greater number of years). Fertility has a large effect by increasing the ratio of workers to retirees, which directly reduces costs. The effects of fertility may be exaggerated in the 75-year financing measure, since it counts all the taxes these new workers pay but not all the benefits that will be owed to them, but it's still a major factor.

    Anyone interested in the relative importance of the different variables should check out the sensitivity analysis in the Trustees Report, which shows the effects of different values for each.

  3. Given that the TF lasts at least 20 more years, is moving (some of) it from US Treasuries to equities a good idea? How does that change with LC instead of IC?

    While I tend to favor private investment (through personal accounts) for purposes of pre-funding – i.e., putting aside more money today to reduce costs in the future – I don't believe switching from bonds to stocks is of major importance. Any investment has a combination of risk and return, and at market equilibrium the total 'value' provided by each investment is around the same. That is, the safety of bonds compensates for their lower returns, while the high returns on stocks compensate for their higher risk. There's reason to believe that stock returns still tend to be higher than is necessary to compensate for their level of risk, meaning there's something of a 'free lunch' to holding them. For that reason, some stock holdings may make sense, especially for low earners who can't save on their own. But overall I've shifted toward CBO's view of things, in which stocks aren't treated as 'free money' for the system, versus the SSA actuaries' view which tends to focus on returns more than risk.

  4. Is solvency relevant?

    If the program becomes insolvent, by law it can no longer pay full scheduled benefits. That will be a pretty important consideration for people at the time. More broadly, we should be thinking in terms of the efficacy of the program and the relative costs and benefits allotted to different generations. Those aren't solvency related, but are nevertheless quite important.

  5. Are the stochastic projections (Appendix E) meaningful?

    The stochastic projections, which attempt to quantify the uncertainty regarding future financing projections, are far superior to the Low Cost/Intermediate Cost/High Cost scenarios the actuaries have traditionally used. CBO uses only a stochastic simulation to portray uncertainty. Why are the Low/High Cost scenarios flawed? Each demographic and economic variable (wage growth, fertility, etc.) is assigned a high and low cost value, in addition to the intermediate value that we generally focus on. The Low or High Cost scenarios combine the low and high cost values for each variable to show how things turn out if everything goes right or everything goes wrong. What's wrong with this? First, there is no probability assigned to the low and high values for each variable – they're just 'low' and 'high'. What's that supposed to mean? Is there a 10% chance of being low or high? 5%? 1%? We don't knows. Second, the Low Cost and High Cost scenarios assume that every variable takes its low or high cost value, regardless of whether it's likely that they would occur together. That is, there's no accounting for the covariation between the different variables.

    How does the stochastic analysis work? Each variable has an average value derived from the Trustees intermediate projections. However, it's also allowed to randomly vary from year to based on how that variable varied in the past. For instance, let's assume that real wage growth has an average value going forward of 1.1% per year (this is the intermediate projection, but also the average over the past 40 years). Based on historical date, real wage growth has a standard deviation of 2.1 percentage points. Using a random number generator, the actuaries create thousands of outcomes for real wage growth in each year going forward. They do a similar exercise for each of the other variables. They then feed each random scenario into the actuarial model to calculate the actuarial balance. Each individual random scenario has no real meaning, but together they give a much better view of the level of uncertainty regarding future Social Security financing. The stochastic model isn't perfect, and work is continuing to improve it. But now that the stochastic analysis is available, I personally believe that relying on the Low and High Cost scenarios is all but indefensible – they're simply inferior to the stochastic analysis.

  6. What do the 1984 and 1985 reports show us about intent?

    This question is a little hard to understand, since the Trustees Reports in general don't discuss Congressional intent. If the question is, did Congress in the 1983 reforms intend to cause a big trust fund buildup over the next four decades, which would then be drawn down over the following four decades or so, the general view is 'no.' Neither the Greenspan Commission nor Congress were very well aware that the system would be running large surpluses through the early 21st century, largely because the SSA actuaries didn't provide them with this information. The Commission's concern was maintaining short-term solvency through the 1980s, since benefits were at risk of not being paid, and then reaching 75-year solvency. They would submit potential reforms to the actuaries, who would then tell them whether they system was solvent or not. As I understand it, the Commission was not provided with the year by-year data we take for granted today. There's a good report from the Congressional Research Service on this episode – very instructive.

  7. Why did the TF ratio outperform expectations in the late 90's? Can it do so again?

    The main reason the Trust Fund ratio (and system financing in general) outperformed expectations in the late 1990s was that the Trustees and actuaries underestimated the strength of the economy at that time. The Trustees are reluctant to jump on board with new trends in economic or demographic thinking, at least in terms of applying them to their long-term projections. This caused their Reports in the late 1990s to look pessimistic. However, most economists have now backed off the most optimistic views of the "new economy" and extremely high future productivity. Had the Trustees bought into that view their Reports in the 1990s would have looked more accurate, but they'd be forced to back off significantly today. Can the system outperform expectations again? Sure, and it almost surely will. But it will almost surely underperform expectations as well. For instance, recent Reports might not have anticipated the current economic slowdown and so the system might underperform for a year of two. But the ups and downs of the business cycle aren't very meaningful with regard to their longer term projections.


Anonymous said...


One quibble with your comment on the stochastic approach used in Appendix E of the Trustees Report. I prefer to distinguish between "uncertainty" and "variability." Variability is due to randomness around an expected value (e.g., around the mean). Uncertainty is by it's nature nonquantifiable because one is unsure what the expected value is in the first place. So variability tells us the variance around the mean, but with uncertainty we don't even know the mean itself. The Trustees use historical data to determine the most likely estimate, so there is no "uncertainty" in their analysis, but there is plenty of variability around that mean estimate.

Personally I think the 40 year average is an unreliable estimate of the mean, but that's a different issue.

I completely agree with your larger point that the stochastic approach, although still in its infancy, is far superior to the Low Cost/Intermediate Cost/High Cost approach used by the Trustees in the past. As you say, there is no valid reason to continue using the deterministic approach.


Anonymous said...


you give a very long answer that would require a very long counter answer. i would only be repeating points i have made elsewhere.

but in very brief:

you don't make a case for "starting now." you simply assert that it would be more efficient and generationallyfairer. a little later you propose paying more taxes now to ease the burden of health care on future generations. this is rather strange. in what way to do I benefit from the health care given to future generations. is there some reason they cannot be expected to make their own care/cost decisions?

what does "deadweight losses associated with tax increases" mean?

why should i think of Social Security as a tax instead of as a way to save my grocery money until i need it, protected from inflation?

since a one dollar per week per year increase in the "tax" will solve the projected deficit, why would we want to spread it out over genrations... is that because we are so fond of the "worthless iou's" we already have in the Trust Fund?

and why would we accept a 250 dollar a month benefit reduction when we can keep benefits above suruvival level with an 80 dollar a month tax increase at a time when we will be making 1200 dollars a month more than we are now?

ah, i see... we want to cut benefits so we can save the money for other taxes. pretty good. the people can find another way to feed themselves in their old age, but we MUST free up money for more submarines for the eternal war on terror?

"encourage delayed retirement say by increasig the early retirement age..." brilliant. you mean you would simply raise the retirement age, so that people whose jobs are less fun than yours can work more that forty years, even if they were willing to pay the "tax" that would pay for their retirement, because your friends need elderly slave labor... i think the plan was to eliminate the retirement age altogether and replace it with a note from your doctor attesting that you had only one year to live before you could retire.

stop here. you have slid from a serious consideration of social security to a bogus fugue on financial schemes... which social security was deliberately designed to avoid.


Brooks said...


Re: #4, I think you missed some of the backstory on the question of "relevancy" of SS solvency, and therefore missing the essence of the question.

The question really comes down to this:

If we assume, just for the sake of argument, that under current policies, SS will be fully solvent forever (as some claim), does that mean that SS spending is unrelated to our projected OVERALL deficits? Does it mean that reducing projected SS spending could not reduce OVERALL projected deficits? Does it mean that the only way we could reduce projected SS spending would be to default on Trust Fund bonds?

My answer to all of the above is CLEARLY NO!

Social Security is part of OVERALL spending, and as such, reducing projected SS spending would reduce OVERALL deficits, ceteris paribus. And if doing so would create excessive SS surpluses, we could simply lower SS taxation. If we did that, and offset the reduction in SS revenues with increases in other (non-SS) taxes, projected OVERALL spending would be lower, projected OVERALL revenues would be unchanged, and therefore projected OVERALL deficits would be lower. And it would NOT mean defaulting on Trust Fund bonds.

I hope you'll comment on the above.

Brooks said...


Just a bit of follow-up to my prior comment.

Those who think that full SS solvency under current policy WOULD mean all those things (see the questions in my previous comment) are implying that, as we seek to reduce our long-term fiscal imbalance, SS should be off the table -- that we should not even consider reducing projected SS spending (by reducing benefits, such as via different indexing, and/or reducing eligibility, such as by further raising the retirement age or means testing). Their reasoning (for lack of a better word) is that if SS is already fully solvent, SS spending is not part of the problem of our overall long-term fiscal imbalance, so on that basis alone we should not even consider reducing projected SS spending.

The above is obviously nonsensical, for reasons I've provided in my prior comment.

I hope you'll comment on it.

JG said...

what does "deadweight losses associated with tax increases" mean?

Every tax imposes a "deadweight loss" that reduces economic activity, so the cost of the tax equals (1) the dollar cost of the tax on the taxpayer, plus (2) the resulting decline in economic activity.

#2 can be conceived simply like this: If a tax imposes a cost of $X on an activity, everybody engaged in the activity for a marginal return of <$X simply will cease the activity without paying any tax. (In fact, they will stop paying any tax they paid from it previously).

A nice dramatic example from real life was the 10% luxury tax on yachts imposed as a budget-gap-closer during the Bush I years. It largely put the boat-building industry out of business. Result: No luxury tax collected, business & income taxes lost, unemployment benefit expenses up,. etc. Congress had to repeal it in a year.

The amount of deadweight loss varies by the elasticity response of the taxed item (high for luxury goods, low for land) but it is always there.

The interesting thing about it is that deadweight loss rises not with the tax rate but with the *square* of the tax rate -- so if you double a tax the deadweight loss from it *quadruples*. This is why a basic principle of public finance is to have several taxes at low rates rather than one tax collecting the same amount of revenue at a high rate. Here's a more rigorous explanation if desired.

People who are happy to plan to impose taxes at ever higher rates to pay for entitlements never, ever consider the deadweight loss they would impose on the economy. But they should. Here's the CBO letter from this week on the cost of paying for Social Security and Medicare (that I previously linked to both here and at Angry Bear)::

"CBO estimates that individual income tax rates would have to be raised by about 90 percent to finance the projected increase in spending between 2007 and 2050 ... Under this scenario, real GNP per person in 2050 could be between 5 percent and 20 percent less than what it would be if revenues and spending in 2050 were the same shares of GDP as in 2007."

That 5% to 20% lower GNP is deadweight loss. So the cost of the tax increase wouldn't be merely tax rates 90% higher than today -- it would be tax rates 90% higher than today plus GNP reduced by as much as 20%.

Anonymous said...


Was the 10% luxury tax under Bush 41? My admittedly fuzzy recollection is that it was under Clinton's 93 tax increase.

Another less rigorous but more intuitive understanding of deadweight loss is that it is a tax that accrues as a benefit to no one...not the government and not private actors. It is a pure economic loss.

Another implication is that, as one of my old public finance profs used to say, the best tax is an old, established tax. That's because over the long run the elasticities of factor inputs increases, so economic actors are able to adjust to the tax. In other words, over the long run some of the deadweight loss due to a tax is dampened as long run elasticities converge toward infinity (at the limit). But if tax rates change frequently, then deadweight loss becomes a persistent problem.


Andrew G. Biggs said...

2 slugbaits: Good point: what the stocahstic forecast shows is variability (or risk), while there's additional uncertainty from other causes. For instance, we don't know what the mean value for a given variable is, nor do we know for certain that the future variance will be the same as the past. The stochastic forecast improves our knowledge versus the High/Low, but it's not perfect.

Coberly said: "You don't make a case for "starting now." you simply assert that it would be more efficient and generationally fairer." Another good point. Implicitly, I make a value judgment to place the same value on future generations as I do the current one. Given that, I want to equalize treatment between generations. The future shortfall is basically a result of very generous treatment of past/present participants -- the so-called 'legacy debt'. I'd like to distribute this proportionately over all future cohorts, based on ability to pay (ie, each pays the same percentage of his earnings). I could explain further, but it would demand a longer post.

Brooks: You're right the Social Security, solvent or not, should be seen in the context of everything else the government is doing. We might be able to afford everything individually, but not when put together. Part of my view on fixing Social Security is informed by my views on health care reform, but I assume you could go further.

Bruce Webb said...

Thank you Andrew. But a couple of comments/questions. Within limits advocates of privatization overlap to a very large degree with advocates of tax cuts on capital (flat tax, estate tax elimination, elimination of corporate income tax, 15% rates on capital gains, etc) even though the net effect of the latter is clearly an inter-generational tax shift forwards via General Fund deficits. Only in the restricted area of future retirement security does this requirement to be solicitous of the interests of the future suddenly manifest itself. This seems somewhat odd. The newborn of today will be the retiree of 2084, why his future interest in retirement security is a societal imperative and yet his health care in the here and now through something like SCHIPS is not seems intellectually inconsistent to me. I have to say I see much of this through the lens of history and the Election of 1936 and for that matter the Election of 1968. Some people never accepted New Deal and Great Society ideals to start with. Consider this rhetorical.

Second as to deadweight loss. I won't repeat the argument here but in an AB post called Calculating the Cost of Inactivity I show that for 10 of the last 12 years the active cost of not moving on Social Security has been negative, in effect the equivalent of a current year tax cut as the cost of the fix actually shrank. In the other two years you have to do some arithmetic to show that inaction netted out as a net short term tax cut but the numbers work out.

I don't know that you have explicitly endorsed LMS, on the other hand it does propose a sharp increase in taxes for workers and so a dead-weight loss for them (assuming JG's formulation is correct), while doing Nothing resulted in the equivalent of about a 2% tax cut in every year from 1997 to 2007.

People who say variants on 'We can't afford to wait' can only do so by ignoring the actual historical record. A 'crisis' that left unaddressed for a decade actually shrank in magnitude projected forward hardly deserves the name.

Andrew G. Biggs said...

There are two 'camps' within the social security reform/personal accounts movement, generally referred to as the 'pain caucus' (personal accounts coupled with tax increases, benefit cuts, retirement age increase, etc.) and the 'free lunchers' (larger person accounts financed with borrowing, no cuts in benefits, usually a guarantee against market downturns). When it comes to taxes, the pain caucus is closer to balanced budget groups like the Concord Coalition while the free lunchers are closer to the supply siders. Not sure that this totally answers the question, but at least gives some context on where people are.

Regarding the cost of waiting, I just think your argument that things have improved over recent years isn't very strong, since it doesn't imply that things will continue to improve in coming years. (Eg, if stocks performed really well in the 1990s, does that imply they'll perform really well forever?) Put it this way: make your best guess of what the 75-year deficit will be. Whatever that guess is, we can expect the deficit will be worse one year from now so long as the balance 76 years from now is worse than the balance today, which it will be with almost 100% certainty. From here, we can calculate the cost of waiting. I don't think most analysts from either side would accept your argument for delay.

Brooks said...

Bruce Webb,

I can't speak to the arguments presented by others who say "we can't wait" to make changes to reduce projected SS spending, but as for myself, I say "we can't wait" to consider ALL possible means of reducing our long-term fiscal imbalance, and reducing projected SS spending is one way we could do that, in conjunction with others (raising projected revenues via tax increases and cutting other projected spending).

That is the context that you and some others seem to be missing entirely. Even if SS is/were fully "solvent" infinitely under current policies, SS spending would still be contributing to our OVERALL fiscal imbalance, and reducing projected SS spending could still reduce our overall fiscal imbalance. And if, arguendo, cutting projected SS spending would create perpetual, undesirably large SS surpluses under current policies, we could just reduce SS FICA taxation and offset that revenue decrease with revenue increases from other taxes -- resulting in lower projected overall spending, unchanged projected overall revenues, and therefore lower projected overall deficits (less long-term fiscal imbalance).

And the longer we wait to reduce our unsustainably large long-term fiscal imbalance, the greater the ultimate pain and the worse off we will be. I don't think there's much controversy among experts on that point. So yeah, "we can't wait" to get started making whatever sacrifices we end up choosing to make to reduce this long-term fiscal imbalance, whether those sacrifices include cutting projected SS spending or not.

See the big picture. See the forest, not just a tree or two. SS is part of our overall fiscal outlook, not something to be viewed in isolation in some fiscal vacuum.

Brooks said...

Bruce Webb,

My last sentence should have read "...not something to be viewed ONLY in isolation..."

Bruce Webb said...

Well you might have to follow the link. My take is that attempts to take this on via 75 year or certainly Infinite Future projections are inherently futile. Because we don't even have consensus on what is happening in Q2 2008. Instead I suggest we take one of two approaches. One is to adopt a specific policy of targeting, instead of predicting what may happen we would establish a particular policy goal and then pursue it.

Say we decided that current CPI measures were just giving future retirees too good of a deal. Or contrawise that as a society we wanted to preserve the current schedule, either way we could then design an economic model using whatever combination of growth, tax increases, benefit cuts and CPI assumptions would be needed to achieve that goal. And then enact whatever measures would be needed on a lagged basis to achieve that goal. At which point we watch and wait on current numbers. If they are better than the model requires we could decide to move the proposed changes up some, if they are worse we could make the reasoned decision about whether to tinker with the proposed changes or accept the new result.

The second approach is just to continue to calculate the Cost of Inactivity on a year over year basis. Because I know what not doing something in 2006 cost me, less than nothing. An equivalent of 2.02% of current and future utility of my payroll dollar was left under my control. Total cost to society? -.05% per year going forward. At that point I was 17 years from retirement and so gained a net combination of .85% of my average annual salary in obligations not taken on. Doing nothing in 2007? -.07% per year going forward with a resulting non-obligation of 1.12%. Nothing in 2008? Well we will have to see. But it takes a really big jump in the payroll gap to offset savings already banked by not acting in years past.

'Nothing' may not be a viable plan long-term, but we cannot and should not take the tack that future utility of that wage dollar automatically is privileged to current utility. I call this the Pampers vs Depends question, I know what it costs a single mother to lose that payroll dollar right here in 2008, whereas her needs fifty years down the road are a lot more uncertain. The numbers suggest that doing 'Nothing' is a viable plan for 2008 just as it has been a proven plan over the last decade. Because in the end the 'pain caucus' is really not proposing to take much of that pain on themselves, instead the net effect of these proposals is to reduce the pace and perhaps the scale of the repayment of the money they borrowed, kind of a 'pain for thee, gain for me' approach.

Bruce Webb said...

Brooks this isn't my blog, it isn't your blog and it isn't Angry Bear where I have posting privileges. Your attempt to drag your grievances over onto third parties is nothing short of rudeness to Andrew Biggs, though I am sure he can speak for himself. I for one have no intention of wasting his time or column inches responding to you here.

Brooks said...

Bruce Webb,
If you don't wish to respond substantively to what I've said (as you never do anywhere), you should not respond at all here. Don't engage in your typical, utterly baseless personal attacks on this blog. THAT is what is inconsiderate to Andrew Biggs and his readers. This isn't the place for you to start yet another food fight through false accusations just because you don't want to engage substantively and wish to fabricate some excuse to save face.

If you believe my points are invalid, try to refute them. Otherwise, our dialogue here should end right now.

Anonymous said...

Biggs said

"the so-called 'legacy debt'. I'd like to distribute this proportionately over all future cohorts, based on ability to pay (ie, each pays the same percentage of his earnings)."

no attempt to quantify that legacy debt in terms of real costs to real people or explain what it means to distribute it proportionately over all future cohorts.

Orszag suggests a parable in which poor son has to pay off his fathers medical bills leaving him with less money to earn interst on for his own retirement, to be made up for by His son contributing to his retirement, and so on.

He doesn't consider the possiblity that the first father actually paid for the Sons medical bills, or education, and that's why he didn't have enough money to earn interest on for his own retirement.

he surely did not look at the generatioinal equity of Father having to spend two years as a draftee, or the father's two percent being on a lower income than the sons, making it actually a larger cost in terms of what was left over for "discretionary" spending.

and of course, what Biggs seems unwilling to understand is that if those future generations are going to be living longer than the present generation why should they NOT pay a greater percentage for their retirements than we will...

you are going to have a retirement that is fifty percent longer than mine, and you are objecting to a 2% tax increase to pay for that??? (of course, the Biggsians will be the first to point out that that 2% tax increase is really a 33% increase on a 6% tax.)

the horror, the horror.


Anonymous said...


re deadweight costs

your theory is just that. one i think not well supported.

are you saying that the government doesn't spend the money that they tax away from you... stimulating the economy?

or are you saying that the money i put away and don't spend today so i can spend it when i am old and have no income... is deadweight loss to the economy?

then what about the fact that the money i save is "really" taxed and used given to the people who are old today who will spend it very much as i would have spent it had i not needed to save it?

i think you must be one of those economists who only looks at one side of the ledger, depending on which way he wants to misdirect the audience.


Anonymous said...

Biggs said

"Whatever that guess is, we can expect the deficit will be worse one year from now so long as the balance 76 years from now is worse than the balance today, which it will be with almost 100% certainty."

and yet Bruce said, that for the most part waiting over the past ten years has resulted in a 75 year balance that is better than the one we started with.

personally i don't think the 75 year balance matters at all. but people who choose to live by them should.. well, live by them.

meanwhile, i just know in my heart that those people 75 years from now will be able to look at how much money they have, and how much of that it is prudent to save via the payroll tax.. to pay for a decent life for the then-reitired, and simultaneously stake their claim to the same deal when they get old. sometimes it's just too hard to explain pay as you go to people who don't want to understand it.


Anonymous said...

Biggs said

" it's relatively easy for people to make up for reduced Social Security benefits by saving more, "

this seems to ignore that putting your money into the payroll tax in return for non reduced Social Security benefits is the same as saving more.

you have to be careful of plausible sounding words that don't actually mean anything.


Anonymous said...

Biggs says

"Demographics are almost certainly more important than productivity"

well, since the "supply side" of Social Security is a product of demographics time productivity, "which is more important" starts out as a nonsense question.

what turns out to be the answer is that some combination of population growth (or not) and real income growth (or not) will determine the extent to which benefits to retirees can exceed the amount they paid in taxes in their turn.

there is no particular reason to be married to any particular "return on investment".. even a negative return could be justified by the CERTAINTY of the return.

nor is there any way that a declining economy (total productivity... again, demographics times per capita productivity) would be able to provide a given number of retirees with a greater return on investment... without cutting the wages of the still working... however you obscure that with inflation in "assets."


Anonymous said...

Biggs says

"I don't believe switching from bonds to stocks is of major importance. Any investment has a combination of risk and return,"

and here i agree with him.

what puzzles me then is why he wants to move the working people away from a zero risk "investment" like Social Security into a stock market which always has risks... even for low risk government bonds.

but he is talking about the Trust Fund, and there the argument was never so much about the rate of return (vs risk) as the problem of government interference in private enterprise. (and the moral hazard of the gummint using the TF special treasuries to spend more money than it would otherwise).

it is however interesting to note that the private enterprise people are not shy about a government program to force people to invest in private securities.


Anonymous said...

Biggs says, "If the program becomes insolvent, by law it can no longer pay full scheduled benefits. That will be a pretty important consideration for people at the time. More broadly, we should be thinking in terms of the efficacy of the program and the relative costs and benefits allotted to different generations. Those aren't solvency related, but are nevertheless quite important."

well, but not a word about why the program should become insolvent. or whether the partial insolvency envisioned by the trustees would be serious (as Rosser says not, and I say yes).

we would have to assume that people, and the congresscritters, after 2030 or so would be stupid enough not to raise the payroll tax one tenth percent per year to avoid "insolvency."

or we could agree with Biggs that the real problems are "not solvency related."

as for the relative costs and benefits alloted to different generations, Biggs does not explain why people who are going to be living 50% longer than I am should not be paying 50% more than I am for their longevity insurance. and remember this "50% increase" is actually a 2% increase. gotta watch those denominators.

or why we should single out Social Security as the one place society cannot tolerate a change in the price of something (insurance) to reflect a change in the cost of providing it.


Anonymous said...

are the stochastic projections meaningful?

probably not. except to statisticians who like to make bets about the best guess for numbers they can't measure.

of course it's very different predicting how many lightbulbs out of a million light bulb production run will be defective.

and what is the "best guess" for the cost of a single year of program out in the science fiction future.

likely the stochastic projections will be harder for someone like me to explain the problems with.

but more importantly, there is NOTHING about Social Security that requires any predicting at all.

We know the input variables cannot change very fast, and we can always change the tax rate to meet this years needs, without seriously affecting the budget of any taxpayer.

yes we would want to pay attention to possible, or probable, changes in those variable that could result in a serious cost challenge in the future.

but so far no one sees any such likely change. and no one sees any way to avoid paying for such a change should it occur... aside from telling granny she is on her own while disguising the money flow with wall street games and really sophisiticated statistics.


Anonymous said...

Biggs says

" But the ups and downs of the business cycle aren't very meaningful with regard to their longer term projections."

once again, I agree. but the ups and downs sure provide a good chance for the privatizers to scream "we're all going to die" on the downs, while chanting "who needs that old Social Security" on the ups.


Bruce Webb said...

My main issue with both the Stochastics and Infinite Future PV calcations is their relative novelty they each appear out of nowhere with the release of the 2003 Report. Neither had ever been thought needed before, why Spring 2003. Suspicious minds then and now speculated whether this was a reaction to Baker/Weisbrot's Phony Crisis which pointed out the dual facts that only were IC assumptions relatively pessimistic when gauged against recent economic performance, but that the actual fix cost was shrinking whether that fix was considered in either growth terms or as a percentage of payroll. The Stochastics allowed privatizers to stave off the 'too pessimistic' charge, Infinite Future allowed the size of the fix to be almost doubled to 3.7% while it boosted 'Unfunded Liability' by a full $10 trillion over that of the 75 year window.

I vividly remember commenting at DeLong's in 2005 when some guy I had never heard of dropped the unqualified claim that the payroll gap was 3.7% rather than what I considered the official 1.92% number. When I inquired "Whence 3.7% Samwick?" he publically and DeLong by e-mail pointed out he was using some newfangled Infinite Future number. Then and now I thought this wasn't exactly Cricket, I had been working for years in the accepted frame, suddenly they changed the rules on me.

Ibam afraid you will just have to excuse me for believing that both changes were defensive in nature, meant to push back on Baker et al, which by that point included me.

Brooks said...

Bruce Webb,

Re: your 6:33pm comment,

Your suspicion of the motives of someone presenting a conceptual/analytical approach is not an argument that the approach is invalid. Granted, if you don't understand the concept/analysis, and if you have reasonable grounds for suspicion of motive, such suspicion is a reasonable source of skepticism. But I don't think you're claiming not lack understanding in this case, so it seems that if you're going to reject the approach in question here, you should come up with an actual refutation, not just a statement of your suspicions of others' motives (which, in my experience, is not reliable anyway, but that's mostly beside the point). Debate the approach on its merits, or lack thereof if that's your contention, rather than focusing on your suspicions, accusations and conspiracy theories.

Brooks said...

Bruce Webb,
typo: should have read "...I don't think you're claiming to lack understanding..."

Andrew G. Biggs said...


Re the stochastic and infinite horizon projects: the 1999 Tech Panel recommended new measures of both uncertainty and of system financing post-75th year, so it's not as if they were just dreamed up by GOP operatives. CBO was doing stochastic projections before SSA, and SSA basically just adopted their model. As I said in my post above, the 1994-96 Advisory Council also urged policies that went beyond 75 year solvency, and they also funded development of the Policy Simulation Group's SSASIM model, which was (I believe) the first Social Security model to do stochastic simulation. Once you know the background and rationale, there's really nothing nefarious about either of these measures.

Bruce Webb said...

Biggs you are granting a little more good faith, New Deal credentials to Clintonian DLC/third way centrists than I am willing to.

Dale!! Come to the Light Side!! You don't need to be doomed to anonymity on Blogger sites. Plus you get a gig of free web space plus a nice free e-mail account to which you can direct all your spam. E-mail me using the account only the ABs know and you will only be a couple of clicks away from being you here and on any other Blogger site. said...


Well, the stochastic method depends on looking at past volatilities. However, you have admitted that the late 90s were not accounted for, and there have been other years since 2000 when the low cost projection was met. You agree that we do not know the probability of it, even though it has done better over the past decade than the others.

So, why not wait and see how it does? Why not wait and see if it looks like we might actually move towards having a deficit in 2017 on the TF as projected by the IC rather than jumping the gun? There is in fact plenty of time to make adjustments if need be, especially as we need to do something about the exploding medical care cost problem now.

Barkley Rosser

Andrew G. Biggs said...

Barley -
"Well, the stochastic method depends on looking at past volatilities. However, you have admitted that the late 90s were not accounted for..."

It's not that the late 90s aren't accounted for in the current stochastic model, it's that they weren't predicted in the intermediate projections at the time -- just as at the time of the 1983 reforms they overpredicted future wage growth. These things are hard to do, especially in the short term. If you run the stochastic model 5,000 times, there will be instances where everything comes up roses or everything comes up mud, but they'll be very rate. Likewise, there are instances of very high economic growth for a few years, such as in the late 1990s, but we can't reasonably expect that to continue for 75 years.

Barkely: "So, why not wait and see how it does? Why not wait and see if it looks like we might actually move towards having a deficit in 2017 on the TF as projected by the IC rather than jumping the gun? There is in fact plenty of time to make adjustments if need be, especially as we need to do something about the exploding medical care cost problem now."

Two related reasons: First, and simply, if we wait until we're in a crisis then solutions are harder, and a crisis isn't a good time to make good policy. Second, if the optimal solution is to smooth tax increases and benefit reductions over as many cohorts as possible, then by waiting we're reducing the number of cohorts that take part in the reform. That's great for them, but worse for everyone else.

Beside, since so many people focus on year-to-year variability, what will a few extra years tell us? Let's say we reach 2017 and it's looking like the intermediate cost projections are correct -- people could still argue that the future will be rosy and there's no need to worry. If we act now we can scale things back if things turn out better than expected, and there are policy approaches that would do that. But if we fail to act, there's a very high level of confidence that we're just making things worse on future generations. That seems negligent to me. said...


If we get to 2017, or even a few years before, and the IC projections are coming on strong, you will not find me saying that "everything will be rosy in the future." The basis of hoping for the latter has been the fact that LC has easily beaten IC over the last decade and not done too badly over even the most recent years. So, it is not completely stupid or super low probability to project something similar forward. That will not be reasonable if over the next five to ten years IC consistently beats LC, or reality is even worse than IC.

Also, if the situation is a balance in 2017 about to turn into a deficit, that is still not a "crisis," not even remotely close. But, there would be more serious credibility about doing something than there is now.

As it is now, this push looks like either an effort to cut social security benefits (so that they do not need to be cut later, maybe), or an effort to further tilt the tax code to the regressive fica away from the income tax, which was the real bottom line of the total of tax system changes during the Reagan era, unless of course one just does an Obama-style "raise the income cap" approach.


Lord said...

The problem with 1 is it is mostly likely not right and it will take quite a while to know how far from right it is. That bears heavily on when it should change. The model accuracy has been limited to around 30 years so attempting changes for what we know the model can't predict anyway are less then helpful. They could well be harmful. Also, the sooner the better argument falls short on macroeconomics. Societies for the most part cannot transfer consumption from the present to the future or across generations. Solutions like these simply don't exist and we shouldn't pretend they do. Both uncertainty of whether change will be necessary and inability to effect change anyway recommends delaying for as long as possible.

Anonymous said...


thank you.

i had to note that my argument that there is no way to fix the "crisis" beforehand, and not much cost to fixing the balance when the imbalance does appear...

have fallen on deaf ears.

or maybe just.. "oh, look over there! see the looming crisis!" ears.

Biggs does a poor job of justifying either the looming crisis idea or the idea that we gain anything by "fixing" it now...

but there is this: i think Biggs et al concern with the plight of future generations is related to the old scam that "personal finance" salesmen run on prospective customers.

they like to show how a dollar invested in 1776 at 10% compounded annually would now be worth 3 billion dollars.

it is only by talking about some infinite future tha Biggs et al can create the illusion of a big enough problem to panic people into failing to realize they pay for their breakfasts one day at a time.

the whole scam depends first on the idea that we can never, ever, increase the payroll tax, and that meanwhile some amount of money invested at compound interest will eventually make us all rich and give us manna from heaven.

a very emotional, non statistical argument against the Biggs view may be found on angry bear. i won't apologize for it. the real reality is emotional, not statistical.


Andrew G. Biggs said...

I'll leave most of Coberly's post aside, but this is worth commenting on: "it is only by talking about some infinite future that Biggs et al can create the illusion of a big enough problem to panic people into failing to realize they pay for their breakfasts one day at a time."

This actually isn't the case. The source of Social Security's deficits isn't shortfalls in the infinite future; it's that past and present participants received extremely high returns relative to what they paid in, and to equalize total benefits paid with total taxes collected future generations have to receive a lot less.

According to this section of the Trustees Report (, past and present participants received $15.2 trillion more in benefits than they paid in taxes. In other words, had they only received an actuarially fair return (ie, IRR at the trust fund bond rate) the trust fund balance today would be $15.2 trillion bigger than it is.

The table also shows that all future participants can expect to receive $1.5 trillion LESS in benefits than paid in taxes, even under full scheduled benefits. However, to balance the system, future participants don't just have to swallow that $1.5 trillion net tax, but also the $15.2 trillion net benefit paid to past/present participants.

So basically we know with certainty how large the shortfall is because it's already occurred and must be paid off in the future. The net tax per person in the future can decline if the population increases (higher fertility) or scheduled benefits get to be an even worse deal (say, lower lifespans so people collect for fewer years).

But we don't NEED the infinite horizon measure to tell us the shortfall; we can see it simply by looking at how past participants have been treated. The fact that the infinite horizon measure matches the past treatment shows how it's consistent in a way that the 75-year actuarial balance isn't.

Bruce Webb said...

Ouch. On Memorial Day of all days that someone would point out that the Greatest Generation was collecting more Trillions of benefits from the country they first protected and the economy they then grew is kind of mean-spirited to start with. This is the last day to pose the question 'What have you done for me lately?'

If slacker Gen-X people get off the couch and grow the economy like the Greatest Generation and their kids the Boomers did that $1.5 trillion gap vanishes like morning dew. said...

Rather than saying "ouch" like Bruce W., I would like to thank Andrew for giving the lie to the so frequently heard complaint from the Gen-X-ers et al that all the problems have been caused by the baby boomers, who, of course have been paying through the nose for fica since the Greenspan Commission reforms, providing that nicely rising surplus that may yet remain with us with a higher probability than Andrew is willing to grant... :-).


Anonymous said...

The answer to the first question is "Wait as long as possible" because the intermediat projections probably are not correct. With the SS surplus presently over $190 billion and growing, the question almost answers itself.

Anonymous said...


I was thinking "won't he hurt himself doing that ?" Bruce tells me you are a professional and can do it without permanent harm to yourself. But I am not so sure.

I have heard of a blind topologist who can mathematically invert the sphere. But I don't think he proposes going around taking every child's balloon and turning it mathematically inside out so the harder the child tries to blow it up the smaller it becomes.

It is the first part of your answer that is the most literally insane. I have found it very hard to explain insanity to those who embrace its seductive charms, but you have provided me with
a topic for a future sermon.

Meanwhile my simple arithmetic does not support the second part of your answer... that future cohorts will pay in more than they get out. Is there something simple I am overlooking, or does this require a sophisticated mathematical imagination to understand?


Andrew Biggs said...


I can't understand all of your comment, but I can answer this:

"Meanwhile my simple arithmetic does not support the second part of your answer... that future cohorts will pay in more than they get out. Is there something simple I am overlooking, or does this require a sophisticated mathematical imagination to understand?"

I basically have two sources: the first is the Trustees Report, which simply states it to be the case that the present value of future participants taxes will be $1.4 trillion or so larger than the PV of their benefits.

Alternately, you can look at projections of future internal rates of return for stylized workers (available at the SSA actuaries' web site: For instance, a medium wage two-earner couple retiring in 2050, which I'm taking to be roughly representative of the population, can expect a real return of around 2.45% under scheduled benefits. Now, the trust fund bond rate is projected at 2.9%. The difference implies that the present value of this couple's total taxes exceeds the PV of their benefits.

Does this more or less answer the question? Thanks.

Anonymous said...


ah, yes. PV.

which would be useful to know if Social Security was the sort of thing that could meaningfully be addressed by PV. my understanding is that it was designed explicitly to be outside the normal risks associated with PV.

Since you cite a real return for that couple at 2.45%, I will assume you agree with me that they will not be getting back less than they put in.


Andrew Biggs said...

Coberly -- These issues are almost always discussed in terms of present values calculated at the trust fund interest rate. For instance, the so-called money's worth ratio equals the present value of benefits divided by the present value of taxes. A value of less than 1 implies you're 'losing money', even if you have some positive return, because an alternative risk-free investment would have gained you more. You won't find many serious analysts disagreeing with this.

However, if you still don't want to believe in present values and think any positive return is ok, I'd be happy to borrow $100,000 from you at 1% interest. You'd still be making money, at least in nominal terms, but you'd be a fool to take the bargain.

Anonymous said...


i guess "always discussed" is the problem. And "serious analysts" appear to be dead wrong.

I wouldn't lend you money at any interest. The risk factor you know.

Social Security was designed to be outside the "risk factor" universe. If you don't know that, you don't know anything about Social Security.

oh, not such a fool as to lend money at below expected rate of inflation. that's why its nice to know that Social Security is protected from inflation by "wage indexing." no need to guess about appropriate "interest rate" or appropriate degree of risk, which are at the heart of any PV calculation.


Anonymous said...

Dear Mr Biggs,

apparently you and i have different ideas about what constitutes a serious discussion. for you a serious discussion begins by accepting your premises and assumptions; and by ruling out any questions that would lead to examining those assumptions.

Serious enough for a blog no doubt, or for one-armed economists. But not all that much fun for me.

I would invite you to examine the seriousness of your own thread of logic in the current discussion.

You began by claiming that future taxpayers would get less in benefits than they paid in.

I replied not according to my arithmetic.

You came back with a citation to a claim about PV and a statement that the real return on Soc Sec would be 2.5% compared to a real return on Trust Fund notes of 2.9%

I replied that PV had nothing to do with Soc Sec, and that a 2.5% real return was evidence that taxpayers would get back more than they paid in.

You replied that all serious analysts use PV and then offered to teach me a lesson about PV with an example loan at 1%.

Of course the loan has nothing to do with PV... making me wonder if you even understand the concept. And your point about the 1% being a way to lose money seems to have lost track of your own projected REAL return to Soc Sec of 2.5%.

Now, if we were having a serious discussion, we could straighten all this out and go on to perhaps discover that my "PV" analysis for Soc Sec invokes different paramaters than yours would. Not a moral problem, just a different evaluation of the risks... but under those circumstances it is unlikely we could close the deal.


Andrew Biggs said...

Coberly -- I apologize if my use of PV was implicit rather than explicit; probably it should have been. How about if I rephrase and say that for future participants Social Security will be actuarially unfair (a term that is based on present values) to the tune of $1.5 trillion?

Anonymous said...


if you said that, you would just be repeating your argument with no evidence that you understood mine, or what was wrong with yours.

you may need to review the concept of PV. or perhaps there has been some concept drift since i learned that it had something to do with comparing the value of investments with different maturities, and different interest rates, and different risk evaluations. and that such comparisons only make sense if one person is doing the evaluating, or if two parties to a deal are each making their own evaluation and then making their offers based on that evaluation.

i see no reason why i should accept your PV. and i am fairly sure you could not persuade an informed public to do so.

a half percent difference, entirely arbitrarily calculated, projected out into the unknowable future, with unknowable risks... and zero evaluation of even the knowable risks... would not persuade me to buy your bag of magic beans in return for a guaranteed as well as anything can be guaranteed adequate income for life when i become to old to be able to work, or to want to work, or to be allowed to work by your friends looking to maximize their returns on my money.


Paul Lawin said...

Mr. Biggs,

I have to admit that I'm not following "generationally fair". Can you give me a definition and a sample calculation?

(I know that you've written a lot on SS. I'll be happy with a link if you don't want to re-type an explanation here.

Andrew Biggs said...


Generational fairness refers to how individuals in different birth cohorts are treated by the program, particularly in terms of the benefits they receive relative to the taxes they pay. A simple way to think about this is in terms of the 'internal rate of return' (IRR) paid to different birth cohorts, though in academic work it's more common to talk about net taxes or net benefits. These equal the present value of lifetime benefits minus the present value of taxes, expressed as a percentage of lifetime earnings.

You can find information on IRRs for different birth cohort at this link ( For instance, a single medium wage male born in 1920 could expect to receive a return of 2.83% above inflation, while a similar person born in 2004 could expect to receive only around 1.17%.

One of the main goals of reform is to smooth things out over generations; this involves current or near-retirees doing a little worse, but future retirees doing a little better.

Paul Lawin said...

Mr. Biggs,

I think I understand your definition. I'll stick to the "IRR to a cohort" concept since it avoids an assumption regarding bond yields.

I accept the notion that the IRR a cohort gets from a pure paygo system with constant tax rates is equal to the growth in the tax base over their working/retired years. Suppose that in such a system cohort A averages 3 children per couple. Their children, cohort B, average 2 children per couple.

A will clearly have a higher IRR than B. (They will also have higher net PVs if you keep the discount rate constant.) By your definition, this qualifies as "unfair". But in my, admittedly normative definition of "fair", it seems perfectly appropriate.

Am I missing something?

Anonymous said...


apparently Biggs idea of generational fairness is to have the current generation do worse

but some future folks do better.

it is worth noting that we can do something about current policies. we can't do much about future conditions.

so he is expecting you to buy a magic bean at a real cost to your welfare, in the hope of making some future person.. whose real income will be two or three times more than yours... marginally better off in his payroll tax rate.

one thing he is not very clear about is just where the generational fairness comes in when granny worked nights scrubbing floors so your mom could have enough to eat during the depression, and later she paid for your operation and contributed to your college fund while paying the taxes that build the colleges and the roads and defended the country from the evil empire...

but she got a better deal on her social security, so you have no choice but to sulk and waste your days thinking 'she owes me' or some equivalent insanity... in - sane un healthy un clean morally repugnant fixation on an imaginary debt of two cents while overlooking a gift of tens of thousands of dollars, both directly and in terms of creating the basis for your own income, which is much much higher than granny's. take my word for it.

and to make the "debt" look big enough not to be laughed out of court he has to project it at an unknowable rate of interest out into the distant future and report the aggregate over a hundred million in poplulation.

by his own numbers the "inequity" he talks about looks like about a hundred dollars per year per taxpayer or less. and for this all he asks you to do is give up a zero risk comprehensive insurance plan (he does not count the insurance value of the plan) in return for a sure thing on the market.


JG said...

Thanks for the pointer to the CRS report. I've been picking that material up a bit at a time by going through the SSA history and oral history sections and various other sources -- and here it all is wrapped up in a neat little package. Kudos!