Tuesday, September 30, 2008

Social Security: Hard Solvency vs Soft Solvency vs Sustainable Solvency

'Sustainable solvency' is a new buzz phrase floating around with 'intergenerational equity' and 'unfunded liability'. But before exposing how the term blurs the real issue we need to examine the concepts of what I call 'hard solvency' vs 'soft solvency'.

'Hard solvency' is used here to describe a Social Security system that can pay 100% of all scheduled benefits over the seventy-five year actuarial window with no changes to current law in the form of tax increases or increases in retirement age. If and when we ever achieve it supporters of Social Security can clearly call 'game over, and we won'. But we are not there, instead the most recent Report would have us being able to pay out full benefits until 2041 and 78% after that. These numbers are subject to change and on balance have been improving (the 2007 Report had a 75% payout at Trust Fund depletion). Still as long as the numbers continue to improve supporters have a good case for pushing a plan consisting of 'Nothing'.

Now 'soft solvency' is conceptually a little more difficult. The current schedule of benefits is set up so that future generations of retirees get a fair share of real wage increases over their working lives, that is it provides improvements in standard of living equivalent to those of people still in the work force. Per Prof. Rosser of JMU this improvement means a benefit in real terms of 160% of what similarly situated retirees get today. And if we use what I call Rosser's Equation we see that 78% of 160% = 125%. Which is to say that after TF depletion future retirees will be able to buy a basket of goods 25% better on average than my mom can today. Now it might be that the future basket will pale somewhat when compared to people still in the workforce or to the basket that retiree was collecting in 2040 but it can hardly be described as some huge crisis. By this standard any time Rosser's equation delivers a result of 100% or better we can say that the overall system is in 'soft solvency'. Another way of describing this would be in terms of 'intergenerational equity'. If the Gen-X retiree of 2041 is still getting a real benefit as good or better than his grandmother did in 2008 there can hardly be any equity issues to resolve.

If we establish soft solvency as a floor and hard solvency as a goal we can say that currently we are 40% of our way home. Whether we really need to take positive steps in the short run to boost the result of Rosser's Equation is a question of policy preference and a balance between current utility of dollars for workers today against future utility for retirees thirty years down the road. And by the way not forgetting that these two populations largely overlap. That is it may be more efficient to hope to close the gap between 125% and 160% by mechanisms outside of Social Security itself.

But in any event both hard and soft solvency look at the outlook from the standpoint of the beneficiary for whom results closer to hard are on balance better than results closer to soft, but not necessarily so much so as to simply accept higher payroll taxes to achieve 'hard', 'soft-plus' might be perfectly acceptable if need be.

When critics of Social Security use language like 'dead broke' and 'bankrupt' all they are really saying is that Social Security is not currently passing the test of hard solvency. Which is fair enough, but it also suggests that any fixes should move the needle away from current levels of soft solvency towards hard, that is getting that 78% number up. Instead they pivot and invoke the concept of 'sustainable solvency'.

'Sustainable solvency' in practice means eliminating 'unfunded liability' which in our terms is the difference between the current projections of soft solvency (78% of 160% = 125%) and hard solvency (100% of 160%), that is it is a gap in benefits. But in their terms 'unfunded liability' is a pure burden on future taxpayers and so seen as a gap in future income vs cost. From that perspective it doesn't matter whether you address the gap on the income side or the cost side, instead the focus is on eliminating it one way or another. What this implicitly does is to kick hard solvency to the curb. Even though in a real sense 'crisis' starts out as a failure to achieve hard solvency, the solutions simply accept that some version of soft solvency is in fact good enough, if that is if it can be used to introduce a system based on Personal Retirement Accounts (PRAs).

So the task for supporters of Social Security as currently configured is to put the privatizers' feet to the fire. Does their solution give a better result then the current level of soft solvency projected? If not why should current workers and near retirees buy in? The system as currently configured is projected to deliver a minimum of soft solvency plus and is trending more and more towards hard solvency. Why should we accept a result that doesn't deliver more solvency at equivalent cost?

Social Security Checkup: Monthly Trust Fund Reports

Near the end of each month the Treasury Dept releases Trust Fund Reports giving balances to the penny for the previous month. By comparing these balances to the projection in the Annual Reports we can get a rough idea of how Social Security is doing year to date. This year's Report was released on October 2. Two caveats:
One, these numbers are not seasonally adjusted and I couldn't tell you the relative impact on total wages of summer employment vs harvest vs holiday. Each are marked by the entry of temporary workers into the system.
Two, Social Security collects FICA on your total check right up to the point you hit the annual cap at which point they stop collecting at all. For example if you have a salary of $200,000 per year you would see full deductions for Jan to Jun, a small deduction for July and nothing thereafter. This should make for the earlier parts of the year seeing relatively higher collections than the latter. But once again I can't quantify the effect.

So this is an imperfect tool. But it is what we have. So lets have some numbers.
The web page for all the Trust Fund Reports (including Medicare HI, Highways etc) is Trust Fund Monthly Reports While generally Social Security is reported for convenience as having a single Trust Fund which will go to depletion (or not) in some future year (currently 2041), in reality there are two Trust Funds which are legally distinct and can have different levels of solvency and so different projected depletion dates. The larger by far (about 10 to 1) is the OAS (Old Age Survivors) TF. Its report can be found at OAS Monthly TF Report. Its smaller companion the DI (Disability Insurance) TF can be found at DI Monthly TF Report These can then be cross checked against the relevant tables in the 2008 Report:
Table IV.A1.—Operations of the OASI Trust Fund, Calendar Years 2003-171 [Amounts in billions] and Table IV.A2.—Operations of the DI Trust Fund, Calendar Years 2003-171 [Amounts in billions]
OAS::Opening balance//Projected year end balance-Intermediate Cost//Y-O-Y Increase//Year end balance-Low Cost//Y-O-Y increase-Low Cost
$2.023 trillion // $2.216 trillion //$193 billion// $2.221 trillion// $198 billion

DI::Opening balance//Projected year end balance-Intermediate Cost//Y-O-Y Increase//Year end balance-Low Cost//Y-O-Y increase-Low Cost
$214.9 billion//$218.7 billion//$3.8 billion//$221.3 billion// $6.4 billion.
Per IV.A2 the opening balance for DI was $214 billion, projected year end under IC $218.7 billion, under LC $221.3 billion

June 30th/Mid-year: OAS $2.140 trillion// DI $220 billion

Aug 31st/Two-thirds: OAS $2.164 trillion// DI $219 billion

I had some calculations up but they all got garbled in my head so I deleted them. Bottom line 2008 is shaping up to be a kind of sucky year for Social Security. What was a decent picture for OAS mid-year now is kind of dim, it is unlikely that we will even hit Intermediate Cost projections. What was a really bright picture for DI is now less shiny, from June to August the balance actually shrank, if that continues at the same rate it too will fail to hit IC projections.

Which just goes to show the ultimate truth about Social Security, it prospers in good times, it shares the pain in bad as receipts react to covered employment and real wages. (We have a parallel here with the late seventies, a reform was put in in I believe 1977, it didn't fare well when stagflation simultaneously choked off revenue while boosting cost. The result was a new crisis and a need for a bigger reform in 1983).

Sunday, September 28, 2008

Social Security Actuaries score the Warshawsky Plan

(Cross posted at Angry Bear)
Andrew Biggs directs our attention to a new detailed PRA plan by Mark Warshawsky, a member of the Social Security Advisory Board: Notes on SS Reform: Actuaries Score New Reform Proposal The post does not link to the plan itself but instead to a detailed scoring of it by the Office of the Chief Actuary in a memo to be found here (PDF) Estimated Financial Effects of “A Reform Proposal to Make Social Security Financially Sound, Fairer, and More Progressive” by Mark Warshawsky

I just came across this and haven't studied it in detail (and boy is there a bunch of detail) but like any plan it raises some standard questions.

1) Is the rate of return assumed on the PRAs actually reasonable under Intermediate Cost assumptions? (The No Economist Left Behind Challenge).

2) The plan assumes a direct transfer from the General Fund to supplement the PRAs starting in 2012 equivalent to .5% of payroll. Given that one current definition of 'crisis' is 'General Fund transfers starting in 2017 to pay partial interest', adding an additional transfer amount starting even sooner seems to undercut the overall message. How do advocates of the plan address this?

3) The Warshawsky Plan assumes a whole range of cuts and adjustments to retirement age, tax on benefits, and to the benefit formula generally. Each is scored individually as seen in Table A (which follows the actual memo page 14). Any combination of those scored cuts and adjustments that add up to 1.7% of payroll would put current Social Security in Long Range Actuarial Balance. What happens if we just take this cafeteria style?

Provision 3 would modestly raise the cap for a net addition of .15% of payroll. Provision 4 would gradually expose all SS benefits to taxation for an additional .24% of payroll. Provision 6 would bring in all new State and Local government employess for a net addition of .22% of payroll. Provision 7 would increase early and full retirement ages for an addition of .56% of payroll. Provision 8 is a little obscure but it would seem to just reduce lifetime benefits for disabled workers for a net addition of .35% of payroll. For a total combination of 1.52%. I don't really get provision 5 but it would give you an additional .65% of payroll for a new total of guaranteeed tax increases and benefit cuts of 2.17 of payroll. Which is to say .47% more than would be needed to simply fix the program as is.

Now Warshawsky sweetens the plan by reducing payroll tax by 1.0% (presumedly translating to a increase in worker pay of .5%) but offsets that with a new transfer from the General Fund of .5%. Now given the different incidence of Income tax and FICA this means a lower income worker would benefit more by the reduction of FICA by .5% from his first dollar than he would from some theoretical increase in income tax to fund that transfer from the GF. But enough to offset the guaranteed increases proposed?

No one really doubts that you could close the proposed 1.7% gap by some combination of tax increases and benefit cuts and Warshawsky's plan does that up front. But in this scheme where do PRAs come in?

This is where the water gets deep. The proposed FICA tax cuts in provisions 1 & 2 total 1.26% of payroll. Hurrah for workers! But the guaranteed cuts in benefits in provisions 3 to 8 equate to 2.17% of payroll. Which means workers are .91% behind right from the get go. Plus you add in whatever their share of that additional .5% of GF transfer. Then you get the whopper, Warshawsky proposes to change the benefits for everyone by an ADDITIONAL 1.46% of payroll under provision 9, meaning the worker is guaranteed a combination of 2.37% plus of average cuts in the face of a gap now scored at 1.7%. But wait, some of that money was steered into private accounts, surely most people will make up the gap from the equity premium. Won't they?

Well maybe. If the economy grows at a rate that allows for assumed returns and if you are willing to take higher levels of risk some people after 2029 or so get a better deal overall. But not everybody and not anything is guaranteed. The actuaries put it this way
The personal account annuity replaces the smallest portion of the reduction in the scheduled benefit for the married couple with only one earner. The annuity would fall somewhat short of covering the PIA reduction for one-earner couples retiring at 65 in about 2030 or earlier. For single workers and two-earner couples retiring after 2039 with low career earnings, however, this approach would generally be expected to provide an overall increase in retirement income.
Translation: Boomers and most Gen-Xers get less net than they would be leaving the system alone.

This is by no means a complete analysis of the plan, more like a skim, and for those with the chops I encourage you to dig in. But I just don't see how the average worker really benefits under this plan given the risk involved. The benefit cuts are guaranteed, the gains from the PRAs are contingent. Plus we haven't even examined the NELB component, can they really get these projected PRA yields at Intermediate Cost assumptions?

Thursday, September 11, 2008

Intergenerational Equity, Unfunded Liability and Selfish Boomers

The newest buzzphrase in the Social Security world is 'Intergenerational Equity'. It is indeed the theme of the new movie IOUSA (to whose webpage I link) which itself is pretty much a documentary of the Concord Coaltion's Fiscal Wake Up Tour. (The fact that Concord was founded by Pete Peterson and rights to distribute the IOUSA film are in the hands of the PGPF: Peter G. Peterson Foundation not being a coincidence at all.)

One of EconomistMom's (who is a/the chief economist at Concord) first posts was entitled The Young People Get It which in turn was plugging the Youth Entitlement Summit 2008 in turn sponsored by Americans for Generational Equity an organization first founded in 2006 and funded by the usual group of conservative foundations.

The idea isn't new exactly, in Googling around today I found this lengthy article from Sept 2003 that probably explains it better than I can (I have only read the first page so far) Generational equity, generational interdependence, and the framing of the debate over social security reform. But before I turn this one over to a discussion of that let me highlight one thing.

If you go to the IOUSA webpage and look for the first part of the description of the overall issue you find it framed as follows (bolding mine) :
I.O.U.S.A. boldly examines the rapidly growing national debt and its consequences for the United States and its citizens. As the Baby Boomer generation prepares to retire, will there even be any Social Security benefits left to collect? Burdened with an ever-expanding government and military, increased international competition, overextended entitlement programs, and debts to foreign countries that are becoming impossible to honor, America must mend its spendthrift ways or face an economic disaster of epic proportions.
Sure they go on to talk about military spending and foreign debt but the discussion ALWAYS starts and mostly ends with Social Security and equally ALWAYS with a dig at Baby Boomers.

Having run into this particular article I want to post this now as Part 1 and get some discussion going on the overall topic and then return to the theme of 'Social Security Crisis = Selfish Boomers' in a latter post.

Sunday, September 07, 2008

Unfunded Liability Bookended

In the last installment of this Social Security series we kind of dug into some of the details of unfunded liability, what it was and what it wasn't and most importantly where the incidence occured: in the past or in the future. Backwards Transfer is Back. In the course of that I think it became pretty clear that none of that liability really was the consequence of overpayments to what Social Security calls 'past participants' that instead it was all due to a gap between future cost and future income for 'current participants'. But in the course of that discussion the role of 'future participants' fell through the crack when instead the numbers have some surprising implications. But before getting to that I want to back up and consider Unfunded Liability more broadly.

Traditionally Social Security has judged solvency over the short term (10 years) and the long term (75 years). Which seems reasonable enough, 75 years being a period that will capture the retirement years of pretty much any current worker. The metrics of solvency were typically expressed as percentage of payroll or percentage of GDP. In 2003 the Reports introduced new measures of solvency which expressed the gap between income and cost in Present Value dollars over the 75 year window and indeed over the Infinite Future. The first set of numbers just divides this into periods: first 75 years, 75 years to Infinite Future, and total and expresses it in dollars.
2003: $3.5 trillion, $7 trillion, $10.5 trillion
2004: $3.7 trillion, $6.7 trillion, $10.4 trillion
2005: $4.0 trillion, $7.1 trillion, $11.1 trillion
2006: $4.6 trillion, $8.6 trillion, $13.4 trillion
2007: $4.7 trillion, $8.9 trillion, $13.6 trillion
2008: $4.3 trillion, $8.3 trillion, $13.6 trillion
What do these numbers tell us? Well not much really. Generally you would expect the unfunded liability over the first seventy five years to increase simply because of normal population growth, we will have more people overall in year seventy-six than we do in year one, as we drop the latter and add the former we can expect an uptick in liability which right now is about .06% of payroll. The relatively small changes from 2003 to 2004 and 2006 to 2007 can be explained in this way. On the other hand the bigger movements from 2005 to 2006 or 2007 to 2008 turn out on examination to be the result of additional changes in assumptions, in the first case in assumed interest and in the latter changes in assumptions about immigration. But other than that the numbers don't really give us much guidance except perhaps to wonder why the future numbers from year 76 on, representing as they do the Infinite Future are not even larger. To get insight into this we need to move to a more granular analysis. Which comes under the fold.

The Table numbers vary a little bit between reports with what was Tables IV.B7 and IV.B8 in earlier Reports become IV.B6 and IV.7 but all are titled Present Values of OASDI Cost Less Tax Revenue and Unfunded Obligations for Program Participants[Present values as of January 1, 2008; dollar amounts in trillions] People who read the last post may remember that the Trustees break down "Program Participants' in kind of an odd way:
'Past participants' would seem to be that group of people who formerly drew benefits but no longer do. In short the dead.
'Current participants' are defined as everyone fifteen and older and so include all current workers and current retirees.
'Future participants' are defined as everyone under fifteen plus those not yet born. The table assigns dollar figures to these groups as follows:
Row 1 = "Present value of future cost less future taxes for current participants" (Over the next 100 years)
Row 2 = "Less current trust fund"
Row 3 = "Equals unfunded obligation for past and current participants" (Note that in this case the contribution of past participants is likely positive overall)
Row 4 = "Plus present value of cost less tax for future participants for through the infinite future"
Row 5 = "Equals unfunded obligation for all participants through the infinite future"
This result can be expressed as an equation. So what does it look like over the same period as above? (Figures in trillions)
2003: $11.9 - $1.4 = $10.5 -$.0 = $10.5
2004: $12.7 - $1.5 = $11.2 -$.8 = $10.4
2005: $13.7 - $1.7 = $12.0 -$.9 = $11.1
2006: $15.1 - $1.9 = $13.3 + $.1 = $13.4
2007: $16.5 - $2.0 = $14.4 - $.8 = $13.6
2008: $17.4 - $2.2 = $15.2 - $1.5 = $13.6 (rounding is Trustees')

What does this tell us? Well actually quite a bit. The increases in column one are mostly I think to be explained by current demographics, fewer people entering the workforce compared to the large cohort of Boomers leaving. And column two is just showing the effects of a Trust Fund in current surplus with column three being the simple sum.

But it is column four that is most interesting to me. Biggs looks at that and sees future participants paying more in taxes than they are projected to get in benefits. I suggest that is the wrong way to look at it, instead turn it around. In 2003 future participants, then defined as all people born after 1988, taken as a whole could expect their benefits to be fully funded by their taxes. Which is to say that long term Social Security is projected to return to pay-go with a surplus and that all of the problem is in fact confined to the next 100 years.

This has some profound implications for policy. Under current projections Social Security is set to pay out 78% of the scheduled benefit starting in 2041, an amount that will shrink to 75% at the end of the 75 year window. But at that point the very youngest of the 'current participants' of 2008 will be 90 and the impact of that cohort will be fading rapidly and we can reasonably expect it will go to zero right at the end of the hundred year period. Meaning that any fixes we choose to put in over the next couple of decades could be reversed later on with no damage to the long, long term outlook for Social Security. In actual practice there is no way we could limit the impacts of any medium term fixes to the current batch of current participants, some of the earlier cohorts of future participants will no doubt be called to sacrifice along with existing current participants. But there is a light at the end of that long tunnel. And if over the next couple of decades we can beat the current economic projections and so reduce the growth of column one and three we can make that light brighter and brighter.

In the light of the above equations Social Security's unfunded liability is more akin to a fixed term mortgage than an infinite burden. We can and should take some efforts to pay it down quicker while knowing that given regular payments it goes away some time in the next hundred years anyway. As the post title notes, that liability is effectively bookended.