Tuesday, June 17, 2008

Rivlin and Kingdon: Tackle Social Security First

Alice Rivlin of the Brookings Institution and John Kingdon, professor emeritus at the University of Michigan, write that Social Security reform should be a top priority for the next president.

The next president and new Congress face a daunting set of challenges come January 2009: Iraq war, troubled economy, global climate change, looming government debt, taxes, health care reform and rebuilding infrastructure, all vying for immediate attention. Such a long "to do" list presents two possible tactics: tackle the hardest problem first or get the easy ones out of the way. We prefer the latter and would start with Social Security.

With the large baby boom generation retiring and Americans living longer, the ratio of workers to Social Security beneficiaries is falling fast. Quite soon the payroll taxes coming into the Social Security system will be inadequate to pay all the benefits promised to retirees. Restoring the system's solvency is crucial to almost everyone's retirement security. The longer we put off setting Social Security on a firm footing, the more expensive the fix will be - indeed, we have waited too long already.

Fixing the Social Security program is relatively simple - far less complex than reforming health care. It requires choosing a combination of revenue increases and benefit reductions, but relatively small changes will do the trick. Emotions run high on which combination is best, but the list of options is short, and political compromise is clearly within reach.

Dealing with Social Security would build rebuild confidence in our elected leaders and give both Democrats and Republicans a legitimate chance to take credit for solving a major national problem. The new president would achieve a major objective, and congressional leaders would score a win as well. That accomplishment would build the momentum and relationships needed for tackling more difficult national problems that require pragmatic cooperation across party and ideological lines.

Of course, everyone has to compromise to accomplish a comprehensive Social Security reform. Republicans have to give up diverting existing revenues into private accounts. But they can preserve private accounts on top of Social Security and strengthen incentives for individual retirement savings without going to "privatization."

Democrats have to accept future benefit cuts, but they need not be drastic and can spare current retirees and lower-income beneficiaries. The package could include future gradual increases in the retirement age and concentrate benefit cuts on higher income people.

As part of the compromise, both parties must agree to revenue increases, but they, too, can be modest and can spare low-income individuals. For example, the cap on income subject to social security tax can be raised in gradual steps.

We know that nothing as politically charged as Social Security can be fixed easily. But saving Social Security is crucial and requires bipartisan agreement. It must be done sometime, and the sooner the better. Let's show that government can function, by solving the Social Security problem early in 2009.


Alice M. Rivlin has directed both the Congressional Budget Office and the Office of Management and Budget, and has served as vice chair of the Federal Reserve. John W. Kingdon is professor emeritus of political science at the University of Michigan.


Steve Selengut said...
This comment has been removed by a blog administrator.
Bruce Webb said...

Opinions vary over whether Social Security is facing a 'crisis' and in particular whether that 'crisis' is defined by a cut in benefits come 2041 or by a drag on the economy by financing the gap between income and cost by repaying the excess dollars borrowed from 1993 (sic) and 2017. And that is fine, everyone has a right to their own opinion. But you don't have a right to define reality. The following is just not supported by the historical evidence. It might be true going forward but it is a false statement looking back, and is particularly egregious because some form of it has been repeated year in and year out for the last decade:

"The longer we put off setting Social Security on a firm footing, the more expensive the fix will be - indeed, we have waited too long already."

The cost of not acting last year was a reduction in the payroll gap from 1.95% to 1.7%.

We have two choices vis a vis the Social Security Reports. One, we can accept that this is the good faith effort of a group of committed professionals to present the best informed projection available and just go with the current year number, or we can nitpick about why 2006 showed an increase in payroll gap from 1.92% to 2.02% while 2008 showed a decrease from 1.95% to 1.7%. If we take the former approach, which I believe is the position of the host here, then what Rivlin said is nonsense, the cost of inactivity since 1997 has left us a less expensive fix going forwards, we are currently down from a 2.23% gap to 1.7%. Will this progress continue? Hard to say. Will the gap increase enough next year to offset the savings (in terms of dollars retained) of not acting this year? Well maybe, though my calculations suggest not. But should we simply accept the simplistic formulation of "We can't afford to wait"? Well I don't see it. We can in fact take this matter one year at a time. If you use real world numbers (and in an economics policy question why wouldn't we?) a policy of Nothing has proven to be a winning bet.

It may in fact be true that the cost of fixing Social Security will be more expensive if we wait an additional year. But to say "We have waited too long already" is simply to ignore the historical record.

Andrew G. Biggs said...

Here's a thought experiment for you. Consider what YOU think the future actuarial deficit will be. As long as that deficit is positive the cost of delaying reform is positive. You may think the deficit is smaller than the Trustees project, others may think it's larger, but either way the deficit will be bigger next year than this year. I guess you can argue that there simply won't be a future deficit -- that Low Cost is certain to happen -- but that's a very, very minority position.

Anonymous said...

Bruce, you are confused.

1. The actual deficit did not go down or up duringhte period you describe. The projections vary from year to year, but the actual deficit is not yet realized. Thta may argue for gradual policy change rather than all at once changes, but if you are hoping the problem can be assumed away you are deeply mistaken.

2. What changed between 2007 to 2008 was primarily a change in the way they project that you could argue should be applied retrospectively to preovious projections. From that perspective, we are in the same place in 2008 as we were in 2007. We think the level of the deficit is smaller than we previously thought, but there is no change in trajectory. You seem to think the "improvement" memans if we wait long enough it will go away, which again is wrong.

3. Why is it younever comment onthe likelihood that the projections are way too optimitics when it comes to brithrates and life spans? You only talk about whether econoic growth will be better than projected?

Finally, the size of the deficit over a given time horizon does not change if we wait (given any single projection). It is exactly the same. What changes is the magnitude of the changes that need to occur in the time remining. We either raise taxes 1.7% today or twice that in 30 years. One can argue that gradual changes are better than sudden ones, and that waiting to implement reforms means only future generations have to pay the inevitable cost.

Anonymous said...


Here is a thought experiment for you. Assume some future level of benefit cut or tax increase is required in th future. Putting that future tax increase or benefit cut on the books today, to be in effect in 2017 or 2041 doesn't make it any cheaper than if it were passed on December 31, 2016 does it.

Great blog by the way.


Anonymous said...

Oh and I'd much rather see the problems at Medicare tackled before OASDI. The problem with Social Security are realtively minor and easy to deal with when comapre to Medicare, where there are no easy solutions.


JG said...

If you don't solve the problems you can because there are problems you can't, you'll not solve many problems in life.

In fact, you'll never solve any of them.

Bruce Webb said...

Well anonymous I can only use the numbers supplied by the Trustees. In the 2007 Report they suggested that the unfunded obligation over the Infinite Horizon was $13.6 trillion, while the gap through 2081 was $4.7 trillion Table IV.B6.-Unfunded OASDI Obligations for 1935 (Program Inception) Through the Infinite Horizon That same table from the 2008 Report shows Infinite Horizon unchanged at $13.6 tn but the gap through 2082 down to $4.3 trillion over that single year.
Table IV.B6.—Unfunded OASDI Obligations for 1935 (Program Inception)
Through the Infinite Horizon[Present values as of January 1, 2008; dollar amounts in trillions]
Given that the change in valuation period generally adds .06 in payroll equivalent all in itself I have to see this 15% improvement in cash outlook over the Long Term Actuarial (75 year) window as proof that we were right to not move on this in 2007, the cost of a 100% fix went down from 1.95% of payroll to 1.7%.

These are not my numbers. I don't have independent numbers to start with. Instead they are the official numbers produced by the Trustees of Social Security. If you think they are skewed somehow I am sure former Deputy Commissioner Biggs knows someone in the Office of the Chief Actuary and can pass your corrected calculations along.

If you accept the numbers of the official Reports at face value then there is no question that the cost of inactivity since 1997 has been negative. Money left behind in taxpayers' wallets, no need to adjust the retirement age, and smaller fix going forward. As I say I am not a number cruncher, instead I am a number pointer.

As to why I never comment on the birthrates and lifespans. Well a couple of reasons. One no one has made any convincing specific argument that the mortality rates are overly optimistic, only vague suggestions that they might be. Well I am no more a demographer than I am an economist and have to presume that the professionals in the Office of the Chief Actuary know what they are doing. Two changes in fertility will have no particular impact over the 10-15 years that will be decisive, those kids won't be hitting the workforce until after this matter has been decided once and for all.

The only numbers really available to me are the year over year changes from one Report to another, and those changes vary much more for the economic numbers than they do for the demographic, 2008 being an exception in that regard (big move due to changes in immigration assumptions).

Bruce Webb said...

"Finally, the size of the deficit over a given time horizon does not change if we wait (given any single projection). It is exactly the same."

Well there is an interesting philosophical argument buried there, the future indeed will be what it will be, in some respects setting policy for the out years is like exploring a dark room using a rope. But using the best tools available to us the size of that deficit did in fact shrink 15% over a single year. Best available information suggests we can afford to wait another year, and then probably the year after that.

It may well be that Low Cost outcomes are indeed at the edge of the probability band, on the other hand if they actually do come about it turns out that tax increases or benefit cuts are not only not needed but actually counterproductive long term. If you like you can join the ongoing discussion at Angry Bear, #22 in the Soc Sec series went up yesterday and is mostly about this same matter.

But your position boils down to assuming that something will have to be done when that is not clear at all. From the perspective of current utility of a worker's payroll dollar doing too much too soon may be as bad or worse as doing too little too late.

Arne said...

I have looked at mortality and fertility.

The IC projection has life expectancy increasing linearly forever. This seems patently ludicrous to me. There has to be a limit. The historical data for women already shows fails a decreasing rate of increase. Even though I believe that the mortality data is pessimistic, I would support a change that increases the retirement age - as long as it was tied to actual data showing people living longer.

If fertility comes out at LC, it eliminates over half of the cash flow shortfall (and that is below the 55 year historical average). IC cost has fertility continuing to drop even though it has gone up in the last decade.

Where I work our marketing folks are required to forecast how many units we are going to sell. The fact that they are professionals at forecasting does not seem to seem to improve their accuracy. My guess is that we will come out about halfway between IC and LC.

The message I would prefer is that ther is a good chance that we will need to make adjustments to SS in the next 20 years. We should be prepared to settle on some numbers about 10 years from now. In the mean time, we better get to work on health care and Medicare, or the big problem will make the time we spend on the little problem a waste of effort.

Andrew G. Biggs said...


Your comment re whether we assume potentially unlimited lifetimes is a good one, although I believe that most demographers now don't assume any particular limitation on how long people can lives.

Second, in practice it doesn't matter much here. The Intermediate mortality assumptions are that it will take until 2053 for US life expectancy at birth to equal Japan's today. That's pretty modest, I think, and we know a life expectancy of 83 is possible because Japan and a number of countries already have it. So in practice I don't think it matters too much.

Arne said...

I've heard some health care folks predict that the rise in the incidence of obesity will cause life expectanies to go down in the future.

I notice you quote the life expectancy at birth where I tend to look at the life expectancy at 65. Just convention - or is there a reason to prefer one over the other?

In order to maintain the same ratio of working years to retired years you need to raise the retirement age by 1 year every 30 (IC) to 80 (LC) years. In 1983 they overshot by raising it by 2 years over 23 years. I am ready to agree that retirement age should be increased as life expectancies increase, but I am not willing to set the age for people who are not born yet. Nonetheless, I would think it useful to have a formula that could be used for the 75-year forecast. A thumbnail view suggests that keeping the ratio of working years to retirement years constant woud be about the same as the difference between IC and LC mortality, so from Appendix D, it would solve about one third of the (IC) shortfall. Does that sound about right? Is there a model that someone like me could use to test such assumptions?

Andrew G. Biggs said...


I used life expectancy at birth just because it was easier to find the cross-country data. I expect there's still a pretty big gap at 65, but don't know the numbers offhand.

The sensitivity analysis re mortality (http://www.ssa.gov/OACT/TR/TR08/VI_LRsensitivity.html#92688) states that "Each additional 0.1‑percentage-point increase in the average annual rate of decline in the death rate decreases the long-range actuarial balance by about 0.13 percent of taxable payroll." Now, the average annual reduction in mortality is projected at 1%. So this implies that if mortality remained the same as today going forward, this would reduce the 75-year shortfall by around 1.3% of payroll. It's a little more complicated than that, but that's a pretty good estimate of how much rising life spans is contributing to the shortfall.

Arne said...

Thanks Andrew, but what about a scenario in which the retirement age is adjusted with observed increases in life expectancy. How do I figure that out?

Andrew G. Biggs said...

if you eliminated the current hiatus in raising the retirement age, then once it reached 70 continued at the pace of 1 month every two years, that would eliminate about 0.7 percentage points of the 1.7% deficit. It's not intuitively obviously how this relates to previous thought experiment I did regarding mortality declines. I suspect the difference may be due to mortality improvements for current beneficiaries, whose benefits wouldn't be affected by raising the NRA.

AK: Sorry for not responding to your question. You're right that if we need to fix a cash deficit of x percent of payroll in year y, it doesn't matter whether we announce it today or announce it later. On the other hand, if we need to fix a deficit of $x trillion (around $15 or so), the longer we wait to reduce benefits or cut taxes the larger those changes will be on the affected cohorts, simply because delay exempts today's cohorts from any changes.