Friday, January 23, 2009

New paper: “Prefunding Social Security Benefits to Achieve Intergenerational Fairness: Can it Be Done in the Social Security Trust Fund?”

Randy Mariger of the Treasury Department is the author of "Prefunding Social Security Benefits to Achieve Intergenerational Fairness: Can it Be Done in the Social Security Trust Fund?" I'm rushing for a flight and so haven't read it yet, but it seems well worth a look. Here's the abstract:

Being fair to future generations requires that Social Security be reformed in a manner that prefunds a significant share of future Social Security benefit payments. All serious reform plans have this property. Prefunding is done exclusively in the Social Security trust fund in some plans, and it is done partly in personal retirement accounts (PRAs) in others.

The consequences of prefunding Social Security in the trust fund are controversial and not well understood. The key question is whether Social Security surpluses are offset by smaller non-Social Security surpluses; if they are, and if the offset is 100 percent, then Social Security surpluses are not truly saved and prefunding intended to make Social Security fair to future generations is neutralized by a non-Social Security fiscal policy that is less fair to future generations. This paper makes this important point concrete by simulating the response of non-Social Security fiscal policy to two alternative Social Security reforms that differ only with regard to the breakdown of prefunding in the trust fund and prefunding in PRAs. The reforms simulated are the Nonpartisan Reform Plan proposed by Jeffrey Liebman, Maya McGuineas, and Andrew Samwick, and a version of that plan that prefunds exclusively in the trust fund. If Social Security surpluses are not saved, it is found that NRP's PRAs increase the net benefits of government to future generations by about 0.6 percent of GDP; that is, future generations enjoy some combination of lower non-Social Security taxes and higher non-Social Security government spending that amounts to about 0.6 percent of GDP in every year.

The paper also reviews budget politics over the past 30 years and concludes that there is a substantial probability that trust fund accumulations are largely offset by reduced non-Social Security surpluses.

The implications of these findings for Social Security reform are explored. If budget politics precludes the possibility that Social Security surpluses are saved, then large dividends would be paid if an alternative means of effectively prefunding Social Security could be found. If politics also precludes that possibility, then it would be rational to compromise other Social Security reform objectives so as to reduce trust fund accumulations. Specifically, relative to a first-best reform with effective prefunding, smaller benefit levels would be appropriate.

We've looked at the issue of "saving the surplus" here.

Also, it seems important to me that Mariger examines the Liebman, McGuinea, Samwick plan, which is aimed at fiscal responsibility. One of the issues with the Bush reform proposal, and many other personal account plans, is that the "transition costs" associated with personal accounts would be financed with new borrowing. When that happens, personal accounts no more pre-fund future benefits than does the current trust fund structure. The devil is in the details. I'm likely to add more later once I have the chance to read more carefully.


Paul Lawin said...

The paper begins with this assertion: "Being fair to future generations requires that Social Security be reformed in a manner that prefunds a significant share of future Social Security benefit payments. All serious reform plans have this property."

I scanned the paper and didn't find any explicit support for this statement. "Fair" may be the most over-used word in political debate. It seems that much debate arises from different definitions of that word.

I think it would be good for the author to explain what "intergenerational fairness" means when different generations have distinctly different fertility rates.

Bruce Webb said...

There seems to be a logical problem in the last paragraph cited. If the issue is that current Social Security surpluses are not truly being saved and that politics precludes any real pre-funding meaning we should seek to "reduce trust fund accumulations" the answer is not to reduce benefits up front. In the absence of any other changes all that does is increase the existing (not really saved) surplus. If you truly want to put the Trust Fund on a trend towards its mandated TF ratio of 100 there are only three avenues to doing so: raise benefits, cut payroll taxes, or invest cash surpluses into other asset classes. My personal preference would be for the third choice, just not in the form of PRAs which to me simply add administrative friction.

In any event under Intermediate Cost assumptions this problem is fast going away, by 2017 we won't have any surpluses to not-save to start with.

Andrew G. Biggs said...


Good point. I think what they're taking as 'fair' is that each generation get roughly the same deal out of Social Security (meaning same average rate of return, etc.). While I more or less agree, you can easily argue the opposite: currently, we feel fine if rich folks get a worse deal than poor folks. In the same way, we could argue that future people, who will be significantly richer than people today, should get a worse deal than today's participants. Since it's all zero sum, you have to trade off between them.


Good points. I think I did a post somewhere (if not, I should) pointing out that the surplus between now and 2017 is around 5% of the total shortfall (the $13 trillion or whatever it is now). While it would be good to save the surplus, it would be very hard in practice and maybe not worth the political capital it would cost to do it. You might get more for your political effort by focusing on changes to taxes, benefits, etc. Hard to say.

Bruce Webb said...

I find that people at both ends of the political spectrum but probably more so on the left vastly overestimate both the size and the role of the SS surplus and the Trust Fund itself. While $2.4 trillion is a big number (and even more so $13.6 tn) when you start looking at how the former actually accumulated on a year by year basis and how the latter is actually projected to come due on that same basis the numbers come back to earth. When I hear people on the right saying that Johnson introduced the Unified Budget so that Social Security could hide the cost of the Great Society program or people on my side of the aisle claiming Reagan funded his military buildup by raiding SS I have to respond "Dude you are not really talking about enough money here". You don't have to go to motive, either man or both might be conceived to have those malign motives. But you can't steal what really never existed.

I am far from being a fan of Ronald Reagan, but the numbers show that he was a very fine steward for Social Security, as was Bush I. Collectively they took TFs that seriously flirted with zero in 82/83 and delivered them to Clinton with a combined ratio of 100. The much ballyhooed raids never happened.

Bill Woessner said...

I agree that "fair" is a loaded word. There are 300 million people in the United States. Good luck coming to a consensus on what's "fair".

What we can talk about is generational equality. But that ship has pretty much already sailed. If you look at the data from the CBO's most recent update on Social Security, you'll see the payable ratios are in steady decline.

It should be noted that the CBO's data only goes back to people born in 1940. But if we go back to 1874, we would see the real inequality. Ida May Fuller famously collected $22,888.92 in benefits but only paid $24.75 in taxes. Adjusted for inflation, her payout ratio was close to 50,000%. In comparison, I can expect about 64%. Yay.

Andrew G. Biggs said...

Good point. A lot of the generational inequity is baked in the cake now -- those folks are gone and there's nothing that can be done to smooth things out. (I posted on this issue here:
But there is still a disparity in treatment between current beneficiaries, who get an ok deal, and future ones who get a pretty rotten one. Acting now on reform can help equalize treatment between these cohorts.

Paul Lawin said...


I can understand “equal IRR”. Part of my concern is replacing that descriptive term with the normative “fair” without notifying the reader. The other part is that in doing so Mariger ignores the fertility issue.

Consider generation A, their children B, and grandchildren C. Suppose A had 3 children per couple, but B had only 2. Why would it be “fair” for B to expect the same retirement replacement ratios as A? B had extra discretionary income because they supported fewer children. Suppose the extra income went toward cruises and SUVs. What should gen B expect from gen C?

Maybe the real issue is that SS is on a course which will provide higher IRRs for B than for C, D, E, … If that is the concern (this seems more likely from your response to Bill than from anything I can get out of Mariger’s paper), it seems the only option at this point is to lower gen B’s consumption, whether by reducing SS benefits, increasing some other tax, or…? You say that “acting now can help equalize…”. You may have explained how you would do this in a prior post, but I’ll confess that I don’t recall your recommendation. Could you help me out?

Andrew G. Biggs said...

A rigorous definition of 'fairness' requires that we be explicit about how we weight the welfares of people of different incomes/family types, etc within a given cohort, and how we weight the welfare of future cohorts relative to today's. It's generally taken for granted that we value future cohorts less than present ones, but by how much? So your point is correct -- 'fairness' involves a lot more than stipulating equal rates of return, or whatever.

Acting today would generally mean increased taxes or contributions to individual savings accounts. That would lower IRRs for current generations, but reduce funding burdens on future generations as well, thereby raising their IRRs. The problem is, it's current generations that make the decisions, and they don't want to give anything up. This was a real problem with the Bush personal accounts plan -- in the end, they wanted to borrow all the money that went into the accounts, which defeats the purpose.

Paul Lawin said...

I think I understand the concept. If we lowered both taxes and benefits for 2010 (only) by equal amounts, that would certainly reduce the SS IRR for currently retired individuals and raise it for current workers. (Some of that transfer might be offset by private transfers in the other direction between generations.)

But most SS "reform" proposals start by saying "no benefit reductions for current retirees or anybody born before 19XX. The numbers matter here. Is there a generation born after 19XX that is getting "too much" from SS? If so, how big a decrease would we need?

Andrew G. Biggs said...


Good comment. And you're right that most reform proposals partially defeat the purpose by taking people 'off the books' -- saying that they won't be affected. This obviously limits how big the trades between generations can be.

While, on average, each succeeding generation does worse under Social Security, there's no clear birth cohort that marks a dividing line. Optimally, you'd redistribute between all cohorts in which you could (reasonably) redistribute, meaning all living ones. By exempting all current retirees, and all people over 55, etc. you're reducing the pool. You can still smooth things out, since (say) 54 year olds will get a better deal from the system than 24 year olds, but you're not getting as much as you could.

Paul Lawin said...


I'm not so clear on "each succeeding generation does worse". Don't we hit a point where the IRR becomes the total growth rate in covered wages? If we freeze taxes at the current level, which has been in effect for about 20 years, and make the program pure paygo, then all cohorts born after 1968 (approximately) should get the same deal.

I suppose the complication is the current trust fund balance. Under current law we'll burn it in the next 30+ years, so cohorts that get to use it up get the better deal.

Maybe your idea amounts to spreading the trust fund forward to reduce the cliff. I've thought of restricting spending from the trust fund to only the real interest, therefore forcing benefit cuts sooner and making the trust fund last a long time. I can't recall a specific proposal from you.

(Sorry for the slow response, I've had travel problems.)

Andrew G. Biggs said...

You're right. While financing continues to get worse over time due to longevity, individual treatment (IRRs, etc.) hits a steady state where average rates of return equal the growth of aggregate wages (labor force growth plus wage growth, more or less). I don't know exactly when this takes place, but could figure it out.
You're right that teh trust fund exhaustion complicates this. But the key point is that current retirees are getting an IRR of around 3% while the long-run steady state IRR is around 1.5%. By reducing benefits for current/near retirees we can sweeten the deal a little but for those in the future. Not a ton, but it's something.

Paul Lawin said...


I agree with your comments, I'm disagreeing with Mariger. He says that "fair" requires more prefunding. I'm thinking that "fair" is going to happen anyway, the primary complication with getting there is the pre-funding that we've already done.

From a political perspective,

Paul Lawin said...

Oops, I hit "publish" instead of "preview".

... From a political perspective, it seems the biggest acceptable decrease would be to prevent the older cohorts from using up all the trust fund. If we use just the real interest on the trust fund, benefits would start decreasing in 2022 (roughly). The youngest current retirees would be 74 then.

A younger cohort, like the 1970 BY, is currently expected to feed off the trust fund until they are 71, then have a sudden drop in benefits. This change gives them smaller benefits from ages 62 through 71, partially offset by a smaller drop. The 1980 BY currently gets no benefit from the trust fund, this change gives them some benefit in their retirement years.

Granted, this isn't perfect. But it's the best idea that I've seen. It prevents the cliff in benefits, puts the IRR on a shallower glide path, and is as politically feasible as anything. Note that it doesn't require any additional prefunding.