Monday, June 2, 2008

Social Security and Global Warming

Arnold Kling posts on the comparison at EconoLog, discussing a post from Megan McArdle's blog. (As usual, Kling is thought provoking and worth reading.) At any rate, this reminded me of some past thinking comparing the Social Security issue to how people think about global climate change. I discussed this as part of a luncheon speech to the 2007 annual conference of the Retirement Research Consortium sponsored by the SSA. The whole text is available here and includes some discussion of the politics of reform that might be worth reading, but in any case I've excerpted the global warming parts below.

The short story is that Social Security (and broadly entitlement) reform and global warming are similar in three ways: first, the costs are potentially large, but also uncertain; second, the costs differ significantly across time, requiring us to make judgments about how to treat different generations; and third, the costs of policies today are inevitably passed on to future generations.

What I would like to do here is put forward two additional areas in which the existing toolbox available to economists may help policymakers better understand Social Security financing, its impact on the public, and the choices we face in coming years.

Interestingly, both of these areas derive from a methodological parallel between Social Security policy and the debate over whether, how and when to confront the challenge of global climate change.

Policy questions over Social Security and other entitlements are in important ways similar to questions of how to tackle global warming. But for those with their political antennae up, there is a certain irony in how climate change and entitlement reform are treated in the political process.

Stereotypically, at least, those on the right claim there is an imminent crisis in entitlements that should have been solved yesterday, while arguing for caution regarding global warming, and waiting until a consensus is reached. Those on the left do the opposite, mentioning the long time horizon and considerable uncertainty regarding Social Security projections but citing the precautionary principle in arguing for immediate action on climate change. Now, it may be that both sides are right – or wrong, for that matter – for reasons going beyond the similarities between the issues. But it is worth pointing out those similarities since the two fields may inform each other.

Both climate change and entitlements are a relatively benign issue for current generations but potentially severe for future ones, with considerable uncertainty regarding the potential effects. These lead to both philosophical and technical questions, which the people in this room are among the best equipped to inform policymakers and the public.

For anyone who follows debate over the economics of climate change, last year there was considerable discussion over the role of the discount rate in the U.K.'s Stern Commission report, which concluded there were large costs to global warming and large gains from averting it. A typical person hearing about this would assume the conclusions derived principally from scientific projections regarding how temperatures would change or how sea levels would rise. Someone in this room, on the other hand, would probably just mutter "low discount rate."

And, in fact, these results were generated in large part from utilizing a low discount rate of around 1.4 percent for intergenerational welfare comparisons, and much debate ensued over whether that was the proper discount rate. What's important here is that few typical policymakers, much less ordinary people outside this room, realize the dominant role the discount rate can play in these very long-term calculations.

For Social Security's internal finances we utilize the trust fund interest rate, which seems appropriate given the important legal role the trust fund plays. And in cases where we're actually transferring assets or debts over time, it also makes sense to think in terms of market rates.

But for calculations balancing intergenerational well-being – which, after all, is what Social Security reform ultimately comes down to – it's not clear to me that the government bond rate is necessarily more appropriate than the rate of wage growth, per capita GDP growth or some such measure. This seems to me to be an area that's both ripe for academic investigation and potentially very useful for presentation to policymakers. People in Washington, and throughout the country, have a gut feeling that entitlement reform comes down to balancing the well-being of your grandparents and your grandchildren, but lack a rigorous framework within which to think these questions through. I'm not saying that we should send every American an index card with the Ramsey formula on it and ask them to fill in their pure rate of time preference and return to Washington for tabulation.

Nor would such a formal framework guarantee success. After all, economists do have a rigorous framework and still they disagree. But they have a much clearer idea about what it is they disagree about, and better knowledge on how these disagreements determine their conclusions. This can only help as we think more about these problems.

There's a second way in which Social Security projections resemble those for climate change, which is the considerable uncertainty inherent in both. A literature survey undertaken by the Congressional Budget Office showed a wide range of estimates for the potential effects of global warming, with the possibility that the net effects on GDP would actually be positive. Of course, there's also the possibility of a – literal – meltdown scenario.

Likewise, as I mentioned previously, since 2003 the Social Security actuaries have included a stochastic analysis of system financing in the annual Trustees report. While some seize on this uncertainty as a reason not to act, in fact uncertainty isn't an argument for delaying action so much as an argument for acting even sooner. There is an insurance value in protecting against an unwanted outcome, even if the chance that this outcome will occur is small.

Remember, the Trustees don't project that either there will a Social Security deficit of roughly 2 percent of payroll or there won't be a deficit at all. Rather, the Trustees projection is the median outcome, with about as much chance the deficit will be double the current project as that there won't be a shortfall at all. So in our policy minds, outcomes in the worst 1 percent of the distribution should play more heavily than those in the best 1 percent.

Finally, like climate change, Social Security policy is effectively permanent. It is difficult to undo changes to the global environment, which is why advocates argue that action should be taken immediately. Likewise, the below market returns to future generations are a function of subsidies to past and present ones. While technically we can change the benefit formula however we wish, we cannot change the underlying financing constraints of a pay-as-you-go program which effectively "invests" in the growth of aggregate wages via labor force growth and productivity increases. The decisions we make about the treatment of participants in the present and near future constrain the treatment of cohorts in the more distant future, just as choices we make about the environment today may dictate outcomes for generations who will follow us.


Bruce Webb said...

Do respective consequences matter?

Global warming: potential 20' rise in sea level putting New York underwater and eliminating Bangladesh altogether.

Social Security 'crisis': a 22% cut in a check meaning a real benefit not as much better than expected compared to that which current retirees get today.

Pretty much every bit of evidence that has come in over the last ten years is that global warming is real and if anything accelerating. On the other hand every bit of evidence that has come in over the last ten years shows that Social Security 'crisis' has been both shrinking and receding out in time.

Conceptually you can draw some parallels but once you reduce this to the data level the comparison becomes a little specious.

Andrew Biggs said...

A valid comment, but...

First, you're comparing a worst-case scenario with global warming to the intermediate cost estimates for Social Security. It's important to be consistent between the two. Moreover, the basic question for both is, how much would you spend to avoid these worst case scenarios? Probably a lot.

Second, the cost comparisons are hard to make conceptually, but it's important to make them on the same basis. When expressed as a percentage of GPD (eg, lost production, costs of prevention, etc.) the mid-case costs of global warming aren't huge for countries like the US. Plus, they would take place a very long time from now, so the assumed discount rate becomes extremely important. My guess is that if you discounted Social Security's long-term deficits at a 1% interest rate, as was used in the Stern Commission report, they'd be extremely large. said...


As one of those naughty people who thinks we should do nothing (for now) about social security, but at least something about global climate, I wish to push Bruce Webb's basic argument a bit further.

First of all, you are simply wrong about changes to social security being "permanent." They are not. They can be changed at any time, although, of course such changes may be politically difficult at any time. That is why it is not that big of a deal to sit a few years and see whether IC or LC is more likely, with plenty of time to do something by 2017, if in fact the more pessimistic forecasts prove likely (and, of course, for reasons that have been stated here previously, I do not buy into the official estimate of the likely stochastic distributional projections).

The second point is that for serious students of this, and I have been involved with climate modeling for over three decades, including close association with the most credible of the "global warming skeptics," that the issue is not the discount rate. That is certainly a garden variety way to look at it, but I am more in the school of Weitzman.

In contrast to social security, where a Gaussian normal distribution may not be unreasonable for the projections over the next half century or so, it is unreasonable for climate projections. The global system is full of nonlinearities and potentially destabilizing positive feedbacks, with these being confirmed by the geological record, notably that the movement of the world in and out of ice age periods occurred very rapidly on geological time scales, probably less than a century going each way. So, what we are dealing with looks more like financial market asset returns (rather than the real economy and labor income returns), namely characterized by kurtosis and "fat tails," with a serious danger of a large fat tail in the direction of far more serious global warming than the median forecast, truly catastrophic outcomes, much harder to offset, and with prevention requiring actions with a much longer time lage, certainly far longer (and harder to implement) than the essentially instant feedback/turnaround available for changing social security if it really goes sour.

So, the precautionary principle is very relevant for the climate change issue in a way that it is not for the social security issue. This lumping together of the two is a sort of game that Washington insiders who like inane consensuses can use to make wisecracks to each other at clubs and institutes and seminars, but off the wall in reality.

Barkley Rosser

Andrew Biggs said...


On the first point, the issue of permancy of changes isn't to do with the tax/benefit formulas, which we can change at any time. It's to do with the distribution of generational burdens. Early participants in Social Security received far more from the program than they paid in, which implies that later participants must receive less than they put in. That's permanent. Now, going forward, the longer we go on with scheduled taxes and benefits, which imply higher net returns than can be maintained over the long term, the lower the net returns to post-reform generations. If our job in reform is to spread the legacy debt from past/present participants as evenly as possible, putting off reform means larger burdens for the future. That's the sense in which I meant permanent.

Second, I understand there's more to the discount rate issue than just the cost of capital, etc. As I understand it, the Stern Commission's use of a low discount rate was implicitly to adjust for the risks of worst-case scenarios. Weitzman makes the point, as you do, that with global warming we don't have a good picture of the shape of possible distributions of outcomes, and so it's hard to ascribe probabilities to different outcomes. All that is true, but it seems that Weitzman and Stern implicitly assume a distribution without specifying it, and then derive a discount rate or course of behavior from it. That seems fishy to me.

But in any case, uncertainty regarding future outcomes means there's an insurance value in preventing them, even if they're quite unlikely. That holds for both global warming and Social Security. As for uncertainty, that's the basic point I was making.


Paul Lawin said...

If I were going to compare some policy issue to global warming, I would pick “capital accumulation” rather than Social Security.

The current generation really does face a trade-off between producing for current consumption and building physical, technical, and human capital to raise consumption for future generations. When we make that decision, the “discount rate” (which is just the importance we give to consuming ourselves vs. the next generation consuming) is very important.

Furthermore, the decisions we make today are irrevocable. If we accumulate too much or too little, the future generation can’t get into a time machine and come back to change our decisions.

And, our projections are definitely problematic. When we’re trying to decide how much effort we should put into capital accumulation, we don’t know how effectively we’ll use that effort. Will we get huge payoffs? Or will we discover that we invested in all the wrong things?

Social Security has none of these characteristics. As an income transfer program, it simply splits current consumption between current producers and former producers. There is no inter-temporal trade-off, everything happens in this year. We can change our decisions every year. There is no need for long term projections when making this year's decisions.

The SS financial projections are useful for informing current workers about the ease or difficulty that future workers will have in providing their benefits. That may help current workers decide how much private savings they will try to accumulate. It might also help us all decide whether we should expand or shrink SS over the long term.

But these projections have nothing to do with “real capital accumulations”. It’s unfortunate that we’ve let the trust fund build up to more than one year’s benefits, because it seems to be confusing the issue.

Since Social Security can influence decisions about capital accumulations, we should recognize some interactions when we make SS decisions. But, we shouldn’t confuse the two. They have more differences than similarities.

Paul Lawin said...


I posted my comment before I saw your reply to Barkley. I can see that if we want to calculate an IRR for a cohort, the taxes they paid as workers will be part of that calculation. Furthermore, once paid, those taxes can’t be changed. So in that sense, the current year tax decisions can’t be changed.

But I don’t understand your enthusiasm for this IRR calculation. The taxes my generation will pay depend on how many children our parents had, how productive we are, and how generous we are. The benefits we receive will depend on how many children we have, how productive they are, and how generous they are. I can calculate an IRR, but once I have it, I don’t know what to do with it. I can’t defend any number as the “right” IRR, so I don’t know how that calculation would impact policy decisions.

Lord said...

Earlier generations may have had a greater return on SS, but that hardly means they had it easier. They also raised more children, developed a greater infrastructure, and otherwise incurred more costs. Similarly, boomers may have a lower return, but raised fewer children, developed less infrastructure, and otherwise incurred fewer costs. SS is much more amenable to solution, it is only about money after all. Increased borrowing over a transition that can be implemented as needed is a small price to pay. There is little intergenerational about it. Any change would have to persist in the future.

Climate change is much riskier because damage may be irretrievable and money may not be able to fix it. We have little to estimate costs and benefits. Only by trying will we be able to arrive at what is possible, what is realistic, and what is not.