Asymmetrical Information - Megan McArdle

01.07.13

You're Doing it Wrong: Social Security Life Expectancy Calculations

The trust fund runs out even sooner than you think

Whenever I point out that old-age entitlements are costing us a lot of money, I get pshawed by folks who point out that the Social Security trust fund doesn't even run out of money until 2033.  Leave aside the arguments over whether the Trust Fund makes Social Security safer (short answer: no); Gary King and Samir Soneji argue that even this estimate is too optimistic.  They say that the actuaries at the Social Security Administration are not taking adequate account of improving longevity trends like the decline in smoking.  

I find this entirely plausible, not because the government bureaucrats who calculate this stuff are stupid, but because institutions like the Social Security Administration are inherently conservative in their methodologies for calculating figures like this.  These estimates are invariably politicized, and one way to minimize the political fallout is to make your calculations as mechanical as possible.  Witness the pushback that the census has gotten whenever it's suggested using advanced statistical techniques to estimate the "missing" people who don't get counted on the census because they live unstable lives.  

Let's say they're right: what does that tell us?  Well, it's part of a larger pattern, which is that the date of the trust fund's exhaustion keeps creeping ever closer.  And I don't just mean because time is passing.  When I started writing about Social Security in 2002, the trust fund was projected to run out in 2041.  Now it's projected to run out in 2033.  If King and Soneji are right, it will run out in 2031.  Which means that over ten years, the projected time-to-exhaustion will have fallen by just about 1/2, to 18 years.  

Moreover, it might well fall still further.  With the exception of the dotcom boom, all of the trust fund surprises have been in the wrong direction.  And I'd argue that this is no accident.  


Here are the things that produce upside surprises:  

1. Massive economic booms

2. People delaying retirement

3. A lot of people suddenly dying unexpectedly  


Now here are the things that produce downside surprises:  

1.  Slower-than-historic-average economic growth

2.  People retiring early (often because of slower-than-expected economic growth)

3. People living longer than expected  


Which of these sets of things seems more likely over the next ten years?  The much-discussed recent study from the Journal of the American Medical Association seems to indicate that Social Security will not be bailed out by our growing national avoidupois.  And all the other trends I can think of favor longer lifespans, not shorter ones.  

Of course, I might be surprised.  We might get a global pandemic that kills a bunch of elderly people, or a vogue for later retirements might suddenly develope.  Predictions are hard, especially about the future.  

But of all the projections we make, Social Security finances are among the easiest. Demographic change is like watching the tide come in: no one drowns at high tide precisely because you have a long, long time to see it coming.  

But demographic change is like the tide in another way: once it's commenced a-rising, there's no way to turn it back.  The 45-year-old peak earners we need to support aging baby boomers in the next ten years already haven't been born; we cannot go back and have them now.  It's largely too late to make the investments that could have boosted their productivity and thereby made it easier for them to support more retirees.  And because immigration is largely a youth phenomenon, we can't really import them, either.  

The tide eventually recedes, of course--but in its own time.  The demographic structure of the United States for the next decade or so is already largely baked in.  And because Social Security benefits are indexed to wages and inflation, so is the likely trajectory of the system's finances.  

It doesn't actually matter very much in exactly what year the nominal Social Security trust fund is "exhausted".  But it does matter very much if retirees are going to be living longer and exerting a greater burden on current workers.  King and Soneji's work argues that they will.  The day when we have to come to some sort of reckoning with our old-age entitlements is rapidly approaching.