December 19, 2004
Ask Dr. Social Security Projections
Matthew Yglesias has some questions for Dr. Social Security Projections. But first, the answers:
- The current low-cost estimate has the 75-year balance at +0.41% of taxable payroll; the intermediate estimate is at -1.89% of taxable payroll; the high-cost estimate is at -4.96% of taxable payroll. The long-run economy-wide productivity growth rates for those three projections are 1.9%, 1.6%, and 1.3% per year, respectively. The crossover point--the 75-year balance of zero--currently comes at a productivity growth rate of 1.85% per year.
- The average productivity growth rate over the 40 years from 1960-2000 was 1.76% per year. Add in 2001-2004 (unless the fourth quarter of 2004 is really bizarre) and find that the average productivity growth rate over 1960-2004 is 1.85% per year. Up until the late 1990s there was a reason to project future productivity growth to be slower than the long-run past average: productivity growth did seem to be gradually slowing down. Since the late 1990s that is no longer the case: there is no longer any good reason to project that future productivity growth will be slower than in the past (unless you believe that we are going to fail to undo the Bush policy mistakes, that is).
- You shouldn't try to compute averages by calculating trough-to-trough numbers. Different business cycle troughs are very different in their characteristics. By contrast, while all peaks are not identical, they are closer along a number of dimensions like capacity utilization and upward pressure on inflation than are troughs.
- The 44 years since 1960 give you an average growth rate of 1.86% per year; the 39 years since 1965 give you 1.69% per year; the 34 years since 1970 give you 1.66% per year. Moving forward: since 1975, 1.58%; since 1980, 1.70%; since 1985, 1.73%; since 1990, 1.91%; since 1995, 2.37%. Personally, if I had to give an estimate of what productivity growth will be over the next 75 years, I'd guess 1.9% per year because it seems to me that 1960-2004 includes a uniquely bad period in 1973-1989 or so. I'd be really surprised if the productivity growth rate averaged less than 1.6% per year. And I'd be really surprised if it averaged more than 2.3% per year. But that probability distribution is worth exactly what you paid for it.
Here are Yglesias's questions:
Matthew Yglesias: Questions, Questions: I was talking to Mark Schmitt and Henry Farrell the other day about the blogosphere's potential as a mechanism for distributed policy analysis, and the SSA's economic assumptions page raises all sorts of questions that for reasons of either mathematical ignorance, ignorance about where to obtain the correct historical statistics, or simple time constraint that I can't work on. If you think you're someone who can do a little math and would be interested in helping to pierce through the web of suppositions and alleged crises, read on.
The first thing that would be nice to calculate if someone can figure out how to do it is the exact average productivity growth over the next 75 years that we need to make sure that the Trust Fund is never exhausted. We know that under the "low cost" projection, the fund stays solvent forever, but the low cost scenario is actually lower in cost than needed. Where's the tipping point? I can't begin to understand how one would do the requisite math.
The other set of questions I have in mind right now pertain to the point I made here about how they estimate producitivity growth. This is done by breaking up the past into a series of economic cycles measured peak-to-peak (with the most recent peak in 2000) calculating the average annual producivity growth for each cycle, and then averaging together the results for the last four cycles. That's obviously a pretty arbitrary method. I'd be interested in knowing what you get if you try some different methods. What happens if you use the last four business cycles measured trough to trough? Or what if you use the last five (or six or...) business cycles? What happens if, instead of using the double-averaging method, you just take average annual productivity 1966-2000? How sensitive, in other words, are the long-term projections of doom to minor changes in how the analysis is done? Has the SSA just happened to pick the most pessimistic possible way to do this? That'd be quite the coincidence.
Posted by DeLong at December 19, 2004 11:43 AM
TrackBack URL for this entry:
Listed below are links to weblogs that reference Ask Dr. Social Security Projections:
» The subject of Social Security reform (UPDATED) from The Glittering Eye
There have been quite a few posts in the last week or so on the subject of Social Security reform. These posts have varied considerably in tone and content based on the interests, knowledge, politics, and point-of-view of the writers.... [Read More]
Tracked on December 20, 2004 04:11 PM
» Productivity and Social Security from The Dead Parrot Society
Matthew Yglesias asks, "The first thing that would be nice to calculate if someone can figure out how to do it is the exact average productivity growth over the next 75 years that we need to make sure that the Trust Fund is never exhausted." Brad DeLon... [Read More]
Tracked on December 21, 2004 03:16 PM
» Victor Does the Heavy Lifting from Vox Baby
2.8 percent, with some caveats. In a comment on a recent post, and in more detail on The Dead Parrot Society, Victor tries to answer this request: Matthew Yglesias asks, "The first thing that would be nice to calculate if someone can figure ou... [Read More]
Tracked on January 4, 2005 12:14 PM
Tracked on January 5, 2005 01:09 AM
I'm just reading a biography of one of my heroes, Glenn Gould, and I find on about page 412 that he hated JFK and rather liked Richard Nixon.
This is rather my attitude, although I am a dyed in the wool lefty. I first learned to sing the Red Flag during some by-election of 1944 when my mother was pulling me over the dales in a milk box on a sled. (Yes, we won the by-election: the score for the National Government during WWII was 0-7.)
A Democratic Party which prides itself on its twee policy accuracy,but cannot relate to the fact that Richard Nixon was, with one or two blatant exceptions, a really fine drinker, is a donkey going nowhere.
In a post here a few months ago I boasted that we had Mississippi and Louisiana under control, and Alabama within sight. I was wrong.
Still, the notion that the Democratic party is the party of the guys with the 'Fed flags on their Ford 150's, and their female friends, is the key to the Democrats' future.
A straight-out appeal to the American working class (Hunh? America doesn't have classes, does it? A Schlitz to the guy who comes up with a good catchy way of saying "working clas" in American English) is the key.
The gravy is that one-payer health insurance and retirement benefits financed out of general revenue will both help America's manufacturing competitiveness.
Damn, how do you paragraph stuff under this new regime?
Posted by: David Lloyd-Jones at December 19, 2004 12:05 PM
>The current low-cost estimate has the 75-year balance
> at +0.41% of taxable payroll; the intermediate estimate
> is at -1.89% of taxable payroll; the high-cost estimate
> is at -4.96% of taxable payroll.
That's very nice: three decimal places of utter garbage. As a physicist I learned never to trust numbers without error bars, and this is why. These numbers are about as good as today's weather prediction for Christmas of next year.
[Do you have a point? These are, roughly, the Social Security actuaries' estimates of the 90 percentile, 50 percentile, and 10 percentile of the probability distribution. That's useful information to have.]
Posted by: Douglas Davidson at December 19, 2004 12:37 PM
>I'd guess 1.9% per year because it seems to me that >1960-2004 includes a uniquely bad period in >1973-1989 or so.
Why uniquely bad? Why so sure that over the course of the projected 75! years there aren't going to be 16 particularly bad ones (especially given your scepticism about the Bush policy consquences, the uncertainties about shifts in international division of labour and demography, still unanswered questions about the truth in growth statistics, the ramifications of a generally higher (certainly perceived) uncertainty than in the period mentioned)?
Posted by: Tobias at December 19, 2004 01:38 PM
If this country were run by the right people -- meaning by labor; meaning by the majority -- there would be nothing critical about this discussion. If Social Security tax rates needed to be raised 15-30 years out, they just would be; no hassle. No big deal to raise a few points of tax on many more points of income (due to per capita output growth).
The current problem is that we need to spend all day every day keeping the bad guys from tricking the nation out of doing what is right. Not a problem in Europe. Big difference: labor unions rule there but is non-existent here. Only come back?: somebody has to make a big issue of mandating labor unions (my answer to everything) -- only plausible way to stop doing the treadmill for the truth.
Posted by: Denis Drew at December 19, 2004 02:26 PM
That would be the Europe of near-zero economic growth, stagnant incomes, and 10% unemployment, right, Mr. Drew?
Posted by: Spanky at December 19, 2004 06:15 PM
Tobias, you are missing the framing here. Sure the future is by definition uncertain, but we have to start somewhere, and past trendlines is the place to do it. The question remains, given likely economic growth is the Social Security Trust Fund over or under funded? The discussion of privatization pivots decisively on that point. Advocates insist that there isn't even a discussion to be had on this point, it is broken and doing something is better than doing nothing. They are trying to sweep two questions under the table. One, "Is it broken" and two "Assuming the economic projections that show it is broken can you produce historic returns on equities".
The second question is hard and results in all kinds of difficult technical questions about the varying yields of stocks and bonds and the effects of channeling huge flows of money from one to another. Grist for the mill of people with PhD's from Harvard and academic posts at UCB, people who don't blanch in the face of equations. But all a little over the head of this former medieval historian.
But the first question is easy and anyone who reads the business pages can grapple with it. Is the economy going to grow at a rate of 2.0% or better? If it does Social Security is fixed and the discussion is moved to an entirely different location. Privatizers are transformed from arguing "something is better than nothing" to "this specific proposal is better than this agreed upon something". Privatizers are desperate to avoid this question, because they would then be reduced to to arguing that we should replace a guarantee with a contingency. And they simply can't deliver.
I simply propose that we move the discussion from "something is better than nothing" to "something is better than something" and then let each 'something' make their best case.
Posted by: Bruce Webb at December 20, 2004 04:25 AM
Bruce you are on the right track, but take it a step futher.
Keep asking the republicans why they are projecting that their economic policies are going to fail.
Keep asking them why they are projecting such strong growth in their budget documents and such weak growth in their SS plans.
Posted by: spencer at December 20, 2004 05:37 AM
Brad wrote, "Up until the late 1990s there was a reason to project future productivity growth to be slower than the long-run past average: productivity growth did seem to be gradually slowing down. Since the late 1990s that is no longer the case..."
But profound changes in the way GDP (and hence productivity) is calculated were introduced inthe 1990s---to wit, hedonistic adjustments, etc.
There's good reason to question the validity of these adjustments. Without them, recent growth numbers are less rosy.
Posted by: liberal at December 20, 2004 06:21 AM
Spencer, actually I am on that one. I call it the double book deal. Privatizers are perfectly willing to work from an infinite horizon gap of 3.5% of payroll to front a problem that needs fixing but simply ignore the 1.6% future productivity number that underlies that projection. I challenged a poster who I didn't know from Adam on this very site on that very topic. Well it turned out that this wasn't any ordinary Samwick. It was a very real Professor (http://www.dartmouth.edu/~samwick/cv.html) Andrew Samwick. It even sparked an e-mail from our host, pointing out that Samwick was pulling a real number from a real table in the 2004 Report of the Social Security Trustees. Well Samwick simply ran by the relevant table in that same report (http://www.ssa.gov/OACT/TR/TR04/V_economic.html#wp159107) Economic Assumptions under the Three Alternatives that pretty much showed that 3.5% number to be total bullshit. ( I accused him of academic dishonesty, which probably will prevent me from getting a Chair in any Economics Department in the country. On the other hand my graduate work was in medieval history and I dropped out of the PhD program anyway and am making money better than I ever would teaching Introduction to World History at the local community college (which is where I was headed in all likelyhood). So I am not losing any sleep over this one.)
I popped up on this site a few weeks ago posing this simple question for privatizers: "Can you present an economic model for future growth that supports an equities based solution to the Social Security "crisis" that does not beat the numbers of the Low Cost Alternative of the Social Security Trustees' Annual Report?" And even people like Samwick, who are obviously familiar with the Report came back with nothing.
Spencer you have put your finger on exactly the right question. And I will keep asking them precisely that. Everywhere.
Posted by: Bruce Webb at December 20, 2004 06:30 AM
>Sure the future is by definition uncertain, but we >have to start somewhere, and past trendlines is the >place to do it.
I agree. And while I wasn't entirely unaware of the framing, thanks for the very concise primer :). Yet my main concern was with indeed with the "unique" - I remember charts explaining the real growth rate of industrialised countries since industrialisation to hover around 1.7%. So maybe there are reasons to believe the transactional structure of modern economies has changed to allow a changed trend in real growth, but this is apparently being debated. And so I wondered what made the 1973-1989 period so special that it would be unfair to include periods of equally weak growth in the calculation.
To add something with respect to a point mentioned by Spencer - what will happen if an even more Orwellian, less hospitable America will attract less foreign human capital than it did in the recent past? I hear the numbers of foreign students and phd candidates are dropping already.
Posted by: Tobias at December 20, 2004 02:57 PM
During the last 6 or 7 years of your slow growth period 1973-1989 corporate America poured billions into personal computer technology without notable improvements in productivity. I remember an article from Fortune magazine from the period to that effect rather vividly. Much of the growth of the last ten or so years have been the latent result of that long fallow period. Other than the persistence theory is there a reason to believe that the product of that investment will continue?
I'm not being snide here. I'd really like to know the answer.
[It's a good question. There are a bunch of guesses, but no real solid answers...]
Posted by: Dave Schuler at December 20, 2004 04:19 PM
"It is a cautionary tale for his generation in many parts of Europe, as aging societies from Sweden to Spain and across Eastern Europe as well confront a demographic time bomb: With ever fewer young people to work and pay taxes, the cost of looking after growing numbers of older people with state pensions akin to American Social Security becomes ever more prohibitive."
One should also keep in mind that the UK has partially privatized its old-age pension system with "personal pensions."
Matt's question reveals a very complicated issue - it isn't just productivity or GDP growth that drives Social Security. It also is driven by how long people live, as well as wage growth.
And there is a political issue. Right now the real rate of return of Social Security to young people is barely 1%. Increasing payroll taxes, or increasing the age of retirement will bring that real rate of return below 0%, the point where you would be better putting the money under your pillow.
At that point, will there still be political support for Social Security? Will people work to avoid paying Social Security payroll taxes through black market labor? More to the point, why even bother?
Posted by: Mr. Econotarian at December 20, 2004 05:56 PM