Tuesday, April 8, 2008

I won’t tell you what investments to pick if you won’t…

Marketwatch’s Irwin Kellner comments on the latest Social Security Trustees Report. You can read the entire column for yourself, but here I’ve pulled some key excerpts followed by commentary.

Kellner's comments are indented:

In 2000, the system’s actuaries thought the assets of this fund would be exhausted by 2032. Two years later it was 2037. Now the projected exhaustion date is 2041. Meanwhile, the Congressional Budget Office, which makes these projections as well, recently thought the system will remain solvent until at least 2052.

Two thoughts: First, the trust fund exhaustion date, while important, is also a volatile measure, since even a small change in assumptions can shift it by a year or two. Second, while CBO “recently thought the system will remain solvent until at least 2052,” they more recently amended that projection to 2046. Given that CBO uses slightly more optimistic assumptions and that their model tends to show smaller shortfalls even with constant assumptions, this is a relatively small difference.

Judging by past history, assumptions underlying the intermediate projection are very conservative -- especially when it comes to economic growth…. The intermediate projection assumes that the economy will grow by an annual rate of 2.3% per year between now and 2085… well below the 3.4% that the economy grew on average between 1960 and 2005.

Please, I’m begging you, stop these comparisons of past to projected GDP – they just miss the point. As far as Social Security is concerned, “GDP” doesn’t matter. The closest thing to GDP growth that matters for Social Security is the sum of real wage growth and labor force growth. The Trustees projected rate of real wage growth (1.1% annually) is – cue the music – slightly higher than the average from 1960-2007. What’s lower is the rate of labor force growth, but this makes total sense: from 1960-2007, labor force growth grew rapidly (around 1.7%) principally due to Baby Boom birth rates and rising female labor force participation. But birth rates have fallen significantly, meaning fewer new workers, and female labor force participation isn’t going to get much higher than it is today. As a result, total GDP growth is projected to slow, even if output per worker – a better measure of the economy’s health – continues along just fine.

And as you might imagine, the speed at which the economy grows has a lot to do with the other variables -- including the interest the fund earns from investing its surplus in Treasuries.

Actually, GDP growth is the product of the other variables, not an input to them. Moreover, the correlation between real GDP growth and real interest rates on the trust fund – around 0.18 – while positive, isn’t terribly strong.

You might ask the question why this more realistic [low cost] projection has escaped politicians from both major parties. I don't know why, but I can only theorize that it's because they haven't taken the time to read the entire report.

Alternately, it may be because they have read the entire report. The arguments here are typical of those you read in the press and see on the web, not what you hear from people who really spend a lot of time with this material.

4 comments:

Bruce Webb said...

The Baby Boom is conventionally dated from 1946 to 1964 meaning that Boomers hit the work force between about 1965 and 1985 with half of that pulse in labor force participation coming after 1975. Which makes the following a wee bit deceptive:
"What’s lower is the rate of labor force growth, but this makes total sense: from 1960-2007, labor force growth grew rapidly (around 1.7%) principally due to Baby Boom birth rates and rising female labor force participation."
Your baseline is suffering from a little time shift problem. Moreover the whole argument that the problem going forward is some shortage of workers ignores what has been the historical American response in such cases-we open the immigration doors.

Intermediate Cost immigration numbers prior to the 2008 Report were kind of a joke, there was no reason why we could, should or would have seen numbers decline so drastically from the 1.2 million figure of 2004 to ultimate 900,000.
http://www.ssa.gov/OACT/TR/TR07/V_demographic.html#wp185380
To accept this result would result in a gradual decrease in the amount of foreign born Americans over time, a rather perverse outcome given a crisis largely framed as one of labor participation.

At least the 2008 Report came to its senses a little on this front. It no longer calls for an absolute decrease of 25% for combined immigration in fairly short order (ultimate in 2030), instead replacing that with a model that phases that decrease in over a longer term (ultimate in 2065) but still suggests that over time immigrants will play a smaller and smaller demographic role in relative terms. To say the least this result is rather odd in historical context.
http://www.ssa.gov/OACT/TR/TR08/V_demographic.html#205410

And this is not quite even half right:
"The arguments here are typical of those you read in the press and see on the web, not what you hear from people who really spend a lot of time with this material."
I would defy you to show any mainstream press reporting reflecting the 'No Crisis' narrative by any reporter or columnist whose name doesn't start with a 'K-r-u-g'. The USA Today is particularly egregious on this topic but the WaPo and the NYT are not much better, they consistently follow the line of the Trustees' Press Release dutifully reporting 2017 as if that date had any particular significance once placed in the context of the entire economy and budget. This particular narrative has gained a foothold in certain sectors of the web, particularly at the center left Econoblogs, but generally bloggers are still repeating the party line as promulgated by Cato and associates.

And please it is just because GDP is a 'product' of other factors that makes it matter. Only a tiny fraction of people could possibly weigh differences in assumptions of labor force participation or real wages on models of solvency, what we do know is that once you grind all those numbers through you end up with a number that we understand and can put in historical context. You might as well argue that '8' doesn't matter because it is the product of 3+3+2.

But thanks for the link, it is nice to know that the Dean Baker message is finally starting to resonate in the wider world.

Andrew G. Biggs said...

Bruce -- You're right about the timing in the Baby Boom's effect on labor force growth. I used the 1960-2007 average because that's the total period available, but had I restricted to the key Baby Boom years the difference would be even larger (e.g., between 1965-85 the average growth was 2.2%, vs. 1.7% overall). Not sure this causes my argument much of a problem.

You're right that immigration may go higher, although some have argued to the contrary. (Mexico is aging rapidly, which may raise wages and reduce incentives to emigrate somewhat. Not sure I buy it, but it's possible.) I any case, going up a few hundred thousand per year doesn't qualitatively change the financing picture.

I should have restricted my references on the press, because you're clearly right there (though not on the web, I think). Besides Krugman, David Francis of the CSM bought into Dean's argument quite some time ago. I stand corrected there, but stand by the argument that those who spend a lot of time with this material rarely make the kinds of claims in Kellner's article.

The point regarding GDP growth is that a seemingly unreasonable GDP number can easily come about through very reasonable changes to its components. This is pretty clearly the case here, since no one seems to come up with alternate values for wage growth, fertility, immigration, etc. that can produce total GDP growth that matches the past.

Anonymous said...

I would suggest life expectancy and birth rates are enormously potent sources of uncertainty in the long-range projections. And I would posit they are more likely to make the financing situation worse. Economic performance could improve and the effect could be offset entirely if we all just live longer. The Trustees assumptions about improvements in life expectancy are quite conservative.

Andrew G. Biggs said...

Anon -- in theory the Trustees assumptions are what they consider the mid-point, but you also have to consider where their assumptions lie relative to other expert opinion.

On mortality (life expectancies), the Trustees are on the low end of the mainstream. (E.g., see this survey here: www.ssa.gov/policy/docs/ssb/v66n1/v66n1p16.pdf). Lower mortality has a big negative effect on solvency, though it has an equal affect on personal accounts (annuity prices would rise) so it doesn't decide the proper policy response.

Re fertility, the US is high relative to other developed countries, it's just a question of whether we stay where we are or revert to the mean. I expect we'll always be higher, but how much higher is hard to say.