> ...That there really weren't all that many people aged 65-87 in the
1950s.
> ...That the baby-boom produced a lot of now-40 year olds--a hump in the
age
> distribution.
> ...That the elderly need more services (especially medical services).
> ...That as a result the effective dependent-to-worker ratio is rising.
and John K Taber responded:
> Brad, are you confusing life expectancy in the 50s with the proportion
> of 65-87 in the population? Life expectancy if I remember correctly was
> about 62. BUT, that's at birth. Each year you make it, your life
> expectancy improves. I have seen the proportion of aged to the
> population, I don't remember the figure, but it was significant in
> 1950 and earlier years.
Personally I would bet a case of Lagavulin that professor DeLong is familiar with the concept of life expectancy. But in any case, why not bring some facts to bear on these questions?
Population (millions) year under 20 over 65 total dependency ratio 1950 53.9 12.8 159.4 .719 1960 73.0 17.3 190.1 .904 1965 80.1 19.1 204.0 .946 1970 80.7 20.9 214.8 .898 1975 78.4 23.3 224.5 .828 1980 74.6 26.1 235.1 .749 1985 73.2 29.1 247.1 .706 1990 75.1 32.0 260.0 .701 1995 79.0 34.3 273.3 .708
The last column is the ratio of those under 20 or over 65 to those in between. As you can see it peaked in 1965. And according to the Social Security Trustees Report, from whence these figures come, the total dependency ratio will never again reach 1965 levels, even under the insanely pessimistic assumptions of their "high cost" scenario.
That's not the end of the story though, because as DeLong points out old dependents are more burdensome than young ones--one estimate, which seems as reasonable as any, has under-20s consuming 72% as much as adults, and over-65s 127% as much as adults. From these figures you can construct a weighted dependency ratio, which rises 11-12% by 2060. Even weighted by consumption, though, dependency ratios never regain the 1965 high.
I'm taking all this from an interesting article by Lawrence Summers and some other folks, which goes on to say, "It is difficult to know whether these estimates represent a large or a small burden spread over 70 years. They correspond to between a 0.10 percent and 0.15 percent reduction in the annual productivity growth rate, which is small relative to the uncertainty in secular productivity growth. They represent three to four times as large a cost as the peace dividend that the United States is likely to enjoy over the next decade [the article is from 1990]. ... A three- to four-year increase in the average age at retirement, or a 19 percentage point increase in female labor force participation [which would bring it even with men's], would ... offset the increase in dependency."
What all this means for Social Security is a different story, of course. Summers and his coauthors conclude, through arguments I sort-of follow--a smaller workforce requires a smaller capital stock; other industrial countries are aging even faster, and therefore will be investing in the US; slower population growth is associated with faster productivity growth--that demographic shifts indicate a lower, not higher savings rate, and in particular that "population aging does not constitute a strong argument for accumulating a large social security trust fund."
So I have to side with Taber and Pollak and against DeLong: Though the ratio of workers to dependents is not constant, the main arguments based on a presumed increase in the dependency ratio are false both in their premises and their conclusions.
Josh