Many interesting comments on my suggestion that science could bail out social security. First, Andrew writes,

Social Security pays retirees a benefit based on their pre-retirement wages, targeting a ‘replacement rate’ rather than a real benefit level. If GDP grows faster so do their wages, all other things equal. Higher wage growth raises payroll tax revenues but, with a slight delay, also raises the benefits SS must pay. As a result, Social Security’s finances are remarkably insensitive to wage growth.

Quite right. Productivity gains lead to higher benefits for recipients as well as higher wages for workers. However, with really rapid productivity gains, workers might be able to stay ahead. For one thing, there could be strong “vintage effects” of human capital, so that as people get close to the retirement age, their wages level off, while wages of young workers rise.

Also, if productivity is high, then the standard of living for everyone can be high, even if the elderly’s share of the pie is disproportionately large. Still, the case for indexing benefits to prices rather than to wages (thereby de-linking benefits from productivity) is very strong. This would be a subtle yet effective way to scale back Social Security benefits and improve the viability of the system.

Finally, the program that really blows up under conventional extrapolations is Medicare. Medicare costs are not linked to productivity growth by any mechanism. So greater productivity would reduce the burden of Medicare.

It was pointed out that if technology increases longevity, but we keep the retirement age constant, then this will increase the burden of entitlement programs. That, too, is quite right. I have argued that the failure to index the retirement age to longevity is a major reason that the entitlement programs threaten to blow up.

On his web log, Steve Verdon writes,

Suppose we could construct a technology index and set say, 1968=100. What would the current value of the index be? 200? 300? 600? Has human productivity increased by that much? No.

This is reminiscent of Robert Solow’s famous comment that “We see computers everywhere but in the productivity statistics.” That was in 1987. Now, we are starting to see them in the productivity statistics. The problem is that until recently, the technological component of the capital stock was small. Thus, even large increases in quality meant very little to the economy as a whole in 1987. They mean more now.

Go through an exercise in which one sector of the economy grows at one percent per year and another sector grows at 20 percent per year. Start with the fast-growing sector being really small–say 0.5 percent of the economy. At first, the the growth rate of the economy as a whole is pretty much the same as the slow-growth sector. But after a number of years, the fast-growth sector becomes more and more important, and it starts to influence the overall growth rate significantly. This is a highly nonlinear process. The influence of technology on productivity can be too small to detect for a while, then it becomes noticeable, and then a few years later it becomes tremendous. One could argue that today we are somewhere in between “noticeable” and “tremendous” today.

Returning to the comments on my post, several people mentioned the law of diminishing returns. Technology optimists reject that law. They argue the opposite–that innovation is a positive feedback process. Ray Kurzweil uses the expression “law of accelerating returns.” Better computers enable us to design better computers. Powerful computers enable us to carry out nanotech and biotech research. Nanotech and biotech will have positive feedback on one another, etc.

Another comment is that humans may not adopt technology quickly or effectively. I think that could be a significant source of friction in the process of turning technological change into economic growth.

Since the original thread seems to have gotten a bit off the subject, continue the discussion here.