Arnold Kling  

Long-term Growth Forecasts

Sowell on Math and Economics... The 1990's Bubble Economy...

What will economic growth look like in the 21st century? Mark Bahner solicited answers to this question from many economists and pundits, but only a few replied. Bahner's predictions are relatively optimistic. He foresees worldwide average GDP per capita rising from $6500 in 2000 to $100,000 in 2070.

In my opinion, any prediction of world per-capita GDP growth that's less than the 2.2% per year that was the average for the 20th century is probably too low. The 20th century had 2 world wars, a Great Depression, and a great many countries mired in communism for most of the century. I don't expect any of those things for the 21st century.

Bahner also believes that if one accepts Ray Kurzweil's extrapolations of Moore's Law, with computer power continuing to double every two years or so, then economic growth has to accelerate. I agree that if we take Moore's Law seriously then economic growth will reach staggering levels in the next few decades.

My own prediction was even more optimistic than Bahner's.

For Discussion. Will improvements in computer power continue to translate into faster overall economic growth?

Comments and Sharing

COMMENTS (14 to date)
Boonton writes:

No, the phrase that is missing here is 'diminishing marginal returns'. 'Computing power' is an input to producing goods and services. Like any other input it is subject to diminishing marginal returns. Being able to run MS Office twice as fast does not mean a business can cut its workforce in half or produce twice as much revenue.

What will happen if Moore's law continues to hold is that goods and services that are easily produced through computing power will become much less expensive. After that, though, the increased computing power will become difficult to translate into higher economic growth.

In some ways I think this is already happening. It is getting very hard to justify spending $300 per person to upgrade everyone to the latest version of MS-Office when Office 2000 appears to be handle 98% of the tasks one needs it to handle.

David Thomson writes:

“It is getting very hard to justify spending $300 per person to upgrade everyone to the latest version of MS-Office when Office 2000 appears to be handle 98% of the tasks one needs it to handle.”

What in heavens name does software improvement have to do with per se computer power? The latter will most certainly makes us a far wealthier nation. I am utterly convinced that less expensive computer power will enrich us in ways that we are barely able to currently imagine.

Steve writes:


As computers become faster, more human-like taks will be performed, replacing the need for humans, lowering costs, increasing profits, etc. Moore didn't just talk about PCs (although he did work for Intel) running Windows. We're talking about everyone having a Deep Blue (tm) ("AI" chess playing IBM box) equivalent for less than $500 someday soon only with the ability to do everything you do, including potificate on econlog. ;-)

Of course, replacing humans on a wholesale basis destroys demand, which is what the entire capitalist society is based on. What we will do when this occurs is hopefully something my children's children will have to figure out.

Don Lloyd writes:

If tomorrow, alien contact produces computer power of infinite speed, zero size, weight and cost, would anyone believe that this would result in a measured benefit for an honestly measured economy with constrained growth of both population and money supply?

Economic progress, in terms of an improving standard of living, has little or no necessary connection with normal financial measurements.

Regards, Don

Scott writes:

One of the other things Moore’s Law gives you is larger and faster databases. As applications catch up with the ability to gather, store and manage data, businesses will make faster, better informed decisions.

Boonton writes:

Question: While 'computing power' continues to increase, what evidence is there that it is being implemented now? It seems getting the fastest chips are among the last variables most businesses consider when they need to make IT investments.

Steve writes:

Very true, Booton. However, when software is written with the express purpose of replacing well-paid humans in the way that prior killer apps have (Office, Acrobat, Lotus 123, Quicken), you can be assured that Intel will be unable to keep up with demand for faster chips.

MSOffice replaced the need for a secretary.
Acrobat replaces typesetters.
123 replaces low level actuaries.
Quicken replaces low level bookkeepers.
What will be written to replace Economics professors? :-P

Witchfinder General writes:

"Will improvements in computer power continue to translate into faster overall economic growth?"

In a word, no.

If computing power follows along the same path as lighting,(1) as his evidence suggests,(2) then (barring hedonics) its contribution to GDP should dim. So the longer answer then depends on what use that computing power is employed within the larger economy.
(1) "William Nordhaus (1997) has analyzed the real price of light: how much it costs in the way of resources and labor to produce a fixed amount of artificial illumination, and has found that the real price of light has fallen by a thousandfold over the past two centuries. A middle-class urban American household in 1800 would have spent perhaps 4 percent of its income on illumination: candles, lamps, oil, and matches. A middle-class urban American household today spends less than 1 percent of its income on illumination, and consumes more than a hundred times as much artificial illumination as did its predecessor of two centuries ago.

"Yet we do not speak of the 'illumination revolution,' or of the 'new economy' generated by the existence of exterior streetlights and interior fluorescent office and store lights. The productivity of illumination-producing technology has increased enormously, but its impact on the economy and on society has been limited. Demand has not grown rapidly enough to offset falling prices. The total share of illumination in total urban spending, and thus the share of illumination production in the urban economy, has shrunk. Our artificial illumination technologies are an enormous boon and source of value--Nordhaus (1997) believes that it has contributed seven percent to the growth of real wages over the nineteenth and twentieth centuries--but its economic salience has been limited."

(2) "William Nordhaus (2002), "The Progress of Computing" (New Haven: Yale University xerox), heroically ventures where angels fear to tread and constructs estimates of the falling price of computation: trillionfold in the past 60 years: 35 percent per year compounded continuously. A halving time of 2 years."

Lawrance George Lux writes:

I agree with Boonton in believing that Computer power, in and of itself, has functionally peaked in generation of additional economic growth. I foresee this Century to bring advance in Materials--both Research and Production. It will be the Century of Design materials, vastly expanded from Drugs and Chemicals. A Plastic, or some such, will replace both timber and steel. The most promising seems to be Compound interlocking bonds, but supercompressions may be the answer, or both.

The Health issue will be the great element of the new Century. I related, some books back, a belief this Century would be one of Declining human populations, leaving the Century at the approximate level of Population We entered the Twentieth Century. I believe We have yet to see the first of the unstopable plagues, only potentials, like AIDS and SARS. These are the ones stopped only by development of immuune resistence.

Per Capita Income will be slow (below 2% in the first third of the Century, too many of both Population and Problems). There will be spurts thereafter. The overall PCI by the next Century should be about $27000 Worldwide. lgl

Cap'n Arbyte writes:

Drat, I got here late and Boonton and Don Lloyd beat me to the most valuable points. :)

The primary benefit of the continuation of Moore's Law will be the new applications that will become practical -- things like designing drugs or super-materials from scratch. Existing applications like word processing have already hit diminishing returns. (Not speaking for my employer, etc.)

While better computers will enormously improve the standard of living, this won't translate into higher GDP figures because GDP is (still? mostly?) a nominal, not real, measure of production.

Mats writes:

>>Will improvements in computer power continue to translate into faster overall economic growth?>>
Yes, but its marginal contribution will decrease, as did once the dramatic and highly important contributions from the textile industry, (global) transport industry, steelmaking etc. Technology in these industries still improve, but with little impact on the overall economy.

Tim Swanson writes:

Computing technologies like CPUs require swaths of capital, long-term capital. There is another law called "Rock's Law" which states in part that the cost of building a semiconductor fab plant doubles every four years.

Ceteris parebus, the regulation and intervention in the business world will not make the production and development of devices like CPUs any cheaper and in fact can and will slow down its development and deployment.

Individuals like Kurzweil continually miss one important factor in their analysis of "the future": the State. It doesn't shrink and in fact increases every year, some more than others.

Additionally, GDP is a poor indicator for economic health because it incorporates State spending which inflates the percentage growth. Ex: if the State began an effort to build 1000 cement skyscrapers in every large city or spend hundreds of billions of dollars, GDP would increase.

Back to Moore's Law, I believe it is more of a case of happenstance than an economic law -- the ability to reinvest and research has outpaced regulations thus far. And I do not see regulations/intervention decreasing anytime in the near or distant future.

Marina writes:

The process of upgrading the computer soft- and hardware is endless. However, once there will a great fall in the world economy, primarily caused by the disappearance of most of the human proffessions.

Mark Bahner writes:


I didn't see this until yesterday (Septmber 18). I was thinking about economic growth as a result of a Tech Central Station article about the levels of economic growth projected by the IPCC (Intergovernmental Panel on Climate Change).

The IPCC apparently projects an overall per-capita world GDP growth rate of less than 2.2 percent for the 21st century, i.e. less than in the 20th century. Way, way low, I think.

In the course of my thinking, I became even more certain that economic growth will be astounding in the 21st century. I think Arnold Kling's predictions on my weblog are more realistic than mine...EXCEPT for the next decade or so. There I think he's a little too optimistic.

A couple people here have questioned why computers should be different than other technologies that have dramatically advanced (e.g. lighting, textiles, steel). The difference is that computers think (like the human mind). That's why they're different than any other technology.

I'll post an update on my own blog. Sometime. ;-)

Comments for this entry have been closed
Return to top