Arnold Kling  

Creative Non-Destruction

NRO is 90 Degrees Off... A Parallel Fallacy...

Amar Bhide writes,

Over 30 years after the introduction of minicomputers and more than 20 years after the introduction of microcomputers, the mainframe remains an important category. Total worldwide revenues of large-scale computer processors (or mainframes) amounted to $16 billion in 1997 compared to $16.2 billion in 1982. But because total demand grew from $38 billion to $183 billion, mainframes’ share of the total computer market dropped considerably, from 42% to about 9%.

He argues that new products and services tend to satisfy new wants rather than immediately displacing existing goods. He says that this complicates issues of macroeconomic theory and policy.

The new want machine, which has an excellent record, does not however create jobs at exactly the same rate as increases in efficiency and outsourcing, or cyclical downturns, causes traditional industries to shed them...

No one can predict when new industries will start adding jobs faster than old industries shed them. Standard macro-economic policies cannot speed things up. Tax cuts and easy money might stimulate ‘old’ economy demand for automobiles and housing, but they cannot overcome the unwillingness of U.S. consumers to use Short Messaging Services on their cell phones.

Thanks to Timothy Taylor in the Journal of Economic Perspectives for the pointer. The Summer 2005 issue of that journal is filled with articles on institutional economics, behavioral economics, and freakonomics. For those of us who believe that the mathematical paradigm has gone well past the point of diminishing returns, this journal seems to offer a better way forward.

Comments and Sharing

CATEGORIES: Growth: Consequences

TRACKBACKS (2 to date)
TrackBack URL:
The author at The Club for Growth Blog in a related article titled Friday's Daily News writes:
    Louisiana Looters in Congress - David Keating, NTU Not One Dime More - Daniel Glover, National Journal Human Action and Human Design - Don Boudreaux, Cafe Hayek Is Hillary The Next Geena Davis? - John Fund, WSJ Bush’s Blank Check... [Tracked on September 30, 2005 8:46 AM]
The author at The Reconstruction in a related article titled People and Curves writes:
    ...the sentiment of many economists that the returns from the mathematical paradigm within economics are diminishing. It's expressed in the point that mathematics are valuable as a tool, and surely the tools should be further developed. The problem man... [Tracked on October 2, 2005 10:53 PM]
COMMENTS (4 to date)
Bruce Cleaver writes:

I can also add the following, at least with respect to computers:

At every small, medium, and large scale company I have worked at, there has been a concerted effort to do away with the mainframes (the 'legacy systems') and consolidate all functions under one machine architecture - naturally, the latest one purchased, since it has the most features.

It never comes to fruition. The cost and effort of rewriting all the daily, weekly, and monthly processes is just too much. Half the programs are black boxes to their maintainers, and truly ferreting out all the sources and their interactions and dependencies with other processes is fantastically time and resource consuming. It turns out to be cheaper to simply let the new functions accumulate on the new machines.

My experience, YMMV.

Timothy writes:

That sounds about right. See also, the bank I work at still uses Office 97. That's right 97, I was 14 what that was released in the fall of '96.

Robert Schwartz writes:

I wanted to book a trip to Chicago so I called Wells Fargo, but they said they are a bank and that the stage coach line closed when the railroad was put in. I called the C&O, but they said they did not deal in less than truckload lots. ...

Schumpeter's discussion of the the devaluation of physical capital by technological innovation included the example of that the railroads had done to the capital of the coach lines.

The computer industry is a bad example. All computers are physical instantiations of a mathematical idea, the Turing Machine. Every computer can do what every other computer can do, just maybe not as quickly.

triticale writes:

Every computer can do what every other computer can do, just maybe not as quickly.

True as a generality about processors. Far less true about computer systems. Mainframe architectures produce central superservers capable of managing huge databases for multiple clients - they were optimized for this long before desktop microprocessor systems were for geeks to experiment with.

Comments for this entry have been closed
Return to top