A Couple of Figures That Will Need to Have Another Data Point Added to Them Tomorrow: Archive Entry From Brad DeLong's Webjournal

October 30, 2002

A Couple of Figures That Will Need to Have Another Data Point Added to Them Tomorrow

The share of total nominal spending (GDP) devoted to investment in information technology equipment and software. The key is that as technology has progressed--as the real price of computation has fallen--the share of GDP devoted to infotech investment has risen extremely rapidly. On top of the secular trend is superimposed the business cycle--the recessions of the early 1970s, of the mid 1980s, of the early 1980s, the stagnation of investment throughout the late 1980s as the growing federal deficit squeezed out capital formation, and then the steep high tech-led recession of the past two years.

Removing "other infotech" from the aggregate, leaving computers, peripherals, and software, sharpens the 1990s boom and arguably creates a better aggregate of "high tech": computers, peripherals, and software contains fewer telephone poles, coils of copper wire, and electromechanical switches. Curiously enough, removing "other infotech" does not increase the relative size of the recent fall in nominal investment--much of the fall in spending is a fall in spending on telecom, as we know well.

FYI, the components of the share of GDP made up by the components of nominal investment spending on information technology equipment and software:

Perhaps the most interesting thing shown by the "components" figure is that the 1990s--the decade that saw the steepest decline in the prices of computers--saw a tremendous growth not in the share of spending on computers but on computers' complement, software.

Posted by DeLong at October 30, 2002 04:58 PM | Trackback

So spending on software has increased while spending on hardware has not.

Could some of this difference be a consequence of the Microsoft monopoly in operating systems and office applications, while there is no corresponding monopoly on the hardware side (not even Intel)? It seems to me that software is a much larger proportion of the price of a new computer than it was ten years ago (but I have no figures to back that assertion up).

Posted by: Tom Slee on October 30, 2002 07:12 PM

Possible, but surely the post below ("the mythical man month") gives a much better expalanation - ie the lack of productivity growth in software manufacture.

This lack of productivity growth could itself, of course, be due to weak incentives to reduce software production and distribution costs (as distinct from incentives to try and create artificial rents).

Posted by: derrida derider on October 30, 2002 10:11 PM

I tried to subscribe to your rss feed but that didn't work. Their are some mistakes in the rss feed, I hope you can correct them, so I can put it in my newsaggregator.

Posted by: Niels Koekkoek on October 30, 2002 10:55 PM

>>the stagnation of investment throughout the late 1980s as the growing federal deficit squeezed out capital formation<<

Isn't this precisely one of the reasons that people are concerned by recent attempts to 'ditch' the EU growth and stability pact, or why Paul Krugman is concerned about the growing US deficits.

On another level exponentials like the IT investment share, or the per capital education levels cannot continue indefinitely. Hasn't the Harrod-Domar idea of coefficients something to teach us here. Not to mention Baumol.

Posted by: Edward Hugh on October 31, 2002 02:53 AM

There is considerable reason to worry about the structural deficit we are developing because of the tax cuts that will be coming for the rest of the decade. However, should interest rates stay quite low in lieu of a deflation possibility I wonder what the effects of the deficit might be. Can a deficit in a low interest rate environment be compared in effects to the deficit of the late 80's?

What can Japan teach us?

Posted by: on October 31, 2002 10:47 AM