G'day Doug, > Devine, James wrote: > > >is there something about counting software as if it were a physical > >investment (and thus part of output and the numerator) that distorts > >productivity numbers? > > Shouldn't, much, though no one knows how to adjust it for quality > either. It does enter into the growth accounting games that are the > supposedly rigorous evidence for the productivity burst (e.g. Oliner > & Sichel and Jorgenson & Stiroh). > > I see no conceptual reason why software shouldn't be considered an > investment - without it, computers are useless, and it lasts a long > time.
Software is definitely investment, but it's been a very regular investment, no? Where I work, people are always getting perfectly functional computers or software suites whipped off their desks and replaced - loads of labour, living and dead, goes into this artificial cycle. If that experience is generalisable, we'd be talking high profits and productivity (which number depends on what price increases you manage squeeze out of your customers as much as it does on your plant and labour costs) for computer, software and corporate network consultancy firms and lower-than-necessary numbers everywhere else. Ain't that what the likes of Gordon and Preissl found in the 90s? Here at home, my Mac is 9 years old and so is the software that came with it. I type just as slowly on this as I do on the one at work, and get my e-mails at exactly the same speed (although Michael Keaney and Charles Brown's posts don't come wrapped). So, yeah, software does last a long time - as long as it takes you to replace it. And the pentium may be faster than the 486, but it's the 286 chip that's embedded in much of our electricity network - it'd run no better with a pentium, would it? Oh, and it'd cost no more to make a computer with a clockspeed of X than it would one of Y, would it? Each might have cost as much as the other to develop at the time, and I can't imagine plugging a pentium into a motherboard (if that's where chips go, I wouldn't have the faintest) is any harder than plugging in a 486. So some pentiums are called upon to do stuff a 486 couldn't have done, but a lot are not. And only some of that difference is a productive difference. How on earth would one find the proportions and numbers in all this? Allowing for the fact I'm missing the point, Rob.
