Marginal Utility

by Tom Bozzo

So I bit the bullet and, after 4-1/2 years with my trusty PowerBook, bought a new computer for the house, an iMac G5 if you must know. Since I'm an Apple partisan of long standing, the decision was mildly complicated by Apple's recent announcement of the switch of the Macintosh line to Intel processors starting next year. The Intel announcement did cost Apple some current-period revenue. I'd probably have opted for a fully-loaded PowerBook had Apple stuck with the PowerPC family; I'd assume margins on the higher-end computers are better, too. Oh well.

New Computer Day is also as good of a time as any to take a reading on computer price deflation. The price premium for portable computers makes a comparison between the old PowerBook and the new iMac unfair. My previous desktop computer purchase, from August 1995, is ancient history as far as the Consumer Price Index goes. The CPI-U series for computer hardware and peripherals didn't even exist at the time — it started at 100 in December 1997 and stands at 13.2 as of May 2005. Splicing in the broader IT hardware and services series for the earlier period, to fill in something for the missing computer series, suggests that computer prices are now around one-tenth what they were in the late summer of '95.

This partly reflects econometric adjustments for the quality of modern computers — a.k.a. "hedonics" — and drives nuts some people who don't like the basic idea of hedonics, that better could be considered equivalent to more. It is sometimes objected that the people who bought $3,000 computers in 1995 aren't buying $300 computers now (*). Or that a fancy new computer won't allow a blog post to materialize any faster than on the old computer (**). Another objection is that quality adjustment of price changes makes the inflation picture (and, by extension, data purporting to describe the health of the "real" economy) look better than they really are, making hedonic price indexes part of <irony>a grand conspiracy to prop up the economy</irony>.

Yet another is that quality improvements shouldn't directly translate into hedonic price reductions, which would be a problem if they did, but they don't, as the following comparison table will show:

Feature	1995 computer ($2,799)	2005 computer ($1,499)
CPU	100 MHz PowerPC 601	2000 MHz PowerPC 970FX
RAM	16 MB	512 MB
VRAM	2 MB	128 MB
Disk capacity	0.5 GB	160 GB
Optical disk drive	CD-ROM	Reads/writes various CD and DVD formats
Display	None	17" LCD
Modem	None	V.92, useless
Wireless networking	None	802.11g, Bluetooth

The main cardinally quantifiable specs have improved by factors ranging from 20 times to a whopping 320x for the HDD, far greater than the 10x deflation factor. In fact, it would be easy to run the price of a 1995 computer into the tens of thousands of dollars trying to more-or-less match 2005 specifications to the extent that would have been technically possible at the time. The upshot is that the constant-quality deflation in Apple computers is much larger than that from hedonically-adjusted price index! Considering that very few people would have paid 1995 prices for 2005 personal computer specifications, that is not a flaw in the measurement.

Addendum (geek note): The most expensive RAM I ever purchased was the 16 KB RAM expansion pack for the Sinclair ZX81, $50 as I recall. At that unit price, 512 MB (***) would be $1.638 million.

---------------------------

(*) The problem with this argument is that they could, or very nearly could, and would get a better computer for $300 now than $3,000 would have bought 10 years ago.

(**) The CPI data actually show much less deflation in software prices than hardware prices, partly reflecting this phenomenon. There are significant counterexamples for highly processor or storage-intensive tasks.

(***) Current retail price around $50.