June 30, 2002
My Earliest Forecast of the Late-1990s Boom: May 18, 1994

I was performing semi-random keyword searches of my hard disk when I found this: my first forecast of the late-1990s new-economy boom. It was made on May 18, 1994--much longer ago than I had remembered.

Of course, it was a hesitant and uncertain forecast: I clearly didn't believe what the numbers were telling me. And in order to understand what I am saying, you have to know that Larry Summers and I wrote a series of articles back around the start of the 1990s showing that the cross-country pattern of economic growth suggested a remarkable role for private-sector investment--equipment investment--as a cause or carrier or perhaps simply as an indicator of rapid forthcoming productivity growth.

Nevertheless, the speed with which investment in high-tech and other forms of equipment jumped as the Clinton administration deficit-reduction program began to take hold is truly astonishing.


May 18, 1994


From: Brad DeLong, Deputy Assistant Secretary, Economic Policy


Isn’t this remarkable? I should figure out how much of it simply the falling price of computers coupled with the 1987 base year. But if we do not see a substantial acceleration of U.S. potential output growth over the next few years, Larry Summers and I will have a lot of explaining to do...

Posted by DeLong at June 30, 2002 08:17 AM

Email this entry
Email a link to this entry to:

Your email address:

Message (optional):


note: a typo early on -- "book" used instead of "boom"

Posted by: David Ballard on July 1, 2002 07:22 AM

I've heard the hedonic deflator described as "magic" and "mirage". What do you think and what should an ordinary person make of statistics that incorporate this feature? (I've done the requistite google search and have plenty of technical material to peruse.) I'm wondering with all the talk of an improving economy, impressive productivity and still faltering stock market (valuations and psychology aside) if government statistics aren't next in line for scrutiny.

Posted by: William on July 1, 2002 10:50 PM

Well, from a "strictly" economic standpoint, hedonic deflators make sense to me. After all, imagine an indifference curve between a higher-quality and lower-quality product that are similar, and hence close substitutes. Surely, the curve will intersect the axis of the higher quality good at a lower level than it intersects the lower quality good: since this is close to a straight line, the goods being close substitutes, I'll claim the marginal rate of substitution is constant... anyway, my point is that to achieve equal utility, you have to have more of a lower quality good, so in terms of whether people would prefer to live now or when the hedonically-adjusted GDP is lower, it makes just as much sense to adjust for higher quality as to adjust for higher quantity.

Two small gripes:
1) It's a tough job, but I'll assume that the national accounts guys know what they're doing. They're probably well trained econometricians and such, and I'm a high school student.

2)GDP is often used, in the short term, not so much as a quality-of-life indicator (which is one reason, IMHO, of why externalities are not factored in) as an indicator of the level of *cyclical* business activity. Saying that GDP is up because quality is up more than quantity is down may indicate that we're better of in a mild recession than at the height of the previous boom (in the nebulously cardinal utilitarian world of "quality of life" accounting), but it also renders GDP somewhat meaningless as a business cycle indicator. After all, product quality is probably *always* improving, unless our civilization falls into a dark age or something, but that tells us nothing about the current direction of our economy in a cyclical sense.

Julian Elson

Posted by: Julian Elson on July 2, 2002 12:09 AM

A couple of examples.

From Grant's Interest Rate Observer (date unknown): But what continues to fascinate most is the flattering effect of hedonic adjustments, particularly in rapidly deflating computers and peripheral equipment. In nominal dollars, the growth in this line item was about flat -- up $1.9 billion. But deflate the numbers (to reflect imputed quality improvements and price declines), and computers and peripheral equipment rose in real terms by $23.5 billion. Expressed another way, nominal outlays rose to $80.4 billion from $78.5 billion. Real outlays climbed to $289.2 billion from $265.7 billion. Note, observes colleague Jay Diamond, the huge disparity between real and nominal after only five years of deflating. In the world of hedonic adjustments, nominal of $80.4 billion equals real of $289.2 billion.

From Jeffrey Saut, Raymond James (3/4/02): Also helping the GDP revision was the use of “hedonic” pricing, which is used to capture the increases in “real” computer power. In essence, by translating the price declines of computers, hedonics correspondingly adds to GDP growth. For example, between 1995 and 2001 businesses actually increased spending on computers by some $23.2 billion. But, when the “hedonic deflator” is applied to this figure, said spending is magnified by a factor of 10 to yield a jump of some $240 billion! Also, remember that whatever adds to GDP growth adds equally to productivity. . . hello the productivity miracle of the 1990s, which evaporates without hedonics [sic].

Then this from NYT: The reasons [for sluggish technology spending] go beyond the weak economy. Perhaps for the first time since the invention of the computer, technology is outpacing customers' needs, industry executives say.

Now, Moore's Law has a new take: customers can spend half for the same processing power every 18 months. So we could see declining nominal outlays deflated into real gains except for the technology vendors for whom it won't feel so real.

Posted by: William on July 2, 2002 05:50 AM

>>the national accounts guys know what they're doing<<

They're certainly trying very hard. And hedonic methods applied to investment produce results that seem reasonable. But be aware that hedonic methods applied on the labor side--trying to see if people are, holding other things constant, paid more for dangerous, noisy, or otherwise unpleasant jobs--routinely fail to produce believable estimates.

I'm a believer in hedonic methods. But it does make the numbers feel a little less solid.

Brad DeLong

Posted by: Brad DeLong on July 5, 2002 06:56 PM

>>They're probably well trained econometricians and such, and I'm a high school student.<<

Albeit it looks like they're also making a higher quality of high school students these days...


Brad DeLong

Posted by: Brad DeLong on July 5, 2002 06:58 PM
Post a comment

Email Address:



Remember info?