Monday, August 23, 2004

Economics: Less Computers, More GDP?

What is the oldman talking about? The oldman has been wiggling through the bowels of the BEA and BLS releases. He's found some interesting accounting "anomalies". There's no single smoking gun, but there are some pretty damning trails of evidence. How damning?

Well if you look at the BEA page publishing their data Prices and Output for Information and Communication Technologies there's a very odd anomaly. How odd? Before 1997 if you subtract the final sales of the computers for each year (measured in chained 2000 dollars) then GDP according them is bigger than if you don't subtract it.

Huh?!?!

Yes, you heard the oldman right. Apparently according to them, if you subtract the Final sales for computers in the country for each year, up to 1996, it actually increases the total GDP.

Don't believe me?

Check out the XLS file source (Excel) yourself.

I'll reproduce the relevant lines here myself:

[Billions of chained (2000) dollars; SAAR]

YearGDPGDP less salesDiffFinal Sales
19907112.57239.8+127.35.0
1991 7100.57220.0+119.55.6
19927336.67446.5+109.97.1
19937532.77632.5+99.88.8
19947835.57928.0+92.510.8
19958031.78094.3+62.616.8
19968328.98355.7+26.826.6
19978703.58693.2-10.340.7
19989066.99022.2-44.758.2
19999470.39389.3-81.083.8
20009817.09707.0-110.0109.9
20019890.79767.0-123.7125.1
200210074.89942.7-132.1136.1
200310381.310216.5-164.8192.0

If you look at the two outer columns, which are just reproduced from the BEA table, then the chained GDP and Final Computer Sales figures make sense. Now the second column from the left, that's the one that doesn't make sense, and is also reproduced from the BEA. In 1990, if you can believe it, the GDP of this country was bigger without the input of the final computer sales. But it isn't some sort of one off screw up.

The numbers gradually improve up to 1997, when they cross over and start acting as one might expect. That less the total computer sales in this country, the GDP would be less. How much less? The oldman has calculated this number, simply by subtracting the second column from the first column of numbers. This is labeled as Diff, standing for the difference between the first two columns. As you can see, it doesn't correspond to the final sales number either.

Now in an ideal world, you could imagine the BEA simply intending to take the Final Sales column and subtracting it from the unmodified GDP column to produce the GDP less sales column. Except that something horrible happened along the way, like the person making this table got blind skunk drunk and put in the wrong numbers. I suppose that people could argue that this is what the BEA must have intended to do. Nothing else makes sense in a rational world. As it turns out however since he is cynical, the oldman has asked precisely the question what if it was intentional, and because of it has been reading the BEA methodology release. As it turns out it seems that such an assumption that this was a mistake would be itself mistaken.

As far as the oldman from reading their methodology release can tell, yes the numbers are intentional and are part of the BEA's hedonic and chain-weighting GDP calculation method. Further scariness ensues if you segue over to the BLS website and read their methodology release. Apparently according to them their productivity figures are calculated using output and input figures from the BEA. Yes, this includes apparently the screwed up computer technology sales adjustment to GDP figure that is noted above.

There is some sort of insane incestuous data fraudulence going around here. And it's all centered around how technology is accounted for in GDP and productivity. The oldman hasn't uncovered a smoking gun yet, but he's been reading some hair-raising methodological errors in their releases - including if you can believe it the assumption that year over year the computing power in desktop computers stays constant.

The oldman has written previously about Moore's law and how it makes a mockery of hedonic or "quality" adjustments. He didn't know just how true it was.

So it's all tangled up in some weird incestuous data feedback going around where the numbers just don't make any sense at all. Nice words. Final conclusions? GDP in 1990 without the computer industry was higher than it was with it.

Scary stuff. This is like people talking themselves into the proposition that jumping off the World's Trade Towers wouldn't hurt them. And no this isn't a spoof. In days to come the oldman will be tracking the trail of this fraudulent accounting. Anderson-Enron accounting couldn't be more screwy.

I would add as a final note that all of these apparently terminally screwed up numbers are behind the neoliberal consensus in the nineties that formed that we were achieving more prosperity because of higher productivity and GDP growth increases. Apparently, it was all a lie. Scary isn't it?

32 Comments:

At August 23, 2004 at 10:58 PM, Anonymous Anonymous said...

Yes, it's the hedonic adjustment. Before the word was invented, old-school accountants called it "fudging". Not fraudulent fudging, mind you, but well-intended self-bullshitting. Like, well, I went to this party yesterday & had two small pizzas, a good amount of potato salad, and 5 beers in excess of hunger and thirst, but I'm increasing my exercise effort (yeah right) and it will come off, so on a party-adjusted basis my "real" weight should be 1 pound less.

But I didn't get your point about Moore's law -- certainly they don't assume computer power stays constant, but it increases, or why would they inflate computer sales numbers?

What this looks to me is that the hedonic adjustment leads to a rising inflation of computer sales numbers (not constant inflation factor, but rising inflation factor). So if you extrapolate an adjustment rising with time into the past, you should end up with negative numbers at some point. Is this not what is happening?

-- cm

 
At August 23, 2004 at 11:54 PM, Blogger Oldman said...

You'd think so, but as a matter of fact I'm just reading straight out of their press releases, publications, and manuals. I'll cite the ref tomorrow. I keep on thinking I'll come to some sort of sanity at last, some sort of sense of the thing, but as I dig digger it just keeps on getting more and more rotten...

Be kind to the oldman. It took him the better part of an afternoon to figure out how to make that HTML table, and the last two days fiddling with html code to adjust the new template format so that it looked right. Tomorrow I will roll out the next revelation of the bizaarities that I've uncovered.

 
At August 24, 2004 at 10:25 AM, Anonymous Anonymous said...

I think I understand what's happening. When you use hedonic (quality adjustments) older computers look awful, so when you're going back, when you go before the baseline quality number you start multiplying by less than zero.

Or at least that's the only guess I have that makes any sense of the numbers. It assumes that hedonic adjustments are made with multipliers that can go negative. I honestly don't know but I'd love to hear exactly what they're thinking.

 
At August 24, 2004 at 10:38 AM, Anonymous Anonymous said...

Actually, never mind that. I don't understand that spreadsheet at all. There is obviously some math going on between the cells that we can't see - whatever it is it's not intuitively obvious what they're doing. Someone's going to have to dig hard to figure out WTF they're doing.

Ian.

 
At August 24, 2004 at 11:20 AM, Blogger Oldman said...

It's sort of eerie since the further you go back in time, the larger the increase you get by subtracting the hedonically adjusted final sales. So in 1990, when the chain-weighted final sales of the entire computer industry was only 5 billion, subtracting that out from the GDP gives you a boost of 127 billion to the GDP. Which is all patent nonsense of course. It's just physically impossible that this be the case.

But I meant what I said about them ignoring Moore's law. It's freaky the kind of bad juju they're practicing down there in the bowels of the BEA and BLS. This truly is voodoo economics. If they just made stuff up it couldn't be weirder.

 
At August 24, 2004 at 11:21 AM, Anonymous Anonymous said...

This "damning evidence" was already in the air four years ago:

"Although much attention has recently been focused on whether real GDP growth in the latter half of the 1990s has been overstated as a result of the use of hedonic-based price estimates for computers and peripherals and for computer software, there are other reasons to suspect that growth — especially that related to high-tech innovations — has been understated."

 
At August 24, 2004 at 11:22 AM, Blogger Oldman said...

Yes but no one bothered to read it, until now.

 
At August 24, 2004 at 1:03 PM, Anonymous Anonymous said...

Interesting paper. These people are making GDP and inflation numbers much more complicated than they need to be. And I really, really don't like the way they are handling hedonic adjustments for computer (I'm suspicious of the whole concept, truth be told, but the cpu cycle model for computers ignores the usefulness of computers to users.) My computer today is vastly more powerful than the Apples I worked with back in the eighties - but I still database and I still spreadsheet and I still play games. The ease of user interface and the interconnectivity of internet and the pretty graphics are the main improvements from a user's point of view. While dramatic (especially the searchable knowledge base of the internet - which I could obtain with a computer from 1995) they do not nearly add up to the percentage drop in price that the BEA is trying to tell me I've experienced.

Ian

 
At August 24, 2004 at 4:55 PM, Anonymous Anonymous said...

Brad DeLong and Co. grappled with this last September.

http://www.j-bradford-delong.net/movable_type/2003_archives/002148.htmlThey so grappled building from a Kurt Richebacher article in Financial Times where he said, "The important thing about hedonic pricing is that it measures dollars that nobody pays and nobody receives. And it grossly distorts international comparisons."DeLong and Co. thought that Richebacher got it wrong..

One comment from someone named Matt(matthew.higgins@ny.frb.org)may be helpful,

Professor DeLong is exactly right: what hedonic pricing giveth, chain weighting taketh away.

Two comments. First, several studies have found that carefully constructed conventional price indices for computers would also show very large declines.

Second, several OECD countries employ hedonic pricing for computers and measure real GDP growth with reference to a base year rather than by chain weighting. It is in those countries where GDP growth is likely to be overstated.
Posted by: Matt on September 8, 2003 09:30 AM

I still don't know whether or not we ought to worry a lot about hedonics in the CPI and elsewhere, but at least I'm not worrying as much as I did about 6 months ago.. Dave Iverson

 
At August 24, 2004 at 5:42 PM, Blogger Oldman said...

My investigations have discovered something similar, but I did not come to same conclusion. I initially had the impression that chain-weighting offset hedonic pricing. The real answer is that it only partially does so. The problem isn't chain-weighting, it's the hedonic adjustments and the hedonic adjustments are truly out of whack. That's what the GDP less sales numbers say. They're saying even if you chain-weight them, the hedonically adjusted numbers don't converge.

 
At August 24, 2004 at 5:44 PM, Anonymous Anonymous said...

In the meantime Kurt Richebacher is undaunted, writing in January,

In real terms, or "chained" dollars, it increased over the full year until the third quarter of 2003 from $297.6 billion to $390.3 billion, that is, by $92.7 billion, of which $35.4 billion occurred in the third quarter. That is the statistical fiction; actual business spending in current dollars on computers increased over the same time by just $11.5 billion, from $76.8 billion to $88.3 billion, of which $5.9 billion was in the third quarter. In reality, measured in current dollars, nonresidential investment over the year increased overall by $46.2 billion. Among this total, computer investment soared by $93.1 billion, of which $81.6 billion came from the hedonic spin.

Each additional dollar spent on computers in the real GDP accounts during the year translated into eight additional "chained" dollars, accounting, by the way, for 26% of real GDP in this time. The difference between the two measures of business computer investments is exploding. So much for the trumpeted investment recovery.

As we have explained many times, these particular dollars are fictitious dollars that nobody has paid and nobody received. Obviously, such dollars inherently add nothing to profits.

Putting it briefly and bluntly: The trumpeted brisk rebound in U.S. business capital investment is another bullish mirage lacking any serious substance.
January 02, 2004
A Bullish Mirageby Kurt Richebächer

Who is right, Brad DeLong and Co. or Richebacher? Both? Neither? I have no idea. And my discomfort level with hedonics is creeping back. Dave Iverson

 
At August 25, 2004 at 9:35 AM, Anonymous Anonymous said...

Reichebacher is right and DeLong is wrong, alas. GDP is supposed to measure the gross amount of money spent in an economy, not some fantasy of how much money would be spent if you took into account how much better products are today. I've thought about it and I've come down on the BS side. Hedonic pricing is simply bullshit. All I want to know from GDP is how much money is being spent. It's not that simple, of course, because you do need a GDP deflator, but it's close to that simple.

We don't try and say a drill with a faster drill speed is a better drill therefore the older drill cost less than it actually cost. The same thing goes for computers. These people are terminally confused.

KISS applies here and the statiticians and the economists have forgotten it.

Ian

 
At August 25, 2004 at 11:00 AM, Anonymous Anonymous said...

I'd go with Richebächer. Interestingly, the case that he makes is exactly reverted by some economic "visionaries" in Europe -- that European countries should also use hedonic indexing so that their economic indicators compare more "fairly" to the US. (Side note: in a different area, there are attempts at (un)employment "reform", apparently to bring the system closer to the US version, thus producing more US-style "prosperity". This is of course not stated explicity, but it does not take a lot of effort reading between the lines.)

While a case could be made that GDP (or some other measure) should measure "total social welfare", and thus purposefully adjust for quality improvements, I don't see how this applies to computers, that are mostly _tools_, not end products. If you get better tools, this will boost GDP in those areas where the tools are applied. Then what is the need to adjust the tools? To express their "potential"? That's all nice and well, but a potential that goes unused has no place in a factual statistic.

As Ian (?) rightfully points out, even though his computers have gotten more powerful by let's say 50-100X (mostly in terms of processing speed, working memory, and "permanent" storage; less so but still significantly in display quality), that has not translated into becoming that much _better_ in terms of work output. He still does correspondence and office-type tasks (including games!) at approximately the same rate & quality, only that now his (Windows?) file search dialog has a tail-wagging dog that it didn't on the less powerful computers. And if he ever produces slides, he can now fiddle around with so many more color schemes, animations, cool effects, etc. that have little to no bearing on the information content and effectiveness of his presentations. Much of the higher capacity goes into gimmicks, and the GDP adjustment tries to sell us the gimmicks as productivity tools.

I don't want to trash computer advances, but while power and user comfort have increased a lot and many new applications have become feasible, 90% or so computers are used for tasks where the productivity enhancement is rather marginal.

-- cm

 
At August 25, 2004 at 11:06 AM, Anonymous Anonymous said...

A thought that should have occurred to me earlier -- effectively a supposed and "unmeasurable" quality improvement where computers are applied is imputed on computers. But why is it unmeasurable? Because it eludes quantification, or because it does not exist?

It's like imputing an "unmeasurable" creative output of R&D workers on their working hours. Then I should get my bonus based on the number of hours I stay in the office, not what I put on the table.

-- cm

 
At August 25, 2004 at 11:28 AM, Anonymous Anonymous said...

Ian,

I too have had my fill of BS hedonics (in the realm of so-called "nonmarket values" in government land mangement agencies. All my economic advice to Forest Service practitioners has been like yours, to track costs and receipts (where collected), owning-up to the fact that this is a very partial view and therefore complimenting narrow-focused finance with broader social/political reasoning.

As per government accounting writ larger, I have often wondered why we don't embrace more qualitative indicators more akin to Genuine Progress Indicators, or others, rather that tying so tightly to GDP, GNP, etc. Only recently did I realize that the GDP folks were into hedonics.

Dave Iverson

 
At August 26, 2004 at 1:47 PM, Anonymous Anonymous said...

Okay. Someone please take pity on a poor, confused non-economist. What is the one-sentence summary here? "GDP understated"? Or "GDP Growth Overstated"? Or both?

-Mithras
Fables of the reconstruction

 
At August 27, 2004 at 12:51 AM, Anonymous Anonymous said...

Mithras: "GDP growth overstated", and by implication "GDP overstated". But then that's a value judgement, to the extent how much hedonic adjustment you accept.

-- cm

 
At August 31, 2004 at 4:39 PM, Anonymous Anonymous said...

A related issue is that computers and software sold to businesses used to be counted as intermediate input (I think) but changed to investment sometime in the 90s. Therefore, the effect of hedonic pricing is much bigger an issue now than it used to be. Does anybody remember when it started to be counted as investment? Was it 1997? Could this change explain the negative contribution of computers to GDP?

 
At October 11, 2005 at 3:01 AM, Blogger St Louis Cardinals BUFF said...

So many blogs and only 10 numbers to rate them. I'll have to give you a 10 because you have done a good job. Great Job,

Free Access To More Information Aboutsell on the internet

 
At October 12, 2005 at 8:35 AM, Anonymous Anonymous said...

Hey, you have a great blog here!

Just thought I would check out your blog and say hello to as many people as I can.

I have a online computer sale site. It pretty much covers ##COMPUTER## related stuff.

Come and check it out if you get time.

Thanks again

 
At October 12, 2005 at 5:01 PM, Anonymous Anonymous said...

Hey, you have a great blog here! I'm definitely going to bookmark you!

I have a ##KEYWORD## site/blog. It pretty much covers ##KEYWORD## errors and related stuff.

Come and check it out if you get time :-)

 
At October 14, 2005 at 1:38 AM, Anonymous Anonymous said...

Dicecities.net
Large selection of Websites for sale, starting as low as $19.99 come by and check it out.

 
At November 7, 2005 at 7:11 AM, Anonymous Sound equipment said...

After reading a good blog I tend to analyze it to see what the person behind it might be like.

I've been in the goal setting business a while now and just like most people with experience in different areas of expertise, I can tell a lot about a person just by speaking to them or reading something they've written.

You haven't wrtten any goals on paper for awhile have you? And if you have I'll bet you haven't looked at them for a long time. Right?

Software is hard work, and harder still if you don't have short range goals, mid range goals and long range goals.

I think you'll agree that few people really take the time to set goals of any kind.

When is the last time you really thought about setting some real goals, or are you like the vast majority of people who just "hope for the best"?

You already know successful people aren't "just lucky", they know how to set effective goals and reach them.

I was like that once, you might be also. How about changing all that.

Come on over to goals+ and learn exactly how to set effective goals, act on them and be the real you.

There's a ton of FREE information and who knows, this little insert in your blog might well change your life!

Happy New Year!

 
At November 23, 2005 at 8:43 PM, Blogger All the best to you! said...

Hey Oldman, you have a great blog here! I'm definitely going to bookmark you!
I have a Laptop Computer Reviews site. It pretty much covers Laptop Computer Reviews related subjects.
Come and check it out if you get time :-)
Best regards!

 
At November 24, 2005 at 2:29 AM, Blogger All the best to you! said...

Hi, Oldman.I was just searching blogs,and I found your site! I like it!
If you have a moment, please visit my site:
Free Computer Backgrounds
It covers Free Computer Backgrounds related contents.
All the best!

 
At December 4, 2005 at 5:38 PM, Anonymous data recovery said...

Hey, you have a great blog here! I'm definitely going to bookmark you!
I have a hard drive recovery site/blog. It pretty much covers hard drive recovery related stuff.
Come and check it out if you get time :-)

 
At December 11, 2005 at 2:25 AM, Anonymous computers sales said...

Oldman,

I took a look at your post, Economics: Less Computers, More GDP?,
regarding computers sales .

You are invited to place a link to
your blog on our website for free. See:

http://www.quickregister.net/infowizards


We get over 18,000 visitors per day.
Many are looking for computers sales
related products and services.

We have a specific category for computers sales .
Your listing will be spidered by the search
engines under computers sales . Our pages
are made to be search engine friendly.
We hope you take a moment to take
advantage of this free advertising.

Cheers,

John

http://www.quickregister.net
Free Search Engine Submission Service.

 
At January 24, 2006 at 11:32 AM, Anonymous disk recovery said...

Hey, you have a great blog here! I'm definitely going to bookmark you!
I have a data recovery site/blog. It pretty much covers data recovery related stuff.
Come and check it out if you get time :-)

 
At January 24, 2006 at 8:15 PM, Anonymous drive recovery said...

Hi - I’m a newbie trying to find some information
relating to technology
However, I managed to find your Blog which I think is interesting and had some relevance to what I was looking for. Thanks for your post and do keep up your good work.!

 
At February 2, 2006 at 12:10 AM, Anonymous joel said...

A recent study of 80 large organizations by Infonetics Research found that overall downtime costs averaged an astounding 3.6% of annual revenue! In another syudy, Forrester estimated the average cost of downtime for e-commerce sites at $8,000 per hour-- at larger sites like eBay,Intel and Amazon, the costs soar to hundreds of thousands of dollars per hour.
Link to my site: cd data recovery free ware

 
At November 9, 2010 at 7:55 AM, Blogger alina said...

Barrister Global Services Scam

 
At July 14, 2011 at 8:59 AM, Blogger John said...

Nice Blogging,
UTAH : Utah Web Design http://www.adaptivitypro.com/utah-web-design/

 

Post a Comment

<< Home