Heinz, the Barron's reporter elaborates extensively on hedonics pricing in this week's edition -- I suppose he got lots of email on last week's article.
In this week's article, he flatly states that the Bundesbank analysis is wrong because it uses the wrong weights to compute the effect of hedonics on the reported statistics. I hate to bring this up again, but his arguments make sense to me, and I think it is important to bring it to the attention of the many reading this thread that believe the CPI and other numbers reported by the government statisticians are bogus, largely because of hedonics.
Here is the link to this weeks article and the part with the hedonics argument.
interactive.wsj.com
"Just go to the Website maintained by the Bureau of Economic Analysis (www.bea.gov), find the August 27 news release and then scroll down to Appendix A.
That's the unsolicited advice I rendered to Kurt Richebacher when I called him in Cannes, France, last week. Dr. Richebacher, author of a famous doom-and-gloom newsletter and former chief economist at Germany's Dresdner Bank, had written a long letter reiterating the claim that "U.S. GDP and productivity data [are] grossly overblown by the method of measuring business computer investment."
But as I tried to explain in this space last week, that simply isn't true, and I offer up Appendix A as my Exhibit A. There you'll find a list of estimates for GDP growth that excludes all final sales of computers, not just to business but to consumers, government and for export. The result: GDP growth including computers generally runs a mere 0.4% greater than GDP growth without them.
Hedonics, the government's method for boosting the contribution made by computer sales, accounted for no more than 0.2% of that 0.4% difference. Hedonics is what Richebacher feels is at the root of the problem.
How can this be, when the analysts at the Bundesbank calculate that with the aid of hedonics, U.S. computer investment soared by 40% a year since 1991? How can such a big number lead to such a small result? Answer: Because several years ago the Bureau of Economic Analysis decided to calculate gross domestic product according to the method known as chainweighting. (For more about the Bundesbank, see this week's Mailbag.)
Here's how chain-weighting works. When BEA says that GDP expanded by 4% in a given year, the statement conjures up a vision of all goods and services growing uniformly at that rate. But alas, the economy doesn't work that way. The output of some goods may be rising by, say, 6%, while that of others may even be shrinking. So what the agency is really saying is that, according to the method by which it averages the flow of goods and services, it has come up with a figure that equals 4%.
Each item is assigned a weight through the only means handy -- prices. The more expensive an item, the greater its weight. But prices vary over time in relation to each other, which means that weights should vary over time.
According to the old method of weighting, you choose a particular year when certain price relationships prevailed, and you use that year's data to assign each good or service its weight. That's known as the fixed-weight system. But with the rise of computer output, BEA realized that fixed weights were causing severe distortions.
Take a simple example. Imagine an economy producing just two products: bread and widgets. Say that in Year One, the price of a loaf of bread is 50 cents, while that of a widget is $1.50, and that 100 units are initially produced of each. Now imagine that BEA decides to use Year One in its fixed weight system. Since $1.50 is three times as great as 50 cents, the increase or decrease in the number of widgets gets three times as much weight as a change in the number of loaves of bread.
Now assume that by the end of Year One, widget output doubles to 200, while bread output rises by only 10%, from 100 loaves to 110. What's happened to GDP?
Well, remember that because of the price ratio established at the beginning, widgets are being assigned three times as much weight. Clearly, then, the jump in widget output has more than compensated for the much slower increase in loaves of bread, and GDP growth has soared -- by 82%, if you do the math.
But now suppose that by the end of Year One, the widget price has plummeted to 50 cents, while the bread price has remained the same at 50 cents. In fact, industry observers know full well that the only reason the widget industry was able to produce and sell three times as many units as a year ago was because it was willing to slash prices.
So why not use those more up-to-date prices to weight the rise in output? Instead of three-to-one, widgets-to-bread will now run one-to-one. And with equal weights, the rise in GDP will be more modest-running 55%.
Which calculation is correct -- 55% or 82%? The answer, says BEA, lies somewhere in between. Instead of using fixed weights, it uses chain weights, which involve a kind of moving average that splits the difference between this year's and last year's prices.
Widgets stand for computers. As computer output has soared, their price per unit has been plummeting. The impact of this trend on GDP is actually exacerbated by hedonics. What Richebacher and the Bundesbank economists assume is that their output is measured according to fixed weights. They're quite mistaken."
Kyros |
