David R. Henderson  

Brett Arends on CPI

Financial Crisis Inquiry Commi... Lindsey on the Great

Brett Arends has an article in the January 26 Wall Street Journal titled, "Why You Can't Trust the Inflation Numbers." His distrust is all in one direction: he thinks the Consumer Price Index understates inflation. My distrust is all in the other direction: I think the CPI overstates inflation. My reasons are given in Michael Boskin's entry, "Consumer Price Indexes," in the Concise Encyclopedia of Economics.

Arends' reasons for thinking the CPI understates inflation are given in his article. In analyzing how the CPI is computed, he focuses on two sources of bias.

Here's the first:

Under the official calculations, if steak prices boom, the government just assumes you buy cheaper hamburger instead. Presto--no inflation!

Arends is getting at what economists call substitution. If the price of one good rises relative to the price of another good that is a substitute, we will, ceteris paribus, buy less of the good whose price has risen and more of the good whose price has not. So, for example, if the price of chicken rises relative to the price of steak, we will buy less chicken and more steak.

Why do I give that example? Because even if the price of chicken rises relative to the price of steak, it will still likely be below the price of steak. So I give that example to get you thinking about substitution in response to changes in relative prices. Why do Arends and the man he quotes--John Williams--give his example of people substituting hamburger for steak? I think it's because they want to get you thinking that hamburger is obviously inferior to steak and that, therefore, you are worse off. Of course, you're worse off. No one ever said you weren't. And "no one" includes the economists and statisticians at the Bureau of Labor Statistics. The point of substitution, as any good basic economics principles textbooks will point out, is not that you're no worse off when the price of steak rises and the price of hamburger doesn't, but that you're not as worse off as you would conclude by seeing what the consumer who doesn't substitute will pay. So, for example, if you conclude that the person would pay 10 percent more if he bought the bundle he bought at the old prices, then allowing for substitution means that he won't pay 10 percent more for the same level of utility but, instead, will pay, say 8 or 9 percent more.

Here's the second source of bias, according to Arends:

Or consider the case of Apple computers. We all know Macs are expensive. And we know Apple doesn't discount. The cheapest Mac laptop today costs $999. A few years ago, it also cost $999. So the price is the same, right?

Ha. Not according Uncle Sam. Using a piece of chicanery called "hedonics," Uncle Sam calls this a price cut. His reasoning? You're getting more for the money. Today's $999 Mac is lighter, fancier and faster than last year's $999 Mac. So the government calculates that the "real" price has actually fallen.

How's that work in the real world? Try it. Go into your local Apple store and ask for 50% off thanks to hedonics. (If you do, please, please video the exchange and put in YouTube. We could all use a good laugh.)

Arends thinks it's "chicanery" to note that today's Mac is better than a Mac produced 3 years ago. He admits it's "lighter, fancier and faster." So, adjusted for quality, the price of a Mac really has fallen. The correct conceptual experiment is not to go to the local Apple store and ask for 50% off a new Mac. The correct experiment is to ask what the price of an old Mac is compared to what it was 3 years ago. If you can find an unused old Mac from 3 years ago, I guarantee it will be priced a lot lower.

HT to Thomas Strenge.

Comments and Sharing

CATEGORIES: Revealed Preference

COMMENTS (7 to date)
Chris R writes:

Note that the BLS published a reply to Williams a couple of years ago and summarized it all here:


Basically the argument about weighting is that we should be using year-1913 weights forever...which would mean that the price level now is infinity since I can't have a buggy whip at any price, but who cares about the price of computers? The argument about geometric growth rates is even more silly...if the price of chicken falls by 50% and rises by 100%, so that it's unchanged, arithmetical weighting would say that the price of chicken as reflected in the price index rose by 50%...even though the price of chicken is ultimately unchanged.

Paul writes:

[Comment removed for supplying false email address. Email the webmaster@econlib.org to request restoring this comment. A valid email address is required to post comments on EconLog.--Econlib Ed.]

andy writes:

While I have some reservations about the iMac getting 'cheaper', it all boils down to the aim you have when computing the CPI. Nobody would build an old iMac if you have the knowledge that more or less allows you to build a better one with the same costs. If you aim to compute how 'well-off' people are than it makes sense to try to guess how much better-off you are if you use the new iMac (I sincerly hope that the adjustments do not resemble the equation 'CPU power + 100%' = 50% price...)

However, I am somewhat dubious how useful would such index be for monetary decisions. Let's say we've got some better knowledge; people start buying new iMacs (the older ones aren't being manufactured anymore) for the same price. CPI would show that they are better off (which is correct) - CPI would hint at deflationary monetary policy; which seems to me an incorrect conclusion.

bill woolsey writes:


I agree. Having the monetary authority target the "cost of living" is a mistake. I think the answer is to target money expenditures.

Jeff writes:

Wow. Are the editors of the Wall Street Journal really that economically illiterate?

Julien Couvreur writes:

For me the problem with CPI measure is that it induces you to think that zero change rate is "stability" or "no inflation".

The problem is clear when you think about monetary policy. Let's say that you had a gold standard. The CPI would probably be falling. But we don't know by how much.
Therefore, I don't know that the official target rate (2% target?) set by the Fed in its monetary policy is as low as it appears, or low at all for that matter.

Dale writes:

"Wow. Are the editors of the Wall Street Journal really that economically illiterate?"

It has gotten worse since Murdoch took over. It did not use to be this bad

Comments for this entry have been closed
Return to top