Does the CPI Overstate Inflation?

by Don Boudreaux on February 28, 2006

in Standard of Living, The Hollow Middle

As Russ announced, Cafe Hayek recently opened a section dealing with the alleged "hollowing out" of America — the alleged disappearance of the American middle-class.

My two latest columns in the Pittsburgh Tribune-Reviewhere and here — offer some reasons why the Consumer Price Index likely overstates inflation — and, hence, causes the measured growth in living standards of ordinary Americans to be lower than it really is.

Be Sociable, Share!



39 comments    Share Share    Print    Email


Aaron Krowne February 28, 2006 at 10:01 am

Alleged or not, the Gini index has been inexorably marching up:

Interpret this as you wish!

John F. Opie February 28, 2006 at 10:07 am

Hi -

Two very good columns on a difficult subject.

But there are two further aspects that come in to play.

1) Retail channel measurement. The basket in use reflects a certain distribution of retail channels at the time of putting the basket together: these are the data sources for price changes, which remain within the individual channel and are then weighted together at the end of the day using the designated weights of the basket. In other words, if you have two retail channels on day 1, both standard retailers, and 1 year later a discounter enters the retail channel, this is not counted.

Hence consumer preferences between full-service retail and discounters is largely ignored: further, the price changes are not weighted according to the current volume of retail sales, but rather according to the fixed rate of the basket. This means that changing consumer preferences from high-price retailers to high-volume discounters is completely ignored.

2) You are, if anything, understating the quality problem. I work in the commercial world and none of my customers are economists, which in some ways is very difficult (I once had a customer who couldn't understand why we would calculate 1995 = 100. Indexes were something completely new to him) and at other times keeps you honest: the quality changes are so enormous in areas like automobiles and electronics that you must use a hedonic deflator, which relatively few economists understand and even fewer non-economists.

This latter is also, as far as I am concerned, the reason for the so-called productivity paradox: we really do have fantastically improved productivity, we just don't have a way of capturing it accurately. I offer my clients products that would have taken 10 people months to do in the 1980s, but I do it in a couple of hours. Massive improvement in productivity, but most of the benefit is in a vastly improved product, rather than massively increased profits (unfortunately).

Best regards,


Don Boudreaux February 28, 2006 at 10:18 am

I know that the following statement identifies me as the intellectual and moral equivalent of a cheap-beer-guzzling, NASCAR-addicted, gun-worshipping, high-school-drop-out, but I have never, ever — not even a teenie-weenie bit — cared about the the Gini index or, indeed, about income inequality in general.

My parents are in town visiting my wife, my young son, and me. They're now retired — my dad from a life-time of working as a welder in a shipyard and my mom from working as a clerk in a hardware store.

We happened on Sunday evening to talk about income inequality. My parents were surprised to learn that one of my siblings, when growing up in the 1960s and 70s, thought of our family as "poor."

My mom and dad both said "We weren't poor. We were lower middle-class." (They're right.) I took this opportunity to ask my parents (who aren't at all political) what they think of the likes of Bill Gates and other business people who are worth billions of dollars.

"It's great!" my dad says. "They earned it." My mom agrees. In their words and expressions are not to be found even a hint of envy or resentment.

My parents went on to add that anyone whose gains are ill-gotten don't deserve them, but my parents both agree in their guts and in their minds that people who earn money through enterprise are to be applauded and not envied.

Are my parents unusual? Maybe. But growing up I knew lots of parents of my many friends. None of these working-class people ever gave me any hint of feeling oppressed, cheated, or hopeless.

Don Lloyd February 28, 2006 at 10:29 am


Setting aside the fact that all price indices are inherently flawed, it just isn't going work to try to use the CPI as both an indication of living standards and as a measure of the purchasing power of the monetary unit at the same time.

While it is almost certainly true that the CPI is overstated when used as an indication of living standards, it is at least as true that the modern day CPI grossly understates the loss of purchasing power of the monetary unit.

While quality adjustments make some sense when talking about living standards, they are largely inappropriate when used to adjust purchasing power.

First, quality adjustments are sterile unless they represent something that people both demand and can make use of.

If I went inside your computer and increased the clock speed by 100X, you might barely notice for most applications, but it wouldn't make economic sense for you to spend much,if any, more for your computer. If Toyota increased the top speed of the Corolla to 300 mph, would it really have any economic significance? Yet all sorts of similar pseudo-quality improvements are used to hedonically adjust the CPI downward.

Quality improvement can only affect relative prices, not overall price levels. If we waved a magic wand and increased the quality of every product in such a manner that all relative prices stayed the same, without increasing the money supply, then all the absolute prices would remain the same as well. In this case, living standards will have subjectively increased, but there will have been no change in the purchasing power of money that needs to be adjusted for in some kind of a COLA.

Regards, Don

spencer February 28, 2006 at 2:36 pm

Before you write about the CPI again I suggest you talk to someone at the BLS.

Take your example of the typical auto. You are completely correct that the current auto is much different from the auto of 25 years ago. But the BLS has a very good way of incorporating these changes in the way they calculate the impact of changes in autos into the index that accounts for the quality changes. Essentially every improvement in auto technology like brakes, airbags, etc,, is first introduced as an extra that the consumer pays for as a separate item. So the BLS can take the price of an improvement in the last year before it becomes a standard feature and calculate how the value and/or price of the new standard car is impacted by the new technology. For your example to be correct you would have to argue that the price of the extra feature the market establishes for new technology is incorrect.

Moreover, if you take the example of the calculater I think that you will find the BLS does a survey every few years to establish what the typical consumers market basket is. So the BLS will miss something like the caluclator in the very early stages when only a few early adopters purchase the caluclator. But by the time something like the calculator becomes widely used or has a significant impact on the consumer market basket it is in the CPI.

While your arguments about the CPI have some validity, you are massively overestimating the size of the problem.

Big Al February 28, 2006 at 3:54 pm

Don Lloyd,

If the Toyota Corolla hit a top speed of 300 mph it would be far mor valuable to a certain kind of customer. It would also kick ass and guzzle gas. If Toyota could give you that kind of power of the same price, it'd be hard keeping them in stock.

Jason February 28, 2006 at 8:13 pm

A recent fed reserve report stated that the average family income has fallen from 2001 to 2004 – this of course recieved a lot of media attention. The report also said that median income had risen – this did not get a lot of play. When these two things are taken together, doesn't it suggest that the richest are not quite as rich but more people in the middle are moving up? Thoughts?

gump March 1, 2006 at 9:45 am

I have heard, though, that the CPI understates the cost on elderly people. Is this true?

joe March 1, 2006 at 10:26 am


This a great site, and I am a huge fan of most of your work. However, you are wrong on this issue, at least in terms of today's measured rate of inflation. Whether the problems I'll describe below are constant over the 30 years you were looking at, I am unsure, but make no mistake, today's CPI is dramatically understating inflation.

1) Owner equivalent rents. This is how the BLS chooses to incorporate the rising cost of housing into the CPI. Housing accounts for approx 40% of the index, and yet home prices have nothing to do with the input. The BLS seeks to calculate what the rent a homeowner could potentially charge if he did not live there. This leads to a huge error. For the last 4-5 years, home prices have risen at a far larger percentage than rents for the equivalent properties. Thus, in a given month, the CPI may report housing costs up .4%, or 5% annualized, but the National Association of Realtors will announce that home prices rose by 1.17%, or 15% annualized. Now, since about 70% of households now own their home, clearly it is a mistake to focus on the 30% renting. This clearly understates the rate of increase in housing costs.

2) Hedonic adjustments. Don, your complaints about quality not being accounted for is actually incorrect. The BLS "hedonically adjusts" prices on goods, especially in areas such as technology, in order to account for the fact that a $1000 computer last year is not the same quality wise as a $1000 computer purchased this year. In all likelihood, these hedonic adjustments are overdone, and again lead to an understatement of rising prices.

3. Medical costs. The CPI assigns a 6 or 7% weighting to medical costs, but these are actually 14% of GDP, so while not completely accurate to extrapolate consumer spending from aggregated national spending, I would surmise that medical costs are underweighted, and since they are growing at double digit rates annually, the CPI again appears to be understating the true rise in costs.

Finally, realize that the government (and its data measurement instrument, the BLS) has every incentive to keep CPI as low as possible. This benefits them for various reasons, principally because it holds the cost of living increases that many entitlement programs are tied to low. Also, it allows the federal reserve the leeway to keep interest rates artificially low, thereby allowing the government to finance its irresponsible deficits through lower cost debt financing, as well as to hold down the interest costs of previously issued debt.

Brian March 1, 2006 at 1:22 pm

I have a couple of questions related to this. I know for a fact that if I use the CPI as an inflation index, real per capita personal income tax revenues have increased over 20% from 1980-2005. They have increased almost 10% over the past two years.

I am reasonably certain that average tax rates have declined from 1980-2005. I also know that the workforce as a % of the total population over that time period has only increased slightly (certainly not 20+%).

Given all of that, how can income tax revenues increased like they have with lower tax rates if in fact average earnings have not increased?

I have always understood average earnings are well up, but median earnings are roughly flat. I am reading here that the BLS reports that average earnings are flat.

Can anyone help with this questin

spencer March 1, 2006 at 4:55 pm

Brian — you are confusing nominal and real earnings — nominal earnings deflated by
various measures of inflation.

Nobody is claiming that nominal earnings are not growing.

Moreover, nobody is claiming that real average income is not growing. What they claim is that an extremely large share of the growth is going to a small segment of the population so that large segments of the population are receiving a very small share of the increase.

Brian March 1, 2006 at 7:49 pm


I have taken into account nominal versus real earnings. That's why I mentioned real per capita tax revenues are up 20+%.

"Moreover, nobody is claiming that real average income is not growing"

When the article in question mentions, "Average hourly wages of private sector workers", I assume (perhaps incorrectly) that does not literally mean only hourly workers. So I guess the question boils down to what percent of the workforce does that stat include? That would determine its meaningfulness.

Morgan March 2, 2006 at 12:03 pm


With regard to your point #1, isn't the true measure of how much housing costs, given that almost no one (in my experience at least) pays cash for a home, dependent on the mortgage rate?

If the price of a home were a component of CPI, wouldn't we also need to include the cost of borrowing as a component – the cost of homes, cars, education, major medical procedures, and everything else that is financed through borrowing fluctuating greatly with this rate?

Imputed rents at least have the benefit of indirectly reflecting changes in interest rates.

Previous post:

Next post: