08-09-11 - Inflation - 3

Some reference :

Response to BLS Article on CPI Misconceptions
Consumer price index - Wikipedia, the free encyclopedia
WSJ piece on hedonic price adjustments
Consumer Price Index
Chapter 11�Money and Its Purchasing Power (continued) - - Mises Institute
Bill Gross Claims the CPI is Understated, But Is He Right - Seeking Alpha
An inflation debate brews over intangibles at the mall
United States Consumer Price Index - Wikipedia, the free encyclopedia
Trudy Lieberman Entitlement Reform Archive CJR
Shadow Government Statistics Home Page
Meet the Fed's Elusive New Inflation Target - TheStreet
Inflation The Concise Encyclopedia of Economics Library of Economics and Liberty
How BLS Measures Price Change for Medical Care Services in the Consumer Price Index
Higher Education Price Indices
God Punishes Us When We (Collectively) Vote Republican, Part 5 Angry Bear - Financial and Economic Commentary
Consumer Price Index, a rant
Charts College Tuition vs. Housing Bubble � My Money Blog
Chained Cpi Social Security, CPI, Michael Hiltzik Using 'chained CPI' to determine Social Security payments would rip off ne
cbloom rants 5-14-05 - 1
cbloom rants 12-27-08 - Financial Quackery
cbloom rants 09-17-07 - 2

Some of these guys have the whiff of crackpottery which should give us a bit of pause. Nevertheless...

We can track down a few of the strange problems that I identified last time.

Education basically is miscounted : "The inclusion of financial aid has added to the complexity of pricing college tuition. Many selected students may have full scholarships (such as athletic), and therefore their tuition and fixed fees are fully covered by scholarships. Since these students pay no tuition and fees, they are not eligible for pricing." discounting financial aid makes some sense if you are trying to measure the consumer's expense, but not if you are trying to measure the cost of the good; just because someone else paid for part of it doesn't make it cheaper. But really I imagine the biggest problem with education cost is that they effectively count college as being free for people who can't afford college. That is, people who can't afford it don't buy it, so it's not in the basket (doesn't contribute to the "quantity" in the CPI metric). A better way to measure inflation would be to assume that everyone would go to college if they could afford it. (also it seems that non-acredited technical school time places are not counted at all)

Health care is simply not counted at all, by design : "The weights in the CPI do not include employer-paid health insurance premiums or tax-funded health care such as Medicare Part A and Medicaid" The only thing they count is out-of-pocket / discretionary health care expenses, which are obviously just a tiny fraction of the total.

Real estate has the funny owner's cost to rent thing which makes it very hard to tell if that is being gamed or not.

Obviously anything based on "core" inflation (without food or energy) is ridiculous. The standard argument that those fluctuate too much seasonally is absurd, you could just use a seasonally-adjusted moving average, you don't need to remove them completely.

The other really obviously fishy parts are :

"Substitution". A while ago the CPI was changed to use geometric averages of prices within a category. This seems pretty innocuous, but it basically causes a down-weighting of higher priced items. And in fact the geometric mean is always lower than the arithmetic mean, so this change can only make inflation seem lower, which is a dirty trick. For example :

(1+1+8)/3 = 3.333

(1*1*8)^(1/3) = 2.0

pretty big difference even though they are both "means". Now, they hand wave away and say that this reflects consumers' ability to choose and substitute cheaper products. But it is totally unscientific.

Furthermore, newer measures like the CPI-U or Fed's PCE also explicitly include substitution. This just seems like it obviously does not reflect inflation. When a product gets expensive and the consumer substitutes for something cheaper, they are by definition getting something of lower utility (because it wasn't their first choice), so you can't say that no inflation happened, they are getting less for their money.

"Hedonics". These are poorly documented pure bullshit ways of pretending inflation is lower by claiming that we got more for our money. This is just pure nonsense for various reasons :

1. The whole definition of "better" is so vague and open to interpretation that it has no business in a metric. For example they consider air travel to be massively improved since the 70's. Sure it's safer, more efficient, but also much much less pleasant. Personally I think that the same trip is actually worth much less now than it was in the past, but they say it's worth much more. Similarly for the quality of buildings and clothing and cars and so on; yes, they're safer, faster, more durable, whatever, but they aren't hand crafted, they aren't made of hard wood and steel and chrome; I think most of those things are actually much crappier now than ever, made more cleverly but also more cheaply. Anyway, it just has no business in there. The idea that you can measure the hedonic quality of some product and say it improved by 0.1% from April to May in 2010 is just absurd.

2. Using quality of goods at all just isn't right to begin with. Inflation should be a measure of the cost to buy a standard set of goods at the expected quality level of the era. Just because technology gets better over time doesn't mean you can discount the inflation! For example if computers get 50% better every year and our money is inflating by 50% would you say the cost of computers is not changing? Of course not, the cost is going up 50% , yes they are also getting better but that is not part of the discussion.

It's just wrong on the face of it. If a median decent car was $5k in the 70's and now is $25k , then the price of cars has gone up by 5X. But oh no they say, you have air bags and more power and fuel economy and so on, the modern car is 5X better, so in fact there has been no inflation at all. Well, wait a minute. I *expect* the quality of life and technology to go up over time. Are you telling me that in a world with 0% inflation that technology does not get better over time? That's a strange way to measure things. And it's not really what you want to know when you ask about inflation. You want to know how much money do I need to afford a decent house, car, food, etc. at the expected standard of the time that I buy it.

Obviously we wish there was some item that had absolute constant value that we could measure against. Also obviously measuring inflation is very complicated and we are only scratching the surface. But it's very fishy. Rotten fishy.


jfb said...

If the car has gotten five times better so that five times the price is a trade I'd be getting a good value to make, can I buy something equivalent to the older car, *in the condition it was when it was brand new at $5K*, for $5K now?

That is to say, is there any trade similar to the old one for $5K? Supposing I were happy with the old quality of car, can I still get it? If a similar car is now $10K, that might indicate something...

Apples to apples really is the only honest way to measure I can think of... A 486 was good enough for many people for word processing etc., they aren't getting 100 times the value on Facebook with their shiny new PC despite the clock rate...

Very odd that such a method of "measuring" has come to be accepted.

mimi igs said...

I agree that the CPI has problems and it might be skewed to give a lower number than 'for real'.

That being said, focusing too heavily on straight CPI doesn't give a very complete picture of inflation for a lot of reasons.

One of the reason why economists don't focus heavily on straight CPI is that it doesn't allow one to separate simple price fluctations from actual inflation trends. Paul Krugman has an explanation of this here http://krugman.blogs.nytimes.com/2008/05/31/embedded-vs-non-embedded-inflation/. Feel free to discount the source if you think he's a leftist pinko mouthpiece I guess, but the concept is fairly uncontroversial.

A couple of other things to point out. Food and energy are excluded because they are _very_ volatile. Here is the a fed article talking about core inflation.


Look at the fluctuation in energy prices in chart 1. Any kind of fiscal or monetary policy that tried to compensate for fluctuations on that scale would be insane. You'll also notice that the fluctuations are not at all seasonal like home prices or employment. Finally, food and especially energy is so volatile partially because they are more-or-less raw commodities and are driven more directly by surges and drops in global supply-and-demand which is inherently much more volatile than say, Federal reserve interest rates.

One other thing I would mention is that there is a bit of a classic perception issue with inflation over the past 12 months where people feel it when food and gas prices go up much more accutely than when they go down. Both food and gas were in big upswings this year for various reasons (i.e. globally bad grain harvests, recovery from recession + emerging economy growth), and it wouldn't suprise me if both food and energy were down over the next 12 months but people don't perceive the down legs as accutely as the upswings.

Finally, if inflation were such an obvious concern over the medium term, it would be hard to explain the financial communities continued willingness to loan the US billions of dollars at an effective yield of just a smidge over 2% over the next 10 years. That is kind of the financial equivalent of 'scoreboard' :)

mimi igs said...

I agree the CPI can be whacky and that it has a bunch of crap in it that is very handwavy, sometimes hard to justify, and _possibly_ rigged (slightly IMO) in the government's favor.

However, there are good reasons why economists don't get too focused on the raw CPI, and why the core CPI is generally considered more 'important'. It has to do with finding the price trend signal in the noise. Paul Krugman referred to this as embedded vs. non-embedded inflation and explained it here:


Feel free to dismiss it as the ravings of a pinko leftist mouthpiece if you want but the basic concept is fairly uncontroversial in the field of economics at this point.

Also, some people may not realize just _how_ volatile food and energy can be. This is another article explaining core CPI:


Check out the volatility of energy in chart 1. Any fiscal or monetary policy targeted at smoothing those fluctuations would be insane and foolhardy. It's also very clear that they are not seasonally driven like employment or housing and can't really be 'adjusted'.

In the end, food and energy are so volatile because they are driven much more directly by short-term global supply and demand.

One thing I also think is going on with perceptions of inflation this year is the psychological phenomena where people feel the pain of a financial loss much more acutely than the elation of a financial windfall. The past 12 months was a very bad year for both food (globally bad grain harversts) and energy (recession recovery + excess demand from emerging markets). If, for example, the prices of staples like gas and bread start reverting to trend this year, people won't 'feel' that nearly as strongly as the sting of the recent inflation.

Finally, it would be hard to reconcile high inflation or pending high inflation with the fact that the US government has been able to borrow billions upon billions of dollars from the financial markets at yields barely north of 2% over a 10 year term. I believe this is the financial equivalent of the 'scoreboard' chant :)

cbloom said...

I have no problem with Krugman's slightly-less-right-leaning-than-most-economists politics, but I do have a problem with his sloppy reasoning. "embedded vs. non-embedded inflation" is just complete hand-wavey nonsense. I can't find one piece of actual information in that entire article.

Obviously if you're making monetary policy you don't want to take drastic action because of temporary fluctuations.

But any economist who doesn't have their head in the sand should know that it is *impossible* to tell the different between transients and trends until they are over! Your only option is to use some kind of time-delayed low pass.

For example, I think this :


is a reasonable and well thought out proposal. Rather than exclude energy, you should exclude high frequency changes in general.

Furthermore, it seems to be a complete myth that food is volatile :


energy in fact is very volatile, food is not (in modern times).

See also :


As for this -

"it would be hard to reconcile high inflation or pending high inflation with the fact that the US government has been able to borrow billions upon billions of dollars from the financial markets at yields barely north of 2% over a 10 year term"

that's an interesting question that I don't know the whole answer to. But here are some ideas :

1. If you have fear of major market collapse you may have good reason to make safe investments whose return is even negative

2. When the fed funds rate is 0% , a 2% profit is not bad.

old rants