Further thoughts on “inflation”

Nick left the following comment on the previous post:

I really like this post … So I’m sorry to snark … But:

‘I concluded inflation is real.’
‘That’s why I keep claiming that inflation is a meaningless concept.’

Me think Econ no fit words good sometime.

What can I say?  I suppose I was thinking about this in two different ways:

1.  The BLS tries to do “hedonic” adjustments to the CPI, i.e. adjust for product quality change.  I showed that when we really did have high inflation you’d see the prices of ordinary items like cars rise very rapidly.  That’s clearly not true today.  I also showed that using what I thought was a plausible hedonic comparison (the 2014 Accord is just as good as the 1986 Legend) you could get zero inflation in new car prices, far less that the 35% assumed by the BLS.  So I’m dubious that the BLS is grossly understating inflation.  Of course I acknowledge that lots of service prices have risen faster than car prices, and thus there has been some inflation.  Even so, using the BLS hedonic approach, the actual BLS numbers seem plausible.

However . . .

2.  When I say inflation is a meaningless concept I’m suggesting that the concept is not well defined, despite the BLS’s attempts to do so.  Here’s commenter Vivian making a very good point:

Taking literally, “hedonic” means relating to pleasure. Did you get more or less pleasure from that 1964 Olds with all its trunk and leg space, steel and chrome and its muscular engine than you would from the Accord? What would it cost today, even with our advances in manufacturing technology, to reproduce that 1964 Olds with the same specs? Are hedonic adjustments confusing functionality with price (or even pleasure)?

These are all debatable questions, and for this reason I doubt the statement “there is no such thing as true inflation rate” is debatable.

In earlier posts I’ve made an argument (similar to Vivian’s and almost the opposite of my previous post)–that if you use a sort of “pleasure” criterion, then price inflation is roughly equal to wage inflation, and living standards haven’t risen at all. Thus people used to get great pleasure from crummy black and white TVs, but now someone with that TV set would be miserable, thinking about the great big flat panel HDTV his neighbor has.  He’d feel poor.  If economists really believe the CPI is supposed to measure a constant utility level, then for all we know there might have been no real wage gains in the past 100 years.  Who’s to say if people are happier than 100 years ago? All of these concepts are so slippery that I’m very skeptical of the notion that there is any “true” rate of inflation.

But my previous post was sort of saying; “if we are going to play the game of trying to seriously estimate inflation using BLS hedonic-type approaches, there is no reason to doubt their claim that inflation has slowed sharply from the Great Inflation period.”  Nice quality cars went from $3600 to $22,500 in 22 years, then to $22,105 in 28 more years.  You can quibble about the models I chose, but the overall pattern is clear.  Inflation has slowed sharply.  Or should I say “inflation” has slowed sharply?  I don’t seem to be able to make up my mind.



48 Responses to “Further thoughts on “inflation””

  1. Gravatar of Vivian Darkbloom Vivian Darkbloom
    11. September 2014 at 07:42

    “When I say inflation is a meaningless concept I’m suggesting that the concept is not well defined, despite the BLS’s attempts to do so.”

    Exactly so. Almost all the arguments over whether and to what extent we have “inflation” and to what extent are really about definitions, but rarely are disagreements directly expressed that way. The “inflation truthers” are essentially arguing that they have a different definition of inflation than do others. This largely revolves around whether hedonic adjustments should even be attempted, etc. It’s pretty easy to win an argument when you get to measure your argument against your own definitions. In essence, each side is arguing how “inflation” should be measured and hence defined, but without directly saying so or giving cogent reasons therefor. That’s not very useful.

    What would be useful would be arguments that center around *why* a particular definition is appropriate for the purpose one proposes it be used. Once that is established, then one should make the case that it is actually measurable within a reasonable time and an acceptable degree of accuracy so as to be useful to policymakers. The second point is often forgotten as, for example, Steve Kopits has reminded us about the major revisions in GDP. Flying without both appropriate and accurate instruments is really flying blind.

  2. Gravatar of Randomize Randomize
    11. September 2014 at 07:49

    Dr. Sumner,

    I like the car comparison but your choice of models is suspect. While the Legend and the Accord are both made by Honda, the Legend (aka the 3.5RL) was sold under the premium Acura branding. The Legend is actually still sold today although it is only called the Legend outside of the United States. The 2015 RL model starts at $48,450, a rate of inflation of about 2.7%.


  3. Gravatar of Dan S Dan S
    11. September 2014 at 07:50

    I think you’re somewhat stuck between a rock and a hard place, and I sympathize. On the one hand you’re totally right to say inflation is not well-defined. That is, there doesn’t seem to be such a thing as a “true” rate of inflation in a Platonic ideal sense (not that I believe in Platonic ideals but you get the idea) and all attempts to “measure” it are very messy weighted averages of price increases with quality adjustments.

    But on the other hand, let’s say NGDP increases 20% next year. There is SOME sense in which that 20% increase is not solely due to an increase in production and is due to stuff getting more expensive.

    To try to put it in a clever way, the only proposition more ridiculous than the existence of true inflation is the NON-existence of true inflation.

  4. Gravatar of Randomize Randomize
    11. September 2014 at 07:53

    According to the BLS inflation calculator, $22,500 in 1986 dollars would be worth $48,910 in 2014 dollars; pretty darned close to the price of that 2015 Acura.

  5. Gravatar of Michael Michael
    11. September 2014 at 07:59

    Randomize, I think part of Scott’s point is that today’s Honda is a far higher quality product than the 80s Acura, never mind that it is no longer considered a luxury product.

  6. Gravatar of Mike Sax Mike Sax
    11. September 2014 at 08:21

    It’s funny but I was going to ask about this last night-but I couldn’t get to my computer-was Friedman actually off in saying that inflation is everywhere a monetary phenomenon?

    I mean by your criteria shouldn’t that be ‘demand side inflation is everywhere a monetary phenomenon?’ or maybe ‘NGDP is everywhere a monetary phenomenon?’

  7. Gravatar of Steve Roth Steve Roth
    11. September 2014 at 08:39

    I tell my kids: “You have no idea. Back in the day, when you got in your car, you had no idea if it was going to start. And that was a NEW car.” 😉

    Viz: Tyler Cowen and low-hanging fruit. The car starting every time has very high utility (and utility per dollar). Heated seats somewhat less so.

    Basically: can we measure the value of things in some absolute (cardinal, not ordinal) unit called “utils”? Obviously not. (Though the better research methods used in psychology and social psychology — mostly ignored and unemployed by economists despite Kahnemann/Tversky and their successors — might make a serious go of it…)

    But if we want to think coherently about how economies work, or compare economies over time using numerical measures, we have no alternative but to try. Hence hedonics.

    This all explains why I think econ textbooks should start with the most basic questions: What is value, and what is money? (That’s how my NYU MBA “Accounting for Non-Accountants” class started out, Day One…)

    Modern price theory — based purely on ordinal utility — sidesteps those fundamental (and fundamentally philosophical) questions entirely, so students never consider the unstated assumptions underlying pretty much everything they’re taught. They can’t look at that ordinal-utility construct from outside the construct.

    The primary unstated assumption being: we can’t measure cardinal utility, so it doesn’t exist, or/hence it isn’t worthy of consideration by rigorous economists — even though it’s fundamental to estimating one of the most fundamental measures we work with: the inflation rate.

    It’s a conundrum I certainly don’t have an answer to. But do think econ students should learn very early on that the conundrum exists.

  8. Gravatar of AbsoluteZero AbsoluteZero
    11. September 2014 at 08:47

    Agreed. Even if it’s defined, as Vivian pointed out, it would have to be limited in many dimensions. And using happiness, even if we can clearly define and measure it, doesn’t solve the problem, as people’s happiness is still constrained by the world they live in, and the world, in general, changes. Today we have routine cures for things that were killing even rich people 100 years ago. So while the people 100 years ago, certainly rich people, might be happy with what they had, they surely would prefer many of what we have, but they couldn’t have. They just didn’t know. Most of what we have and do today were not part of their world.

    And even with seemingly very simple and trivial things it can be tricky. In Japan they have these things called Gashapon capsule toys. They’re like gumball machines. You insert some coins, and out comes a small plastic capsule with a little toy in it. There’s a huge variety of such things, from simple key chain type stuff to idol goods to tiny animals. Usually there are rare items mixed in with the usual ones and those are highly valued by collectors. It’s really just a small bit of plastic. They haven’t changed, essentially, for decades. But if you know how they’re made, you’ll know they’ve changed quite a bit. The new ones are made of a much better material that won’t crack or fad over time, and they now use pigment impregnation instead of paint. And the manufacturing processes are completely different. Overall it’s much nicer, even if it’s still just a small plastic animal. So even with such a simple and trivial thing, we cannot just take the price now and compared it with the price from say 30 years ago.

    And even with basic materials it’s not easy. Take copper. Copper is surely basic enough. There are many applications. In some, it has been replaced by other alloys. The reason is the new alloys are cheaper (compared to copper now, or at least 10-15 years ago when these new alloys were invented), and they actually work better than copper. So in a way we cannot use those applications to gauge the change in price for copper, as it’s not used any more. But using the applications also doesn’t work, as these applications didn’t exist 30 or 40 years ago. The world is different, what we do is different, and even for the same functions or purposes we now do things differently.

  9. Gravatar of TallDave TallDave
    11. September 2014 at 09:08

    Good points Scott. As I like to say, you can’t really boil down a multivariate, multidimensional, highly subjective concept into one number that applies to everyone in a socety. I think this is especially true in today’s highly diversified economy — if you buy a lot of tech, you experience a different rate of inflation than someone who avoids tech and prefers, say, renting oceanfront property, or consumes more of something else that has become much more expensive.

    There’s also some very interesting revealed preferences — people who have to choose between cell phones and indoor plumbing choose the former surprisingly often. That and similar results suggest inflation is probably overstated.

    ..and that implies CB policy is actually even tighter than we think.

    On the question of hedonics, you have to consider “relative utility” and evaluate preferences, but in the end there’s still a lot of guesswork.

  10. Gravatar of Major.Freedom Major.Freedom
    11. September 2014 at 09:08

    “You can quibble about the models I chose, but the overall pattern is clear. Inflation has slowed sharply. Or should I say “inflation” has slowed sharply? I don’t seem to be able to make up my mind.”

    This is no good. The overall pattern is clear yes, but reported inflation has slowed sharply in large part because of the hedonic adjustments.

    The most useful way to track price inflation is to select as large a sample of homogeneuous goods whose quality changes are as relatively the smallest.

    Track things like gasoline, oil, sugar, metals, wheat, printer paper, pens, ink, cotton, housing by square foot, salt, bottled water, candy bars, men’s underwear, books, concrete, plastics, etc.

    If we select widely used, generally homogeneous, and generally similar to today as say up to 40 years ago goods, then the pattern is very clear. Price inflation is much higher than what is being reported by the BLS.

    The “truthers” are right.

  11. Gravatar of TallDave TallDave
    11. September 2014 at 09:12

    The most useful way to track price inflation is to select as large a sample of homogeneuous goods whose quality changes are as relatively the smallest.

    That would bias you away from deflationary technological advancement.

  12. Gravatar of Don Geddis Don Geddis
    11. September 2014 at 09:20

    MF/Geoff: “The most useful way to track price inflation is to select as large a sample of homogeneuous goods whose quality changes are as relatively the smallest.

    That’s an intriguing procedure that you’ve recommended, and it will surely result in some answer. But why do you think that answer is useful for anything? By definition, you’ve deliberately excluded exactly the goods that society has improved in quality the most. So the greatest improvements to people’s lives over the years, are left out of your calculation.

    Why would the resulting number, whatever it is, be of any significance?

  13. Gravatar of TravisV TravisV
    11. September 2014 at 09:25

    Dear Commenters,

    Question re: Lorenzo’s new essay:


    Do Robin James, David Harvey, Leigh Johnson and/or Stuart Hall have a name for the philosophies they espouse?

    If utilitarianism isn’t their philosophy, then what is it?

  14. Gravatar of Don Geddis Don Geddis
    11. September 2014 at 09:37

    Another good example: seafood. A century ago, lobster was viewed as worthless by-catch. From wikipedia: “lobster was considered a mark of poverty or as a food for indentured servants or lower members of society … servants specified in employment agreements that they would not eat lobster more than twice per week. Lobster was also commonly served in prisons, much to the displeasure of inmates. American lobster was initially deemed worthy only of being used as fertilizer or fish bait.

    Lobster is now one of the most expensive seafoods, only available to rich elites. But also society’s tastes have changed, and it is now valued in a way it wasn’t before. What is the “correct” rate of inflation, of lobster? What hedonistic utils does society derive from lobster fishing now, vs. in the past? Lobster used to be extremely cheap, and also nobody wanted it. Now the same lobster is very expensive, and also very desired. Has there been “inflation”?

    For commercial fishing, various species became “overfished”. The common species available at the fish market a century ago, are different than the common ones available today. If you wish to demand the purchase of a specific species, the nominal (and real) price may be vastly higher today. But if you measure, instead, in $/calorie of generic edible fish, you’ll wind up with a very different estimate of inflation. On the other hand, now your tuna comes with mercury, and it didn’t use to. How does that change “the real price” of tuna over time?

    It’s useful to have a fuzzy concept of “inflation”, because some portion of price increases are merely nominal increases in the overall price level, and it is useful to attempt to estimate this kind of change. But there is no “fact of the matter” about what the “real” rate of inflation is.

  15. Gravatar of maxk maxk
    11. September 2014 at 09:59

    My impression is that cars have stayed about the same (real) price over the years, while becoming much better technically – more reliable, better gas mileage, safer, more features, etc.

    Perhaps people have a notion about the “right” price to pay for a car, and manufacturers stay in that window.

    I think other technical products have typically become both better and cheaper. TV’s, for example.

    Tennis balls are an example of something that is manufactured but with no real scope for improvement. They have become much cheaper. I think I paid the same nominal price for a can of balls in the 70’s as I do today, certainly same price from 80’s to today. Back then I’d use the same can a lot longer. Major Freedom must not be including tennis balls in his basket, although they fit his definition quite well: homogeneous good with very little quality change.

  16. Gravatar of Nick Nick
    11. September 2014 at 10:05

    Quibbling about the car models you chose, sure the new Honda and the old Acura have the same power, driveability, safety, etc. (although the new car actually is probalby better on all fronts), but what you don’t get from the current Honda is that you would have a better car than your neighbors with the old Acura. Although that just further supports the method in this post of indexing to general wage gains.

  17. Gravatar of Randomize Randomize
    11. September 2014 at 10:29


    I think you have to consider that one is a luxury model and the other is not. Revealed Preference shows that customers get utility (no matter how irrational it may seem) from having an Acura badge instead of a Honda badge on their hood.

  18. Gravatar of Don Geddis Don Geddis
    11. September 2014 at 13:09

    Nick & Randomize: Yes, there is lots of signalling and status-seeking and all sorts of zero-sum relative behavior in the value that humans assign. Sumner even mentioned it in the post already: “if… then… living standards haven’t risen at all“.

    But surely the absolute improvement in quality matters as well. Not all subjective value is only relative to your neighbors.

    Is humanity wealthier than a century ago? Or not?

  19. Gravatar of benjamin cole benjamin cole
    11. September 2014 at 15:15

    Excellent blogging. The key takeaway is that is poor policy for the Federal Reserve Board to obsess over small changes in measured inflation between 1.5 and 2.0 percent. Egads. Obviously the emphasis should be on robust growth—and if inflation is mild, then that is good policy.
    I prefer market monetarism, but if a central bank is going to inflation target, perhaps a band, such as between two and four percent inflation, is a good way to go.

  20. Gravatar of ssumner ssumner
    11. September 2014 at 15:39

    Randomize, I think you should reread this post, as we are talking past each other. I addressed the “perceived luxury” issue.

    But it makes no sense to compare a 1986 and 2015 Acura from the perspective of the BLS’s methodology (hedonics.) If one wants to argue we are no better off than cave men because they were just as happy, you won’t get an argument from me. I have no idea whether we are happier. As I pointed out in my post the guy with a black and white TV felt rich in 1950, and poor today. People want to keep up with the Joneses.

    Dan, As I said in the previous post, major changes in inflation like the Great Inflation to the more recent 2% are pretty obvious, despite the fact that inflation is poorly defined. I probably shouldn’t have said “meaningless,” I get carried away.

    Mike, Friedman later clarified that he meant persistent inflation. He accepted the fact that supply shocks could create temporary price level increases.

    Steve Roth, Good observations.

    Absolute zero, Good points.

    TallDave, Good points.

    Don Geddis, Good points.

    maxk, OK, but the BLS does use hedonics, so I was following their technique.

    Nick, I addressed that point, I agree that people care about comparisons.

  21. Gravatar of Doug M Doug M
    11. September 2014 at 16:32

    I think that inflation is one of the better defined concepts in economics. Fundamentally, inflation is what the BLS says it is.

    It is similar when one of the developers of the IQ tests was asked “what is intelligence?” he sais, “It is what my test measures.”

    Recursive argument see circular reasoning.

    But, if the BLS defines inflation, and the Fed targets that measure, then from the point of view of monetary policy, you are done.

    I think the economics profession has many other poorly defined concepts.
    Such as:
    Aggregate Demand
    and Resources….

  22. Gravatar of Major.Freedom Major.Freedom
    11. September 2014 at 17:21


    “That would bias you away from deflationary technological advancement.”

    How so? Don’t technological advancement affect the production methods of pretty much everything?

  23. Gravatar of Major.Freedom Major.Freedom
    11. September 2014 at 17:38

    Don Geddis:

    “That’s an intriguing procedure that you’ve recommended, and it will surely result in some answer. But why do you think that answer is useful for anything?”


    “By definition, you’ve deliberately excluded exactly the goods that society has improved in quality the most. So the greatest improvements to people’s lives over the years, are left out of your calculation.”

    Yes, that is the point. It is to minimize the possibility of making errors in such “quality adjustments.”

    “Why would the resulting number, whatever it is, be of any significance?”

    Well, I think it should be emphasized that increases in the money supply, while not affecting the prices of everything equally, does affect the prices of everything in the long run, specifically, it makes prices of almost all goods higher than they otherwise would have been.

    I am assuming that in the long run, the average price inflation of the types of goods I listed are roughly on par with the price inflation of goods subject to high quality volatility. Not in the sense of percent or rate of price increase, but in the sense of nominal demand paid for the supplies of whole classes of goods subject to quality improvements. In other words, if we take as wide a range of homogeneous goods whose quality changes the least, the price increases of these goods is a good proxy for what we mean by “prices of goods rising.”

    I think there has to be a constancy in the concept of “goods” if we associate one number to it at one point, and another number to it at a later point. Saying “the price of good X rose 5%” I think requires us to be talking about the same specific good. To me that means we should be looking at those goods that are most similar in quality over time. For all other goods, whose quality significantly changes, price inflation is less meaningful, but we can garner what effect inflation of the money supply is having on those goods by assuming that in the long run, inflation affects prices of all goods in a similar way (I don’t think that is an accurate assumption, so perhaps we can leave it not fully answerable, and rest on the idea that IF the quality of those other goods did not change much like our goods of interest, then they would have increased in price at roughly the same rate.

    Ultimately, we’re trying to say something about money and prices, so to me that means we have to make the real side of things as constant as possible in terms of modelling.

  24. Gravatar of Lorenzo from Oz Lorenzo from Oz
    11. September 2014 at 17:39

    TravisV: I have responded over at Econlog.

  25. Gravatar of A.W. Carus A.W. Carus
    11. September 2014 at 20:41

    “Inflation has slowed sharply. Or should I say “inflation” has slowed sharply? I don’t seem to be able to make up my mind.”

    Bravo! Sometimes you sound almost like Locke; takes a certain amount of courage. (That’s meant as a compliment, by the way, which you may not appreciate as Rorty presumably disapproved of Locke — the ur-empiricist etc.)

    By the way, when at Chicago did you take Fama’s two-quarter finance sequence? (In my day, at least, a decade or so after yours, it was cross-listed in Econ and there were a lot of Econ students there.) He made skeptical noises about “inflation” very similar to yours here, though less worked out; macro only came into the course around the edges. I’ve tried to track down whether he (or anyone else) actually put any of these ideas down anywhere, but couldn’t find anything. It’s been a while, though; do you meanwhile have any allies on this issue in (that corner of) the academic literature?

  26. Gravatar of fds fds
    12. September 2014 at 00:18


  27. Gravatar of James in London James in London
    12. September 2014 at 06:01

    You can file this under “incredible but true”, only the US does any significant amount of hedonic adjustment for CPI. The UK is a distant second, most countries do little, many do none. It is a shock. It means CPI is a very poor measure of anything other than what it measures. A useful tool survey with which to beat Inflation Targeters over the head, nonetheless. Target income or wages!

  28. Gravatar of Vivian Darkbloom Vivian Darkbloom
    12. September 2014 at 06:28

    @James in London

    That’s interesting. It also means that international comparisons of CPI are suspect. It’s like comparing the income of a company using accounting under GAAP with the income of a company using an entirely different set of accounting rules. Of course, the same international non-standardization exists with other measures, too, but I suspect CPI might be even more non-standard than measures of other things. Not only that, but there are significant temporal comparison problems that occur when standards change. Despite all attempts to correct for these things, I’m more and more of the view that accurate accounting and measurement is the achilles heal of economics generally. As far as targeting is concerned, there should be a lot of weight given to targeting something for which the measurement is *relatively* more accurate and reliable over something that might be theoretically superior but practically useless. I also suspect that measurements of income or wages would be relatively more reliable than other targets.

  29. Gravatar of Saturos Saturos
    12. September 2014 at 06:30

    “ECB unanimous in commitment to take more measures if necessary to fight low inflation” – This guy: http://www.cnbc.com/id/101993078

  30. Gravatar of James in London James in London
    12. September 2014 at 07:43

    The goods subject to hedonic adjustment in the UK currenly account for just 7.3 parts in 1000 (page 4). TVs aren’t even in the list. It would be funny if it weren’t so sad. Most important cost for most people is the cost of their internet connection. Not in.

    A rider to Scott’s philosophical musings on the CPI, consumption and happiness: if you are not paying for something you are spending time enjoying or consuming, are you really consuming it for “consumption” measurement purposes? How do you pay for wikipedia? How can the cost of wikipedia be put in the CPI? How can the quality of a visit to the library c.1960 be compared to the results of a google search today?

  31. Gravatar of mpowell mpowell
    12. September 2014 at 08:36

    I think it’s a lot easier to measure inflation on a short time scale though because you get far fewer apples to oranges comparisons. And that’s the time scale that you need for monetary work. I respect your arguments for using NGDP instead, but I don’t think this is really that strong of a criticism of inflation targeting.

    For longer time scales inflation is useful for measuring whether people are better off or not. This is primarily useful for general policy debates. Unfortunately, we are stuck with doing the best job we can for this purpose even though there are substantial issues with the methodology.

  32. Gravatar of EconStudent EconStudent
    12. September 2014 at 15:14

    I don’t understand inflation.
    In school we’re taught inflation is “bad” because it results in an increase of prices.
    In the media we recently hear that inflation isn’t an issue until wages are rising.
    Seems to me if wages are rising with prices, there are no issues as (for the most part) purchasing power isn’t greatly impacted. It seems the “bad” inflation would be when wages aren’t rising.
    I’m obviously missing a lot, but there’s a disconnect between what people teach and what they believe.

  33. Gravatar of TravisV TravisV
    12. September 2014 at 15:56

    Lorenzo, thanks!

  34. Gravatar of TravisV TravisV
    12. September 2014 at 15:57

    Lars Christensen on Prof. Sumner:

    “The Fed’s un-announced 4% NGDP target was introduced already in July 2009”


  35. Gravatar of TravisV TravisV
    12. September 2014 at 16:00

    What are the flaws in this analysis?

    104-page report on the giant bond mega-bubble:

    “Reid says we also need to look at inflation, because if we’re in an environment in which inflation remains, “structurally lower for an extended period it might help justify lower bond yields.”

    Looking deep into history, Reid examines inflation dynamics dated from both 1210 and then from 1800, and he finds that the modern world has generally seen positive inflation as monetary systems became less tied to precious metals.

    Reid writes that “the longer-term investor has evidence that we live in a world with a positive inflation bias and as such must approach the current low levels of bond yields with extreme caution.””


  36. Gravatar of ssumner ssumner
    12. September 2014 at 17:27

    Doug, OK, but then can we agree that the inflation truthers are wrong when they criticize the CPI? Like the Pope, the CPI is infallible.

    AW, Unfortunately I didn’t take that course.

    James, Yes, the US is ahead on that issue.

    On your second point, just assume the cost of encyclopedias has fallen to zero.

  37. Gravatar of Nick Nick
    12. September 2014 at 17:32

    I think it’s really helpful to break the yield curve down when thinking about these things. The curve is pretty flat 10-30 and this kind of very long run analysis makes you wonder about how very certain some investors appear to be that dynamics we see today will persist for decades and decades. Even demographic trends might change substantially.
    As for the low yield on the 5 year and the flatness 5-10, a sweeping view of interest rates throughout history is just too much.

  38. Gravatar of James in London James in London
    12. September 2014 at 21:55

    The point I was trying to make, badly, was: if the price of something is zero then we aren’t consuming it from the point of view of the CPI. It has zero weighting in the basket.

    And worse, what if that thing we are not officially consuming keeps improving, hedonically-speaking, and we keep using it more? It should have more weight in the basket, but can’t.

  39. Gravatar of TravisV TravisV
    13. September 2014 at 04:44

    Noah Smith had two great recent posts:



  40. Gravatar of ssumner ssumner
    13. September 2014 at 05:44

    James, Yes, improvements are hard to deal with, as with any new product. I was considering the case of a shift from paper encyclopedias to electronic encyclopedias. The CPI can handle that shift.

  41. Gravatar of James in London James in London
    13. September 2014 at 08:27

    It can’t handle that switch if the electronic encyclopaedias are effectively free, and worse for the CPI basket, used more. And worse again, hedonically-speaking, are 100x better.

  42. Gravatar of James in London James in London
    13. September 2014 at 08:30

    The basket should really reflect what people do, rather than what they spend mere money on. I walked past a street-sleeper the other morning, lying in his sleeping bag on his cardboard bed, checking the news on his smartphone.

  43. Gravatar of ssumner ssumner
    13. September 2014 at 09:03

    James, The CPI can handle the switch if free, as they weren’t (by assumption) free in the base year. So they are a part of the base year consumption basket.

  44. Gravatar of James in London James in London
    13. September 2014 at 12:50

    Scott, I understand the method, I don’t value the result. Just because I’m not paying for something doesn’t mean I’m not consuming it.

  45. Gravatar of ssumner ssumner
    13. September 2014 at 19:36

    James, I agree, I was making a narrower point about the CPI.

  46. Gravatar of James in London James in London
    13. September 2014 at 22:27

    If I may, one last point. If we end up questioning the value of CPI then the value of the PCE deflator comes into question too. And with that, RGDP itself.

    Hence, the need to focus on NGDP stability. But then you know that, of course.

  47. Gravatar of ssumner ssumner
    14. September 2014 at 05:29

    James, Yes, that’s why I’ve argued that employment is the best indicator of the business cycle, not RGDP.

    Of course even there you have issues (part time work, changes in LFPR, etc.)

  48. Gravatar of TallDave TallDave
    16. September 2014 at 11:29

    How so? Don’t technological advancement affect the production methods of pretty much everything?

    Well remember, you said “The most useful way to track price inflation is to select as large a sample of homogeneuous goods whose quality changes are as relatively the smallest.” For many goods, the smaller the quality change, the smaller the technological impact was.

Leave a Reply