I’ve yet to find someone who has been able to reproduce the claims made by Shadow Government Statistics about the extent to which government agencies are grossly misreporting the U.S. inflation rate. Apparently, neither has the Bureau of Labor Statistics, as detailed in an article by BLS economists John Greenlees and Robert McClelland in the latest issue of Monthly Labor Review.
First, some of the bolder claims by Shadowstats:
The Boskin/Greenspan argument was that when steak got too expensive, the consumer would substitute hamburger for the steak, and that the inflation measure should reflect the costs tied to buying hamburger versus steak, instead of steak versus steak. Of course, replacing hamburger for steak in the calculations would reduce the inflation rate, but it represented the rate of inflation in terms of maintaining a declining standard of living. Cost of living was being replaced by the cost of survival. The old system told you how much you had to increase your income in order to keep buying steak. The new system promised you hamburger, and then dog food, perhaps, after that….
The BLS initially did not institute a new CPI measurement using a variable-basket of goods that allowed substitution of hamburger for steak, but rather tried to approximate the effect by changing the weighting of goods in the CPI fixed basket. Over a period of several years, straight arithmetic weighting of the CPI components was shifted to a geometric weighting. The Boskin/Greenspan benefit of a geometric weighting was that it automatically gave a lower weighting to CPI components that were rising in price, and a higher weighting to those items dropping in price.
Once the system had been shifted fully to geometric weighting, the net effect was to reduce reported CPI on an annual, or year-over-year basis, by 2.7% from what it would have been based on the traditional weighting methodology. The results have been dramatic. The compounding effect since the early-1990s has reduced annual cost of living adjustments in social security by more than a third.
And here’s the response by Greenlees and McClelland:
To begin, it must be stated unequivocally that the BLS does not assume that consumers substitute hamburger for steak. Neither the CPI-U, nor the CPI-W used for wage and benefit indexation, allows for substitution between steak and hamburger, which are in different CPI item categories.
Instead, the BLS uses a formula that implicitly assumes a degree of substitution among the close substitutes within an item-area component of the index. As an example, consumers are assumed to respond to price variations among the different items found within the category “apples in Chicago.” Other examples are “ground beef in Chicago,” “beefsteaks in Chicago,” and “eggs in Boston”….
The quantitative impact of the CPI’s use of the geometric mean formula also has been grossly overstated by some, with one estimate exceeding 3 percent per year. It is difficult to identify real-world circumstances under which geometric mean and Laspeyres indexes could differ by such a large amount. The two index formulas will give the same answer whenever the prices used in an index all change by the same percentage. The bigger the differences in price changes, the more the Laspeyres index will tend to exceed the geometric mean. For the growth rate of the Laspeyres index to exceed the growth rate of a geometric mean index by 3 percentage points, however, the differences in individual price changes have to be quite large.
To see this point, consider another very simplified example. Suppose that the CPI sample for ice cream and related products in Boston consisted only of an equal number of prices for ice cream and frozen yogurt and that, between one year and the next, all the prices of ice cream in Boston rose by 8.6 percent while all the frozen yogurt prices fell by 4.2 percent. In that case, the geometric mean estimate of overall annual price change would be 2.0 percent, only slightly less than the Laspeyres estimate of about 2.2%. In order to come up with a difference of 3 index points, one has to assume a much more dramatic divergence between ice cream and frozen yogurt prices than the one hypothesized. For example, if ice cream prices rose 30 percent in one year, while frozen yogurt prices fell by 20 percent, the overall geometric mean index would still rise by 2 percent, but the Laspeyres index would rise 5 percent, for a difference of 3 index points. However, such a large annual divergence would be quite uncommon within CPI basic indexes– between ice cream and yogurt, between types of candy and gum, between types of noncarbonated juices, or between varieties of ground beef. Moreover, for a 3-percentage-point divergence to continue year after year, the divergence between the individual component prices would have to continue to widen. For example, if, by contrast, during the next year ice cream prices increased by the same amount as frozen yogurt prices, then the two index formulas would give the same inflation estimate for that year. Although such a divergence might plausibly occur in one component for 1 year, it is beyond belief that such sharply divergent price behavior would continue year after year across the whole range of CPI item-area components.
Finally, and most importantly, there is rigorous empirical evidence on the actual quantitative impact of the geometric mean formula, because the BLS has continued to calculate Laspeyres indexes for all CPI basic indexes on an experimental basis for comparison with the official index.
These experimental indexes show that the geometric mean led to an overall decrease in CPI growth of about 0.28 percentage point per year over the period from December 1999 to December 2004, close to the original BLS prediction that the impact would be approximately 0.20 percentage point per year.
There’s much more in the BLS article on this and related questions such as hedonic price adjustment and owner’s equivalent rent.
Why do people continue to give credibility to an operation like Shadowstats? Now that’s something that I’d like to hear explained.
Technorati Tags: CPI,
You ask, “Why do people continue to give credibility to an operation like Shadowstats? Now that’s something that I’d like to hear explained.”
Most importantly, many people have a world view about politics, government, and the economics profession. They deny the value of traditional economic analysis, professional training, and peer review.
As a result, they do not appreciate the importance of normal academic practices like sharing data and providing results that can be falsified if wrong.
Since so many pundits and fund managers find these results helpful in making the case for their trading positions, the results get wide publicity. Mainstream media, with two notable exceptions, have been cautious in embracing the Shadow Stats results.
I also plan to write about the BLS article and I am delighted that you have confronted the topic head-on.
>Why do people continue to give credibility to an
>operation like Shadowstats? Now that’s something
>that I’d like to hear explained.
Easy – but first, why would anyone assume that the BLS has any credibility either ? Shadowstats is not the only person/group who have noted the glaringly obvious fact the the BLS numbers are wholly out of whack with everyday public experience.
You should judge a person/group by their body of work. And it is absolutely clear that the BLS changes always led to biases towards less CPI. There is no way the what we see around us corresponds to a 4% give or take inflation that BLS purports.
Shadowstats gets credibility in that their model reproduces the BLS numbers prior to the changes, and in that the numbers they claim for now seem much more in keeping with people are experiencing. Can I confirm the model from my desk – no, but neither can I judge the BLS’.
Using the BLS as the source to judge Shadowstats criticisms of the BLS is silly.
The reason people give it credibility is because thats what they feel the situation is on the ground, when you’re paying 4+ dollars a gallon for gas, when heating bills soar, when the cost of bread rises. People can’t be expected to believe that inflation is just 4 or 5%. All statistical models have flaws, they make numerous assumptions and the 5.6% CPI is just the mean, we don’t hear about the 95% confidence interval it lies in. It’s always a good idea to average the shadowstats data and the BLS data too get a fairer picture.
Very misleading. This isnt a debunking of shadow stats, this is a defense of the calculation of substitutions within the CPI. They might not calculate substitutions to the degree that shadow stats assumes, but why would they do any substitutions. If you are trying to measure the rate of inflation then you should not add in a measurement for the rate of substitution. Conflating two different variables is simply bad science.
Why would you allow for substitution in calculating the CPI? In other words, why does the BLS argues that it can determine baskets of goods that have different price variations yet provide the same standard of living?
Isn’t it great that ALL changes made to CPI calculations by the government always results on lower CPI (substitutions, quality imprisonments etc) so that there is no need to increase payments for retired workers? What a system!
1. It is highly suspicious’ that when they recreated the new CPI, it reduced the amount of money government needed to pay people (of course, government shouod NEVER have indexed some spending to CPI in the first place, but that is another discussion).
2. Because when people say “… such a large annual divergence would be quite uncommon within CPI basic indexes”, they sound just like the Hedge fund managers right before a ‘Black Swan’ comes along and destroys all their models… Making predictions about the future using historical ‘past performance’ behavior data for something as significant as future price estimates is simply a “fools folly”. “Yes”, most of the time they will be right, but when they are wrong, they will be VERY VERY VERY wrong (as experts always inevitibly are).
3. Trust is in short supply these days
I am stunned at these comments. The BLS tracks imaginably more prices than any one consumer, and doesn’t depend on a gut check (“inflation is WAY over 4%, just look at gas prices!”) but actually *computes* the number (and computes confidence intervals). Methods used are all posted online, and conform to international best-practice. (Guess what, people actually do substitute. Come on guys, this is econ 1. To see the upward bias of the “ordinary” index, check out the superlative index.) Academic experts are free to challenge these methods, and from time to time do – although it should be noted that the Boskin commission relied heavily on the findings of internal BLS researchers. These outside-expert, PhD economist critics would be the first to laugh at the sorts of objections raised in these comments, and by these bond-trader types. Why don’t you read the MLR article yourselves, and become informed? Shadowstats is completely and utterly demolished. Bill Gross is also trashed. You should be ashamed of your ignorance.
The BLS authors could perform a useful service if they shows the CPI calculation prior to the changed methodology and the current, and allowed people to compare the net differences for themselves.
Shadowstats claims to perform this calcuation. Is it incorrect? If not, all the words these fellows expend are meaningless, and as someone noted above, explations for changed methods that many are rejecting.
CPI has been significantly altered over the past twenty years through “pollyanna creep” and we all know it despite any hair-splitting rationales
frankly, the entire debate makes little sense because there is no ‘average price’ in the economy. the whole concept is nonsensical. why should adding up the prices of cars, haircuts and pork chops result in a number that makes any sense whatsoever?
however, as to the alleged ‘debunking’ of shadowstats by the BLS, i don’t buy that either (especially not when the BLS itself does it!). Williams claims to simply use the same methodology that was used prior to the changes introduced by the Boskin commission. the fact that this commission was a political project designed to lower the government’s COLA outlays and make the government’s economic policies look better is to me beyond doubt. many of the changes that were introduced are highly subjective (like hedonic indexing – how does one put a value on increasing CPU speeds for instance? the so called ‘real’ numbers from BLS and Commerce in this particular area are so distorted as to be way beyond what one would accept by mere application of common sense).
Anyone up for yet *another* area to slam the BLS on?
Take a look at how the surveys measuring median salaries by profession have “evolved” from 2000 to today.
There has been a wholesale aggregation of professional categories that in essence make it impossible to perform an apples-to-apples comparison of salary changes since 2000.
Wonder why that is?
By polluting the category definitions, the BLS has utterly castrated the longitudinal value of their surveys.
And take a look at how non-stop redefinition of metro areas has also made it impossible to judge employment growth (yeah, right) by metro since 2000.
It takes more of a leap of faith to believe that the BLS is operating in the interest of the public than it does to believe that they are intentionally crippling the longitudinal value of their data.
The first thing that stood out to me about Shadow Stats is that William’s figures are usually a constant spread from the official government figures. The derivatives of the graph almost never vary. In order to figure out why this is so, I went hunting through his “about” pages to find his methodology. It turns out that this is not an accident; the similarity of the graphs is the result of his methodology.
It seems that John Williams does not actually recompute CPI etc using the old techniques. What he does is take the current measurements, and then offsets them with a correction factor. This correction factor is derived from some internal government document which was created around the time that the measure is changed (e.g. someone writes a memo claiming that the new CPI will be 2% lower). Hence, his system has the same problems that the new government measurements have, namely that they cannot meaningfully be compared with the original government measurements. They are simply an alternate metric, and one of dubious quality.
This is a really shame. We desperately need a site that actually does what Shadow Stats claims (but fails) to do. If I hear another mathematically illiterate economist/pundit numerically compare 2000s CPI with 1980s CPI (or 2000s unemployment measurements with 1930s unemployment measurements), my head is going to explode.
Why doesn’t the fed (and other policy makers) demand the original statistics? How can you possibly make sound decisions an an adjusted number without knowing the “raw number”?
What is the “unadjusted” number? Do you know? Why not?
Shadowstats (or organizations like it) will always exist to fill a data need. They gain credibility because government policy makers choose not to publish a number they have available.
More data should be good, then we can focus the discussion on the “is the data meaningful”.
Why allow substitution? Because consumers substitute when the price levels of substitute goods diverge.
For example: suppose I buy wool pants and cotton pants. If the price of cotton pants rises and the price of wool pants falls, stays the same, or rises by less than cotton pants, I would (most likely) be better off substituting some of my usual cotton-pants purchases for wool pants purchases. That would affect (indeed, reduce) the price change I actually experience.
Using substitution helps the CPI reflect the price level changes I actually experience (instead of the price level increase I would experience if I were an idiot who refused to substitute).
Yes, you’re right. We have learned no economics since the 1970s when the “old” ways were being used. There have been no advances in measurement. The BLS should use the old method of measuring homeowner costs; that the rest of the world has rejected these old methods only PROVES that these old methods are watertight and ironclad. Because after all, the old methods show that inflation is higher, and we KNOW that higher inflation is the case — because we just know, that’s all. (Don’t confuse me with data.) All the PhD economist experts who insisted that the BLS change its methods are lunatic idiots, morons, imbeciles. We should definitely put faith in a shoestring operation which “corrects” the CPI in a way that implies that we were in the worst depression imaginable only a few decades ago.
You guys should think about something. Undermining confidence in official statistics leads to a dismissal of DATA. If data does not inform policy, then only politics will.
If you have a valid criticism of the CPI, try to write a serious paper, present it before experts, and see if you have a case. So far, no one here has raised any credible issue; it is the wildly uninformed misleading the even more wildly uninformed. Please become informed, and actually read the MLR article before commenting.
Measurement is an active area of research in economics. There are top PhD’s writing papers in measurement all the time. Their papers are peer-reviewed and published in academic journals. Quoting Marginal Revolution, “Being pro-science means being pro economic science.”
Or: see Jeff’s comment above.
Jesse — the BLS does show what the CPI would be without the revisions. They publish it every month. It is called the CPI-RS and can be downloaded at BLS.
What it shows that the difference in the CPI as it is now calculated and what it would have been under the old methodology would be about one percentage point.
The bulk of the difference stem from the use of
home owners equivalent rent. Over the long run rents and home prices move together which is what you would expect since they are both driven by the same factors — land and construction cost on the supply side and income on the demand side.
When they changed the system the experts did not expect it to make much difference. Many wanted the change because it took mortgages rates out of the CPI and eliminated the perverse effect of the Fed tightening causing higher reported inflation through the interest rate impact. But in fact we got a housing bubble that no one expected.
I lived through the inflation of the 1970s and in the 1970s we got the same thing of people claiming that the inflation data was wrong.
Yes, of course. Substitution does reduce the impact of price increases. This is not disputed.
The question is, do you get the same utility from wool and cotton pants? You may say yes, I may say no, and the BLS decides for the both of us.
Further, the idea that inflation is best measured once people have adjusted to price variations is not trivial. Why not before? Instead of reflecting the price level I “actually” experience (as decided by the BLS), the CPI would reflect variation in purchasing power if I bought the same things as last month (yes those “same things” as decided by the BLS).
Nice post JDH; however, you just had to know that it would bring out the conspiracy theorist moonbats in droves. I’ll be hanging out on the grassy knoll with my fellow members of the Illuminati, talking to the aliens that escaped from Area 51.
Two wrongs do not make a right. BLS and Shadowstats are both wrong. The BLS has many more resources to do it right. Shadowstats is a one person(??) operation, I don’t expect Shadowstats to do the job that BLS should be doing, but I do appreciate Shadowstats as a gadfly.
I look at CPI with suspicion because of ‘guilt by association’: The Fed points us to ‘core’ and asks us to ignore ‘headline,’ the GDP deflator gives us unbelievable results at times (Q2 08), OER did not reflect the rocketing home prices in the run-up and BLS includes auto prices but not home prices, and, a bit further out, the birth-death adjustment to the establishment employment report was unbelievable over these last 12-18 mos. and the stopping of reporting of M3 raised suspicions.
Sorry, BLS, you use suspect assumptions on housing and employment and ‘hang out’ with suspect folks (The Fed). That is why we do not accept your pronouncements at face value.
why ? Two words: elaine chao.
For those that read, that’s Mitch McConell’s (R-KY) wife. See her performance at United Way for further inspection of credibility. Not saying this validates Shadowstats just that a certain amount of curiosity about elaine chao’s appoinment and work is completely justified.
Short version: Williams versus Elaine Chao should be the credibility comparison. Not BLS vs SS.
Using real micro price data from a country that is not the US (the same data used in the construction of the CPI), I computed what would be the average monthly increase using arithmetic and geometric mean and the differences are absolutely brutal: Arithmetic – 0.8809%/month; Geometric 0.1998%/month. These numbers do not use any type of weighting, but the weighting would be common to both means and they do not take into account product quality or substitution, but again this would be common to both means.
The bottom line here is that Greenlees and McClelland from the BLS are not correct, the differences between geometric and arithmetic mean can be very large. Base on my numbers, the difference can be of the order of 1 to 4 (arithmetic mean 4 times larger than geometric mean).
I asked, “Why do people give credibility to an operation like Shadowstats?”, to which many of you offer answers along the lines of, “because we don’t believe the BLS.” But that’s not a logical answer. Just because you doubt A is not a valid reason for accepting the truth of B.
And as for the question, “Why should we believe any BLS economists?”, here’s one answer: because you can check their math. Let me walk you through how Greenlees and McClelland’s ice cream vs. yogurt example works. The arithmetic mean is
0.5(1.086 + 0.958) = 1.022
whereas the geometric mean is
(1.086)(0.958)^0.5 = 1.020
for a difference between the two measures of 1.022 – 1.020 = 0.002 or 0.2%. For those of you whose position is that “we believe Shadowstats because it’s consistent with what we see around us,” here is your homework assignment. Give me some plausible numbers, which you think are consistent with what you see around you, under which geometric averaging could reduce the reported inflation rate by something on the order of 3% per year, year after year.
And please show your work.
Anonymous, let’s see your spreadsheet.
Un Estran wrote:
>The question is, do you get the same utility from wool and cotton pants? You may say yes, I may say no, and the BLS decides for the both of us.
The superlative index answers this definitively. The data show how people did in fact substitute. (That index is only available with a lag, though.)
>Further, the idea that inflation is best measured once people have adjusted to price variations is not trivial. Why not before? Instead of reflecting the price level I “actually” experience (as decided by the BLS), the CPI would reflect variation in purchasing power if I bought the same things as last month (yes those “same things” as decided by the BLS).
You have in mind a Laspeyres index, which is the old index that is (way) upward-biased due to substitution. The current CPI index is Laspeyres except within very narrowly-defined items (e.g., apples in New York), where unit-elasticity substitution is assumed. I believe this is discussed in the MLR article, which no one apparently has any interest in reading, evidently willfully preferring to remain ignorant, but still very confident in their beliefs.
Please everyone, read my previous post regarding DATA. Don’t mislead people about the CPI. In fact, you might want to guard your tongue entirely, and tell others to do the same, unless you know what you are talking about.
Tyaresun: what is your data source? Read the article. Shadowstats is totally destroyed by a very simple computation.
JG: the BLS does not take a stand on whether one should ignore “headline.” OER shouldn’t reflect rocketing home prices, unless these are reflected in increases in the price of housing services; read up on “the user cost of housing” before you simply assume that house prices should be plopped into a CPI. (It is not as if current homeowners whine when prices rise, for example. You have to address all those issues and get all of them correct. It’s more complicated than you might think.) Using new car prices instead of user cost of autos is an approximation, but not as bad as in housing, since autos depreciate so much faster.
My spreadsheet is 2 million plus observations!!
“It is useless to attempt to reason a man out of a thing he was never reasoned into.” – Jonathan Swift.
Jim – you had to expect these reactions. Nevertheless there is much wisdom in Jeff and Randy’s comments, and of course, in your initial post.
Another interesting post on inflation measurement; having been initially sceptical, I am gradually being convinced that the cpi is a reasonable measure of the cost of living. I was particularly impressed to read that quality adjustment makes so little difference.
I would still question, though, whether a cost of living index is the measure of inflation that is most useful. I would favour a measure of the underlying change in prices, calculated by a stochastic approach. Much of the scepticism about the cpi arises because the idea of the cost of living (at a certain welfare level) inevitably involves more subjectivity. To some extent, by committing to index benefits, and then delegating (albeit with the occasional “nudge”) measurement of the cost of living to the BLS, the politicians are avoiding responsibility for denying benefit recipients the “proceeds of growth”.
Also, the article says that the BLS methods continue to be reviewed by outside commissions and advisory panels, without actually saying what those organisations are. As far as I know, there is no independent US statistics watchdog; the introduction of one should support public confidence in US statistics.
Finally, I wonder whether the article makes too much of the fact that the price changes for steak and hamburger are not geometrically averaged. Presumably – assuming that there is substitution from steak to hamburger – price changes for steak are given less weight over time as the reference basket changes.
I have no stance on which mean is the correct one, arithmetic or geometric, but the differences I showed are based on real price data.
I also know that in the computation of the CPI, there are two layers of computation. First, for narrowly defined areas (think of zipcodes or counties) the prices indexes are calculated using a geometric mean. Based on these “local” price indexes, the overall price index is computed based on arithmetic averages of the local indexes.
Anonymous, there’s no reason to hide both your name and your data from us. Surely you could give us one annotated sheet of a subset of your calculations to illustrate numerically how geometric and arithmetic averaging can produce big sustained differences.
This is a really shame. We desperately need a site that actually does what Shadow Stats claims (but fails) to do.
to my mind, the pretext of what the BLS is attempting to do is to model reality as closely as possible.
but this frankly is a futile activity — the pre-hedonic, pre-substitution CPI was well within the margin of error. greenlees and mcclelland are in effect arguing as much, minimizing the nature of the difference as minor or non-existent.
if one accepts their rationale, the question then becomes: if the error around either method of calculation is far greater than the difference between them in any given month, why have made the change at all?
i deeply suspect (but can not prove) that the aggregation of small differences over time are in fact intended to substantially minimize COLA in government subsidy programs. even a cursory reading of history shows that such government agencies as the BLS are across time and place reliably vulnerable to political pressure, regardless of whatever ennobled image of applied economic science we would prefer to be true.
the BLS could easily rebut by producing the original CPI series alongside the revised CPI series. that is doesn’t opens the door for shadowstats. hopefully someone better-equipped, inside or outside the BLS, takes up the challenge.
the BLS could easily rebut by producing the original CPI series alongside the revised CPI series. that is doesn’t opens the door for shadowstats.
sorry — i missed spencer’s comment above. spencer, thank you for the illumination!
is the position of the pro-BLS contingent here then that the changes made over time to the CPI function not to introduce significant changes over the longer term but to merely more accurately reflect reality?
gauis marius — don’t you read the other comments.
The BLS does publish the original CPI data along side the revised CPI exactly as you propose.
It is called the CPI-RS and can be downloaded at BLS.
see my comments above.
Randy – good comments.
Rather than simply wondering what John Williams of ShadowStats was up to and what he is doing, I asked him.
He says that there is nothing in the BLS article that has not been raised and answered before. He will prepare and publish an article on the BLS positions.
To all you blowhards that accept that the BLS has the last say in debunking criticisms of itself (WTF?), wait for what ShadowStats has to say.
gauis marius — don’t you read the other comments
sorry spencer — i missed yours on the initial read-through.
but i am a bit confused.
In 1999, the BLS released a new series, the Consumer Price Index Research Series Using Current Methods (CPI-U-RS).(1) The CPI-U-RS is an index of inflation from 1978 to the present that incorporates most of the improvements in methodology made to the CPI-U over that time span into the entire series. Among other improvements, the CPI-U-RS makes quality adjustments for the aging of housing units and for the prices of used cars, personal computers, and televisions, and it employs a geometric mean formula to account for consumer substitution within CPI item categories. Although the research series has some limitations, including being subject to annual revisions, the BLS states that it is the most detailed and systematic estimate available of a consistent CPI series.
CPI-U-RS is not the old series. it is the new methodology applied to the old data.
is there a different CPI-RS you are intending to reference?
lastly — i don’t think anyone seriously debates, do they, that the old computation resulted in a higher CPI? that’s really what a proper reading of the CPI-U-RS demonstrates.
it seems to me that one can argue forcefully that there is a wide gap between shadowstats methodology and the widespread perception of what it is doing. and one can argue perhaps less convincingly that CPI-U is more “accurate”, and that is was not intended to reduce COLA.
but what one cannot do is argue that CPI-U did not effectively result in the diminishment of COLA in government entitlement payments. it did, per dr. perry and data he cites from the BLS itself.
i admire those who believe that academic pursuit of truth rules the day even in government departments run under political administrations. but my reading of history suggest that, even if CPI-U is “better”, it never would have been enacted had it not also reduced government welfare payouts. as far as that goes, regardless of its theorietical merits as an abstraction, that is the purpose of the implementation of CPI-U.
As Pater pointed out above, a national-average number is interesting, but meaningless, as my personal rate of inflation is different from everyone else’s. The rate of inflation for my age-group, city, education level, etc., etc., etc., is different from all others, and so on.
Having said that, I am inclined to believe that an “official” government inflation rate constructed with a minimization preference or bias is in the government’s interest of keeping cost-of-living adjustments, especially for social security, as low as possible so as to keep budget deficits and national debt borrowings below what they otherwise would be.
Which comes back to the concept of developing inflation rates based on demographic factors. Taking retirees as an example, if one were to reasonably construct a “retiree” basket of goods and services and calculate the inflation rate for that basket over the last 20 years, intuitively one might expect that the Retireee Inflation Rate would be considerably higher than the average, given that a theoretical retirees’ basket of goods and services would contain elements that demonstrably have experienced greater cost increases (health care, property taxes, utilities, food, gasoline, insurance, entertainment, travel) while goods such as computers, electronics, furniture and others that have deflated over the years, would represent a proportionally smaller component of a retirees’ basket of goods and services compared with other groups.
Hence the dilemma and hence the uselessness of an “average.”
The informed versus the uniformed. Much laughter was the result.
Try this: The informed against the informed justifies the uninformed.
Read 3 economy blogs and you’ll get 3 different viewpoints. Depending on your agenda you will validate your opinion. Add in data and equations and you get informed dissent among the informed.
Can we agree the science of economics is worthy of skepticism. Send all the spreadsheets you want but your average person will pay their bills every month and know prices have risen faster then the CPI claims.
I’ve read enough of the Greenlees McClelland report to see that it is articulate, logical, and clear.
The primary fuel for the shadow movement and its adherents is economic illiteracy. It’s that simple.
From the G/M report:
“criticisms of the CPI may arise from a distinction between the express goals of that index and the uses that some critics wish to make of it.”
And so it is with so many other examples.
Witness the wretched mangling of the GDP deflator concept in recent blog discussions.
This site plus Economist’s View and Macroblog were heroic in defending the truth against the dissembling barbarians. But it is a hard fight.
There is a reason, this data is confidential. You can ask Pete Klenow or Emi Nakamura to perform the same exercise I did but with US data. My guess is that are not allowed to do that.
This is depressing. Most CPI critics are so, so sure of themselves, but they don’t produce any coherent arguments or data, or even bother to understand the problem. They mostly just claim over and over that it is just *obvious* that inflation is higher. Why is it obvious? How would you know?
Sorry, anonymous, I just don’t see the need for smoke here. This is a numerical claim about the difference between a geometric and arithmetic mean. You and Shadowstats should be able to support the validity of that claim without resorting to reference to 2 million secret numbers.
Greenlees and McClelland give some simple numerical examples to illustrate their point. Why can’t you?
I agree. It is depressing. But don’t reach for the blade just yet.
There should be a counter-revolt against such a massive cultural wave of vacuous criticism directed at substantive economic analysis.
I’ll get back with plans.
All I can say as a citizen my observations are the following.
In the last year-
1. My food is 10-20% higher
2. My gas is over 20% higher.
3. My haircut is 10% higher.
4. My car repairs and parts are 10% higher.
5. Any furniture I look at is 5-10% higher.
6. My clothing is about 5% higher.
7. My healthcare is 7% higher.
8. My heating was >10% higher this winter.
Given the high FREQUENCY of transaction for several of those, inflation is running about 10% in my world. Not 2%!
That’s the facts Jack!
i wouldn’t presume to speak for others, but i would note that there seem to be different debates going on here.
many of the critics of CPI-U are essentially arguing that — regardless of what you think of shadowstats — the series isn’t reflective of reality. this is what G/M are pointing out when they say “criticisms of the CPI may arise from a distinction between the express goals of that index and the uses that some critics wish to make of it.”
but some of the advocates of CPI-U are also missing much of the meat of the subject. the much-defended improvements in the “accuracy” of the metric are really quite beside the point, as is the academic merit of CPI-U and/or its mathematical transparency. many concepts of great academic merit are ignored in policymaking. why was this one implemented and applied to COLA?
the fact is that CPI-U results in lower inflation readings than the old CPI metric — and that is why political decision makers have allowed it to be adopted into general use, as it significantly diminished real government outlays over time.
this would be very, very far from the first time some meritorious academic production was cherrypicked to reach policy goals. one is right to wonder if, had meritorious adaptations not been available, some devoid of merit would have been adopted.
none of this makes shadowstats “right”, but the critics have a point on their own terms even if they are not framing the debate well. the science of CPI-U could even be presumed unassailable (which of course it isn’t); it only saw the light of day because it was politically expedient. this rightfully engenders skepticism of the process and its product.
“many of the critics of CPI-U are essentially arguing that — regardless of what you think of shadowstats — the series isn’t reflective of reality.”
I heard exactly the same argument in the GDP deflator debate.
The problem with the argument is that reality is open to analytical decomposition. Different people have different experiences based on different consumer selections and different ways of remembering price information based on their experiences. Reality is subjective to the extent that that individuals experience different samplings of reality. The most vociferous may experience the least pleasant realities. But some of us don’t. The G/M piece here has quite a good section on this.
Well said Gaius.
Now about your use of capitalization.
How much weight to these categories get in your total expenditures? Are there other categories that have stayed the same or gone down? If you actually did a careful weighted average of all the prices you face, what number would you get for *your* personal CPI?
And how do these numbers compare with the BLS reported increases for these categories? The BLS data used to compute the CPI also show very high increases for things like fuel. Do you see any large divergence between your numbers and theirs that would lead you to think they are cooking the books somehow?
What about the BLS’ use of ‘hedonics’? Take a look at their index for the cost of televisions. I think the index has dropped about 90% over the last ten years. That’s absurd. I paid $550 for a nice Sony about seven years ago, and I paid $850 for a decent flat-screen a few months ago. The BLS may have good intentions, but it’s pretty clear to me that they are way off base with regards to the actual cost of living. OER is so BOGUS. Hedonics is bull$hit. A lot of the substitution is bogus. Just give me a basket of goods and tell me how much that same basked of goods costs through time.
Fine!! I can just repeat what I did.
Using a micro price CPI dataset, I computed the monthly price variation for each item (item here means product in the store, think of a gallon of milk in a grocery store).
With these monthly variations I used the “ameans” command in STATA which produces the arithmetic, the geometric and the harmonic mean.
The result for the arithmetic mean was a 0.8809%/month.
The result for the geometric mean was 0.1998%/month.
In the past, the BLS has allowed researchers to use their micro price datasets for academic research (Bils and Klenow, Klenow and Kryvtsov, Nakamura and Steinsson, are examples). If the BLS has nothing to hide, then it should allow people to run these experiments with their raw data and answer this question very clearly.
In Europe, around 15% of prices change every month. Of these, 60% are increases and 40% are decreases. The average price decrease is 10% while the average price increase is 8% (Dhyne et al.)
Geometric average (please correct me here if I make a mistake):
This is obviously a back of the envelope calculation, but it illustrates how such differences can occur.
This is really much ado about nothing.
JDH/BLS/academic establishment is arguing that the CPI measures what its model measures.
But has this model any relation to reality?
In the U.S. counts food as only __ 8% __of the CPI index. Whereas, it counts for about 10% in the United Kingdom, about 15% in the rest of Europe and more than 18% in Japan (food@home = 7.66 in BLS)
Interestingly, if you look at the proportion of U.S. household spending on food, by income quintile, all but the top 20% of earners spend at least __ 20% __ of their paychecks on food.
In the CPI, Education is 6%, health is 6%.
Who has that kind of pay? One which requires only 6% for healthcare? University professors and economists?
For most of the population, the CPI has become irrelevant.
I know it from personal experience. I keep accounts of my spending, and I see the rise in the monthly bills.
The only expense that has come down — since 1997 — is the telcom/internet and electronics/hardware related expenses.
Even infrequent items. A car – what used to be 12K is now 17K. That may be a better car, but to me, I still want the old one at 12K. It’s not available
Clothes. I used to buy pants/shirts on sale for $10. You wont see that kind of prices on sale today. You’re lucky if you see the same quality $19 pants or $15 shirts.
The CPI measures what it is modeled to do. But it does not model the experience of mine or 95% of the population.
You want to know what the real inflation rate is? That’s easy. Go find your nearest retiree or college student living on their own and ask them. The closer one is to “the ground”, the more one notices that increases. And no, inflation is nowhere near the reported government figures. It is much, much worse.
Abacus, you are right that there has been inflation since 1997. In fact, the CPI shows a 37% increase since 1997. So it makes sense that most items have gone up significantly.
I bought a car for 11K in 1990 and one for 16K in 2006. That means that using the CPI to adjust for inflation, they cost about the same. But here are a few things that are better about my 2006 car: airbags, anti-lock brakes, stereo, cd-player, more powerful engine, alloy wheels, 4-wheel disc brakes, better suspension. And yes, there *were* cheaper cars available in 2006. So overall, I’d say inflation in car prices has been lower than the CPI.
I’ve criticized Shadowstats for producing implicit predictions that do not match reality. Any model should be carefully reality-tested before being rolled out, and Shadowstats did not in my opinion do that.
But I have to say, BLS could have–and should have– produced this paper a lot earlier.
Also, it’s… well, maybe not unfair but disproportionate to criticize a member of the public who challenges government statistics. We, the public, need to feel free to question. How huge sums of money are deployed hangs on the third significant figure of the inflation calculation, so close scrutiny is indicated.
Furthermore, we do not have the resources to hand that the BLS does, and so if we come to erroneous conclusions, that’s not exactly unlikely.
I will be reading Greenlees and McClelland slowly and seriously. I will also look for John Williams’ response.
Thanks for posting the link.
“Geometric average (please correct me here if I make a mistake):
You used three parentheses at the beginning of your equation? Just spitballing here…
People give credibility to Shadow Stats because nobody really believes you can buy books for ten cents today, despite what the hedonic regression adjusted price of books in the CPI currently is. People tend to ignore all the BS arguments for why the model is good, and focus on whether it produces accurate results, which is this case the answer is obviously ‘no’.
I would agree that there is nothing particularly wrong with the model: it’s designed to understate inflation for the government’s benefit, and it does an admirable job of it. However, when the BLS continues to insist it’s a measure of real inflation, it just gives sites like Shadow Stats more credibility. It turns out lying about something which is obviously false discourages people from trusting other things you claim, and at this point claims from the BLS are pretty laughable. At this point their credibility is pretty shot, and that’s a problem for an organization tasked with producing reliable information about the economy.
BLS should publish the raw numbers, and let organizations like Shadow Stats compute the resulting CPI’s, IMO. Then at least there wouldn’t be the assumption of intentional deception which corrupts all their data and assertions currently.
There are two issues going on regarding Shadow Stats. One is the “who most closely models reality battle”. As you said, this is very difficult to measure. Furthemore, as you point out, what is important is how this stuff gets used in policy decisions.
How do we determine that a particularly CPI is bad? Well, if you look at policy sites, they all have arguments of the form “CPI is much less now than it was in year X when it really sucked, so things should not be as bad now”. This is therefore the underlying crime of changing CPI, and is why we need a site that achieves what Shadow Stats claims it achieves.
Is the old CPI better or the new one? Hard to tell, and I do not care. But the fact that CPI measurements are a moving target means that any analysis of comparing year X to year Y is meaningless. And if you look at any pundit talking about the economy, everything comes down to cross-year comparisons.
For example, on Matt Yglesias’ site the other day, there was a time series graph comparing CPI over the years. This time series graph was built up from numerical values computed from changing CPI metrics. Whomever made this graph is so mathematically illiterate that they should have never been allowed to pass a freshman math class.
But this is the type of information that is used to make policy. Policy makers think that because it looks mathematical, it must be meaningful. But it isn’t.
I made some serious criticism here (see my second post, which I can’t deep link), and I’d expand them into an academic paper if I had the time.
Long story short, they will drop the price index because of alleged quality increase, but rarely increase the index because of quality decreases. But paying the same for a worse product IS inflation, exactly what needs to be measured.
To measure quality changes, you would need to do laboratory testing, which the BEA and BLS do not do. I have noticed numerous quality decreases but can’t find any quality increases except in computation and its output. My cereal boxes are harder to hold, shorter caps on soda bottles are harder to open, pepperonis and soup have more filler content, wait times are longer, and on and on.
I also can’t find their insulin price index, which would be an ideal measure since it has steady, inelastic demand; many buyers; and it’s impossible to degrade it in response to higher costs.
There’s more: allowing for substitution again misses the whole point of measuring inflation: what I get for my dollar. When I can buy with less consumer surplus, that’s inflation. When higher prices make me switch to a different product, that means my consumer surplus dropped, and to fail to account for this is to understate inflation. In the extreme, if food became so expensive that everyone had to literally eat dirt, as some have to in Cuba and Haiti, you would be forced to call that “not inflation” because “hey, people don’t eat that luxury hamburger anymore, they prefer dirt these days which is — heh — dirt cheap! No inflation here!”
(Here’s another test: if you look at classes of households with “the same” real income, does the incidence of desperate measures to make money — like prostitution, second jobs, shoplifting — go up in line with ShadowStats’s claims about their real wealth?)
Charles, the Monthly Labor Review article in fact does not identify the source of the claims about hamburgers and geometric averaging, perhaps out of deference to the concern you raise. I am the one who put the two side-by-side for comparison. BLS may feel it needs to hold its punches, but I don’t.
Following a discussion of arithmetic versus geometric means in a previous post (21/7), I ran a simulation exercise to look at the difference between arithmetic and geometric means and was surprised at how different they are. My purpose was to test the two averages as an estimate of the population mean of a lognormally distributed variable (ie price ratios). I must say though that the simulation exercise convinced me that the geometric mean was most appropriate. Actually, I am puzzled as to why the BLS does not use this as their justification for using geometric means rather than substitution.
I would be interested to see the distribution of your micro price changes, if you are allowed to do that.
I would also be grateful if you would give more precise references (eg Dhyne et al, Bils and Klenow etc?) as I might follow them up.
>You used three parentheses at the beginning of your >equation? Just spitballing here…
Yeah, but that doesn’t change the result. My only concert was if I was writing the formula properly. I rarely used geometric means in my work.
Can just copy paste to excel
I just read the MLR paper, and the authors admit that the BLS does exactly what critics claim. Not literally substituting hamburgers for steak, but for example substituting peanut candy for chocolate candy in response to price changes. Here the BLS is confusing the rise in prices (due mainly to currency debasement) with consumers’ response to changes in price.
If you want to measure how prices of goods are changing, you must not change the composition of the basket of goods in mid-stream. With geometric weighting, the variable being measured (the overall price for a basket of goods of a given composition) is being changed in a data-dependent way – based on data measured after the fact. Any statistician will tell you that is a no-no.
to Rebel Economist
I think the justification for using a geometric mean and not an arithmetic mean is because increases are not bounded from above whereas price decreases are.
For example, using arithmetic mean and computing the price variation by means of log(pt/pt-1) solves this problem. Using the geometric mean is an equivalent way to doing this.
I am not sure what do you mean by “the simulation exercise convinced me that the geometric mean was most appropriate”, can you explain better, please? That is not surprising, in the case of the log-normal distribution the maximum likelihood estimator of the mean is the geometric average:
Dhyne et al (Journal of Economic Perspectives 2006)
Bils and Klenow (Journal of Political Economy 2004)
I don’t need to show you my distribution, you can find a bunch here:
Many of these papers show graphs similar to that.
I believe that Etienne Gagnon does something similar with data from Mexico (his data is actually public)
Can look at http://www.klenow.com for 2 papers on this subject.
Again, I don’t know which estimator is more adequate, geometric average or arithmetic average, the point is that the arithmetic average tends to be larger. As mentioned above, the geometric average is ML estimator for the mean of a lognormal distribution, and therefore it is statistically more adequate. I just don’t know if this is the relevant measure.
I think you are being a little unfair to John Williams. I don’t know him, but it appears from the Shadow Stats website that he sells his product, so he cannot afford to be completely open about his methods. I do know that he has been around since long before the present concern about inflation. He might well be happy to argue his case in a more academic style if he had state support for his research as I presume you do.
From this week’s Business Week:
Argentina’s former central bank president warned Tuesday that the country’s official inflation rate is inaccurate, joining a chorus of critics who claim the figures are regularly manipulated for political gain.
Annual inflation is likely much higher than the 9.1 percent reported by the national statistics institute in July, Mario Blejer said. Most independent analysts say annual inflation tops 25 percent.
Blejer said the real pace of price gains falls somewhere between those two figures — but he argued that it is impossible to know exactly where.
“The truth is, I don’t know how much real inflation is, and that’s a problem,” he told a forum of economists and bankers in Buenos Aires. “It’s important to avoid the inflation mystery.”
Argentina’s national statistics institute has been accused of manipulating inflation data to mask weaknesses in the economy, in effect protecting President Cristina Fernandez from the political consequences of what is likely Latin America’s second-highest inflation rate.
That kind of machination would also save the government money: every one-percentage point increase in official inflation means an extra US$600 million owed in interest payments on inflation-indexed bonds.
To Silas Barta
I absolutely agree with you. Hedonic adjustments try to correct for quality changes but not all goods are equally easy to adjust.
If a car has a/c now and it didn’t have before it is possible to observe that difference and therefore adjust the new price to the increased quality. If I buy a polo shirt and after washing it once I find a hole in the material (as it happened to me today) because the materials now are of less quality, how can I adjust prices to this??
Same problem with product availability (I believe someone mentioned this before). I would like to buy a new car without power steering, but that is no longer available. How can a price be adjusted to this specific characteristic if such product does not exist anymore?
I have no doubt that perception and reality are not necessarily the same, and therefore what people think of inflation based on their own experience does not necessarily reflect the true inflation. But, with all the adjustments to prices either by quality or by substitution, it will create other problems in measuring cost of living or prices in general.
One idea that has been in my head for a while is about the impact of these new techniques of measuring prices (and therefore measuring quantities) on the so called great moderation. When did these hedonic adjustment, geometric averages or product substitution started? Professor Hamilton, any thoughts?
To RebelEconomist (and Anonymous), on arithmetic vs geometric weighting.
The actual amount a consumer would pay for a given basket of goods is a linear combination of the price of each good weighted by the quantity of each.
This corresponds to the (weighted) arithmetic average.
The geometric average does not correspond to a quantity that is so meaningful to an actual consumer paying for a given basket of goods.
Thanks for the references. Also Klenow and Kryvtsov, and Nakamura and Steinsson?
My aim was to simulate price changes in an economy with a known underlying price change and see which average would reveal this best. The answer was the geometric average for the reason you give – it was the maximum likelihood estimator given the price change distribution I simulated.
In other words (Tim), the use of a geometric average can be justified without arguments about substitution (which are highly stylised).
Sorry, Anonymous at September 5, 2008 01:18 PM was me.
Also Klenow and Kryvtsov,
and Nakamura and Steinsson?
Sorry, Anonymous at September 5, 2008 01:18 PM was me.
Posted by: RebelEconomist at September 5, 2008 01:20 PM
Silas Barta = Rebel Economist?
But the purpose of the average is to find the underlying price change, not the expenditure change.
I agree with you. I had the same idea: my budget constraint is a linear combination of prices and quantities. The question I want to answer is about the variation in money spent (budget) keeping quantities fixed.
I don’t see how the geometric mean relates to this.
Thanks for the references, Anonymous at September 5, 2008 01:22 PM.
No, I am not Silas Barta.
If you think this is contained to the US – think again.
Here is the best rant I’ve yet seen and it’s based in the UK:
Compare a 2008 Toyota Corolla with a 1978 Ford Pinto.
Is the value equal? Excluding patriotic concerns, which would you want? Compare the nominal prices adjusted with CPI.
This line of thinking will lead one to determine that the cost of automobiles is going down.
The government is understating inflation to help big corporations hide real ROI!
It is simply being accepted without question that the BLS will implement any improvement which reduces CPI inflation, and otherwise will ignore advances, or that political considerations are what bring about the changes in the CPI. In times past, the BLS seems to have dragged its feet a bit in implementing changes that the profession agreed were necessary. Today, it appears to be quite responsive to academic criticism. Changes are implemented all the time, but they don?t make the headlines. For example, the aging-bias adjustment is the biggest use of hedonic methods in the CPI; it corrects for decline in quality of rental units as they age, and this adjustment increases the inflation rate. The hedonic methods, in general, generate almost identical movements to ?matched-model? indexes which compute inflation based only upon units which stay in the population (i.e., no quality change at all). The fact is, the price of any given mode. of TV drops like a rock. If the new TV?s weren?t better, no one would buy them. The position of this pro-BLS commenter is that all the changes that have been made to the CPI have increased the accuracy of the CPI, and there have been no changes which would have increased the accuracy of the CPI which have been rejected by the BLS. ?The government? is not a monolithic entity, acting in perfect harmony to screw over seniors. If improvements reduce biases (most of which have been upward), I think we can all agree that that is a good thing. If you want to be more generous to seniors, fine: just pass a law stating that you will increase benefits by CPI inflation plus one percent. (And on inflation facing poor vs rich, I think Christian Broda has a recent paper on that.)
The divergence of opinion about where the economy is headed has to do with forecasting, and spin. There is almost no divergence in informed opinion about measurement issues.
Read the CPI computation manual on the BLS website. First of all, many commenters are confusing geometric means (a way of averaging lots of different inflation rates into one number, which is recommended best practice by international standards) with something more like computing growth rates propertly (computing a 5-year growth rate as the 5th root, rather than dividing by 5). Geometric means are not applied at a zipcode level; the CPI is not even constructed that way. Rather, it is constructed on an area-item basis, where area is big (like ?Chicago?) and item is small (like ?apples?); so we are talking things like Chicago apples, Chicago orange juice, etc. Geometric means are used to compute the inflation at the level of ?Chicago orange juice,? a practice which assumes unit elastic substitution across different brands of orange juice. It assumes this for good reason: it would be crazy to think that consumers did not move away from one brand to another in response to large price changes. Econ 1 tells you that this is the right thing to do. It would be crazy not to use economics to guide measurement, for the same reason that we would not want to go back to 19th century medicine. Geometric means are not used across items, even in the same city; Laspeyres is still used to ?combine? Chicago apples and Chicago orange juice into the Chicago inflation rate.
Subtle product quality decline is an interesting issue and, if you have something more than anecdotes to offer, that would be very interesting. In principle, hedonic methods would adjust for either quality increase or quality decrease. An unmeasured attribute change, positive or negative, would have to end up in the error term.
from the BLS article: “the CPI’s objective is to calculate the change in the amount consumers need to spend **to maintain a constant level of satisfaction.** ”
so the BLS does NOT view the CPI in the manner in which it is portrayed almost everywhere, that is, a measure of the increase in prices. So it is not surprising that many do not agree with it – they think it is measuring something else !
I mean, really, **to maintain a constant
level of satisfaction.** is open to so many subjective abuses that it makes the measure meaningless. and they think they know it to a decimal place or two ? come on – an order of magnitude, _maybe_
September 5, 2008
James Hamilton of Econbrowser passes on a debunking of Shadowstats … I confess, I have not examined the issues raised very closely, but this is the sort of thing I like to see. The internet has offered looney-tunes a pulpit, which is generally a …
Marcello says that the phrase “to maintain a constant
level of satisfaction” is “open to so many subjective abuses that it makes the measure meaningless.”
I don’t think the concept is meaningless at all. As the article explains very clearly, we know that the answer to that question has to be a lower inflation rate than the one that would describe how much your income would need to go up in order for you to be able to buy exactly the same items as you did before. To repeat the argument here, if your income went up by that amount, and you in fact did choose to buy exactly the same items as before, then you would have achieved exactly the same level of satisfaction. Insofar as you chose to do anything other than exactly the same thing– and it’s very awkward to write down any coherent model of preferences in which you would not– then if your income went up by exactly the Laspeyres inflation rate, you would be categorically better off than before, because you could (and would) choose to buy some things that you couldn’t have afforded before.
An economist is a trained professional paid to guess wrong about the economy. An econometrician is a trained professional paid to use computers to guess wrong about the economy.
The BLS’s defence of the OER consisted entirely of “proof by repeated assertion”.
Forgive me for approximate quotes, it’s a pdf.
“it’s a myth that OER reduced the CPI shelter index”
Oh yeah? Prove it. Rents have increased more slowly than housing prices for *over* *15* years. If this is smoothing a 30 year trend, then it’s a pretty crappy way to measure inflation.
“it’s not clear that the OER’s impact has been down”
Really? But you said it was a myth? It’s not clear? I thought you guys were scientists?
“OER is used by most countries that matter”
That doesn’t actually prove anything, of course. Just that other countries like to massage data, too.
Wonder why people give so much undeserved credit to ShadowStats? It’s because their competition’s lost most of it’s credibility. Articles like this don’t help, either.
Please note – The reworded comments above was the sum entirety of their addressing the OER.
I’m still waiting for some kind of defense of the OER that makes any sense if you want to measure inflation, rather than inflation ex-interest rates.
Jim D, the article lays out the case for OER as follows:
(By the way, any recent version of Acrobat Reader should give you the ability to cut and paste).
What’s your alternative to OER?
Yeah right, anyone who says they are seeing inflation greater than what the BLS tells us is a moonbat! Anyone who sees even the price of hamburger increasing faster than what the BLS says is just a nut!
IMO an honest measure of inflation measures at a constant standard of living. For me, that constant noticeably exceeds what the BLS tells me I am experiencing. But what do I know? I’m just an irrational loon.
I’m late to review (skim) this paper, but I am uncomfortable with this paragraph:
That “BLS must” depends on their goals. If this is truly about cost of living, then a disappearing product does raise costs.
They are not computing “benefits of living.”
I have my doubts as to the good character John Williams’ writings. In the past, I have seen him support his arguments with misleading historical charts.
Williams used a chart covering 200 years in constant dollars. It showed a spike rise in value for this last decade. I cannot remember what the chart was about now, but that does not matter as the principle was what bothered me. Any discriminating and knowledgeable mathematician would instantly know that this kind of representation was meaningless and misleading to use.
Only Log Charts, that represent percentage changes, show the realistic price relationships over long periods of time. As an example, You would think that the 1921-1934 run-up and crash was minor on a DJIA arithmetic chart. But with a log chart, you can clearly see the huge percentage changes in price.
Now I assume John Williams knew what he was doing. So I am of the opinion that I cannot trust his analysis if he is presenting his argument in a manner that is grossly deceptive.
Please examine Chart 1. OER was introduced into the CPI-U in January 1983. At that point the CPI-U rose faster than the shelter index of the CPI-W, which continued to use the old method. Therefore, introduction of OER increased the measured inflation rate of shelter costs. This data is available at the BLS website, so you may confirm this yourself.
The article also points out that OER has risen more slowly than measures of monthly payments available from the NAR.
>> But the purpose of the average is to find the underlying price change, not the expenditure change.
Tim is right.
The CPI should be measuring Inflation, not the ( presumed ) citizen’s RESPONSE to Inflation. The BLS was not tasked to MODEL HUMAN ECONOMIC BEHAVIOR.
As to Quality measurements: how about my Sears Kenmore Elite ( made in china ) refrigerator. 7 breakdowns in 7 years. Is the CPI reflecting the massive move of US companies to Import JUNK from China?
How about my heating oil “lock in” price increase of 83.6%? From 2.58 last year to 4.70 today?
As for substitutions when will the BLS get around to lowering the inflation rate because instead of buying a new car, I’ll have to keep my current car for up to 20 years. Is that a Quality Improvement?
It looks like it’s the BLS that’s been debunked.
Marcello says that the phrase “to maintain a constant level of satisfaction” is “open to so many subjective abuses that it makes the measure meaningless.”
I don’t think the concept is meaningless at all. As the article explains very clearly, we know that the answer to that question has to be a lower inflation rate than the one that would describe how much your income would need to go up in order for you to be able to buy exactly the same items as you did before.
But “satisfaction” is a metaphysical quantity, because you cannot compare the satisfaction of the goods of today with the satisfaction of the goods of yesterday, because the goods of yesterday are no longer available for purchase, and satisfaction depends a whole lot of context, which varies with time.
Measuring the cost of someone’s lifestyle with a precise measure of something purely imaginary like “satisfaction” or with an approximate measure using a proxy like a constant basket of goods, the latter seems enormously preferable if the goal is to see how prices evolve.
To measure quality changes, you would need to do laboratory testing, which the BEA and BLS do not do.
But that is irrelevant too, because quality often does not enter into purchasing decisions; availability is far more important. For example, chinese stuff today may be less expensive because it is much lower quality, but there is simply no alternative; if you want to buy a t-shirt in effect you can only buy either a low quality but cheap chinese one or a very expensive luxury one. There is no infinite number of markets for an infinite number of quality grades for an infinite number of goods.
If a car has a/c now and it didn’t have before it is possible to observe that difference and therefore adjust the new price to the increased quality. If I buy a polo shirt and after washing it once I find a hole in the material (as it happened to me today) because the materials now are of less quality, how can I adjust prices to this??
The answer as you know is that you should not adjust either way. In countries where statistics are more reasonable a basket of essentials is designed by actually looking at what a number of *median* working class or middle class consumers buy over a year, and then keeping that constant, and then perhaps every 15-25 years updating the basket as consumer habits do change.
In the U.S. counts food as only __ 8% __of the CPI index. Whereas, it counts for about 10% in the United Kingdom, about 15% in the rest of Europe and more than 18% in Japan (food@home = 7.66 in BLS). Interestingly, if you look at the proportion of U.S. household spending on food, by income quintile, all but the top 20% of earners spend at least __ 20% __ of their paychecks on food. In the CPI, Education is 6%, health is 6%. Who has that kind of pay? One which requires only 6% for healthcare? University professors and economists?
There is a very nice article in BusinessWeek discussed here:
about some poor middle class people struggling to make to the end of the month (because of the vicious taxation of the government that steals money from them to give free Cadillacs to welfare queens and t-bone steaks to strapping young buck) on a $300k/year family income.
Out of a modest $25k/month budget $1.6k go towards food, or around, guess what, 6-7%.
Like all countries, the USA surveys consumer expenditure by using surveys:
and these are the latest data by
The typical LOSERS (you know, the 80% that don’t work hard enough to earn more than $70k/y) spend between 11% and 17% of their income on food.
But expect these stats to disappear or be massaged by using “satisfaction”; in the USA the Republican party has a “blind the beast” strategy…
LOL! And who said that time series econometricians didn’t have a sense of humour? hehe
Calling all behavioural economists! Calling all behavioural economists!
I can offer two attempts at an explanation
1. There is clearly a strong consensus to not ‘screw’ with the data collected by national statistical agencies in rich, developed countries. But all other sources of public agency information and analysis are viewed as fair game for manipulation, spin, deliberate omission, and so on. War, national security, natural resources: no controversial public policy field is immune from the purposive and strategic use of information and analysis by elected political leaders, civil servants, and social activists.
I’m guessing that this robust social contract to deliver the best data possible given budget constraints and the current state of economic and statistical science is not widely known or understood. People have good reason to mistrust the State.
2. Periods of economic growth driven by rapidly changing general production technologies (GPTs) tend to polarize income distributions and have significant wealth redistribution effects. Neoclassical substitution arguments, although correct at both the micro and macro levels, inevitably gloss over these redistribution effects.
For example: Early adopters of information technologies who display a high-level of sophistication with respect to lifestyle choices–where to live, what food to eat, leisure choices, e.g, backpacking versus living out of an RV, as well as household budget risk management choices–will experience significant improvements in material standards of living while others will experience stagnate or declining real standards of living.
Apologies for not providing a couple of dozen references. I love this blog but feel super guilty every time I type a comment as I have so much bloody work to do to attain a fraction of what Professors Hamilton and Chinn have already professionally accomplished!
I went out to the BLS website and did a search on CPI-RS…. The BLS website query turned up NOTHING. I then Googled “CPI-RS” and found an article by Satyendra Verma, Ph.D.
Satyendra wrote that “CPI-RS” (CPI research series) were the historical recalculations of 1978-1998 CPI-U which applied the Rental Equivalence, Geometric Mean, and Hedonics implemented by the Boskin Commission. This excercise concluded that the Boskin Commission changes had the effect of lowering the CPI-U about 0.45 percent/year over the 1978-1998 sample period.
MarkS: If you look back to my post on government statistics, you’ll see the link under my graph of CPI-U and CPI-RS inflation.
Interesting discussion. Reminds me of the “proof” of WMDs in Iraq.
Randy’s point is absolutely correct. The usage of geometric means is only used at the product level and its purpose is to have a starting product price. From there the CPI is average arithmetically. I missed this fact completely.
What this means is that the downward impact of a geometric mean on inflation rate is much smaller. Nevertheless, it exists. If I am not mistaken, CPI has started to include (correctly in my opinion) sales, this means that the usage of geometric mean will account in the same way a 50% reduction and a 100% increase (price drops from $50 to $25 and it increases again to $50). This is a situation which makes sense to weight decreases and increases in a similar and comparable way.
My personal conclusion, after all this discussion is that the usage of geometric averaging in the process of producing the CPI makes sense, and it actually is desirable. Regarding Shadowstats, I personally don’t give them much credit.
Another point that I am not sure if it was brought to discussion is the fact that nobody ever said that the CPI is supposed to reflect each individual’s cost of living. I rarely eat bread, wheat price increase does not affect me much. I drive quite frequently, and therefore gas price increase has an impact on me.
I found the link to the CPI-U-RS link provided by Anonymous…. This new series operates as Dr. Verma described “CPI-RS”.
It appears to me, that the BLS began these “research series” in 1981 with the CPI-U-X1 experimental index that first incorporated the equivalent rent estimation for housing cost. In 1999 a new experimental series was published that included equivalent rent, substitution and hedonics and was labeled CPI-U-RS and was calculated for the 1978-1998 time period. This series has appearently been maintained and updated since 1998 to express (as much as possible) historic data (beginning in 1978) with current CPI methodology.
I’m much obliged to you for bringing this BLS series to my attention.
ED- I think your auto example is perfect for explaining why hedonic adjustments are BS. You claim that the 2006 car is better because of ABS, airbags, etc. etc. This misses the point that the utility provided is the same, basic transportation. Folks seeking basic transportation can’t go out and buy a Model T, the buy that Corolla and get the same utility out of it.
Another good example is the computer. If I have a 166mhz Pentium in 1996 running windows 95 that I use to surf the web and respond to blogs I get the exact same utility as a quad core 3 GHZ computer that I use to respond to blogs. You can’t assume because horsepower or processing power increased by X% while the price remained the same that there is deflation.
Randy’s comments, along with Jim Hamilton’s post, are by far the most informative and informed — the latter meaning by their knowledge of economics and statistical matters — in this lengthy thread.
And Randy and a few others are probably right on target when they note that the conspiratorial types who seem to suspect all governments in the democratic world as fraudulent manipulators of data (among other frauds they perpetuated) — in the EU, Japan, and North America — haven’t bothered in the main to read the BLS report in question. It doesn’t matter, by the way, that some of these governments that use a geometric means — endorsed by the IMF and other international organizations, manned by professional economists and statisticians — are socialist, others middle-of-the-road, others yet conservative . . . no! no! they’re all out to bewilder and lead astray full-fledged radicals of the left or Ayn Rand liberterian zealots on the right, plus whatever disgruntled types fall in between on the ideological spectrum.
Hence if the US CPI-U rose 50% faster than that of the other 6 member-countries’ average index between 1997 and 2007 — 2.9% vs. their 1.9% — or rose faster than 16 of the 29 OECD members’ average index, no matter. The other 6 members of the G-7 are apparently even greater frauds than the BLS and BEA here, and the government agencies of the 13 OECD countries like Turkey and the new East European members of the EU haven’t learned yet how to manipulate to beat the band the way the more advanced swindling government agencies have in the richer G-7 countries.
The three major changes adopted by the post-Boskin BLS indexes — a geometric means for the use of substitution effects (a standard micro-economic concept dozens of decades old by now), an updated version of the 1981 measure of the cost of housing shelter for home-owners (a rental equivalent: guess what, 69% of Americans own their homes and aren’t looking yearly to buy another), and a new view of quality changes — are explained clearly and with simple examples that refute the conspiratorial charges as well as those that lead others to yell “cheaters!” and “con-men frauds!”.
It’s doubtful, to repeat, if most of these critics have bothered to read the report . . . not just carefully, but even in a cursory manner.
None of these comments in support of the BLS mean that there are not legitimate technical issues that engage professional economists and statisticians. Here’s one example: an updated assessment of the Boskin Commission’s recommendations by a former member — Robert Gordon of Northwestern — click here:http://faculty-web.at.northwestern.edu/economics/gordon/P376_IPM_Final_060313.pdf
Michael Gordon, AKA, the buggy professor
PS A personal aside. I once had the intellectual benefit of studying with Prof. John Hicks of Oxford who eventually won the Nobel Prize for economics . . . including for his IS/LM interpretation of Keynes’ work and for his book “Value and Capital” — the latter published in the 1930s, which was the first in English (drawing on the work of Marshall and others in Britain, plus Walrus in France, some Austrian economists, and a few in Scandinavia) — that set out clearly, for the first time, substituion and income effects using indifference curves as prices of goods, services, labor, and capital changed.
To question the use by the BLS of close subsitutions as prices rise is to question one of the bases of modern micro-economics, along with the use of marginal analysis, opportunity costs, information problems, agent-principal problems, the Pareto optimal and Pareto-improvements or -infringements (a minimal equity measure), cost-benefit analysis, market-failure, and government’s ability or not to correct them or make things worse.
>> To question the use by the BLS of close subsitutions as prices rise is to question one of the bases of modern micro-economics,…
So what? The use of substitutions is a game economists play and have played for years to fool the public and themselves is not a justification for it’s continued use.
Again, you were not tasked to Model Citizen Response to Inflation, but, to measure the actual yearly change in inflation. A separate report on how citizens responded to inflation seems to fall into an economic-phycology area.
Why don’t we also play this game in general accounting? Why use SAME STORE SALES when we can manipulate sales as we like by dropping the number of stores between year 1 and 2?
Why don’t we give the Phillies a 5 run lead at the start of baseball games? Just this year, because they haven’t won in a long time.
A personal aside. I once had the intellectual benefit of studying with Prof. John Hicks of Oxford who eventually won the Nobel Prize for economics . . . including for his IS/LM interpretation of Keynes’ work and for his book “Value and Capital” — the latter published in the 1930s, which was the first in English (drawing on the work of Marshall and others in Britain, plus Walrus in France, some Austrian economists, and a few in Scandinavia) — that set out clearly, for the first time, substituion and income effects using indifference curves as prices of goods, services, labor, and capital changed.
But IS/LM and Marshall and Walras (and Arrow/Debreu/Lucas that follow) are completely bogus ranting with colossal mathematical errors and contradictory assumptions, and none of the “effects” and “curves” you mention above have any basic in theory or practice.
They are just masses of circular mistakes that get taught in universities with the sole purpose to support the central verity of Economics, that the income distribution depends solely on productivity, because all those crass mathematical mistakes and contradictory assumptions are the only way to get to that verity.
Then of course the astute dissemblers at the BLS prevaricate about a constant “level of satisfaction” as if it was a number that could be measured precisely and unambiguously even across time and different structures of the economy, and thus preferable to using a constant basket as an imperfect but acceptable proxy to a sraffian “standard” commodity.
Guys as soon as some guys talk of a “level of satisfaction” as a fully cardinal quantity with an time/configuration space invariant metric, just laugh, laugh, laugh at that bunch of clowns.
all governments in the democratic world as fraudulent manipulators of data (among other frauds they perpetuated) — in the EU, Japan, and North America — haven’t bothered in the main to read the BLS report in question. It doesn’t matter, by the way, that some of these governments that use a geometric means — endorsed by the IMF and other international organizations, manned by professional economists and statisticians — are socialist, others middle-of-the-road, others yet conservative
That’s more or less correct, all these guys have their bread buttered the same way, and the IMF and OECD are used to share the “technologies” that result in the “best” (lowest) results.
I am also familiar with the UK and italian situations, and a lot of the statistics tricks are the same, some were even invented there and adopted in the USA.
Yet the italian CPI FOI is actually more honestly called “price index at retail for families of wage and salary earners”, and it is (or rather used to be) constant-basket based.
Which is the best approach: monitor the prices paid for a semi-representative basket of good bought by the median (that is, low income) worker.
Except that in order to keep the index low, some of the prices of the basket components have been kept constant (those products are effectively no longer available), and more recently the basket has been accused by many to be unrepresentative of the typical expenditures of low income earners (e.g. housing cost weight is 9%).
BTW, I also find the Shadowstats “add X%” approach a bit suspicious, but rather less than the official numbers.
ABS and airbags were options that I CHOSE to pay more for. So by revealed preference, they provided me with utility. Perhaps you get no utility out of the features I listed, but many do, that’s why they’re willing to pay more. As I said, cheaper models were available.
As far as your 1996 pentium, I’m sure you can get one for very cheap on Craigslist. You might even be able to find someone giving one away for free. If you get the same utility from it that you always did, your cost has gone down by almost 100%.
As to the “cleverness” of using “level of satisfaction” as the magic inflation shrinking machine, the kicker in all this argument is that the USA CPI used to be a fixed basket updated every 10 years, and guess what, the establishment mouthpieces were complaining that it *overstated* inflation, seen as the increase in the cost of (a declining standard of) living.
Consider this random article from a web search, article dated 1997:
The CPI, which originated in 1919, is simply a measure of the average change over time in the prices paid for a fixed market basket of consumer goods and services. [ … ] Reference base periods are updated about every 10 years.
“The best measure of inflation for a given application depends on the intended use of the data. The CPI is generally the best measure for adjusting payments to consumers when the intent is to allow them to purchase, at today’s prices, the same market basket of consumer goods and services that they could purchase in an earlier reference period. It is also the best measure to use to translate retail sales and hourly or weekly earnings into real or inflation-free dollars.”
–From: Understanding the Consumer Price Index:Answers to Some Questions
US Department of Labor/ Bureau of Labor Statistics, July 1996 (Revised)
As important as what the CPI is, is what it is not; and herein lies the primary reason for the current controversy about the index and why some economists allege that the CPI overstates inflation. The CPI is an index of price changes only and does not reflect changes in buying or consumption patterns. It does not reflect changes in spending patterns over time. Since it is not affected by changes in consumer buying habits, the CPI makes it easy to compare the costs of the same market basket items from month to month or year to year.
Four areas of technical concern remain with the CPI, particularly since it continues to be used as a proxy cost of living index:
1) The substitution effect does not allow for the substitution of a cheaper product for a more expensive product in the fixed market basket when relative prices change.
2) The sample rotation effect arises because the current procedures for introducing new items and outlets tends to give disproportionate weight to changes made if, in the month of introduction, they are temporarily high or low.
3) Outlet substitution, similar to the substitution effect above arises when consumers substitute a less expensive outlet if they do not factor in the lower level of customer service provided by such stores.
4) Lastly, the quality-adjustment bias, considered by many to be the most serious shortcoming, arises simply as improvements in the quality of the goods and services unreflected in price.
The timeliness of an every-decade revision of the market basket composition is also an issue of concern.
The current media spotlight on the CPI results primarily from the Boskin Commission, a Presidential Commission charged with determining what, if any, bias existed in the CPI. Commission members, primarily academics, rendered the option (not unanimous) that primarily due to the quality and substitution issues present in the current CPI, there was an overstatement of 1.1 percent annually.
Except that add that 1.1% to the other changes done before 1996 and to those since, including weights etc., and you get to something like 2% or even 3%, and that’s not too far from ShadowStats.com; note also that even just lopping 1.1% off compounds to a non trivial effect over time.
Also note that in this chart:
the weights for things like food etc. are very different from the ones reported above by .
Note: the above is just one paper, with authentic quotes from the BLS of 1996.
What does the BLS of 2008 have to say about the BLS of 1996? That the goals then (measure the cost of a given standard of living, not just living at a declining standard) were wrong? That the adjustments have been smaller than 1.1%?
But of course! 🙂
Also consider some other random search results:
the Commission greatly understated the magnitude of upper-level substitution bias. This retrospective evaluation suggests that the Boskin bias estimate for 1995-96 should have been 1.2 to 1.3 percent, not 1.1 percent.
Current upward bias in the CPI is estimated to have declined from the revised 1.2-1.3 percent in the Boskin era to about 0.8 percent today. Yet the Boskin report, like most contemporary studies of quality change, failed to place sufficient value on the value of new products and on increased longevity. Allowing for these, today’s bias is at least 1.0 percent per year or perhaps even higher.
Because it means a smaller federal deficit, trimming the rate of growth in the CPI would be welcome on Wall Street. Lower rates would help fuel increases in bond and stock prices. “If this makes it easier to address the deficit issues, all the better,” said Micah Green, executive vice president of PSA The Bond Market Trade Association. On the other hand, a lower rate of growth in a new index would mean smaller annual increases in myriad government benefit programs, including Social Security and veterans’ cost-of-living adjustments, that are tied to the index. And because income tax brackets are indexed to inflation, many Americans might see their tax bills rise.
If the BLS would simply rename the CPI as the “Consumer Price and Hypothetical Substitution Effects Index” (CPHSEI), it would do much to resolve concerns about whether the index is being correctly calculated.
Jim – thank you, thank you. And thank you again.
Presumably your closing question was intended to be wry, rhetorical and humorous. Or so I though until finding myself scrolling one of the largest collections of comments on your posts I’ve seen in a long while.
Let me cause some more trouble – from your excerpt the BLS is not substituting between baskets but within them, i.e. not only aren’t steaks sub’d for hamburgers but only hamburger in Boston! 🙂
How anyway can argue that one truly escapes me and that’s not rhetorical. But a very nice try anyway and appreciated by some of us.
Don’t know if anyone’s still here, but I’m working late at the factory so I have time.
For me, the shadowstats argument breaks down over the long term. THe most direct way to prove this is a simple compound interest spreadsheet.
If inflation understatement is as large as 2%, then after 28 years a chained-CPI dollar should buy only 57% of what it is claimed to.
Year Compound 2% error
Except for housing and college education, nothing seems to be so grossly out of proportion to me.
It is easy to use ShadowStats numbers and annecdotal observations in the short run, but in the long run it does not work. Especially for the hard core conspiracy theorists.
Projecting a 3% underestimation from the end of WWII would leave people of today facing 5x higher costs than their grandparents. How can anyone look at the issue on that time scale and see anything but nonsense.
Never program and blog at the same time…
Of course the comparison would not “leave people of today facing 5x higher costs than their grandparents”, but would leave people leave people of today facing 5x higher relative cost of living than reported by the government.
And then those grandparents could rightly claim they deserve 5x the social security benefits.
That would be a sweet deal.
Time to leave the factory.
Sure I worked late, but there is Starbucks and 52 inches of Monday Night Football in my future.
Expecting a very short Friday too.
Consider another practical measure of inflation, cruder but less manipulated: the wage cap on Social Security (FICA) taxes. It is supposedly an inflation measure but is tied to average wages. It still understates inflation because real wages have been falling for most people, but the wage cap has still risen at just over 4% annually for the last ten years. The CPI shows an annual rate of only about 2.9% for the same period. The CPI is clearly understated.
Anyway, all the fancy argumentation in earlier comments is beside the point.
The reason there’s nothing wrong with hedonic analysis in theory is that there’s nothing wrong with it in theory, only in practice. The BLS’ manipulates its hedonic adjustments purely to keep the nominal CPI down. When some good becomes cheaper, it is given no hedonic adjustment. When any goods cost more, they are given risible, even brazenly fraudulent hedonic adjustments. Somehow those adjustments always produce a CPI which understates the real change in cost of living.
The BLS has already shown itself to be utterly corrupt. When Congress mandated “oxygenates” in gasoline (at the time a choice between MTBE and ethanol, now only ethanol), a technical change which caused MPG to drop by ~5% for the unavoidable technical reason that there are fewer joules available from the combustion of “oxygenated” gasoline; and which also *increased* pump prices because oxygenates cost more than actual fuel, and *increased* auto repair costs because oxygenates are hydrophilic and accellerate corrosion of fuel-system and engine parts, the BLS decided the change was a ‘quality improvement’ so it would not count the increase in fuel prices even though prices increased on both a nominal per-gallon basis and (even more) on a real per-mile basis. So much for BLS’ vaunted independence and commitment to accuracy.
Oh, and when questioned about it later, BLS said that oxygenated gasoline reduced pollution so that made it better, even though it was actually worse by every measure. BLS’ was lying then. The EPA had already reported that oxygenated fuel did NOT reduce air pollution at all (since modern fuel-injected cars have “oxygen sensors” and automatically adjust the fuel-air mixture to maintain the same oxygen ratio regardless of how much the fuel has been adulterated), even, to the contrary actually INCREASED air pollution since the oxygenates were more volatile than straight gasoline (RVP) and promoted fuel evaporation leading to more photochemical smog, and INCREASED water pollution– a problem so severe with MTBE in Southern California due to leaks from tanks corroded by MTBE, followed by MTBE’s amazing ability to seek out water to bond with and poison, that many SoCal cities had to shutdown their groundwater wells and purchase aqueduct water from the Colorado river via the MWD, so the “oxygenate” mandate actually drove up water rates as well.
I could multiply examples. As other commenters have pointed out, the role of BLS’ for a couple of decades now has been to declare anything which actually goes up in price an “outlier” or a “quality improvement” and discount it in the CPI. Why anyone should read and credit “white papers” from such charlatans is beyond my capacity to comprehend.
The BLS does not quality adjust gasoline.
According to their “fact sheet”: “There are no explicit quality adjustments made for changes in fuel or service quality. Adjustments are not made for switches in gasoline content due to mandated air quality requirements.”
Guys, by the very statements of the BLS and its advocates, two facts are indisputable:
* Ten years ago the CPI went from measuring the price of a stable basket to measuring a “level of satisfaction”.
* That and other measures were made to reduce the index by around 1.2% on an ongoing basis.
That 1.2% is just the impact of the Boskin commission and “level of satisfaction” at the time. Then there have been quite a few other adjustments in the 10-20 years before Boskin and in the 10 years after, so 1.2% is the FLOOR for the value of the adjustments made to reduce the “overestimation” of inflation by the CPI. Some people estimate that the total is at least 2% and may be as high as 3% over the past 20 years. Or perhaps the adjustements are worth not a fixed % but something like a 30-50% reduction in the index (proportional rather than fixed).
Now, there may be dispute whether the index in 1975 was overestimated or the index in 2005 is underestimated, but there cannot be any dispute that there is a pretty significant difference between the two indices, because a large part of that difference is precisely what the Boskin commission was created to do, under the guide of preventing the index from overestimating inflation.
BTW, as to that 1.2%, even if it were just a 1.2% adjustment, its impact would still be very remarkable.
In the past 10 years Alan Greenspan and others have often based their arguments on the idea that productivity growth has been remarkable; but if inflation in the prices of good and services delivered in that has been underestimated by 1.2% in the past 10 years, then a lot of those productivity improvements was sort of illusory.
When Congress mandated “oxygenates” in gasoline [ … ] caused MPG to drop by ~5% [ … ] *increased* auto repair costs because oxygenates are hydrophilic and accellerate corrosion of fuel-system and engine parts,[ … ] prices increased on both [ … ] on a real per-mile basis
Adjustments are not made for switches in gasoline content due to mandated air quality requirements.
First of all, everybody please note the overly clever switch in topic: the “oxygenates” mandate was not a mandate for “air quality improvements”, so this point is probably irrelevant.
Or else thanks for finding the quote that confirms what said: that even if the quality of gasoline went down, and the price for MPG went up, the BLS did not regard that as an increase in inflation and because it is an example that the BLS does not regard decreases in quality, even if uncontroversial, as increases in price (but it surely has confessed to doing the opposite).
As the Boskin commission said, quality adjustments were part of the techniques used to to ensure that the CPI was less “overestimated”.
The use of geometric averaging in price indexes violates one of the most basic rules of economic analysis.
In a free market, prices are set by sellers to maximize profits, and sellers are responsive to changes in consumer preference.
For each brand of each good, there is an optimal profit-maximizing price.
If the price of Pepsi rose versus Coke, it would signal a
shift in consumer preference from Coke to Pepsi which had
shifted the profit-maximizing price of Pepsi relative to Coke.
For the central planners in the BLS to assume that consumers
could then substitute Coke for Pepsi with no loss in satisfaction
is to betray a deep ignorance of free market economics.
Here’s the link to the BLS itself to confirm what I wrote before: the BLS did treat the adulteration of gasoline as a “quality improvement.” As that BLS report states: “In addition, beginning in late 1994, quality adjustments were made for the introduction of reformulated gasoline, which was required in selected areas for compliance with the Clean Air Act Amendment of 1990.”
Now, as you can also read at that link, once oxygenated gasoline was fully phased in the BLS announced (partly in response to severe criticism), that they would not automatically make the same adjustment if gasoline were degraded again but the BLS did not correct the CPI. So the hit from the oxygenates mandate was simply omitted from the CPI.
In fact, similar dirty tricks are applied frequently to keep the CPI down.
Yikes. Failed to paste the money quote from that BLS report (confirming my earlier comment):
“The motor fuel component, whose index rose 7.5 percent between December 1993 and December 1997, would have increased by an estimated 15.4 percent over that period if adjustment for environmental quality change had not been made.”
“Environmental quality” is BLS’ speak for the oxygenates mandate, also called “reformulated gasoline”. Also note that while the BLS decided to ignore per-gallon price increases, it also ignored reduced MPG from oxygenates. An honest “hedonic” adjustment would have shown gasoline prices going up by MORE than the (estimated) 15.4%. As I wrote, the problem with BLS’ hedonic adjustments is not the theory, it’s the practice.
As at least a few honest economists have recognized for a long time (1912 was the earliest reference I could find), all price indices are basically non-sensical. Each person’s perception of the value of a unit of money over time is entirely subjective and unique–each of us constructs his or her own mental picture of what’s going to happen to the prices of various goods, and what importance to attach to each of those prices in estimating how fast the value of cash balances are depreciating or in estimating what inflation premium is appropriate for a credit transaction.
There is in fact no way to objectively quantify an array of changing prices and changing goods in a manner that has any practical significance. There is no non-arbitrary way to resolve the arguments over hedonic adjustments, weightings of different prices, or even what mathematical averaging method is to be employed.
Given that the BLS’s procedure is as arbitrary as any other, Shadowstats is performing an invaluable service in pointing out how the changes in methodology have led to significant declines in the numbers being reported, and what political motivations lie behind such fudging of the numbers.
Moreover, many of us who actually pay attention to lots of real world prices while forming our own personal views about how fast the dollar is crashing find ourselves coming up with subjective estimates much closer to John Williams’s figures than to the BLS’s. Frankly, the BLS richly deserves to lose its credibility; their numbers are entirely political and have nothing to do with the realities of people trying to make ends meet.
but the BLS did not correct the CPI. So the hit from the oxygenates mandate was simply omitted from the CPI.
In fact, similar dirty tricks are applied frequently to keep the CPI down.
But the level of satisfaction of its beneficiaries with the BLS CPI is probably constant 🙂
For those watching CPI cleverness, another bits of excellent news about hedonics improvements that reduce inflation:
Food in America is shrinking. Last month, Unilever’s Skippy brand peanut butter reduced the contents of a typical jar from 18oz to 16.3oz by increasing the depth of the indentation on the bottom of the plastic tub. It kept the price unchanged.
Kellogg has done the same thing with its Apple Jacks breakfast cereal: while the front panel size – and the price – remain the same, the contents have shrunk from 11oz to 8.7oz. Cans of Del Monte’s StarKist tuna now contain 5oz of fish rather than 6oz, while a redesigned bottle of PepsiCo’s Tropicana orange juice contains 89 fluid ounces compared with the previous 96 fl oz.
The food companies involved mostly blame the impact of rising commodity prices for what are in effect unit price increases being passed to the consumer, although a few creatively attribute content reductions to more “consumer friendly” packaging designs.
Ed Dworsky, the editor of Mouseprint.org, a website that tracks consumer issues, says the tactic, known as downsizing, is as old as the packaged goods industry itself and takes advantage of the fact that consumers are “price sensitive, but not net weight sensitive”. But over the past year, he says, the tactic has become increasingly prevalent, driven by soaring fuel prices – since downsizing also reduces transport costs per unit, adding to the savings from having less of the product in the pack.
“I can’t remember a time when I’ve seen this many major brands and this many categories involved in downsizing,” he says.
Are these price increases? Well, the package price is the same, and an expert says that consumers are not sensitive to weight, so there is NO INFLATION, and arguably they are quality improvements, as the new packages are more “consumer friendly”, INCREASING the level of consumer satisfaction.
I wonder which way the BLS will go on this :-).
especially as the benefits of this don’t stop there:
For the companies, the results can be extremely positive. General Mills reduced the depth of packets of its leading breakfast cereals such as Cheerios while maintaining the dimensions of the front of the box and the price. As a result, the number of packs sold increased by 6 per cent, yet the weight sold was virtually unchanged from the previous year.
See about corporate innovation? LOWER INFLATION and higher corporate sales and profits, all in one packages. Clearly higher profits is another quality improvement, as consumers feel more satisfaction buying from a supplier with increasing earnings :-).
Shadowstats responds to the BLS! Enough said.
Er, he admitted that the whole steak/hamburger criticsm is wrong, and he admists that the BLS is supposed to quality adjust prices. He totally ignored the fact that falling home prices imply that the old method would yield lower, not higher inflation. Not much of a response.
M-3 No reporting since 2006?
My analysis shows average annual increases as follow:
Reagan-320 B per Year
Bush I-59 B per Year
Clinton-330B per Year
BushII-720B per year first six years
Many dollars=dollar devaluation
What gives? Why not reported?
Shadowstats adjustments just don’t seem to reflect reality.
Look at the first chart here
Look how big it considers inflation understatements to be since the mid 90s. We would have been having negative growth almost every year from then to now. Even during the tech stock bubble we would have been shrinking.
The total economic contraction would be larger than the great depression was in the US.
That’s just ridiculous.