How accurately can we measure INFLATION?

Discuss all general (i.e. non-personal) investing questions and issues, investing news, and theory.
User avatar
grayfox
Posts: 4836
Joined: Sat Sep 15, 2007 4:30 am

How accurately can we measure INFLATION?

Post by grayfox » Sun Apr 15, 2018 2:40 pm

Looking at the latest (Apr 13, 2018) inflation report at the Department of Labor Bureau of Labor Statistics inflation report:

Consumer Price Index rose 2.4 percent over the year ending March 2018

They show a chart with 12-month percentage change in Consumer Price Index, not seasonally adjusted. All Items oscillates slightly above zero, maybe between 0 and 5%. Energy fluctuates between about -30% and +30%. They show the CPI-U rose 2.4 percent from Mar-2017 to Mar-2018.

For investors, measuring inflation is important. It determines:

What the cost-of-living increase is for social security recipients.
How much do Inflation-indexed annuities pay out each month.
How much do your inflation-indexed bonds pay back when they mature?
Real GDP growth. Real GDP is calculated from Nominal GDP and an inflation measure.
Calculation of real returns and many other numbers that are adjusted for inflation.
All those SWR studies use inflation numbers.
Etc, etc.

This got me thinking just how accurate is the year-over-year inflation number. They show it to 1 decimal place. Does that imply that it is known to +/- 5 bps? Not 2.3%, not 2.5%, but 2.35% to 2.45%.

Consider how the number is computed. It is an annual rate. They take the difference between the CPI-U level, 12 months apart. In the full new release, CPI-U is shown to 3 decimal places.

Mar-2018 249.554
Mar-2017 243.801
Change 5.753
Pct Change 2.360%

So why not show 2.360%, 3 decimal places? Why round off to 1 decimal, 2.4%

:?: How accurate can we measure INFLATION?
Last edited by grayfox on Sun Apr 15, 2018 2:42 pm, edited 1 time in total.

User avatar
dm200
Posts: 15626
Joined: Mon Feb 26, 2007 2:21 pm
Location: Washington DC area

Re: How accurately can we measure INFLATION?

Post by dm200 » Sun Apr 15, 2018 2:42 pm

It is very complex - and subject to many complications and assumptions.

Then - add the politics ..

User avatar
Pajamas
Posts: 4592
Joined: Sun Jun 03, 2012 6:32 pm

Re: How accurately can we measure INFLATION?

Post by Pajamas » Sun Apr 15, 2018 2:47 pm

grayfox wrote:
Sun Apr 15, 2018 2:40 pm

So why not show 2.360%, 3 decimal places? Why round off to 1 decimal, 2.4%

:?: How accurate can we measure INFLATION?

If you spend $100,000 a year, the difference between 2.36% inflation and 2.4% inflation is 0.0004 x $100,000 = $40. My conclusion is that it is a meaningless difference in practice for individuals.

Research has been done on this. For example:

https://www.bls.gov/ore/abstract/ec/ec060090.htm

User avatar
grayfox
Posts: 4836
Joined: Sat Sep 15, 2007 4:30 am

Re: How accurately can we measure INFLATION?

Post by grayfox » Sun Apr 15, 2018 2:49 pm

Since this is the BHs board, we can neither add not subtract the politics.

What I recall from math class, when a number is given to 1 decimal place, without any error bars, that implies some precision.
2.4 would imply that the number lies in the range 2.35 to 2.45. No?

My question is: How accurately can we measure INFLATION?

Does the True Value of inflation lie in the range (2.4 +/-0.05) ?

Regarding that paper, it is from Aug-2006. It looks to me like BLS change CPI-U from 1 digit to 3 digits in Jan-2007
In that case, does the True Value of inflation lie in the range 2.360 +/- 0.005 ?

Or does the True Value of INFLATION lie in a wider range?
Last edited by grayfox on Sun Apr 15, 2018 3:02 pm, edited 2 times in total.

User avatar
dm200
Posts: 15626
Joined: Mon Feb 26, 2007 2:21 pm
Location: Washington DC area

Re: How accurately can we measure INFLATION?

Post by dm200 » Sun Apr 15, 2018 2:55 pm

grayfox wrote:
Sun Apr 15, 2018 2:49 pm
Since this is the BHs board, we can neither add not subtract the politics.

What I recall from math class, when a number is given to 1 decimal place, without any error bars, that implies some precision.
2.4 would imply that the number lies in the range 2.35 to 2.45. No?
No advanced math expert, but the challenges are assumptions:

1. What is the "market basket" being measured - different folks have different ones, such as older vs younger, higher vs lower income, etc.

2. The mix changes as prices change - and how to measure - so, if beef goes up 20% but chicken goes up 1% - people may eat more chicken and less beef - how do you change the market basket mix?

3. Quality improvements - how to measure? So, a basic car might increase 3.00% in price in one year. However, the car may have improved safety features and better fuel economy. That justifies a figure under 3%. How much?

User avatar
Pajamas
Posts: 4592
Joined: Sun Jun 03, 2012 6:32 pm

Re: How accurately can we measure INFLATION?

Post by Pajamas » Sun Apr 15, 2018 2:59 pm

grayfox wrote:
Sun Apr 15, 2018 2:49 pm
Since this is the BHs board, we can neither add not subtract the politics.

What I recall from math class, when a number is given to 1 decimal place, without any error bars, that implies some precision.
2.4 would imply that the number lies in the range 2.35 to 2.45. No?
In this case there is a margin of error that is discussed in the press release. Reporting additional decimal places would imply a degree of accuracy that is not present.

alex_686
Posts: 3206
Joined: Mon Feb 09, 2015 2:39 pm

Re: How accurately can we measure INFLATION?

Post by alex_686 » Sun Apr 15, 2018 3:16 pm

A big issue is how to measure changes in quality. Compare a car from the 1970s to one today. Today's car is more expensive. Part of that is from inflation. Part of that is that today's car are larger, safer, more fuel-efficient, have more features (like GPS), are more durable, etc. From year to year the changes are small. Over 10 years the changes are huge.

User avatar
grayfox
Posts: 4836
Joined: Sat Sep 15, 2007 4:30 am

Re: How accurately can we measure INFLATION?

Post by grayfox » Sun Apr 15, 2018 3:21 pm

OK, I see below Table 7. Consumer Price Index for All Urban Consumers (CPI-U): U.S. city average, by expenditure category, March 2018, 12-month analysis table

Footnote 2
A statistic’s margin of error is often expressed as its point estimate plus or minus two standard errors. For example, if a CPI category rose 2.6 percent, and its standard error was 0.25 percent, the margin of error on this item’s 12-month percent change would be 2.6 percent, plus or minus 0.5 percent.
All Items
Unadjusted percent change 2.4
Standard error 0.07

The margin of error = 2.4 +/- 2 * 0.07
The range is 2.26 to 2.54

So that answers the question "How accurately can they measure change in CPI-U."

:arrow: According to BLS CPI-U, 12 month INFLATION was between 2.26% and 2.54%
Last edited by grayfox on Sun Apr 15, 2018 3:33 pm, edited 3 times in total.

User avatar
dm200
Posts: 15626
Joined: Mon Feb 26, 2007 2:21 pm
Location: Washington DC area

Re: How accurately can we measure INFLATION?

Post by dm200 » Sun Apr 15, 2018 3:29 pm

alex_686 wrote:
Sun Apr 15, 2018 3:16 pm
A big issue is how to measure changes in quality. Compare a car from the 1970s to one today. Today's car is more expensive. Part of that is from inflation. Part of that is that today's car are larger, safer, more fuel-efficient, have more features (like GPS), are more durable, etc. From year to year the changes are small. Over 10 years the changes are huge.
Very true (in many ways). If you want a cheap car, though, you cannot get a 1960's or 70's VW beetle.

User avatar
dm200
Posts: 15626
Joined: Mon Feb 26, 2007 2:21 pm
Location: Washington DC area

Re: How accurately can we measure INFLATION?

Post by dm200 » Sun Apr 15, 2018 3:31 pm


JBTX
Posts: 2968
Joined: Wed Jul 26, 2017 12:46 pm

Re: How accurately can we measure INFLATION?

Post by JBTX » Sun Apr 15, 2018 4:16 pm

It is 100% accurate, based on the given assumptions. Of course, those assumptions probably won’t 100% apply to everybody, or any one individual.

The big variables are:

Substitution - how much will you change your consumption from product A to product B, which are close but not exact substitutes if product A rises and B falls. If you stick with A, your inflation will be worse. If you already used B, you will have deflation. The more dynamic the substitution allowed in the index, the lower it is.

Quality/Technology - the CPI attempts to adjust for the fact that a car or a TV you buy now is better than ten years ago. You can certainly see it with cars, certain features which were luxury 5-10 years ago are standard now. However, I’m not sure that is how most people behave. If you buy luxury cars you will likely continue to do so.

Also, CPI includes rent for housing, but not house prices, because houses are considered a long term asset/investment. So if you own a house, the rent portion of the CPI is somewhat inapplicable to you. Instead you will have inflation of certain components of housing.

User avatar
grayfox
Posts: 4836
Joined: Sat Sep 15, 2007 4:30 am

Re: How accurately can we measure INFLATION?

Post by grayfox » Sun Apr 15, 2018 4:21 pm

Consumer Price Index Frequently Asked Questions (FAQs)
14. Is the CPI the best measure of inflation?
Various indexes have been devised to measure different aspects of inflation. Inflation has been defined as a process of continuously rising prices or, equivalently, of a continuously falling value of money. The CPI measures inflation as experienced by consumers in their day-to-day living expenses; the Producer Price Index (PPI) measures inflation at earlier stages of the production process; the International Price Program (IPP) measures inflation for imports and exports; the Employment Cost Index (ECI) measures inflation in the labor market; and the Gross Domestic Product (GDP) Deflator measures inflation experienced by both consumers themselves as well as governments and other institutions providing goods and services to consumers. There are also specialized measures, such as measures of interest rates.
So there are various measures of inflation that we can look at. I looked up PPI.

The Mar-2018 reports 12 month change in final demand index +3.0%.
The latest PPI variance estimates are for 2016 Data. 2016 Standard Error = 0.33 percent
If I use 2016 SE, the range is 2.34 to 3.66

CPI showed 2.26 to 2.54
PPI showed 2.34 to 3.66 (using Standard Error from 2016)

I wonder what the GDP Deflator would show.

User avatar
grayfox
Posts: 4836
Joined: Sat Sep 15, 2007 4:30 am

Re: How accurately can we measure INFLATION?

Post by grayfox » Sun Apr 15, 2018 4:51 pm

More inflation measures from BLS: CPI-W. This used to annually adjust social security benefits.

CPI-Urban Wage Earners and Clerical Workers (Current Series)

U.S. All Items +2.433%
Northeast Region +2.043%
Midwest Region +1.947%
South Region +2.328%
West Region +3.336%
NY +1.874%
DC +1.834%
CHI +1.692%
LA +3.886%

So depending on where you live, inflation you experience was anywhere from about 1.7% to 3.9%.
(Just regional variance, not including statistical variance, i.e. standard errors)

User avatar
grayfox
Posts: 4836
Joined: Sat Sep 15, 2007 4:30 am

Re: How accurately can we measure INFLATION?

Post by grayfox » Sun Apr 15, 2018 4:59 pm

alex_686 wrote:
Sun Apr 15, 2018 3:16 pm
A big issue is how to measure changes in quality. Compare a car from the 1970s to one today. Today's car is more expensive. Part of that is from inflation. Part of that is that today's car are larger, safer, more fuel-efficient, have more features (like GPS), are more durable, etc. From year to year the changes are small. Over 10 years the changes are huge.
How do you quantify changes in quality like that?
GPS, add the price of a standalone GPS unit? At one time there was no GPS. Add the price of a map and compass? :)
Larger, normalize to the number of cubic inches?
Safer, add the savings in insurance?

It seems very subjective. Which is bad, because then we are not just dealing with statistical sampling errors, but with the biases and prejudices of whoever is gathering the data and making the calculations.

delamer
Posts: 4142
Joined: Tue Feb 08, 2011 6:13 pm

Re: How accurately can we measure INFLATION?

Post by delamer » Sun Apr 15, 2018 5:22 pm

grayfox wrote:
Sun Apr 15, 2018 4:59 pm
alex_686 wrote:
Sun Apr 15, 2018 3:16 pm
A big issue is how to measure changes in quality. Compare a car from the 1970s to one today. Today's car is more expensive. Part of that is from inflation. Part of that is that today's car are larger, safer, more fuel-efficient, have more features (like GPS), are more durable, etc. From year to year the changes are small. Over 10 years the changes are huge.
How do you quantify changes in quality like that?
GPS, add the price of a standalone GPS unit? At one time there was no GPS. Add the price of a map and compass? :)
Larger, normalize to the number of cubic inches?
Safer, add the savings in insurance?

It seems very subjective. Which is bad, because then we are not just dealing with statistical sampling errors, but with the biases and prejudices of whoever is gathering the data and making the calculations.
There are detailed procedures in place that the individual data collectors use to report changes in the items being priced.

It isn’t as each data collector who prices cars or houses or bananas or whatever makes up their own rules as to how to report the prices or changes in the items.

Of course there is some subjectivity in developing the procedures. If it was easy and totally objective, then there’d be no discussions or controversy.

talzara
Posts: 446
Joined: Thu Feb 12, 2009 7:40 pm

Re: How accurately can we measure INFLATION?

Post by talzara » Sun Apr 15, 2018 5:26 pm

grayfox wrote:
Sun Apr 15, 2018 4:59 pm
alex_686 wrote:
Sun Apr 15, 2018 3:16 pm
A big issue is how to measure changes in quality. Compare a car from the 1970s to one today. Today's car is more expensive. Part of that is from inflation. Part of that is that today's car are larger, safer, more fuel-efficient, have more features (like GPS), are more durable, etc. From year to year the changes are small. Over 10 years the changes are huge.
How do you quantify changes in quality like that?
GPS, add the price of a standalone GPS unit? At one time there was no GPS. Add the price of a map and compass? :)
Larger, normalize to the number of cubic inches?
Safer, add the savings in insurance?
The BLS does not do quality adjustment on car prices. They only do it for a few products: https://www.bls.gov/cpi/quality-adjustment/home.htm

Auto insurance is already included in CPI. The price of the car does not have to be adjusted for safety.

Gasoline is already included in CPI. The price of the car does not have to be adjusted for fuel efficiency.

It wouldn't be hard to do quality adjustment for GPS units. Just look at what it costs to add GPS as an optional upgrade.

User avatar
grayfox
Posts: 4836
Joined: Sat Sep 15, 2007 4:30 am

Re: How accurately can we measure INFLATION?

Post by grayfox » Sun Apr 15, 2018 5:47 pm

Here is another inflation measure The Billion Prices Project

Looking at the bottom chart, US Online Annual Inflation Rate.
Some of the biggest differences I see

Date.............Pricestats (blue) ...... BLS (red) ..... Pricestats-BLS
21jul2010....... 2.83 ................ 1.05 ................ +1.78
10aug2010....... 2.83 ................ 1.24 ................ +1.59
07oct2015....... -1.38 ............... -0.04 ................ -1.34
07mar2017....... 3.58 ................ 2.74 ................ +0.84

It looks like PriceStats has bigger swings in the annual inflation rate than BLS.
In round numbers, since 2008 Pricestats has been within about +/- 1.5% of BLS.

I think this is good information, because it is an independent measure of inflation.
Last edited by grayfox on Mon Apr 16, 2018 5:55 pm, edited 1 time in total.

AlohaJoe
Posts: 3056
Joined: Mon Nov 26, 2007 2:00 pm
Location: Saigon, Vietnam

Re: How accurately can we measure INFLATION?

Post by AlohaJoe » Sun Apr 15, 2018 7:15 pm

grayfox wrote:
Sun Apr 15, 2018 2:40 pm
For investors, measuring inflation is important. It determines:
It is worth pointing out that CPI is not "cost of living", so things like SWR studies and calculations of real returns that use CPI are somewhat misleading. Traditionally people think of inflation as "increase in cost of living" so the CPI is (sort of, kind of) the wrong this for measuring that as well.

phantom0308
Posts: 11
Joined: Thu Aug 25, 2016 6:52 pm

Re: How accurately can we measure INFLATION?

Post by phantom0308 » Sun Apr 15, 2018 7:50 pm

There's a joint effort by Harvard and MIT to measure inflation in different countries using online prices called the billion prices project. Their data shows around 1% tracking error with CPI over 8 years. http://www.thebillionpricesproject.com/datasets/

I'd guess 0.01 precision is too high from skimming the website and other inflation explanations, but I haven't done the research to support this claim. I'm fairly convinced that theories I've read on line that CPI is actually really high and the real data is being hidden by bad people are wrong though.

AlohaJoe
Posts: 3056
Joined: Mon Nov 26, 2007 2:00 pm
Location: Saigon, Vietnam

Re: How accurately can we measure INFLATION?

Post by AlohaJoe » Sun Apr 15, 2018 8:11 pm

phantom0308 wrote:
Sun Apr 15, 2018 7:50 pm
There's a joint effort by Harvard and MIT to measure inflation in different countries using online prices called the billion prices project.
This isn't an effort to measure inflation per se: as the name says it just measures prices. All of those prices need to be combined somehow and as far I can tell they still rely on the government (via consumer expenditure surveys) to tell them what a "representative basket of goods" is and the relative weights of each.

The biggest thing they offer is the ability to sample a lot more prices on a daily basis and to do so in countries where the official statistics bureau is manipulating inflation.

Nowadays (in the US and other similar countries) the argument (from those who argue) isn't that the prices are wrong but that the "representative basket of goods" is wrong. This is the argument that the CPI crazies make: that the government has (subtly) twisted the "basket of goods" (via substitution) to destroy the meaningfulness of CPI. That is, they would say that the data from the Billion Prices project doesn't disprove their claims because they never claimed the government was actively lying about the actual price of a TV.

dknightd
Posts: 214
Joined: Wed Mar 07, 2018 11:57 am

Re: How accurately can we measure INFLATION?

Post by dknightd » Sun Apr 15, 2018 8:18 pm

Maybe somebody decided to show 3 decimal places instead of one when "inflation" got lower?
There are so many different ways to measure inflation. I'm not sure any of them are right.

Jags4186
Posts: 1839
Joined: Wed Jun 18, 2014 7:12 pm

Re: How accurately can we measure INFLATION?

Post by Jags4186 » Sun Apr 15, 2018 8:27 pm

What’s more important is your personal inflation rate rather than the CPI inflation rate. Oil prices will matter more to a person with oil heat in the midwest than it will to someone who lives in San Diego. Increased housing costs will matter more to the permanent renter than to the home owner with a paid off home. Lifestyle creep will beat any inflation number.

Do you know what your personal inflation rate is? What did you spend in 2016 vs 2017?

User avatar
grayfox
Posts: 4836
Joined: Sat Sep 15, 2007 4:30 am

Re: How accurately can we measure INFLATION?

Post by grayfox » Mon Apr 16, 2018 2:45 am

BHs are often interested in "Real" returns. Take the nominal return and adjust for inflation.
OK, which measure of inflation? It's always the 12-month change in CPI-U.

Is that really the best choice to adjust for inflation? Everyone using the exact same inflation adjustment?
Based on some made-up basket of goods that no one actually buys.

Eg. An investor in NY and an investor in LA each buy a Treasury that returns 3%.
Both adjust for 2.4% annual CPI-U inflation. Real Return = 0.59%

:idea: Instead adjust for CPI-W in their city
NY = 1.874%; RR = 1.11%
LA = 3.886%; RR = -0.85%

The guy in LA had negative real return. :x
The guy in NY had higher real return than CPI-U would indicate. 8-)
Same investment, same nominal return, different real return.

The previous poster says use your individual inflation rate. Then, for the same nominal return, the real return varies from person to person.

:?: So why does everyone use the same inflation rate to calculate real return?

Valuethinker
Posts: 34529
Joined: Fri May 11, 2007 11:07 am

Re: How accurately can we measure INFLATION?

Post by Valuethinker » Mon Apr 16, 2018 6:09 am

dm200 wrote:
Sun Apr 15, 2018 3:29 pm
alex_686 wrote:
Sun Apr 15, 2018 3:16 pm
A big issue is how to measure changes in quality. Compare a car from the 1970s to one today. Today's car is more expensive. Part of that is from inflation. Part of that is that today's car are larger, safer, more fuel-efficient, have more features (like GPS), are more durable, etc. From year to year the changes are small. Over 10 years the changes are huge.
Very true (in many ways). If you want a cheap car, though, you cannot get a 1960's or 70's VW beetle.
You can buy an old one-- but they have scarcity value?

The minimum specs for new cars have increased:

- most of us wouldn't want to drive the safety standards of, say, the early 70s. Rigid steering columns. No airbags. No anti lock braking. More to the point we wouldn't want *other* drivers to be without anti-lock braking?

Presumably there are flightworthy Boeing 707s out there somewhere. Would we want to fly in them, if they did not meet modern standards of instrumentation and control?

- none of us, I presume, would want to go back to the air pollution of Los Angeles or Tokyo in the early 1970s. Thus, we would not want *other* drivers to be driving vehicles that polluted as those ones did, or gas pumps without vapour locks, etc.

We will see the same effect as and when autonomous vehicles come in-- in time, we will not want other drivers to be on manual control in situations where they could harm us. I have no doubt that this will cause great teeth gnashing, complaints about government interference in the private sphere, etc. I also have no doubt that eventually we will succumb on this one-- too many of us have lost friends and family to car accidents.

User avatar
AtlasShrugged?
Posts: 504
Joined: Wed Jul 15, 2015 6:08 pm

Re: How accurately can we measure INFLATION?

Post by AtlasShrugged? » Mon Apr 16, 2018 6:20 am

One thing I have never understood. When inflation numbers are presented, they are usually presented with this caveat, "....these numbers strip out food and energy, because of their volatility", or something to that effect. My question is why strip out Food and Energy? Food and Energy are huge parts of our lives. And we do not live our lives in neat linear lines.

Maybe an academic can address that question. It is one I have often wondered about.
“If you don't know, the thing to do is not to get scared, but to learn.”

AlohaJoe
Posts: 3056
Joined: Mon Nov 26, 2007 2:00 pm
Location: Saigon, Vietnam

Re: How accurately can we measure INFLATION?

Post by AlohaJoe » Mon Apr 16, 2018 6:38 am

AtlasShrugged? wrote:
Mon Apr 16, 2018 6:20 am
One thing I have never understood. When inflation numbers are presented, they are usually presented with this caveat, "....these numbers strip out food and energy, because of their volatility", or something to that effect. My question is why strip out Food and Energy? Food and Energy are huge parts of our lives. And we do not live our lives in neat linear lines.

Maybe an academic can address that question. It is one I have often wondered about.
If you type it into Google you won't have to wonder any longer :D

User avatar
AtlasShrugged?
Posts: 504
Joined: Wed Jul 15, 2015 6:08 pm

Re: How accurately can we measure INFLATION?

Post by AtlasShrugged? » Mon Apr 16, 2018 7:08 am

If you type it into Google you won't have to wonder any longer :D
AlohaJoe....Funny. :happy :sharebeer

More to the point, I guess this is a philosophical difference. I would maintain that our inflation measure should include both, as a more true reflection of how we actually live our lives. Our lives are variable, they are not perfectly linear. Inflation is variable, it is not perfectly linear. In my view, it would create a much more nuanced way to look at inflation. But that is just me. So how is my point actionable?

I have thought about calculating my personal rate of inflation. For instance, in BigCorp, we recently went through our benefits selection process. My DW and I made a series of trade-offs in coverage and benefits (more on some, dropping child on others, stepping down life insurance for DW, etc). I mean, it is a real mish-mosh. How do I parse out 'inflation' from my selections? Does a lower premium but higher deductible equal inflation? Not sure. Layer an FSA on top of that, then how does the calculation change? This idea of calculating my personal rate of inflation is one I am really grappling with. Because it is my personal rate of inflation that will really determine the viability of my IPS and retirement planning.

However, calculating one's personal rate of inflation is not so easy, as I have come to discover. Alas, I am probably overthinking it. :oops:
“If you don't know, the thing to do is not to get scared, but to learn.”

User avatar
SimpleGift
Posts: 2882
Joined: Tue Feb 08, 2011 3:45 pm
Location: Central Oregon

Re: How accurately can we measure INFLATION?

Post by SimpleGift » Mon Apr 16, 2018 9:08 am

AtlasShrugged? wrote:
Mon Apr 16, 2018 6:20 am
One thing I have never understood. When inflation numbers are presented, they are usually presented with this caveat, "....these numbers strip out food and energy, because of their volatility", or something to that effect. My question is why strip out Food and Energy?
My understanding is that it is just to smooth out the volatility of the CPI-U data series (in blue, chart below) — which is much more volatile with energy especially included in it.
The CPI-U series less food and energy (in red) gives a better sense of trend inflation, without the gyrations of the oil markets.
Cordially, Todd

User avatar
dm200
Posts: 15626
Joined: Mon Feb 26, 2007 2:21 pm
Location: Washington DC area

Re: How accurately can we measure INFLATION?

Post by dm200 » Mon Apr 16, 2018 9:12 am

How do you (or would you) factor in such items as cable TV or Cell phones? Would we regard these as part of out personal inflation?

Yes - they are a part of "normal" expenditures for most of us. However, they are the result (in my opinion) of financial choices we make to spend more money.

User avatar
grayfox
Posts: 4836
Joined: Sat Sep 15, 2007 4:30 am

Re: How accurately can we measure INFLATION?

Post by grayfox » Mon Apr 16, 2018 9:14 am

AtlasShrugged? wrote:
Mon Apr 16, 2018 7:08 am

However, calculating one's personal rate of inflation is not so easy, as I have come to discover. Alas, I am probably overthinking it. :oops:
It is not easy.

I have accurate records of my spending for every year. I save all my receipts and enter everything into gnucash and libreoffice. I know how much I spent, to the penny, in 2017, 2016, 2015...

2013 -6.06%
2015 -12.80%
2016 +13.88%

Wow, huge personal deflation in 2013 and 2014.
Then huge inflation in 2016.

Is that how it works? I don't think so.

E.g. one year I bought plane tickets to Europe for about $600. Then next year about $1200.
Plane ticket inflation +100% ?
No, one year I went in January (low-season) and the next year I went in July (high season)

Unless you buy the exact same items every year, at the exact same time and place, it is not easy to determine personal inflation, even if you have records of every expenditure.

adam1712
Posts: 311
Joined: Fri Jun 01, 2007 5:21 pm

Re: How accurately can we measure INFLATION?

Post by adam1712 » Mon Apr 16, 2018 9:30 am

grayfox wrote:
Mon Apr 16, 2018 2:45 am
BHs are often interested in "Real" returns. Take the nominal return and adjust for inflation.
OK, which measure of inflation? It's always the 12-month change in CPI-U.

Is that really the best choice to adjust for inflation? Everyone using the exact same inflation adjustment?
Based on some made-up basket of goods that no one actually buys.

Eg. An investor in NY and an investor in LA each buy a Treasury that returns 3%.
Both adjust for 2.4% annual CPI-U inflation. Real Return = 0.59%

:idea: Instead adjust for CPI-W in their city
NY = 1.874%; RR = 1.11%
LA = 3.886%; RR = -0.85%

The guy in LA had negative real return. :x
The guy in NY had higher real return than CPI-U would indicate. 8-)
Same investment, same nominal return, different real return.

The previous poster says use your individual inflation rate. Then, for the same nominal return, the real return varies from person to person.

:?: So why does everyone use the same inflation rate to calculate real return?
I'm of the opinion we all spend dollars so we all have the same inflation rate. Where you choose to live and what you choose to spend your money on are spending choices. But this all comes down to semantics. I would propose your year to year change in spending depends on three factors:

Change in annual spending = monetary inflation + personal inflation + change in goods purchased

-monetary inflation is the change in the dollar in aggregate for everybody who uses dollars
-personal inflation is how much the prices in your personal basket of goods changes, I personally don't like the term personal inflation but people seem to like it
-change in goods purchased, changes in your basket of goods.

We have pretty reasonable measures of monetary inflation with CPI-U and billion price project. One can track their spending so the hard part is dissociating personal inflation from changes in goods purchased. I'm not sure that separating the two to have a precise estimate of each is meaningful.

User avatar
grayfox
Posts: 4836
Joined: Sat Sep 15, 2007 4:30 am

Re: How accurately can we measure INFLATION?

Post by grayfox » Mon Apr 16, 2018 9:38 am

Image
Source: FRED


That FRED graph looks wrong.
Click through for the original chart and data.
It shows CPI-U Change from year ago was 5.745% in March 2018. ?!?

BLS reports showed only 2.4%.

Did FRED made an error? Somebody explain how they are showing 5.745% change from year ago.

Update: OK, that is not the percent change. They just subtracted the CPI-U this year from last year.
Nevermind.
Last edited by grayfox on Mon Apr 16, 2018 9:49 am, edited 1 time in total.

User avatar
grayfox
Posts: 4836
Joined: Sat Sep 15, 2007 4:30 am

Re: How accurately can we measure INFLATION?

Post by grayfox » Mon Apr 16, 2018 9:45 am

adam1712 wrote:
Mon Apr 16, 2018 9:30 am

I'm of the opinion we all spend dollars so we all have the same inflation rate. Where you choose to live and what you choose to spend your money on are spending choices. But this all comes down to semantics. I would propose your year to year change in spending depends on three factors:

Change in annual spending = monetary inflation + personal inflation + change in goods purchased

-monetary inflation is the change in the dollar in aggregate for everybody who uses dollars
-personal inflation is how much the prices in your personal basket of goods changes, I personally don't like the term personal inflation but people seem to like it
-change in goods purchased, changes in your basket of goods.
We all spend dollars so we all have the same inflation rate.

Yes.

So the inflation part of your equation is associated with the money we all use. The rest of the changes have to do with a person's habits, spending choices that vary from year to year, timing of purchases, choosing to buying stuff only when on sale, etc.

But since we all use the same money, the inflation part is the same for everybody.

talzara
Posts: 446
Joined: Thu Feb 12, 2009 7:40 pm

Re: How accurately can we measure INFLATION?

Post by talzara » Mon Apr 16, 2018 10:32 am

dm200 wrote:
Mon Apr 16, 2018 9:12 am
How do you (or would you) factor in such items as cable TV or Cell phones? Would we regard these as part of out personal inflation?

Yes - they are a part of "normal" expenditures for most of us. However, they are the result (in my opinion) of financial choices we make to spend more money.
CPI already includes the cost of cable television and telephone service. It's a measure of what consumers are actually spending their money on.

User avatar
dm200
Posts: 15626
Joined: Mon Feb 26, 2007 2:21 pm
Location: Washington DC area

Re: How accurately can we measure INFLATION?

Post by dm200 » Mon Apr 16, 2018 10:36 am

talzara wrote:
Mon Apr 16, 2018 10:32 am
dm200 wrote:
Mon Apr 16, 2018 9:12 am
How do you (or would you) factor in such items as cable TV or Cell phones? Would we regard these as part of out personal inflation?

Yes - they are a part of "normal" expenditures for most of us. However, they are the result (in my opinion) of financial choices we make to spend more money.
CPI already includes the cost of cable television and telephone service. It's a measure of what consumers are actually spending their money on.
Sure, it now includes them. At one time, these did not exist - that is my point.

User avatar
Oicuryy
Posts: 1201
Joined: Thu Feb 22, 2007 10:29 pm

Re: How accurately can we measure INFLATION?

Post by Oicuryy » Mon Apr 16, 2018 10:38 am

How accurately can we define "inflation"?
On the Origin and Evolution of the Word Inflation

Ron
Money is fungible | Abbreviations and Acronyms

delamer
Posts: 4142
Joined: Tue Feb 08, 2011 6:13 pm

Re: How accurately can we measure INFLATION?

Post by delamer » Mon Apr 16, 2018 10:42 am

dm200 wrote:
Mon Apr 16, 2018 10:36 am
talzara wrote:
Mon Apr 16, 2018 10:32 am
dm200 wrote:
Mon Apr 16, 2018 9:12 am
How do you (or would you) factor in such items as cable TV or Cell phones? Would we regard these as part of out personal inflation?

Yes - they are a part of "normal" expenditures for most of us. However, they are the result (in my opinion) of financial choices we make to spend more money.
CPI already includes the cost of cable television and telephone service. It's a measure of what consumers are actually spending their money on.
Sure, it now includes them. At one time, these did not exist - that is my point.
As consumer expenditures change over time, the CPI changes its weights to reflect those changes.

So cell phones are added in and buggy whips are taken out. It is the change in prices being measured.

User avatar
grayfox
Posts: 4836
Joined: Sat Sep 15, 2007 4:30 am

Re: How accurately can we measure INFLATION?

Post by grayfox » Mon Apr 16, 2018 11:07 am

adam1712 wrote:

"We all spend dollars so we all have the same inflation rate."

Bravo! I am going to take that as an axiom:

Axiom 1. For any currency, e.g. dollars, Euros, there is only one inflation rate.

All the differences have to do with a one's habits, spending choices, timing of purchases, location of purchase, etc.


And yet, all the many measures of inflation:

CPI-U 2.26 to 2.54
PPI 2.34 to 3.66
CPI-W U.S. All Items +2.433%
CPI-W Northeast Region +2.043%
CPI-W Midwest Region +1.947%
CPI-W South Region +2.328%
CPI-W West Region +3.336%
CPI-W NY +1.874%
CPI-W DC +1.834%
CPI-W CHI +1.692%
CPI-W LA +3.886%
etc.

show different values for inflation. Anywhere from about 2.3 to 3.9
BPP has differed from CPI-U by +/- 1.5%, not even considering the error bars.

Which one is the true inflation. Any? None? Some average?
Why would CPI-U be better than BPP or CPI-W?

User avatar
grayfox
Posts: 4836
Joined: Sat Sep 15, 2007 4:30 am

Re: How accurately can we measure INFLATION?

Post by grayfox » Mon Apr 16, 2018 11:51 am

Oicuryy wrote:
Mon Apr 16, 2018 10:38 am
How accurately can we define "inflation"?
On the Origin and Evolution of the Word Inflation

Ron
Wow! That opens up a can of worms!

I think that the ideas in that article can explain why all these price indexes show different values for inflation. It's because they are just showing the price changes for whatever market basket the index uses. They all use different market baskets, so each number comes out different.

Only part of the price changes they report comes from inflation. The rest comes from other things.

In other words, we are conflating price changes with inflation.

User avatar
Oicuryy
Posts: 1201
Joined: Thu Feb 22, 2007 10:29 pm

Re: How accurately can we measure INFLATION?

Post by Oicuryy » Mon Apr 16, 2018 11:53 am

The Consumer Price Index (CPI) does not measure inflation. It measures the change in the price of a basket of consumer goods and services.

The price of a thing, in dollars, is the ratio of the value of the thing to the value of a dollar. A price of five dollars for some thing means the value of the thing is five times the value of a dollar. A price could increase because the value of the thing went up and/or because the value of a dollar went down.

Axiom 1 implies that inflation is a decline in the value of a dollar. The trick to measuring inflation is to figure out how much of a change in prices is due to a change in the value of things and how much is due to a decline in the value of a dollar.

Ron
Money is fungible | Abbreviations and Acronyms

PVW
Posts: 343
Joined: Mon Dec 29, 2014 11:01 am

Re: How accurately can we measure INFLATION?

Post by PVW » Mon Apr 16, 2018 12:52 pm

grayfox wrote:
Mon Apr 16, 2018 9:45 am
adam1712 wrote:
Mon Apr 16, 2018 9:30 am

I'm of the opinion we all spend dollars so we all have the same inflation rate. Where you choose to live and what you choose to spend your money on are spending choices. But this all comes down to semantics. I would propose your year to year change in spending depends on three factors:

Change in annual spending = monetary inflation + personal inflation + change in goods purchased

-monetary inflation is the change in the dollar in aggregate for everybody who uses dollars
-personal inflation is how much the prices in your personal basket of goods changes, I personally don't like the term personal inflation but people seem to like it
-change in goods purchased, changes in your basket of goods.
We all spend dollars so we all have the same inflation rate.

Yes.

So the inflation part of your equation is associated with the money we all use. The rest of the changes have to do with a person's habits, spending choices that vary from year to year, timing of purchases, choosing to buying stuff only when on sale, etc.

But since we all use the same money, the inflation part is the same for everybody.
Money has no intrinsic value, it is worth what you can trade it for. When people put inflation in the context of the changing value of money, they are really talking about the average change in monetary value of the goods and services that you could buy with your money. There are different inflation rates among the goods and services and your personal inflation rate depends on what goods and services you buy with your money.

adamthesmythe
Posts: 1894
Joined: Mon Sep 22, 2014 4:47 pm

Re: How accurately can we measure INFLATION?

Post by adamthesmythe » Mon Apr 16, 2018 1:04 pm

OP has confused precision and accuracy.

Any well-defined measure of inflation can be calculated to essentially arbitrary precision. You collect the prices you have decided are representative and you do high-precision arithmetic.

We can define accuracy only when the quantity "inflation" is well-defined. Then we can choose an experiment (a market basket) that we think approximates "inflation." After the experiment is done we can argue about whether the right approach was taken.

In the scientific literature quantities that are measured are well defined and the measurements converge to a consensus value. The measurement of inflation does not appear to be science.

User avatar
grayfox
Posts: 4836
Joined: Sat Sep 15, 2007 4:30 am

Re: How accurately can we measure INFLATION?

Post by grayfox » Mon Apr 16, 2018 1:05 pm

PVW wrote:
Mon Apr 16, 2018 12:52 pm

Money has no intrinsic value, it is worth what you can trade it for. When people put inflation in the context of the changing value of money, they are really talking about the average change in monetary value of the goods and services that you could buy with your money. There are different inflation rates among the goods and services and your personal inflation rate depends on what goods and services you buy with your money.
Suppose you make a $100 investment for 1 year. At the end of the year you get back $103.
The nominal return R is 3%

I believe that the real return RR can be calculated from the the nominal return R and the inflation rate i using this formula:

1 + RR = (1 + R) / (1 + i)

Which of the different inflation rates you mention would you use for i to calculate the real return RR?

Should everyone use the same i or a different i?

adam1712
Posts: 311
Joined: Fri Jun 01, 2007 5:21 pm

Re: How accurately can we measure INFLATION?

Post by adam1712 » Mon Apr 16, 2018 2:06 pm

Oicuryy wrote:
Mon Apr 16, 2018 11:53 am
The Consumer Price Index (CPI) does not measure inflation. It measures the change in the price of a basket of consumer goods and services.

The price of a thing, in dollars, is the ratio of the value of the thing to the value of a dollar. A price of five dollars for some thing means the value of the thing is five times the value of a dollar. A price could increase because the value of the thing went up and/or because the value of a dollar went down.

Axiom 1 implies that inflation is a decline in the value of a dollar. The trick to measuring inflation is to figure out how much of a change in prices is due to a change in the value of things and how much is due to a decline in the value of a dollar.

Ron
The CPI does try to remove some of the effects of the change in value of items in the basket of goods by reweighting the basket based on substitution effects and adding/removing items over time. Many economists however think there is still some effect from the change in value of items and CPI overstates inflation that is purely from the decline in the value of the dollar.

User avatar
Phineas J. Whoopee
Posts: 7046
Joined: Sun Dec 18, 2011 6:18 pm

Re: How accurately can we measure INFLATION?

Post by Phineas J. Whoopee » Mon Apr 16, 2018 3:15 pm

AtlasShrugged? wrote:
Mon Apr 16, 2018 6:20 am
One thing I have never understood. When inflation numbers are presented, they are usually presented with this caveat, "....these numbers strip out food and energy, because of their volatility", or something to that effect. My question is why strip out Food and Energy? Food and Energy are huge parts of our lives. And we do not live our lives in neat linear lines.

Maybe an academic can address that question. It is one I have often wondered about.
Here are the present weightings. Have a look for yourself. You will find food and energy in there.

The number based on the all-inclusive weightings is called CPI, or CPI-U or CPI-W (they have slightly different weightings). CPI-U is also referred to as headline inflation.

You're referring to "core CPI," which does indeed strip out food and energy because they are the most volatile components. It's useful for tracking longer term trends.

CPI does include both food and energy. See for yourself.

You may also be interested in Why the Published Averages Don't Always Match An Individual's Inflation Experience.

PJW

PVW
Posts: 343
Joined: Mon Dec 29, 2014 11:01 am

Re: How accurately can we measure INFLATION?

Post by PVW » Mon Apr 16, 2018 3:18 pm

grayfox wrote:
Mon Apr 16, 2018 1:05 pm

Suppose you make a $100 investment for 1 year. At the end of the year you get back $103.
The nominal return R is 3%

I believe that the real return RR can be calculated from the the nominal return R and the inflation rate i using this formula:

1 + RR = (1 + R) / (1 + i)

Which of the different inflation rates you mention would you use for i to calculate the real return RR?

Should everyone use the same i or a different i?
Your personal inflation rate is the only one that matters for your personal income and expenditures. Imagine your only expenditure is food and you get it all from a farmer that never raises his prices. If you get a 3% raise, then you can buy 3% more food. It doesn't matter that the average cost for all goods and services went up by some amount or that the value of money has deflated, you are still getting 3% more food than before so your personal inflation rate is 0%.

It's not so clear when figuring returns on your investments - you need the current inflation rate of the goods and services that you will buy in the future with your investments. In this case you might assume your personal inflation rate is the same as the average inflation rate, but defining a suitable average keeps a lot of smart people busy for entire careers.

User avatar
Phineas J. Whoopee
Posts: 7046
Joined: Sun Dec 18, 2011 6:18 pm

Re: How accurately can we measure INFLATION?

Post by Phineas J. Whoopee » Mon Apr 16, 2018 3:21 pm

PVW wrote:
Mon Apr 16, 2018 3:18 pm
...
Your personal inflation rate is the only one that matters for your personal income and expenditures. ...
There is no CPI-PVW any more than there is a CPI-Phineas J. Whoopee.

One's personal expenses vary, one's personal basket of goods and services is unlikely to resemble the one used for the nationwide average measure, and it's likely to change over time.

Of course personal expenses matter, but that's not what CPI measures.

PJW
Last edited by Phineas J. Whoopee on Mon Apr 16, 2018 3:31 pm, edited 1 time in total.

User avatar
Phineas J. Whoopee
Posts: 7046
Joined: Sun Dec 18, 2011 6:18 pm

Re: How accurately can we measure INFLATION?

Post by Phineas J. Whoopee » Mon Apr 16, 2018 3:25 pm

The OP asked a question that's more properly about precision rather than accuracy, examining the reported number of decimal places.

The thread has changed to discussing accuracy, which is in part based on what one means by the word inflation, as Ron posted upthread.

PJW

PVW
Posts: 343
Joined: Mon Dec 29, 2014 11:01 am

Re: How accurately can we measure INFLATION?

Post by PVW » Mon Apr 16, 2018 3:55 pm

Phineas J. Whoopee wrote:
Mon Apr 16, 2018 3:21 pm
PVW wrote:
Mon Apr 16, 2018 3:18 pm
...
Your personal inflation rate is the only one that matters for your personal income and expenditures. ...
There is no CPI-PVW any more than there is a CPI-Phineas J. Whoopee.

One's personal expenses vary, one's personal basket of goods and services is unlikely to resemble the one used for the nationwide average measure, and it's likely to change over time.

Of course personal expenses matter, but that's not what CPI measures.

PJW
The Bureau of Labor Statistics has not started reporting the CPI-PVW, but my only concern about the BLS measures of inflation are how they affect my government benefits.

There is a CPI-PVW and CPI-Phineas J. Whoopee. In figuring my expenditures, I only care what CPI-PVW is. Maybe your point is that I can't calculate CPI-PVW, and I agree, but that doens't change the fact that your personal inflation rate is the only one that matters for your expenses.

User avatar
Phineas J. Whoopee
Posts: 7046
Joined: Sun Dec 18, 2011 6:18 pm

Re: How accurately can we measure INFLATION?

Post by Phineas J. Whoopee » Mon Apr 16, 2018 4:18 pm

PVW wrote:
Mon Apr 16, 2018 3:55 pm
...
The Bureau of Labor Statistics has not started reporting the CPI-PVW, but my only concern about the BLS measures of inflation are how they affect my government benefits.

There is a CPI-PVW and CPI-Phineas J. Whoopee. In figuring my expenditures, I only care what CPI-PVW is. Maybe your point is that I can't calculate CPI-PVW, and I agree, but that doens't change the fact that your personal inflation rate is the only one that matters for your expenses.
Individual expenditures is not what CPI means at all.

If I like collecting vintage sports cars, and the price of vintage sports cars goes up, that's inflation.

If I stop spending money on vintage sports cars and start spending on cardiac care, stents for example, that is not inflation.

Inflation is one thing. Personal expenses are another.

PJW
Last edited by Phineas J. Whoopee on Mon Apr 16, 2018 4:23 pm, edited 1 time in total.

Post Reply