CPI: Precision ex-Accuracy


Consumer prices rose more than expected last month, with big price gains coming in food and medical care. Fortunately, these two items, along with energy prices, are totally irrelevant to our economy and personal consumption, and so don’t matter all that much to the Federal Reserve.

The increases in these two "non-essential items" were the biggest in 15 years.

CPI rose 0.2%, following the 0.4% gain in December. Core prices rose 0.3%.

Inflation ex-inflation was, as always, flat.

On a related note, the BLS has changed the way they are reporting CPI data.  Be sure to put this into the file marked Precision ex-Accuracy:

"The consumer price index for the first time will be carried out to three decimal places instead of one, a change that will allow inflation to be monitored more precisely. Previously, rounding sometimes led to misleading readings, as it could mean the difference between a change of, say, 0.2% and 0.3%."

Reporting inaccurate inflation data to 3 decimals does run the slight risk of creating a false impresion of accuracy where none in fact exists; however, the benefits of this change could avoid some of the goofier rounding errors we have seen over the years.

Mathematically, it is possible to be precise but not accurate (and vice versa).

Lets call this a modest improvement in BLS reporting and move on . . .



BLS, Wednesday, February 21, 2007

CPI Now Gets Three Decimals
WSJ, February 21, 2007; Page A2

Print Friendly, PDF & Email

What's been said:

Discussions found on the web:
  1. wally commented on Feb 21

    “hubris” is the word, I believe. Being more precisely wrong is another way to state it. Being a .50000 wit is yet another.

  2. david foster commented on Feb 21

    “Measure with a caliper, mark with a crayon, cut with an axe”

  3. Mike M commented on Feb 21

    Jeez, it’s always either the food or the energy or the medical or the housing or the tuition……Come on Fed, can’t you control these greedy businesses from raising prices?!

  4. Mr. Beach commented on Feb 21

    Anyone notice their restaurant checks going up in size?

    Used to be that a random tuesday night out for two at a local small italian place (entrees, soda, no appetizers,alcohol,or dessert) was about $35.

    In the past 12 months, I’ve noticed that same meal rise to where I paid $46 dollars last night.

  5. Joe S commented on Feb 21

    Thinking in terms of markets… since people on the street are starting to talk about how prices are inflating no matter what their government says… isn’t there a market for a “private” estimate of the CPI? I’m not in the biz so I wouldn’t even know where to begin. But I gotta wonder – how much longer can the BLS says inflation is so low when people are talking about how difficult it is (esp. in food/medical) before the average person stops believing these numbers?

  6. rtalcott commented on Feb 21

    What’s the 95% confidence limits on these numbers? +/- what?


  7. BCS commented on Feb 21

    My favorite weeding out interview question: “Can you tell me the difference between precision and accuracy?”

  8. Robert Coté commented on Feb 21

    They don’t even know the definition of precision. 0.2% v. 0.3% is one decimal place reporting with a huge margin of error. Surely they understand three decimal places is worse than random in the last digit? THere’s this thingy called CPI. CPI is estimated by extrapolating a granular non-representative aggregation of of non-uniformly distributed snapshot datums in a sparse matrix. The matrix is adjusted for seasonality and portents in chicken entrails. The number is then disaggrated into core and non-core and reported subject to massive revision. Three digits is not accurate but in this case neither is it precise.

    The great Issac Asimov predicted this in 1957 with his short story “A Feeling of Power.”

  9. JKH commented on Feb 21

    Better to be vaguely right than precisely wrong – Keynes

  10. Eddie commented on Feb 21

    Once again I ask:

    What can the fed do about oil prices?

    Nothing. Therefore, why should it be considered for determining inflation policy?

  11. howard commented on Feb 21

    as warren buffett likes to say, i’d rather be approximately right than exactly wrong.

  12. howard commented on Feb 21

    JKH, now that’s interesting! so buffett stole it from keynes?

  13. Mike M commented on Feb 21

    Eddie, The fed cannot control the supply or demand of oil. They can, however, control (or at least substantially influence) the supply of dollars. We pay for oil in dollars, just like everything else we buy. Monetary inflation is inflation, period! It will result in higher prices somewhere, which might be housing, oil, financial services executive’s pay, services, etc… Stagflation, anyone?

  14. Michael C. commented on Feb 21

    The Fed pumps liquidity into the money supply via open market operations creating continuous inflation, propping the asset markets.

    Then hawks on about inflation that they’ve created. Maybe at some point, they will raise rates when the economy can handle it. Giving themselves room for rate cuts when the economy needs it.

    What genius.

  15. JKH commented on Feb 21

    “so buffett stole it from keynes?”

    A lot of people have probably stolen from Keynes.

    Buffet’s no exception.

    Though he seems to be a pretty good student – nice 40 year investment results.

  16. super-anon commented on Feb 21

    I discovered the Cleveland Fed CPI today, which I think I like better. I agree with the claim that inflation is understaded, but I don’t think inflation pressures will get worse with a giant debt-driven asset bubble going bust, unless something else takes it’s place.

    What surprises me about the Cleveland Fed data is how much inflation lags the business cycle – it looks like about a one year lag.

    I think it also helps explain why the bond market didn’t over-react to today’s CPI numbers, but the gold market did.

    I think the bond people are just a little bit brighter.


  17. Lauriston commented on Feb 21

    i think i know what the “BS” in BLS stands for, but not sure what the “L” stands for…

  18. MarkTX commented on Feb 21


    L stands for LIARS

  19. Frankie commented on Feb 21

    Since certain items included in the cost of things are volatile, they shall be excluded entirely. So spoke the Oracle of the Go-Vermin.

    While typing the above, I’m looking at a graph; run-of-the-mill price vs time market quote graph. A couple of lines hover below the candles.

    They’re called moving averages. These things help me getting an idea of the general trend over a time frame “x”. Very useful when the variable called “price” exhibits volatility.

    Is there any Law of Economics that prevents the Oracle to dig this rather simple concept? Replace quote by item price, then compute…CQFD as Pascal was fond to say.

    Shouldn’t be THAT difficult to do, no?

    One more thing: Can someone explain to me why no one in Congress or the Media ever take the Oracle’s servants to the wood and spank them silly in front of the village? And I mean direct, hard-hitting simple, no BS questions about inflation ex-inflation: Crack the frigging whip until they’re forced to admit the mistakes of their ways.


  20. squik commented on Feb 21

    My favorite weeding out interview question: “Can you tell me the difference between precision and accuracy?”

    That’s kind of bad since my online dictionary says:

    precision |priˈsi zh ən| noun – the quality, condition, or fact of being exact and accurate

    accurate |ˈakyərit| adjective 1 (of information, measurements, statistics, etc.) correct in all details; exact : accurate information about the illness is essential.

    In other words, you might want to lighten up a bit in your interviews since these words are synonymous in popular usage.

  21. DavidB commented on Feb 21


    When the horse gets out of the pen she sure can fly like the wind can’t she?

  22. Fred commented on Feb 21

    Precision is a calculator that gives an answer 1.99999999 out to as many as thirteen places while accuracy is a slide rule that gives the answer 2 to three places.

    I still have my slide rule. Slide rules have wonderful transparency. That is, you can see clearly how you get to a particular number and thereby know if it makes sense. I think slide rules will be the next big thing.

  23. RW commented on Feb 21

    Popular usage of the words precision and accuracy is irrelevant in an interview of technical or scientific personnel: To not understand the technical distinction between precision and accuracy is to not understand what a significant digit is and that ignorance leads to terrible errors, from the collapse of bridges to the collapse of currencies. Only those in quasi-disciplines where theory is tenuously related to reality and error detached from causality can afford to “lighten up a bit” when assessing acolytes.

  24. BCS commented on Feb 22


    Google “precision versus accuracy”. Wikipedia has a pretty good entry on it. It’s a great question because it can immediately identify whether or not a candidate has the stats experience they may be claiming to have.

    This post just reminded me of the question. I am sure others who read this blog understand why it is a favorite interview question. I mean no offense, but your response was illustrative of why it is effective.

  25. Bluzer commented on Feb 22

    I concur. The BLS figure will be neither accurate nor precise. It will simply have greater resolution. The BLS will continue putting out the same crap – only from now on it will be in High Definition!

    Reminds me of the old joke of the prison warden who has good and bad news for the inmates. The bad news is that they’ll be fed sh*t from now on. The good news? There’ll be lots of it.

  26. cm commented on Feb 22

    Bluzer: The “same crap in HD” paradigm applies to a many other things that are sold to us as “progress” and “improvements”.

  27. Patu commented on Feb 22

    Yes but the market keeps on making new highs in spite of the housing crash and higher inflation ex infation. this so called disconnect is well into its second year. Maybe its not a disconnect at all. Maybe the bigger picture is bigger than an american deflating housing bubble and some arbitrary way of measuring inflation.

  28. teraflop commented on Feb 22

    Precision vs. Accuracy in military terms:
    Precision is a tight grouping in target practice. Accuracy is being able to hit the target in the right location. In the Cold War, the Soviets were feared for their purported ability to blanket Europe with their inaccurate & imprecise warheads. They didn’t need a 3rd decimal place to threaten the West.

    With a variety of multi-factor models and computing tools, we can be completely precise but inaccurate. Think: *H*aving *S*orrow *B*reaking *C*ovenants and “But there wasn’t any history for adjustable-rate mortgages after something like seventeen interest rate rises…”

Posted Under