Wednesday, August 20, 2008

It's Almost Like a 2% Inflation Target Since 1993

The chart above shows annual inflation since 1973, calculated from the BEA's monthly personal consumption expenditures price index, less food and energy (data here), showing 15 years of extremely stable inflation at around 2%. In fact, it's almost as if the Fed has been following a 2% inflation target since 1993 (based on this price index), and provides more evidence that core inflation is low and stable, and nothing like the inflationary 1970s.

Update: A few people have pointed out that CPI inflation (both core and overall) is calculated differently today by the BLS than in the 1970s, distorting a comparison of today's inflation levels to those of that period. But core inflation calculated from the BEA's "personal consumption price index, less food and energy" (see chart above and the post below) tells exactly the same story as the BLS's core inflation: low and stable for the last ten years, and nothing close to the 1970s.

Q: Did both the BLS and BEA change price index calculations, distorting a comparison to the 1970s inflation by either measure?


At 8/20/2008 11:38 PM, Anonymous Anonymous said...

This is fine except that items not counted in this metric account for a significant, if not the majority, share of households' budget. For every extra dollar spent on energy and food, there is one fewer dollar spent on other items.

At 8/21/2008 8:33 AM, Blogger thomasblair said...


The BLS revised its methodology for calculating CPI when Clinton took office. They changed from a fixed basket of goods to a variable basket using first arithmetic weighting and then geometric weighting. Additionally, food and fuel, which account for 23% of the average household's budget, are excluded under the auspice of tamping out volatility in the index. The geometric weighting has the effect of damping effects caused by rising prices and exaggerating effects caused by falling prices. There are other tricks employed by the BLS in the CPI calculations. Because people tend to substitute similar items when the cost of something gets relatively high, the CPI uses a variable basket to adjust for this. Instead of comparing say, steak to steak, the CPI adjusts the later CPI to reflect the price of ground beef, the item people have substituted for steak. The argument is that the CPI should take into account the items that people are actually consuming at that time, rather than comparing a fixed basket of goods over time to track prices. The problem is that when prices rise, demand falls and substitutions happen. To change the definition of CPI to erase the effects of inflation is not only disingenuous, but outright fraud.

At 8/21/2008 9:04 AM, Anonymous Anonymous said...

For higher income households, substitution is legitimate -- if a/c is too $$, install more insulation; if gas prices too $$, fly or reduce vacations. At lower income levels, there may be fewer options: one's already taking public transportation, using lower cost food, living in a rented dwelling without option to improve. Producer price stats are probably a better LT measure.

At 8/21/2008 10:27 AM, Anonymous Anonymous said...

Not only has inflation stayed at about 2% over the last 15 years implied inflation over the next 15 years is also about 2% (Treasury bond yield minus the same maturity TIP). Like Shakespeare said, "...full of sound and fury, signifying nothing".

At 8/21/2008 10:52 AM, Blogger thomasblair said...


You don't get it. The way CPI is calculated has changed, which makes comparisons between eras or even years irrelevant. It'd be like comparing batting averages over time when the statisticians decide after the year is over whether or not to include walks, hit-by-pitches, sacrifices, or defensive interferences as at bats.

For example, in 1887, MLB counted a walk as a hit, which resulted in batting averages in excess of .500. Would it be accurate to compare a hitter from 1887 with, say, Ted Williams .400 year and conclude that the 1887 batter was better?

At 8/21/2008 12:10 PM, Blogger David Foster said...

What is your justification for excluding food and energy from the calculation? If you were running a company which had highly-volatile sales in one of your divisions, do you think the SEC and FASB would allow you to simply exclude the results of this division from your P&L?

At 8/21/2008 12:18 PM, Blogger spencer said...

Virtually everything Thomas Blair said is factually incorrect.

The change in the calculation of the CPI was done after the Boskin commission during the Clinton administration, not when he took office.

The biggest change int he CPI was the use of home owners rental equivalence to replace actual home prices and mortgage rates in the CPI. Secondary changes were made that improved the way the CPI captured quality changes.

Food and fuel are not excluded from the revised index.

If you want to see the difference between the changed CPI and the old CPI go to the BLS and look at the CPI-RS. The changes to the CPI from the Boskin commission were not
applied retroactively to the old CPI. But the BLS did this to construct the CPI-RS so you could see what the CPI would have been if the Boskin Commission changes were applied retroactively.

The difference in the 1970s was about one percentage point so our good host is right that the changes did not alter his conclusion. Moreover, the changes were not made to the GDP deflators so the argument about the changes to the CPI do not apply to the PCE deflator that was never changed in the way the CPI was changed.

At 8/21/2008 1:20 PM, Blogger thomasblair said...


I was incorrect about the timeline, so cut out those four words: "when Clinton took office."

Speaking of accuracy, you're not doing much better. The change in CPI calculation substituting Owner Equivalent Rent for home prices was the result of a 1983 BLS decision.

Some have even traced a link between the 1983 BLS decision and the current subprime mortgage crises:

As Robert Hardaway, a University of Denver professor, pointed out last fall, the subprime lending crisis "can be directly traced back to the (1983) BLS decision to exclude the price of housing from the CPI. … With the illusion of low inflation inducing lenders to offer 6 percent loans, not only has speculation run rampant on the expectations of ever-rising home prices, but home buyers by the millions have been tricked into buying homes even though they only qualified for the teaser rates."


Food and energy are indeed excluded from "core" inflation, which is the most commonly cited version and is the brainchild of Nixon's second Fed Chairman, Arthur Burns. They wanted to be able to overlook troublesome categories. The difference then, as now: food and energy.

At 8/21/2008 2:47 PM, Blogger the buggy professor said...

Thank you, Mark, for all these informative charts the last few days about different measures of the money supply and different measures of the CPI . . . both helping to illuminate whether overall price inflation is under way or not.


1) At Econbrowser --- where the analyses by James Hamilton and Menzie Chin are always stimulating, though they require a decent grounding in economics --- there has been an ongoing discussion for weeks about the various measures of the CPI, the GDP price deflator, and inflationary tendencies.

Those interested in Mark's post today, plus the exchanges here, might find Menzie Chin's article on various CPI (consumer) inflationary measures. Menzie Chin


2) There are four she singles out:

CPI-U (the one consumers worry about most, the U referring to urban consumers);

CPI-R, referred to in informative fashion by Spencer: it traces CPI-prices between 1979 and 1998 in current CIP measures, changed especially in the mid- and late-1990s. That should help Thomas Blair make more meaningful comparisons (without, of course, settling the dispute about the new post-Boskin indexes taking into account substitution and quality changes).

The PCE-deflator (personal consumption expenditures deflator), which measures the total costs of consumer goods and services incurred by producers, though these are not necessarily the prices that household consumers actually pay.

The CPI- chained, which (like all chained measures) continually changes the weights of the prices it tracks. The CPI itself (the first measure in this list) uses fixed weights that are adjusted every two years.


3) What about CORE CPI, which omits highly volatile goods (however significant as Blair notes to household budgets) from the CPI-U measure?

Note that in this particular article. Menzie does not deal with CORE-CPI until the very end, and briefly. That's because she has dealt with it extensively before. And her summary of it parallels Mark's in recent weeks:

"What about core measures? Now, before everybody gets into a frenzy about how core measures are a conspiracy to confuse people, I think careful analysts should be interested in core measures to the extent that energy and food have historically been more variable than other components. Whether monetary authorities should target core versus complete, I leave to others to debate. But I don't think more information can be worse than less, when just looking at the data. Figure 4 depicts core CPI and core PCE over the last few years (for a longer tracking, see this post).


4) Given all that, note quickly now that Menzie arrives at a different view of price trends than Mark, even though she avoids using the term "inflation" . . . rather, all four measures, she notes from her chart, show a clear upward movement the last few months. (Remember, she is referring to the four measures listed earlier here, not to Core CPI).


5) One final point, which comes closer to my own view: there is obviously a gap between how government authorities at the BLS and BEA and lots (probably most) economists understand price rises, on the one hand; and on the other, how the average household consumer and worker understand and maybe worry about price rises --- not least in key areas like energy and food and housing.

In turn, economists will differ among themselves depending on whether they accept a monetarist view of overall inflation or not.

And finally, in an election year --- with the election itself only 2.5 months off --- it would be wrong for economists advising any of the two candidates for the presidency to fall back on technical BEA- and BLS (Bureau of Economic Analysis and Bureau of Labor Statistics) as a way of calming voter concerns.


6) How serious those concerns are will hinge partly, it goes without saying, about what happens to the price of oil in the next 2 months or so; but not entirely and for obvious reasons.


Michael Gordon, AKA, the buggy professor


PS Menzie refers to both a Paasche and Laspeyres mathematical model used for tracking price changes. They use different bases for tracking those price changes. Neither base assumption is wrong, only different. In the upshot, though, the Paasche index systematically UNDER-states price rises and the Laspeyres OVER-states it.

Those who take a conspiratorial view of the various measures used by the BEA and BLS should consider that not only can you compare price changes between 1979 and 1998 by going to the BLS and tracking CPI-R, you have to consider the technical mathematical problem of which base for comparison to use.

There is no easy answer to that base issue.


At 9/01/2008 9:57 AM, Anonymous Anonymous said...

Menzie its a he, not a she. lol


Post a Comment

<< Home