« July 2006 | Main | September 2006 »
August 31, 2006
Still Not As Good As It Looks
Once again, I find myself not so convinced by what at first appears to be good news on the inflation front. From Reuters:
In another sign of moderating inflation, the government reported earlier Thursday a smaller-than-expected 0.1 percent July rise in the core personal consumption expenditure index, the Federal Reserve's preferred inflation gauge. The core PCE advanced 0.2 percent in June.
That core personal consumption expenditure index, of course, is the PCE exlcuding food and energy components. But the Dallas Fed provides an alternative look, that is not so wonderful:
The trimmed mean PCE inflation rate for July was an annualized 3.1 percent. According to the BEA, the overall PCE inflation rate for July was 4.1 percent, annualized, while the inflation rate for PCE excluding food and energy was 1.7 percent.
That 3.1 percent represents a slight increase over the 3.0 percent annualized increase in June. I can really do no better than repeat what I said in my post about about the Consumer Price Index inflation report a few weeks ago: Unlike the traditional core inflation measure, which simply excludes food and energy components, the trimmed-mean measures excludes both high and low price changes, no matter what they might be. The idea is to get a picture that is not distorted by items with prices that are increasing by a lot, or decreasing by a lot. And that picture just isn't improving that much:
Almost 60 percent of expenditure-weighted price increases -- 59.16 percent to be exact -- were over 3 percent annualized in July. That is a slight deterioration relative to the year for a whole -- the corresponding figure for the first 7 months of 2006 is about 52 percent -- but, more importantly, the shift in the distribution is clearly toward higher rates of increase.
I'm not surrendering, but it is clearly too early to declare victory on the basis of a good PCE ex food and energy signal, even adjusting for the usual caveat to not get overly exxcited about one month's data.
August 31, 2006 in Inflation | Permalink
TrackBack
TrackBack URL for this entry:
http://www.typepad.com/services/trackback/6a00d8341c834f53ef00d834e2855469e2
Listed below are links to blogs that reference Still Not As Good As It Looks:
Comments
Posted by:
knzn |
August 31, 2006 at 04:44 PM
As the right side non-discretionary spending rises, the left side discretionary spending falls putting a hard ceiling on prices. The larger the right gets, the less there is remaining to spend on the left side.
Nice graphic - a picture really is worth 1k words.
Posted by:
RP |
September 01, 2006 at 06:23 AM
knzn -- The idea is similar to the idea of the ex food and energy measure: To filter out the high frequency noise to get a truer picture of the underlying trend in the inflation series. Rather than arbitrarily excluding a couple of components that may or may not be distorting the picture in a given month, trimmed-mean estimates are agnostic about what to exclude -- anything that experiences a particularly volatile price change in a given month gets knocked out. The Dallas Fed uses a technique that chooses the optimal amount to trim off the high and low tails of the distribution -- where optimal is defined as coming up with measure that tracks the true underlying inflation trend as closely as possible.
The Dallas Fed has a very nice, and very accessible description of the trimmed mean here: http://dallasfed.org/research/swe/2005/swe0503b.html
There are also links to more technical information.
Posted by:
Dave Altig |
September 01, 2006 at 07:21 AM
uncertain contends comparable efforts
Posted by:
mayblack |
April 01, 2009 at 04:49 PM
I’m not sure I understand the rationale for the trimmed mean. In principle, you want to exclude prices that are unlikely to exhibit persistent rates of change. There is a good case for excluding energy because energy is storable, and with nominal interest rates still relatively low, there would be a large incentive to “buy now and sell later” if (recent large positive) rates of energy price increase were expected to be persistent. Note that this argument applies whether or not energy is at the extreme of the distribution. Except for discrete changes when new information arrives, energy prices should be expected to rise slowly no matter what the rest of the distribution looks like, provided that nominal interest rates and storage costs remain relatively low (and provided that substantial inventories exist; otherwise you could have expected declining prices). The case for food is a bit weaker, since storability is limited, but it has been observed that food prices are volatile. Similar arguments do not automatically apply to things found at the extremes of the price change distribution. For example, prices of certain computer components have been consistently near the bottom of the distribution because they benefit disproportionately from technological progress. That will most likely continue to be the case. It is not reasonable to exclude such things just because they are at the extremes of the distribution.