19 Aug 2009

Is the BLS Suppressing Producer Price Inflation?

All Posts No Comments

As long-time readers know, I am very skeptical of the big deflation scare. The raw CPI numbers have risen at an annualized rate of 4.2% since December. The “seasonally adjusted” CPI numbers–what the press reports–are up only 2.4% (at an annualized rate) over the same period. Now it’s possible that it will all even out by December, but I have my doubts. (More precisely, I think they will revise the earlier “adjustments” so that the official headline inflation numbers won’t need to be as big in the last half of 2009 in order to reconcile the two series.)

Things are also fishy if we turn to the Producer Price Index (PPI). According to the latest press release, both the raw and the seasonally adjusted PPI fell 0.9 percent from June to July. I’m wondering when the positive seasonal adjustment is going to start kicking in, because here are the NSA and SA numbers for the year:

Monthly Changes in PPI Index
Month NSA SA Difference
January +0.9 +0.8 -0.1
February -0.1 +0.1 +0.2
March -0.7 -1.2 -0.5
April +0.6 +0.3 -0.3
May +0.5 +0.2 -0.3
June +1.9 +1.8 -0.1
July -0.9 -0.9 0.0
Cumulative -1.1

Unfortunately FRED doesn’t have both the seasonally and non-seasonally adjusted PPI series, so to generate the above table I had to go through the BLS archives of the PPI press releases each month.

Just to make sure you understand the significance of the above table: So far this year, the BLS has underreported the actual changes in the PPI by a cumulative 1.1 percent. (It’s probably not correct to just add those differences up, but you get the idea.) The whole rationale of seasonal adjustment is that price changes in one direction tend to be concentrated in certain months. So if, say, prices typically rise very quickly in the beginning of every calendar year, and then remain flat during the third and fourth quarters, the BLS doesn’t want to shock people by reporting the big jumps in the first quarter. So instead they smooth out the yearly increase, by adjusting the early months down and bumping up the later months. Since there had already been a cumulative 1.1 percent deficit going into this month, that’s why I was surprised to see no bumping up of the July numbers.

While looking through the BLS archives of previous PPI releases, I confirmed another suspicion I had had, regarding this month’s press release. I thought the 0.9 percent drop from June to July seemed excessive, so I wanted to look at the breakdown of the overall index into its categories. For example, I could look at what the BLS said happened to fuel costs, and I could compare that with the EIA’s records of actual gasoline prices.

Now in previous months, this would have been easy. Near the bottom of each press release, you can find Table 1 which has all kinds of information. (Check out July’s press release, for example.)

But even though the press releases from January through July 2009 all have a Table 1 with this non-seasonally adjusted breakdown, for some reason this month’s press release doesn’t have it.

That information is presumably still available somewhere in the bowels of the BLS website, but if it’s difficult to find, it discourages people from looking. Is that the point?

Last issue: I was also puzzled as to why the CPI numbers came out first this month, followed by the PPI. In previous months I thought the PPI always came out a day ahead of the CPI. I called the BLS and the guy there told me that yes, the PPI generally comes out first, but they are independent groups working on these numbers, and for whatever reason the PPI people were delayed this month.

Comments are closed.