Others and myself on this site have maintained that the CPI numbers don't reflect reality for the typical american. Thus their shorthand for summarizing an economic condition is useless. Heres the latest from the experts. We should go back to the previous 1985 CPI numbers and method of calculation, and adjust all data since 1985 to the old calculation methods.
Here's the story----
Quirk in Calculation
May Distort Inflation Gauge
By GREG IP
February 21, 2006 4:27 p.m.
If inflation appears to accelerate in coming months, investors might want to play down its significance. Some economists say that quirks in the calculation of the consumer-price index may be causing inflation to be overstated during the winter and understated in the rest of the year.
Analysis by the Bureau of Labor Statistics, which calculates the CPI, seems to confirm the presence of the quirk, but the BLS is unsure how the known gaps in its seasonal-adjustment process could give rise to it.
The quirk appears only in the total index, not the core index, which excludes food and energy. While that means the quirk is less likely to mislead the markets or the Federal Reserve, which focus on the core, it could still leave a misimpression of high inflation in the early months that affects both consumer expectations and pricing in the inflation-indexed Treasury bond market. If it "goes on long enough, there's the possibility of unhinging inflation expectations," says Chris Varvares of Macroeconomic Advisers, in St. Louis.
Goldman Sachs economist Edward McKelvey finds a strong tendency for the CPI to overshoot its trend in January, March and September, and to undershoot in June, July, August and December. Goldman expects a 0.6% increase in the total CPI for January, compared with 0.5% for the consensus. It expects a 0.2% increase in the core CPI, in line with the consensus. The January index is released Wednesday.
If you break the year into two periods -- three months ending in March, and nine months ending in December -- the annualized inflation rate has been higher in the first period than the second period in each of the last six years. On average, inflation was 4.3% in the first three months of the year compared to 2.1% in the last nine months.
No one is sure why this occurs, but it probably crops up as a result of the BLS's "seasonal adjustment" process. This process tries to determine, over a period of years, what movements in prices recur seasonally and aren't evidence of any underlying economic trend. For example, gasoline prices rise regularly in May at the start of the summer driving season, then fall in October as the season ends. The BLS thus makes a downward adjustment in the May index and an upward adjustment in October to account for this. This means any changes you see in those months should be the result not of such seasonal factors but underlying supply and demand. By design, seasonal adjustment factors cancel each other out over 12 months.
However, the BLS doesn't seasonally adjust the entire index. Rather, it seasonally adjusts the components and adds them up for the total. But it is unable to adjust all the components because some don't have a strong enough seasonal pattern. These unadjusted components make up about 20% of the weight of the index, and that can give rise to "residual seasonality," the BLS says, although it says it isn't clear what effect this would have.
Another potential source of the quirk is the BLS's use of "interventions" in its seasonal adjustment process. It uses a statistical program to look for seasonal patterns, but then manually intervenes to see if any fluctuations are due to anomalous events. In gasoline, the BLS has regularly intervened to prevent the program from treating one-time events as recurrent, such as the price spike that occurred when Hurricane Katrina shut down refinery capacity.
The logic is as follows: If left alone, the program would read part of the Katrina-related spike in gasoline prices as a recurring September event. The following September, it would then depress the index accordingly in anticipation of another price jump, while compensating with a slightly elevated index in all the other 11 months. This would tend to understate the inflation rate in that month and overstate it in others. The BLS's interventions prevent that.
In recent years, though, the BLS has had to intervene frequently on gasoline prices because so many events, often geopolitical in nature, have had an impact. Conceivably, some of these events in fact are seasonal, or coincidentally are occurring in a way that gives the appearance of being seasonal.
To analyze these issues, BLS economist Patrick Jackman recently seasonally adjusted the entire index, rather than the component indexes, and didn't use interventions. The analysis significantly reduces the tendency of inflation to speed up in the winter. It is markedly higher in just three of the last six years, rather than all six. On average, it is 3.3% in the first three months and 2.5% in the last nine months.
But here's the irony. Using this alternative approach, Mr. Jackman says gasoline prices would have been even more depressed last December and elevated in January, exacerbating the recent pattern of elevated inflation at the start of the year.
"Seasonal adjustment is an art, not a science," he says. He says gasoline historically was so volatile that it didn't even meet the normal criteria for seasonal adjustment. But its large, 4% weight in the CPI meant it had to be adjusted to make the overall, seasonally adjusted index meaningful. However, in the last two years it has met the normal criteria for seasonal adjustment, Mr. Jackman says.



Reply With Quote

Bookmarks