August Price Statistics
The Consumer Price Index (CPI) fell at a 1.7 percent annualized rate in August, decreasing for the first time since October 2006. A 32.3 percent (annualized) drop in energy prices accounted for most of the decrease. CPI excluding food and energy prices rose a modest 1.8 percent (annualized), coming down from longer-term trends. Also receding from longer-term trends, the 16 percent trimmed-mean inflation measure rose 1.3 percent (annualized), its smallest increase in nine months. The median CPI rose 2.3 percent (annualized) in August, down considerably from its 2006 average of 3.1 percent. All of the measures point to inflation moderating, at least for the moment.
|Percent change, last|
|Consumer Price Index|
|Less food and energy||1.8||2.5||2.0||2.1||2.0||2.6|
|16% trimmed meanb||1.3||1.8||2.2||2.3||2.3||2.7|
|Producer Price Index|
|Less food and energy||2.3||2.5||1.6||2.2||1.6||2.1|
b. Calculated by the Federal Reserve Bank of Cleveland.
Sources: U.S. Department of Labor, Bureau of Labor Statistics; and Federal Reserve Bank of Cleveland.
Long-run inflation trends have also abated recently. The 12-month trend in the core CPI and the 16 percent trimmed-mean CPI have been decreasing since February and now range between 2.1 percent to 2.3 percent. The 12-month trend in the median CPI has fallen from May's recent high of 3.2 percent and now stands at 2.7 percent.
While energy prices have fallen over the past three months, contributing to an easing in the overall CPI, oil prices have started to creep up recently. On September 21, 2007, the spot price of West Texas Intermediate crude oil was $83.36 a barrel, jumping nearly $10 over August's average spot price to a record high in nominal dollars. We are closing in on a real record, however. After adjusting for inflation (in current dollars), $80 a barrel is only about $20 shy of the record spot price of $101.43 set in April 1980. This could feed through to an elevated headline CPI number in the months ahead.
Frequent transitory swings in components such as energy and food lead to volatility in the headline inflation measure. In order to get a clearer picture of the underlying inflation trend, it is useful to trim certain unstable components out of the index. This is why the Federal Reserve Bank of Cleveland produces the median and 16 percent trimmed-mean measures of inflation. However, in the same way that core CPI (which excludes food and energy from the CPI) picks up fleeting movements in other components like medical care, the median CPI was being unduly influenced by Owners' Equivalent Rent (OER). Because of the relative size of OER (it comprises 24 percent of the index, while the next-largest component, food away from home, comprises only 6 percent), and its stability, OER was often the median good. Very recently (September 19, 2007), the way both of these measures are constructed was revised to lessen the influence of OER. The revised measures disaggregate OER into four regional subcomponents (Northeast, Midwest, South, and West). The result is a clearer near-term indicator of inflation trends.
The difference between the revised series (with the regional OER components) and the original series is easily viewed in the context of the component price-change distribution. OER was the median good for the original series in August, and rose 2.8 percent (annualized). The revised series rose 2.3 percent (annualized), and the median component was the regional OER for the West (OER: West).
The differences emerging from the revision can be seen in longer-term trends as well. In January 2002, the 12-month percent change in the original median CPI was 3.8 percent, 0.6 percentage point higher than the revised median CPI's growth of 3.2 percent. A more in-depth look at the changes to the median CPI can be found at Inflation Central.
Another indicator of these estimators' performance is root mean-squared error (RMSE). It is frequently used to measure the difference between an estimator's predicted values and the values actually observed (in this case, the growth in the CPI over the next three years). The estimator with the smallest RMSE is judged to have superior predictive capabilities, and in the case of these CPI-derived measures, that means it is a better measure of future inflation trends. The change in methodology improved the median's RMSE, while the change seems to have improved only slightly the performance of the 16 percent trimmed-mean measure over the one- to three-month range.