Why Are the CPI and the PCE Giving Different Estimates of Inflation?
In January 2012, the Federal Open Market Committee (FOMC), the Federal Reserve's monetary policymaking body, announced that its long-run objective for inflation was 2 percent as measured by the Personal Consumption Expenditures price index (PCE). At the time, the PCE appeared to be moving toward this target. Moreover, "underlying" measures, like the PCE excluding food and energy and a trimmed-mean version of the PCE, which tend to provide a less noisy signal of the inflation trend, were already close to the 2 percent target. But since then, PCE-based measures have drifted down. Toward the end of 2011, the year-over-year change in the PCE was close to 3 percent, but by March, when the PCE fell 0.1 percent, the 12-month percent change in the index had fallen to just under 1 percent. This has prompted some FOMC participants to suggest recently that the Committee may need to act to address this disinflation.
Another price measure, however, the Consumer Price Index (CPI), has generally hewn more closely to 2 percent since the beginning of 2012. In April, the CPI fell 0.4 percent and has risen 1.1 percent from the previous April. But prior to that, year-over-year changes in the CPI have essentially moved in a half-percentage-point band between 1.5 percent and 2.0 percent. CPI-based "underlying" measures have remained even more closely tethered to the 2 percent target during this period. For instance, year-over-year changes in the median CPI have registered 2.1 or 2.2 percent in every month since the fourth quarter of 2011 with only one exception.
Why are these two price measures providing such different signals of the inflation trend? Analysts at the government agencies that produce the two indexes have identified three major factors: differences related to their respective formulas, the weights attached to their price components, and the breadth or scope of coverage. Interestingly, one of the elements that doesn't differ much between the two measures is price index data; where their coverage overlaps, the two measures tend to use the same underlying information on the prices of products.
With respect to formula differences, the CPI is constructed using a formula which assumes that consumers purchase the same set of goods in the same quantities every period. This "market basket" is updated periodically, typically every two years in the case of the CPI. The PCE, by contrast, uses a formula that allows the quantities of its goods to vary from one period to the next. As a result, the PCE is able to more accurately capture how consumers adjust their purchases in response to price changes than the CPI. One implication of this is that, because consumers are likely to shift their consumption toward products whose prices are rising less rapidly, the PCE often registers less inflation than the CPI.
On the weighting factor, the two indexes take their weights from different sources. The CPI uses a survey that asks households about their expenses to determine the weights in its market basket. The PCE, by contrast, relies largely on information from businesses detailing what they've sold to households to produce its weights.
Finally, regarding the third factor, scope differences, the two indexes were constructed to cover slightly different things. The CPI covers the out-of-pocket spending of urban households. The PCE, however, was designed to be somewhat broader, covering consumer purchases by, as well as for, all U.S. residents. Specifically, some items, such as medical care, may be purchased by firms for their workers, or the government or a nonprofit organization on behalf of other consumers. In these cases, any cost not covered out-of-pocket by the consumer wouldn't be factored into the CPI's calculations but would be included in the PCE. Unpriced or "free" financial services are another example of a price the PCE attempts to measure and incorporate into its calculations, which wouldn't appear in the CPI. Roughly, the CPI covers about 75 percent of the PCE's scope. There are also some items that the CPI covers that are not included in the PCE.
Even before the different formulas are applied, the differences in item weights and scope are evident in the two measures' market baskets. For example, the PCE's broader treatment of medical care and financial services shows up as much larger market-basket shares for these items (where the latter is included in the "other goods and services" category). Notably, medical care has a market-basket share that's almost three times larger in the PCE than in the CPI. Differing answers from firms and households on surveys also generate meaningful differences between the two measures for housing and transportation. Owners' Equivalent Rent of Primary Residence (OER), which attempts to estimate the rent for owner-occupied properties, is almost twice as large in the CPI as it is in the PCE.
In the first quarter of 2013, these three factors, as well as other unspecified factors, produced a gap between the annualized percent changes in the two price measures of about 0.5 percentage points. Over the last four quarters, the gap between the two measures has been similar - about 0.6 percentage points. Among the factors driving these differences, the variation in item weights has been important in holding the CPI above the PCE over this period. Indeed, in the first quarter, about half of the 0.5 percentage-point gap is attributable to the differences in item weights, with much of this being driven specifically by differences in OER.
|Food and beverages||15.3||14.0||1.2|
|Education and communication||6.8||4.9||1.9|
|Other goods and services||3.4||14.8||-11.4|
Source: Bureau of Economic Analysis; Bureau of Labor Statistics.