This update, prepared by Dallas Fed Senior Economist Jim Dolmas, provides an in-depth analysis of the latest personal consumption expenditures (PCE) inflation data. Updates will be posted monthly, following the release of the official PCE data by the Bureau of Economic Analysis.

Energy prices drove the May headline PCE rate into negative territory and seem poised to do so again in June. Meanwhile, both the core PCE and trimmed mean PCE rates for May were slightly higher than their average rates over the past several months.

The core rate reflects, in part, large increases in the prices of some financial services and in the cost of hotel and motel stays. Owner’s equivalent rent (OER) was a significant force pulling in the downward direction.

OER—the largest item in the core and, when included, in the trimmed mean—does appear to be leveling off, at least for the time being. That development, together with relative constancy of the trimmed mean rate at six- and 12-month horizons and slight upticks in the core rate at those horizons, suggests some leveling off in the underlying inflation trend, though at a very low level.

Gasoline Pulls Down the Headline Rate
The headline PCE price index fell 0.4 percent at an annualized rate in May, pulled down by a sharp decline in the price of gasoline. The 12-month headline rate ticked down to 1.9 percent from 2.0 percent in April. The six-month PCE inflation rate fell more noticeably, from an annualized 1.4 percent in April to just 0.9 percent in May.

As noted above, gasoline was the prime culprit behind May’s headline decline. The index for gasoline and other motor fuel fell 5.3 percent on a month-to-month basis in May, or almost 48 percent at an annualized rate. Fuel oil, natural gas and electricity services also registered declines, though of more modest size.

As of right now, Department of Energy (DOE) weekly data on retail gasoline prices point to a June decline of comparable magnitude to May’s. With three weeks’ worth of data in, the average retail price of a gallon of gasoline is tracking down 5.6 percent compared with May (monthly rate). Estimates of the normal seasonal price movement for June have been somewhat volatile lately—the typical seasonal effect has trended down (in absolute size) from about a 2 percent decline a few years ago to just a 0.4 percent decline last June. Using that most recent seasonal effect of –0.4, the weekly DOE data predict a 5.2 percent decline in the seasonally adjusted price that will go into June’s headline PCE number.

Assuming the trend in the weekly gasoline data holds up through the final week of June (and assuming no surprise increases from other PCE components), we would not be surprised to see another negative headline number in June. That would pull the 12-month headline rate down to at most 1.3 percent, and the six-month rate down to an annualized 0.6 percent—those being the numbers resulting from a zero headline rate in June.

Food a Mixed Grocery Bag
Food prices were a nonfactor in the May release, with food as a whole close to unchanged (down just 0.2 percent at an annualized rate). The prices for the various components that go into food are typically all over the map—any given month will see large price increases for some items together with large price declines for others. We can slice these data in different ways, but one way we find useful is to group items into the categories “less processed” and “more processed” (think fresh produce or meats versus snack foods or soft drinks). Items in the latter category typically have prices that are less volatile and perhaps more informative about underlying inflation trends. In May, less-processed items, taken as a group, fell about 3.5 percent at an annualized rate (pulled down mainly by price declines for eggs and fresh produce), while more-processed items increased about 1.1 percent at an annualized rate (led by bakery products, sugar and sweets, and beer). 

May’s pattern—prices for less-processed food items being generally down and prices for more-processed items being up—is different from what we’ve seen over the past several months. For example, although food as a whole is up an annualized 1.9 percent over the past six months, the less-processed group is up an annualized 7.5 percent, while the more-processed group is down an annualized 0.2 percent.

Inside the Core: Declining Goods Prices, Some Pickup in Core Services Prices
Core PCE increased in May at an annualized rate of 2.0 percent, its largest one-month reading since last October. The 12-month core rate ticked up to 1.3 percent from 1.2 percent in April, and the six-month core rate increased to 1.1 percent, annualized, from 0.9 percent.

Underlying May’s 2.0 percent core rate was a 3.1 percent annualized gain in core services prices together with a 1.3 percent annualized decline in core goods prices. On a 12-month basis, core services inflation is now running at 1.9 percent, its highest rate since April 2009, while goods inflation is running at –0.6 percent, its lowest rate since October 2007.

The impact a particular item has on either the core or headline index depends on both the size of the item’s price change and the item’s weight in the index. One way to gauge an item’s impact on this month’s core rate is to calculate what the core rate would’ve been if that item had been excluded. That’s what we do in the following discussion.

Big-Impact Items in May’s Core Number
As is often the case, tobacco was the biggest-impact item among core goods, increasing at an over 16 percent annualized rate in May and contributing roughly 0.15 annualized percentage point to the May core rate. Personal care products were the most significant drag among core goods, subtracting about 0.1 annualized percentage point from the May core rate.

The biggest positive impacts, though, were among core services, in particular the prices for some financial services and the price index for hotels and motels. The price indexes for “financial service charges, fees and commissions” and commercial bank services furnished without payment each increased at a nearly 20 percent annualized rate and, together, contributed over 0.6 annualized percentage point to May’s core rate—that is, a core index excluding those items would’ve been up an annualized 1.4 percent in May, rather than the 2 percent core rate we actually observed. The hotel and motel category was up nearly 43 percent at an annualized rate and contributed about 0.2 annualized percentage point to the May core rate.

We never want to make too much out of one month’s data, and that’s even more the case with these three items. All three are among the most volatile included in core PCE. As a result, they’re excluded from the trimmed mean PCE about 80 to 85 percent of the time.

Perhaps the most significant development within the core, though, is the slight growth in OER in this release, after six months of declines. While the increase was very slight, just 0.3 percent at an annualized rate, OER is a ship that turns quite slowly, and this data point suggests a leveling off for the time being. Likewise, rent of tenant-occupied stationary homes has increased (again, very slightly) in each of the past three months. At the very least, these series look like they’ve bottomed out. They’re still likely to increase more slowly than the rest of core PCE and, hence, are still likely to be a drag on the core rate, but probably less so going forward than we’ve seen over the past several months.

The Trimmed Mean
One can argue about how much attention to pay to particular components, or which ones to disregard, but essentially that’s a somewhat subjective exercise. Indexes like the trimmed mean take a more objective approach, discarding all extreme price movements each month. May’s trimmed mean came in at 1.1 percent, annualized, after four straight sub-1-percent readings. The six- and 12-month trimmed mean rates held steady at 0.7 percent and 0.9 percent, respectively.

Number of Falling-Price Components Still Elevated
Finally, looking at all 178 components that are candidates for inclusion in the trimmed mean, the fraction of those registering price declines was 42 percent (74 components), which is near the average level over the past year and elevated by historical standards. When items are weighted by expenditure share, the percent falling dropped from 37 percent in April to 28 percent in May—largely the consequence of moving OER, and its very large weight, from the minus column to the plus column. Apart from that shift, there was little change in the underlying distribution of price increases.

—Jim Dolmas
    June 28, 2010

Inflation Update Glossary Annualized rates: A change in a price index (or of the price of a particular item) from one month to the next can be expressed at a monthly rate (just the percentage change from one month to the next) or at an annualized rate—the percentage change we would obtain if the one-month change were repeated every month for a year. This is analogous to rates on credit card balances, which can be quoted in daily terms or as an annual percentage rate (APR). Even though you may not carry a balance for an entire year, you probably still use the APR to describe the rate you’re paying. In our Inflation Update, we like to put most numbers in the form of annualized rates, since this facilitates comparison of price changes over different horizons. Components of personal consumption expenditures (PCE): In the Inflation Update, we often mention what fraction of items fell in price in a given month or what fraction had annualized price increases in a given range, say 0 percent to 2 percent. What are the underlying data these statements are based on? The data are for the 178 components used to construct the trimmed mean. From the Dallas Fed’s Trimmed Mean PCE webpage (www.dallasfed.org/data/pce/index.html), follow the link “Components included and excluded from this month’s trimmed mean” to see an Excel spreadsheet listing all 178 components. Core goods and core services: Goods are items taking the form of a physical commodity, like a TV, a steak or a gallon of gasoline. Core goods are goods apart from food and energy items. Services are, well, services—like a haircut (the economist’s favorite example), a stay at a hotel, an airline trip or the electricity services you obtain from your utility provider. Core services are services other than energy services. In PCE, dining out at a restaurant is treated as a core service. This is not the case in the consumer price index (CPI), which groups meals at restaurants together with other food items—and thus excludes them from the core. Department of Energy (DOE) weekly retail gasoline price data: The DOE’s Energy Information Administration collects and publishes weekly data on the average retail price of a gallon of gasoline. By the time PCE data covering a given month are released, we usually have three or more weeks worth of DOE price data for the subsequent month. Given the large weight gasoline carries in the headline PCE index (and the large volatility of gasoline’s price), the weekly retail price data are useful for getting a rough sense of what we can expect for headline PCE inflation when data for the subsequent month are released. The DOE data can be found at www.eia.doe.gov/oil_gas/petroleum/data_publications/wrgp/mogas_home_page.html. Headline, core and trimmed mean PCE inflation: “Headline” PCE refers to the all-items PCE index (that is, the index including all items of personal consumption expenditure, excluding none). “Core” PCE refers to the index excluding food and energy items. This is the common usage, though some economists take core inflation to mean the underlying trend in headline inflation; for them, the “ex food and energy” inflation rate is one particular estimate of core inflation. The “trimmed mean” inflation rate is constructed by excluding the items that registered the biggest increases or decreases in a given month, regardless of whether those items are food, energy or something else. Taking core inflation in the sense of underlying trend inflation, the trimmed mean is an alternative estimate of that trend. The headline and core (ex food and energy) numbers are produced by the Bureau of Economic Analysis (BEA) and available at www.bea.gov/national/index.htm#personal. Using BEA’s data, the Dallas Fed produces a trimmed mean PCE inflation rate (www.dallasfed.org/data/pce/index.html). The Cleveland Fed produces a trimmed mean and median CPI inflation rate (www.clevelandfed.org/research/data/us-inflation/mcpi.cfm), using data from the Bureau of Labor Statistics, the official source for CPI data. More processed, less processed: Most current theoretical models of inflation (and monetary policy’s impact on inflation) suggest that central banks should focus their attention on prices that exhibit “stickiness”—that is, prices that don’t adjust instantaneously to changing supply and demand conditions. The price of a cup of coffee at Starbucks, for example, may be changed only once or twice in the course of a year, even though the price of one of the primary inputs—coffee beans—is perhaps changing on a daily basis. Economists would characterize the price of coffee beans as flexible and the price of the cup of coffee at Starbucks as sticky. In general, among food items, the less processed the item, the more likely its price is flexible (and conversely—the more-processed items probably exhibit more price stickiness). We thus might want to pay a bit more attention to what’s going on with the price of, say, snack foods or breakfast cereals, than with the price movements in a relatively unprocessed category like fresh fruit. Owner’s equivalent rent (OER): In the PCE (as well as the CPI), the cost of the housing services consumed by someone who owns his own home is measured by OER. In principle, OER answers the question: How much rent would I be paying to live here if I didn’t own this home? In practice, OER is not measured directly, but is imputed from data on rents paid by actual renters. OER is the single largest item in both the PCE (with a share of about 12 percent) and CPI (with a share of about 24 percent). Seasonally adjusted: The prices of some items have a regular rhythm to them over the course of a year. Gasoline prices, for example, typically go up in March, April and May, and fall in July, August and September. Seasonal adjustment subtracts an estimate of the normal seasonal variation in an item’s price to isolate movements that are either higher or lower than we would normally expect, given the seasonal cycle. For example, based on the seasonal price pattern of the past few years, a roughly 5 percent increase in the price of gasoline in March is normal. If we then observe a 5 percent increase in price in March, the seasonally adjusted data would register “no change” (a 0 percent rate). If we observe a 7 percent increase, seasonal adjustment would count 5 percentage points of that as normal variation and just register a 2 percent increase. And if we saw only a 3 percent increase—2 percentage points short of what we would normally expect—the seasonally adjusted data would register a 2 percent decline.