A Few Bad Apples Spoil June’s Price Statistics
Until recently, the debate between the “inflation is too high” crowd and the “subdued” inflation adherents had centered on the use of headline and core measures of inflation. Core measures exclude food and energy prices, and energy prices had been rising sharply through the first four months of the year, pushing up the headline growth rate relative to the core. In June, however, energy prices reversed course, food prices posted modest gains, and the core CPI jumped up markedly, perhaps causing angst to some debaters. Fortunately at inflection points like these, we have a few alternative price change indicators that may shed some light on the underlying inflation trend.
The headline CPI fell at an annualized rate of 2.6 percent in June, due largely to a sizeable decline in gasoline prices, though declines in household energy prices helped as well. Food prices rose 2.4 percent in June, the smallest monthly increase in the series so far this year. But the unexpected (and perhaps somewhat worrisome) aspect of the recently released figures was that the core CPI (the CPI excluding food and energy prices) jumped up 3.1 percent, and has now risen at an annualized rate of 2.9 percent over the past three months. This is an entirely different signal (and more than 1.0 percentage point higher) than that of the median CPI (which increased just 1.7 percent in June, a slight deceleration from its 3- and 6-month growth rates). This raises the question: What gives?
|Percent change, last|
|1 mo.a||3 mo.a||6 mo.a||12 mo.||5 yr.a||2010 average|
|Consumer Price Index|
|Less food and energy||3.1||2.9||2.5||1.6||1.8||0.6|
|16% trimmed meanb||1.2||2.4||2.8||2.0||2.1||0.8|
b. Calculated by the Federal Reserve Bank of Cleveland.
c. Author’s calculations.
Source: Bureau of Labor Statistics.
Well, it appears that the core CPI was affected by a few usually large price increases in June. These “bad apples” were lodging away from home, auto prices, and apparel prices. The index for lodging away from home followed up a 40 percent spike up in May (its largest price increase since October 2005) by increasing 42.6 percent in June. Car and truck rental, a particularly noisy series, rose 51 percent in June, more than rebounding from a 42 percent decrease in May. New vehicle prices, which jumped up 14 percent in May, rose 7.5 percent in June and have risen 8.3 percent over the past six months. That compares to a growth rate of -0.5 percent over the prior six months. Also, used car prices jumped up 22 percent during the month, the largest monthly increase in the series since December 2009. Finally, apparel prices jumped up 18.3 percent in June (their largest monthly increase since mid-1990), in part because the seasonally adjusted index for men's apparel posted its largest one-month jump up in the history of the series (which dates back to 1947), rising 35.4 percent.
A few relative price changes of such a large magnitude most likely indicate idiosyncratic shocks, mismeasurement, or issues with the seasonal factors. Importantly, these relatively large price changes tend to impart noise into the underlying inflation measure and are not useful indicators of future inflation. Indeed, one might suspect that the recent increases in new auto prices are due to temporary supply chain disruptions. Used auto prices could have been buoyed by a dearth in supply stemming from a prolonged period of dampened production during the recession and the government's CARS program. The increase in apparel prices may reflect pass-through from earlier cotton price and other commodity price increases. If these rationales happen to be the root causes of these relative price increases, we could simply exclude these categories in June in an attempt to uncover underlying inflation. However, we don’t know this for certain, and excluding the components on an ad hoc basis could easily yield a poor signal of future inflation.
Fortunately, trimmed-mean measures—such as the median CPI and the 16 percent trimmed-mean CPI—remove sources of noise in a way that does not rely on judgment and story-telling on a monthly basis. These measures trim the largest absolute relative price changes from the price statistic, lessening the amount of noise in the index. The only judgment involved, apart from how much to trim, is the decision to assume that large monthly price swings in either direction do not reflect the underlying inflation trend. (These measures say nothing about which component will impart the noise, unlike the core, which always excludes food and energy categories).
Perhaps adding credibility to the price signal stemming from the median and trimmed-mean measures, the sticky price CPI—which is a composite measure of prices in the consumers’ market basket that change infrequently—rose just 1.0 percent during the month, marking a slight deceleration from its three-month growth rate (1.5 percent). Meanwhile, the three-month growth rate in the core CPI has continued to climb in recent months.
Interestingly, the upward impulse in the core CPI over the past few months appears to be flexible in nature and, according to Bryan and Meyer (2010), that suggests it has very little useful information on future inflation. The core flexible CPI—composed of items in the core CPI that change price frequently—has jumped up 11.6 percent over the past three months (the swiftest growth rate in the series since the early 1980s).
Incidentally, June’s “bad apples” (lodging away from home, autos, and apparel) all happen to be flexible-price goods, which as a set, do not appear to be a useful predictor of future inflation. Also, these “bad apples” almost exclusively comprised the upper tail of the price-change distribution, and, as outliers, were trimmed out of the median CPI and the 16 percent trimmed-mean CPI for the most part. Together, these observations suggest that the snapback in core CPI over the past three months has likely been driven in part by noisy relative price movements, which are biasing up its signal on the underlying inflation trend.