What Can We Glean from October’s Report on Retail Prices?
The CPI rose at an annualized rate of 1.8 percent in October, as gasoline prices posted a modest decrease and general price pressure elsewhere in the retail market basket was fairly tame (though rents did post sizeable increases). On a year-over-year basis, the CPI is up 2.2 percent.
|Percent change, last|
|1 mo.a||3 mo.a||6 mo.a||12 mo.||5 yr.a||2011 average|
|Consumer Price Index|
|Excluding food and energy (core CPI)||2.2||1.5||1.8||2.1||1.7||2.2|
|16% trimmed meanb||1.7||2.1||1.8||1.9||1.9||2.6|
|Sticky CPI excluding shelter||1.9||1.5||1.9||2.2||2.2||2.3|
b Calculated by the Federal Reserve Bank of Cleveland.
c Author’s calculations.
Source: Bureau of Labor Statistics.
The “core” CPI, which excludes food and energy prices, rose 2.2 percent during the month, outpacing its near-term (three-month) growth rate of 1.5 percent, though it came in relatively close to its year-over-year growth rate of 2.0 percent. Measures of underlying inflation produced by the Federal Reserve Bank of Cleveland, the median CPI and 16 percent trimmed-mean CPI, rose 2.3 percent and 1.7 percent, respectively. Over the past year, the median is up 2.2 percent, while the trimmed-mean is up 1.9 percent. However, there does appear to be an upward nudge on October's data, stemming from rising shelter costs, which may be more indicative of a relative price change in housing prices than an indication of inflation.
Shelter prices jumped up 3.2 percent in October, their sharpest monthly increase since March 2008. A significant chunk of this was rent of primary residence, which spiked up 5.1 percent in October, well above its 12-month trend of 2.8 percent. Also, owners' equivalent rent (OER) rose 2.6 percent in October and has risen 2.8 percent over the past three months, accelerating over its 12-month growth rate of 2.1 percent. Shelter costs comprise a little over 30 percent of the market basket (with OER accounting for roughly 25 percent alone) and have the propensity to influence the measured underlying inflation trend.
As evidence of OER’s, perhaps undue, influence on our read of inflation in October, excluding it from the median CPI calculation pulls the increase in the median CPI down from 2.3 percent to a mere 0.4 percent. This large a difference between the median CPI with and without OER is a marked shift from recent months. Over the prior three months, the difference is only 0.2 percent. Moreover, the median CPI with or without OER is up 2.2 percent over the past year, suggesting that relative price changes in OER haven’t clouded our perception of underlying inflation yet.
In fact, over the past 12 months, nearly every inflation indicator we track is trending within a few tenths of a percent of each other. This is somewhat unusual compared to the last 15 years or so. Large differences between the growth rates of the CPI and the underlying inflation measures, which are symptomatic of relative price swings, often lead to arguments about the underlying inflation trend.
One element related to the differences in growth rates between the CPI and the underlying inflation measures is the cross-sectional volatility in the overall consumer market basket, which reflects the change in the dispersion of prices from month to month. This volatility, as measured by the weighted cross-sectional variance of price changes across the goods and services in the retail market basket, has increased markedly since the late 1990s, making it harder to gauge underlying price pressure. Smoothing the changes in this variance over rolling 5-year periods helps distinguish whether there have been any marked changes in it. As hinted at by the sharp spike up in the cross-sectional variance in mid-2008, volatility has largely been tied to energy price swings. Excluding food and energy prices from the market basket eliminates much of this volatility. Interestingly, core-market-basket volatility hasn’t increased appreciably since the onset of the Great Recession. If anything, the core price-change distribution is a little more uniform than it was in the early 2000s.
This pattern is also evident when examining the volatility of month-to-month (time-series) variance of the CPI and the underlying inflation measures. Sharp price swings in energy and food prices since the mid-2000s have markedly pushed up the month-to-month variance in the CPI relative to the underlying inflation measures. This suggests that attempting to gauge inflation pressure by solely paying attention to the CPI is a futile exercise, as the series is likely to increase sharply in one month only to be followed by an equally sizeable decrease in the next.
Comparing time-series variances of the CPI and the underlying inflation measures is informative, but comparing the two trimmed-mean CPI measures to the core CPI during the Great Recession is perhaps more so. This period marks the first time since 1990 that the variances in the median CPI and the 16 percent trimmed-mean CPI have risen above that of the core CPI. One interpretation of this state of affairs is that month-to-month volatility has increased since the recession in such a way that the core CPI cannot capture it. The conclusion from this line of thought would be that underlying inflation has become harder to gauge. However, the month-to-month volatility in the median CPI, which did increase sharply following the depth of the last recession, has ebbed back in line with the core CPI, while the variance in the 16 percent trimmed-mean CPI has stayed elevated. This difference may indicate that the 16 percent trimmed-mean CPI isn’t an aggressive enough trim and is allowing too much relative price noise to seep in. This conclusion dovetails with the recent work of Meyer and Venkatu (2012), which shows that the aggressive (more than 20 percent) and symmetric trimmed-mean measures tend to perform better in forecasting future inflation.