Prices rose 3.7% in the year to September. Core inflation, the number economists watch for the trend, says 2.6%. Tuesday’s report (Nov. 10) will update both. So which one tells you where prices are going?
We tested that against every month since 1985. When headline runs ahead of core, as it does now, core usually pointed closer to where inflation ended up a year later.
What is core inflation?
Core inflation is the consumer price index without food and energy. That sounds odd, since groceries and gas are what most people notice first. The reason is that food and energy prices jump around with weather, wars and harvests, then often fall back. Leaving them out shows the slower trend in rent, cars, insurance, medical care and other services.
This year the swing is in energy. Energy prices were up about 20% from a year earlier in September, after the conflict with Iran pushed up oil and gas. Food was up only 2.5%. Take both out, and inflation drops by more than a point.
Core was the closer guess two times out of three
Since 1985 there were 77 months when headline inflation ran at least a point above core, like today. For each one we asked: if you’d guessed next year’s inflation using today’s headline number, how far off would you have been? And using core?
Each dot below is one of those months. It shows how far that guess landed from the actual inflation rate a year later, so dots further left are better guesses.
Core was usually the closer guess a year later.
Core was closer about 64% of the time. In a typical month, core missed by about 1.4 percentage points and the headline number by about 1.8. Energy spikes mostly faded, and inflation drifted back toward core.
Here’s how the biggest splits played out:
Four of five big splits closed toward core within a year; Dec. 2021 didn’t.
| Month | Headline | Core | Headline a year later |
|---|---|---|---|
| Sept. 2005 (Hurricane Katrina) | +4.7% | +2.0% | +2.1% |
| July 2008 (oil peak) | +5.6% | +2.5% | -2.1% |
| Sept. 2011 | +3.9% | +2.0% | +2.0% |
| Dec. 2021 | +7.0% | +5.5% | +6.5% |
| June 2022 | +9.1% | +5.9% | +3.0% |
When does core get it wrong?
Core misses when the spike spreads. In late 2021, energy wasn’t the only thing rising; shipping, rents and wages were too. Core said 5.5% and headline a year later was still 6.5%.
The same thing explains the 1970s. Oil shocks then fed straight into wages and everything else, so headline inflation didn’t fall back toward core. Count those years too, back to 1958, and core was the closer guess barely half the time.
So the test isn’t whether energy is high. It’s whether energy costs spread into other prices. In tomorrow’s report I’d look past gas to things that use a lot of fuel, like airfares and delivery.
Which number I use, and for what
I use both, for different jobs. For my own budget, the number that matters is what I pay. My living costs went up about 12% over the past year, mostly groceries and eating out, far more than either number here.
My bill rose far faster than either official number.
Both indexes average thousands of prices, including many that barely moved, and core leaves out food entirely. That’s why my bill and the official numbers feel so far apart, and why people feel prices more than the inflation rate.
I use core to read where interest rates go. If oil falls but core doesn’t, I don’t take it as a sign rate cuts are coming. The Fed also watches a core measure: its 2% goal uses a different price index, but it strips out food and energy to see the trend. After a strong jobs report, markets are asking whether the Fed will raise rates again in December. A calm core number tomorrow would make that harder to justify.
In 2022 I believed inflation would last, and that meant rates would stay high too. So in 2023 I took a 30-year fixed-rate mortgage instead of a variable or hybrid one. If rates fall from here, that choice will have cost me, and I accept that.
I never trade on report days. In the evening I check just two numbers: headline and core, against what was expected. Tomorrow, if core stays near 2.6%, this year’s energy spike is likely to fade like 2008’s. If core moves up, it’s starting to look more like 2021.
Sources
Method: We used the monthly consumer price index and the index without food and energy (core) from the Bureau of Labor Statistics via FRED, and measured each one's change over 12 months. For every month since January 1985 in which headline inflation was at least one percentage point above core, we compared two guesses for headline inflation a year later: today's headline rate and today's core rate. We counted which guess was closer and how far off each was. Data: BLS via FRED; calculated by MKViz. Code and data
- Bureau of Labor Statistics, Consumer Price Index
- Bureau of Labor Statistics, CPI release schedule
- Federal Reserve Bank of St. Louis (FRED), Consumer Price Index for All Urban Consumers
- Federal Reserve Bank of St. Louis (FRED), CPI less Food and Energy
- Federal Reserve Bank of St. Louis (FRED), CPI Energy