The most important cut of any inflation number is headline versus core, and getting the logic of it right is what separates someone who reads the newspaper from someone who reads the economy.
- Headline inflation is the whole basket, every item, including food and energy.
- Core inflation excludes the most volatile components: in the US, food and energy; in India, food and fuel (roughly, CPI ex "food and beverages" and ex "fuel and light," about 47% of the basket, with a further "core-core" that also strips petrol and diesel out of transport).
Why do central banks watch core inflation?
Why on earth would you throw away food and energy, the prices households feel most viscerally? Because a central bank is not trying to measure how much households hurt this month; it is trying to measure the underlying, persistent trend of inflation so it can set a policy rate that bites 12–24 months later (M7.02's long-and-variable lags). And food and energy prices are (a) extremely volatile, whipping around on weather, harvests, and geopolitics; (b) usually supply-driven (a monsoon or an OPEC decision), i.e., exactly the cost-push shocks that monetary policy cannot fix and should often look through; and (c) frequently mean-reverting, since this month's food spike often reverses next month, so reacting to it would have the central bank yanking the rate up and down chasing noise. Core strips out the noise to reveal the signal: the "sticky," demand-driven, broad-based inflation that reflects the true state of the economy's supply-demand balance and that monetary policy can influence. That is why the Fed operationally watches core PCE above all else, and why the RBI, its analysts, and the whole market watch core CPI momentum even though, crucially, the RBI is accountable to headline CPI (see CPI vs WPI in India). Hold that nuance: watching core and targeting headline are different, and the RBI does both for good reasons.
Advertisement
What are supercore and trimmed-mean measures?
Two refinements a professional keeps in the toolkit:
- The volatile components are not "ignored"; they are watched differently. A central bank looks through a food/energy spike only as long as it stays a relative price move that does not leak into core and into expectations. The moment a sustained energy shock starts pushing up core (via transport, packaging, and wage demands) or un-anchoring expectations (Engine 4 in causes of inflation), the central bank must react. Looking through is a judgment, not a rule. In 2022–23 the Fed narrowed its focus even further, to "supercore" inflation (core services excluding housing), on the theory that this component best reflects the tight labour market and wage pressure, a reminder that "core" is not one fixed number but a family of ever-refined attempts to find the persistent signal.
- Better "core" measures than simple exclusion exist. Rather than mechanically dropping food and energy (which sometimes discards genuine signal and sometimes keeps volatile non-food items), statisticians build trimmed-mean and median measures that, each month, drop whatever categories happen to be the outliers. The Cleveland Fed's median CPI and the Dallas Fed's trimmed-mean PCE are the well-known US versions, and the Atlanta Fed's "sticky-price CPI" isolates categories whose prices change infrequently (and therefore embed expectations). When these robust core measures and the simple core measure agree, you can trust the read; when they diverge, the underlying trend is genuinely ambiguous.