How to read inflation data without getting fooled

One morning a couple of months back I had two news sites open while the coffee brewed. Same day, same inflation report. One headline: inflation's heating up. The other: inflation continues to cool. I stared at them for a second and thought — somebody's lying to me.

Turns out, mostly, nobody was. They were just describing different things and counting on me not to notice. That annoyed me enough to build myself a little field guide, so I could read this stuff like a person who isn't being sold something. Here it is.

The four numbers people mean when they say "inflation"

Headline CPI, year-over-year. The Consumer Price Index, all items, compared to twelve months ago. This is the one the government reports each month. "CPI rose 3.2% in March."

Core CPI. Same thing, but with food and energy pulled out — because those two bounce around a lot and can hide the underlying trend.

Month-over-month CPI. How much prices moved just last month. Good for spotting turning points, easy to over-read.

Specific item prices. What an actual thing costs now versus then. Eggs went from this to that. Gas went from this to that. This site is mostly about that last one.

So a news story can say "inflation is high" using the first number and "inflation is cooling" using the third, while a politician waves the fourth one around and shouts about eggs. None of them is technically lying. They're just standing in different spots.

Once I understood that, the two coffee-morning headlines stopped feeling like a betrayal and started feeling like a card trick I could now see through.

What "inflation is X%" actually means

When you see "CPI rose 3.0% year-over-year," that does not mean every single thing got 3% pricier. It's a weighted average of about 80,000 specific prices, with each item weighted by how much normal households actually spend on it.

In practice:

Housing is roughly a third of the whole index. Rent and "owners' equivalent rent" together are about 33% of CPI. If rent inflation flattens out, the headline number flattens — even if other stuff keeps climbing.

Food is about 13%, energy about 7%. That's why "core" — which drops both — is sometimes a cleaner read on the trend.

Eggs are about 0.1% of the index. A 50% egg spike, by itself, barely moves CPI. Which is exactly why the vibe and the number disagree on egg-spike weeks. I'm not experiencing the index weights. I'm experiencing the cart Jenna sent me to fill.

The four cheap tricks to watch for

When somebody uses a number to win an argument, I check these four things now.

Cherry-picked start date. "Gas is up 90% since 2020!" — technically true, because gas was freakishly cheap in 2020 when nobody was driving. Picking April 2020 as your baseline is like measuring a kid's growth from the morning they had the flu. Always ask: why that start date?

Cherry-picked endpoint. "Egg prices are down 40% from their peak!" — also true, but the peak was a bird-flu crisis. Coming down off a crisis isn't the same thing as being affordable.

Year-over-year versus cumulative. "Inflation is back to 2%." That means prices are still rising — just slowly. The total price level is up a lot from a few years ago. "Back to 2%" doesn't mean prices fell. Inflation is the speed, not the price.

The wrong index for the question. CPI is consumer prices. PPI is wholesale. PCE is the one the Federal Reserve actually steers by. A lot of arguments about Fed policy use CPI when they should use PCE, and just... don't mention it.

How to read this site honestly

A few things we try to do here — and how I'd want you to push back if we get sloppy:

We anchor to January 2019 for percent-change comparisons. It's pre-COVID, past the 2018 trade-war noise, and reasonably stable. It's an opinionated pick. If you've got a better one, tell us.

We show prices, not just rates. A line going from $2.99 to $4.99 is more useful to me than "+67%" floating in a headline. Both are on every page.

We mark estimates as estimates. Some data series don't publish state-level numbers, so we sometimes apply regional adjustments to estimate a state price. When we do, we say so, right on the chart.

We try not to imply causation we can't prove. "Carpooling won't lower national gas prices" is something we'll keep repeating, because the household savings are real even when the macro effect isn't.

If you read inflation coverage with those four tricks in mind, most of it gets a lot less confusing. The headline number is real — but it's a summary. Your grocery cart is also real, and it's not a summary.

Both are true. Both are useful. The card trick is anyone telling you to pick one.