Inflation 1

We're seemingly going to need to do a lot on inflation this year, so this is just a primer to get started.

Why are we interested in this right now? Because the U.S. has started putting up inflation rates that have not been seen in 30-40 years. We are not alone in this, amongst comparably developed countries.

What is inflation? Statistically, it's a rate of change of prices.

But what prices? That depends. A rate of change of one price would be easy to figure out. But what about more than one? For this we need to take some sort of average of prices. But because some goods are more important than others, it should be a weighted average, so that each one can have a different weight.

A weighted average of prices is called a price index. There are many of these, with each including different goods and services.Some goods and services are excluded from some price indices, so their weight would be zero. For the ones that are included, the weights can be different, but should be expected to be similar: they don't just make up the weights, but instead rely on actual decisions made by buyers in the macroeconomy.

But, there's different mathematical methods for calculating indices (chain weighting is one of those). And they make a difference in the final numbers.

Also, prices sometimes go up because quality improves. In the background, good indices correct for this. But it's not easy to judge quality, so everyone is aware this may be a bit sketchy. We put up with that because there isn't much alternative to doing the best measurement job that you can. 

Another thing they correct for is changes in packaging size: a smaller package at the same price (as my daughter noted Taco Bell has recently done with their quesadillas) is a trick to hide inflation. The statisticians are on the lookout for this, but it's a big country to catch all of these.

There are many price indices, depending on statistical needs. Two are primarily used for calculating inflation for the U.S.: the consumer price index (CPI), and the personal consumption expenditures (PCE) price index. The CPI will tend to show higher inflation rates. 

A weird but important quirk that I learned about this year is that, unlike almost all other economic data series, the CPI is never corrected retroactively. When most data series are revised and updated, this is done retroactively as far back as needed. So there's an old version that ends at the revision, and a new version that overlaps it, continues forward, and is used from there on. You can imagine in the parts that overlap that there are dates when the new and old series are closer together and further apart. Not so with the CPI. With it, revision and updates are introduced, but only from that point going forward. So imagine something like line segments that touch where one ends and another begins, but which might have a kink at that point. I believe the reason it's done this way is that the CPI is commonly used in labor contracts, and if they changed the numbers retroactively after a contract was signed, one side or the other might sue to claw back some of the differences.

There are also indices for wholesale prices and producer prices. These get watched because it's thought that inflation in those will show up in inflation in consumer prices a few months later. 

One last thing: you sometimes see discussions of "core" inflation.† Price changes can be very volatile. And volatility can make it hard to spot trends. So the idea here is that if you remove the most volatile prices, what you'll be left with is a cleaner view of inflation. The most volatile prices faced by consumers are for food and energy, so core inflation means those are excluded.

† Maybe my personal biases are bleeding over into my professional viewpoint here, but I hate the use of core inflation. To me it seems like it's excluding the things that people actually notice the most.