On June 10, the government released its report on the Consumer Price Index (CPI) for May of 2021. These numbers can be sliced and diced in many ways. I am going to show you one reasonable way to look at the figures, and this indicates an annual inflation rate of 8 percent. This is the highest since the summer of 1982. That is thirty-nine years ago, perhaps even before you were born.

How should one measure the annual rate of inflation? When I was younger, the media reported the annualized one-month rate of inflation. If the CPI went up by 1 percent from one month to the next, reporters would multiply by 12 to get a 12 percent annualized rate of inflation. In fact, because of compounding, the true annualized rate would be even higher, about 12.7 percent.

For May of 2021, the government reported a one-month increase of 0.6 percent. Multiplying by 12 gives 7.2 percent as the annualized rate of inflation. But I have not seen any news media reporting this number.

Most media instead report the rate of inflation from May 2020 to May of 2021, which is 5 percent. That is called “year-over-year” inflation, and it is another measure of annualized inflation. Most news stories are quick to point out that May 2020 consumer prices were “depressed,” which they suggest artificially makes the jump to May of 2021 large.

But if the year-over-year inflation rate of 5 percent is biased upward, why not use the monthly annualized rate of inflation instead? The monthly annualized rate of inflation does not suffer from “base bias.” But the monthly annualized rate of inflation of 7.2 percent is even higher than the “biased upward” year-over-year rate.

In terms of arithmetic, the “base bias” story is wrong. It turns out that the year-over-year inflation rate is being held *down *by data from the pandemic, which exerted an influence most of the latter part of 2020, not just May. Looking at inflation year over year gives a picture that is artificially low, not artificially high.

### Special Factors

One argument against relying on a monthly inflation rate is that it is affected by special factors. For many years, sophisticated analysts have looked at the “core CPI,” which excludes food and energy, two categories where prices bounce around a lot from month to month. Looking at the core CPI removes transitory blips that have taken place over recent decades.

But food and energy were not the drivers of the recent run-up in the CPI. In contrast to past transitory upticks in inflation, in April and May the core CPI increased by more than the overall CPI. The food and energy sectors actually exerted a slight moderating influence.

If we take March, April, and May together, the core CPI rose by 2 percent over this period. Multiplying by 4 to get an annual rate gives **8 percent**. This could very well be the most reasonable estimate of the current rate of inflation.

This annualized three-month increase in the CPI of 8 percent is a dramatic breakout. The last time we saw a number that high was July of 1982.

The 8 percent three-month annualized core inflation is *more than double* anything that we have seen for almost the entire period of the last 25 years. The three-month annualized core inflation rate was below 4 percent for the entire period from April of 1995 through July of 2020. Last August and September saw three-month core CPI increases of 1.14 percent and 1.08 percent, but this uptick was followed by a sharp drop-off.

What about special factors other than food and energy? It is true that used car prices will not continue to rise as rapidly as they have recently. But on the other hand, the CPI is very slow to pick up on the increased cost of housing, which is evident to anyone tracking the residential real estate market. Double-digit increases in home prices have yet to be reflected in the CPI, but they almost surely will show up eventually.

Inflation has never been easy to measure, and it is particularly difficult today. (See Is the Economy Illegible?) Some industries are booming, and some are depressed. More and more economic activity takes place in hard-to-quantify sectors, such as health care and education, where the challenge is to divide spending totals between increases in real output on the one hand and increases in prices on the other.

But for this purpose, I am just looking at the CPI as reported by the government statisticians. If you rely on media reports, you are not getting the real story told by these numbers. Reporters are claiming that inflation is only 5 percent, and they add that even this number is overstated. Nothing to see here, move along.

I am looking at the same numbers but taking a three-month average multiplied by 4, and using the core CPI. Such a calculation informs me that inflation is actually 8 percent.

For the past 70 years (yes, since 1951), I have been watching my fellow Argentinians talk about inflation (less than 10% per year), high inflation (around 20%), very high inflation (around 50%), hyperinflation (around 50% but per month). I have also watched and studied the inflationary experiences of other Latin American countries.

My first rule is that if a country has an inflation rate of over 10% per year over a period of at least three years regardless of what price index is used, then we can conclude that the national government is financing all or part of its deficit by relying on the inflation tax.

My second rule is that if a country has a sharp increase in all price indexes in a period of not more than 3 months (say over 20% from January to April) followed by much lower increases (say monthly rates of no more than 1%) then there was a need to adjust relative prices through increases in some nominal prices, and the adjustment doesn't amount to inflation.

My third rule is that any combination of an initial sharp increase in all price indexes with monthly rates of more than 1% over a period longer than 3 months, then the relative-price adjustment has failed and be ready for another adjustment.

My fourth rule is that in any other situation one has to pay close attention to mix signals from price indexes. Check first for relative-price adjustments and their effects on price indexes. Then check the convergence of all price indexes to the same inflation rate. Always remember that there are significant differences in how nominal prices of the many goods and services change over time.

We can apply those rules to discuss the details of all past experiences in all countries after WWII. To apply them to the understanding of what is going on in the U.S., we should start by looking at what has been happening with all price indexes, separating between indexes in which a few goods account for at least 40% of the basket and all others. A final judgement must be delayed until at least January 2022.

Also, remember that the available data on government expenditures and their financing are not reliable and hardly relevant to determine the relative importance of the inflationary tax and debt financing. That's true in Argentina, and everywhere, including the U.S.

Looking at the 3-month-over-3-month inflation rate is a useful way to smooth out unusual factors, so the suggestion of a near-term 8% annual rate of inflation rate is apt. The question is whether high inflation (e.g. greater than 2.5% or 3% annual rate) will be with us for a while because, if so, then interest rates will have to go up more than otherwise. (And currently interest rates are low mainly because of central bank financial repression around the world.) Lots of arguments on both sides of the question of persistent high inflation to come. My sense is that cat is already out of the bag and inflation fighting will be the next phase of monetary policy, sooner than central bankers are saying publicly.