What measures of inflation are important?

Stop me if you've heard this before:

In a few days, we will have a couple of important reports on inflation, which could clarify the debate on whether we can have a soft landing, that is,

.

US consumer inflation slowed in March, according to US data for April 12, 2023. (Photo by Patrick T. Fallon / AFP)

Of course, if you've been following this debate, you've heard this before.

Clarity in the inflation debate has been very hard to find, even if you ignore the people who fill my inbox with statements that the dollar is doomed and hyperinflation is just around the corner.

I have likened the debates among economists over each new inflation data to ancient Roman priests looking for omens in the entrails of slaughtered animals, a comment that curiously does not seem to have won me many friends among my colleagues.

It is also disturbing that many economists always seem to fall on the same side in these debates:

Optimists are always optimists, pessimists always pessimists.

Surely I have been guilty of the same.

So I thought I'd dedicate today's column to a probably doomed attempt to bring some clarity to this debate.

And to be honest, I'm also trying to pre-register my opinion on how to look at the upcoming data, especially the labor cost index data due out on Friday, so that I can't be accused of picking and choosing which numbers to emphasize

. of the facts.

One thing that all serious participants in these debates agree on is that you cannot

, which fluctuate wildly in response to unpredictable short-term events.

Instead, some measure of

is needed .

However, it is surprising that many economists use this term without defining what "underlying" means.

What I think it means is that there is a lot of

in some prices, but not all.

Many prices are sticky in the sense that firms do not change their prices every day or even every month;

hardly anyone changes salaries frequently, for example.

Instead, companies change prices or wages at intervals, say once a year.

How is inflationary inertia thus created?

When inflation has been relatively high for a while, companies are likely to raise their prices

each time they are revised, even when demand for their products is slack, for two reasons: They are trying to catch up with rising costs ( and competitors' prices) since their last revision, and they try

of the inflation they expect until the next time they change their prices.

This is an old idea in economics, going back at least to a

1968

I exposed an informal version in 2008;

Ivan Werning of MIT did a rigorous presentation last year (warning: highly technical).

So what I think people mean by "core inflation" is this inertial or embedded inflation, which can be

without putting the economy in trouble, perhaps with a period of high unemployment.

Implicit inflation, in turn, should reflect an average of recent inertial price increases and

inflation

But, how to put a number to implicit inflation and follow its evolution?

Before it was relatively easy.

From the 1990s until the eve of the COVID pandemic, inflation expectations seemed fairly stable.

Yes, consumer predictions for next year's inflation fluctuated wildly, but that was highly dependent on fuel prices and probably had little impact on price action.

And we seemed to have two viable measures of inertial inflation.

Core inflation, which excludes volatile food and energy prices, seemed a pretty good measure of inertial prices;

The same was true for average wages, not because labor costs necessarily drive inflation, but because wages tend to be especially sticky and thus a good indicator of price stickiness in general.

Focusing on core inflation really helped policymaking around the time of the 2008 financial crisis, helping the Federal Reserve keep its cool

oil and food price fluctuations caused big swings in inflation numbers. gross.

Unfortunately, these measures have not adapted well to the quirks of the economy since the crisis broke out.

There have been

in the prices of things other than food and energy, like used cars.

The increase in working from home caused a huge increase in demand for space and therefore rents, but it can take a year or more for market rents to be fully reflected in official measures of shelter inflation, which in turn This time makes up about 40% of core inflation, so much of the reported inflation these days reflects a rent increase that ended many months ago.

As for

, they have been affected by changes in the composition of employment.

Low-wage workers were disproportionately laid off during the worst of the pandemic, so their removal from the mix caused a spurious increase in median wages;

the opposite occurred when laid-off workers were rehired, and these composition effects may still be

the wage data.

Finally, we know that consumer inflation expectations one year ahead is a poor measure, which has led many economists, myself included, to focus on longer-term expectations - say, three or five years ahead. years - which have been much more stable.

But no one sets prices three years in advance, which raises some

about what these measures actually reflect.

So, looking at what I've seen, where do I look now?

As for

's excellent blog

has convinced me that the best measure available is probably the Atlanta Fed's

, which asks companies how much they expect their costs to rise over the next year (a figure they can really tell something about).

This number is still high from pre-pandemic levels, but not by much, and has been coming down.

As for prices, I have been looking at what has been called

inflation , which excludes housing and used cars, as well as food and energy.

Lately, the Federal Reserve has been looking at a conceptually related measure: basic services excluding housing.

This figure will be known on Thursday.

But, with all due respect, that figure

and much of what's left is hard to measure, so it feels a bit like the aforementioned gut reading to me.

As for wages, while there have been several attempts to adjust for the compositional effects I mentioned earlier, there is enough controversy among these estimates that I still rely primarily on the Bureau of Labor Statistics Labor Cost Index, which is assumed to be It is independent of the employment mix.

Unfortunately, this index is only published every three months.

This means that Friday's release will be a big event.

But beyond the question of what to look at, there remains one important consideration:

How often?

Everyone worth listening to in this business agrees that the changes in the last year are too far behind the times, while the monthly numbers are too loud.

This is why many people, myself included, have tried to extract the signal from the noise by looking at quarterly rates of change.

But lately I've been looking at those quarterly numbers, and they still seem awfully loud.

The three-month variances continue to show large dips and spikes that don't seem to be related to anything real going on in the economy.

As a result, for now, I'm leaning towards six months and plan to stick with that on Friday, even if the three-month number is closer than I want to hear.

Overall, the data seems to point to a gradual decline in inflation, even if unemployment has not risen.

I hope that as more data becomes known, this view will be increasingly confirmed.

But I am going to try not to let my hopes influence my analysis, and that is why today I have recorded my preferred measures.

Lets wait and see what happens.

c.2023 The New York Times Company

look also

The international currency madness strikes again

What's wrong with New York?