Paul Krugman is desperate to declare victory over inflation and that a soft landing has been achieved. His latest victory lap op-ed cites the Atlanta Fed’s survey of Business Inflation Expectations (BIE), which has settled down to 2.2 percent for the coming 12 months.
The BIE reflects about 200 Atlanta-area business executives’ responses to survey questions. The graph above shows the mean response to “Projecting ahead, to the best of your ability, please assign a percent likelihood to the following changes to unit costs over the next 12 months.” They are given a set of options that include “Unit costs down” up to “Unit costs up very significantly.” An average inflation expectation is then calculated based on the responses.
Krugman says that this is a superior measure of price inflation expectations compared to consumer expectations: “You see, surveys of consumer expectations are fine, but ordinary consumers don’t have strong incentives to think hard about how much prices will rise in the near future. […] Business expectations, however, are both well-informed and important.”
Once again, just a quick look under the hood reveals that Krugman is reaching for evidence to back up his own views instead of doing impartial macroeconomic analysis. He claims that this measure is “a better predictor,” but that is dubious.
Consider the fact that this measure forecasted 3.0 percent price inflation for June 2022, when official measures of price inflation turned out to be 8.9 percent (CPI), 22.4 percent (PPI), and 7.1 percent (PCE).
The Atlanta Fed likes to compare the BIE to a GDP-based price index, shown below:
But we must be careful here, because there are two y axes: one for the BIE and one for the GDP price index. While they have calculated a correlation of .96, the realized price inflation data is more exaggerated than the BIE data. The left axis ranges from one to five while the right axis ranges from zero to 8. The data may be correlated, but that doesn’t mean that Paul Krugman can imply that the BIE is an accurate, one-to-one predictor of price inflation.
The BIE data seems to be consistently muted compared to all the official inflation measures. This is most plainly seen when we compare the BIE to PPI. (You may wonder why Krugman didn’t just use the current PPI to make his case that inflation is dead. PPI has been negative for the past few months!)
It’s easy to see why this measure is consistently muted when you look at the way the data is calculated based on the survey responses.
As a reminder, the respondents are asked to “assign a percent likelihood to the following changes to unit costs over the next 12 months.” Notice that “Unit costs up very significantly” contains everything greater than five percent. This means that if a respondent thinks that there is going to be a ten percent increase in unit costs, it is coded as five percent. There’s no way for a respondent’s high inflation expectation to be translated into anything over five percent. This is why the minimum calculated mean is 1.4 percent and the maximum is 3.8 percent over the 13-year series.
Krugman has chosen an inflation expectations measure that is restricted by design. It will hover around two percent no matter what the respondents’ real inflation expectations are.
If you look at other measures of inflation expectations, you’ll see why Krugman chose the one he did.
The University of Michigan measure has recently spiked up to 4.5 percent.
Source: “Surveys of Consumers, University of Michigan, University of Michigan: Inflation Expectation© [MICH], retrieved from FRED, Federal Reserve Bank of St. Louis https://fred.stlouisfed.org/series/MICH/
Another well-cited measure from The Conference Board indicates 5.6 percent consumer inflation expectations as of December 2023.
But 4.5 percent and 5.6 percent don’t help Krugman’s case.