Consumer Inflation Uncertainty Is Rising

Inflation Uncertainty at Longer and Shorter Horizons

The relative stability of median longer-run inflation expectations is not, by itself, a clear sign that consumers’ inflation expectations are firmly anchored. It is important to consider how uncertain consumers feel about their inflation forecasts. One might expect consumers with well-anchored expectations not only to report a long-run inflation forecast that is near the Fed’s target, but also to have significant confidence in their forecast. If consumers report an inflation forecast of 2 percent, for example, but believe that there is a good chance that inflation could be much higher or much lower, then those consumers do not have a lot of confidence in a near-target outcome. Moreover, high uncertainty about longer-run inflation can make it very hard for households to make sound financial decisions.

Measuring inflation uncertainty is not entirely straightforward. The MSC, which has been conducted monthly since 1978, asks consumers for only their point forecast of inflation; it does not ask them how certain they are about their forecast. The SCE, introduced in 2013, gauges the level of respondents’ uncertainty. Specifically, the SCE asks for respondents “density forecasts” of inflation. That is, respondents assign probabilities to various ranges, or “bins,” of future inflation. These probabilities should add up to 100 percent. If respondents assign all probability to a single bin, one can infer that they are certain about the inflation outcome that they expect. By contrast, if respondents distribute probability across many bins, they are uncertain.

The researchers at the Federal Reserve Bank of New York who run the SCE use each respondent’s bin probabilities to estimate the parameters of a smooth distribution representing the respondents’ density forecast of future inflation. From those parameters, they compute the interquartile range of the respondents’ density forecast. The width of this interquartile range can be used as a proxy for the respondents’ inflation uncertainty.

An Alternative Measure of Inflation Uncertainty

In 2017, I wanted to find a way to measure inflation uncertainty using the MSC data, given that they are available for a much longer period than those of the SCE and given that the 5- to 10-year forecast horizons are of great interest to policymakers. I wanted this measure also to not rely on the assumptions the Federal Reserve Bank of New York needs to fit distributions to respondents’ bin forecasts. In creating such a measure, I discovered that respondents have a notable tendency to report round-number point forecasts for inflation—that is, multiples of 5 percent. This rounding tendency varies substantially over time. On the basis of the literature in psychology, cognition, and communication, I inferred that at least some respondents use round number forecasts to convey their high uncertainty about their responses.

My 2017 paper provides the technical details of how I used that rounding behavior to construct inflation uncertainty indices for the 5- to 10-year forecast horizons and how I verified that the rounding behavior is a good proxy for uncertainty. In short, the indices capture the percentage of respondents who are likely to be highly uncertain about inflation, given their survey responses. A round number forecast alone is not enough to categorize a respondent as uncertain, but if a respondent provides a round forecast—say, 10 percent—in a month when very few respondents report expectations of neighboring nonround numbers (9 percent and 11 percent), then I assigned the respondent a high probability of being highly uncertain. In figure 2, I plot both the SCE 2- to 3-year inflation uncertainty measure, which corresponds to the median respondent’s forecast interquartile range, and the 5- to 10-year inflation uncertainty index using my methodology based on rounding behavior. The two series have a very strong positive correlation coefficient of 0.85.