Non-technical summary

Research Question

Survey-based measurements of inflation expectations are of growing importance for policy makers and researchers, especially in times of heightened inflation volatility. It is common to measure expectations via probabilistic question, when respondents are given a set of intervals (bins) and are asked to assign the probability to the intervals that represent their beliefs about future inflation. This method has serious drawbacks, such as scale-dependence and anchoring, which coincidentally become particularly relevant in times of high and volatile inflation.

In this paper we propose a novel method of measuring inflation expectations in a survey, which is rooted in decision-making theory, and at the same time is easy to implement in a survey. We conducted a survey of households, where respondents either answered a variation of the standard probabilistic question, or reported their inflation expectations using a variation of the newly proposed survey method.

Contribution

The most important advantage of our method is that it allows to elicit the full distribution of respondent’ inflation expectations, but does not use a fixed scale. It is therefore robust to the state of the economy and allows comparison over time and across different countries.

Additionally, it does not impose high cognitive demands on the respondent when answering the question, in contrast to the standard density forecast method, which requires individuals to at least understand the concept of probability.

Finally, the proposed method can be applied to measure beliefs about different macroeconomic concepts.

Results

We demonstrate that our method delivers a well-defined measure of inflation expectations with central tendencies close to the corresponding point forecast.  At the same time, measured uncertainty about future inflation is lower than resulting from the standard method. The reported difficulty of the questions in the survey and the objectively measured time spent on answering them is below or similar to the currently used standard method.

Our results suggest multiple benefits from using the proposed method to measure inflation expectations.  From the perspective of measurement bias, our method does not suffer from the exogenously imposed structure or anchors to the respondents’ beliefs. From the perspective of the analysis, flexibility of the method allows comparison over time and economies. Additionally, this new method allows to measure and compare expectations about economic variables of very different magnitudes, such as low economic growth and high inflation.