A choice-based approach to the measurement of inflation expectations
What this paper finds — and why it matters
Standard survey-based measurement of inflation expectations relies on density forecasts in which respondents assign probabilities to pre-specified inflation bins; this method has been found to induce biases through its bin structure (suggesting that values near zero are more likely), to impose cognitive demands that raise dropout rates, and to become uninformative during high-inflation episodes when responses cluster in open-ended extreme bins—making cross-time and cross-country comparisons unreliable. This paper proposes a new choice-based elicitation method rooted in decision theory that uses a bisection process: respondents first state a minimum and maximum inflation level for which they see almost no chance of actual inflation falling outside the range, avoiding external anchors, and then answer a series of binary choices from which the relevant percentiles of their subjective distribution can be inferred. Two large surveys (UK and US) and a laboratory experiment demonstrate that the method leads to well-defined expectations that fulfil both subjective and objective quality criteria, that it is neither perceived as more difficult nor more time-consuming than the density forecast standard, and that—unlike density forecasts—it is robust to differences in the state of the economy, enabling comparisons across time and countries. The method is portable and can be applied to elicit distributions over other macroeconomic variables beyond inflation.
Summary of a forthcoming paper, AI-assisted and human-reviewed. See the linked original for the authoritative claims and full conditions.
Q1. What specific failures of density forecasts motivate the new method?
The paper identifies four problems with the standard density forecast format: (i) bin-structure bias—narrower bandwidths around zero may lead respondents to infer that near-zero inflation is more likely by design, biasing responses toward zero; (ii) cognitive demands that raise dropout rates and may introduce selection bias; (iii) sensitivity to question wording and response-scale changes; and (iv) loss of informativeness during high-inflation episodes when responses bunch in extreme open-ended bins, compounded by the incompatibility of adjusted bin structures across survey waves. During the recent surge in inflation these problems became especially visible, motivating a method more robust to the state of the inflation environment.
Q2. How does the choice-based bisection method work?
The method, building on Baillon (2008), elicits respondents’ subjective inflation distribution via a series of binary choices structured as a bisection algorithm that partitions the state space into equally likely subevents, allowing the relevant percentiles of the distribution to be recovered without imposing external anchors. The procedure begins by asking respondents for a minimum and maximum inflation level for which they believe there is “almost no chance” actual inflation falls outside the interval—avoiding the bin-structure bias of the density forecast by letting respondents define their own relevant range. Subsequent binary choices then narrow down the median, quartiles, and further quantiles according to a strict algorithm.
Q3. What do the field surveys and laboratory experiment show?
Two large surveys—one in the UK testing feasibility across multiple protocol variants, one in the US—and a laboratory experiment demonstrate that the choice-based method produces well-defined expectations fulfilling both subjective and objective quality criteria, and is neither perceived as harder nor more time-consuming than the standard density forecast. The UK survey compared two variants of the proposed “Midpoint method” against existing density forecast formats. The convergence on quality criteria across different samples and settings supports the method’s potential for adoption in large-scale central bank surveys.
Q4. What makes the method robust to the state of the economy, and why does that matter for monetary policy?
In contrast to density forecasts, the choice-based method is robust to differences in the level and volatility of inflation because respondents define their own relevant range rather than choosing among fixed pre-specified bins, so the method does not become uninformative when actual inflation is far from the bins’ central mass. This robustness allows comparisons of inflation expectations distributions across time (including across high- and low-inflation regimes) and across countries—a feature density forecasts cannot deliver without adjusting bin structures in ways that compromise comparability. For monetary authorities that use survey expectations as both an indicator and a policy tool, this portability is a key advantage.
Key concepts
density forecast : the standard survey format in which respondents assign subjective probabilities to pre-specified inflation intervals (bins); the format used by the Federal Reserve Bank of New York’s Survey of Consumer Expectations and widely adopted by central banks.
choice-based elicitation (Midpoint method) : the paper’s proposed alternative; a bisection procedure in which respondents first report a subjective min/max range and then answer binary choices, yielding quantiles of the subjective inflation distribution without imposing an external bin structure or anchors.
bisection process : the algorithmic structure in which each binary choice partitions the remaining probability mass so that successive responses identify the median, quartiles, and further quantiles of the respondent’s subjective distribution.
bin-structure bias : the distortion introduced by the density forecast’s pre-specified bins when narrower intervals near zero suggest to respondents that near-zero inflation is considered more likely by the survey designers, biasing their reported probabilities toward zero.