(Mis)measuring Uncertainty: Bunching in Probabilistic Expectations -- by Carola Binder, Laura Crespo, Carlos Gento, Luis M. Guirola, Ernesto Villanueva

TEXT START: Probabilistic expectation questions are often used to measure subjective uncertainty, but respondents frequently assign all probability to one outcome.

The Dissection

This paper is not measuring the future. It is auditing the instrument used to measure uncertainty. Its finding is concrete: degenerate answers are partly produced by cognitive burden and survey administration—especially among less financially literate respondents, with longer panel tenure, nonneutral interviewers, and flawed questionnaire design.

The empirical contribution is real. Reported certainty can be manufactured by the measurement process.

The Core Fallacy

Relative to the Discontinuity Thesis, the danger is mistaking measurement correction for structural correction. Replacing a pile of zero-one answers with a more diffuse probability distribution does not restore wages, jobs, bargaining power, or productive necessity.

The paper improves the map of human expectations. DT concerns the destruction of the economic road system itself. Better measurement of uncertainty cannot reverse P1–P3. It can only reveal whether people are genuinely uncertain or merely struggling with the survey instrument.

If the findings are used to imply that improved expectations data preserves the existing macroeconomic order, that inference is false. Genuine uncertainty can coexist with AI dominance, labor displacement, and collapse of the wage-consumption circuit.

Hidden Assumptions

• Respondents can articulate calibrated probabilities once the questionnaire is improved.
• Lower bunching represents latent uncertainty rather than new forms of guessing, strategic response, or noise.
• Interviewer neutrality and redesign produce responses closer to an underlying “true” belief distribution.
• Financial literacy differences primarily reflect task burden rather than deeper differences in beliefs or incentives.
• Panel-tenure effects are survey artifacts rather than selection, learning, fatigue, or changing economic conditions.
• More accurate uncertainty measurement will materially improve policy or forecasting outcomes.
• Better information can preserve institutional control even when the productive participation of the majority is collapsing.

The final assumption is the one DT rejects. Measurement quality is not economic viability.

Social Function

Classification: partial truth with technocratic transition management and prestige signaling.

The paper performs useful bureaucratic maintenance. It helps central banks and researchers stop treating survey-induced certainty as genuine belief. But this is still maintenance of the dashboard, not repair of the engine. It reinforces the institutional reflex that better instruments and cleaner data keep the system governable, even when the underlying labor architecture is being hollowed out.

The Verdict

This is a credible measurement autopsy and no rebuttal to DT. Its strongest result is that some apparent certainty is survey-produced. Its systemic limit is absolute: correcting the artifact merely exposes underlying uncertainty; it does not restore human economic necessity. The institutions may gain a longer statistical and social-management lag. They do not regain the mass employment circuit. The dashboard becomes more accurate while the machine it monitors continues toward obsolescence.