Real-Time Inflation Expectations: Households vs. Generative AI

Last registered on July 27, 2026
View Trial History

Pre-Trial

Trial Information

General Information

Title
Real-Time Inflation Expectations: Households vs. Generative AI
RCT ID
AEARCTR-0019243
Initial registration date
July 25, 2026

Initial registration date is when the trial was registered.

It corresponds to when the registration was submitted to the Registry to be reviewed for publication.

First published
July 27, 2026, 7:04 AM EDT

First published corresponds to when the trial was first made public on the Registry after being reviewed.

Locations

There is information in this trial unavailable to the public. Use the button below to request access.

Request Information

Primary Investigator

Name
Yoshiyuki Nakazono
Affiliation
Yokohama City University
Contact Primary Investigator

Other Primary Investigator(s)

PI Name
Gaku Tanabe
PI Affiliation
Yokohama City University
Contact Investigator
PI Name
Jun Takahashi
PI Affiliation
Yokohama City University
Contact Investigator
PI Name
Kento Tango
PI Affiliation
Yokohama City University
Contact Investigator

Additional Trial Information

Status
In development
Start date
2026-08-10
End date
2026-08-31
Keywords
Behavior
Additional Keywords
JEL code(s)
Secondary IDs
Prior work
This trial does not extend or rely on any prior RCTs.
Abstract
We compare real-time inflation expectations between households and generative AI using identical survey settings.
External Link(s)

Registration Citation

Citation
Nakazono, Yoshiyuki et al. 2026. "Real-Time Inflation Expectations: Households vs. Generative AI." AEA RCT Registry. July 27. https://doi.org/10.1257/rct.19243-1.0
Sponsors & Partners
Experimental Details

Interventions

Intervention(s)
Intervention Start Date
2026-08-10
Intervention End Date
2026-08-31

Primary Outcomes

Primary Outcomes (end points)
Posterior inflation expectations (point forecasts of the inflation rate over the next 12 months and over a longer horizon), elicited after information provision from (i) household respondents and (ii) generative AI (large language model) respondents surveyed under identical instruments; revisions of inflation expectations (posterior minus prior forecasts).
Primary Outcomes (explanation)
Prior inflation expectations are elicited before the information intervention, and posterior inflation expectations are elicited immediately after. Respondents are randomly assigned to one of three groups: (1) a treatment group informed of the Bank of Japan's 2% inflation target, (2) a treatment group informed of the Bank of Japan's inflation outlook (forecast), and (3) a placebo group informed of Japan's population growth rate. The same survey instrument and information treatments are administered to generative AI agents. The primary endpoints are: (a) the level, bias (deviation from realized/benchmark inflation), and cross-sectional dispersion of posterior inflation expectations; (b) the revision of expectations (posterior minus prior) in each treatment arm relative to the placebo arm; and (c) differences in (a) and (b) between household and AI respondents.

Secondary Outcomes

Secondary Outcomes (end points)
Secondary Outcomes (explanation)

Experimental Design

Experimental Design
We conduct an online survey experiment on inflation expectations in Japan. Household respondents are randomly assigned to one of three groups: two treatment groups that receive publicly available information about monetary policy, and one placebo group that receives information unrelated to inflation. We elicit inflation expectations before and after the information provision. In parallel, we administer the identical survey instrument, including the same random assignment to information treatments, to generative AI (large language model) agents. We compare inflation expectations and their responses to information between households and generative AI.
Experimental Design Details
Not available
Randomization Method
Randomization is conducted by the survey company using a computer.
Randomization Unit
Randomization is conducted at the individual level. The sample is stratified by gender and 10-year age brackets to match the population distribution, and random assignment to treatment and placebo groups is carried out within these strata.
Was the treatment clustered?
No

Experiment Characteristics

Sample size: planned number of clusters
1,000 individuals (250 individuals per experimental arm)
Sample size: planned number of observations
1,000 individuals (household respondents)
Sample size (or number of clusters) by treatment arms
250 individuals in the treatment group receiving the Bank of Japan's inflation target information; 250 individuals in the treatment group receiving the Bank of Japan's inflation outlook information; 250 individuals in the placebo group receiving information on Japan's population growth rate; 250 individuals in the control group receiving no information.
Minimum detectable effect size for main outcomes (accounting for sample design and clustering)
Supporting Documents and Materials
IRB

Institutional Review Boards (IRBs)

IRB Name
Yokohama City University
IRB Approval Date
2026-05-25
IRB Approval Number
N/A
Analysis Plan