Experimental Design Details
The experiment will be conducted online (using Prolific) and will consist of six parts. In the first part, potential respondents must pass a reCAPTCHA question and are asked three standard financial literacy screening questions (Lusardi and Mitchel, 2014). The text of the screening questions is presented as a picture to help prevent non-human respondents. Participants must correctly answer all three questions to proceed to the main experiment.
In the second part of the experiment, participants are given a brief tutorial on Treasury bills. They are told what bills are issued for and how they are sold. After the tutorial, participants are asked a set of comprehension questions and required to answer correctly before proceeding. At this stage, participants continue to answer each question until they submit the correct answer.
In the third part of the experiment, we elicit participant beliefs about inflation. They are told that we are interested in their expectations and do not want them to look it up -- for this reason (and the fact that there is no ``right'' answer to the question), we do not incentivized these expectations. We use a version of the recent Gonzalez-Fernandez et al., (2025) procedure for eliciting the first two moments of an individual's expectations.
Participants are then randomized to treatment and control. In the control condition, respondents are reminded of their inflation expectations and asked to bid on two 12-month T-bills. The first is a nominal bill paying \$100 in one year and sold at a discount. The second bill (a TIPS) pays 100 inflation-adjusted dollars in a year. For each bill, the Becker-DeGroot-Marshack procedure is used to incentivized truthful bidding. Participants are informed that one participant will be chosen at random for each bill to either receive the nominal or real face value in one year, if their bid is over a randomly chose ``market price'' or the endowment they are given to bid on the bills if their bid is below the market price.
The only difference between control and treatment is that treated participants learn the March 6, 2026 mean core inflation prediction published by the Survey of Professional Forecasters. These participants are then asked to compare their estimate to the forecaster's estimate. When bidding on the two bills, ``corrected'' participants see the forecaster's prediction instead of their own estimate.
All participants then do an incentivized portfolio allocation task in part five of the experiment. They are each endowed with \$100 that they can allocate to four assets: Cash (which earns 0\%), Nominal bill (that earns 5\%), A TIPS (that earns 2\% plus inflation) and a Stock that earns +20\% with half probability and loses -10\% with half probability. One person, chosen at random, will then earn the value of their portfolio in one year.
As part six of the experiment, all participants are asked a series of questions about (i) the extent to which they find the Survey of Professional Forecasters prediction credible, (ii) further financial literacy questions, (iii) basic demographics, and (iv) information about their time and risk preferences.
The experiment is expected to last 15 minutes and participants are paid a fixed completion payment of \$3. They also have a chance to have one of their three incentivized choices be paid, resulting in a bonus of around \$100.