Stated vs. Revealed Preference Invariance in Time Valuation
RCTClinical Trial
Economists often question whether hypothetical survey-based stated preference choices reflect actual real-world labor decisions, creating doubt regarding the empirical validity of survey data in structural model calibrations.
Picture this
Think of testing whether people's food preferences on a written questionnaire match what they actually order at a restaurant menu; if both methods lead to identical ordering patterns, researchers can trust quick survey questionnaires for future menu planning.
What the evidence says
Implied flow value of non-work time $\tilde{z}$ estimated from real field choices ($0.58$, $s.e. = 0.04$) matches hypothetical choice estimates from the UAS panel ($0.61$, $s.e. = 0.03$) within statistical confidence intervals.
- Who was studied
- N = 1,152 applicants in a real-stakes field experiment compared against N = 2,354 hypothetical job choices made by 785 hourly workers in the Understanding America Study (UAS) web panel.
- How
- Methodological comparison of real job selection choices in field applications against hypothetical vignette choices with randomized wage multipliers in a nationally representative panel.
What to do
Validate hypothetical choice experiment designs by benchmarking stated preference estimates against field-experimental revealed preference parameters.
From the source
"We get similar estimates when we weight the sample to match the characteristics of hourly workers in the United States and when we use a hypothetical choice experiment in the nationally-representative Understanding America Study (UAS)."
Labor_Supply_and_the_Value_of_Non_Work_Time_Experimental_Estimates.pdf