Employed vs. Unemployed MVT Divergence
RCTClinical Trial
Applying non-work time valuation models across full applicant pools without distinguishing employment status can skew estimates, because currently employed workers evaluate new work hours against secondary job earnings rather than pure leisure.
Picture this
Think of offering overtime hours to two people: one who is currently jobless and sitting at home, and one who already works a second side-hustle paying $15/hour during those same hours; the employed person demands a higher wage because taking the new hours means giving up real earnings from their second job, not just giving up TV time.
What the evidence says
The overall sample (including employed workers) yields a higher implied flow value of time ($\tilde{z} = 0.64, s.e. = 0.03$) compared to the unemployed sample ($\tilde{z} = 0.58, s.e. = 0.04$), reflecting inter-job hours substitution costs among employed applicants.
- Who was studied
- N = 2,658 total applicants (1,152 unemployed, 1,506 employed) across 80 U.S. metropolitan areas.
- How
- Subgroup econometric analysis using logit discrete choice models to compare MVT across work hour bundles between currently employed workers and unemployed jobseekers.
What to do
Restrict sample selection to unemployed jobseekers when estimating pure leisure and home production values, or explicitly control for secondary employment wage rates when evaluating employed workers.
From the source
"On the other hand, for employed jobseekers, MVT may include not only the value of leisure and household production, but also the wage in another job if workers substitute hours between jobs."
Labor_Supply_and_the_Value_of_Non_Work_Time_Experimental_Estimates.pdf