Dynamic Covariate Updating in Treatment Selection
Static assignment models evaluate jobseeker characteristics only upon initial registration, ignoring how unemployment duration and changing local labor market conditions alter optimal treatment timing. Static systems risk assigning costly interventions prematurely or missing critical intervention windows.
Picture this
Imagine a digital GPS that updates route recommendations every two minutes based on real-time traffic jams rather than relying on a static printout generated before starting the drive. As driving conditions change, the recommended route changes to ensure the driver stays on the fastest path.
What the evidence says
Biweekly prediction recalculations dynamically shifted optimal recommended interventions over the unemployment spell, demonstrating that optimal programme timing varies substantially across individuals as unemployment duration accumulates.
- Who
- Field predictions generated biweekly for N = 18,713 jobseekers in treatment groups across 21 Swiss employment offices.
- How
- Biweekly re-estimation of individual potential employment outcomes incorporating updated time-varying covariates X_it, specifically tracking current elapsed unemployment duration.
What to do
Re-calculate individual potential outcome predictions biweekly using updated elapsed duration and labor market indicators to adjust treatment recommendations dynamically.
From the source
"The predictions were updated every second week by incorporating new information on time varying covariates (in particular unemployment duration). This is a big advantage vis-à-vis simple profiling models as it takes into account that the optimal time when a labour market programme should start may also vary across individuals."
Targeting_Labour_Market_Programmes_Results_from_a_Randomized_Experiment.pdf