Sequential 'No Programme Now' Dynamic Baseline
Evaluating active labour market programmes using a static non-treatment baseline ("no programme for 12 months") introduces selection bias because jobseekers who remain untreated for a full year are disproportionately those who found employment quickly. In dynamic counselling, the true counterfactual baseline is "no programme in the current period, but possible participation later."
Picture this
Think of deciding whether to take an umbrella when leaving home. "No umbrella now" does not mean promising never to use an umbrella for the rest of the year; it just means stepping outside right now without one, while keeping the option open to grab an umbrella later if it starts raining.
What the evidence says
Defining Option 0 dynamically eliminated survivor selection bias in counterfactual estimation, showing that "no programme now" was the statistically optimal choice for 21–23% of jobseekers in Basel and Bern.
- Who
- National population sample of N = 460,442 Swiss jobseekers aged 25 to 55 from administrative databases (2001–2003); biweekly treatment evaluation.
- How
- Counterfactual potential outcome estimation where Option 0 is defined dynamically as $E[Y^0_{i,t+\tau}|X_{it}]$ ("no active programme in month $t$, but possible participation in $t+1$ or later").
What to do
Define baseline comparison choices in dynamic evaluation models as "no intervention at current period $t$" rather than permanent non-treatment to eliminate survivor selection bias.
From the source
"The first category 'No programme' means that the jobseeker is not allocated to any ALMP in this month, but leaving the option for the future, if still unemployed then. This category could therefore also be labelled as 'waiting' or 'no programme now but perhaps later'."
Targeting_Labour_Market_Programmes_Results_from_a_Randomized_Experiment.pdf