Equilibrium Search Modeling via Indirect Inference
Partial microeconometric cost-benefit analyses evaluate policies based solely on individual participant benefit reductions, ignoring market-wide congestion and vacancy costs. Consequently, naive evaluations incorrectly predict positive outcomes for interventions that actually decrease total social welfare upon full expansion.
Picture this
Evaluating a policy using only local experiment data is like deciding to build a nationwide highway based on one smooth toll lane, completely overlooking that opening the entire highway creates massive traffic bottlenecks elsewhere.
What the evidence says
A full rollout ($\tau = 1.0$) decreases overall social welfare by 0.13% and increases aggregate government spending due to high program costs (2,122 DKK per worker) and market congestion. While microeconometric methods wrongly estimate a 4,094 DKK net gain per participant, structural modeling proves welfare is maximized at 20% participation and government expenditure is minimized at 30% participation.
- Who
- Structural Diamond-Mortensen-Pissarides model calibrated with micro-data moments from 89,466 Danish benefit spells and baseline treatment intensity set at 30% ($\tau^e = 0.3$) [5, 26, 27].
- How
- Discrete-time search model with urnball matching function estimated via indirect inference to simulate full-scale program rollouts ($\tau = 1.0$) [26-28].
What to do
Simulate large-scale policy rollouts using structurally estimated equilibrium search models before scaling active labor market interventions nationwide.
From the source
"This model shows that a large scale role out of the activation program decreases welfare, while a standard partial microeconometric cost-benefit analysis would conclude the opposite."
Estimating Equilibrium Effects of Job Search Assistance