In Simple Words
When a corporate headquarters issues new customer service rules, different branch managers enforce those rules with different levels of strictness. As a result, customers in one branch experience frequent check-ins while customers in another experience looser oversight, effectively turning one rule into two distinct operational environments.
Treated workers observed across Danish counties
- Storstrom county
- About 40% attended a meeting each week
- Southern Jutland county
- About 30% attended a meeting each week
Among workers still unemployed, mainly between the first job-search and second programs. Approximate delivery shares; county implementation was not randomized.
See source →Research question
Centralized social experiments and national labor policy mandates encounter varying levels of compliance and discretionary enforcement across local job centers and caseworkers.
Participants
N = 4,513 unemployed UI recipients split across Storstrøm and Southern Jutland counties in Denmark.
Study Design
Separate parametric duration model estimations for each county to capture implementation differences and avoid pooling bias.
Results
Meeting intensity reached ~0.4 (once every 2.5 weeks) in Storstrøm compared to ~0.3 (once every 3 weeks) in Southern Jutland against a weekly protocol target, yielding distinct meeting risk coefficients (0.47 in Storstrøm vs 2.30 in Southern Jutland).
Takeaway
Evaluate decentralized trial data by estimating separate models for each administrative region rather than pooling data across regions with unequal enforcement rates.