Welfare Contrasts in Policy Choice

In Simple Words

Think of evaluating a new school lunch menu against the old default menu. Welfare contrast 1 checks if adapting choices during the school year fed kids better than flipping a coin every day. Contrast 2 measures how much health improves if every single student gets their exact personalized ideal meal instead of the old default. Contrast 3 measures how much health improves if every student gets the single best overall group meal instead of the old default.

Research question

Evaluating experimental interventions requires distinguishing between adaptive learning gains realized during trial execution, total potential welfare gains from an optimal targeted policy, and welfare gains from an optimal non-targeted policy compared to baseline controls.

Participants

N = 3,770 Syrian refugees and Jordanian jobseekers in urban Jordan.

Study Design

Bayesian hierarchical posterior expectations evaluating three formal welfare contrasts ($\Delta_1$: adaptive vs equal random assignment; $\Delta_2$: optimal targeted policy vs control; $\Delta_3$: optimal non-targeted policy vs control) on 6-week wage employment.

Results

Contrast $\Delta_1 = 0.002$ (95% CS: [0.000, 0.004]), showing negligible in-sample adaptive gain due to muted 6-week treatment effects. Contrast $\Delta_2 = 0.017$ (95% CS: [0.001, 0.034]), demonstrating a 35% gain in employment for the optimal targeted policy over control. Contrast $\Delta_3 = 0.006$ (95% CS: [-0.015, 0.027]), showing half the gain for an un-targeted policy.

Takeaway

Calculate formal welfare contrasts ($\Delta_1, \Delta_2, \Delta_3$) comparing observed adaptive allocations, optimal targeted policies, and non-targeted policies against baseline controls to isolate targeting gains from adaptive design gains.

Related findings

• Policymaker Targeting Efficiency Dilemma Under Stigma Mitigation →Economics (general)

  Adjusting a program's message to reduce social shame acts like renovating a low-income community center into a high-end fitness club. While it attracts motivated members who achieve high success rates, it unintentionally displaces the low-income families the facility was originally built to serve.

• Administrative Social Security Data Linkage for Earnings Measurement →Economics (general)

  Instead of asking individuals to remember and report exact yearly income numbers from a decade ago, researchers obtain signed consent to link survey profiles directly to official government tax records. This creates an unalterable, objective earnings ledger covering years before and after a financial event.

• Agnostic Machine Learning Heterogeneity Detection in Field Trials →Economics (general)

  Evaluating a program using only average results is like measuring the average temperature of a room with one frozen wall and one burning wall and concluding the room is comfortable. Machine learning acts like a thermal camera, splitting the room into sections to reveal the extreme cold and extreme heat hiding behind the average.

• Asymmetric Wealth Effect of Subsidized Index Contracts →Economics (general)

  Imagine buying a lottery ticket that costs $1, but pays out $1,000 if a major storm hits. Paying $1 during normal weather is so cheap that it does not change weekly spending or force extra work hours to make up for the cost.