Right to Work and Refugee Economic Integration: Experimental Evidence from Ethiopia

We will provide treated refugees with a comprehensive support package to help them apply for work permits or business licenses. This package includes administrative assistance to obtain required documentation, facilitation during application procedures, and payment of administrative fees associated with the application.

We will provide treated refugees with a comprehensive support package to help them apply for work permits or business licenses. This package includes administrative assistance to obtain required documentation, facilitation during application procedures, and payment of administrative fees associated with the application.

In preparation for the intervention design and delivery, we engaged in extensive coordination with the RRS at both the head and field office levels. This collaboration aimed to identify and address procedural bottlenecks in the issuance of work permits and business licenses for refugees. Through this process, we mapped the full administrative pathway for permit acquisition, identified inconsistencies across locations, and worked with RRS to standardize and streamline procedures.

We also conducted a diagnostic assessment of operational constraints, including gaps in equipment and staffing capacity at key administrative points. Based on these findings, we procured necessary equipment and provided logistical support to ensure smoother processing of applications during the intervention period.

The intervention package provided to treated individuals includes:
- Tailored administrative assistance to gather and complete required documentation (biometric ID, passport photos, support letter, tax ID).
- Direct facilitation during application procedures with RRS and the Ministry of Labor and Skills.
- Full coverage of administrative fees associated with the application process.

2025-08-01

2025-09-30

We have 4 primary outcome groups for individual-level outcomes: Labor Market Integration, Economic Outcomes, Mental Health and Subjective Well-being, Perceived Safety and Trust. See pre-analysis plan filed below for further details and explicit hypotheses for each of these outcome groups and their components.

For each outcome group, we will construct an index following Kling, Liebman, and Katz (2007). See pre-analysis plan filed below for further details.

We have two groups of secondary outcomes, measured at the household-level: economic outcomes and intra-household dynamics. See pre-analysis plan filed below for further details and explicit hypotheses for each of these outcome groups and their components.

Economic outcomes at the household-level will be similarly measured in an index, as described in the primary outcomes and the pre-analysis plan. For intra-household dynamics, we intend to look at hours worked in different paid and unpaid activities across working-aged members of the household. Please see the pre-analysis plan for specific hypotheses for margins of adjustment.

The study employs individual-level randomization across registered refugees from four refugee camps in Ethiopia, split evenly between treatment and control groups. Randomization will be stratified by camp, gender, and skill profile to ensure balance across characteristics that may influence outcomes.

We will examine heterogeneous treatment effects by gender, baseline mental health and trauma, skills-market match, and household position of the respondent, with particular attention to differential barriers and outcomes across these dimensions.

The study employs individual-level randomization across registered refugees from four refugee camps in Ethiopia, split evenly between treatment and control groups. Randomization will be stratified by camp, gender, and skill profile to ensure balance across characteristics that may influence outcomes.

Treatment includes comprehensive administrative support for obtaining work permits or business licenses, covering documentation assistance, facilitation of administrative processes, and payment of fees. Control group members receive only rights awareness information but retain the ability to pursue permits independently.

Our population of interest is recognized refugees living in four refugee camps across Ethiopia. Three camps are located in Benishangul-Gumuz region (Bambasi, Sherkole, and Tsore) and one camp is located in the Somali region (Sheder).

Baseline data collection is scheduled from May to July 2025. Randomization and implementation will follow immediately after baseline data collection. We will conduct twelve monthly high-frequency phone check-ins with selected participants and an in-person endline survey approximately October 2026.

We will examine heterogeneous treatment effects by gender, baseline mental health and trauma, skills-market match, and household position of the respondent, with particular attention to differential barriers and outcomes across these dimensions.

We aim to implement a randomization protocol that is acceptable to the Government of Ethiopia Refugees and Returnees Services (RRS) and local refugee representatives. Our preferred option is a computer-based randomization though we may have to resort to a public lottery at the household level if need be.

Our study employs a two-stage randomization design within each camp. First, we randomly select households from our eligible sample within each camp. Second, within each selected household, we randomly choose one eligible individual to receive the treatment offer. This two-stage approach ensures both statistical power and equitable distribution of opportunities across the refugee population.

The randomization proceeds as follows:
1. Household-level randomization: We randomly assign households to treatment or control groups.
2. Individual-level selection within households: Within treatment households, if multiple members are eligible (aged 18-65 and interested in work permits), we randomly select one individual to receive the support package.

We aim to stratify the treatment by gender of eligible individuals.

No

N/A -- not a cluster randomized design. Trial implemented across four different sites.

Target sample size of 8,000 households with 16,000 individuals (2 individuals surveyed per household) across all four sites.

Target sample size of 4,000 households receiving the treatment, 4,000 households in control (across all four sites)

In each camp location, we use a single-stage, individual-level randomization with two arms (treatment and control, equal allocation). We focus power calculations on an individual labor market outcomes (an indicator for whether respondents have been engaged in productive work in the past 7 days) and one household-level outcome (total annual aggregate food and non-food consumption expenditure per capita). We report power for each camp, assuming a total sample size of N=2,000 per location and an equal split between treatment and control. We note that power would be significantly larger when pooling all locations, so the detectable effect sizes below should be seen as conservative. We can detect a 0.073 (0.061) difference in the proportion of refugees in productive work in each camp, assuming compliance with the treatment of 75 percent (90 percent). For total consumption expenditure of the household per year, we could detect a 65.38 birr (54.49 birr) difference in each camp, assuming compliance with the treatment of 75 percent (90 percent). This represents 0.381% (0.318%) of the baseline mean.