Beyond Box-Ticking: Bureaucrat-Firm Relationships and Policy Success

Last registered on January 06, 2026
View Trial History

Pre-Trial

Trial Information

General Information

Title
Beyond Box-Ticking: Bureaucrat-Firm Relationships and Policy Success
RCT ID
AEARCTR-0017482
Initial registration date
December 28, 2025

Initial registration date is when the trial was registered.

It corresponds to when the registration was submitted to the Registry to be reviewed for publication.

First published
January 06, 2026, 7:07 AM EST

First published corresponds to when the trial was first made public on the Registry after being reviewed.

Locations

There is information in this trial unavailable to the public. Use the button below to request access.

Request Information

Primary Investigator

Name
David Wu
Affiliation
Georgia State University
Contact Primary Investigator

Other Primary Investigator(s)

PI Name
Philipp Barteska
PI Affiliation
The University of Hong Kong
Contact Investigator
PI Name
Ritwika Sen
PI Affiliation
Ludwig-Maximilian University of Munich
Contact Investigator

Additional Trial Information

Status
In development
Start date
2025-11-10
End date
2026-12-31
Keywords
Firms & Productivity, Governance
Additional Keywords
JEL code(s)
O12, D73, M50, J23
Secondary IDs
Prior work
This trial does not extend or rely on any prior RCTs.
Abstract
Many active labor market policies in low- and middle-income countries struggle to reduce youth unemployment (McKenzie 2017; Card et al. 2018). A growing literature suggests that the challenge may lie not only in policy design, but also in how policies are implemented by street-level bureaucrats (Besley et al. 2022; Bandiera et al. 2021). In this project, we focus on the performance of bureaucrats tasked with delivering these programs and the potential importance of misaligned incentives. Bureaucrats are frequently evaluated on easily observable, “box-ticking” outputs—such as the number of vacancies collected—rather than on harder-to-monitor outcomes like sustained job placements. This misalignment may skew effort toward the easily observable parts of the job. We study this question in Addis Ababa (Ethiopia), where the Ministry of Labor and Skills operates a large-scale labor market matching program. Local district-level (“woreda”) officials collect vacancy information from firms and recommend registered job seekers, yet program effectiveness has been limited. We ask: can improved measurement of bureaucrat tasks strengthen the effectiveness of an active labor market program? We work with around 110 woredas, surveying around 450 bureaucrats and 1760 firms. We will conduct a cluster-randomized controlled trial at the woreda level. For a random subset of woredas, bureaucrats will be provided with two additional follow-up checklists—one to record referrals of job seekers and another to document actual placements. These tools will be designed to allow measurement of the respective tasks and hence allow for performance evaluation in line with sustained job placements. We will compare the impacts on policy implementation and endline firm outcomes between treated and control woredas.
External Link(s)

Registration Citation

Citation
Barteska, Philipp, Ritwika Sen and David Wu. 2026. "Beyond Box-Ticking: Bureaucrat-Firm Relationships and Policy Success." AEA RCT Registry. January 06. https://doi.org/10.1257/rct.17482-1.0
Sponsors & Partners

There is information in this trial unavailable to the public. Use the button below to request access.

Request Information
Experimental Details

Interventions

Intervention(s)
We collaborate with most woredas in Addis Ababa, Ethiopia. Each woreda Office of Labor and Skills implements a standardized labor market matching policy mandated by the Ministry of Labor and Skills. The policy aims to identify firm vacancies and unemployed jobseekers and to facilitate matches between the two. We focus on three core program components: (1) collecting vacancy information from firms, (2) referring jobseekers to these vacancies, and (3) verifying whether referred jobseekers are ultimately hired by the firms.

We will provide a digitized version of an existing checklist to all woredas – control and treatment. This digital checklist is focused on component (1): collecting vacancies from firms. In the status quo, bureaucrats fill out a pen-and-paper version of this checklist, which they later have to copy into excel for reporting purposes. The digital checklist simplifies this as data entered digitally is immediately exportable into an excel sheet that follows the existing format and reporting requirements. Both team leaders and ground-level bureaucrats (“experts”) have a personal login to the dashboard that highlights key information and allows for downloading of the excel sheet.

For a random subset of woredas, bureaucrats will also be provided with two new (digital) checklists. These checklists are focused on steps that are crucial to the policy’s goal, but which are not measured consistently. In the treated woredas, team leaders and experts will have access to an upgraded dashboard that allows tracking (2) which vacancies have jobseekers referred to them and (3) whether the referred jobseekers end up working at the firm.
Intervention Start Date
2025-12-29
Intervention End Date
2026-05-31

Primary Outcomes

Primary Outcomes (end points)
The primary outcomes for this study will be measured under three key domains:

0. Take Up of Digital Checklist by Bureaucrats
- Adoption of digital checklists: (i) any checklist submitted by a woreda? (ii) number of checklists submitted; (iii) share of assigned firms for which a (relevant) checklist was completed.
Note: Vacancy checklist for both control and treatment woredas; Referral and Follow-up checklist only for treatment woredas.

1. Policy Implementation Metrics Focused on Firms:
- Firm-reported frequency of site visits by woreda officials
- Share of firms that post at least one vacancy through the woreda
- Share of firms with vacancies where >0 jobseekers recommended by woreda
- Share of vacancy types where >0 jobseekers recommended by woreda

2. Firms’ willingness to take up the policy:
- Firms’ willingness to hire (suitable) jobseekers from the woreda in the future, including wage offers for job seekers
- Firms’ perceptions of the woreda office including: levels of trust in bureaucrats, belief that woreda office will find suitable jobseekers, and satisfaction with prior interactions.
Primary Outcomes (explanation)

Secondary Outcomes

Secondary Outcomes (end points)
The secondary outcomes for this study will be measured under two key domains:

1. Downstream firm outcomes
- Share of firms where >0 woreda-recommended jobseekers start working
- Share of firms with vacancies where >0 woreda-recommended jobseekers start working
- Share of vacancy types where >0 woreda-recommended jobseekers start working

2. Policy Implementation Metrics Focused on the Organization:

- Embeddedness measures: (i) share of time spent interacting with firms; (ii) number of firms whose phone number a bureaucrat can name;

- Bureaucrat beliefs: (i) how many jobs are created per vacancy collected?; (ii) how many vacancies need to be collected to create n jobs?

- Goals/Evaluation content: (i) is one of the top 3 KPIs related to post-vacancy collection tasks?

- Adherence with pre-specified policy guidelines (0/1) linked to: vacancy collection, job seeker referrals and follow up.
Secondary Outcomes (explanation)

Experimental Design

Experimental Design
We collaborate with most woredas in Addis Ababa, Ethiopia.
For a random subset of woredas, bureaucrats will also be provided with two new checklists. Team leaders from the treated woredas will be able to monitor the progress and measure bureaucrats’ performance for both vacancy collection and the subsequent follow-ups.
After the intervention, we will conduct an endline survey of bureaucrats and firms involved in the intervention, to compare the impacts on policy implementation and outcomes between treated and control woredas.
Experimental Design Details
Not available
Randomization Method
Randomization done in office by a computer using the software Stata.
Randomization Unit
The unit of randomization is the woreda (local district level in Addis Ababa)
Was the treatment clustered?
Yes

Experiment Characteristics

Sample size: planned number of clusters
Target: 110 woredas

Sample size: planned number of observations
Target: 110 woredas, 450 bureaucrats, 1760 firms
Sample size (or number of clusters) by treatment arms
Target: 55 clusters (880 firms over two waves) control; 55 clusters (880 firms over two waves) treated
Minimum detectable effect size for main outcomes (accounting for sample design and clustering)
Supporting Documents and Materials
IRB

Institutional Review Boards (IRBs)

IRB Name
The Ethics Commission, Department of Economics, University of Munich
IRB Approval Date
2025-10-15
IRB Approval Number
2025-12
Analysis Plan