Finding and Lifting Up the Poor and Vulnerable: A Field Experiment on Targeting Mechanisms and Effects of a Public Works Program in Rural DRC

How to cite

Mani, Subha; Mvukiyehe, Eric. 2021. "Finding and Lifting Up the Poor and Vulnerable: A Field Experiment on Targeting Mechanisms and Effects of a Public Works Program in Rural DRC." AEA RCT Registry. AEARCTR-0007874. https://doi.org/10.1257/rct.7874-1.0

Download: BibTeX · RIS — for reference managers without a browser translator. Zotero and most others pick up this page's citation metadata directly.

Subha Mani* , Eric Mvukiyehe

* Primary investigator

Registry ID
AEARCTR-0007874
DOI
https://doi.org/10.1257/rct.7874-1.0
First registered
Last updated
Status
On going
Trial dates
2016-06-15 – 2022-12-31
Countries
Private
JEL codes
I3, D14, J24, C93, H41
Registry record
http://www.socialscienceregistry.org/trials/7874

Abstract

Targeting social protection programs, including public works has been a central issue for resource strapped developing countries facing unreliable data on target beneficiaries’ socioeconomic status. Impacts of public works program is further complicated by selective targeting. We design a unique two-stage randomized control trial and use experimental variation in beneficiary selection procedures to examine two primary research questions of interest. First, how effective are the targeting mechanisms (chief selection, community selection, and public lottery) in ensuring poor and vulnerable households’ participation in public works programs? Second, what is the impact of having access to public works program (through public lottery) on individual labor market outcomes and household welfare? We will exploit experimental variation in both program targeting and access to public works programs to produce rigorous impact estimates on both material and non-material outcomes. In the registered pre-analysis plan we provide a detailed summary of the key chosen outcomes and associated variable definitions to be used for computing the intent-to-treat effects, local average treatment effects, and sub-group average treatment effects, accounting for both Type I and Type II error in statistical analysis. We also present our plans for handling missing values, outliers and measurement error in data.

Intervention

Between 2017 and 2020, selected rural areas in eastern DRC were targeted to receive public works projects through community selection, public lottery, or chief selection procedures. A total of approximately 10,000 beneficiaries received cash-for-work opportunities through participation in local public works projects. The key public works activities include garbage collection, street cleaning and road maintenance. Beneficiaries received $3 a day (minimum wages in eastern DRC) for working five days a week for four months (75 to 100 working days). Priority was given to women, youth at risk, former combatants, and displaced or returned youth.

Experimental design

This paper is built to answer the following primary and secondary research questions of interest :
(1) Primary and secondary research questions related to program targeting:
Primary research question (Q1a):
How effective are the different targeting mechanisms (chief selection, community selection, and public lottery) in ensuring poor and vulnerable households’ participation in public works programs?
Secondary research question (Q1b):
What is the impact of the different targeting mechanisms (chief selection, community selection, and public lottery) on individual (e.g., labor market outcomes and income) and household outcomes (e.g., consumption expenditure, debt, savings)?
(2) Primary and secondary research questions related to access to public works projects:
Primary research question (Q2a): What is the impact of public works projects on individual and household material (e.g., employment, earnings, consumption) outcomes?
Secondary research question (Q2b): What are the non-material impacts (e.g., mental health, social cohesion), distributional impacts and unintended consequences of public works programs?

Design details

Randomization method
Cluster level randomization was conducted remotely in office on a computer (by the DIME team at the World Bank) whereas the second stage randomization in the public lottery villages was conducted via public lottery in these villages
Randomization unit
Two levels and units of randomization - village level randomization for targeting and individual level randomization for roll out of PWPs.
Clusters
endline surveys to take place in 250 villages
Observations
cover approx. 4850 individuals
Arms
We plan to target 4,850 respondents for the detailed household survey across the 250 villages. The distribution of target respondents across the three targeting mechanisms is described below: o 2,400 respondents from the public lottery arm (20 surveys in each of the 120 villages in the public lottery arm) o 975 respondents from the chief selection arm (about 12-13 surveys in each of the 65 villages assigned to the chief selection arm) o 975 respondents from the community selection (about 12-13 surveys in each of the 65 villages)
Primary outcomes
(1) individual labor supply, (2) household earnings, (3) consumption expenditure, (4) assets, and (5) poverty. (1)-(4) are relevant for addressing research questions Q1b and Q2a. (5) is relevant for addressing research question Q1a.
Secondary outcomes
Employment and earnings for other household members, financial inclusion index, human capital/skills accumulation, job-search behavior, psychosocial wellbeing, community and social capital, civic engagement, political participation and social cohesion Indices, conflicts and violence, victimization, women's empowerment, physical disability, limitations and illness, and migration. These are relevant for addressing research question Q2b. See attached pre-analysis plan for variable definitions.

Cited by

No citation data available for this registration yet.

Keywords