Public works programs, well-being, social cohesion and women’s empowerment: Evidence from a Randomized Control Trial in Egypt

How to cite

Mvukiyehe, Eric; Cottin, Raphael. 2019. "Public works programs, well-being, social cohesion and women’s empowerment: Evidence from a Randomized Control Trial in Egypt." AEA RCT Registry. AEARCTR-0005179. https://doi.org/10.1257/rct.5179-1.0

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

Eric Mvukiyehe* , Raphael Cottin

* Primary investigator

Registry ID
AEARCTR-0005179
DOI
https://doi.org/10.1257/rct.5179-1.0
First registered
Last updated
Status
On going
Trial dates
2015-08-15 – 2020-12-31
Countries
Private
JEL codes
D13, I38, J12, 012.
Registry record
http://www.socialscienceregistry.org/trials/5179

Abstract

Public work programs (PWPs) constitute a popular form of providing social protection for households in low- in middle-income countries, and middle-income countries. While the growing number of empirical studies has focused on assessing the impact of PWPs on material outcomes (e.g., labor market; economic welfare; etc.), rigorous evidence about their impact on non-material outcomes is scarce.

This paper analyses a large-scale randomized control trial (RCT) of a PWP conducted in Egypt between 2015 and 2017 to provide rigorous evidence on three sets of issues: (ii) non-material aspects of the community’s life such as social capital, trust collective action and violence; (ii) subjective well-being; and (iii) women’s empowerment and intimate partner violence. One defining characteristic of this PWP is that the jobs created were supposed to be “good jobs”, providing decent working conditions and rewards, and that these jobs were predominantly targeted towards women and people of low social status. The program was randomized across villages and within villages, allowing us to measure both its individual and spillover effects.

We exploit a rich survey on non-material outcomes to assess the impact of the ELIIP PWP at the individual level as well as at the community level. At the individual level, we evaluate the impact of ELIIP on the perceptions of economic conditions, income security, subjective well-being, psychological health, and gender norms. At the community level, we examine whether the program has been deemed as “fair” by the community at large, and whether it has led to an improvement in the perceptions of the economic conditions, social capital, trust, conflict in the community, and confidence in government.

Intervention

The Community Social Services component of the ELIIP evaluated in this IE focuses on social services and youth employment activities that are fostered through grants to non-governmental and/or community-based organizations that employ youth, especially females, to provide social services such as cleanliness, maternal health and environmental awareness campaigns in local communities. A distinguishing feature is that sub-projects are lasting relatively long between 12-18 months and thus provide employment and security for a longer period. Further, sub-projects are required to be labor-intensive: at least 60% of project costs must be on labor. Other criteria are that 80% are between 18 and 29 years old, at least 70% is female, and the beneficiaries should be considered the “poorest of the poor” within their community. The projects are implemented through NGOs, with which the SFD has worked with in the past. To give an example: community health care projects will create job opportunities for girls from the age of 19, who will be trained to provide health education programs and administer home visits to expand access to women, thus contributing to improved maternal and child health. Other NGO projects include: (a) cleanliness and environmental awareness campaigns; (b) early childhood education; (c) mother and child health awareness home visit programs; (d) illiteracy eradication activities; and (e) youth engagement in community initiatives in rural and urban areas, among others.

Experimental design

Our impact evaluation is designed to shed light on these questions through the use of randomization at two levels: the village and the individual level. The villages were randomly allocated into treatment and control groups, using two different methods (see PAP for details). In the end, we retain 78 were villages in which projects were signed and implemented, against 86 control villages. In addition to that, there are 70 “overflow” villages, in which project were implemented without randomization at the village level.

In each treated village, NGOs were asked to provide worker lists of twice the number of workers needed in order to enable randomization at the worker level. Worker lists provide detail on the name, gender, age, national ID, type of work, telephone number, residence and official residence. Workers are only excluded from the list if the village registered on their National ID is a control village. We have made the distinction on the workers’ list registration form between where someone lives and the residence information listed on their National ID, since for many people there is a discrepancy between this information.

The double randomization at the village and individual level will allow for the identification of direct effects on program beneficiaries (including consumption, assets, labor market outcomes, and human capital accumulation amongst others) as well as general equilibrium changes in local economic activity. The project activities taking place in overflow villages as selected by NGOs will still allow for randomization at the worker level.

Design details

Randomization method
We carried out a single round of data collection for individual level outcomes through a survey instrument upon completion of the project. In addition to data collection for individuals, we also carried out a community level survey interviewing local community leaders. For the community level survey component we interviewed two local community leaders (the official/ traditional leader and a secondary leader) in all 234 villages. For the household-level survey, data collection involved surveying households in both treatment and control communities as well program participants and non-participants. There are three distinct samples: the first, a sample of program participating individuals and (randomly selected) non-participating individuals in “treated” communities (about 15 individuals per village in all treated villages). The second is a synthetic control sample of individuals in control communities who have the similar characteristics of the program participating individuals in “treated” communities (about 5 household per village in all control villages). The third will be a random representative sample of non-participants across treatment and control villages (about 5 households per village in all 234 villages in the study).
Randomization unit
Unit of randomization is at the individual and village level.
Clusters
234 villages.
Observations
2348 individuals (1308 treatment and 1041 controls). 1170 households (740 treatment, 430 control).
Arms
The trial only has one treatment, which is assigned both at the village-level and at the individual level.
Primary outcomes
1. subjective well-being index 2. mental health index 3. perception of economic conditions index (individual level) 4. income security index 5. fairness of allocation process index 6. Perception of economic conditions index (household level) 7. social capital index 8. citizenship and confidence in government index 9. victimization index 10. Attitude towards marriage index 11. Attitude towards fertility index 12. Women decision making index 13. Women control over household resources index 14. Intimate partner violence index 15. Violence against women index
Secondary outcomes
1. Perception of economic conditions: o Economic conditions of household compared to previous year o “poverty ladder” question 2. Income security: o How long are you able to pay your expenses with your savings o Comfortable paying for unexpected or emergency expenses 3. Fairness of allocation process: o Conflict about program about how workers were chosen o Satisfied with way beneficiaries identified and served o Efficiency of NGO implementation of SFD program o Poverty of participants in SFD program o How hardworking are beneficiary households o Presence of inclusion errors o Presence of exclusion errors o Fairness of selection process 4. Social capital o Acceptance by other households o Membership in village committees o Social divisions in village o Collective action in village o Involvement in community initiatives o Trust

Cited by

No citation data available for this registration yet.

Keywords