Social registry targeting in Mauritania

We partnered with the government and the World Bank to study the improvements introduced by an updating of the social registry. Mauritania was one of the first countries of the Sahel region to implement a social registry and to reach national coverage by 2020. The social registry includes 28% of the population and is used by more than 20 programs, including its national food and cash transfer programs. The government has started to update and expand the social registry in 2023, which should be completed by the end of 2024. In doing so, the government introduced a series of important improvements of the social registry process:
- While the identification of beneficiaries was based on community-based targeting (CBT), it has now moved largely to a system based on scoring combined with community selection.
- The number of beneficiaries per village (quota) is now determined by the actual score of households, while it was previously based on imprecise poverty estimates at the commune level.

2024-12-22

2025-02-28

We are focusing on outcomes of targeting efficiency and legitimacy.
For targeting efficiency, we will focus on targeting errors and poverty reduction simulations based on the following primary outcomes:
- Per capita consumption
- Food security (FCS)
- Multidimensional poverty
For legitimacy, we will measure the satisfaction with the targeting process and outcomes.

We will consider a wide range of secondary outcomes to understand the determinants of targeting (especially community selection). These outcomes include:
- connections (e.g. elite capture)
- socio-demographic variables
- sources of income (including agricultural and non-agricultural activities)
- assets
- mental health
- psycho-socio measures

This research will assess whether these improvements lead to changes in targeting in terms of efficiency, legitimacy and impacts. We will test three main hypotheses:
a. The union of scoring (60 to 80%) and CBT (20 to 40%) reduces exclusion errors compared to traditional hybrid methods that selects the intersection of PMT and CBT, while improving the legitimacy of the targeting scheme.
b. Targeting thresholds have important effects on targeting outcomes, with higher selection rates leading to lower levels of targeting errors and higher legitimacy.
c. The two previous improvements translate into larger impacts of social programs that rely on the improved targeting mechanism of the updated version of the social registry.
To test the efficiency, legitimacy and impacts of these innovations of the social registry, we will conduct a randomized control trial (RCT) comparing different targeting mechanisms. Specifically, we will test the improvement in terms of selection quota by randomizing this innovation at the commune level. Then, at the locality (village) level, we will randomly allocate targeting as a scoring mechanism, a community-based targeting method, or a combination of both (with the union approach).

Not available

Randomization is conducted on a computer by the research team.

Randomization is conducted at the level of the localité (equivalent to the village in Mauritania).

Yes

We randomized 330 localités (villages).

We plan to collect data among 3960 observations (12 households per localité).

We have 55 localité per treatment arm (330 localités randomly allocated to 6 treatment arms of equal size).

We conducted power calculations based on nationally representative data (2019 LSMS data) and using similar studies (such as Premand & Schnitzer, 2021) for calibrating Minimal Detectable Effects. These calculations indicate a required sample size of approximatively 4000 households for the baseline survey. With this sample size, we will be able to detect a difference of 0.2 sd in per capita consumption between two treatment arms.