Impact Evaluation of Delivering Support as Cash versus Mobile Money in South Sudan

How to cite

Christian, Paul; Jospe, Eric; Manea, Roxana Elena; Heirman, Jonas; Waidler, Jennifer; López, Agustina. 2026. "Impact Evaluation of Delivering Support as Cash versus Mobile Money in South Sudan." AEA RCT Registry. AEARCTR-0018256. https://doi.org/10.1257/rct.18256-2.0

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

Paul Christian* , Eric Jospe , Roxana Elena Manea , Jonas Heirman , Jennifer Waidler , Agustina López

* Primary investigator

Registry ID
AEARCTR-0018256
DOI
https://doi.org/10.1257/rct.18256-2.0
First registered
Last updated
Status
In development
Trial dates
2026-04-27 – 2027-12-31
Countries
South Sudan ()
JEL codes
O1, I38, O33
Registry record
http://www.socialscienceregistry.org/trials/18256

Abstract

In South Sudan, the delivery of assistance is shaped by constraints on infrastructure, access to technology, and financial literacy. This study evaluates the comparative effectiveness of delivering transfers via mobile money relative to traditional cash transfers in a low-connectivity, fragile setting. While digital transfers have the potential to reduce transaction costs, improve the safety of delivery, offer recipients greater flexibility in managing their resources, and expand financial inclusion, limited access to mobile networks, low penetration of mobile phones, and poor mobile vendor infrastructure may constrain their impact.

Using an experimental design, we compare outcomes across delivery modalities, focusing on take-up and use of mobile money, welfare, financial inclusion, and social capital. We further examine how take-up and impacts vary with exposure to mobile money infrastructure and transaction costs.

The results provide evidence on the feasibility and effectiveness of digital transfer systems in environments with weak infrastructure and can inform the design of assistance programs in fragile states.

Intervention

This impact evaluation will test two alternative methods to deliver support, cash and mobile money, in the context of two WFP programmes, social safety nets (SSN) and asset creation and livelihoods (ACL), in South Sudan.

SSN consists of 8 months of unconditional transfers, while ACL is relatively smaller and amounts to 3 months of conditional support. The sample for the impact evaluation includes mostly ACL villages (approximately 90%).

The ACL programme makes transfers to households against their commitment to work on creating or rehabilitating communal assets. The goal is to improve food security, restore livelihoods, while contributing to peace. The assets created by the programme include land restoration, water management solutions, feeder road construction or rehabilitation, cultivation of agricultural land, and the creation of infrastructure for livelihood diversification. The SSN programme, too, aims to help households meet their basic needs and improve resilience. The programme targets eligible urban and peri-urban areas based on their relative poverty and food insecurity.

WFP’s Country Office in South Sudan has traditionally been delivering support as either cash or in-kind (e.g., cereals and pulses), the modality being informed by county-level market assessments. Moving forward, however, the Country Office aims to switch to mobile money where cash is a feasible delivery method. A significant expansion is thus planned for 2026, with further scale-up efforts thereafter until the switch to mobile money is complete where possible.

Experimental design

We initially planned to implement a two-arm cluster-randomized controlled trial in approximately 126 villages. Villages were randomly assigned, in equal proportions (and stratified at the payam level) to receive either cash transfers or mobile money transfers. The village-level randomization reflects the operational scale of program delivery and mitigates concerns about within-community spillovers. Treatment effects are to be estimated by comparing mean outcomes across arms, with inference adjusted for intra-cluster correlation.

As the mobile money scale-up depended heavily on the network operator, the RCT was ultimately implemented in the strata (payams) where the mobile money expansion was possible. Out of the initially planned sample, 5 strata were dropped because the network coverage proved insufficient for registration and the use of mobile money. As the treatment arm assignment was stratified at the payam level, the sample has remained balanced: 41 villages continuing to receive Cash and 41 villages switching to Mobile Money.

Data collection schedule: Ultra-brief baseline (less than 20 questions) in May 2026, follow-up round 1 in September 2026, and follow-up round 2 in December 2026.

Estimation specifications:
Y = cst + a MoMo + d Strata + e (round-based)
Y = cst + a MoMo + b Round2 + c MoMo x Round2 + d Strata + f Strata x Round2 + e (pooled)

Alternative specification where baseline values of Y are available (e.g., HHS, phone use, MoMo use):
Y = cst + a MoMo + b Round2 + c MoMo x Round2 + d Strata + f Strata x Round2 + g BaseChar + e

Design details

Randomization method
Randomization done in office by a computer.
Randomization unit
The randomization is conducted at the village level, i.e., all households within a village will benefit from the same transfer modality.
Clusters
Our research will include approximately 82 clusters. The clusters are villages across three counties and one municipality in two different states in South Sudan. The clustered randomization was stratified at the payam level, which is two levels above the village and one level below the county.
Observations
There will be two waves of data collection with a targeted number of 15-20 surveys per village in each round, which leads to a total of approximately 1,650 household observations per round.
Arms
The sample will be divided equally between the two arms: approximately 41 villages will receive their support as cash-in-hand, and 41 villages will switch to mobile money.
Primary outcomes
1. Primary outcomes collected to compile evidence on whether the distribution modalities were correctly rolled out: a. Receipt of transfers b. Cash receipt in the cash arm of the experiment c. Mobile money (MoMo) receipt in the MoMo arm of the experiment d. Phone receipt in the MoMo arm of the experiment 2. Primary outcomes collected to compile suggestive evidence of experimental success: a. Phone use (last 7 days) b. MoMo use (ever and last 30 days) 3. Primary outcomes collected to compile upstream evidence of differences between the two modalities: a. Fees and cost of accessing transfers (South Sudanese Pounds SSP) b. Time spent collecting cash (including walking to, waiting, and processing) (Minutes) c. Amount received in the last 90 days as sharing and supporting (SSP) d. Amount sent in the last 90 days to support others (SSP) e. Amount received in the last 90 days as formal borrowing (e.g., VSLA; SSP) f. Fearful of theft and/or harassment as they relate to the receipt of/accessing transfers (binary) g. Capacity to meet urgent needs in the last 30 days (binary) h. Capacity to obtain cash from transfers/programme participation when it was timely (binary)
Secondary outcomes
4. Other exploratory outcomes: a. # of trips to cash out transfers b. # of trips to the market (last 30 days) and average amount spent on most recent trips c. Speed of spending transfers d. Transfer spending preferences (SSP amount for various categories) e. Yes/No to saving and amount saved (last 90 days) f. Food security as measured by FIES and HHS (last 30 days) g. Female decision power (e.g., her opinion matters at least equally with that of male household members on 4 broad household decision matters: food, non-food, transfer use, and expenses related to children) h. Trust in various institutions: WFP and its cooperating NGOs, mobile service providers and mobile money, financial intermediaries, and one’s community i. Feelings of safety as they relate to the receipt of/accessing transfers (binary for any concerns, sum of concerns such as theft, harassment, long queues) j. PHQ4 (last 14 days) 5. Furthermore, we will assess (where relevant): a. Whether the gender, educational level or literacy of the recipient/household head are sources of impact heterogeneity b. Whether having a migrant in the family is a source of impact heterogeneity c. Whether MoMo knowledge/understanding and openness towards the technology (i.e., willingness to receive a debt payment via MoMo) mediates MoMo use d. Whether experiencing shocks (and their nature) in the last 90 days is a source of heterogeneity explaining impact differentials, especially as they regard sharing practices, borrowing, and PHQ4

Cited by

No citation data available for this registration yet.

Keywords