Impacts of Targeted Covid-19 Cash Transfers in Togo

How to cite

Bellue, Suzanne; Blumenstock, Joshua; Aiken, Emily; Udry, Christopher; Karlan, Dean. 2021. "Impacts of Targeted Covid-19 Cash Transfers in Togo." AEA RCT Registry. AEARCTR-0007590. https://doi.org/10.1257/rct.7590-1.1

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

Suzanne Bellue* , Joshua Blumenstock , Emily Aiken , Christopher Udry , Dean Karlan

* Primary investigator

Registry ID
AEARCTR-0007590
DOI
https://doi.org/10.1257/rct.7590-1.1
First registered
Last updated
Status
On going
Trial dates
2020-11-01 – 2021-07-01
Countries
Togo ()
Registry record
http://www.socialscienceregistry.org/trials/7590

Abstract

In response to COVID-19, a third of social protection measures have taken the form of cash transfers reaching more than 1.1 billion people --- a 240% increase in coverage from pre-COVID levels. In the aftermath of the COVID-19 pandemic, direct cash transfers are an effective tool to protect vulnerable households. We conduct a randomized controlled trial (RCT) of a targeted cash transfer program implemented in rural Togo between November 2020 and May 2021. In collaboration with GiveDirectly, the government of Togo secured sufficient funding to provide benefits to roughly 57,000 of the approximately 580,000 citizens living in the poorest 100 cantons. Using mobile phone and satellite data, we identified the poorest cantons and poorest people living in them. We randomized the beneficiaries among the poorest phone owners. After registration, every month and for five months, eligible women receive a cash transfer of 8,620 F CFA ($15.5 US) and eligible men, a transfer of 7,450 F CFA ($13.5 US). We conduct a telephone survey at the end of the intervention to measure a wide range of outcomes, including consumption, food security, labor supply, access to health care, education, psychological well-being, financial inclusion and the perception of poverty. We also have access to administrative data of mobile phone companies in Togo, which will allow us to exploit phone usage behaviors and build other types of outcomes, such as adoption and use of the mobile money services, migration or predicted poverty.

Intervention

In March 2020, and over the course of just a few weeks, the Government of Togo built and deployed ``Novissi,'' a contactless, digital system that provided over half a million individuals with cash. Beneficiaries registered using their mobile phones; after entering basic information into a USSD menu, they were then immediately sent mobile money transfers of approximately $20/month, lasting for three months. This program was the first of its kind, described as an “exemplary case of social protection in response to the coronavirus pandemic in Africa.”

In an expansion phase, the government decided to expand Novissi to rural areas of the country, where extreme poverty is most severe. However, the government did not have a comprehensive social registry that would allow them to directly identify and prioritize its poorest people. Using rich data from satellites, a nationally-representative household survey and the deep learning pipeline described in Chi et al. (2021), we first identified the poorest 100 cantons (admin-3 regions) in the country. In a second step, building on recent advances in the research literature (Blumenstock et al., 2015; Aiken et al., 2020), we developed methods that use mobile phone metadata to help identify the individuals with the greatest need.

Individuals must self-register for Novissi using a mobile phone. When an individual registers, they provide a valid voter ID and phone number. In the Novissi expansion that we study, eligibility is restricted to individuals who have registered to vote in one of the 100 poorest cantons in Togo. In addition, individuals must have a predicted daily consumption of less than $1.29, where the prediction is calculated by applying machine learning to mobile phone data.

Experimental design

Individuals must self-register for Novissi using a mobile phone. When an individual registers, they provide a valid voter ID and phone number. In the Novissi expansion that we study, eligibility is restricted to individuals who have registered to vote in one of the 100 poorest cantons in Togo. In addition, individuals must have a predicted daily consumption of less than $1.29, where the prediction is calculated by applying machine learning to mobile phone data. We define the ``eligible" population as those individuals who successfully register for Novissi between November 1st 2020 and January 31 2021, and who meet both of the above criteria. There are an estimated 50,000 such eligible individuals. In addition, about 43,000 individuals signed up using a newly purchased SIM card.

Eligible individuals are then randomly assigned to one of two conditions: The `Treated' group begin to receive Novissi benefits immediately after registration. Cash is transferred every 30 days for 5 months, starting on the day of registration. All `Control' individuals who registered with an existing SIM card will receive delayed payments, end of May 2021. A random fraction of `Control' individuals who registered with a newly purchased SIM card will receive those delayed payments. 56% of eligible individuals (with an old-SIM card) assigned to Treatment; the remainder to Control. In addition, 16% of new SIM cards are assigned to Treatment.

Design details

Randomization method
The sample is stratified by type of SIM cards (new or old) and whether or not the individuals are registered to vote in the Savanes region. The randomization is done by a computer.
Randomization unit
Individuals are assigned to either the treatment (cash transfers starting in November) or the control group.
Clusters
There is no cluster.
Observations
An endline survey is being conducting beginning in mid-April 2021, roughly five months after the start of the registration period, and will last approximately three weeks. We expect to complete phone surveys with approximately 12,500 people.
Arms
There are two treatment arms: Treatment and Control. Among registered individuals, 34,940 are in the Treatment group and 57,597 are in the Control group. We plan to survey 6,050 individuals of the Control group and 6,450 individuals of the treatment group.
Primary outcomes
Food consumption and food security Financial stability Financial inclusion Adoption and use of mobile money services
Secondary outcomes
Individual labor supply Health care access Psychological well being Self-perception of poverty Predicted poverty (from admin data) Registration to Savanes-Novissi program Mobility and migration

Cited by

No citation data available for this registration yet.

Keywords