Improving Women’s Access to Digital Mobile Technology in Sub-Saharan Africa

How to cite

Schaner, Simone; Pande, Rohini; Riley, Emma; Barasa, Laura Nelima; Jorgensen, Erik. 2025. "Improving Women’s Access to Digital Mobile Technology in Sub-Saharan Africa." AEA RCT Registry. AEARCTR-0016908. https://doi.org/10.1257/rct.16908-4.0

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

Simone Schaner* , Rohini Pande , Emma Riley , Laura Nelima Barasa , Erik Jorgensen

* Primary investigator

Registry ID
AEARCTR-0016908
DOI
https://doi.org/10.1257/rct.16908-4.0
First registered
Last updated
Status
In development
Trial dates
2025-09-30 – 2027-08-31
Countries
Kenya ()
JEL codes
O33, D14, C93
Registry record
http://www.socialscienceregistry.org/trials/16908

Abstract

Mobile phones can enable economic and social empowerment by lowering communication costs and reducing information frictions. Yet, gender gaps in smartphone ownership in sub-Saharan Africa are substantial. The high up-front cost of smartphone devices remains a primary barrier, especially for women. Credit is one way to overcome this barrier, but it is often difficult for lenders to extend credit to low-income borrowers, who lack collateral and therefore pose a risk to default on their loans. Recent research indicates that a “pay-as-you-go” (PAYG) structure -- which relies on technology that locks the asset being lent in the event of non-payment -- could be an effective means of lending to the large numbers in low- and middle-income countries with no credit history. However, the rigid nature of these contracts may pose problems, especially when loans finance income-generating assets like smartphones, and could be especially challenging for groups earning irregular income that is subject to shocks. Working with a large asset-financing platform providing PAYG smartphones to individuals in Kenya, we conduct two randomized controlled trials to assess (a) whether offering payment flexibility enhances digital inclusion and raises economic wellbeing and (b) how marketing under these flexible contracts affects selection. In the first experiment, we randomly surprise existing customers into one of 3 experimental contracts offering different types of flexibility and repayment incentive; this lets us estimate the impact of the new contracts holding selection constant. In the second experiment, we assess selection by comparing borrower demographics and repayment outcomes among those surprised with the experimental contracts versus those marketed the contracts. Findings will inform both product design for financial service providers and consumer protection policies amid the rapid growth of digital credit markets in Sub-Saharan Africa.

Intervention

We will conduct two randomized controlled trials in Kenya to understand how payment flexibility affects borrower repayment, financial stress, smartphone use and livelihoods, and how marketing under these contracts affects adverse selection.

Under the partner’s standard lending model, customers receive a smartphone after making a down payment. They must then make 365 daily payments to clear their loan. The partner remotely locks the phone on days that a payment is due but the customer does not pay. Working with our partner organization, we will study the impact of three contracting variations designed to ease borrower liquidity constraints:
- Under “fixed flex”, customers receive a non-renewable bank of flex days, which can each be redeemed for a day of phone usage on a day the customer cannot repay their loan.
- Under “earned flex”, customers accrue one flex day for every 10 days they repay and, once earned, may redeem their days whenever they choose throughout their contract.
- Under “earned rewards”, customers accrue a cash reward equal to one daily payment for every 10 days they repay. Once earned, customers may redeem their rewards whenever they like.

Experimental design

Our study includes two experiments. In the first experiment, we will individually randomize 3,764 new smartphone customers into one of four arms (941 in each), where they will either remain with the status quo contract or be “surprised” with one of the three contracts discussed above (fixed flex, earned flex, earned rewards; S1-S3). These arms allow us to isolate the behavioral effect of each contract, holding constant selection.

The second experiment involves two layers of randomization. First, we worked with our partner to identify 140 geographically relevant sales markets serviced by a set of sales agents with a common supervisor. We will randomly assign the sale contract available to agents in these markets for a predetermined special offer period. The markets will be randomly assigned to selling conditions such that the number of clusters marketing under each treatment is as follows:
- M0: Status quo (60 markets)
- M2: Earned flex (40 markets)
- M3: Earned rewards (40 markets)

The marketing period length will be set to target enrolling 40 randomly-selected customers per market into the study, for a total enrollment of approximately 5,600 borrowers.

Second, customers who received their devices under the status quo contract (M0) will be individually randomized to be “surprised” with one of two contracts discussed above (earned flex or earned rewards; S2-S3). 1200 customers will be assigned to each of the contracts (S2 and S3). By comparing those who selected into each contract type (M2 and M3) to those surprised (S2 and S3), we isolate the selection effect of being offered the contract.

Design details

Randomization method
Randomization will be conducted via STATA in an office computer, using administrative data from our partner.
Randomization unit
Marketing conditions are randomized at the market level. Surprise offers are randomized at the individual level from amongst those marketed the status quo contract.
Clusters
Experiment 1: no clustering. Experiment 2: 140 clusters.
Observations
The total number of observations across both experiments is 9,364 (3,764 in experiment one and 5,600 in experiment two).
Arms
Experiment one: S0: customers will remain with status-quo contracts, 941 individuals. S1: customers will be surprised with fixed flex, 941 individuals. S2: customers will be surprised with earned flex, 941 individuals. S3: customers will be surprised with earned rewards, 941 individuals. Experiment two: M0: Agents will market the status-quo contract. 60 clusters, 2400 individuals - M0-S2: customers will be surprised with earned flex, 1200 individuals. - M0-S3: customers will be surprised with earned rewards, 1200 individuals. M2: Agents will market contracts with earned flex. 40 clusters, 1600 individuals M3: Agents will market contracts with earned rewards. 40 clusters, 1600 individuals
Primary outcomes
We will further detail our primary and secondary outcomes for the RCT in a pre-analysis plan, which will be posted on the registry before endline data collection is complete. Our primary outcomes include: - Loan repayment - Phone lockout by the lender - Smartphone use - Income generation and labor supply (experiment 1 only)
Secondary outcomes
Secondary outcomes include: - Use of other financial services (saving and debt - experiment 1 only) - Psychosocial well being (experiment 1 only) When studying adverse selection (experiment 2) we will also consider selection in terms of a small number of customer demographics, including gender and education.

Cited by

No citation data available for this registration yet.

Keywords