Source research: Angrist, N. et al. Mobile education builds resilience during shocks in five countries. Nature https://doi.org/10.1038/s41586-026-10990-x (2026).

Messages for policy

• Education systems can be made more resilient by building mobile-phone tutoring into emergency responses, to ensure that learning continues when schools are closed because of disease, conflict or climate shocks.

• Prioritize live phone-call tutoring over text messages alone: calls worked in every country, whereas message-only support helped only in countries with long school closures.

• Deliver through government teachers with confidence: gains were as large under government guidance as when delivered by non-governmental organizations, showing that the approach can be scaled in public systems.

• Expect the largest returns where closures are longest and alternatives fewest. However, the low cost of US$11 per child makes phone tutoring attractive even outside emergencies.

The policy problem

Emergencies often close schools. Disease outbreaks, conflict and climate events disrupt education for at least 200 million children annually¹. Emergency school closures worsen an existing learning crisis in which less than half of primary-school-age children (around 6–14 years old) in low- and middle-income countries can read a simple story or do basic arithmetic². When classrooms close, education systems need alternative ways to keep children learning that are cheap, quick to deploy and workable across various settings. Yet rigorous evidence on which approaches work during emergencies is scarce, and programmes that succeed as small pilots often fail when scaled across contexts³ and when delivered by government systems. Proven, affordable tools that sustain learning across contexts when schooling is disrupted are needed.

The findings

Weekly phone-call tutorials that covered basic numeracy skills targeted to each child’s level produced large learning gains across five countries: India, Kenya, Nepal, the Philippines and Uganda. On average, students learnt roughly three times more when they received phone calls tailored to their learning levels alongside weekly text messages containing numeracy content than with the average effective education intervention. At about US$11 per child for eight weeks of instruction, the phone-call approach yields gains of 3.6 high-quality years of schooling per $100 spent, ranking in the top-ten most-cost-effective educational interventions in a review⁴. Text-message-only interventions worked in settings with the longest school closures, but had inconsistent effects elsewhere, revealing that phone calls are probably necessary to achieve effective results across contexts consistently. Crucially, learning gains from phone-call tutorials delivered by government teachers were as large as those provided by instructors from non-governmental organizations, demonstrating that the programme is scalable in government systems.

Compared with a proof-of-concept trial in Botswana⁵, the learning gains and the implementation of the phone-call tutorials improved over time: the trials in Uganda and the Philippines, which were the last of the trials, showed the largest educational improvements (Fig. 1). This improvement was probably due to the detailed monitoring data and mechanisms used, which enabled implementers to learn from experience and improve delivery over time. Results also indicate that learning assessments conducted over the phone were robust, providing a low-cost, easy-to-deploy approach at scale.

The study

We ran five coordinated randomized controlled trials in India, Kenya, Nepal, the Philippines and Uganda. These included roughly 16,900 primary-school children (covering students in grades 1–5; aged 6–11 years) who were out of school during closures in 2021–22 because of the COVID-19 pandemic. Families were randomly assigned to a comparison group that received business-as-usual interventions according to the nation’s education system; to a group that receive weekly educational text messages; or to a group that received weekly text messages plus eight weeks of 20-minute phone-call tutorials.

During the phone calls, tutors used simple, frequent assessments to pitch each lesson to the child’s actual learning level rather than to their school grade. In Nepal and the Philippines, we also randomized whether government teachers or non-governmental instructors delivered the calls, directly testing whether the model works in government systems. Our five-country study provides evidence that is not only relevant to each context, but also supports a broader generalizability of the interventions’ effects across diverse settings. We assessed the robustness of measuring learning remotely over the phone, enabling real-time measurement of educational outcomes during disruptions.

Funding statement We thank the UBS Optimus Foundation, Mulago Foundation, Douglas B. Marshall Foundation, Echidna Giving, Stavros Niarchos Foundation, Jacobs Foundation, JPAL IGI, Northwestern’s ‘Economics of Nonprofits’ class, the Peter Cundill Foundation and the World Bank. C. Cullen received funding from the UKRI GCRF and Oxford’s OPEN Fellowship.

Competing Interests

N.A. and M.M. co-founded Youth Impact, the non-governmental organization that first developed and tested the ConnectEd phone tutoring programme evaluated in this study. C. Cullen heads Youth Impact’s research team. Youth Impact provided training and technical assistance to implementing partners across the five trials but did not itself deliver the intervention in the study countries and received no financial benefit contingent on the study’s findings. All other authors declare no competing interests.