What We Learned Implementing an Adaptive Program to Support Refugees

Humanitarian crises are increasingly protracted and complex, lacking clear solutions and paths to reach the most-affected individuals and communities. Implementers need to constantly reflect on what is and what is not working within, and adapt accordingly. Our Re:Build project has been attempting to work in an adaptive way over the past two years. This blog summarizes a new paper that details what we’ve learned and the way forward.

Understanding adaptive management

Adaptive management is an approach to humanitarian and development programming. It recognizes that achieving impact is rarely a linear process, particularly when operating on complex challenges within uncertain contexts. Holding implementers accountable to a rigid theory of change at the outset of a program is therefore counterproductive.

Instead, adaptive programs build in opportunities for reflection on what is and what is not working, allowing the program to flexibly course-correct if the original activities are no longer going to deliver the desired impact. It’s particularly useful when the solution isn’t clear, as it allows programs to test multiple potential activities and approaches. Changes are both expected and encouraged.

Operating in this way is more difficult than delivering a standard humanitarian or development program. In particular, it requires a conducive enabling environment consisting of four things. Firstly, donors need to provide multi-year grants that allow for regular changes to budget lines. Secondly, staff at all levels need to be equipped with the skills to critically examine the program and report issues to management. Thirdly, flexibility needs to be passed down to local partners, freeing them from rigid sub-contracts. Finally, new monitoring, evaluation, and learning (MEL) tools that prioritize tracking outcomes and impacts (rather than activities or outputs) need to be developed and used.

Applying these principles within Re:Build

We have been trying to implement adaptive management principles within Re:Build (Refugees in East Africa: Boosting Urban Innovations for Livelihoods Development), a five-year program aiming to support urban refugees in Kampala and Nairobi. The program started in 2021, funded by the IKEA Foundation. The International Rescue Committee (IRC) is the lead partner, CGD is supporting the research, and a whole host of local organizations are supporting implementation.

From the outset, the IKEA Foundation recognized this is a challenge characterized by a high degree of uncertainty. There is little evidence on what works for urban refugees, and the system itself is incredibly complex and interdependent. We therefore did three things; we developed hypotheses around promising solutions; created channels for generating evidence and learning (including several large randomized controlled trials (RCT)); and developed a clear framework for governance and decision making.

After two years, we wanted to know whether these things have worked, in a bid to inform the remaining three years of the program. We issued a structured survey to Re:Build staff and conducted semi-structured interviews, exploring their views on how adaptive Re:Build has been. Here are our four key learnings:

1. Adaptive management requires programs to plan differently. Re:Build has largely been able to be flexible, especially in responding to COVID-19 and subsequent lockdowns. But there is an opportunity to improve timeliness by further decentralizing decision making and further clarifying accountability for decisions.
2. Adaptive programming de-risks creativity and innovation. In standard programming, implementers are locked into delivering the program as designed, making it risky to pursue novel solutions. Working in an adaptive way has allowed Re:Build staff to “fail fast and pivot if needed,” though some worried that too much trial and error could undermine impact if not controlled carefully.
3. A focus on targets can limit adaptive action. Re:Build has a bold target of 20,000 clients served. This ambitious target has created competing incentives: What’s more important? Staying the course and meeting targets, or learning? Some staff felt trialing new actions would undermine this overall goal.
4. Tailored learning methodologies help to balance the need for timely information and rigorous evidence. Nesting RCTs within an adaptive program has presented challenges (traditional RCTs require fidelity to the original design, while adaptive programming encourages flexibility). There has therefore needed to be frequent communication and compromises between the research and implementation team.

The way forward

This analysis came at a good time: the learnings will allow Re:Build to better live out adaptive management principles in the final years of the grant. Yet it also provided some recommendations for those keen to implement new adaptive programs.

For implementers, we recommend socializing adaptive management approaches and expectations with both program and operations functions early on, ensuring they know when and how reflection will take place and how to best feed in learnings into decision making processes. This requires frameworks on how to make decisions, better data and learning capacity, and an incentive structure that supports learning.

For donors, we recommend providing multiyear and flexible funding that values learning outcomes and anticipates adaptation. This means working with implementers on a framework for joint decision making in lieu of a detailed activity plan, holding them accountable for adapting when new learnings are surfaced as opposed to valuing strict adherence to a pre-agreed workplan.

Adaptive programs are difficult to implement, but Re:Build has shown (to date) the value of using an approach focused on reflection and updating activities. If we truly want to have an impact on the livelihoods of urban refugees (and many other communities!) it will be essential to break out of the log-frame mold and embrace a little uncertainty.