Refugee Economies in Uganda: What Difference Does the Self-Reliance Model Make?
Uganda’s refugee policies have been widely recognised as among the most progressive in the world. Through its self-reliance model, it allows refugees the right to work and freedom of movement. It has sustained this approach virtually since independence despite currently hosting more refugees than any other African country.
Uganda’s model has three core elements that distinguish it from most other refugee-hosting countries. First, its regulatory framework: it lets refugees work and choose their place of residence. Second, its assistance model: it allocates plots of land for refugees to cultivate within its rural settlements. Third, its model of refugee–host interaction: it encourages integrated social service provision and market access.
This report explores the question: what difference does Uganda’s self-reliance model make? How do its different elements influence welfare outcomes for refugees and for host communities? These questions matter both for Uganda and for refugees around the world. Given that Uganda’s model has become an exemplar for development-based approaches to refugees, an evidence based understanding of the conditions under which self-reliance policies lead to enhanced welfare outcomes matters for policy, programming, and advocacy.
Assessing the impact of the self-reliance model is methodologically challenging. It relies upon being able to compare outcomes for refugees and hosts within the model with refugees and hosts outside the model. In order to do that, we compare welfare outcomes for refugees and host communities in Uganda with those in neighbouring Kenya. We choose this comparison because the countries have contrasting legal and policy frameworks relating to refugees, and yet both are in the same region and host refugee populations from the same countries. The comparison can also help answer questions of broader interest like ‘what difference does the right to work actually make?’
Drawing upon quantitative and qualitative research, including a survey of over 8,000 refugees and host community members in urban (Kampala and Nairobi) and camp (Kakuma and Nakivale) contexts, we provide a nuanced account of the impact that different aspects of the Ugandan model have on particular groups in relation to particular welfare outcomes. We focus on outcomes for Congolese and Somali refugees. The data in the report is representative of our focus populations and selected sites but not for all refugees or host communities in Uganda and Kenya.
We argue that there is a need to go beyond a romanticised view of the Ugandan model, and identify the conditions under which particular self reliance policies actually lead to improved welfare outcomes. The picture that emerges is mixed. It shows that aspects of the Ugandan model are highly effective for some populations, but that other aspects may be less effective than is commonly assumed.
On the positive side, refugees in Uganda enjoy greater mobility, higher incomes, lower transaction costs for economic activity, and possibly more sustainable sources of employment than those in Kenya. However, refugee employment levels in Uganda are surprisingly lower compared with either refugees in Kenya or Ugandan nationals, educational access is more limited, and our data raises questions about the viability of current land allocation practices as the basis for self-reliance.
One of the biggest advantages of the Ugandan model comes from the country’s regulation. Freedom of movement appears to result in higher levels of mobility.
This enables refugees to adopt economic strategies that might not otherwise be possible, including split-family strategies. Mobility is particularly important for Somalis whose commercial activities are often connected to national and transnational supply chains. In Kenya, refugees are working and moving outside the camps. However, they incur far higher transaction costs as a result of doing so.
In terms of employment, overall levels are surprisingly higher among refugees in Kenya than Uganda. However, in Kakuma, this is largely due to the availability of international organisation and NGO-supported ‘incentive work’, whereas refugee employment in Nakivale comes from self-employment in agriculture and marketbased sources, and hence may represent more sustainable sources of employment. In Nairobi, the difference may reflect that the city offers a larger labour market.
Refugees in Uganda generally have higher incomes than those in Kenya, even though there is not a significant difference between the surrounding host communities. However, there is one notable exception: Congolese refugees in Nakivale are worse off than Congolese in Kakuma. This pattern reflects that Congolese in Nakivale are mainly engaged in subsistence agriculture, while Congolese in Kakuma are mainly employed as incentive workers by NGOs. Somalis generally engage in commercial activities, and are able to earn higher incomes across our research sites in Uganda than those in Kenya.
The allocation of plots of land for cultivation is a defining feature of the Ugandan model. However, our data calls into question the sustainability of the current approach. First, the approach does not benefit all communities: although many Congolese households take up the opportunity to cultivate land, Somalis refrain from agricultural activity. Second, there is insufficient land for newly arrived refugees: the overwhelming majority of land is cultivated by families who arrived before 2012. Third, although Congolese refugees who have access to land do better than those who do not, and more land is associated with better food security outcomes, subsistence agriculture is inherently limited as a pathway to high income levels.
Aside from land allocation, levels of assistance in Uganda and Kenya are broadly comparable. This suggests that the most important explanation for refugees in Uganda’s generally better welfare outcomes is the different regulatory environment rather than the assistance model. Indeed, in some areas, notably education, there is evidence that public service provision may actually be weaker in Uganda than in Kenya. For example, in regression analysis and controlling for other variables, being in Nakivale is associated with three years less education than being in Kakuma for refugees who arrived before the age of sixteen. This may be partly due to the greater involvement of the international community in parallel service provision in Kenya compared with direct national government provision in Uganda.
In terms of refugee–host interaction, host communities in Kenya are slightly more likely to have positive perceptions of refugees than in Uganda. This is especially the case for the Turkana around Kakuma and ethnic Somali Kenyans in Eastleigh. In both cases, the difference seems to be based on a perception that their presence brings a positive economic contribution, notably through employment. In the camp context, this difference may be because whereas the economic activities of refugees and hosts are complementary in Kakuma, refugees and hosts undertake similar economic activities in Nakivale, making competition more likely.
Overall, the report provides strong support for the value of ensuring that refugees have access to the right to work and freedom of movement. However, it offers a nuanced account of the aspects of the Ugandan model that lead to improved welfare outcomes and those that do not. Based on these insights, we outline a series of recommendations for refugee policy in Uganda and globally.