Child poverty among refugees

Theresa P. Beltramo, Rossella Calvi, Giacomo De Giorgi, and Ibrahima Sarr

World Development, Volume 171 (2023), Article 106340 


This article estimates the intra-household allocation of consumption in refugee settlements and surrounding communities in Kenya and Uganda, and examines the implications for child poverty. Uganda is the largest refugee hosting country in Africa, with more than 1.5 million refugees and asylum seekers as of June 2022. Kenya is the third-largest refugee-hosting country in Africa, after Uganda and Ethiopia, with over 555,000 refugees and asylum seekers as of June 2022.  

The authors estimate the share of household expenditures for individual household members by identifying personal expenditures that are assignable (exclusively to either men, women, or children) or otherwise allocatable to individual members. Individual shares of household expenditures are then used to compute poverty rates at the individual level. The authors also apply a supervised machine learning algorithm to identify the most critical predictors of child poverty. 

The analysis is based on data from: (1) the 2018 Uganda Refugee and Host Communities Household Survey, which covers households in the largest refugee settlements and in surrounding host communities in the West Nile and South West regions of Uganda; (2) the 2018–2019 Kalobeyei Socio-Economic Assessment covering refugee households in the Kalobeyei refugee settlement in Kenya; and (3) the 2015–2016 Kenya Integrated Household Budget Survey covering Kenyan households. Descriptive statistics reveal: 

  • Refugee households tend to be larger in size relative to host households and have higher youth dependency ratios (share of children to adults). 
  • In both Uganda and Kenya, refugee households are twice as likely than hosts to have female heads. 
  • The average per-capita household expenditure is higher in Kalobeyei relative to its surrounding host communities, whereas the average per-capita household expenditure in refugee settlements in Uganda is lower relative to host communities.  

Main empirical findings: 

  • Children in both refugee settlements and host communities are allocated a disproportionately low share of household consumption relative to adults. In all refugee settlements but with varying intensity, adults’ consumption is above the household per-capita consumption, while children’s consumption is substantially below. Intra-household consumption inequality (the degree to which household resources are shared unequally within households) is widespread in refugee camps and the surrounding hosts communities and in all regions. 
  • Children in both refugee and host communities are more likely to be poorer than adults due to the unequal allocation of resources within households. The poverty rate among refugee children ranges from 39 percent in Kenya to 69 percent in the South West region of Uganda. Among hosts, child poverty ranges from 27 (in the West Nile region of Uganda) to 69 percent (in the South West region of Uganda).  
  • Refugee children suffer a disproportionately high poverty risk compared to children in surrounding host communities. The total poverty gap for children in refugee settlements is estimated to be as much as five times larger than in the surrounding non-refugee communities. 
  • Up to almost half of poor refugee children live in non-poor households. These children would not be reached by anti-poverty programs that ignore intra-household consumption inequality. 
  • A small set of observable traits can predict child poverty in refugee settlements and surrounding host communities. These include a child’s age and gender, household composition (such as the age of the household head, household size, number of children in the household), measures of household food insecurity, the head of household’s education and employment status, the number of rooms in the dwelling, and access to sanitation and clean water.   

Overall, the results show that intra-household consumption inequality is widespread in both refugee and host communities in Uganda and Kenya, with children facing a particularly elevated risk of poverty. The authors conclude that household-level poverty measures based on per-capita expenditure may be weak proxies for child poverty in and around refugee settlements. Without explicitly accounting for the high incidence of children who are poor in non-poor households, existing humanitarian and development programs targeting assistance to poor households will miss these children, making them especially vulnerable. The authors suggest several econometric approaches to predict and target child poverty.