1. Introduction
Approximately 85 percent of the global refugees are hosted in developing countries (UNHCR, 2020). Sub-Saharan Africa (SSA) hosts about one-third of them (Ruiz & Vargas-Silva, 2017).
UNHCR (2019) asserts that the number of refugees in the region increased threefold between 2010 and 2019. This rise has been mainly attributed to persistent conflicts in the region (Verwimp & Maystadt, 2015; Kasozi, 2017 & Ivanova et al., 2018). Protracted conflict has also led to long refugee stays; an average of 9-21 years according to Hunter (2009). Hosting refugees could have far-reaching consequences in areas already struggling to ameliorate their own economic situation (Maystadt et al., 2019).
There has been a booming literature assessing the consequences of hosting refugees (Meyer et al., 2011; Ruiz & Vargas-Silva, 2017; Maystadt et al. 2019). Although the literature highlights that refugees can have a positive effect on economic development but with likely distributional consequences, the evidence from individual studies is mixed (Ruiz and Vargas-Silva 2017; Maystadt et al. 2019; Verme and Schuettler 2021). In their review, Verme and Schuettler (2021) argue that the direction of impact depends on which economic dimension is studied.
For instance, they find that beneficial impacts are less likely if the outcome of interest is host employment or wages. By contrast, it is more likely if the outcome of interest is well-being measured in terms of income, consumption or wealth (Verme & Schuettler, 2021).
Furthermore, they stress that few studies have employed panel data to study the impact of hosting refugees on the host communities.
Based on panel data collected between 2009 and 2012, we assess the impact of hosting refugees in Uganda on material welfare, measured by the consumption per adult equivalent. Our main outcome variable differs from studies such as Alix-Garcia and Saah (2009), Alix-Garcia et al. (2012), and Loschmann et al. (2019), which focus on market prices, host employment and household assets. Alix-Garcia et al. (2018) is one exception since they use cross-sectional data on consumption to validate their results based on night light indexes. Other studies on SSA using longitudinal data on consumption include Maystadt and Verwimp (2014), Ruiz and Vargas-Silva (2017), and Maystadt and Duranton (2018) on the Kagera region of Tanzania; and Alloush et al. (2017) on the Congolese refugees in Rwanda. However, all these studies investigate the economic impacts of refugees living in camp settings.
Uganda is an interesting case study. According to the unique Ugandan refugee policy, refugees are not settled in camps but rather live in settlements. According to Betts et al. (2017), Betts et al. (2019), Kreibaum (2016) and UNDP (2017), refugees enjoy a certain freedom of movement, the right to work and are encouraged to engage in agriculture towards attaining self-reliance by availing them with plots of agricultural land and seeds for planting. The World Bank Group (2016) further contends that this progressive refugee policy also supports local integration of refugees. Verme and Schuettler (2021) argue that restrictions on right to work and movement for refugees can significantly determine the direction of impacts on the host communities.
To the best of our knowledge, Kreibaum (2016) and d’Errico et al. (2021) are the closest studies. Kreibaum (2016) examines the effect of refugee presence on household welfare in terms of consumption in Uganda. The author uses three repeated cross-sections of the UNHS1 data and employs a difference-in-difference strategy to determine the effect of refugee presence, in particular in districts hosting Congolese refugees. d’Errico et al. (2021) find that the proximity to refugees, considered as a measure of inter-group interactions, increases the welfare of the hosting population. Our paper complements these studies in several ways. First, we exploit nationally representative surveys, while d’Errico et al. (2021) focus on a few settlements and cross-sectional data collected by FAO in their surroundings. Second, we use panel data and can therefore exploit within district and household variations to better deal with unobserved heterogeneity. The longitudinal nature of our data also allows us to adopt a more dynamic perspective by investigating possible coping strategies at the household level.
Our study utilizes the LSMS-ISA data spanning 3 waves from 2009 to 2012, to quantify the effect of the refugee presence on households’ welfare. We consider refugees from various source countries and residing close to local communities (clusters in LSMS). We construct a refugee index which weights the number of refugees in the closest refugee settlements by the inversed distance from those settlements to the clusters. In order to limit endogeneity concerns, we instrument this variable of interest with a shift-share instrumental variable which is based on the distance of the refugee settlements to the closest border crossing points for each source country.
Our findings with regards to household consumption are similar to those found in Kenya (AlixGarcia et al., 2018), Rwanda (Alloush et al., 2017), and Tanzania (Maystadt & Verwimp, 2014). Our results indicate that rural households, living close to the refugee settlements benefit from the presence of refugees. Similarly, Alix-Garcia and Saah (2009) find that rural households closer to refugee camps experience a positive wealth effect which could be resulting from production and supply of non-aid food products in response to the upward demand and price shifts. We also investigate the heterogeneity of the average impact, its distributional consequences and further discuss coping strategies in the labor and commodity markets. Education level of the household head does not seem to explain the effects of the refugee presence. However, we find that households who are able to change their main source of income to commercial farming benefit more from the refugee influx. This is in line with Whitaker (2002) who argues that it is the relatively wealthier farmers, thus not reliant on subsistence farming, who take advantage of the price dynamics in addition to availability of cheap labor. We also find that the type of crop produced matters in this context. Despite the differences in research designs, it is also interesting to observe that d’Errico et al. (2021) report a similar shift in economic activity. They find a significant reduction in the value of crop sales and an increase in wage income for host-households living closer to the refugee households. d’Errico et al. (2021) points to a shift towards wage employment as an important adaptation mechanism. Similarly, we unearth changes in households’ main source of income as a potential coping strategy. However, though we find that households who change to wage employment seemingly benefit, potentially bigger welfare returns are observed for households who move to commercial farming.
The paper is organized as follows: The next section summarizes the background of the study.
Section 3 describes the data used in this study and presents the descriptive statistics. Section 4 covers the empirical strategy employed. Section 5 discusses the main results of the study and the assumptions underlying the identification strategy used. Section 6 presents insights into the potential coping strategies on the labor and commodity markets. The final section concludes with a summary of the findings and possible recommendations for policy and future research.
Source: World Bank