Policy Research Working Paper 9862 Forced Displacement, Gender, and Livelihoods Refugees in Ethiopia Yeshwas Admasu Gender Global Theme November 2021 Policy Research Working Paper 9862 Abstract This study uses the Ethiopia Skills Profile Survey (2017) to economic activities, conversely, higher numbers of children examine the gender differences in livelihood opportuni- in the household significantly reduce women’s opportuni- ties and activities between refugees and host communities. ties. Higher education attainment boosts both male and The results show that refugees are significantly less likely female refugees’ participation in wage employment. Among to be in employment, and that household characteristics refugees, Somali refugees have relatively better access to influence women’s economic opportunities. While having a employment opportunities compared to other refugee female household head, access to agricultural land, and the groups, especially refugees from South Sudan and Sudan. number of female adults increased female participation in This paper is a product of the Gender Global Theme. It is part of a larger effort by the World Bank to provide open access to its research and make a contribution to development policy discussions around the world. Policy Research Working Papers are also posted on the Web at http://www.worldbank.org/prwp. The author may be contacted at ayeshwas@gmail.com. The Policy Research Working Paper Series disseminates the findings of work in progress to encourage the exchange of ideas about development issues. An objective of the series is to get the findings out quickly, even if the presentations are less than fully polished. The papers carry the names of the authors and should be cited accordingly. The findings, interpretations, and conclusions expressed in this paper are entirely those of the authors. They do not necessarily represent the views of the International Bank for Reconstruction and Development/World Bank and its affiliated organizations, or those of the Executive Directors of the World Bank or the governments they represent. Produced by the Research Support Team Forced Displacement, Gender, and Livelihoods: Refugees in Ethiopia Yeshwas Admasu Keywords: refugees, livelihood, gender, forced displacement, Ethiopia JEL classification : D74, J16, O5 Unit: HGNDR (Gender Unit) Project: Gender Dimensions of Forced Displacement funded by “Building the Evidence on Protracted Forced Displacement: A Multi-Stakeholder Partnership” TTLs: Diana J. Arango and Lucia C. Hanmer Forced Displacement, Gender, and Livelihoods: Refugees in Ethiopia Yeshwas Admasu The World Bank, Washington D.C. (E-mail: ayeshwas@gmail.com) Acknowledgments: The author is grateful for the guidance and comments received from Amalia Hadas Rubin, Diana Jimena Arango, Jeni Klugman, and Lucia Hanmer at the World Bank. We gratefully acknowledge Christian Wieser, Matthew Stephens and Mattias Lundberg for their comments and suggestions. We also thank participants of the Gender Dimension of Forced Displacement workshop. All errors remain our own. Funding Disclaimer: This work is part of the program “Building the Evidence on Protracted Forced Displacement: A Multi-Stakeholder Partnership''. The program is funded by UK aid from the United Kingdom's Foreign, Commonwealth and Development Office (FCDO), it is managed by the World Bank Group (WBG) and was established in partnership with the United Nations High Commissioner for Refugees (UNHCR). The scope of the program is to expand the global knowledge on forced displacement by funding quality research and disseminating results for the use of practitioners and policy makers. This work does not necessarily reflect the views of FCDO, the WBG or UNHCR. Data availability statement: The data used in this paper is publicly available at the World Bank Microdata Library at https://microdata.worldbank.org/index.php/catalog/3445/study-description Disclosure statement: There is no conflict of interest. 1 Introduction Forced displacement, long considered as a humanitarian crisis, is also a development challenge. There are 26 million refugees, of whom 85 percent are hosted in developing countries (UNHCR, 2020c). Ethiopia is one of the largest refugee-hosting countries in the world and home to the third-largest refugee population in Africa (UNHCR, 2020a). Fragility and conflict in the Horn of Africa have led to refugee inflows from South Sudan, Somalia, Eritrea, and Sudan. Although Ethiopia maintains an open-door policy towards refugees, a strict encampment policy was in place until early 2019, that limited livelihood opportunities and economic inclusion. Refugee families and communities encounter a range of vulnerabilities and insecurities in new environments where their livelihoods and social protection have been eroded or disappeared. Regaining access to traditional livelihoods, like farming, is often impossible in situations of forced displacement (Cagoco-Guiam, 2013). Despite the efforts of governments and partner agencies to promote economic opportunities for refugees and support their inclusion within the host communities, they often face legal and practical barriers to entering the formal labor market (Crawford et al., 2015). For example, until recently refugees did not have the right to work in Ethiopia. These barriers not only affect refugees’ livelihoods and well-being, but they also create a wide range of costs and foregone benefits for the host country (see Alix-Garcia & Saah, 2010; Alix-Garcia, Walker, Bartlett, Onder, & Sanghi, 2018; Kreibaum, 2016; Taylor et al., 2016; Verme & Schuettler, 2021; World Bank, 2011). Ethiopia is in the process of reforming its policies toward refugees. Pledges at the 2016 Leaders’ Summit on Refugees held in New York led to the adoption of a range of policies and administrative instruments, including the Refugee Proclamation of 2019 with significant revision to its existing national refugee law. The Refugee Proclamation provides refugees the right to work and reside out of camp, open and register business in their own name, access social and financial services, and register births and marriages (UNHCR, 2020a). However, it remains the case that refugees in Ethiopia still have limited rights as the implementation has been slow; for example, in 2020, Out-of-Camp permits 1 (OCP) were issued to a large number of Eritrean refugees, but while residence permits have been issued for some refugees, the process still lacks clarity (UNHCR, 2020d). In addition, there are specific gender-related differences or barriers in terms of policies and/or their implementation in accessing services and economic opportunities; including the existence of discriminatory provisions of work permits, limited access to sexual and reproductive health services, weak participation of women in community-based leadership structures, and significant drop in girls’ primary school attendance (UNHCR, 2020d). The lack of developed markets around camps also limits economic opportunities for refugees (Dempster et al., 2019; Vemuru, Sarkar, & Fitri 1 The Out-of-Camp permits (OCP) provide regular residency outside refugee camps, which is issued by the Administration for Refugee and Returnee Affairs (ARRA) to refugees on the condition that they can support themselves either through self-reliance (sponsorship) or holding a work permit allowing them to work legally. 1 Woodhouse, 2020). This is further exacerbated by the fact that refugees are largely located in poor areas of Ethiopia where poverty rates are high and livelihood opportunities are limited. Numerous studies from around the world have documented the constraints faced by displaced people in accessing livelihood opportunities, including unfavorable political and policy context, poverty, lack of access to capital and markets, negative gender stereotypes and discriminatory attitudes (Campbell, Crisp, & Kiragu, 2011; Jacobsen, Ayoub, & Johnson, 2011; Jacobsen & Fratzke, 2016; Samuel, 2014). However, studies on the gender dimensions of access to livelihoods opportunities have been more limited. This paper examines livelihood opportunities among refugee and host community households in Ethiopia. For this purpose, we use four livelihood activities including wage and salaried employment, self-employment, participation in family farm activities, and participation in family businesses. First, we examine the gender differences in livelihood activities among refugees and host communities. Second, we compare gendered livelihood outcomes between the four refugee groups (South Sudanese, Somalis, Eritrean, and Sudanese) 2 and comparable host community households. Finally, we analyze livelihood constraints and opportunities that are available to advance the well-being of forcibly displaced women. Our study aims to increase understanding of livelihoods in one of the largest and most complex refugee settings in the world, with a focus on refugee women and girls. The refugee groups in Ethiopia are diverse, and we are able to compare four refugee-hosting regions that are geographically and ethnically distinct. This study contributes to the empirical literature by providing a closer look at the gender differences in livelihood activities among forcible displaced population and host communities. While there are studies that have looked at gender and displacement (Cagoco-Guiam, 2013; Pape, Petrini, & Iqbal, 2018), there is no empirical study that examines gender differences and livelihoods in a forced displacement situation. We build on these studies to investigate the gender differences among refugees and hosts and compare the two groups, where displacement is increasing, and the literature thus far has focused on the impacts on hosts. The results both contribute to the empirical knowledge on the gender dimensions of forced displacement and inform governments and humanitarian agencies on supporting appropriate livelihood strategies for refugees, across gender lines and refugee groups. The remainder of the paper is organized as follows. Section 2 provides background information on the four main refugee groups in Ethiopia. Section 3 describes the data used in the analysis 2 While the paper used data from the Skills Profile Survey 2017, the situation for refugees in Tigray region is completely changed since first drafting the paper. The conflict in Norther Ethiopia is creating a full-scale humanitarian crisis with many internally displaced individuals as well as having a direct impact on Eritrean refugees. According to the UNHCR, two refugee camps (Shimelba and Hitsats) were destroyed following the conflict, leading Eritrean refugees who lived in the two camps prior to the crisis to flee and leaving them in urgent need of support and safety. Thus, any findings related to Eritrean refugees should be interpreted with caution as they do not reflect their current situation. 2 and presents descriptive statistics. Section 4 discusses the empirical strategy and section 5 presents the empirical results. The final section concludes. 2 Country Context Ethiopia is one of the largest countries in Africa, characterized by low levels of urbanization, high rates of poverty (23.5% of the population lived below the national poverty line in 2016) and low levels of human development, with only about half the adults being literate and only one-third of youth (aged 15 -24) having completed primary school (World Bank, 2020). There are also major gender gaps. Women are less likely to own land alone or jointly (40% compared to 48% of men, with men owning a larger proportion of land than women) and to use a bank account (15% compared to 25% of men), and almost one in four women have experienced physical violence since the age of 15 (23%) and sexual violence (10%) (CSA & ICF, 2017). Access to health care service is also a problem for women due to poverty and distance to the health facility. Female rates of agricultural labor participation are 55%, compared to 83% for men (Hasanbasri et al., 2021). As elsewhere in the world, women and girls do most of the unpaid household and care work (Admasu, Crivello, & Porter, 2021; Pankhurst, Crivello, & Tiumelissan, 2016), which reduces their time available to engage in education and/or paid employment. Ethiopia traditionally had an open-door policy towards refugees and is among the largest hosts of refugees in Africa. 3 In the Horn of Africa, refugee movements have been driven by the conflict in South Sudan, the prevailing political environment in Eritrea, and the conflict and drought in Somalia (World Bank Group & UNHCR, 2015). The number of refugees in Ethiopia has increased over the last decade, exceeding 735,000 in 2020 (UNHCR, 2020a). The majority of refugees live in the 24 government managed camps, near the borders of their respective country of origin (Figure A2), located in Tigray region and the four emerging regions: Afar, Benishangul-Gumuz, Gambella, and Somali (UNHCR, 2020a). The four regions that host refugees are also the least developed regions in the country, characterized by poor infrastructure, high poverty rate, low administrative capacity and poor development indicators (Pape et al., 2018; UNHCR, 2020a). While Ethiopia has long history of hosting refugees, their prospects for durable solution has been hindered by the traditional encampment policy (Zetter & Ruaudel, 2016). Refugees were disadvantaged because of legal restrictions on their right to work, restriction on movement and residence, and inability to obtain business licenses, which prohibits them from taking up most forms of formal employment or establish their own enterprise. These legal restrictions together with the isolation of refugees living in camps limited their livelihood opportunities (Samuel, 3 In addition to refugees, Ethiopia has a large number of internally displaced persons (IDPs) with more than 1.8 million IDPs in 2020 (IDMC, 2020; IMO, 2020). The main causes of displacement are conflict, drought and floods (IDMC, 2020). 3 2014), and meant that they were highly dependent on humanitarian assistance. The United Nations High Commissioner for Refugees (UNHCR) works closely with the Agency for Refugees and Returnee Affairs (ARRA), to coordinate the delivery of protection and assistance for refugees in Ethiopia (UNHCR, 2020b). Recent changes to refugee policy mean that Ethiopia now has among the most progressive regimes in Africa (UNHCR, 2020a). The 2019 refugee law 4 introduces several provisions that will expand refugees’ socio-economic opportunities – given: • the right to freedom of movement - the right to choose their place of residency and freedom to leave the country at any time. • the right to work – that is, the right to engage in wage earning employment in the formal sector and the right to open and register businesses in their own name. In this regard, the government is working with industrial partners to build on industrial parks to create over 100,000 jobs, of which at least 30% are reserved for refugees. In this regard, by the end of June 2020, more than 2,800 residence permits were issued by ARRA. In addition, the new law ensures that the allocation of irrigable land (10,000 hectares) to benefit refugees and hosts on an equal basis. • access to housing and property rights – which provides the right to purchase, lease, or use housing and immovable property. • access to education and health care services – which provides the right to enroll in pre- primary and primary education under the same condition as host communities as well as the right to enroll in secondary and tertiary education. It also provides refugees access to health services including sexual and reproductive services for women and girls. • access to financial and administrative services – which provides the right to open bank accounts, financial services, and use mobile money services. The proclamation also provides refuges the right to obtain a national driver’s license and recognizes driver’s licenses as well as refugees’ foreign academic and vocational qualifications. • access to civil registration and documentation – including registration of births, deaths, marriages, and divorces. • the right to local integration – for refugees in protracted displacement (those who lived in Ethiopia for over 20 years). This 2019 law thus removes various legal hurdles that had limited refugee’ socio-economic rights and opportunities. This expands the potential for refugees to become self-reliant and build their livelihoods, still income-earning opportunities remain limited for both refugees and hosts across the refugee hosting regions of Ethiopia (Vemuru et al., 2020). 4 See Refugees Proclamation No.1110/2019. https://data2.unhcr.org/en/documents/details/68964 4 2.1 Refugees in Ethiopia Most refugees in Ethiopia originate from four countries: South Sudan, Somalia, Eritrea, and Sudan. Figure 1 provides the official data since 1970 and illustrates the rise in the refugee population over the last decade. Source: UNHCR’s Operational Portal Ethiopia Figure 1: Number of refugees in Ethiopia by country of origin We briefly review the characteristics of these different groups, to provide important context for the empirical analysis which follows. Refugees from South Sudan South Sudanese comprised about 45% (329,123 persons) of all refugees in Ethiopia in 2020 (UNHCR, 2020b). Among South Sudanese refugees, about 20% were adult women compared to only 10% of adult men (Figure A3). Of South Sudanese refugees 66% are children under the age of 18, including significant numbers of unaccompanied or separated children (20%). They primarily reside in the Gambella region and a small number of new arrivals in Benishangul- Gumuz. Most of these refugees have arrived since 2013. In Gambella, the refugee population outnumbered the host population in 2017 (Abebe, 2018). The presence of South Sudanese refugees (mostly ethnic Nuer) has thus altered the region's social, economic, and demographic context, and exacerbated the existing tensions between the two largest ethnic groups in Gambella, the Nuer and Anywaa (Pape et al., 2018; Vemuru et al., 2020). The security situation in the Gambella region has been volatile due to ethnic tensions and competition among refugees and host communities over resources, and environmental degradation including deforestation 5 caused by displacement (Vemuru et al., 2020). At the same time, studies suggest that the creation of refugee camps and the associated large-scale humanitarian operations have benefited host communities through infrastructure projects and expanded services (Pape et al., 2018; Vemuru et al., 2020). In Gambella, refugee livelihoods are more constrained than other refugee hosting regions of Ethiopia. The three primary livelihood activities are cutting down wood or collecting grass for sale, brewing local alcohol, and selling items that they receive in their rations (Vemuru et al., 2020). Moreover, insecurity constrains livelihood options, and mobility, which can especially affect female refugees. Refugees from Somalia Somali constitute about one-fourth of registered refugees in Ethiopia (UNHCR, 2020b). Again, there are more adult women versus men (19% versus 13%) while two-thirds are children. Most of the refugees are concentrated in five Dolla Ado refugee camps in which refugees outnumber the host population of 141,000 (Betts et al., 2019). Somali refugees arrived in waves over several decades, which begin after the 1987/88 civil war in Somalia. The greatest influx was recorded during 2011 due to drought and food insecurity compounding ongoing conflict (Pape et al., 2018). In the Somali region of Ethiopia, refugees and host communities share the same ethnic group, culture, language, and religion. The socio-cultural proximity between the two groups has reportedly allowed a peaceful coexistence and greater integration compared with other displacement settings (Betts et al., 2019). While refugees in Ethiopia have traditionally been highly dependent on humanitarian assistance, only 66% of Somali refugees are aid dependent, compared to 83% of other refugee groups (Pape et al., 2018). In the region, refugees can pursue various livelihoods including in services and retail, and agriculture. Around 21% of refugees have an income generating activity, compared with 29% of the host community (Betts et al., 2019). The largest sources of employment for both refugees and host communities are humanitarian NGOs, ARRA, and international organizations. In Jijiga, refugee camps have developed into a vital economy where refugees conduct entrepreneurial activities and participate in traditional keeping and trading of small ruminants. The camps also provide markets, shops, tailors, kiosks, barbers, and laundry services run by the host community. However, livelihood opportunities for refugees within and outside the camps are reportedly limited. 6 Refugees from Eritrea Ethiopia currently hosts about 174, 000 Eritrean refugees and asylum seekers, representing 22.5% of registered refugees in the country (UNHCR, 2020b). Since 2000, Eritreans have fled political persecution, open-ended military conscription, systemic human rights violations, and economic hardship (World Bank Group, 2019). Some also seek family reunification with relatives in Ethiopia. Among the Eritrean refugees, 28% are adult women and 25% adult men, with 44 percent children (Figure A3). Large numbers of unaccompanied and separated children among new arrivals have strained humanitarian and reception services, including immediate care arrangements, emergency education, and individual counseling (UNHCR, 2020a). Eritrean refugees lived in four camps in Tigray region and the two camps and settlements within host communities in Afar region. However, due to the conflict in Northern Ethiopia (within and outside Tigray region), 5 which broke out in November 2020, a significant number of Eritrean refugees who were hosted in Shimelba and Hitsats camps (in Tigray region) have been fleeing to other locations, with some relocating to Mai Aini and Adi Harush camps and other regions of Ethiopia (UNHCR, 2021). According to the UNHCR, the two refugee camps (Shimelba and Hitsats) have been destroyed and entirely deserted following the conflict. 6 The ongoing conflict and insecurity have continued to affect the delivery of services and provision of humanitarian assistance to Eritrean refugees as well as the internally displaced persons (IDPs). Moreover, the conflict has had a considerable impact on the deteriorating refugee-host relationships. Many Eritrean refugees also live outside the camps following policy changes, which relaxed movement restrictions for Eritrean refugees. In 2010, Ethiopia implemented the out-of-camp scheme allowing Eritrean refugees to live outside camps on the condition that they are able to sustain themselves (Samuel, 2014). According to Ruaudel and Morrison-Métois (2017), there were more than 8,000 refugees living in Addis Ababa under this scheme in 2015. For Eritrean refugees in camps, there is heavy dependence on humanitarian assistance, reflecting limited livelihood opportunities in the camps. The main types of jobs accessible to refugees are casual labor in the construction sector, self-employment in the business sector, and working with NGOs and ARRA. The camps have tended to not develop much economic activity because of high rates of exit. There have been high rates of secondary movement of Eritrean refugees out of the camps, particularly from Tigray, to urban centers within Ethiopia and to Europe through Sudan and Libya (UNHCR, 2020a). 5 The ongoing conflict in Northern Ethiopia is creating a full-scale humanitarian crisis with an ever-increasing IDP population. The limited access to cash, food, fuel and other essential supplies has severely hindered operational activities, which has created a complex humanitarian situation (UNHCR, 2021). 6 According to the UNHCR, work is progressing in all seven sites in Dabat (in Amhara region) for the new Ale mwach refugee camp, which is planned to accommodate an expected 20,000 refugees to be relocated from Mai Aini and Adi Harush camps. 7 Refugees from Sudan The number of Sudanese refugees in Ethiopia was around 42,000 in 2020, residing in five camps in the ethnically mixed region of Benishangul-Gumuz (UNHCR, 2020b). There are similar shares of adult women and men, while children are 60 percent of the total. Sudanese refugees arrived in successive waves since the 1980s, with the first group escaping from the Blue Nile areas of Sudan and arriving in 1987-89 (Vemuru et al., 2020). Most of the current refugees arrived in 2011. This includes double refugees, individuals who arrived as refugees in the late 1980s and returned to Sudan in the 2000s only to return as refugees to Ethiopia in 2011 (Vemuru et al., 2020). Sudanese refugees, like South Sudanese, have limited livelihood opportunities, and primarily engage in activities like artisanal gold mining, wood cutting, and daily farm labor (Vemuru et al., 2020). Access to the local market helped refugees to generate income by running businesses including small shops, restaurants, and cafes. Cross-border trade is also important in the region, which benefits both refugees and locals by providing access to more dynamic markets in Sudan. 3 Data The Skills Profile Survey 2017 (SPS 2017) is an excellent source of data to examine and compare refugee and host community profiles and economic activities – although the data pre- dates the most recent reforms which liberalized economic opportunities for refugees and strengthened their economic and social rights on various fronts, as outlined above. While the situation of refugees may well have improved since the survey, the results are still important and instructive, especially in casting light on constraints associated with camp-based locations in poor developing country settings. SPS 2017 is a household survey carried out by the World Bank that covered refugees in camps and members of surrounding host communities. The sample consisted of four strata based on the main refugee hosting regions: Tigray-Afar (Eritrean), Gambella (South Sudanese), Benishangul-Gumuz (Sudanese and South Sudanese), and Somali (Somali). Using a multistage stratified sampling, refugee households were randomly selected from 82 enumeration areas (EAs) in each stratum. 7 Since only refugees living in camps were surveyed, the sample is only representative of refugees living in camps in Ethiopia (Pape et al., 2018). The samples for the host communities were selected by classifying Ethiopian non-displaced households residing within a 5 km radius of a camp (in the four stratum) as host community. Then, host community households were randomly selected from 42 EAs in each stratum (World Bank Group, 2019). Figure 2 shows the sample composition across refugee and host communities by gender of the 7 Details on the sampling procedure can be found at https://microdata.worldbank.org/index.php/catalog/3445/study-description#metadata-sampling 8 household head and refugee hosting regions. The total survey sample was 5,317 households with 3,627 refugee households (837 South Sudanese, 871 Somalis, 893 Eritreans, and 1,016 Sudanese) and 1,690 host community households. 100 90 80 70 60 50 40 30 20 10 0 Refugee Host South Sudanese Somali Eritrean Sudanese Community Male-headed HH Female-headed HH Source: Author’s calculation using Skills Profile Survey (2017), Ethiopia Figure 2: Percentage of refugee and host community households: by gender of head and refugee groups The SPS 2017 contains detailed information on household characteristics, food and non-food consumption, assets, livestock, education and economic conditions, access to services, and perceptions. It also includes individual-level information on employment, barriers to labor force participation, livelihood structures of refugees before displacement, movement and return intentions. The data combines detailed household questionnaire information with displacement-specific information including drivers of displacement, access to resettlement mechanisms, and return intentions. 3.1 Livelihood Activity Variables For the empirical analysis, we use four types of livelihood variables, defined as follows: • Wage employment refers to work where the employee receive payment regularly in the form of cash or in kind. This includes paid work for an NGO, private company, paid farm work, or paid domestic work. • Self-employment refers to people who work on their own account including shop owners, family non-farm business, street traders, or people who own non-farm business. • Family business refers to any non-farm business owned by a family member. This includes helping to prepare food or drinks to sale, helping to sell at the shop. • Farm activities include working on a farm owned by the household or member of the household, raising the household’s livestock, herding, fishering, or planting fields. 9 3.2 Household Demographics The four refugee groups and host communities differ in terms of gender of household head, age, and gender composition. Overall, it is notable that there are more adult women than men from each of the refugee groups – except Sudan. At the same time the majority of refugees are children, meaning very high dependency ratios as well as childcare responsibilities. In general, host communities have a higher proportion of adults than the four refugee groups. All refugee groups except Eritreans have higher percentages of children than host communities. The South Sudanese refugee household are distinct because more than 90% refugee households are headed by women, whereas the majority of Eritrean, Somali, and Sudanese refugee households are headed by men (Figure 2). This arises because only 6% of South Sudanese refugees are adult men (Figure 3). According to Pape et al. (2018), the missing men among South Sudanese could be partly explained by the fact that adult men may be more likely to join the conflict or be recruited by armed groups; indeed 30% of survey respondents identified recruitment by armed groups as the main factor for family separation. However, similar patterns were also observed among the Sudanese (World Bank Group, 2019). Another important difference is that Eritrean refugees have more working age men (who could be fleeing conscription), and there is a higher adult share among Eritreans. 60 50 40 30 20 10 0 Male Female Male Female Male Female Male Female Male Female Male Female Refugee Host Community South Sudanese Somalis Eritreans Sudanese Children (0-14) Youth (15-24) Adults (25-64) Elderly (>65) Sources: Author Calculation using Skills Profile Survey (2017), Ethiopia and Pape et al. (2018) Figure 3: Demographics of refugee population in Ethiopia 10 3.3 Sources of Income and Labor Participation Refugees in Ethiopia tend to be dependent on humanitarian assistance to a larger extent than host communities. In 2017, some refugees were earning incomes from wage employment, agriculture, and running business along with remittances. We also see differences across refugee groups (Figure 4 and Table 1), • a significant number of Somali refugees earn most of their income from wage employment (24% of male-headed households and 12% of female-headed households). Also, about 15% of refugees obtains most of their income from other sources including self-employment (services and retail), agriculture, and remittances. This is further confirmed by their high percentage of participation in the labor market, we see that men refugees are significantly more likely to participate in wage employment (27%) compared to women (9%). Similarly, among hosts, 33% of men are wage employed, while only 7% of women in hosts are participating in wage employment. • among Eritrean refugees, only 13% obtain most of their income from wage and salaries, retail and services, and remittances. Looking at participation in employment, 7% of women wage employed compared to 12% men. The gender difference is also observed among their hosts, with 36% of men and 14% of women in wage employment. • livelihoods are the most constrained among Sudanese and South-Sudanese refugees, with 11% of South-Sudanese obtaining most of their income from agriculture, self- employment, and wage employment. This is further reflected by the low percentage of South-Sudanese men and women participating in wage employment (6%). While South Sudanese refugees derived incomes from agriculture and wages before their displacement, currently they are highly dependent on humanitarian assistance. This is due to low labor market participation among refugees, with more women (more than 40%) compared to men (around 21%) are inactive and not attending any education. There is also significant gender difference in education among South Sudanese refugees, while about 50% of female have no education compared to only 12% of men. • Sudanese refugees relied primarily on agriculture (more than 70%) as their source of income before their displacement, currently they are heavily dependent on humanitarian assistance for their livelihood. In terms of participation in employment, 13% are participating in wage employment compared to 6% of women. Interestingly, among hosts, only 19% of men and 7% of women are wage employed, while over half of the hosts are participating in agriculture. This suggests the limited availability of employment opportunities in the region. Generally, labor participation is low among refugees - more women (close to 60%) compared to men (around 25%) are not participating and not attending any education. Most women (52%) and about one-fourth of Sudanese male refugees report having no education at all. 11 Table 1: Participation in employment by gender and refugee hosting regions Gambella Somali Tigray-Afar Benishangul-Gumuz Men Women Men Women Men Women Men Women Refugees Labor participation (all) 0.19 0.22 0.40 0.20*** 0.20 0.16* 0.16 0.09*** Wage employment 0.06 0.06 0.27 0.09*** 0.12 0.07*** 0.13 0.06*** Self-employment 0.08 0.09 0.12 0.10 0.06 0.06 0.04 0.02** Unpaid employment 0.13 0.16 0.03 0.03 0.03 0.04 0.02 0.02 Agriculture 0.02 0.02 0.05 0.03* 0.02 0.02 0.01 0.02 No. of observation 361 610 920 1053 1496 1022 1420 1519 Hosts Labor participation (all) 0.66 0.29*** 0.72 0.50*** 0.80 0.65*** Wage employment 0.33 0.07*** 0.36 0.14*** 0.19 0.07*** Self-employment 0.16 0.12 0.16 0.20 0.17 0.08*** Unpaid employment 0.06 0.05 0.03 0.04 0.07 0.07 Agriculture 0.23 0.09*** 0.25 0.13*** 0.54 0.51 No. of observation 329 341 381 487 993 1109 Notes: Mean comparison across refugees by gender. Asterisks indicate statistical significance at 1% ***, 5% **, and 10% * levels. Overall, female-headed households are more likely to depend on aid and less likely to participate in the labor market than male-headed refugee households (Figure 4). The gender differences are striking, with only 13% of female-headed refugee households obtaining most of their income from sources other than aid compared to 26% of male-headed refugees. There are large differences between female- and male-headed households in the sources of their income. Whereas 19% of male-headed households participate in agriculture and wage employment, only 5% of female-headed households participate in these activities. Female- headed are more engaged in services (6%) than male-headed households (4%). Gender differences in labor force participation are significant - most women report being economically inactive and not enrolled in school or college. Summary statistics of variables by the gender of the household head is presented in Table A1 in the appendix. 12 Female-headed Male-headed Refugee - Current Refugee - Before disp Host community 0 10 20 30 40 50 60 70 80 90 100 Agriculture Self Employment Paid Employment Aid Remittances Source: Author’s calculation using Skills Profile Survey (2017), Ethiopia Figure 4: Source of livelihood by gender of the household head and refugee status To examine the factors influencing the gender differences in livelihood activities among the four refugee hosting regions, we undertook an in-depth empirical study. We therefore focus on the empirical work to estimate the gender gaps among refugees and hosts, understand what drives the gender gaps, as well as compare refugees and hosts across regions. Thus, the gendered analysis will help to determine the various needs and capacities of refugee women and men. 4 Empirical Strategy In our empirical analysis, we use two different models to examine the gender differences in livelihoods among the forcibly displaced and host communities in Ethiopia. We begin by examining the gender differences in the probability of being in employment (formal or informal). A probit model is used where the dependent variable takes a value of one if an individual is in employment and zero otherwise. = ( + 1 + 2 )--------------------------------------------------- (1) The model is estimated separately for female- and male-headed households as well as female and male individuals to see the differences at both household and individual levels. Where is the probability of an individual/household in year being in employment (regular and casual), are camp fixed effects that capture local labor market level. is a vector of individual/household head characteristics, which contains education attainment (primary, secondary, and higher education), and marriage status. is a vector of household characteristics, including age and sex of household head, household size, household head education level, number of children, access to land, and livestock ownership. 13 We have seen above that women and men significantly differs in livelihood activities both among refugees and host communities. However, to assess the gender differences more rigorously in livelihood activities among the forcibly displaced and host communities, we use matching techniques to pair refugee households with host community households based on observable characteristics. In theory, in a randomized experiment where households are randomly assigned, the impact of displacement and of gender can be obtained by comparing the means of refugee households and host community households. However, refugee households are generally not similar to households in the host community. Thus, a simple comparison of these two groups of households can be confounded by selection issues. To address this and to enable greater rigor, we use matching to identify a set of comparable households. We employ a propensity score estimation to balance the distribution of the covariates (individual and household level) across refugee and host community households, thereby reducing bias in the measurement of the effect of forced displacement based on observables. To identify a valid comparable counterfactual, we start by estimating the probability of being a refugee household using a probit model: = + + --------------------------------------------------------- (2) Where, is the probability a household is a refugee, is a vector of variables, and is the error term following logistic distribution. The set of explanatory variables included in are demographic characteristics of the household head, education, household size, number of children in the household, land ownership, number of adult females in the household, and remittances. We estimate propensity scores using logistic regressions and derive the comparable sample of households using kernel matching. The kernel method improved the quality of matching and reduces bias by using only households with most similar characteristics. Since the weighting function differs for the different matching methods available, we present results using the nearest neighbors matching, full matching, and kernel matching. The main assumption of the matching technique is that there are no differences in unobservable characteristics once we have matched on observables (Hill, Reiter, & Zanutto, 2004; Rubin & Thomas, 1996). The use of cross-sectional data limits our ability to match on characteristics prior to displacement. Thus, we acknowledge that some selection bias issues remain. Therefore, we are not interpreting the estimated differences as impact rather we are only comparing the two groups in terms of livelihood activities. Using the predicted values from the above equation, we generate the propensity score for all households, and then ensure common support by plotting densities of the propensity scores among refugees and host community households. Figure A1 confirms common support of the propensity scores. This indicates that for each refugee household we are able to find a host community household with similar propensity scores. Further, we assessed the quality of the 14 constructed comparison group by testing whether the distribution of the covariates between the two groups given the predicted propensity score. Table A2 presents the result of the balancing test and suggests the balancing property is satisfied. 5 Estimation Results We begin our analysis at the household level by comparing female and male employment among refugees and host communities. Estimation results from Equation 1 are reported as average (marginal) effects for female and male-headed sub-samples by displacement status. The average effects show the change in the probability of being in employment which is associated with a one-unit change in the explanatory variables; for indicator variables it is the difference with the reference category. The results confirm significant gender differences in both the refugee and host community populations. Specifically: • Adult refugees living in female headed households are significantly - 7.5 percentage points - less likely to participate in the labor market compared to their male headed counterparts: • Adults from host communities living in female headed households are also less likely to participate in the labor force – the gender gap is 3.1 percentage points. There are also important differences between refugees and host communities: in particular, education appears to have a stronger effect on the probability of being in employment among refugees, male and female. This is true for primary, secondary, and tertiary education. In contrast, we do not find strong significant effect of education on the probability of being in employment among people in host communities, except for tertiary education. This points to the importance of expanding access to education to enable refugees to gain technical and vocational skills. Education can thus play a significant role in increasing refugee women’s participation in the labor market by providing them with the skills needed for the labor market (as Table 2 shows women with secondary or tertiary education are more likely to be employed). Access to land also significantly increases the probability of female-headed households’ participation in employment – by 16 percentage points among host communities. Table 2: Participation in employment: probit model Refugees Hosts All Female Male All Female Male HHH HHH HHH HHH Female head -0.075*** -0.031* (0.019) (0.017) Age of household head 0.001 -0.001 0.001 0.000 -0.001 0.001 15 (0.001) (0.001) (0.001) (0.000) (0.001) (0.000) Married 0.090*** 0.022 0.131*** 0.011 0.007 0.025 (0.019) (0.023) (0.036) (0.017) (0.036) (0.022) No. of children below age 6 -0.017** -0.020** -0.018 0.003 -0.033 0.013* (0.007) (0.010) (0.012) (0.008) (0.020) (0.007) No. of female adults 0.037*** 0.027** 0.022 0.047*** 0.081*** 0.014 (0.010) (0.012) (0.019) (0.014) (0.030) (0.013) No. of male adults 0.019*** 0.079*** -0.001 0.033*** 0.065** 0.026** (0.006) (0.012) (0.008) (0.011) (0.027) (0.011) Head Education (reference: no education) Primary 0.079*** 0.068*** 0.096*** 0.018 0.060* 0.006 (0.019) (0.026) (0.029) (0.017) (0.035) (0.017) Secondary 0.152*** 0.184** 0.152*** 0.026 0.019 0.023 (0.033) (0.073) (0.038) (0.021) (0.061) (0.018) Tertiary 0.288*** 0.391*** 0.270*** 0.054 *** 0.000 0.022 (0.058) (0.146) (0.062) (0.019) (.) (0.020) Remittance -0.072** -0.097** -0.049 -0.129 *** -0.144 *** -0.132*** (0.028) (0.043) (0.039) (0.021) (0.041) (0.026) *** *** Access to land 0.000 -0.029 0.039 0.057 0.160 0.016 (0.027) (0.037) (0.038) (0.017) (0.037) (0.017) Livestock 0.156*** 0.120*** 0.178*** -0.000 -0.001 0.008 (0.017) (0.024) (0.025) (0.015) (0.034) (0.015) Refugee hosting regions Gambella 0.016 0.031 -0.003 (0.029) (0.035) (0.074) Somalia 0.114*** 0.091*** 0.116*** -0.011 -0.031 0.003 (0.024) (0.034) (0.035) (0.018) (0.043) (0.017) *** *** Benishangul-Gumuz -0.133 -0.096 -0.192*** 0.060*** 0.105*** 0.045*** (0.022) (0.032) (0.031) (0.016) (0.040) (0.015) Number of Obs. 3546 1828 1718 1676 512 1138 Notes: The dependent variables are participation in employment. Robust standard errors are in parentheses. Asterisks indicate statistical significance at 1% ***, 5% **, and 10% * levels. Marital status has no impact on employment for female-headed refugee households, but a strong positive effect of marital status on male-headed refugee households. We would expect high dependency ratios to constrain women’s economic opportunities, especially in the absence of affordable childcare and uneven distribution of care responsibilities at home. In this case, we find that the number of young children has significantly reduces female - but not male - employment among refugees. And we find that children have a positive significant effect on male employment among hosts. This suggests that gendered divisions of unpaid labor adversely affect women’s employment opportunities. 16 Receiving remittances significantly reduces the probability of employment across the board – for both men and women - refugees and hosts. In our sample, 10% of refugees (9% of female headed households) and 5% of hosts (11% of female headed households) receive remittances from their relatives. The impacts of gender appear to vary across refugee hosting regions, especially for refugees. In particular, refugee women from Sudan are 10 percentage points less likely to work than refugee women from Eritrea and Somali women 9 percentage points more likely to work than Eritrean women. In contrast, refugee Somali men are 12 percentage points more likely to work than Eritrean men and Sudanese men 19 percentage points less likely to work than Eritrean men. Gender patterns also vary among hosts, interestingly men and women in Benishangul- Gumuz region are more likely to work than men and women in Tigray region. The regional differences indicate that livelihood opportunities are generally more limited in Benishangul-Gumuz and Gambella regions. Refugees in Somalia region appear to have better access to livelihood opportunities compared to other refugee hosting regions. Moreover, refugees and hosts in the Somali region share both language and culture. 5.1 Employment Participation: Matched Data In this section, we further examine different livelihood activities using matched data. We first estimate a fixed effect model at household level to compare female and male employment as well as total employment between refugee and host communities. For this purpose, we use participation in different employment activities by household members between the age of 15 and 65. Table 3 presents the estimation results using matched data for female and male employment. This confirms the result that refugees are significantly less likely to be employed than individuals in host community households. However, behind relatively lower employment rates for refugees, there is a huge variation across refugee groups and gender, specifically, comparing matched households: • Refugees in Benishangul-Gumuz region (Sudanese and South-Sudanese refugees) are 65 percentage point less likely to be in employment, underlining that refugees have very limited access to livelihood opportunities in the region. Somali and Eritrean refugees are 22 and 35 percentage points less likely to be in employment compared to their hosts. • The overall differences among women refugees and women hosts are large: 24 percentage points. The difference is largest in Benishangul-Gumuz region, where refugees are 41 percentage points less likely to have an employed female member compared to their hosts. Interestingly, the differences in the probability of having an employed female member among refugees and their hosts in Eritrea and Somalia regions are not statistically significant. 17 • In contrast, the result for having an employed male member shows a 35-percentage points difference between male refugees and male hosts. This difference is observed in all refugee hosting regions ranging from 19 percentage points in Somali region, 29 percentage points in Tigray and Afar region, and to 46 percentage points in Benishangul-Gumuz region. • Comparing the observed differences between refugees and hosts by gender, the results indicates that women refugees show a relatively smaller difference with their hosts (24 percentage points) compared to men refugees who show the largest difference with their hosts (35 percentage points). Unpacking the results further, we see that the differences among female refugees and host are driven by difference in employment in Benishangul-Gumuz region. In contrast, among men, the differences are observed in all refugee-hosting regions. Table 3: Employment participation: matched results All Female Male *** *** *** Refugee -0.461 -0.238 -0.348 (0.033) (0.034) (0.037) Refugee - Somali -0.218*** -0.039 -0.193*** (0.068) (0.060) (0.073) Refugee - Eritrea -0.349*** -0.098 -0.294*** (0.066) (0.064) (0.070) Refugee - Sudanese -0.651*** -0.413*** -0.463*** (0.043) (0.052) (0.052) Number of Obs. 2002 2002 2002 2002 2002 2002 Notes: The dependent variable is participation in employment. All regressions include camp fixed effects. Robust standard errors are reported in parentheses. Asterisks indicate statistical significance at 1% ***, 5% **, and 10% * levels. Overall, the results suggest that Somali refugees have better employment opportunities than other refugee groups in Ethiopia. As noted above, refugees and hosts in the Somali region share both language and culture. This has allowed a peaceful coexistence and greater integration compared with other displacement settings, and better opportunities for socio-economic integration (Betts et al., 2019). Moreover, some Somali refugees are also benefited from crop irrigation schemes in Melkadid, which was implemented in line with CRRF pledge with the aim of benefiting refugees and hosts on equal basis (UNHCR. 2020b). We check the robustness of our results by expanding the sample to include all households in the SPS. Table A4 shows that when we include all households, our estimated results remain robust. 18 5.2 Employment Participation: By Type We next examine the effect of displacement status on the likelihood that an individual will participate in different types of work, namely: (a) wage employment; (b) self-employment (non-farm business), (c) own or family farm activities, and (d) unpaid work in the family business. Interestingly, there is no statistically significant difference in female participation in wage employment between refugee and host community households across regions. We find similar results for male participation in wage work in Somalia and Benishangul-Gumuz regions, while Eritrean male refugees are less likely to participate in wage employment compared to their hosts (significant at 10%). However, there are significant differences between refugees and hosts in participation in self- employment, household farm activities, and family business (results for family business are presented in Table A3). Table 4: Participation in different types of employment by gender: matched results Wage employment Self-employment Family Farm Female Male Female Male Female Male Hosts - reference Refugee - Somali 0.051 0.034 -0.002 -0.097 -0.048* -0.176*** (0.040) (0.069) (0.033) (0.062) (0.029) (0.053) Refugee - Eritrea -0.008 -0.118* -0.059 -0.146** -0.012 -0.049 (0.042) (0.063) (0.051) (0.058) (0.025) (0.035) Refugee - Sudanese -0.030 -0.026 -0.140*** -0.234*** -0.232*** -0.231*** (0.027) (0.033) (0.033) (0.048) (0.034) (0.034) Number of Obs. 2002 2002 2002 2002 2002 2002 Notes: The dependent variables are participation in paid employment, self-employment, and family farm activities. All regressions include camp fixed effects. Robust standard errors are reported in parentheses. Asterisks indicate statistical significance at 1% ***, 5% **, and 10% * levels. Sudanese and South-Sudanese refugees, both male and female, are significantly less likely to be self-employed and participate in family farm activities compared to their hosts. This is partly due to refugees are less likely to have access to land and own their own business (see Table 1). While we find no significant difference in participation in family farm activities between Eritrean refugees and their hosts, male Eritrean refugees are less likely to be self-employed by 15 percentage points compared to their hosts. Female participation in self-employment shows no significant difference between Somali refugees and their hosts, while the difference in participation in family farm activities is only 19 significant at 10% level. In contrast, male refugees from Somalia are significantly less likely to participate in family farm activities (18 percentage points) compared to their hosts. In general, the disaggregated results suggest that self-employment and agricultural employment are driving the observed differences among refugees and hosts, reflecting the limited availability of paid employment opportunities in refugee-hosting regions. Among women, we only see significant differences in Benishangul-Gumuz region where refugees are less likely to participate in self-employment and/or farm activities compared to women in hosts. This difference directly links with refugees’ lack of agricultural land, while women in hosts have greater access to land and about half participates in farm activities. Individual-Level Participation We now turn to examine the determinants of wage employment for working age individuals (aged 15-65). Estimates are reported in as average (marginal) effects for women and men by refugee status. The results from Table 5 suggests that there are significant gender differences, at individual level, both among refugee and host community population. The gender gap is large among hosts where women are 15 percentage points less likely to participate in wage employment compared to men. Similarly, there is a 9-percentage points gender difference among refugees, in which women are significantly less likely to be employed. Overall, our results suggest that women (both refugee and hosts) are more likely to suffer from lack of employment opportunities compared to men. These results are consistent with findings in the descriptive statistics (Table 1), which show large differences in employment between women and men. Again, at the individual level, education increases the probability of participation in wage employment for both women and men, both among hosts and refugees. While secondary and tertiary education are strongly associated with female employment, tertiary education has a large and strong effect on male employment. Displacement duration also significantly increases participation in employment, with women who are displaced for at least three years more likely to participate in the labor market compared to those displaced for less than three years. In contrast, among men, duration only plays a role for those who are displaced for more than six years. Lack of physical safety – measured as feeling moderately or very unsafe when alone at home or walking around during the day or after dark – significantly reduces wage employment for women (but not men) in the host community and have no effect on refugees. Table 5: Individual participation in wage employment: probit model Hosts Refugees All Female Male All Female Male 20 Female -0.149*** -0.094*** (0.012) (0.008) Household characteristics Female household head 0.022 0.047*** -0.026 -0.004 0.034*** -0.042*** (0.015) (0.015) (0.029) (0.008) (0.010) (0.013) Married household head 0.010 -0.014 0.067* 0.030*** 0.006 0.052*** (0.019) (0.017) (0.036) (0.009) (0.010) (0.015) No. of young children 0.005 -0.008 0.019* -0.007** -0.014*** 0.001 (below 6) (0.006) (0.006) (0.011) (0.003) (0.004) (0.005) No. of female adults -0.019** -0.016 -0.035** -0.006 -0.013*** -0.010 (0.009) (0.010) (0.017) (0.004) (0.005) (0.008) No. of male adults -0.033*** -0.002 -0.047*** -0.024*** -0.001 -0.030*** (0.007) (0.007) (0.012) (0.003) (0.004) (0.005) Education (reference: no education) Primary 0.018 0.014 0.025 -0.004 0.020** -0.026** (0.013) (0.013) (0.023) (0.008) (0.009) (0.013) Secondary 0.054*** 0.046** 0.075** 0.068*** 0.099*** 0.056*** (0.018) (0.020) (0.030) (0.015) (0.027) (0.020) Tertiary 0.403*** 0.406*** 0.447*** 0.179*** 0.143* 0.199*** (0.033) (0.057) (0.041) (0.038) (0.076) (0.046) Internet access 0.038** 0.026* 0.056** 0.003 0.010 -0.007 (0.015) (0.015) (0.028) (0.010) (0.012) (0.015) Physical Safety -0.043** -0.068*** -0.022 -0.011 -0.007 -0.015 (0.018) (0.022) (0.030) (0.009) (0.010) (0.014) Remittance -0.074** -0.036 -0.106 -0.014 -0.004 -0.032* (0.038) (0.031) (0.079) (0.012) (0.015) (0.019) Access to land -0.102*** -0.063*** -0.132*** -0.028** -0.003 -0.056*** (0.012) (0.013) (0.021) (0.013) (0.014) (0.021) Displacement duration (reference: less than three years) Three to six years 0.027*** 0.033*** 0.020 (0.010) (0.012) (0.015) Greater than six years 0.030*** 0.027** 0.033** (0.010) (0.013) (0.016) Refugee hosting regions Gambella -0.036** 0.013 -0.110*** (0.017) (0.018) (0.031) Somali -0.031* -0.028* -0.029 0.036*** 0.024* 0.043** (0.017) (0.017) (0.030) (0.011) (0.014) (0.017) Benishangul-Gumuz -0.038** -0.001 -0.086*** -0.025** -0.010 -0.039** (0.015) (0.016) (0.027) (0.011) (0.014) (0.018) Number of Obs. 3559 1896 1663 8205 4111 4094 Notes: The dependent variables are male and female adult individual participation in paid employment. Robust standard errors are reported in parentheses. Asterisks indicate statistical significance at 1% ***, 5% **, and 10% * levels. Looking at household characteristics, having a female household head, access to land, and number of children aged below 6 in the household all significantly affect women’s participation in wage employment. Women in households headed by female are more likely to participate in 21 wage employment. Refugee households headed by women are 3 percentage points more likely to have a female member participating in wage employment. In contrast, access to land significantly reduces the probability of having a female member participating in wage employment among hosts, while it has no effect for female refugees. This result is further confirmed by the finding that access to land significantly increases women participation in farm employment among hosts (results not reported). This suggests that access to land plays an important role in ensuring women participation in employment activities, particularly in areas where wage employment opportunities are limited and women access to wage work opportunities are constrained. While having internet access shows significant positive effect on both women and men wage employment in host communities, it has no effect on refugees. The number of young children in the household reduces female participation in paid employment among refugees but has no significant effect on male participation. Being married boosts the likelihood of men being employed, but not women. These results reflect the unpaid caring responsibilities of married women and the high dependency ratio among female headed refugee households. Finally, the results by refugee groups confirm our previous results (in Table 2), which indicate that Somali refugees are relatively better off compared to other refugee groups. For example, Somali women and men are 2- and 4-percentage points more likely to be employed compared to Eritrean women and men, respectively. In contrast, South Sudanese men (11 percentage points) and Sudanese men (4 percentage points) are less likely to be employed compared to Eritrean men. 6 Conclusion This paper has presented the first ever examination of gender gaps in economic opportunities in Ethiopia, which hosts one of the largest refugee populations in Africa. While our data predates the 2019 national refugee proclamation which grants refugees access to a range of socio-economic rights, the results remain relevant to Ethiopia and to forced displacement settings around the world. We found that while refugees are more likely to rely on humanitarian assistance, more detailed econometric analysis provides deeper and robust insights than simple descriptive statistics. Several factors appear to drive gender gaps which face both women in refugee and host communities. For women, the number of young children and lack of access to land limit opportunities, although they are more likely to work if the household is female headed. Very low levels of education among many refugee women limit their economic opportunities. Although women refugees and hosts have limited access to employment opportunities, there are significant differences between these two groups. For example, refugee women are more likely to live in households with larger family size, high dependency ratio and high poverty rate compared to women in host communities. In addition, women in host communities are more likely to have secondary or above education, whereas refugee women are less likely to 22 go beyond primary school. Moreover, there is a huge difference in access to land where women in host communities have an advantage over refugee women. We found important differences between refugees and hosts in the types of work pursued. The results from a fixed effect model on matched data suggest that refugees are significantly less likely to be self-employed and participate in family farm activities. Particularly women refugees in Benishangul-Gumuz have a low rate of participation even compared with women refugees in other regions. There are also significant differences across regions, with refugees in the Somali region having better access to livelihood opportunities compared to the Benishangul-Gumuz and Gambella regions where Sudanese and South-Sudanese refugees are hosted. These findings point to the importance of expanding access to education, especially for young refugees. Particular efforts are needed to close gender gaps in education, bearing in mind the diverse range of cultural and religious backgrounds. As refugees’ access to public technical and vocational trainings are limited, providing greater access and skill development is particularly important for refugees to run their own businesses. The high dependency ratio that constrains women’s economic opportunities underlines the importance of access to sexual and reproductive health services so that women can decide when and how many children to have. There is also a need to expand childcare services in refugee shelters and host community neighborhoods. This could be an area for support for small business start-ups, to provide startup capital and technical training. Increasing economic inclusion of refugees through joint refugee-host cooperatives, investment in agriculture and livestock, and employment creation is vital, as there are limited livelihood opportunities in the refugee-hosting regions. In this regard, the government’s efforts in industrial parks and irrigation schemes to benefit refugees and hosts are good progress. However, gender should be given special attention as women are particularly affected by the legal and practical barriers to access the labor market. Refugee women typically face greater challenges and have lower rates of employment compared to refugee men. Similarly, among host communities, women have lower rates of employment compared to men. This gender gap has likely been exacerbated by the disproportionate effect of the pandemic. Thus, livelihood support activities should be tailored to the specific challenges that both refugee and host women face. Promoting peaceful coexistence between refugees and host communities combined with community security emerges as an important fact. Because Somali refugees have a positive relation with their hosts, this has helped to close the gaps between refugees and hosts observed elsewhere, as Somali refugees are better able to pursue local economic opportunities. 23 The 2019 proclamation is a welcome reform and could help to address many of the underlying constraints faced by refugees in Ethiopia. The rights to work, mobility, access to land and other relevant resources and services are all key aspects to facilitating the economic inclusion of refugees. While formal rights are important, it may remain difficult to access the labor market and business opportunities where refugees lack networks in the market. Proactive efforts to disseminate information and provide support may be needed to support newly emerging opportunities. Finally, rigorous evaluation of livelihood programs using experimental and quasi-experimental methods helps to produce evidence on the effectiveness of these programs in increasing self-reliance, decent work, and income. 24 References Abebe, T. T. (2018). Ethiopia’s refugee response-focus on socio-economic integration and self- reliance. ISS East Africa Report, 2018(19), 1-20. Admasu, Y., Crivello, G., & Porter, C. (2021). Young women’s transitions from education to the labour market in Ethiopia. Alix-Garcia, J., & Saah, D. (2010). The effect of refugee inflows on host communities: Evidence from Tanzania. The World Bank Economic Review, 24(1), 148-170. Alix-Garcia, J., Walker, S., Bartlett, A., Onder, H., & Sanghi, A. (2018). Do refugee camps help or hurt hosts? The case of Kakuma, Kenya. Journal of Development Economics, 130, 66-83. Betts, A., Bradenbrink, R., Greenland, J., Omata, N., & Sterck, O. (2019). Refugee economies in Dollo Ado: development opportunities in a border region of Ethiopia. Cagoco-Guiam, R. (2013). Gender and livelihoods among internally displaced persons in Mindanao, Philippines. Brookings Institution: Brookings-London School of Economics Project on Internal Displacement. Campbell, E., Crisp, J., & Kiragu, E. (2011). Navigating Nairobi: A review of the implementation of UNHCR’s urban refugee policy in Kenya’s capital city. Geneva: UNHCR. Central Statistical Agency of Ethiopia, & ICF. (2017). 2016 Ethiopia Demographic and Health Survey Key Findings. Addis Ababa, Ethiopia, and Rockville, Maryland, USA. CSA and ICF. Retrieved from https://www.statsethiopia.gov.et/wp- content/uploads/2019/06/Ethiopia-Demographic-Health-Survey-Report-Key- Findings-2016.pdf Crawford, N., Cosgrave, J., Haysom, S., & Walicki, N. (2015). Protracted displacement: uncertain paths to self-reliance in exile. London: Overseas Development Institute. Dempster, H., Ble, M. G., Graham, J., & Zimmer, C. (2019). Refugees’ Right to Work: Progress and Remaining Challenges. Retrieved from https://www.cgdev.org/blog/refugees-right-work-progress-and-remaining-challenges Hasanbasri, A., Kilic, T., Koolwal, G., & Moylan, H. (2021). LSMS+ Program in Sub-Saharan Africa: Findings from Individual‑Level Data Collection on Labor and Asset Ownership. Washington, D.C.: World Bank Group. Retrieved from https://openknowledge.worldbank.org/handle/10986/35544 25 Hill, J. L., Reiter, J. P., & Zanutto, E. L. (2004). A comparison of experimental and observational data analyses. Applied Bayesian Modeling and Causal Inference from Incomplete‐Data Perspectives: An Essential Journey with Donald Rubin's Statistical Family, 49-60. Jacobsen, K., Ayoub, M., & Johnson, A. (2011). Remittances to transition countries: the impact on Sudanese refugee livelihoods in Cairo. Cairo papers on migration and refugees; no. 2. Jacobsen, K., & Fratzke, S. (2016). Building livelihood opportunities for refugee populations: lessons from past practice: Migration Policy Institute Washington, DC. Kreibaum, M. (2016). Their suffering, our burden? How Congolese refugees affect the Ugandan population. World Development, 78, 262-287. Pankhurst, A., Crivello, G., & Tiumelissan, A. (2016). Children's work in family and community contexts: examples from Young Lives Ethiopia. Pape, U. J., Petrini, B., & Iqbal, S. A. (2018). Informing durable solutions by micro-data: a skills survey for refugees in Ethiopia. Retrieved from Ruaudel, H., & Morrison-Métois, S. (2017). Responding to Refugee Crises: Lessons from evaluations in Ethiopia and Uganda as countries of destination. Rubin, D. B., & Thomas, N. (1996). Matching using estimated propensity scores: relating theory to practice. Biometrics, 249-264. Samuel, H. (2014). Living Out of Camp–Alternative to Camp-Based assistance for Eritrean Refugees in Ethiopia. Commissioned by NRC. Taylor, J. E., Filipski, M. J., Alloush, M., Gupta, A., Valdes, R. I. R., & Gonzalez-Estrada, E. (2016). Economic impact of refugees. Proceedings of the National Academy of Sciences, 113(27), 7449-7453. UNHCR. (2020a). Ethiopia country refugee response. Retrieved from https://reliefweb.int/report/ethiopia/ethiopia-country-refugee-response-plan-2020- 2021 UNHCR. (2020b). Ethiopia Fact Sheet (June 2020). Retrieved from https://data2.unhcr.org/en/documents/details/78059 UNHCR. (2020c). Global Trends: Forced Displacement in 2019. Retrieved from https://www.unhcr.org/5ee200e37.pdf UNHCR. (2020d). Refugee Policy Review Framework Country Summary as at 30 June 2020, Ethiopia. 26 UNHCR. (2021). Ethiopia Country Refugee response plan January 2020 - December 2021. Retrieved from https://reporting.unhcr.org/node/31204 Vemuru, V., Sarkar, A., & Fitri Woodhouse, A. (2020). Impact of Refugees on Hosting Communities in Ethiopia: A Social Analysis. Retrieved from https://openknowledge.worldbank.org/handle/10986/34267 Verme, P., & Schuettler, K. (2021). The impact of forced displacement on host communities a review of the empirical literature in economics. Journal of Development Economics, 102606. World Bank. (2011). The impacts of refugees on neighboring countries: A development challenge. In: World Bank. World Bank. (2020). Ethiopia Poverty Assessment : Harnessing Continued Growth for Accelerated Poverty Reduction. World Bank, Washington, DC. Retrieved from https://openknowledge.worldbank.org/handle/10986/33544 World Bank Group. (2019). Informing Durable Solutions for Internal Displacement in Nigeria, Somalia, South Sudan, and Sudan : Country Case Studies. World Bank, Washington, DC and UNHCR, Geneva. Retrieved from https://openknowledge.worldbank.org/handle/10986/32626 World Bank Group, & UNHCR. (2015). Forced Displacement and Mixed Migration in the Horn of Africa. World Bank, Washington, DC and UNHCR, Geneva. Retrieved from https://openknowledge.worldbank.org/handle/10986/22286 Zetter, R., & Ruaudel, H. (2016). Refugees’ right to work and access to labor markets–An assessment. World Bank Global Program on Forced Displacement (GPFD) and the Global Knowledge Partnership on Migration and Development (KNOMAD) Thematic Working Group on Forced Migration. KNOMAD Working Paper. Washington, DC: World Bank Group. 27 7 Appendix Table A1: Test of Covariate Balance across refugee groups by gender of the household head South Sudanese Somalis Eritrean Soudanese Male Female Male Female Male Female Male Female Heads characteristics Age 30.65 33.35 41.01 35.98*** 31.66 30.95 33.36 32.48 *** *** *** Early marriage 0.10 0.83 0.06 0.54 0.10 0.60 0.26 0.69*** Married (Monogamous) 0.50 0.80*** 0.75 0.57*** 0.57 0.41*** 0.61 0.62 *** Married (polygamous) 0.15 0.07 0.17 0.04 0.01 0.01 0.11 0.05*** Separated/ Divorced 0.03 0.02 0.02 0.23*** 0.08 0.34*** 0.05 0.22*** Widow 0.00 0.10* 0.01 0.15*** 0.01 0.08*** 0.01 0.09*** Single 0.33 0.02*** 0.05 0.02* 0.34 0.16*** 0.21 0.03*** Household composition Household size 5.10 5.84* 6.69 5.89*** 4.99 4.73 4.89 5.60*** No. of adults 2.70 2.18** 2.49 2.04*** 3.34 2.20*** 2.29 1.96*** No. of female adults 0.82 1.46*** 1.15 1.36*** 0.78 1.68*** 0.87 1.28*** No. of male adults 1.88 0.72*** 1.34 0.68*** 2.56 0.53*** 1.43 0.68*** No. of elderly 0.05 0.03 0.02 0.05* 0.01 0.02 0.01 0.05** No. of children (=60 Male >=60 Eritrea Somalia South Sudan Sudan Source: Author’s calculation using data from UNHCR’s Operational portal Ethiopia Figure A3: Demographics of refugees in Ethiopia by Country of Origin 33