Policy Research Working Paper 9826 A Multi-Country Analysis of Multidimensional Poverty in Contexts of Forced Displacement Yeshwas Admasu Sabina Alkire Uche Eseosa Ekhator-Mobayode Fanni Kovesdi[ Julieth Santamaria Sophie Scharlin-Pettee Gender Global Theme October 2021 Policy Research Working Paper 9826 Abstract Despite the many simultaneous deprivations faced by relationships between multidimensional poverty, displace- forcibly displaced communities, such as food insecurity, ment status, and gender of the household head. The results inadequate housing, or lack of access to education, there reveal significant differences across displaced and host com- is little research on the level and composition of multi- munities in all countries except Nigeria. In Ethiopia, South dimensional poverty among them, and how it might Sudan, and Sudan, female-headed households have higher differ from that of host communities. Relying on house- MPIs, while in Somalia, those living in male-headed house- hold survey data from selected areas of Ethiopia, Nigeria, holds are more likely to be identified as multidimensionally Somalia, South Sudan, and Sudan, this paper proposes a poor. Lastly, the paper examines mismatches and overlaps Multidimensional Poverty Index (MPI) that captures the in the identification of the poor by the MPI and the $1.90/ overlapping deprivations experienced by poor individuals day poverty line, confirming the need for complementary in contexts of displacement. Using the MPI, the paper measures when assessing deprivations among people in con- presents multi-country descriptive analysis to explore the texts of displacement. 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 authors may be contacted at atybogale@worldbank.org, sabina.alkire@qeh.ox.ac.uk, uekhator@worldbank.org, fanni.kovesdi@qeh.ox.ac.uk, juliethsa@iadb.org, and sophie. scharlin-pettee@qeh.ox.ac.uk. 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 A Multi-Country Analysis of Multidimensional Poverty in Contexts of Forced Displacement Yeshwas Admasu[†], Sabina Alkire[‡], Uche Eseosa Ekhator-Mobayode[†], Fanni Kovesdi[‡], Julieth Santamaria[§], and Sophie Scharlin-Pettee[‡] JEL classification: I32, J16, D63, D10, D74, O55 Keywords: forced displacement, multidimensional poverty, monetary poverty, Sub-Saharan Africa, refugee, IDP, internal displacement, gender inequality, female-headed households The authors of this paper conducted their research under Gender Dimensions of Forced Displacement project. The project is co-led by Lucia Hanmer and Diana Arango under the guidance of Hana Brixi, Global Director, Gender Unit, The World Bank Group. Acknowledgments: The authors are grateful for the guidance and comments received from Amalia Hadas Rubin, Andrew Peter Brudevold-Newman, Bilal Malaeb, Diana Jimena Arango, Jeni Klugman, Kathleen Beegle, and Lucia Hanmer at the World Bank, as well as Monica Pinilla-Roncancio and Corinne Mitchell at OPHI. The authors would also like to acknowledge the help of Utz Pape with the data used in this study. All errors remain our own. Funding: 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. The data that support the findings of this study are openly available at the World Bank Microdata Library at https://microdata.worldbank.org/index.php/home. 1. Introduction As of December 2019, 1 UNHCR estimates that nearly 80 million people were forcibly displaced due to conflict and violence, or as a result of natural disasters. Among them, close to 46 million people are internally displaced in their home country, and 26 million are living as refugees 2 in a host country (UNHCR 2020). Some have moved for short periods, while others have been forcibly displaced for years or decades. In 2019, fewer than half a million forcibly displaced individuals had returned to their homes – signaling the chronicity and longevity of displacement that has increased throughout the last decade, partly due to political instability in North Africa and the Middle East, Sub-Saharan Africa and in countries such as the República Bolivariana de Venezuela, Afghanistan or Myanmar. And while the displaced population only amounts to 1% of the global population, these communities are disproportionately affected by deprivation, often chronic. UNHCR (2020) estimates that over 80% of all displaced people reside in countries and areas affected by acute food insecurity and malnutrition, and many of the internally displaced and refugee families live in temporary housing or camps with basic living conditions and limited access to services and employment. Many governments and organizations are searching for solutions to respond to the growing number of displaced people within and across countries and to address concerns around human well-being and safety. Such attempts require timely and representative data on internally displaced persons (IDPs) and refugees that accurately reflect their unique situation and capture specific issues related to forced displacement. While various studies have looked at poverty among refugees and IDPs from a monetary perspective, they do not capture the overlapping deprivations experienced by many in forced displacement, such as lack of access to education, food insecurity, and inadequate housing. Forcibly displaced individuals may be more likely to live in overcrowded houses with poor infrastructure, leading to poor health outcomes (IOM 2013). They also might have difficulty accessing social services or schooling because of a lack of identification or other barriers. These challenges also differ based on the gendered experience of the individual. Labor market integration and social integration – two major considerations for forcibly displaced communities – feature long-standing gendered concerns (Blau, Kahn, & Souza 2003, Blau, Kahn, & Papps 2011, OECD 2018). Women may struggle more than their male counterparts to find opportunities for decent work (ILO 2009). Women and girl IDPs may also become more vulnerable to other forms of exploitation and marginalization as a result of their deprivations, such as human trafficking, child marriage, and sexual-based violence (IOM 2020, Kelly 2021, Kelly et al. 2021). Despite this, to date, there have been very few studies that have looked at multidimensional poverty among forcibly displaced populations (Temgoua Noumedem et al 2020, Loaiza Quintero et al 2018), and even fewer that then apply a gendered lens. This paper adds to the emerging scholarship by developing a tailored Multidimensional Poverty Index (MPI) for a multi-country study of refugees and internally displaced populations in subnational regions of Ethiopia, Nigeria, Somalia, South Sudan, and Sudan. The aim is to better understand differences between the forcibly displaced communities and host communities with [†] World Bank, Washington, D.C [‡] Oxford Poverty & Human Development Initiative [§] Inter-American Development Bank, Washington, D.C, USA 1 Provisional estimates from UNHCR suggest it is likely that the 80 million figure has been surpassed as of mid- 2020. 2 Throughout the paper we refer to refugees as inclusive of refugees and people in refugee-like situation, as per the classification by UNHCR. 1 regards to who is poor, how poor they are, and the composition of their poverty. Based on the Alkire-Foster (AF) method, the index provides a summary measure of poverty for the population that can be disaggregated by displacement status and gender of the household head to analyze the variation in deprivations. The MPI can be further broken down by indicator to show the proportion of the population who are poor and deprived in each area. These features of the MPI can inform better policy responses, with interventions and programs targeting the most deprived communities and indicators with the highest headcount ratios. The paper proceeds as follows. Section 2 of the paper reviews some of the existing literature to provide the background and motivation for the analysis, including a summary of the different country contexts covered by the data analyzed in this paper. Section 3 outlines the Alkire- Foster method and the selected dimensions and indicators used to construct the MPI, followed by Section 4, which introduces the data. Section 5 presents the findings, first for results at the national level and then results disaggregated by displacement status. Section 6 analyzes differences in multidimensional poverty by gender of the household head to improve understanding of the gendered aspects of multidimensional poverty in these contexts. Section 7 compares the MPI results with monetary poverty, with some concluding remarks discussed in Section 8. 2. Background and literature review 2.1 Poverty and forced displacement By 2019, there were about 51 million IDPs across the world, most of them – 46 million – displaced by conflict and violence, with around five million displaced due to natural disasters (IDMC 2020). According to estimates by UNHCR (2020), the number of refugees reached over 20 million as of the end of 2019. While in many cases, conflict and natural disasters have been temporary, resulting in fluctuations in the number of people fleeing their homes in any given country, the global number of IDPs and refugees has grown almost every year over the last two decades. UNHCR estimates that the number of refugees has doubled over the last ten years. Nearly all IDPs live in low- and middle-income countries, and many have experienced secondary displacement. Overall, about half live in urban areas, with one-fourth in major urban areas (i.e., populations exceeding 300,000). Since almost all IDPs are in developing countries, governments are often resource-constrained in terms of providing assistance and access to services, and in some cases, government authorities may be a cause of displacement (World Bank Group 2020). For refugees, the situation is more varied, with most staying close to their country of origin while a smaller minority fled to countries further away. UNHCR (2020) estimates that three-quarters of all refugees were hosted by neighboring countries. To reflect the increase in forced displacement over the last decade and enable sustainable and long-term solutions to refugee situations, the UN Statistical Commission approved a new indicator, SDG Indicator 10.7.4, in early 2020 to measure and track the “proportion of population who are refugees, by country of origin” (UNHCR 2020). While the specific challenges for displaced communities depend on the country or host community context, often, in new locations, key challenges confronting IDPs and refugees include food insecurity, lack of livelihood opportunities, and tensions and competition over resources with host communities. The multiplicity of deprivations faced by displaced 2 communities and their host neighbors warrants a multidimensional lens on the experience of poverty, as a complement to a monetary poverty analysis. While most studies note the varied socio-economic issues facing displaced populations, they still tend to discuss poverty largely in monetary terms (Chaaban et al 2013, Oruc 2015, Verwimp 2012, Zetter and Ruaudel 2014) or in terms of single or specific dimensions of non- monetary well-being. However, a small number of recent studies have attempted an analysis of multidimensional poverty in these communities. For instance, Temgoua et al. (2020) analyze the relationship between internal displacement and poverty status according to the global MPI for Iraq. Loaiza Quintero et al. (2018) develop an MPI for Antioquia in Colombia to study the situation of internally displaced persons due to armed conflict. This paper builds on those studies by taking a cross-country MPI approach to explore patterns of multidimensional poverty among IDPs and refugees in five African countries. 2.2 The role of gender in poverty and forced displacement Gender plays a prominent role in contexts of forced displacement. Although men typically suffer higher mortality rates due to conflict, the demographics of internal displacement still reflect overall gender parity, with the global male-female ratio of IDPs remaining largely half- and-half over the last decade (UNHCR 2020). Moreover, this increased mortality of men and displacement of women can lead to a higher percentage of female-headed households among forcibly displaced populations (Buvinic et al 2013), and these households may be particularly vulnerable, especially in terms of income and social networks (Brück and Schindler 2009, Hanmer et al 2020). That women constitute half the IDP population but are nonetheless exposed to increased vulnerabilities and deprivations deserves serious attention. Forcibly displaced and refugee women may suffer greater impacts on health and more instances of gender-based violence than their male counterparts (Kelly 2021, Kelly et al. 2021). Gendered challenges with food insecurity, lack of livelihood opportunities, and inaccessibility to land ownership, public services, and community structures also characterize poverty and forced displacement. Gender-specific challenges to obtaining livelihoods in protracted displacement settings – due to norm discrimination in the labor market, risk of sexual or physical violence, educational disparities, inter alia – provide a double burden for women hoping to integrate into their host community and lift themselves and their families out of poverty (Brück & Stojetz 2021). Exclusion from the formal labor market paired with the need for economic security may also lead to displaced women and girls engaging in sex work and transactional relationships that come with additional risks, such as violence, prejudice (against any form of commercial sex), and exposure to HIV/AIDS and other health concerns (ODI 2019b). The disadvantages faced by women in developing countries are well-established, with women typically having fewer opportunities in the labor market (World Bank 2011) and being less likely than men to own land (World Bank 2011) or to have access to formal credit markets (King 2009). In addition to the disadvantages faced by women in general, female-headed households may be particularly disadvantaged (Klassen 2013), as female heads may carry a double burden of domestic work and income-generating activities (World Bank 2011). Empirical evidence on the relationship between poverty and female headship is ambiguous. Many studies find that female-headed households are poorer than male-headed ones (Buvinic and Gupta, 1997; Lampietti and Stalker, 2000), others find no evidence that female-headed households are disproportionate among the poor, and in many cases results are mixed (Quisumbing et al 2001, Munoz-Boudet et al. 2018, Brown & van de Walle 2020). When 3 looking at multidimensional poverty, studies in Burkina Faso and Togo have found large inequalities by gender – with women more deprived than men in the six dimensions studied (housing, basic utilities, assets, education, employment, and access to credit) – albeit coupled with regional variation (Agbodji et al 2013), though these results did not consider displacement status. The lack of consensus concerning the poverty situation of female-headed households is due in part to the use of the inconsistent definition of headship and the heterogeneity of female-headed households (Milazzo and van de Walle, 2015). Understanding the routes that a family had towards having a female head is important to characterize this heterogeneity, especially in displacement situations where widowhood and separation from family members are common (Hanmer et al, 2021). As a result, the literature further disaggregates female-headed households to de jure and de facto heads. De facto female heads are households headed by a married female with a husband living either in the household or working outside the family. De jure female heads are households headed by widows, divorced, separated, and single women. Based on the disaggregated analysis, widow-headed households are frequently found to be significantly impoverished (Appleton 1996; Horrell and Krishnan 2007; van de Walle 2013; World Bank, 2001). In contrast, de facto female-headed households who receive transfers from a migrated male member are better off than de jure ones (Chant, 2010; Klassen et. Al, 2015; World Bank 2011). This suggests that distinguishing households headed by de jure and de facto females has the potential to improve the targeting of policies. This is particularly relevant in a forced displacement situation where female-headed households often have higher dependency ratios, fewer able-bodied adults, and more caregiving responsibilities. The experience of children and adolescents growing up in forcibly displaced households also intersects with gender and multidimensional poverty. Displacement can disrupt cultural and gendered norms to present women with new opportunities for household decision-making and leadership (ODI 2019a). Even if gender norms and attitudes do not change on a structural level within IDP households, there is evidence that the behaviors of men and women in these households do indeed shift in response to the pressures that poverty and displacement status create (Assad et al. 2021, Rubiano-Matulevich 2021). Children and adolescents’ long-term exposure to households with a more equalized division of domestic labor, as a result of violent or political conflict, warrants further investigation. 2.3 Country contexts The countries with subnational regions covered in this study, using data from 2017 or 2018, are Ethiopia, Nigeria, Somalia, South Sudan, and Sudan. All are located in Sub-Saharan Africa, have undergone or are currently involved in armed conflict, and are affected by environmental issues such as drought, famine or flooding. Despite some commonalities, each faces a unique set of social, political and economic challenges, which cannot be accurately covered in this study. However, to contextualize the findings, a brief introduction of the country context is presented alongside the poverty estimates by the $1.90/day measure and the global Multidimensional Poverty Index (MPI). 3 3An international measure of acute multidimensional poverty, aligined with the 2030 Agenda, that captures deprivations in health, education, and living standards for more than 100 countries (Alkire and Jahan 2018; Alkire, Kanagaratnam and Suppa 2020). 4 Displacement in Ethiopia has been driven by historical conflicts such as civil wars during the 1970s and 1980s, and more recently by the border conflict with Eritrea and the current crisis in the Tigray region. Droughts and severe weather also contribute to displacement, with over 80% of the population dependent on agriculture for their livelihoods. More than 10 million people were affected by the 2015-16 drought, with lasting problems around food security and subsequent droughts leading to widescale displacement from the worst affected northern and central regions. 4 Climate change, and scarcity of pastures, food and water have escalated displacement and fueled conflict over resources. Many IDPs have moved to cities in search of better opportunities, leading to rapid urbanization and secondary displacements. Although the situation somewhat stabilized in 2019 following a comprehensive peace deal between regions that enabled the return of an estimated 1.2 million IDPs, a new violent conflict in the Northern part of Ethiopia in late 2020 has led to a wave of new displacements. According to estimates, there were over 1.8 million internally displaced people in Ethiopia by the end of 2019 (IDMC 2020), making up roughly 1.6% of the population (World Bank 2020a). According to the latest estimates, 32.4% of the population was living on less than $1.90 per day in 2015 (World Bank 2020b) and 83.5% were multidimensionally poor in 2016 according to the global MPI (Alkire, Kanagaratnam, and Suppa 2020). As of 2016, 25.6% of the population resided in households headed by women. 5 Regarding violence against women and girls, estimates from UN Women show a 40% prevalence of child marriage and 65% of females subjected to genital mutilation or cutting, 6 while nearly a third of women and girls have experienced physical or sexual violence. 7 Besides the large number of internally displaced people, Ethiopia has welcomed refugees fleeing conflict and droughts in surrounding countries. Due to its open-door policy, the country hosted millions of refugees escaping the violence of the war in South Sudan, as well as people fleeing increasing droughts in Somalia (Nguyen 2019) or persecution in neighboring Eritrea. While the country adheres to international law, the rights and freedom of movement among refugees have been limited until the January 2019 revision of the national refugee law, which enables them to access work, primary education, obtain work permits, driver’s license, register legal events such as births and marriages, and access banking. 8 In Nigeria, large parts of the country have seen violence and instability since the emergence of Boko Haram in 2009, which have staged attacks, bombings, and kidnappings in the northern regions of the country. Boko Haram’s terrorist insurgency has also committed targeted violence against women and girls – infamously in the 2014 Chibok schoolgirls’ kidnapping – and instrumentalized women’s social roles and norms in their ideological and strategic frameworks (HRW 2014, Zenn & Pearson 2014). Northern Nigeria has also faced troubles due to the dramatic shrinking of Lake Chad over the last 45 years, leading to large-scale migration and concentration around remaining water sources in the region. As violence heightened under Boko Haram, millions of people left the northern states for other parts of Nigeria, and surrounding countries. Additionally, inter-ethnic conflict and conflict between farmers and semi-nomadic herders have led to further displacement in recent years as violence has 4 https://www.un.org/africarenewal/news/drought-ethiopia-10-million-people- need#:~:text=Ethiopia%20is%20in%20the%20grip%20of%20its%20worst%20drought%20in%20recent%20histor y.&text=Humanitarian%20needs%20in%20Ethiopia%20have,failures%20and%20widespread%20livestock%20deat hs. 5 https://data.worldbank.org/indicator/SP.HOU.FEMA.ZS. 6 https://evaw-global-database.unwomen.org/fr/countries/africa/ethiopia. 7 https://www.unodc.org/easternafrica/en/addressing-violence-against-women-and-girls-in-ethiopia.html. 8 https://www.unhcr.org/uk/news/press/2019/1/5c41b1784/unhcr-welcomes-ethiopia-law-granting-rights- refugees.html 5 escalated. As of December 2019, the number of internally displaced people is estimated to be more than 2.7 million (IDMC 2020), around 1.3% of the total population (World Bank 2020b), while an additional 300,000 Nigerian refugees are hosted in surrounding countries (UNHCR 2020). The World Bank (2020b) estimates that 39.1% of the population was living on less than $1.90 a day in 2018, while 46.4% were estimated to be poor by the global MPI in the same year (UNDP & OPHI 2020). In Nigeria, 18% of the population resided in households headed by women as of 2018. 9 The UN Women database on gender-based violence shows that over 17% of women and girls are subject to lifetime physical and/or sexual intimate partner violence, over 43% of girls are subject to child marriage, and the prevalence of female genital mutilation is over 18%. 10 The total number of IDPs in Somalia as of December 2019 was more than 2.7 million – roughly 17.5% of the total population (World Bank 2020a), almost all of whom were displaced by natural disasters, conflict and violence (IDMC 2020). In 2020, UNHCR reported that floods and droughts explained about 80% of the internal displacement, while conflict explained 18%. 11 The country has seen nearly three decades of recurrent conflict since the collapse of the government in 1991. Recent years have seen the rise of the terrorist militia al-Shabaab, leading to large-scale displacement in the country. Somalia is heavily dependent on international assistance and humanitarian aid, and economic opportunities for the population are scarce. As a result, and due to the impacts of climate change such as increased flooding and severe droughts, the last decade has observed worsening acute food insecurity and famines that have forced many people to leave their homes. Migration of the rural population, who are dependent on agriculture, land and livestock, has led to rapid urbanization and many informal settlements. As more people are fleeing their homes to find refuge in other parts of the country, gender- based violence has seen an increase, with 99% of survivors being female and 76% of all survivors being internally displaced. The most common forms of violence against women include female genital mutilation – prevalent among 98% of the Somali population – as well as forced marriage, rape, and physical, verbal and emotional abuse, among others (UNFPA Somalia 2019). South Sudan seceded from Sudan in 2011, following more than two decades of civil war that led to widespread economic instability. A political conflict in 2013 sparked the beginning of the armed conflict in Juba, which – following a failed peace treaty in 2015 – has since spread to the northern part of the country, forcing millions to flee their homes. Coupled with continued inter-communal conflict over the years in other parts of the country, the violence in South Sudan was followed by acute food insecurity, which further aggravated displacement as people have fled famine. A renewed peace agreement in 2018 has seen a decrease in violent conflict, however as of December 2019, there were still 1.6 million internally displaced people (IDMC 2020), equal to around 14.4% of the total population (World Bank 2020a). In addition, UNHCR (2020) reports nearly 2.2 million South Sudanese refugees and asylum seekers in surrounding countries as of November 2020. According to the most recent monetary poverty estimates for South Sudan from 2009, 44.7% of the population was living on less than $1.90 a day (World Bank 2020b) while in 2010, 91.9% of people were poor by the global MPI (UNDP and OPHI 2020). Sudan has seen violent conflict since its independence in 1956, with two civil wars which have lasted for decades. The War in Darfur erupted in early 2003 between armed groups and 9 https://data.worldbank.org/indicator/SP.HOU.FEMA.ZS. 10 https://evaw-global-database.unwomen.org/fr/countries/africa/nigeria. 11 https://data2.unhcr.org/en/dataviz/1?sv=1&geo=192 6 government forces and has lasted for 17 years until a recent peace treaty in 2020. The war caused tens of thousands of civilian casualties, and millions were forced to flee their homes due to the conflict and general economic instability due to years of fighting that inflated commodity prices. Moreover, much of Sudan is covered by desert and arid land, posing further difficulties as most people rely on agriculture for their livelihoods. Overuse of the land, droughts and flashfloods have contributed to displacement as people escape food insecurity and natural disasters. As of December 2019, there were 2.4 million IDPs in Sudan (IDMC 2020), around 5.6% of the total population (World Bank 2020a). Sudan also has one of the largest refugee populations in Africa, with many people from surrounding countries such as Ethiopia and South Sudan, and even as far as the Syrian Arab Republic and the Republic of Yemen, seeking safety from violence and conflict. As of 2014, 12.2% of the population was living on less than $1.90 a day (World Bank 2020b) and 52.3% were MPI poor in the same year (UNDP and OPHI 2020). Widespread displacement and poverty following decades of conflict has also contributed to gender-based violence, particularly against the most vulnerable population groups such as IDP girls and women. According to the UN Women database, the prevalence of child marriage in Sudan is 34%, while 87% of the population are subject to female genital mutilation. 12 3. Methodology 3.1 Alkire-Foster method The Multidimensional Poverty Index (MPI) constructed for this analysis is based on the Alkire- Foster (AF) method of multidimensional poverty measurement (Alkire and Foster 2011). The AF method allows for the construction of individual- and household-level deprivation profiles that can then be used to identify multidimensionally poor people. There are three key statistics for any MPI: incidence or headcount ratio (H), which is the proportion of the population who are multidimensionally poor; intensity (A), which is the average share of weighted deprivations experienced by the poor; and the MPI or adjusted headcount ratio, which is the product of the incidence and intensity (MPI = H × A). The AF method uses a dual-cutoff counting approach to poverty measurement. First, households (or individuals) are classified as deprived or non-deprived according to deprivation cut-offs, specific to each indicator. The deprivations for each individual are weighted and aggregated into a weighted deprivation score. Next, a cross-dimensional poverty cut-off k is used to identify people with a deprivation score equal to or greater than the threshold as multidimensionally poor. The MPI can be broken down by indicator to show the composition of multidimensional poverty, a feature of dimensional detail that brings added policy relevance to the analysis. The censored headcount ratio is the percentage of the population that is both poor and deprived in a given indicator. The MPI can be calculated by adding up the weighted censored headcount ratios of each indicator: MPI = sum [(ℎ)] for all , where add up to 1. The percentage contribution of a given indicator to the MPI is calculated as the indicator’s censored headcount ratio, multiplied by its weight, and divided by the MPI. In addition, the MPI can be disaggregated by different population groups, such as urban or rural areas, age groups, subnational regions or, in the case of this work, host communities and 12 https://evaw-global-database.unwomen.org/fr/countries/africa/sudan 7 forcibly displaced communities. Formally, the AF method can be used to analyze differences by different population subgroups (e.g. displaced and non-displaced) by having the value of the of society = {1,…, }, denoted as (), which can be disaggregated by ℓ mutually exclusive and exhaustive subgroups ℓ = 1, …, as: ℓ ℓ ( ) = � ( ) ℓ ℓ where is the population in society , and � � denotes the of subgroup ℓ in society ℓ with a population sized . For notational convenience, we omit the parameters of the poverty identification function in the above equation to highlight on which data a particular estimate depends. The equation above states that society-level can also be obtained as a population weighted average of the disaggregated subgroup-specific s. In turn, can be disaggregated following the same procedure. Moreover, can also be disaggregated in a similar way replacing the society and subgroup population sizes by the number of poor people in the corresponding levels. Most importantly, the AF method allows flexibility around the selection of dimensions, indicators, deprivation cut-offs, weights and poverty cut-offs to reflect different contexts and priorities. This enables users to create poverty measures tailored to the specific needs and conditions of the population of interest. As such, it is possible to create global, regional, national or even state level multidimensional poverty indices, as well as ones designed specifically for children, women, or in this case, forcibly displaced populations. 3.2 Structure of the measure Multidimensional poverty measures can cover a range of non-monetary areas such as education, health, employment, living standards, or financial security. The design of the MPI for forcibly displaced communities (shown in Table 1) was guided by discussion with experts in both multidimensional poverty and forced displacement, but constrained by the available data. The final structure was agreed following the construction of five trial measures, and includes the same three dimensions as the global MPI (and Human Development Index), with an additional dimension of financial security. The MPI measure analyzed in this paper includes a total of 15 indicators that cover important non-monetary aspects of poverty among displaced populations. Given the importance of education in reducing the intergenerational transmission of poverty and increasing the likelihood of employment, two indicators were included to capture educational attainment in households: i) completed years of schooling among adolescents and adults, and ii) school attendance among primary school age children. Related to the presence of universal basic provisions, we include indicators to capture lack of access to clean drinking water and sanitation (SDG 6), inadequate housing (SDG 11), access to electricity (SDG 7), clean fuel for cooking (SDG 7), and safe disposal of household waste. Such basic conditions are often not available to refugees or IDPs living in temporary accommodation or settling in places outside their place of origin. Many of the living standards indicators also proxy other conditions such as indoor air pollution, respiratory illness as a result of unclean fuels, transmission of diseases due to poor sanitation, waste management, overcrowding or housing conditions, that were not included due to the lack of sufficient data. 8 Besides capturing basic deprivations associated with acute poverty, the MPI for forcibly displaced communities includes additional indicators specific to the problems faced by IDPs and refugees. Unemployment is included as people forcibly displaced often face multiple barriers to entering the labor market due to the inability to obtain legal identification or work permits, as well as language or cultural barriers. While unemployment is not a perfect measure, data on informality, subemployment or sub-optimal working conditions, which are also common among displaced populations, was not available. IDPs and refugees often lack documentation to obtain access to basic services, thus the MPI includes two indicators to reflect such deprivations: legal identification, and access to banking. The cutoffs were set to ensure that households have at least one member with legal identification or access to banking who can help provide access to employment, social assistance and services such as health care and education. Food insecurity is a key indicator of deprivation among IDP and refugee communities, causing possible adverse impacts on health and well-being of individuals, especially children under the age of 5, as well as direct deprivations of inadequate foodstuffs. Additionally, given the interest in gendered dynamics of forced displacement and poverty, the MPI includes indicators on deprivations affecting women, such as pregnancy care (combining information on prenatal care and assisted delivery) and early marriage, and indicators with a gendered dynamic such as physical safety. The indicator on early marriage was considered an important addition, with young girls being increasingly vulnerable as a result of displacement, and due to the close links with early pregnancy, both of which can limit opportunities for education and employment, and increase reliance on family ties. Some additional indicators, such as underemployment, access to roads and transportation, land and livestock ownership and internet access, were also considered, but following conceptual discussions, review of the available data, and redundancy analysis, they were dropped from the list of indicators. The deprivation cut-offs identify individuals as deprived or non-deprived according to set criteria; however, for some of the indicators (years of schooling, school attendance, pregnancy care, early marriage and unemployment), data is only collected from a subset of the sample, such as women with young children or household members aged 15 and above. Eligibility for the MPI indicator is thus determined based on this available subsample, while the size of eligible populations is considered carefully in indicator selection. For MPI indicators with a subsample, households with no eligible members (who were asked the question) are considered non-deprived (see Alkire et al 2015, ch.7). For example, a household with only male family members cannot lack pregnancy care, thus, they are not deprived in that dimension. Table 1: Structure of the Measure Dimension Indicator Household is Deprived if… Weight Education Years of Schooling No eligible household member has completed at least 6 1/8 years of schooling 13 School Attendance Any child of primary school age is not attending school up 1/8 to class 6. Health Food Security In the past 7 days, there was ever a time when there was not 1/16 enough food or money for food 13Eligibility is determined by primary school starting age in the country. Those aged 6 years or older than the starting age are considered to be eligible. For starting the age in each country, see http://uis.unesco.org/. 9 Pregnancy Care A woman who gave birth in the last 2 years did not visit a 1/16 clinic while pregnant or have a trained assistant during delivery Physical Safety Any member feels unsafe at home or walking alone 14 1/16 Early Marriage A member was married before age 19 1/16 Living Garbage Disposal Main method of solid waste disposal is dumping, burying in 1/24 Standards own compound, burning, or other Drinking Water Main source of drinking water is unsafe, or it takes more 1/24 than 20 minutes (round-trip) to get water 15 Electricity It does not have electricity 1/24 Cooking Fuel Main energy source for cooking is solid fuels 1/24 Housing It is an unimproved housing type 1/24 Sanitation Main toilet facility is unimproved, or shared with other 1/24 households 16 Financial Unemployment Any member 15 or older is unemployed and looking for 1/12 Security work 17 Legal Identification No member has a form of legal identification 1/12 Bank Account No member has a bank or mobile money account 1/12 The MPI presented here uses equal nested weights with all four dimensions considered to be equally important, and all indicators within a dimension receiving an equal share of the total weight. The cross-dimensional poverty cut-off is defined as k=50%, with those deprived in half or more of the weighted indicators identified as multidimensionally poor. 4. Data Data on forcibly displaced populations are scarce, with many household surveys excluding refugees and IDPs from the sample framework. To ensure that MPI results are representative of these communities and that they can be disaggregated for comparative analysis, an initial review of possible data sets was conducted. Feasibility was determined based on the availability of sufficient sample sizes for forcibly displaced persons for quantitative analyses, as well as inclusion of many of the indicators (on health, education, living standards, etc.) 14 A household is deprived if the respondent reports feeling moderately or very unsafe when alone at home, walking alone after dark, or walking around during the day. In Sudan, the indicator on the ‘feeling safe from crime and violence when at home’ was not available, and the indicator only considers answers to the questions on safety when walking alone. 15 Unprotected dug well, unprotected spring, carts with tank, tanker-truck, surface water, or other are considered as unsafe water sources according to international guidelines. See https://washdata.org/monitoring/drinking-water. 16 Pit latrine without slab, bucket, hanging toilet, and no facility (open defecation) are considered as unimproved sanitation facilities according to international guidelines. See https://washdata.org/monitoring/sanitation. 17 According to the ILO definition, those who did not participate in employment in the last four weeks (and have no work to return to) are actively looking for work and are available to start, or those currently waiting to start work are classed as unemployed. See https://www.ilo.org/ilostat-files/Documents/description_UR_EN.pdf. 10 usually used in MPIs, or which were determined as relevant to this context. One advantage of the chosen surveys was that they were intended to look at forced displacement, so they oversampled that population. After review, five surveys were selected that best suited the proposed analysis (see Table 2). 18 It is essential to note that the classification of “IDP” and “non-IDP/host community” of the survey respondents varies by country. In Ethiopia and Sudan, the comparator groups include refugees and IDPs living in camps: in Ethiopia, refugees in camps, and their host communities within a 5-km radius; in Sudan, IDPs in Abu Shouk and El Salaam camps and host communities in the nearby city of Al-Fashir. In Nigeria, the comparator groups refer to the IDPs and host communities (non-displaced population in the enumeration areas) of six Northeastern states. In Somalia, the sample is representative of the whole Somali population living in secure areas and covers the IDP sample, which considers households living inside and outside settlements, and the non-displaced sample (which, unlike the other four surveys, does not require the household to be in the vicinity of the displaced sample households). In South Sudan, the sample covers IDP and host communities in urban areas of seven of South Sudan's ten prewar states. Data was also collected on reason for displacement, albeit with differences across surveys. In Ethiopia, refugees in camps identified armed conflict, increased violence, crime and insecurity, ethnic/political/religious violence and drought, famine, and floods as the main reason for displacement. In Nigeria, the majority of respondents indicated armed conflict between Boko Haram and armed forces in their village or nearby as the most important reason for displacement, although some migrated due to lack of access to home, land or livestock. In Somalia, new respondents who reported being IDPs were asked about why they left their permanent residence. Most fled due to drought, famine, flood or other environmental issues, or armed conflict, violence, and increased insecurity in their area of residence. This was also the case in South Sudan. In Sudan, the majority indicated lack of safety as the most important reason for displacement, with lack of access to education services and health services also being an important reason. Table 2: Summary of the Datasets Country Survey Sample Design Geographical coverage Population coverage Retained sample Ethiopia Skills Multi-stage Refugee camps and Refugees of four main 26,517 Profile stratified proximity in Tigray Afar, nationalities (Eritrean, (96.5%) Survey random sample Gambella, Benishangul South Sudanese, (2017) Gumuz, and Somali regions Sudanese, and Somali) N.E IDP Multi-stage Six Northeastern states IDPs and host 17,543 Nigeria Survey stratified (Adamawa, Bauchi, Borno, communities (97.7%) (2018) random sample Gombe, Taraba, and Yobe) Somalia High Multi-stage Somalia (within secure IDPs and host 27,287 Frequency stratified areas) communities (82.3%) Survey random sample (2017) 18For more information on survey coverage, design and the data for each survey, see https://microdata.worldbank.org/index.php/catalog. 11 Sudan IDP Stratified cluster Abu Shouk and El Salam IDPs and host 17,645 Profiling sampling camps, and neighboring and communities (95.2%) Survey non-neighboring Al-Fashir (2018) South High Stratified two- Urban areas of seven of the IDPs and host 4,554 Sudan Frequency stage cluster ten pre-war states (Western communities (92.8%) Survey design Equatoria, Central Wave 4 Equatoria, Eastern (2017) Equatoria, Northern Bahr- El-Ghazl, Western Bahr-El- Ghazal, Warrap, Lakes state). Data was available for all selected MPI indicators in every country except South Sudan, where the question on access to a bank account was not included in the survey. The MPI for South Sudan is thus constructed without this indicator, and the remaining indicators in the financial security dimension are re-weighted accordingly, receiving 1/6 of the weight each. 5. Findings 5.1 Key findings The focus of this study is to analyze the level and composition of multidimensional poverty among forcibly displaced communities across countries and compared to host communities, respecting each of the survey designs. Table 3 shows the headline figures for the MPI disaggregated by displacement status. Overall, the highest level of poverty can be found in the Somalia population covered, where over 63% of IDPs and nearly 45% of the host community are identified as multidimensionally poor, while the lowest overall poverty is in the population covered in N.E Nigeria, with 23% of IDPs and 17% of non-IDPs suffering from poverty. As expected, many countries show large inequalities, with displaced persons being poorer across all five countries according to MPI and incidence of multidimensional poverty, with statistically significant result for all countries but Nigeria and South Sudan. In addition, there is large variation in the level of poverty between the two groups across countries. For instance, the difference in the incidence between displaced and host communities ranges from 6 percentage points in Nigeria (not significant) to 15 and 19 percentage points in South Sudan and Somalia, and to over 30 percentage points difference in Ethiopia and Sudan, where 46% of refugees and 44% of IDPs are poor compared with 13% and 9% of the host community, respectively. Interestingly, we find that the intensity of poverty (average share of weighted deprivations experienced by the poor) is similar across displaced and non-displaced populations in all countries, with the only significant difference between IDPs and host communities appearing in Nigeria. Table 3: Main results disaggregated by displacement status Country Displacement status MPI Confidence Incidence Intensity Population interval (H) (A) share (sample) Ethiopia Host Community 0.073 (0.057, 0.090) 12.8% 57.3% 12.0% Refugee 0.270 (0.233, 0.307) 45.5% 59.4% 88.0% 12 N.E Nigeria Host Community 0.107 (0.045, 0.169) 17.0% 62.7% 71.5% IDP 0.139 (0.103, 0.175) 23.3% 59.8% 28.5% Somalia Non-IDP 0.286 (0.245, 0.327) 44.8% 63.8% 64.3% IDP 0.396 (0.337, 0.454) 63.3% 62.5% 35.7% South Sudan Host Community 0.119 (0.057, 0.180) 19.5% 60.9% 86.7% IDP 0.203 (0.004, 0.402) 34.1% 59.5% 13.3% Sudan Host Community 0.054 (0.033, 0.074) 9.4% 57.0% 67.5% IDP 0.255 (0.232, 0.277) 43.6% 58.5% 32.5% Note: Own elaboration based on data from the High Frequency Surveys of Ethiopia (2017), Nigeria (2018), Somalia (2017), South Sudan (2017) and Sudan (2017). Another important trend concerns the Ethiopia and Sudan samples, which surveyed refugee and/or IDP households in camps, whereas the Nigeria, Somalia, and South Sudan data sets surveyed IDP populations distributed throughout the country. Of the five countries, Ethiopia and Sudan have the largest differences in MPI and indidence between the refugee/IDP populations and their host communities. Ethiopia has a nearly 33 percentage point difference in incidence between refugee households in camps and host community peers, while Sudan has a 34 percentage point difference in incidence between the IDP households in camps and their host community neighbors. Curiously, only in Ethiopia and Sudan do we see higher intensities of poverty among the refugee and IDP populations, whereas in Nigeria, Somalia and South Sudan, we observe, on average, a higher intensity of deprivations among the poor in the host community populations. More, the included areas of Ethiopia and Sudan also have the lowest rates of multidimensional poverty for the host communities, with incidences of 12.8% and 9.4%, respectively. Delving further into the depth of poverty, Table 4 shows a breakdown of the poor population by intensity, ranging from 50% to 100%. In Sudan, 72% of the poor in host communities and 62% of the poor IDPs are in the lowest intensity band, relatively close to the poverty line. In Somalia, the country with the highest average intensity across both groups, the distribution is less concentrated, with only one-third of IDPs in the lowest band. Nearly 18% are deprived in 70% to 79% of the weighted indicators, and a further 6% of poor IDPs face deprivation in 80% or more of the weighted indicators. Table 4: Intensity of deprivation among MPI poor by displacement status (%) Ethiopia N.E Nigeria Somalia South Sudan Sudan Intensity Host Refugee Non-IDP IDP Non-IDP IDP Non-IDP IDP Non-IDP IDP 50-59% 65.5 57.3 59.2 49.1 45.8 34.0 54.8 57.0 71.6 61.8 60-69% 31.0 34.2 35.0 35.7 33.6 42.4 21.4 36.7 21.6 29.7 70-79% 3.0 7.4 5.9 13.6 17.1 17.6 21.3 6.3 6.8 7.8 80-89% 0.5 1.1 1.7 3.1 5.6 2.6 0.8 90-100% 0.3 0.4 13 Source: Authors’ calculations based on data from the High Frequency Surveys of Ethiopia (2017), Nigeria (2018), Somalia (2017), South Sudan (2017) and Sudan (2017). 5.2 Composition of poverty Unpacking the headline numbers further, important patterns emerge about the composition of multidimensional poverty among forcibly displaced and host communities in these countries. Overall, the censored headcount ratios (proportion of people who are poor and deprived in a given indicator) are lower among non-displaced communities than among refugees and IDPs, but there are large differences in which indicators are the most salient in different countries. The indicators with the largest difference between the two populations are bank account and cooking fuel in Ethiopia, years of schooling in Somalia, electricity in Sudan, drinking water in South Sudan, and legal identification in Nigeria. These findings reinforce the need for policies and programming that take into account the measured experiences of IDPs and refugees. In this way, the MPI can function both as tool to monitor, track, and bear witness to the lived experiences of forcibly displaced communities, as well as advise on evidence-based interventions that address the needs of the local population. Figure 1 shows the censored headcounts of each indicator in Sudan’s MPI, with large differences appearing by displacement. MPI of IDP communities is considerably higher than that of the host communities, so it follows that the censored headcount ratios display a similar gap. The difference by indicator is particularly noticeable in the living standards dimension, where the electricity, cooking fuel, housing, and bank account indicators show over 34 percentage point difference between the censored headcount ratios for the IDP and host communities. Figure 1. Censored headcounts of each indicator in the MPI, by displacement status in Sudan (2018) 14 Source: Authors’ calculations based on data from the High Frequency Surveys of Ethiopia (2017), Nigeria (2018), Somalia (2017), South Sudan (2017) and Sudan (2017). As the Sudanese sample surveys IDPs living in the Abu Shouk and El Salam camps, we must understand these deprivations with the background that these settlements were created as emergency and crisis responses rather than durable, long-term solutions (Sudanese Government’s Joint Mechanism for Durable Solutions 2019). Although Sudan does have a national electric grid that supplies electricity to the urban and peri-urban areas of the nearby city El Fasher, IDP communities living in the camps report limited connection to the city’s electricity supply, reflected in the high deprivations in the electricity and cooking fuel indicators. The ad-hoc construction of dwellings in the two camps explains why 71% of the IDP households in Abu Shouk and 65% in El Salam live in tukuls or other permanent mud or wood structures (Sudanese Government’s Joint Mechanism for Durable Solutions 2019: p.50), both of which register as unimproved housing types. Bank branches in El Fasher have limited capital for small businesses as their headquarters in Khartoum regard the area as too great a risk and IDPs themselves as riskier investments than their host community peers (UN- HABITAT 2009: p.8). Exposure to unclean cooking fuels and inadequate housing can lead to poor health outcomes, while lack of access to electricity and a bank account further excludes individuals from labor market integration and livelihood opportunities that would empower forcibly displaced persons to overcome their multiple, overlapping deprivations. Clearly, displacement status puts individuals at a greater risk of poverty than their host community neighbors, and we can unpack those risks in greater detail using the MPI. Results can also be broken down to show the percentage contribution of each indicator to multidimensional poverty (see Figure 2). Among refugees in Ethiopia, lack of a bank account is the largest contributor to poverty, while among host communities, the largest contributor is years of schooling. While in absolute terms, the proportion of refugees who are MPI poor and experience deprivations in years of schooling is 27.9%, compared to the host community’s smaller proportion of 10.0%, the indicator contributes more to poverty among the host community population, due to the weighting structure of the index and the lower levels of deprivation among the other indicators. This distinction reveals the complex, varied, and overlapping deprivations experienced by the poor when displacement status is taken into account – a complexity that is otherwise overlooked in monetary headline figures. Physical safety, early marriage, and legal identification are also larger contributors to poverty among refugees than among host communities. The gap in deprivations of physical safety by displacement status is particularly stark; in absolute terms the proportion of refugees who are MPI poor and experience deprivations in physical safety is 30.2%, compared to the host community’s near-zero proportion of 0.9%. Figure 2. Percentage contribution of each indicator to the MPI, by displacement status in Ethiopia (2017) 15 Source: Authors’ calculations based on data from the High Frequency Surveys of Ethiopia (2017), Nigeria (2018), Somalia (2017), South Sudan (2017) and Sudan (2017). 6. Gender differences in multidimensional poverty Next we examine differences in multidimensional poverty outcomes by the gender of the household head. Existing literature points out the limitation of household level MPI analysis in masking the intrahousehold distribution of deprivations, and thus being less sensitive to gender based differences in individual outcomes within the family unit, which might lead to underestimation of inequality and gender gaps (Espinoza-Delgado and Klasen 2018; Franco 2017; Klasen and Lahoti 2020, Rodriguez, 2016). However, as the MPI identifies poverty at the household level, our initial analysis focuses on disaggregated results by the gender of the household head. 19 We acknowledge that this approach has several limitations since most women reside in male-headed households, and the composition of households can change after displacement due to separation of family members, and widowhood. Regardless, the analysis at the household level remains relevant given the high prevalence of female-headed households that emerge after displacement, with the analysis showing large differences across countries between households based on the gender of the head. Following Hanmer et al (2021), we also show the results disaggregated by type of female head and by households’ earnings profile, building on other work detailing the interaction between female headship and marital status (Brown and van de Walle 2020). Analysis of the individual-level and intrahousehold differences among forcibly displaced and host communities living in multidimensional poverty is detailed in subsequent papers. 20 Figure 3 shows the MPI for people living in female- and male-headed households across the populations surveyed in all five countries. Those in female-headed households are significantly more likely to be poor compared to those in households headed by males in Ethiopia, South 19 That MPIs are generally household-level – and indeed, as this one is – is largely due to data constraints (Klasen and Lahoti 2020). 20 See forthcoming paper by Adamasu, Alkire, and Scharlin-Pettee (2021). 16 Sudan and Sudan, while the opposite is true for Somalia, where people living in male-headed households have a higher MPI. In Nigeria, there is no statistically significant difference between the poverty levels of the two groups. These differences also hold for incidence of multidimensional poverty, where female-headed households are 39 percentage points poorer in Ethiopia, and 17.6 and 10.3 percentage points poorer in South Sudan and Sudan, respectively. Figure 3. MPI for male- and female-headed households by country 0.450 0.400 58.5% 0.350 61.7% 0.300 41.5% 0.250 42.1% 0.200 35.7% 65.9% 0.150 38.2% 64.3% 0.100 34.1% 58.0% 0.050 0.000 Ethiopia Nigeria Somalia South Sudan Sudan MPI Male-headed household Female-headed household Source: Authors’ calculations based on data from the High Frequency Surveys of Ethiopia (2017), Nigeria (2018), Somalia (2017), South Sudan (2017) and Sudan (2017). Note: Population shares of headship included above each bar. With the above results, it is important to consider the gender differences by displacement status as the gender of the household head may be correlated with disruption in family structures that occurred after displacement. Looking at the sample composition by headship at the country level, 45% of the population in Ethiopia, 34.3% in Nigeria, 49.8% in Somalia, 44.7% in South Sudan, and 41.6% in Sudan live in female-headed households. Disaggregation by displacement status shows that there is a higher proportion of the population living in female-headed households in displaced communities in Ethiopia (51.4 % of refugees vs 32.2% of hosts), South Sudan (53.3% of IDPs vs 43.6% of non-IDPs), and Sudan (47% IDPs vs 30% non-IDPS). Table 5 shows the MPI, incidence (H) and intensity of poverty (A) by the gender of the household head and displacement status for the five countries. Disaggregating the household gender differences by displacement, we find interesting results across four countries. In Ethiopia, the observed gender difference that hurts female-headed households is driven by the difference in MPI among refugees. The difference in MPI between displaced households is largest in Ethiopia, with an MPI of 0.355 for those in female-headed refugee households and an MPI of 0.111 for those in male-headed refugee households. Further, there is a 40 percentage point difference in the incidence of poverty among refugees, with 57.9% of those in female- headed households and only 19.1% of those in male-headed households being multidimensionally poor. Interestingly, there is no statistically significant difference between the poverty levels of the two groups among host community households. In South Sudan, individuals living in female-headed households are more likely to be poor compared to their male-headed counterparts both in IDP and non-IDP households. Female- 17 headed IDP households are worse off than their male-headed counterparts with regards to the incidence of poverty by about 29.2 percentage points, although weakly significant. The corresponding difference in incidence between the two groups is 15.1 percentage points among non-IDP households. In Somalia, we find significant differences in MPI and incidence of poverty between female- and male-headed households among both IDPs and non-IDPs. The differences observed are in favor of female-headed households. Looking at the incidence of poverty, there is a 17.2 percentage points difference among IDPs with 70.7% of those in male- headed households and 53.6% of those in female-headed households being MPI poor. Similarly, there is a 19.9 percentage points difference among non-IDPs in Somalia, with 52.8% and 32.9% of those in male- and female-headed households live in multidimensional poverty, respectively. In Sudan, considering the whole population, the results indicate the presence of a statistically significant difference by gender of the head in incidence of multidimensional poverty (10.3 percentage point) and MPI (0.063). However, disaggregating the results by displacement status reveals no significant difference between male- and female-headed households among both IDPs and non-IDPs. Finally, individuals in Nigeria have no significant difference in their poverty incidence, intensity, or MPI based on the gender of the household head. Although the difference is insignificant, we can see that people living in female-headed IDP households have a higher MPI (0.171) compared to their male-headed IDP counterparts (0.138), while the opposite is true for non-IDPs in Nigeria, where female-headed households have a lower MPI (0.080) than male-headed households (0.144). The finding that individuals living in female-headed displaced households are more likely to be multidimensionally poor raises questions related to the gender composition of the household. The sex ratio 21 of female-headed households is lower than that of male-headed households among refugees in Ethiopia and IDPs in South Sudan, suggesting a higher proportion of females are residing in female-headed displaced households in the two countries. This further suggests that a higher share of women and girls are identified as poor in households headed by females. Table 5: Multidimensional Poverty outcomes by gender of the household head and displacement status Refugees/IDPs Hosts/Non-IDPs All HH headed by HH headed by HH headed by Female Male Diff. Female Male Diff. Female Male Diff. MPI Ethiopia 0.355 0.111 0.244*** 0.073 0.058 0.014 0.341 0.099 0.242*** N.E Nigeria 0.171 0.138 0.033 0.080 0.144 -0.065 0.107 0.143 -0.035 - - - Somalia 0.335 0.443 0.108*** 0.206 0.34 0.134*** 0.254 0.375 0.121*** South Sudan 0.298 0.119 0.179* 0.183 0.084 0.099*** 0.201 0.088 0.113*** Sudan 0.262 0.230 0.033 0.080 0.046 0.034 0.158 0.095 0.063*** H (%) Ethiopia 58.1 19.1 39.1*** 12.7 10.0 2.7 55.9 17.0 38.9*** 21 The sex ratio is obtained by dividing the number of males by the number of females in the household. 18 N.E Nigeria 29.0 22.5 6.5 14.0 22.5 -8.6 18.5 22.5 -4.0 Somalia 53.6 70.7 -17.2** 32.9 52.8 -19.9*** 40.7 59.0 -18.3*** South Sudan 49.7 20.5 29.2* 29.3 14.2 15.1*** 32.5 14.9 17.6*** Sudan 44.7 39.5 5.2 13.7 8.3 5.4 27.0 16.6 10.3*** A (%) Ethiopia 61.1 58.1 3.0*** 57.4 58.6 -0.013 61.1 58.1 2.9*** N.E Nigeria 59.0 61.2 -2.2 57.1 64.1 -7.0 58.0 63.3 -5.3* Somalia 62.5 62.6 -0.1 62.4 64.4 -2.0** 62.5 63.0 -1.3 South Sudan 59.9 58.3 1.7 62.5 59.1 3.4** 61.9 59.0 2.9** Sudan 58.7 58.2 0.6 58.2 55.2 3.0 58.6 57.1 1.5* Source: Authors’ calculations based on data from the High Frequency Surveys of Ethiopia (2017), Nigeria (2018), Somalia (2017), South Sudan (2017) and Sudan (2017). H is the incidence of multidimensional poverty; A is the intensity of multidimensional poverty and MPI=H*A. Asterisks indicate statistical significance at 1% ***, 5% ** and 10% * levels. The results reveal that in all countries, female-headed IDP/refugee households have a higher incidence of poverty and MPI than female-headed households that are not displaced. We also note the emerging pattern that, for female-headed households in camps (Ethiopia, Sudan) multidimensional poverty appears higher and more intense than for their female-headed counterparts living outside camps. In Ethiopia and Sudan, there is a 28 and 18 percentage point difference in incidence between female-headed households in camps and in the host community, respectively, whereas the differences between the female-headed IDP households and female-headed host community households of Nigeria, Somalia, and South Sudan are 12 percentage points or smaller. This finding among female headship is particularly notable in Sudan, considering the fact that its neighbor South Sudan sees overall bigger differences in magnitude of multidimensional poverty between male- and female-headed households within the country overall, as well as when disaggregated by displacement status. This suggests that, in determining a household’s level of multidimensional poverty, the gender of the household head interacts not only with displacement status, but type of displacement status (e.g. living in camps or resettled) and that our multi-country comparison succeeds in capturing the diversity of IDP experiences. Looking at the breadth of poverty, Figure 4 reveals that male- and female-headed households in Somalia are likely to suffer a similar intensity of poverty (average deprivation share among the poor), which suggests that the observed difference in MPI is generally driven by the difference in the incidence of poverty. In contrast, there is a significant difference in intensity between the two groups in Ethiopia and South Sudan, with only 46% of the poor female-headed households being in the lowest intensity band in the latter, compared to 63% of male-headed households. More, in South Sudan, nearly 24% of the multidimensionally poor female-headed households are deprived in 70% or more of the weighted indicators, while only 17% of poor male-headed households have such high intensity. The direction of the overall difference in Ethiopia and South Sudan can be explained by the difference in the share of poor households with an intensity of 70% or more by female- versus male-headed households, with a higher proportion among the former (9.5% vs 3.5% in Ethiopia, and 23.5% vs 17% in South Sudan). Figure 4: Distribution of intensities among MPI poor male- and female-headed households: by country 19 Ethiopia N.E Nigeria Somalia South Sudan Sudan 60 60 60 60 60 Perecentage of Multidimensionally Poor Perecentage of Multidimensionally Poor Perecentage of Multidimensionally Poor Perecentage of Multidimensionally Poor Perecentage of Multidimensionally Poor 50 50 50 50 50 40 40 40 40 40 30 30 30 30 30 20 20 20 20 20 10 10 10 10 10 0 0 0 0 0 90-100% 50-59% 60-69% 70-79% 80-89% 50-59% 60-69% 70-79% 80-89% 50-59% 60-69% 70-79% 80-89% 50-59% 60-69% 70-79% 80-89% 50-59% 60-69% 70-79% 80-89% Male-headed HH Female-headed HH Source: Authors’ calculations based on data from the High Frequency Surveys of Ethiopia (2017), Nigeria (2018), Somalia (2017), South Sudan (2017) and Sudan (2017). Considering the relative contribution of each indicator to multidimensional poverty between the two types of households, we see that being deprived in the indicators of bank account and school attendance in Ethiopia; bank account and years of schooling in Sudan; years of schooling and school attendance in Nigeria; and legal identification and school attendance in South Sudan are the main contributors of poverty to both types of households. In South Sudan, early marriage contributes more to poverty in female-headed households, while school attendance contributes more to poverty in male-headed households. In Nigeria, the relative contributions of early marriage and lack of legal identification are greater in female-headed households. The source of deprivation is slightly different for male- and female-headed households in Ethiopia, where lack of physical safety, early marriage, and lack of legal identification contribute more to female-headed households, while years of schooling and school attendance are much larger contributors to poverty in households headed by men. Table 6 refines the above analysis by looking within displaced households, and presenting data at the individual level for selected indicators. This is important as household level deprivations can mask inequalities within the household unit (for instance, a household where four children are attending school and one is not would be considered deprived, similar to households where no children are in school). Looking at the three countries, years of schooling and early marriage are key indicators in analyzing the gender differences in poverty among forcibly displaced individuals. with a significantly higher percentage of displaced females deprived in years of schooling compared to males. In terms of marriage, females are significantly more likely to get married at early age compared to males. The prevalence of child marriage in South Sudan and Sudan is particularly high with 75% and 50% of displaced female individuals getting married before they turn 18, respectively, which is significantly higher than the national pre-conflict average of 45% in South Sudan (Buchanan, 2019) and 38% in Sudan (Nagar et. al., 2017). While child marriage has been common in these countries, the extended conflict might have interacted with this practice as child marriage is seen as a negative coping mechanism for families to rising poverty and food insecurity, and protection against sexual violence (Buchanan, 2019). This often directly affects girls’ educational attainment as many drop out of school, due in part to early marriage. 20 Table 6: Percentage of displaced individuals deprived in selected indicators by gender Ethiopia South Sudan Sudan Male Female Male Female Male Female Years of schooling 55 78*** 36 63*** 32 46*** School attendance 16 19** 21 29 23 23 Early marriage 3 13*** 8 75*** 6 50*** Unemployment 7 5*** 2 0* 3 3 Legal id 45 46 48 74*** 10 10 Source: Authors’ calculations based on data from the High Frequency Surveys of Ethiopia (2017), Nigeria (2018), Somalia (2017), South Sudan (2017) and Sudan (2017). Asterisks indicate statistical significance of mean differences between male and female at 1% ***, 5% ** and 10% * levels. Returning to patterns of household headship, Figure 5 breaks down the variation in censored headcount ratios among refugee households in Ethiopia, depending on the gender of the household head. The largest differences between households headed by men and women is for early marriage (42.3 p.p.), waste disposal (37.2 p.p.), and housing (37.1 p.p.), whereas the pregnancy care and school attendance have the smallest differences (3.3 p.p and 3.0 p.p.). Considering that girls and women of reproductive age are more likely than their male counterparts to live in income-poor households below the international poverty line (Boudet et al. 2018), the interaction of multidimensional poverty, gender of the household head, and displacement status in Ethiopia cannot be neglected. The observed gender difference disadvantaging female-headed households in terms of the MPI and incidence in Ethiopia derives from the difference among refugees. The breakdown of indicators here opens up questions for policy makers to ask how and why these differences in waste disposal and housing are occurring for households with different genders in headship, given that both contemporaneous features of displacement rather than indicators such as early marriage that denote deprivation which may have occurred before or as part of displacement. Thus, to see this interaction further, we extend our analysis by type of female heads and household earnings profile. Figure 5. Censored headcount ratios of refugee households in Ethiopia, by sex of the household head 21 Source: Authors’ calculations based on data from the High Frequency Surveys of Ethiopia (2017), Nigeria (2018), Somalia (2017), South Sudan (2017) and Sudan (2017). To uncover the main drivers of the observed gender based differences in multidimensional poverty at country level, we study the absolute contribution of the gender difference in each indicator to the overall household gender gap (see Figure 6), calculated as the difference between the censored headcount ratio for males versus females. We find that in Ethiopia, the gender gap that disadvantages female-headed households is mostly driven by the difference in financial insecurity measures (lack of legal ID and bank account) and health measures (early marriage, physical safety, and food insecurity), which is further reinforced by the differences in the living standard and education measures. Female-headed refugee households are more food insecure, live in unimproved housing, have lower access to electricity, are more likely to be married at an early age, and have lower access to legal identification and a bank account. In South Sudan, gender gap that disadvantages female-headed households is mainly explained by the differential in the financial insecurity and health measures, but cumulative gaps in the living standard and education indicators also contribute to the overall gap at the household level. In South Sudan, IDPs in general do not feel safe in camps, in particular female-headed households are more likely to feel unsafe. In addition, early marriage and lack of legal ID are issues among female-headed households. Figure 6: Absolute contribution of the gender gap in each indicator to the overall gap, by country 22 .03.01 .02 Absolute contribution -.01 0 -.02 g el l re y n y nt e er g ce d ty sa t t irt in tio en li g n fu fe ci u ca at an ria si po ga ol cu co tri ta sa m w ou g ho nd y ar ec is Se Le ni n ac oy nc g al H ki D m Sc Sa n tte El pl oo ic na nk ki od te rly m a ys rin of C Ba eg as Fo ne ol Ea Ph D s Pr W ho ar U Sc Ye Ethiopia Somalia South Sudan Sudan Source: Authors’ calculations based on data from the High Frequency Surveys of Ethiopia (2017), Nigeria (2018), Somalia (2017), South Sudan (2017) and Sudan (2017). Notes: A positive number means that the gender gap in that indicator is in favor of male-headed household (female-headed households are worse off), and a negative number means that the gap is in favor of female-headed households. The sum of all indicator gaps gives the overall gender gap. Similar patterns are observed in Sudan, except that lack of legal ID does not contribute to the household level gender gap. In contrast, in Somalia, the gender gap that favors female-headed households is driven by the differential in education and living standard measures. In general, in countries where female-headed households are worse off, the gap in lack of legal ID, lack of bank account, early marriage, and physical safety are larger than the gap in other indicators. In Somalia where the gap is in favor of female-headed households, the gap in education is a much larger contributor to the gender gap, followed by gap in lack of legal ID. As noted above, female-headed households are generally poorer than their male-headed counterparts in Ethiopia, South Sudan and Sudan. Existing literature shows heterogeneity among households headed by married women living with a husband or a nonresident (de facto) and those headed by divorced, separated, and single women (de jure) (Klasen et al., 2015; Milazzo and Walle, 2015). Thus, to explore heterogeneity of multidimensional poverty among households headed by female, we further disaggregated by de facto and de jure female heads (see Table 7). Table 7: Multidimensional Poverty outcomes by De facto and De jure female-headed households and displacement status Refugees/IDPs Hosts/Non-IDPs Female household head Female household head De facto De jure Difference De facto De jure Difference MPI Ethiopia 0.392 0.238 0.154*** 0.085 0.064 0.021 23 N.E Nigeria 0.158 0.190 -0.032* 0.082 0.068 0.015 Somalia 0.353 0.263 0.090 0.211 0.185 0.026 South Sudan 0.227 0.399 -0.172 0.194 0.161 0.033 Sudan 0.269 0.233 0.036 0.092 0.055 0.037 H (%) Ethiopia 64.2 38.5 25.7*** 15.0 11.2 3.8 N.E Nigeria 25.9 33.5 -7.6 14.3 12.1 2.1 Somalia 55.8 44.8 11.0 33.9 29.3 4.6 South Sudan 37.5 67.0 -29.5 31.2 25.3 5.9 Sudan 45.7 40.2 5.5 15.3 10.5 4.7 A (%) Ethiopia 61.0 61.7 -0.7 56.9 57.8 -0.9 N.E Nigeria 61.0 56.8 4.2* 57.4 55.4 2.0 Somalia 63.3 58.7 4.5* 62.3 63.2 -0.9 South Sudan 60.5 59.5 1.0 62.1 63.6 -1.6 Sudan 58.9 58.0 0.9 60.3 52.3 8.0*** Source: Authors’ calculations based on data from the High Frequency Surveys of Ethiopia (2017), Nigeria (2018), Somalia (2017), South Sudan (2017) and Sudan (2017). De facto - female-headed households whose head is a married female to a husband in the household or working away from the family. De jure - female-headed households whose head is a female widow, divorced or unmarried woman. H is the incidence of multidimensional poverty; A is the intensity of multidimensional poverty and MPI=H*A. Asterisks indicate statistical significance at 1% ***, 5% ** and 10% * levels. We find differences, with de facto female-headed households being worse off in Ethiopia and de jure being worse off in Nigeria. In Ethiopia, refugee individuals in de facto female-headed households (MPI of 0.392) are significantly more likely to be poor than de jure female-headed households (MPI of 0.238). Looking at the incidence of poverty, there is a 25.7 percentage points difference among female heads, with 64.2% of those in de facto and 38.5% of those in de jure households being MPI poor among Ethiopian refugees. In contrast, the results in Nigeria suggest that individuals in de facto female-headed IDP households are less likely to be poor compared to de jure female-headed IDP households, although the difference is only significant at the 10% level. For non-displaced households, we find no significant differences in poverty between de facto and de jure female heads across the five countries. In South Sudan, de jure IDP households have a higher MPI and incidence of poverty compared to de facto and although the difference is large, it is not statistically significant due to the small sample size. Next, we turn to an analysis by earning profiles. Although women are increasingly able to access labor markets, households that mostly depend on women’s contribution may be more likely to fall into poverty. Gender norms, wage gaps, and their disproportional presence in lower-paying occupations are among the factors driving this result. Following Hanmer et al (2021), we classify households according to the number and gender of the income contributors. A household member can contribute to the family income through paid employment, self- employment, farm-business work, trade, remittances or aid. Based on this definition, we classify households in seven mutually exclusive categories: no earners; dependent on remittances only; single earner (male or female); multiple earners (majority female, equal contribution or majority male). Table 8 shows the disaggregation of poverty headcount ratios by earnings profile and displacement status, highlighting that households without income contributors are more likely to be poor compared to other types of families, regardless of the country. Among IDPs, households with a single female earner also emerge as vulnerable. For example, in Sudan, while 24 55% of female-single IDP earners are considered multidimensionally poor, the equivalent figure is 16 percentage points lower (39%) in IDP households dependent on multiple male earners. Similarly, in Somalia, 70% of female-single earner households are poor, which is the highest poverty rate among IDP and non-IDP households. Households that depend mostly on women’s earnings seem more likely to fall into poverty. For example, in Ethiopia, poverty rates among female-single earners and multiple-female earners in refugee households are the highest (52% and 57%, respectively). These rates are strikingly high and pass even the poverty rate of households without earners. In contrast, 23% of refugee households whose income depend on a single male earner and 16% of those that depend on multiple male earners are poor. However, it is worth noting that regardless of the earnings category, the odds of falling into poverty are larger among refugee or IDP households with at least one earner compared to their host communities. Table 8. Poverty headcount ratios by earnings profile and displacement status Ethiopia N.E Nigeria Somalia Sudan Host Host Host Host Refugees IDP IDP community community community community IDP No earners 40% 51% 62% 50% 32% 64% 7% 45% Remittance recipients only 7% 8% 76% 69% 16% 37% - - Female single earner 12% 52% 23% 37% 35% 70% 24% 55% Male single earner 8% 23% 4% 14% 24% 50% 8% 39% Majority female earners 3% 57% 9% 28% 9% 28% 15% 45% Equal contribution 10% 24% 17% 16% 23% 63% 8% 41% Majority male earners 5% 16% 15% 19% 20% 52% 9% 39% Source: Authors’ calculations based on data from the High Frequency Surveys of Ethiopia (2017), Nigeria (2018), Somalia (2017), South Sudan (2017) and Sudan (2017). Overall, the results show that besides gender, displacement status and the number of household contributors plays a key role in the identification and level of poverty. In comparison with female-headed non-displaced households, more female-headed displaced households are classified as multidimensionally poor. According to the number and gender of the household income contributors, we find that refugee and IDP households are the most vulnerable regardless of the family classification. Among displaced populations, households dependent on a unique earner or on female earners are disproportionally more likely to fall into poverty, compared to other households with at least one earner. These findings indicate that differentiation by gender of the head, displacement status and subgroups of headship have important implications for policy and targeting. 7. MPI and monetary poverty We now compare the results using monetary and multidimensional poverty measures, as both can provide us with different insights into the short-term and long-term hardships that households face in displacement situations. Monetary poverty is calculated using the international poverty line of US$1.90 per capita per day and an aggregate of consumption per capita (Pape et al 2017) from the sum of expenditures on food items, non-food items and the 25 value of consumption flow of durable goods. 22 While monetary poverty can measure temporal resource holdings, multidimensional poverty, as a more comprehensive measure, includes chronic and exacerbating sources of poverty. This difference explains the existence of mismatches between individuals identified as monetary versus MPI poor, which are often more prominent in poorer countries (Evans et al 2020). This section examines these differences in the contexts of displacement. Table 9. Percentage of the sample in each poverty category: Rows sum to 100% Only Monetary and Non-poor by both Only Multidimensional multidimensional measures Monetary Poor Poor poor Ethiopia 38% 23% 12% 27% N.E Nigeria 13% 69% 4% 15% Somalia 20% 32% 14% 34% Sudan 33% 47% 4% 17% Source: Authors’ calculations based on data from the High Frequency Surveys of Ethiopia (2017), Nigeria (2018), Somalia (2017), South Sudan (2017) and Sudan (2017). This table presents the distribution of households in each of the categories in the columns. Thus, each row adds up to 100%. South Sudan is excluded from this analysis as monetary data is not available for the country. Given that monetary and multidimensional poverty identify different populations, a household can be classified as: non-poor according to both measures; poor only by the monetary measure; poor only by the multidimensional measure; or poor by both measures (see Table 9). The results confirm the expectation that monetary and multidimensional poverty measures do not always capture the same people. Ethiopia has the largest overlap between the two measures, with 65% of individuals receiving the same classification (38% are non-poor, and 27% are poor according to both measures). This considerable overlap could be explained by Ethiopia having the most prolonged conflict and largest number of fatalities among the countries, leading to a loss in assets and livelihoods, which in turn affected long-term welfare and decrease employment and economic opportunities, that impact available short-term resources. Thus, the overlap may reflect the long-lasting as well as the short-term effects of conflict and economic hardships. On the other hand, Nigeria is the country with the smallest overlap, where only 15% of the sample is classified as poor by both measures and 13% as non-poor. The majority, 69% of people, are only monetary poor, and 4% are multidimensionally poor only. The lack of overlap might be explained by the relatively recent start of the displacement situation in 2014, when Boko Haram appeared in the north-eastern part of the country. Pape et al (2018) identify two groups of IDPs in this situation: one group representing 74% of the IDP population that was more engaged in wages and non-farm business before displacement, and another group representing about 26% of the population, that had significantly more unemployed women. Most of the displaced populations from the first group live in host communities with good access to basic services such as sanitation and water, and safety nets. However, they are disproportionally more likely to be female-headed households and lack access to education, health services, and may face more stringent labor-market barriers. In other words, this group has relatively better housing conditions, but may lack short-term resources that reduce their consumption expenditure. 22In summary, expenditure in these three categories is computed based on the quantities and prices of a selected list of items in each category. See more details about the computation of the consumption aggregate in Appendix A of the Somali Poverty Profile (Pape et al, 2017). A similar procedure was followed in the other countries of analysis. 26 Next, we analyze poverty rates by demographic category and by displacement status (Table 10). In Ethiopia, the monetary measure shows evidence of large differences in poverty rates between refugees and the host community. For example, while 63% of refugee households with a female household head are monetary poor, only 7% of the female-headed households in the host community are identified as poor by the $1.90 a day measure. This gap between the refugee community and the host community is also large among households headed by men. When turning to the multidimensional measure, a similar pattern emerges, with 55% of refugees in female-headed households being poor in Ethiopia, while only 14% are poor using the multidimensional measure in the host community. These results are consistent in Nigeria among male household heads. Among female household heads, we find no differences between IDPs and non-IDPs using the monetary poverty measure; however, there is a sizable gap between these two groups when using the MPI. To explore the role of possible disruptions to family structures as part of the displacement, Table 10 also presents descriptive analysis by de jure and de facto head. We find that regardless of the type of female head, IDP households are significantly more likely to be classified as poor according to both the monetary and the multidimensional measures. Another striking finding is that de facto headed households are more likely to be monetary and multidimensionally poor in Ethiopia, while the opposite is the case in Nigeria. The legal environment, as well as the disproportionate role in caring and housework responsibilities, may act as a disadvantage for households with a relatively larger number of women contributing to the household income. The third set of poverty rates shown in Table 10 explores this hypothesis. In Ethiopia, for example, households where women are contributing more to the household income and those with no earners are the most vulnerable groups in both the refugee and the host community. This pattern is consistent using both measures of poverty. On the other hand, households where most earners are male and where the primary income source is remittances are less likely to be poor, especially among refugees. Overall, monetary poverty and multidimensional poverty identify different households as poor. In spite of these salient differences, people living in refugee/IDP households are consistently poorer according to both the monetary and the multidimensional poverty measures. When we compare the rankings of the population subgroups according to each measure, we find that households with a single female earner or no earners are consistently identified as poorer according to each of the measures. Table 10. Poverty by demographic characteristics for Ethiopia and Nigeria Panel A. Ethiopia % of monetary poor % of multidimensional poor Host Host Refugee Community All Refugee Community All Female hhh 63% 7% 58% 55% 14% 53% Male hhh 49% 20% 38% 18% 10% 17% De jure female hh 47% 17% 39% 34% 11% 32% De facto female hh 68% 19% 66% 61% 17% 60% 27 No earners 64% 29% 63% 51% 40% 50% Remittance recipients only 19% 4% 16% 8% 7% 8% Female single earner 47% 15% 41% 52% 12% 47% Male single earner 54% 27% 44% 23% 8% 19% Majority female earners 66% 1% 43% 57% 3% 50% Equal contribution 46% 6% 27% 24% 10% 20% Majority male earners 41% 4% 16% 16% 5% 13% Panel B. N.E Nigeria % of monetary poor % of multidimensional poor IDPs Non-IDPs All IDPs Non-IDPs All Female hhh 91% 91% 91% 28% 8% 14% Male hhh 90% 76% 80% 21% 22% 22% De jure female hh 99% 77% 96% 32% 9% 20% De facto female hh 85% 80% 89% 25% 7% 11% No earners 83% 98% 94% 50% 62% 58% Remittance recipients only 100% 99% 96% 69% 76% 73% Female single earner 98% 91% 92% 38% 24% 32% Male single earner 96% 93% 94% 14% 4% 7% Majority female earners 88% 86% 83% 28% 9% 14% Equal contribution 93% 81% 83% 16% 17% 17% Majority male earners 78% 56% 57% 19% 15% 16% Source: Authors’ calculations based on data from the High Frequency Surveys of Ethiopia (2017), Nigeria (2018), Somalia (2017), South Sudan (2017) and Sudan (2017). 8. Conclusion This paper contributes to the literature by analyzing multidimensional poverty among refugees and internally displaced populations. We observe that forcibly displaced communities are poorer than host communities in each of the five countries’ sub-populations covered in the surveys, with the difference in incidence between displaced and non-displaced population ranging between 15 and 19 percentage points in South Sudan and Somalia to over 30 percentage points in Ethiopia and Sudan. Displaced communities also experience greater deprivations in nearly every indicator, although there is significant variation in which indicators are the most salient, with having a bank account and cooking fuel in Ethiopia, years of schooling in Somalia, electricity in Sudan, drinking water in South Sudan, and legal identification in Nigeria showing the largest differences between the two populations. The results also indicate gender differences in the experience of multidimensional poverty, with female-headed households more likely to be poor than male-headed households in most of the countries. In addition, displaced households headed by women have a higher incidence of poverty and MPI than non-displaced female-headed households. Particularly, female-headed households in camps have higher multidimensional poverty and intensity compared to their counterparts living outside camps. Dissaggregating further, we find heterogeneity among de facto and de jure female heads. This variation lends itself to further research questions about 28 targeted policies and interventions and the evidence-based approaches required to address the deprivations faced by men and women and boys and girls living in forcibly displaced households. The comparison with monetary poverty shows the differences and complementarity of the measures, can provide a first step to better understand the different conditions and challenges faced by forcibly displaced communities, and suggests potential policy responses for addressing them. The findings demonstrate the value added of the multidimensional monetary poverty measure to analyze the overlapping deprivations, particularly in contexts of forced displacement. The MPI’s indicators allow policy makers to identify the arenas for intervention, and further analysis decomposing each country’s MPI in greater detail and disaggregating by geographic and administrative areas, as well as age groups, would benefit targeting measures. In general, the results offer compelling evidence in support of a more disaggregated multidimensional poverty analysis. Our findings indicate that differentiation by gender of the household head, displacement status and subgroups of headship have implications for policy and targeting. This is important with the existing limitation of data (particularly in situations of displacement) to capture gender differences in multidimensional poverty. Due to data limitations, our analysis of differences is limited to the gender of the household head and type of headship. Although the analysis at the household level remains relevant given the high prevalence of female-headed households in displacement situations, individual level analysis is invited as individual level data becomes available. References Agbodji, A. E., Batana, Y., M., and Ouedraogo, D. (2015). Gender Inequality in Multidimensional Welfare Deprivation in West Africa. International Journal of Social Economics, 42(11): 980-1004. Alkire, S. and Foster, J. (2011). Counting and Multidimensional Poverty Measurement. Journal of Public Economics, 95(7): 476-487. Alkire, S., Foster, J. E., Seth, S., Santos, M. E., Roche, J. M., and Ballon, P. (2015). Multidimensional Poverty Measurement and Analysis. Oxford: Oxford University Press. Alkire, S. and Jahan, S. (2018). ‘The new global MPI 2018: Aligning with the Sustainable Development Goals’, OPHI Working Paper 121, University of Oxford. Alkire, S., Kanagaratnam, U., and Suppa, N. (2020). ‘The global Multidimensional Poverty Index (MPI) 2020’, OPHI MPI Methodological Note 49, Oxford Poverty and Human Development Initiative, University of Oxford. Appleton, S. (1996). Women-headed households and household welfare: An empirical deconstruction for Uganda. World Development, 24(12), 1811-1827. Assad, R., Krafft, C., & Pastoor, I. (2021). Gender Role Attitudes and Gendered Outcomes: Syrian Refugee Adolescents in Jordan. World Bank and Policy Research Working Paper, forthcoming. World Bank Group. 29 Blau, F. D., Kahn, L. M., and Souza, A. P. (2003). The Role of the Family in Immigrants' Labor-Market Activity: An Evaluation of Alternative Explanations: Comment. American Economic Review, 93(1): 429-447. Blau, F. D., Kahn, L. M. and Papps, K. L. (2011). Gender, Source Country Characteristics, and Labor Market Assimilation among Immigrants. Review of Economics and Statistics, 93(1): 43- 58. Boudet, A. M. M., Buitrago, P., Leroy de la Briere, B., Newhouse, D., Matulevich, E. R., Scott, K., and Suarez-Becerra, P. (2018). Gender Differences in Poverty and Household Composition through the Life‐cycle: A Global Perspective. Policy Research Working Paper 8360. Washington, DC: World Bank Group. Brown, C. and van de Walle, D. (2020). Headship and Poverty in Africa. Center for Global Development: Working Paper 531, April 2020. Washington, D.C.: Center for Global Development. Brück, T. and Schindler, K. (2009). The Impact of Violent Conflicts on Households: What Do We Know and What Should We Know about War Widows? Oxford Development Studies, 37(3): 289-309. Brück, T. & Stojetz, W. (2021). Gendered Dimensions of Protracted Forced Displacement in Sudan. World Bank and Policy Research Working Paper, forthcoming. World Bank Group. Buchanan, E. (2019). ‘Born to be Married’: Addressing child, early and forced marriage in Nyal, South Sudan. Retrieved from https://www.oxfamamerica.org/explore/research- publications/born-be-married/ Buvinić, M., & Gupta, G. R. (1997). Female-headed households and female-maintained families: are they worth targeting to reduce poverty in developing countries?. Economic development and cultural change, 45(2), 259-280. Buvinic, M., Das Gupta, M., Casabonne, U., and Verwimp, P. (2013). Violent Conflict and Gender Inequality: An Overview. The World Bank Research Observer, 28(1): 110-138. Chaaban, J. M., Seyfert, K., Salti, N. I., El Makkaoui, G. S. (2013). Poverty and Livelihoods Among UNHCR Registered Refugees in Lebanon. Refugee Survey Quarterly, 32(1): 24-49. Espinoza-Delgado, J. and Klasen, S. (2018) ‘Gender and multidimensional poverty in Nicaragua: An individual based approach’, World Development, 110: 466-491 Evans, M., Nogales, R., and Robson, M. (2020). Monetary and Multidimensional Poverty: Correlations, Mismatches, and Joint Distributions. OPHI Working Paper 133. Oxford: University of Oxford. Hanmer, L., Rubiano, E., Santamaria, J., and Arango, D. J. (2020). How Does Poverty Differ among Refugees? Taking a Gender Lens to the Data on Syrian Refugees in Jordan. Middle East Development Journal, 12(2): 208-242. 30 Hejoj, I. (2007). ‘A Profile of Poverty for Palestinian Refugees in Jordan: the Case of Zarqa and Suhkneh Camps’, Journal of Refugee Studies, 20(1): 120-145. Horrell, S., & Krishnan, P. (2007). Poverty and productivity in female-headed households in Zimbabwe. Journal of Development Studies, 43(8), 1351-1380. Human Rights Watch. (2014). “Those Terrible Weeks in their Camp”: Boko Haram Violence against Women and Girls in Northeast Nigeria. Retrieved from: https://www.hrw.org/sites/default/files/reports/nigeria1014web.pdf. Internal Displacement Monitoring Centre. (2020). Global Report on Internal Displacement (2020). Retrieved from https://www.internal-displacement.org/global-report/grid2020/. International Labour Organization. (2009). Report VI: Gender equality at the heart of decent work. Geneva: International Labour Office. Retrieved from https://www.ohchr.org/Documents/Issues/Racism/IWG/Session7/GenderEquality.pdf. International Organization for Migration. (2020). World Migration Report 2020. Geneva: International Organization for Migration. Retrieved from https://publications.iom.int/system/files/pdf/wmr_2020.pdf. International Organization for Migration. (2013). Internal Displacement in Iraq: Barriers to Integration. Baghdad: International Organization for Migration. Kelly, J. (2021). The Risk that Travels with You: The Links between Forced Displacement, Conflict and Intimate Partner Violence in Colombia and Liberia. World Bank and Policy Research Working Paper, forthcoming. World Bank Group. Kelly, J, Holmes, M., & Voors, M. (2021). Gender-based Violence among Displaced and Host Populations in Eastern Democratic Republic of Congo. World Bank and Policy Research Working Paper, forthcoming. World Bank Group. King, E. M., Klasen, S., & Porter, M. (2007). Women and development. Copenhagen Consensus Center. King, E., & Mason, A. (2001). Engendering development: Through gender equality in rights, resources, and voice. Klasen, S., & Lahoti, R. (2020). ‘How Serious Is the Neglect of Intra-Household Inequality in Multi-dimensional Poverty and Inequality Analysis?’, The Review of Income and Wealth, 67(3), 705-731. Klasen, S., Lechtenfeld, T., & Povel, F. (2015). A feminization of vulnerability? Female headship, poverty, and vulnerability in Thailand and Vietnam. World Development, 71, 36- 53. Klasen, S., & Povel, F. (2013). Defining and measuring vulnerability: State of the art and new proposals. Vulnerability to poverty, 17-49. 31 Loaiza Quintero, O. L., Muñetón Santa, G., and Gabriel Vanegas, J. (2018). Forced Displacement and Multidimensional Poverty in Antioquia, Colombia: An Assessment by Means of a Seemingly Unrelated Regression. Journal of Regional Research, 41: p167-190. Milazzo, A., & Van De Walle, D. (2017). Women left behind? Poverty and headship in Africa. Demography, 54(3), 1119-1145. Munoz-Boudet, A.M., Buitrago, P., De La Briere, B., Newhouse, D., Matulevich, E.R., Scott, K., & Suarez-Becerra, P. (2018). “Gender Differences in Poverty and Household Composition through the Life-Cycle: A Global Perspective.” Policy Research Working Paper No. 8360. World Bank, Washington, DC. https://openknowledge.worldbank.org/handle/10986/29426 Nagar, S. E., Bamkar, S., & Tønnessen, L. (2017). Girls, Child Marriage, and Education in Red Sea State, Sudan: Perspectives on Girls’ Freedom to Choose. Sudan Report. Nguyen, K. (2019) ‘As drought compounds security woes, Somalis flee to Ethiopia’, UNHRC UK. 20 September 2019. Retrieved from https://www.unhcr.org/news/stories/2019/9/5d8495984/drought-compounds-security-woes- somalis-flee-ethiopia.html. Overseas Development Institute. (2019a). The Impact of Displacement on Gender Roles and Relations: The Case of IDPs from FATA, Pakistan. London: Overseas Development Institute. Retrieved from: https://www.odi.org/sites/odi.org.uk/files/resource-documents/12610.pdf. Overseas Development Institute. (2019b). Gender in Displacement: The state of play. London: Overseas Development Institute. Retrieved from: https://www.odi.org/sites/odi.org.uk/files/resource- documents/201911_gender_in_displacement_hpg_working_paper_web.pdf. OECD. (2019), "Gender differences in immigrant integration", in Settling In 2018: Indicators of Immigrant Integration, OECD Publishing, Paris, https://doi.org/10.1787/9789264307216- 10-en. Oruc, N. (2015). Urban IDPs and Poverty: Analysis of the Effect of Mass Forced Displacement on Urban Poverty in Bosnia and Herzegovina. Croatian Economic Survey, 17(1): p. 47-40. Quisumbing, A. R., Haddad, L., & Peña, C. (2001). Are women overrepresented among the poor? An analysis of poverty in 10 developing countries. Journal of Development Economics, 66(1), 225-269. Rodríguez, L. (2016). Intrahousehold inequalities in child rights and well-being. A barrier to progress? World Development, 83, 111-134. Rubiano-Matulevich, E. (2021). Do Gender Norms Relax with Conflict-Induced Displacement? The Case of Colombia. World Bank and Policy Research Working Paper, forthcoming . World Bank Group. 32 Sudanese Government’s Joint Mechanism for Durable Solutions. (2019). Progress Towards Durable Solutions: In Abu Shouk And El Salam IDP Camps, North Darfur, Sudan. United Nations – Sudan. Retrieved from https://www.jips.org/uploads/2019/12/JIPS-Sudan- profilingreport-2019.pdf. Temgoua Noumedem, C., Sharma, D., and Wai-Poi, M. (2020). Multidimensional Poverty Assessment of Internally Displaced Persons in Iraq. Policy Research Working Paper 9203. Washington DC: World Bank Group. UN-HABITAT. (2009). Darfur: Profile of El Fasher Town and Abu Shouk IDP Camp. Nairobi:UN-HABITAT. Retrieved from https://unhabitat.org/sites/default/files/download- manager-files/El%20Fasher%20and%20Abu%20Shouk%20Profile.pdf. United Nations High Commissioner for Refugees. (2020). Global Trends: Forced Displacement in 2019. Retrieved from https://www.unhcr.org/5ee200e37.pdf. United Nations Populations Fund Somalia. (2019). Gender Equity: Hit or miss in the Somali population. December 2019. Retrieved from https://somalia.unfpa.org/en/publications/gender- equity-hit-or-miss-somali-population. Van de Walle, D. (2013). Lasting welfare effects of widowhood in Mali. World Development, 51, 1-19. Chant, S. H. (Ed.). (2010). The international handbook of gender and poverty: concepts, research, policy. Edward Elgar Publishing. Verwimp, P. (2012). Returning Home after Civil War: The Consequences of Forced Displacement for Food Security, Nutrition and Poverty among Burundese Households. IDEAS Working Paper Series from RePEc, ECARES Working Paper 2012-033. World Bank. (2011). World development report 2012: Gender equality and development. The World Bank. World Bank (2020a) World Development Indicators: Population dynamics. Last updated: 16 December 2020. Retrieved from http://wdi.worldbank.org/table/2.1# World Bank (2020b) World Development Indicators: Poverty rates at international poverty lines. Last updated: 16 December 2020. Retrieved from http://wdi.worldbank.org/table/1.2# Zenn, J., and Pearson, E. (2014). Women, Gender, and the evolving tactics of Boko Haram. Journal of Terrorism Resarch, 5(1): https://doi.org/10.15664/jtr.828. Zetter, R., and Ruaudel, H. (2014). Development and protection challenges of the Syrian Refugee Crisis. Forced Migration Review, 47:6-10 33 Appendices Table A1: Population share by displacement status and marital status of female headship (%) IDPs/Refugees Hosts/non-IDPs De facto De jure De facto De jure Ethiopia 72.73 22.25 1.97 3.0% N.E Nigeria 17.65 12.51 57.71 12.13 Somalia 29.71 7.75 49.00 13.55 South Sudan 9.39 6.66 56.91 27.05 Sudan 34.87 8.13 37.77 19.23 Table A2: Censored headcounts of each indicators in the MPI, by the gender of the household head in Ethiopia (2017) Ethiopia Male-headed HH Female-headed HH Difference h0 SE h0 SE Years of Schooling 14.0 0.013 33.3 0.02 -19.2*** School attendance 10.9 0.011 12.7 0.01 -1.8 Food Security 14.2 0.013 49.4 0.04 -35.2*** Pregnancy care 1.8 0.006 5.2 0.01 -3.4*** Physical safety 4.3 0.008 41.8 0.04 -37.5*** Early marriage 4.1 0.008 47.6 0.02 -43.5*** Waste Disposal 8.1 0.011 47.4 0.03 -39.3*** Drinking water 2.0 0.004 7.2 0.01 -5.2*** Electricity 14.4 0.013 46.3 0.04 -31.9*** Cooking fuel 17.0 0.014 55.7 0.03 -38.7*** Housing 14.0 0.013 53.3 0.04 -39.3*** Sanitation 13.5 0.013 29.6 0.04 -16.1*** Unemployment 5.9 0.010 15.4 0.03 -9.5*** Legal id 4.6 0.008 40.4 0.05 -35.8*** Bank account 16.8 0.014 55.6 0.03 -38.7*** Notes: h0 is the proportion of male- and female-headed households deprived in various indicators, SE- standard errors. Asterisks indicate statistical significance at 1% ***, 5% **, and 10% * levels. Table A3: Censored headcounts of each indicators in the MPI, by the gender of the household head in Nigeria (2018) Nigeria Male-headed HH Female-headed HH Difference h0 SE h0 SE Years of Schooling 17.7 0.042 12.5 0.016 5.2 School attendance 19.7 0.055 13.6 0.021 6.1 Food Security 20.0 0.053 13.9 0.023 6.1 Pregnancy care 9.1 0.024 5.5 0.028 3.7 Physical safety 1.1 0.004 1.1 0.006 0.0 Early marriage 7.8 0.036 10.9 0.051 -3.2 34 Waste Disposal 21.3 0.052 17.2 0.037 4.1 Drinking water 9.1 0.055 5.7 0.028 3.4 Electricity 20.4 0.053 7.8 0.036 12.7* Cooking fuel 22.5 0.051 18.4 0.041 4.1 Housing 14.2 0.061 5.4 0.026 8.8 Sanitation 17.4 0.032 17.0 0.037 0.5 Unemployment 7.1 0.038 3.0 0.016 4.2 Legal id 4.5 0.013 9.4 0.046 -4.9 Bank account 22.4 0.051 18.1 0.040 4.3 Notes: h0 is the proportion of male- and female-headed households deprived in various indicators, SE- standard errors. Asterisks indicate statistical significance at 1% ***, 5% **, and 10% * levels. Table A4: Censored headcounts of each indicators in the MPI, by the gender of the household head in Somalia (2017) Somalia Male-headed HH Female-headed HH Difference h0 SE h0 SE Years of Schooling 49.6 0.033 32.7 0.030 16.9*** School attendance 45.7 0.028 33.3 0.030 12.5*** Food Security 33.6 0.030 23.1 0.027 10.6*** Pregnancy care 26.9 0.025 14.6 0.022 12.3*** Physical safety 7.6 0.016 6.9 0.016 0.8 Early marriage 33.0 0.027 26.8 0.032 6.3* Waste Disposal 55.7 0.030 38.9 0.032 16.9*** Drinking water 43.2 0.033 22.2 0.023 20.9*** Electricity 49.6 0.034 28.5 0.028 21.1*** Cooking fuel 55.5 0.036 39.3 0.033 16.1*** Housing 54.6 0.031 34.6 0.032 20.0*** Sanitation 49.8 0.035 26.7 0.029 23.0*** Unemployment 5.2 0.011 5.0 0.012 0.2 Legal id 46.3 0.034 32.1 0.032 14.1*** Bank account 25.8 0.038 20.0 0.023 5.9 Notes: h0 is the proportion of male- and female-headed households deprived in various indicators, SE- standard errors. Asterisks indicate statistical significance at 1% ***, 5% **, and 10% * levels. Table A5: Censored headcounts of each indicators in the MPI, by the gender of the household head in South Sudan (2017) South Sudan Male-headed HH Female-headed HH Difference h0 SE h0 SE Years of Schooling 7.9 0.018 17.9 0.033 -10.0*** School attendance 10.0 0.026 16.7 0.036 -6.7 Food Security 13.3 0.023 30.3 0.044 -17.0*** Pregnancy care 4.6 0.016 7.3 0.024 -2.7 Physical safety 12.8 0.025 30.0 0.043 -17.2*** 35 Early marriage 4.3 0.021 24.6 0.041 -20.3*** Waste Disposal 14.6 0.028 29.7 0.040 -15.0*** Drinking water 3.5 0.017 7.9 0.029 -4.4 Electricity 13.2 0.026 32.5 0.044 -19.3*** Cooking fuel 14.9 0.028 32.5 0.044 -17.6*** Housing 13.9 0.026 31.2 0.042 -17.2*** Sanitation 12.6 0.028 28.3 0.044 -15.8*** Unemployment 1.5 0.007 3.2 0.012 -1.6 Legal id 9.0 0.021 23.0 0.032 -14.0*** Notes: h0 is the proportion of male- and female-headed households deprived in various indicators, SE- standard errors. Asterisks indicate statistical significance at 1% ***, 5% **, and 10% * levels. Table A6: Censored headcounts of each indicators in the MPI, by the gender of the household head in Sudan (2018) Sudan Male-headed HH Female-headed HH Difference h0 SE h0 SE Years of Schooling 2.5 0.004 5.6 0.009 -3.1*** School attendance 10.8 0.015 17.7 0.019 -6.9*** Food Security 12.1 0.014 21.8 0.021 -9.7*** Pregnancy care 3.6 0.009 5.6 0.011 -2.0 Physical safety 6.6 0.011 12.1 0.017 -5.5*** Early marriage 12.9 0.015 19.3 0.021 -6.4*** Waste Disposal 15.7 0.017 25.6 0.025 -9.9*** Drinking water 13.2 0.016 21.6 0.018 -8.4*** Electricity 14.0 0.016 24.5 0.021 -10.5*** Cooking fuel 16.6 0.018 27.0 0.024 -10.3*** Housing 16.6 0.018 26.9 0.024 -10.2*** Sanitation 13.2 0.015 20.8 0.023 -7.6*** Unemployment 7.0 0.012 11.0 0.022 -4.0* Legal id 0.0 0.2 0.001 0.0 Bank account 16.0 0.017 26.2 0.025 -10.2*** Notes: h0 is the proportion of male- and female-headed households deprived in various indicators, SE- standard errors. Asterisks indicate statistical significance at 1% ***, 5% **, and 10% * levels. 36