Policy Research Working Paper 9818 Differences in Household Composition Hidden Dimensions of Poverty and Displacement in Somalia Lucia C. Hanmer Eliana Rubiano-Matulevich Julieth Santamaria Gender Global Theme October 2021 Policy Research Working Paper 9818 Abstract Little is known about how gender inequality influences characteristics are strongly associated with poverty rates poverty rates of forcibly displaced people. This paper for internally displaced people but not for non-displaced uses a nationally representative survey to analyze poverty people. Having more income earners reduces poverty risk among internally displaced people and non-displaced for all households. For internally displaced people’s house- people in Somalia. More than half of internally displaced holds, the largest decrease in poverty risk is associated with people’s households and 47 percent of non-displaced having more female earners, while having more male earn- people’s households are female headed. Although poverty ers is associated with the lowest poverty for nondisplaced rates are higher among internally displaced people than people’s households. The analysis highlights that poverty non-displaced people (77 versus 66 percent), male-headed reduction policies and programs must cover all households households are poorer than female-headed ones among and lift barriers to women’s economic opportunities. Pro- both groups. Extending the analysis beyond headship to grams that respond to women’s care responsibilities and demographic characteristics and by the gender and number address barriers to women’s economic opportunities are of earners provides a more nuanced picture. Demographic especially important for internally displaced people. 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 lhanmer@worldbank.org (corresponding author), erubiano@worldbank.org, and juliethsa@iadb.org. 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 Differences in Household Composition: Hidden Dimensions of Poverty and Displacement in Somalia* Lucia C. Hanmer[†], Eliana Rubiano-Matulevich[‡], Julieth Santamaria[§] Keywords: households, conflict, internally displaced people, host communities, gender. JEL codes: D10, O15, D74, J16. 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. * 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. 1. Introduction Globally, the number of forcibly displaced people is unprecedented. At the end of 2020, the number of people forced to flee had reached 82 million, including 26 million refugees and 48 million internally displaced people (IDPs) 1 (UN High Commissioner for Refugees [UNHCR, 2021]). 2 Forcibly displaced women and men face specific barriers to taking up economic opportunities and accessing services compared to their host communities, which create higher risks of being trapped in poverty than many of their hosts (World Bank, 2017). Qualitative research underlines that women and men experience and respond differently to forced displacement (Forced Migration Review, 2018). However, little is known about how gender inequality influences poverty rates of forcibly displaced people, largely due to the scarcity of representative data adequate to investigate these differences. We use the 2017-18 Somalia High-Frequency Survey to examine the role of gender-based disadvantage in poverty rates experienced by IDPs and non-IDPs. Three metrics of gender disadvantage are investigated: first, we use the traditional approach to measuring the impact of gender on poverty, via household headship; second, we classify households according to their demographic characteristics; and third, by the number and sex of earners in the household. To our knowledge, this is the first study to go beyond headship in analyzing the gender dimensions of poverty in situations of internal displacement. [†] Lead Economist, Gender Group, World Bank Group. [‡] Economist, Poverty and Equity Global Practice, World Bank Group. [§] Economist, International American Development Bank, Washington, D.C. 1 Internally displaced persons (IDPs), are “persons or groups of persons who have been forced or obliged to flee or to leave their homes or places of habitual residence, in particular as a result of or in order to avoid the effects of armed conflict, situations of generalized violence, violations of human rights or natural or human-made disasters, and who have not crossed an internationally recognized state border” (UN 1998 Guiding Principles on Internal Displacement E/CN.4/1998/53/Add.2 11 February 1998) 2 Forcibly displaced people include refugees, internally displaced people, and asylum seekers (see https://migrationdataportal.org/themes/forced-migration-or-displacement#definitions) 1 Our findings underline the importance of differentiating among different types of households using other metrics beyond simply male and female headship. Overall, 65 percent of female-headed households are poor compared with 70 percent of male-headed households. Poverty is higher among IDPs than non-IDP communities (77 versus 67 percent); however, male-headed IDP households are still poorer than female-headed ones (83 versus 71 percent). The relative disadvantage of male-headed households in both communities still holds when female heads are divided between de jure and de facto female heads. De jure female household heads are those who are divorced, separated, or widows. De facto household heads are married or in a civil union often with their husband or partner living away from the family home. We find that, controlling for household and individual characteristics, IDP families with children, especially female single caregivers, experience much higher poverty rates than families without children. Interestingly, these demographic characteristics are not strongly associated with poverty rates for non-IDPs. For both IDPs and non-IDPs poverty risk is significantly lower for households with more income earners of either gender. An important finding for IDP households is that the largest decrease in poverty risk is associated with having more female earners. In contrast, for non-IDPs more male earners are associated with the greatest poverty risk reduction. Our paper contributes to the growing literature that suggests that many forcibly displaced people face high and persistent barriers to achieving adequate living standards for themselves and their families and that these barriers are higher than those faced by other poor communities. It extends the literature by analyzing the association of gender-based disadvantage with poverty for displaced and non-displaced populations, respectively. Forced displacement can impair people’s abilities to escape poverty in a way that sets them apart from non-displaced populations (Christensen & Harild, 2009). Loss of assets and trauma resulting from displacement lead to higher poverty rates among IDPs than non-displaced populations. For example, poverty rates among internally displaced persons in Azerbaijan are 25 percent, compared with 20 percent among the non-displaced (Bussolo & Lopez-Calva, 2014). Over 2 the medium term, IDPs may not be able to replace their assets. For example, in Afghanistan, 60 percent of IDPs in urban areas live in a tent or temporary shelter, and even after five years in displacement, 61 percent of them remained in temporary housing (World Bank & UNHCR, 2011). Asset losses may also impact people’s standing in the community; for example, the loss of cattle meant a change in social status among forcibly displaced pastoralists in the Sahel (World Bank, 2017). The rest of this paper is structured as follows: Section 2 provides a brief overview of the literature related to poverty and gender. Section 3 describes the Somali context and provides some background to the drivers of forced displacement and poverty. Section 4 describes the data. Section 5 presents the empirical approach, before describing the results in Section 6. Section 7 concludes and discusses the policy implications of our findings. 2. Poverty and Gender Research examining why women as individuals or households dependent on women’s income may be particularly vulnerable to poverty often draws attention to how underlying norms about gender affect paid and unpaid work (Braunstein et al., 2011; Folbre, 2006; Nelson, 1995). Household allocation of time to paid work or employment or otherwise remunerated employment such as entrepreneurship is partly the outcome of an intra-household bargaining process. Power inequality between men and women combined with social norms leads to a gendered division of labor whereby women undertake the majority of unpaid work. Hence, the demand for labor for unpaid household-services restricts women’s ability to participate in labor markets (Floro, 1995; Folbre, 2006; Folbre et al., 2005) and restricts girls’ ability to attend school and acquire labor market skills (Arora, 2015; Bardasi & Wodon, 2010; DeGraff et al., 2017; Gammage, 2010). Compared to men, women have less access and control over financial, digital, land and property assets putting them at a disadvantage as entrepreneurs and self-employed workers (Deere & Doss, 2006; Deere et al, 2013; Morton et al. 2014; Kantor, 2002; Horrell & Krishnan, 2007; Klugman & Tyson, 2016). 3 A body of evidence shows that across the world, compared to men, women are less likely to participate in the labor market (Klasen, 2019). Legal and social norms shaping attitudes about women’s role in society underlie this outcome. These norms are often reflected in discriminatory laws, employment practices, gender wage penalties and social preferences that reinforce the gender division of labor (Connelly & Kongar, 2017; Rubiano-Matulevich & Viollaz, 2019). In developing countries, when women work, they are more likely to be informal workers, earn less, and are less likely to receive pensions and other work-related benefits (Bosch & Maloney, 2010; Ñopo et al., 2011; World Bank, 2011). Empirical evidence of women’s time constraints, gender norms, and barriers to women’s access to economic opportunities have led researchers to examine whether households headed by a woman or their income are more vulnerable to poverty than male-headed households. Many studies have found that poverty rates among female-headed households are higher than those with a male head, but this finding is far from universal (Buvinić & Gupta, 1997; Chant, 2003; Lampietti and Stalker 2000; Quisumbing et al. 2001). Analyses based on headship have been criticized because the concept hides substantial differences between the types of households in which individuals live, such as income transfers from male migrant workers, social norms surrounding widows, divorcees and lone parents and distribution of care for elders and children (Chant, 1997, 2003, 2004, 2008; Lampietti & Stalker, 2000; Quisumbing et al., 1995; Rosenhouse, 1989). De jure versus de facto distinctions are often used to capture differences between female headship. De jure female household heads are those who are divorced, separated, or widows. De facto household heads are married or in a civil union often with their husband or partner living away from the family home. This distinction carries important, often culturally specific implications for poverty risk (for a recent review see Hanmer et al. 2020). Further, depending on their education, class, race, ethnicity, and other markers of social identity, including possibly displacement status, female-headed households have different degrees of access to income- 4 generating opportunities, asset ownership and use, and entitlements - for example, support from other family members, receipt of charitable giving from religious institutions or government transfers targeted to vulnerable groups like single parents or widows. In turn, these opportunities determine people’s vulnerability to poverty and their ability to cope with adverse events. Results of research which focuses on subgroups of female-headed households show that distinguishing between different types of female-headed households is critical to understand the links between gender inequality and poverty (Appleton, 1996; Horrell & Krishnan, 2007; Klasen et al., 2015; van de Walle, 2013). Often de facto female heads have access to earnings from migrant male workers and are not particularly disadvantaged compared with male-headed households (Appleton, 1996; Horrell and Krishnan, 2007). De jure female-headed households, on the other hand, are more likely to be among the poorest of the poor. For example, in Mali, households with a widow head have significantly lower living standards than male or other female-headed households and the adverse effects of widowhood persist even after widows are absorbed into male-headed households (van de Walle, 2013). Another strand of the literature has focused on subgroups of female-headed households using demographic classifications. Milazzo and van de Walle (2017) examine how female-headed households have fared in the context of Africa’s remarkable progress in reducing poverty since the mid-1990s. Their classifications include households with and without a male adult and with and without a married head, as well as widow-headed households. They find that households with a married woman head and a male adult performed the best, whereas households with an unmarried female head and no male adult presence experience the least poverty reduction. Further, a number of studies show that allowing for economies of scale in consumption reverses conclusions about the incidence of poverty based on the headship comparison (Drèze & Srinivasan, 1997; Lanjouw & Ravallion, 1995; van de Walle, 2013). This is partly because female-headed households are, on average, smaller. Hence, the traditional per capita measures tend to underestimate their poverty and overestimate that of larger male-headed households (Brown & van de Walle, 2020). 5 Other research has classified households by the number of earners, their sex and the share of their contribution to the household income (Fuwa, 2000; ECLAC, 2004; Grown & Valodia, 2010; Rogan, 2013). For example, ECLAC (2004) classifies households based on the presence and number of people who contribute to the household income and finds when both spouses contribute to household income, households are less likely to be poor, compared to households with a single earner of either sex. Our study builds on the foregoing literature. We use household headship, demographic characteristics, and gender and number of people contributing to household income to explore links between poverty and gender. 3. Somali Context In the aftermath of the war with Ethiopia and the ousting of General Siad Barre, rebellions over most of the territory led to the collapse of the Somali State in 1991 and to the separation of the former British colony of Somaliland. 3 The civil war is complex and has multiple causes, including competition for natural resources and power, large numbers of unemployed youth, and a politicized clan identity (Elmi & Barise, 2006). Thousands of Somali civilians have been killed, died of starvation or been forcibly displaced. To date, Somalia remains marked by high levels of insecurity exacerbated by the Al-Shabaab terrorist group and the weak capacity of state institutions (World Bank, 2018a). Somalia is one of the poorest countries in the world. Gross domestic product (GDP) per capita is US$500, and more than 70 percent of its population is poor, living on less than US$1.90 PPP (2011) per day. With the absence of an effective government structure in most of the country since 1991, clan affiliation became a critical source of social, financial, and human protection. 3 Currently, Somaliland is an autonomous region in northern Somalia that has held independent elections since 2003; it considers itself an independent country, which seeks international recognition of its sovereignty (Council on Foreign Relations, 2018). The adoption of a Provisional Constitution and establishment of the Federal Government in 2012, as well as the subsequent formation of four new Federal Member States, provides space for peace and stability. 6 Half of Somalia’s estimated 12 million people live in rural areas, and many are pastoralists, highly vulnerable to climate shocks. The agriculture sector remains the backbone of the economy and accounts for about 75 percent of GDP, among the highest in the world (World Bank, 2019). The 2011 East Africa drought resulted in the deaths of a quarter million people, half of them children under the age of five years, as well as the displacement of nearly 1 million people. In 2016/17, the country experienced another large-scale drought, leaving an estimated 6.7 million people in urgent need of humanitarian assistance. As of 2020 there were over 2,000 recorded IDP settlement sites in Somalia[7] but information on the exact location of sites, living conditions, access to services and exact numbers of IDPs in each location is sparse (UN Office for the Coordination of Humanitarian Affairs [OCHA] & REACH 2019). UNHCR (2021) estimates that there were 3 million people (about 17 percent of the population) internally displaced at the end of 2020. Many IDPs have been displaced during different phases and types of armed conflict, others by clan conflict, drought or famine or by other disasters triggered by natural hazards, such as recurrent floods. Still others had to move due to forced evictions (Government of Somalia, 2017). It is estimated that the 2017 drought, followed by extreme floods in 2018, displaced more than 926,000 people (World Bank, 2019). The majority of IDPs settle in informal and unplanned settlements characterized by overcrowding, poor shelter, unsanitary conditions, and limited access to basic services (UNDP, 2018). Access to humanitarian assistance and services provided by private contractors are often disrupted. One barrier to services delivery is that most settlements are controlled and dominated by gatekeepers who are often community leaders, members of militias, or landowners (Drumtra, 2014; Government of Somalia, 2017; Yarnell, 2019). IDPs outside settlements who have been disconnected from their networks and IDPs from minority groups tend to be excluded from access to jobs and opportunities and may have been denied permission to live in clan-controlled settlements (Drumtra, 2014). Also, high insecurity can restrict the delivery of services and monitor programs (Refugees International, 2019). In response to these challenges, the federal government 7 created a Durable Solutions Secretariat (DSS) and adopted a national IDP policy alongside National Evictions Guidelines. This response aims to protect the rights of IDPs, ensure that any evictions are carried out in legal way, and provide alternative housing options (Refugees International, 2019). Prolonged conflict has had far reaching effects, including the emergence of a large number female-headed households, with reports of nearly half of all households being female headed going back for nearly a decade (Food Security and Nutrition Analysis Unit [FSNAU], 2012; United Nations Development Program [UNDP], 2012). This is often attributed to the conflict related death of male family members or the inability of men to flee to certain areas (World Bank, 2014). As has been observed in other conflict settings, family splitting is also a strategy to cope with lack of economic opportunities (Brück & Schindler, 2009; Ibáñez, 2008). In Somalia sometimes men opt to return to their areas of origin and women and children are left behind with the intention of returning if and when the situation improves (World Bank, 2014). Somalia ranks among the bottom dozen countries on the Women, Peace and Security Index, which measures women’s inclusion, justice, and security. 4 It has one of the lowest female labor market participation rates in Sub-Saharan Africa with only 43 percent of women in the labor force compared to 67 percent of men. However, women have always contributed to work on family farms. Gender roles have changed; increased numbers of female-headed households have led to changes in herd management practices and women’s greater involvement in livestock trade (El- Bushra & Gardner, 2016). Currently, women are engaged in the informal sector and micro- enterprises and work in agricultural production and livestock activities. Women are heavily involved in crop production, provide more than 60 percent of labor in subsistence farming, and are mainly responsible for marketing fruits and vegetables for domestic consumption and account for 60-80 percent of domestic meat and milk trade, hides and skins, as well as petty trade (UNDP, 4 See Georgetown Institute for Women, Peace and Security/Peace Research Institute of Oslo “Women Peace and Security Index”. https://giwps.georgetown.edu/wp-content/uploads/2019/12/WPS-Index-2019-20-Report.pdf 8 2018, World Bank, 2017). In urban areas, women rely on low skilled, low paying service jobs, including cleaning, washing, and childcare (Federal Government of Somalia, 2018). Women’s voice in society and their ability to exercise leadership are limited. They remained virtually excluded from the political and judicial structures that emerged in different parts of the country after the collapse of the state in 1991 (Gardener, 2007; Mahmood, 2018; Parke et al., 2017). Recent reforms have addressed women’s representation in a number of ways including through the introduction of quotas in national parliamentary elections (United Nations Development Program (UNDP), 2019). Nevertheless, the politicization of clan identity and stigma attached to women entering government and politics remain barriers to women’s political participation and adoption of leadership roles (Parke et al., 2017; UN Women, 2018). Permeating women’s lives are high levels of gender-based violence (GBV). Conflict- related violence against women and girls has been extensively documented (Perrin et al., 2019; UN Women, 2018; United Nations, 2020). GBV is particularly prevalent in IDP settlements, among minority clans, and in conflict zones where all parties have reportedly perpetrated sexual abuse and violence (Perrin et al., 2019; United Kingdom: Home Office, 2017). Increasing rates of violence in the home have been linked to changing gender roles as men experience changing dynamics within the household and more women move into breadwinner positions (OCHA, 2019). For women, heightened risk of sexual and gender-based violence is exacerbated by overcrowded makeshift facilities without adequate safety and security measures – from lighting to safe water, and sanitation and hygiene facilities with privacy. Child marriage, another form of GBV is common throughout the country, with 35 percent of women aged 20 to 24 married before the age of 18. 5 An estimated 98 percent of Somali women have undergone female genital mutilation (United Nations General Assembly, 2018). 5 World Bank’s Gender Data Portal based on Somalia Demographic and Health Survey (2020). 9 4. Data and Descriptive Statistics The first wave of the Somali High Frequency Survey (HFS) 6 was conducted in 2016. The second wave conducted in 2017 extended coverage to include nomads and households in insecure areas. The sample of IDPs and non-displaced Somalians is representative of the Somali population living in secure areas. 7 We use the second wave of the HFS which samples 4,780 households (26,317 individuals). IDPs account for 33 percent of the survey population, most of whom reside in IDP settlements in the South West and in urban Jubbaland. Table 1. Household and household head characteristics, by displacement status and gender Settlement IDP Non-settlement IDP Non-IDP Total Female Male All Female Male All Female Male All Female Male All Female-headed (%) 55.4 41.7 51.0 50.5 De jure 31.0 31.0 25.6 25.6 26.7 26.7 27.0 De facto 68.9 68.9 74.4 74.4 73.2 73.2 72.9 Household characteristics Urban (%) 38 32 34 79 69 74 74 63 68 Rural (%) 62 68 66 21 31 26 26 37 32 Household size 6 5 5 6 6 6 5 5 5 5 5 5 Number of children 3 3 3 3 4 3 3 3 3 3 3 3 Dependency ratio 2 1 1 1 2 1 1 1 1 1 1 1 Able bodied adults (%) 46 48 47 54 43 47 52 49 50 51 48 49 Household head characteristics Age 36 38 37 41 43 42 37 40 38 37 40 38 Education (%) No education 88 64 76 90 69 77 51 38 44 68 50 58 Primary or below 9 24 16 8 21 16 24 26 25 17 25 21 Secondary and above 3 12 7 1 10 7 25 36 31 15 26 21 Number of households 226 184 410 186 263 449 1,998 1,923 3,921 2,410 2,370 4,780 Observations (number of individuals) 1,254 970 2,224 1,009 1,426 2,435 10,969 10,689 21,658 13,232 13,085 26,317 Source: Estimated by authors using Somali HFS 2017-18. Note: IDP stands for internally displaced persons. Notes HH = household head 6 Surveys were conducted by the World Bank in collaboration with the Somali statistical authorities. They can be accessed in the World Banks micro data library see https://microdata.worldbank.org/index.php/catalog/2738 and https://microdata.worldbank.org/index.php/catalog/3181 7 Rates of poverty and other markers of well-being may thus be different from those in different regions inaccessible to the HFS. 10 Poverty incidence was calculated using a consumption aggregate from the survey’s consumption module and the inflation-adjusted international poverty line of US$1.90 per person per day (Pape et al., 2019, Vol C). Household heads were identified by the survey respondent who is asked who s/he considers to be the household head. About half of all households are headed by women in Somalia, although for IDP households living outside settlements, female headship is less common (36 percent). All IDP settlements are located in urban areas and about three-quarters of non-IDP households live in urban areas; however, a large share of IDPs living outside settlements (66 percent) are located in rural areas. The average household size is 5 people, three children and two adults. There is no difference between settlement IDPs and non-IDPs household size. Non-settlement IDP households are slightly larger than average with 6 people and settlement IDP households that are female- headed are also larger - 6 people on average (Table 1). Figure 1 shows that most female heads of both IDP and non-displaced households are married, hence, they are de facto heads. About two-thirds of female household heads are de facto in non-IDP communities compared to just over half of female-headed IDP households living outside settlements. There is a slightly higher proportion of de jure female heads in the non- settlement IDP sample (43 percent) than in the settlement and non-displaced samples (34 and 30 percent, respectively). Figure 2 shows that education levels are strikingly low, particularly among IDPs, and there are large gender gaps. On average 93 percent of all IDP heads have primary or less education compared to 69 percent of non-IDP household heads. For both IDPs and non-IDPs the gender gap is large. Female heads in settlements are significantly less educated. Only about 1 in 10 female heads in IDP settlements has some secondary education compared to nearly 1 in 4 male heads in settlements. The gender gap is similar for non-settlement IDPs. In contrast, for non-IDPs, the gender gap emerges at the tertiary level and secondary schooling rates are similar for male and female heads at around 25 percent. 11 Figure 1. Headship and marital status, by displacement status Source: Prepared by authors based on Somali HFS 2017-18. Note: IDP stands for internally displaced persons. Figure 2. Percentage of household heads, by education level, sex, and displacement status Source: Prepared by authors based on Somali HFS 2017-18. Note: IDP stands for internally displaced persons. 12 Almost half of IDPs (48 percent) ages 15–64 are economically active, meaning that they have worked (45 percent) or are unemployed but have looked for work (3 percent) in the last seven days. This is similar to the economically active share of the urban population (49 percent) and rural population (48 percent) (Pape et al, 2019 Vol B). However, there are large gender gaps in economic activity rates, especially for IDP households living in settlements (Figure 3). Pape et al. (2019 Vol B) find that people living in female-headed households are much more likely than those in male- headed households to be economically inactive because they are caring for their families or households. Men, in contrast, are much more likely than IDP women to not be working because of illness or disability. Among IDPs 59 percent of women but only 24 percent of men are economically inactive because of family and household care responsibilities. Figure 3. Share of economically active members, by sex of household head and displacement status Source: Prepared by authors based on Somali HFS 2017-18. Note: IDP stands for internally displaced persons. Pape et al. (2019 Vol B) find that about 40 percent of IDPs have salaried or in-kind labor as their main income source, and one in five relies on small family businesses. In addition, Figure 4 shows that there are other differences including between sources of income according to gender and displacement status of the household head that are worth noting. First, unsurprisingly, IDPs rely more on aid/zakat than non-IDPs. Second, non-settlement IDPs rely on agriculture for a larger share of their income than both settlement IDPs and non-IDPs. Third, all households receive remittances and there is no marked difference in the share of remittance income by displacement status. The share of income accounted for by remittances is slightly larger for female-headed 13 households that are not displaced or live in IDP settlements than for other types of households. Fourth, income from work accounts for a larger share of income for female headed IDP households living in settlements than for any other group. Figure 4. Main sources of income, by gender of the household head and displacement status Source: Prepared by authors based on Somali HFS 2017-18. Note: IDP stands for internally displaced persons. 5. Empirical Approach The summary statistics point to some differences in poverty risk factors between female and male household heads, and between IDPs and non-IDPs. To deepen the analysis, we analyze differences in risk factors beyond the traditional comparison of the gender of the household heads. First, we use the survey data of self-reported male and female headship to compare poverty rates between IDPs and non-displaced households, then disaggregate by de jure and de facto female headship. Second, to examine whether caregiving responsibilities increase poverty risk, we categorize households by demographic composition. Third, households are categorized by income profiles: incomes sources are earnings from salaries, family businesses or farms and remittances. Table 2 shows the categories of demographic and income profiles – each household in the data set falls into one of the categories for each profile type (full definitions can be found in Annex 1). 14 Table 2. Household demographic and income profile categories Demographic Composition Income profile Male single caregiver No earners, no remittances Female single caregiver Remittance recipients only Couple with children Female single earner Multigeneration with children Male single earner Family without children Equal number of male & female earners Majority female earners Majority male earners To examine the role of gender-based factors in poverty rates among IDPs and non-IDPs, we estimate Linear Probability Models (LPMs) with robust standard errors for both groups and a pooled regression model, controlling for the effect of different household characteristics. 8 Our model is: = ′ + ′ + ′ + (1) Where i represents a household; =1 if household i is below the poverty line and =0 if the household is on or above the poverty line; represents a vector of household characteristics, including its size, dependency ratio, whether it has access to land and whether it has access to identity documents; is a vector of characteristics of the household head including gender, age, economically active, whether a widow or not and education; is a vector of household types – gender of the household head, the demographic categories and the employment categories described above; and is a normally distributed error term. Thus, is the vector of coefficients of interest since it captures the relative importance of the family types on the risk of falling into poverty. We draw on the economic literature in this field (Appleton, 1996; Buvinic & Gupta, 1997; Chant, 1997; Hanmer et al 2020; Milazzo & van de Walle 2020) to select other correlates that are 8 A Kolmogorov-Smirnov test rejects the hypothesis that the total per capita consumption for IDP and non-IDP households is drawn from the same distribution; hence, we run separate regressions to allow for heterogeneous effects of the independent variables on the incidence of household poverty for both groups. 15 expected to be important. 9 Our choices are supported by results of a Lasso algorithm run with a wide selection of household and household head characteristics. Some of the strongest predictors of poverty are large household size, having no male earners, lack of education, and being an IDP household living outside settlements (these results are available upon request). We run separate regressions for IDPs and non-IDPs to allow for heterogeneous effects of the independent variables on the incidence of household poverty for IDP and non-IDP households. 10 Finally, we assess whether the household classifications add value to an analysis based on headship alone. We use regression models to estimate the correlation between the different family types to assess whether specific family categories are complementary or substitutes of the others. 6. Results Poverty rates are high in Somalia, and highest among IDPs. Seven out of 10 households live below the poverty line. IDP households living in and outside settlements are 10 percentage points more likely to be poor than non-IDPs (77 compared to 67 percent) (Table 3). Poverty in Somalia is higher in rural than in urban areas (76 versus 64 percent). Almost all IDPs living in rural areas are poor (83 percent). Poverty rates of non-settlement IDPs living in urban areas are about the same as those for the rest of the urban population. 9 Lasso is a simple machine learning algorithm that selects the most relevant variables by minimizing the prediction error of the model, while penalizing the coefficients of the regression variables shrinking the coefficients of the variables that are least predictive of poverty to zero. (Zou & Hastie, 2005). 10 Pooled results with interactions are available on request from the authors. The definition of the intensity of poverty is called the normalized poverty gap in Foster, J. et al. (2013). 16 Table 3. Poverty rates by geographic area, headship, and displacement Poverty rate Observations Non- Non- Settleme Non- Settleme Non- settlement All settlement All nt IDP IDP nt IDP IDP IDP IDP All 77 77.3 66.5 70 2,220 2,427 21,610 26,257 Geographic area Urban 62.7 64.1 64 1,562 17,469 19,031 Rural 83.5 72.8 76.3 865 4,141 5,006 Headship Female 70.8 70 64.8 66.7 1,254 1,009 10,969 13,232 de jure 68.8 64.4 67.7 67.5 383 292 3,124 3,799 de facto 70.1 72.2 63.2 65.7 856 690 7,627 9,173 Male 84.7 81.1 68 72.9 966 1,418 10,641 13,025 Source: Own elaboration based on Somali HFS 2017-18. Note: IDP stands for internally displaced persons. Observations correspond to the number of individuals in the dataset. 6.1. Are female-headed IDP households poorer than male-headed IDP households and male- and female-headed non-IDP households? The incidence of poverty is higher for male-headed households than female-headed ones (73 versus 67 percent) (Table 3). This result is driven by the large gender gap in poverty (against male- headed households) among IDPs (11-14 percentage points), as well as the smaller but significant difference in poverty between male- and female-headed non-IDP households (3 percentage points). Even when distinguishing between de jure and de facto female-headed IDP households poverty remains lower than male-headed households. For non-IDPs there is no significant 17 difference between de jure female heads and male-headed households and poverty rates remain lower for de facto female heads. Turning to differences between female-headed households, poverty rates are higher among de facto female-headed IDP households than de facto female-headed non-IDP households (71 versus 63 percent). For de jure female heads poverty rates are similar regardless of displacement status. In sum, de facto female-headed IDP households are poorer than their counterpart non-IDP households, but male-headed households are poorest regardless of displacement category. We explain below why this is the case. Figure 5 shows the intensity of poverty experienced by poor households. Poverty intensity is measured as the household per capita consumption as a percentage of the poverty line (Foster et al, 2013). The farther the household from the poverty line, the more intensely the household experiences poverty. Thus, a score of 100 percent implies the household is located exactly at the poverty line. A 50 percent score means that the household consumption is 50 percent of that of the average household situated on the poverty line. Figure 5. Poverty intensity, by gender of the household head and displacement status Source: Own elaboration based on Somali HFS 2017-18. Note: IDP stands for internally displaced person. Poverty intensity bands are displayed in this graph. For poor households, poverty intensity is measured as the household income as the percentage of the poverty line. A 100% implies the household is located exactly at the poverty line. A 50% means that the household income is 50% that of the household situated in the poverty line; thus, the poverty intensity is higher. 18 Female-headed poor IDP households located outside settlements experience poverty more intensely than any other group: 64 percent of these households spend less than 50 percent of the amount an average household on the poverty line spends per capita. Poverty intensity experienced by non-IDPs is similar for male- and female-headed households although there are more female than male-headed households with expenditures close to the poverty line (12 versus 9 percent). For settlement IDPs poverty intensity is higher for male-headed household. 61 percent spend less than 50 percent of the average expenditure per capita of households at the poverty line and only 7 percent spend an amount within 10 percent of the poverty line compared to 17 percent of female- headed households. 6.2. Beyond the gender of the household head, are specific types of IDP and non-IDP households more likely to be poor than others? We examine the correlation between poverty rates and the demographic and income categories of households controlling for household characteristics, characteristics of the household head and displacement status (Table 4). Model 1 presents the results for IDPs, controlling for whether they live outside or inside settlements. Model 2 presents results for non-IDPs and model 3 presents results for all households controlling for displacement status. As a robustness check we pool all the data and interact the explanatory variables with IDP status. Table 4. Poverty Correlates: IDP and Non-IDP Households (1) (2) (3) Variables IDPs Non-IDPs All Household characteristics Size (number of members) 0.084*** 0.075*** 0.080*** (0.007) (0.006) (0.005) Dependency ratio -0.026* 0.056*** 0.024** (0.015) (0.016) (0.011) Separated members (1=yes) -0.052 -0.129** -0.120*** 19 (0.050) (0.051) (0.040) Access to land (1=yes) -0.008 -0.070 -0.067** (0.029) (0.048) (0.033) Access to mechanisms to obtain/replace lost docs (1=yes) -0.070*** -0.076*** -0.090*** (0.026) (0.027) (0.022) Characteristics of the household head Male 0.119*** -0.001 0.060*** (0.024) (0.028) (0.020) Age -0.005 0.003 -0.003 (0.005) (0.006) (0.005) Age squared 0.000 -0.000 0.000 (0.000) (0.000) (0.000) Active in the last 12 months 0.037 -0.021 0.013 (0.027) (0.034) (0.025) Widow/er 0.178 -0.397*** -0.431*** (0.154) (0.097) (0.097) Female*Widow -0.082 0.422*** 0.484*** (0.162) (0.113) (0.106) Education (base: No education) Primary or below -0.025 -0.000 -0.009 (0.026) (0.034) (0.028) Secondary and above -0.130** -0.011 -0.030 (0.054) (0.029) (0.026) Household type Demographic classification (base: Families without children) Male Single Caregivers 0.055 0.148 0.135 (0.108) (0.122) (0.094) Female Single Caregivers 0.203** 0.094 0.139** (0.084) (0.080) (0.063) Couples with children 0.183*** 0.060 0.093** (0.050) (0.050) (0.040) Multi-generation with children 0.170*** 0.119** 0.121*** (0.049) (0.050) (0.040) 20 Income classification (base: No earners) Remittance recipients only 0.058 -0.202*** -0.161*** (0.069) (0.065) (0.052) Female single earner -0.095* -0.176*** -0.201*** (0.055) (0.060) (0.049) Male single earner -0.212*** -0.230*** -0.273*** (0.040) (0.049) (0.043) Majority female earners -0.403*** -0.237*** -0.328*** (0.054) (0.065) (0.055) Equal contribution -0.201*** -0.246*** -0.268*** (0.043) (0.055) (0.046) Majority male earners -0.360*** -0.308*** -0.400*** (0.065) (0.065) (0.053) Displacement status (base: settlement IDP) Non-settlement IDP -0.048* -0.076*** (0.028) (0.028) Non-IDP -0.100*** (0.023) Constant -0.256** 0.114 0.314 (0.116) (0.205) (0.196) Observations 1,902 6,341 8,243 R-squared 0.583 0.299 0.368 Source: Estimated by authors based on Somali HFS 2017-18. Note: Robust standard errors in parentheses. Estimations include regional dummies. *** p<0.01, ** p<0.05, * p<0.1. Note: Following the Durable Solutions Analysis Guide (Joint IDP Profiling Service [JIPS], 2018), the variable mechanisms to obtain or replace lost documents refers to access to support systems that grant or restore personal or other documentation such as IDs, land and housing titles. We find some association between gender of the household head and poverty emerges for IDPs, controlling for other individual and household characteristics. Male headship remains associated with higher poverty for IDPs but not for non-IDPs. Among non-IDPs widow-headed households emerge as a group that with high risk of poverty but this result does not hold for IDPs. 21 The demographic and income categories are significant across all three models, which shows there is an association between poverty risk and specific household types after controlling for the gender of household head and other individual and household characteristics. However, the strength of the association varies significantly between IDPs and non-IDPs, especially for demographic household types. Results shown in Table 4 support our assumption that caregiving responsibilities impact poverty risk for IDP households. Compared to families without children, IDP households consisting of female single caregivers, couples with children, and multiple generations with children are between 17-20 percentage points more likely to be poor. In contrast, among non-IDPs, only households consisting of multiple generations with children are more likely to be poor than families without children. Household income profiles affect poverty risk for both IDPs and non-IDPs. Multiple earners, regardless of their gender are associated with lower poverty risk. For example, compared with households of no earners, IDP households with multiple female or male earners are between 36-40 percentage points less likely to be poor. However, households with a male single earner or a female single earner are only 21 and 9 percentage points less likely to be poor, compared to no earners. Household type aside, other household characteristics are associated with poverty risk. Some of them are worth highlighting: ● Having an additional household member in the household increases the odds of falling into poverty by about 8 percentage points. This risk factor is slightly higher for IDP households. ● For IDPs, higher dependency ratios reduce the likelihood of being poor, possibly as a high dependency ratio is a criterion for humanitarian assistance (UNOCHA & Reach Initiative, 2019). In contrast, higher dependency ratios are linked to higher poverty risk for non-IDPs. 22 ● Having one or more members separated from the household reduces the probability of experiencing poverty for non-IDPs, but not for IDPs. Remittances for absent household members could explain this result. ● Having an ID or a means of legally identity is strongly associated with reduced poverty risk for both IDPs and non-IDPs, possibly as an ID enables access to public services and more formal labor markets. ● Compared to having no education or only primary education, having secondary education or above reduces poverty risk by 13 percentage points for IDPs but not for non-IDPs. As unemployment is high and there are few economic opportunities in Somalia, the lack of association is not surprising, however the association between education and poverty reduction among IDPs deserves further exploration. In sum, IDP and non-IDP households in Somalia face large and specific risks of poverty. Many of these differences are associated with differences in the demographic characteristics and income profiles of their households rather than gender of the household head, although widows emerge as a vulnerable group of non-IDPs. The demographic features of households have strong associations with poverty risk for IDPs but not for non-IDPs: specifically, among IDPs female single caregivers have the highest poverty risk and followed other types of families with children (couples and multi-generational families). The income profile of the households is associated with poverty risk for both IDPs and non-IDPs, having more earners is associated with a reduced risk of poverty for both IDPs and non- IDPs and for IDP living in a majority female earner household is associated with more poverty risk reduction than living in a majority male earner household. 6.3 Do household classifications add value to the analysis of poverty? Our results show that different household types are associated with different degrees of poverty risk. But do these categories add value to the traditional comparison between male and female 23 household heads? We estimate the correlation coefficients between all categories used in this analysis to investigate this question. If two of the household categories are highly correlated (correlations close to 1 or -1), the two variables are likely collinear and hence including both variables in a regression will not improve the precision of the estimates because both variables are proxies of the same dimension of gender-based disadvantage. We use linear regressions to estimate the correlations as we have categorical variables formed by multiple mutually exclusive categories. High joint significance tests between gender of household heads and demographic or income categories would suggest that these classifications do not add value to the gender analysis by household headship. The correlation coefficients estimates are reported in Table 5. The results show that the income classification is weakly linked with the gender of the household head; correlations are low (below 40 percent) indicating that these different household types are only slightly related to the gender of the household head. In contrast, the household types in the demographic composition are highly correlated with the headship classification, suggesting that both definitions capture similar dimensions of vulnerability for households in Somalia. Hence, while the income classification provides information not captured by headship definition, there is less value added by using the demographic categories and the differentiation between female and male heads together. The larger correlation between demographic categories and headship may be partially explained because demographic household types are not associated with poverty risk for non-IDPs. Columns (2)-(7) present the results of the same exercise restricting the sample to only female- headed households to examine the extent to which alternative household classifications capture the heterogeneity of this group. Overall, the coefficients confirm that the income and demographic classifications provide complementary information in analyzing the vulnerability of female- headed households. Distinguishing between de jure and de facto female heads adds value relative to the employment classification but appears to somewhat overlap with some of the demographic 24 features. For instance, de facto female-headed households and couples with children are highly correlated as well as de jure female-headed households and multigeneration families with children. Table 5. Regression results for correlations between headship and demographic composition and income composition of households. (1) (2) (3) (4) (5) (6) (7) Single Couples Families De facto De jure Caregiver with Multigeneration without Variables FHH FHH FHH s children with children children Income composition Remittance recipients -0.079** -0.161*** 0.161*** 0.253*** 0.071* -0.178*** -0.146*** (0.040) (0.035) (0.035) (0.033) (0.041) (0.043) (0.035) Female single earner 0.122*** -0.206*** 0.206*** 0.387*** 0.058 -0.219*** -0.225*** (0.036) (0.032) (0.032) (0.029) (0.036) (0.038) (0.031) Male single earner -0.229*** 0.033 -0.033 0.054** 0.245*** -0.097*** -0.203*** (0.031) (0.029) (0.029) (0.027) (0.033) (0.035) (0.028) Equal contribution -0.120*** 0.016 -0.016 0.043 0.245*** -0.076** -0.212*** (0.032) (0.029) (0.029) (0.026) (0.033) (0.035) (0.028) Majority female earners 0.024 -0.167*** 0.167*** -0.005 0.012 0.060 -0.067* (0.041) (0.035) (0.035) (0.033) (0.041) (0.043) (0.035) Majority male earners -0.196*** (0.039) No earners -0.143*** 0.143*** 0.243*** 0.097** -0.213*** -0.127*** (0.035) (0.035) (0.032) (0.040) (0.042) (0.034) Demographic composition Single Caregivers 0.834*** 0.271*** 0.729*** (0.035) (0.033) (0.033) Couples with children 0.583*** 0.991*** 0.009 (0.030) (0.027) (0.027) Multigeneration with children 0.686*** 0.474*** 0.526*** 25 (0.032) (0.029) (0.029) Families without children 0.633*** 0.648*** 0.352*** (0.034) (0.030) (0.030) Female headship De facto FHH -0.061** 0.591*** 0.203*** 0.267*** (0.024) (0.031) (0.032) (0.026) De jure FHH 0.228*** -0.054 0.517*** 0.309*** (0.027) (0.033) (0.035) (0.028) Observations 4,749 2,359 2,359 2,359 2,359 2,359 2,359 R-squared 0.556 0.872 0.714 0.504 0.782 0.297 0.145 Source: Own elaboration based on Somali HFS 2017-18. Note: ***, **, and * denote statistical significance at the 1, 5, and 10 percent levels, respectively. Standard errors in parentheses. 7. Conclusions Our findings show that even when poverty rates are very high, there are important differences between poverty risk according to gender and displacement status. Distinguishing between different types of households beyond the gender of the household head improves our understanding of poverty in situations of forced displacement. The association between a household’s demographic composition, its income profile and poverty risk shows that women’s lack of economic empowerment and their caring responsibilities elevate poverty risk. If we had simply stopped at female headship, we may have concluded that – aside from widows - women are better off in Somalia. Our analysis extends the gender and poverty literature by establishing a link between the demographic and income profiles of households and the risk of experiencing poverty in situations of displacement. Using data that is typically available in household surveys, we showed how this can be done. Our starting point is that women as individuals or households dependent on women’s income may be at greater risk of poverty due to two interrelated gender norms. These two norms are first, the norms assigning women a disproportionate responsibility for care work and the provision of household services, and second the norms that restrict women’s ability to access paid 26 economic opportunities. We show that, for both IDP and non-displaced households, the risk of experiencing poverty decreases with more earners, both male and female. Interestingly, IDP households with majority female earners have a lower likelihood of poverty than other IDP households, but this result does not hold for non-displaced households. This may be because some of the normative constraints to women’s employment in waged or paid in kind work outside the household are lifted during displacement. We also show that there are some marked differences between IDPs and non-IDPs and the link between a household’s likelihood of being poor and its demographic composition. For IDPs their household’s demographic composition is linked to poverty, but this does not hold for non-IDPs. Specifically, compared to IDP families without children, IDP households consisting of female single caregivers and couples with children are between 17-20 percentage points more likely to be poor. These results are consistent with the view that disruptions to family structure during displacement increase IDPs risk of poverty. Our analysis highlights the importance of poverty reduction policies and programs for all households, especially those with children. The Government of Somalia’s new policy framework on social protection – which post-dates the data analyzed here – appears to respond to this need. It aims to create a formal safety net delivery system that provides predictable, reliable, and scalable assistance to households facing chronic food insecurity, complementing systems of humanitarian assistance that provide emergency relief to prevent famine and save lives (World Bank, 2019). The Baxnaano program is the first government cash transfer program in Somalia developed to provide a shock responsive and scalable social safety net system. It provides a benefit set worth USD20 per household per month to poor and vulnerable households that suffer chronic food insecurity. Its focus is on rural areas that are not covered by humanitarian assistance programs, which mainly target urban areas and IDP settlements where populations suffer emergency and crisis levels of food insecurity. Within each state, targeting is conducted in three stages, namely, selection of districts, selection of communities within districts, and cash-based targeting of households using 27 specific eligibility criteria. Key features of the targeting include using a ‘distress index’ consisting of nutrition, trends of shocks, etc.; criteria for geographic targeting favoring rural districts; and criteria specific per household composition, with a strong focus on women with children under five. Recognizing the gender gap in access to income and livelihood opportunities, women are the main recipients of the transfer. In addition, the program provides emergency cash assistance to households affected by the locust crisis to protect their food security and productive assets (Gentilini et al., 2020; World Bank, 2019). In Somalia mobile money has potential to provide the platform needed to scale up social protection and humanitarian assistance (Daniels & Anderson, 2018). In other FCV settings, cash transfers delivered through electronic payments to address mobility and security constraints have had promising results. For example, in Niger, Aker et al. (2016) find that recipients of electronic transfers (m–transfer) were able to save up to 20 hours in terms of travel time and wait–time and had more flexibility on which day to travel to get their transfers relative to the women who received cash transfers. This “extra” time translated into a 7-13 percentage point higher probability of growing crops. In Somalia almost three-quarters of the population ages 16 and older use mobile money. Penetration rates are highest in urban areas (83 percent) and camps for internally displaced people (72 percent), and 55 percent of the population uses mobile money in rural areas. Mobile money is now the main transaction instrument used by both individuals and businesses in Somalia (World Bank, 2018b). However, government regulation is needed to address concerns and uncertainties about risk and vulnerabilities of the system and promote greater competition and innovation (World Bank, 2018b), and lack of IDs and limited rural coverage are challenges that need to be addressed (Daniels & Anderson, 2018). Moving beyond cash transfers to other social protection policies, our findings suggest that the design of work programs accounts for the occupations and sectors where forcibly displaced women and men work, and whether specific activities are needed to increase women’s earnings and productivity. For example, public works programs with a heavy manual labor component often 28 put women at a disadvantage (Palmer-Jones & Jackson, 1997). Such constraints can be alleviated by including specific activities that are suitable for both men and women, e.g., as material preparation and planting trees. Equally important is to ensure that livelihood programs are based on market analysis and participatory approaches that proactively include women. In some cases, as is the case of Liberia, forcibly displaced women have crossed over into non-traditional activities such as reforestation and construction projects (Fuest, 2008). Livelihood programs that aim to increase women’s economic empowerment should incorporate complementary components to address barriers preventing women from benefitting from these interventions. For instance, embedding support services such as community-childcare facilities, flexible working hours, and facilitating transport arrangements or assigning women to work close to their residences in large-scale programs can increase women's participation in income-generating activities (de Silva et al., 2020). Equally important is to ensure equal wages between women and men in public works programs. Many humanitarian and social protection programs in Somalia use community-based targeting, often with women as the main beneficiaries. Extending this type of programming by targeting men and community leaders with communication campaigns and engaging them in all stages of the program can contribute to women’s empowerment. For example, in Burundi, a microfinance program consisting of village savings and loan associations combined with discussion groups on the role of women in household decision-making improved financial autonomy for women, reduced exposure to and acceptance of violence among beneficiaries, and increased the consumption of household goods relative to luxury goods such as alcohol (Iyengar & Ferrari, 2016). Ultra-poor graduation programs that combine multiple interventions at once also hold promise to lift several constraints affecting IDP and non-IDP women at once. These programs combine safety nets, life skills training, saving strategies, as well as support for income generation activities such as seed capital, and business mentoring. These programs have the potential to 29 increase self-employment opportunities, improve women’s access to income, help them better manage their money through financial education and access to credit. For example, BRAC’s ultra- poor graduation program in Bangladesh combines large-scale business asset transfers and skills training to target the most disadvantaged women in selected communities that were not recipients of anti-poverty government transfers or microfinance lending. An impact evaluation revealed that as a result of the program, targeted women shifted their working hours from casual wage labor towards livestock rearing, increasing both total hours worked and earnings. After four years, beneficiary women worked 22 percent more hours and 25 percent more days, while their earnings had increased by 37 percent (Balboni et al., 2015). Finally, IPV is a significant challenge in FCV contexts, and particularly in Somalia. Displacement linked to conflict and disaster can worsen the risk of violence. Programs that address norms and beliefs that underlie gender inequality, while providing women with economic opportunities and enhancing life skills, have proven successful in reducing intimate partner violence in many low income and conflict settings (Arango et al, 2014). At the same time, it is critical to ensure that poverty reduction interventions do not exacerbate the situation and are developed with good understanding of available GBV response services and community advocates and partners. 30 Annex 1: Definitions of household categories The demographic profile divides households into five mutually exclusive categories: • Male single caregiver: One male adult living with children of his/her own below the age of 18; and/or children of others (classified by UNHCR as separated children) and/or one or more elderly and/or disabled persons. • Female single caregiver: One female adult living with children of his/her own below the age of 18; and/or separated children; and/or one or more elderly and/or disabled persons. • Couple with children: Married or cohabiting couple (at least one of whom is over the age of 18, not disabled and/or elderly) with children (below the age of 18) of their own and/or children of others. • Multigeneration with children: Includes extended family households, households of siblings, and polygamous households. • Family without children: Includes married or cohabiting couples (at least one of whom is over the age of 18, disabled and/or elderly) without children and single person households reporting no other dependents or cohabitees. The income profile divides households into seven mutually exclusive categories: • No earners: Household where none of its members are engaged in economic activity and do not receive remittances. • Remittance recipients only: Household whose only source of income is remittances. No household member is economically active. • Female single earner: Household composed of two or more people, one of whom is a woman engaged in at least one economic activity. This woman is 15-64 years old and either employed, a farm worker or self-employed. Her contribution to the household income can also come from remittances she receives. • Male single earner: Household composed of two or more people, one of whom is a man engaged in at least one economic activity. This man is 15-64 years old and either employed, a farm worker or self-employed. His contribution to the household income can also come from remittances he receives. 31 • Equal contribution earners: Household with more than one member who receives remittances or engages in economic activity through paid employment, farm employment or self-employment. Exactly 50 percent of those who contribute to the household income are women. • Majority female earners: Household with more than one member who receives remittances or is engaged in economic activity through paid employment, farm employment or self-employment. More than 50 percent of those who contribute to the household’s income are women. • Majority male earners: Household with more than one member who receives remittances or is engaged in economic activity through paid employment, farm employment or self- employment. More than 50 percent of those who contribute to the household’s income are men. 32 References Appleton, S. (1996). Women-Headed Households and Household Welfare: An Empirical Deconstruction for Uganda. World Development, 24(12), 1811–1827. Arango, D., Morton, M. Gennari, F., Kiplesund, S, and Ellsberg, M. (2014). Interventions to Prevent and Reduce Violence Against Women and Girls: A Systematic Review of Reviews. Women's Voice, Agency, and Participation Research Series. Washington, DC: World Bank. Arora, D. (2015). Gender Differences in Time-Poverty in Rural Mozambique. Review of Social Economy, 73(2), 196–221. Connelly R., & E. Kongar (Eds.) (2017) Gender and Time Use in a Global Context. Palgrave Macmillan US. Balboni, C., Bandiera, O., Burgress, R., & Kaul, U. (2015). Transforming the economic lives of the ultra-poor [Growth Brief]. International Growth Centre. https://bracultrapoorgraduation.org/wp-content/uploads/2017/02/Transforming-the- Economic-Lives-of-the-Ultra-Poor.pdf Bardasi, E., & Wodon, Q. (2010). Working Long Hours and Having No Choice: Time Poverty in Guinea. Feminist Economics, 16(3), 45–78. Bosch, M., & Maloney, W. F. (2010). Comparative analysis of labor market dynamics using Markov processes: An application to informality. Labour Economics, 17(4), 621–631. Brown, C., & Van de Walle, D. (2020). Headship and Poverty in Africa (Working Paper 531). Center for Global Development. Brück, T., & 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, Braunstein, E., van Staveren, I., & Tavani, D. (2011). Embedding Care and Unpaid Work in Macroeconomic Modeling: Structuralist Approach. Feminist Economics, 17(4), 5–31. Buvinic, 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. Bussolo, Maurizio; Lopez-Calva, Luis F. 2014. Shared Prosperity: Paving the Way in Europe and Central Asia. Europe and Central Asia Studies. Washington, DC. Chant, S. (1997). Women-Headed Households: Poorest of the Poor?: Perspectives from Mexico, Costa Rica and the Philippines. IDS Bulletin, 28(3), 26–48. Chant, S. (2004). Dangerous Equations? How Female-headed Households Became the Poorest of the Poor: Causes, Consequences and Cautions. IDS Bulletin, 35(4), 19–26. Chant, S. (2008). The ‘Feminisation of Poverty’ and the ‘Feminisation’ of Anti-Poverty Programmes: Room for Revision? The Journal of Development Studies, 44(2), 165–197. 33 Christensen, Asger, & Harild, N. (2009). Forced displacement: The development challenge (Conflict, Crime & Violence Issue Note). The World Bank. Council on Foreign Relations. (2018). Somaliland: The Horn of Africa’s Breakaway State. https://www.cfr.org/backgrounder/somaliland-horn-africas-breakaway-state Daniels C. Mike, and Georgina Anderson (2018). Evaluation of the 2017 Somalia Humanitarian Cash-Based Response Commissioned by the Somalia Inter-Agency Cash Working Group https://www.careevaluations.org/wp-content/uploads/Final-CWG-EVAL-REPORT- PRINT-READY-LOW-RES-FOR-EMAILING.pdf DeGraff, D. S., Levison, D., & Dungumaro, E. W. (2017). Environmental Chores, Household Time Use, and Gender in Rural Tanzania. In R. Connelly & E. Kongar (Eds.), Gender and Time Use in a Global Context (pp. 407–434). Palgrave Macmillan US. de Silva, S., Hasan, A., Ouedraogo, A., & Rubiano-Matulevich, E. (2020). Getting it Right: Strengthening Gender Outcomes in South Sudan (No. 2002; Social Protection & Jobs Discussion Paper). World Bank. Drumtra, J. (2014) Internal Displacement in Somalia, Brookings Institution, Washington, DC. Drèze, J., & Srinivasan, P. V. (1997). Widowhood and poverty in rural India: Some inferences from household survey data. Journal of Development Economics, 54(2), 217–234. El-Bushra, J., & Gardner, J. (2016). The impact of war on Somali men: Feminist analysis of masculinities and gender relations in a fragile context. Gender & Development, 24(3), 443–458. Elmi, A. A., & Barise, A. (2006). The Somali Conflict: Root causes, obstacles, and peace- building strategies. African Security Review, 15(1), 32–54. Floro, M. S. (1995). Women’s well-being, poverty, and work intensity. Feminist Economics, 1(3), 1–25. Folbre, N. (2006). Measuring Care: Gender, Empowerment, and the Care Economy. Journal of Human Development, 7(2), 183–199. https://doi.org/10.1080/14649880600768512 Folbre, N., Yoon, J., Finnoff, K., & Fuligni, A. S. (2005). By What Measure? Family Time Devoted to Children in the United States. Demography, 42(2), 373–390. Foster, James, Suman Seth, Michael Lokshin, and Zurab Sajaia. 2013. A Unified Approach to Measuring Poverty and Inequality: Theory and Practice. Washington, DC: World Bank. Food Security and Nutrition Analysis Unit [FSNAU], 2012 Gender in Emergency Food Security and Nutrition in Somalia. https://reliefweb.int/report/somalia/gender-emergency-food- security-livelihoods-and-nutrition-somalia Fuest, V. (2008). “This is the Time to Get in Front”: Changing Roles and Opportunities for Women in Liberia. African Affairs, 107(427), 201–224. Fuwa, N. (2000). The Poverty and Heterogeneity Among Female-Headed Households Revisited: The Case of Panama. World Development, 28(8), 1515–1542. 34 Gammage, S. (2010). Time Pressed and Time Poor: Unpaid Household Work in Guatemala. Feminist Economics, 16(3), 79–112. Gentilini, U., Almenfi, M., & Dale, P. (2020). Social Protection and Jobs Responses to COVID- 19: A Real-Time Review of Country Measures. World Bank. Grown, C., & Valodia, I. (Eds.). (2010). Taxation and gender equity: A comparative analysis of direct and indirect taxes in developing and developed countries. Routledge. Hanmer, Lucia Eliana Rubiano, Julieth Santamaria & Diana J. Arango (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. Horrell, S., & Krishnan, P. (2007). Poverty and Productivity in Female-Headed Households in Zimbabwe. 43(8), 1351–1380. Iyengar, R., & Ferrari, G. (2016). African successes (S. Edwards, S. Johnson, & D. N. Weil, Eds.). The University of Chicago Press. Lanjouw, P., & Ravallion, M. (1995). Poverty and Household Size. The Economic Journal, 105(433), 1415. Klasen, S. (2019). What Explains Uneven Female Labor Force Participation Levels and Trends in Developing Countries? The World Bank Research Observer, 34(2), 161–197. 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. Milazzo, A., & van de Walle, D. (2017). Women Left Behind? Poverty and Headship in Africa. Demography, 54, 1119–1145. Ñopo, H., Daza, N., & Ramos, J. (2011). Gender Earnings Gaps in the World (IZA DP No. 5736; IZA Discussion Paper Series). OCHA. (2019). Gender in Humanitarian Action. OCHA on Message. https://www.unocha.org/sites/unocha/files/OOM%20Template%20gender%20- %2005Aug2019%20%28002%29.pdf Palmer-Jones, R., & Jackson, C. (1997). Work intensity, gender and sustainable development. Food Policy, 22(1), 39–62. Pape, U.J., Sharma, A., Nezam, T., Petrini, B.; Ebrahim, M.F., Udo-Udo, J., Appler, F. K., Woodhouse, A. F.; Phipps-Ebeler,V., Meckelburg, A. B.; Iqbal, S. A. (2019). Using Micro-Data to Inform Durable Solutions for IDPs (Vol. 2): Volume B : Country Case Studies (English). Washington, D.C. : World Bank Group. Pape, U. J., Sharma, A., Kaplan, L. C., Walsh, J. S. (2019). Using Micro-Data to Inform Durable Solutions for IDPs (Vol. 3) : Volume C : Technical Aspects (English). Washington, D.C.: World Bank Group. Perrin, N., Marsh, M., Clough, A., Desgroppes, A., Yope Phanuel, C., Abdi, A., Kaburu, F., Heitmann, S., Yamashina, M., Ross, B., Read-Hamilton, S., Turner, R., Heise, L., & 35 Glass, N. (2019). Social norms and beliefs about gender based violence scale: A measure for use with gender based violence prevention programs in low-resource and humanitarian settings. Conflict and Health, 13(1), 6. Refugees International. (2019). Durable Solutions in Somalia: Moving from Policies to Practice for IDPs in Mogadishu [Field Report]. Rogan, M. (2013). Alternative Definitions of Headship and the ‘Feminisation’ of Income Poverty in Post-Apartheid South Africa. Journal of Development Studies, 49(10), 1344– 1357. Rubiano-Matulevich, E., & Viollaz, M. (2019). Gender Differences in Time Use: Allocating Time between the Market and the Household (Working Paper No. 8981; Policy Research Working Papers). The World Bank. UN Women. (2018, September 18). Executive Director remarks at the United Nations Security Council meeting on Somalia. News and Events. https://www.unwomen.org/en/news/stories/2018/9/speech-ed-phumzile-unsc-meeting-on- somalia UNDP. (2018). Somalia Drought and Needs Assessment. https://www.undp.org/publications/somalia-drought-impact-and-needs-assessment United Kingdom: Home Office. (2017). Country Policy and Information Note—Somalia (South and Central): Security and humanitarian situation. https://www.refworld.org/docid/5979e68c4.html United Nations. (2020, June 3). Somalia. Office of the Special Representative of the Secretary- General on Sexual Violence in Conflict. https://www.un.org/sexualviolenceinconflict/countries/somalia/ United Nations General Assembly. (2018). Intensifying global efforts for the elimination of female genital mutilation (Seventy-Third Session) [Report of the Secretary General]. https://undocs.org/en/A/73/266 United Nations High Commission for Refugees [UNHCR] 2021 Global Trends in Forced Displacement 2020. https://www.unhcr.org/flagship-reports/globaltrends/ United Nations Office for the Coordination of Humanitarian Affairs [OCHA] & Reach Initiative, 2019 Severity of needs in IDP-hosting districts Somalia, Detailed Site Assessment. September 2018 - January 2019. Originally published 31 January 2019. https://reliefweb.int/report/somalia/severity-needs-idp-hosting-districts-detailed-site- assessment-somalia-september-2018 van de Walle, D. (2013). Lasting Welfare Effects of Widowhood in Mali. World Development, 51, 1–19. Warsame, A. (2004). Crisis or Opportunity? Somali Women Traders and the War. In J. Gardner & J. E. Bushra (Eds.), Somalia—The Untold Story: The War Through the Eyes of Somali Women. Pluto Press. 36 World Bank. (2011). World Development Report 2012: Gender Equality and Development. The World Bank. World Bank Group. (2014). Analysis of Displacement in Somalia. Washington, DC. World Bank. (2017). Forcibly displaced: Toward a development approach supporting refugees, the internally displaces, and their hosts. World Bank Group. World Bank. (2018a). Country Partnership Framework for the Federal Republic of Somalia for the Period FY19-22. World Bank Group. World Bank (2018b). Federal Republic of Somalia: Somalia Economic Update. Rapid Growth in mobile money: Stability of Vulnerability? August. Macroeconomics, Trade and Investment Africa, World Bank Group. https://documents1.worldbank.org/curated/en/975231536256355812/pdf/REPLACEMEN T-PUBLIC-Somalia-Economic-Update-3-FINAL.pdf World Bank. (2019). Shock Responsive Safety Net for Human Capital Project (Project Appraisal Document PAD3421). World Bank. https://documents1.worldbank.org/curated/en/989731565575232490/pdf/Somalia-Shock- Responsive-Safety-Net-for-Human-Capital-Project.pdf Zou, H., & Hastie, T. (2005). Regularization and variable selection via the elastic net. Journal of the Royal Statistical Society: Series B (Statistical Methodology), 67(2), 301–320. 37